Development of Nominal Inflection in First Language Acquisition: A Cross-Linguistic Perspective 9783110217117, 9783110188400

Citation preview

Development of Nominal Inflection in First Language Acquisition

≥

Studies on Language Acquisition 30

Editor Peter Jordens

Mouton de Gruyter Berlin · New York

Development of Nominal Inflection in First Language Acquisition A Cross-Linguistic Perspective

Edited by Ursula Stephany Maria D. Voeikova

Mouton de Gruyter Berlin · New York

Mouton de Gruyter (formerly Mouton, The Hague) is a Division of Walter de Gruyter GmbH & Co. KG, Berlin.

앝 Printed on acid-free paper which falls within the guidelines 앪 of the ANSI to ensure permanence and durability.

Library of Congress Cataloging-in-Publication Data Development of nominal inflection in first language acquisition : a crosslinguistic perspective / edited by Ursula Stephany, Maria D. Voeikova. p. cm. ⫺ (Studies on language acquisition ; 30) Includes bibliographical references and index. ISBN 978-3-11-018840-0 (hardcover : alk. paper) 1. Languages, Modern ⫺ Inflection. 2. Grammar, Comparative and general ⫺ Inflection. 3. Grammar, Comparative and general ⫺ Noun. 4. Language acquisition. I. Stephany, Ursula. II. Voeikova, M. D. PB104.D48 2009 4151.95⫺dc22 2009020703

Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.d-nb.de.

ISBN 978-3-11-018840-0 ISSN 1861-4248 쑔 Copyright 2009 by Walter de Gruyter GmbH & Co. KG, D-10785 Berlin. All rights reserved, including those of translation into foreign languages. No part of this book may be reproduced in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Cover design: Sigurd Wendland, Berlin. Printed in Germany.

Contents List of abbreviations

vii

Contributors

ix

Introduction Ursula Stephany and Maria D. Voeikova

1

Early nominal morphology in Turkish: Emergence of case and number F. Nihan Ketrez and Ayhan Aksu-Ko¸c

15

Acquisiton of case and plural in Finnish Klaus Laalo

49

The acquisition of numeral classifiers and optional plural marking in Yucatec Maya Barbara Pfeiler

91

The early development of case and number in Estonian Reili Argus

111

The acquisition of case, number and gender in Croatian Melita Kovaˇcevi´c, Marijan Palmovi´c, and Gordana Hrˇzica

153

Acquisition of case and number in Russian Natalia Gagarina and Maria D. Voeikova

179

The emergence of nominal inflection in Greek Ursula Stephany and Anastasia Christofidou

217

The acquisition of number and case in Austrian German nouns Katharina Korecky-Kr¨oll and Wolfgang U. Dressler

265

The emergence of nominal number in Italian Sabrina Noccetti

303

Number morphology in Spanish first language acquisition Carmen Aguirre and Victoria Marrero

341

vi

Contents

Relations between the development of the category of nominal and verbal number in two French-speaking children Marianne Kilani-Schoch

371

Noun plurals in early Palestinian Arabic: A longitudinal case study Dorit Ravid and Rola Farah

411

Subject Index

433

List of abbreviations used in the present volume 1. Grammatical codes 1P 1S 2P 2S 3P 3S ABESS/ABE ABL ABS ACC ADESS/ADE ADJ ADV ALL ANI AOR ART ASS AUX CLFR COLL COM COMP COP DAT DC DEF DEM DET DIM DIST ELAT/ELA ERG ESS

1st person plural 1st person singular 2nd person plural 2nd person singular 3rd person plural 3rd person singular Abessive Ablative Absolutive Accusative Adessive Adjective Adverb Allative Animate Aorist Article Assurative Auxiliary Classifier Collective Comitative Completive Copula Dative Declension Class Definite Demonstrative Determiner Diminutive Distal Elative Ergative Essive

FEM FILL GEN ILL IMP INANI INC IND INDEF INESS/INE INF INSTR INSTRUC IPFV LIAIS LOC MASC MOD N NEG NEUT NOM NP NR NUM OBL ONO OPT PARTIT PASS PAST PERS PFV PL

Feminine Filler Genitive Illative Imperative Inanimate Incompletive Indicative Indefinite Inessive Infinitive Instrumental Instructive Imperfective Liaison Locative Masculine Modal Noun Negation/Negative Neuter Nominative Noun phrase Nominalizer Numeral Oblique case Onomatopoeia Optative Partitive Passive Past tense Person Perfective Plural

viii

List of abbreviations used in the present volume

POSS POSTPOS PREP PRES PRO/PRON PROG PROX PTL Q Q1 Q2 Q3 REL

Possessive Postposition Preposition/al Present tense Pronoun Progressive Proximal Particle Question particle 1st degree of quantity 2nd degree of quantity 3rd degree of quantity Relational

SF SFP

Sound feminines Sound feminine plurals SG Singular SM Sound masculines SUBJ Subjunctive TER/TERM Terminative TNS Tense TRANSL/TRL Translative TRANSNUM Transnumeral U Umlaut V Verb VOC Vocative

2. Other abbreviations ADS CDS CHI coll. CRO CS EST FINN Form. FR GER

Adult-directed speech Child-directed speech Child Colloquial Croatian Child speech Estonian Finnish Formulaic French German

GRK IT MLU MOT MSA PA PBF RUS SPAN TURK YUC

Greek Italian Mean length of utterance Mother Modern Standard Arabic Palestinian Arabic Percentage of base forms Russian Spanish Turkish Yucatec Maya

*0 %mor %eng %sit

Ungrammatical omission Morphology/gloss English translation Situational context

3. Symbols & + > 0

Morphological fusion Compound marker Morpheme boundary Commission error Omission of the following morpheme

Contributors

Carmen Aguirre Universidad Complutense, Fac. Ciencias Informaci´on, Dpto. Lengua espa˜nola III AVDA COMpltr s/n 28040 Madrid, Spain [email protected]

Wolfgang U. Dressler University of Vienna, Department of Linguistics Austrian Academy of Sciences, Department of Linguistics and Communication Research Kegelgasse 27 A-1030 Vienna, Austria [email protected]

Ayhan Aksu-Ko¸c Yeditepe University, Department of Psychology, Kayı¸sda˘gı, Istanbul Bo˘gazi¸ci University, Department of Psychology Bebek, Istanbul 34342, Turkey

Rola Farah Tel Aviv University, School of Education, Dept. of Communications Disorders Tel Aviv 69978, Israel

[email protected]

[email protected]

Reili Argus Tallinn University, Institute of Estonian Language and Culture Narva mnt 2 Tallinn, Estonia

Natalia Gagarina ZAS Zentrum f¨ur Allgemeine Sprachwissenschaft Sch¨utzenstr. 18 D-10117 Berlin [email protected]

[email protected]

Anastasia Christofidou Academy of Sciences, Athens University of Athens, Pedagogical Department Solonos 84 GR-10680 Athens, Greece

Gordana Hrˇzica University of Zagreb, Laboratory for Psycholinguistic Research Zvonimirova 8, 10000 Zagreb, Croatia

[email protected]

[email protected]

x

Contributors

F. Nihan Ketrez Yale University, Near Eastern Languages and Civilizations Hall of Graduate Studies P.O. Box 208236 New Haven, CT 06520-8236, USA [email protected]

Klaus Laalo University of Tampere, School of modern languages and translation studies Academy of Finland, Research council for culture and society Maistraatinkatu 5 A 19 00240 Helsinki, Finland [email protected]

Marianne Kilani-Schoch University of Lausanne, School of French as a Foreign Language (EFLE) Anthropole CH-1015 Lausanne, Switzerland [email protected]

Victoria Marrero Universidad Nacional de Educaci´on a Distancia, Dpto. Lengua Espa˜nola y Ling. Gral. C/Senda del Rey, 7 28040 Madrid, Spain [email protected]

Katharina Korecky-Kr¨oll Austrian Academy of Sciences, Department of Linguistics and Communication Research Kegelgasse 27 A-1030 Vienna, Austria katharina.korecky-kroell@ oeaw.ac.at

Sabrina Noccetti University of Pisa, Dept. of English Studies Via Santa Maria, 67 I-56126 Pisa, Italy [email protected]

Melita Kovaˇcevi´c University of Zagreb, Laboratory for Psycholinguistic Research Zvonimirova 8, 10000 Zagreb, Croatia

Marijan Palmovi´c University of Zagreb, Laboratory for Psycholinguistic Research Zvonimirova 8, 10000 Zagreb, Croatia

[email protected]

[email protected]

Contributors

Barbara Pfeiler Universidad Nacional Aut´onoma de M´exico, Centro Peninsular en Humanidades y Ciencias Sociales Ex Sanatorio “Rend´on Peniche”, Calle 43 s/n, entre 44 y 46 Col. Industrial C.P. 97150 M´erida, Yuc., Mexico [email protected]

Dorit Ravid Tel Aviv University, School of Education, Dept. of Communications Disorders Tel Aviv 69978, Israel [email protected]

Ursula Stephany University of Cologne, Institute of Linguistics D-50923 Cologne, Germany [email protected]

Maria Voeikova Russian Academy of Sciences, Institute of Linguistic Studies Tuchkov per. 9 199053 Saint Petersburg, Russia [email protected]

xi

Introduction∗ Ursula Stephany and Maria D. Voeikova 1. Typological considerations of nominal inflection The present volume is a collection of studies on the early stages of the acquisition of nominal morphology and its most important grammatical categories, number and case. These categories are grammaticized in most languages studied in the present volume so that these represent a suitable sample for studying the influence of morphological typology on language acquisition. This allows both an intra-typological and a cross-typological analysis capturing the similarities of typologically different languages and the differences between typologically similar ones (Slobin 1997: 5; Voeikova 2002: 26). This volume is closely connected with a similar publication on the acquisition of verb inflection edited by D. Bittner, W. U. Dressler, and M. Kilani-Schoch (2003). Both studies result from work in the “Crosslinguistic Project on Preand Protomorphology in Language Acquisition” which has been coordinated by Wolfgang U. Dressler on behalf of the Austrian Academy of Sciences since 1993. In that research project, child language is studied from the point of view of the development of morphological oppositions in relation to the children’s input languages and their typological characteristics. A detailed investigation of the input in typologically different languages has shown that child-directed speech does not differ as strongly as adult-directed speech (Laaha and Gillis (eds.) 2007). The twelve languages included in this volume have eight genetic affiliations (table 1) and belong to different morphological types (see figure 1). The data come from longitudinal case studies of children from the first (or second) half of their second year of life well into their third (or even fourth or fifth) year.1 The development of contrasts between inflectional forms of nouns will be studied in the light of typological differences between languages and the influence of morphological richness on the order and speed of acquisition of inflectional patterns (see also Bittner, Dressler, and Kilani-Schoch 2003: xvi–xix). In spite of the fact that the categories of case and number may both be expressed inflectionally, they differ functionally in important ways: while the function of number is mainly referential, that of case is rather semanto-syntactic and therefore strongly relates to the structure of the sentence. With the exception of Yucatec Maya, which mainly follows the ergative system, all languages studied in the present volume belong to the syntactic type of nominative-accusative

2

Ursula Stephany and Maria D. Voeikova

Table 1. Languages and their genetic affiliations Language families Indo-European

Finno-Ugric Turkic Semitic American Indian

Branches Germanic Romance Slavic Greek

Languages German French, Italian, Spanish Croatian, Russian Greek Estonian, Finnish Turkish Palestinian Arabic Yucatec Maya

languages. The studies of the individual languages concentrate on case and number synthetically marked on the noun rather than on determiners. The analysis of case is limited to morphologically marked case, so that the expression of ‘abstract case’ or case roles by syntactic means such as word order has not been considered. Since this study is mainly concerned with the early inflectional development of the noun and languages differ regarding the role played by grammatical agreement (e.g. within the noun phrase), varying degrees of attention are given to agreement in the different language chapters. As far as richness of grammatical categories in the nominal domain is concerned, there are more extensive differences to be found in the case systems than in the number systems of the languages studied. Some of the most important parameters by which the systems of nominal inflection of the languages studied may be compared are the following: fusion vs. agglutination and typical number of inflectional markers per noun form (zero, one, several); root- vs. stem-based inflection and as a consequence addition vs. substitution of inflectional markers; gender classes and declensional classes; degree of syncretism of case forms; stem alternation and phonological form of markers (vowel vs. consonant, one syllable vs. more than one syllable). Important differences among languages consist in the number of cases and the distinction of grammatical vs. semantic cases. As far as the category of number is concerned, numeral languages are opposed to transnumeral ones, with a consequent distinction between obligatory and optional number marking (see Corbett 2000). Both the difference between numeral and transnumeral languages and that between two- and three-term plural-marking systems (singular/plural vs. singular/dual/plural) will play a minor role in our sample of languages (for details see Voeikova 2002; Stephany 1998, 2002). As far as noun morphology is concerned, the languages in our sample may be assigned to a scale of ideal language types reaching from agglutinating to isolating ones with an intermediate position of inflecting-fusional languages

Introduction

3

agglutinating strongly inflecting weakly inflecting isolating TURK–FINN–YUC–EST–CRO–RUS–GRK–GER–IT/SPAN–FR Figure 1. Typological order of noun inflection

(figure 1) (see Dressler 2007: 5).2 Palestinian Arabic, which follows the nonlinear root inflecting pattern as well as the linear suffixing one, cannot easily be placed within this scale. Although Turkish, as well as Finnish,Yucatec Maya, and Estonian all tend towards the agglutinating pole, Turkish does so most distinctly, while Estonian has historically moved towards the inflecting-fusional type (for Finnish and Turkish see P¨ochtrager et al. 1998). As pointed out by Dressler (2007: 4), strongly inflecting Croatian, Russian, and Greek “approach the ideal inflecting type fairly well, Croatian more so than Russian and Greek.” In contrast to these languages, the weakly inflecting ones have a less rich case system and typically mark grammatical relations periphrastically. As far as nominal inflection is concerned, Greek is intermediate between strongly and weakly inflecting languages since it expresses case distinctions both on the noun and the article. In the weakly inflecting languages tending toward the isolating type (Italian, Spanish, and French), nominal inflection is limited to number. In French, where the isolating tendency is strongest, number distinctions are almost entirely marked on the determiner rather than the noun. While agglutinating and strongly inflecting languages mark inflectional categories synthetically on the noun, weakly inflecting ones may also mark them periphrastically, i.e. on the determiner. As is typical for inflecting languages, nouns carry a single ending at most, which may express both case and number. This feature is characteristic of the fusional nature of these languages so that different grammatical categories are “fused together to give a single, unsegmentable morph” (Comrie 1981: 41). In contrast to this, agglutinating languages typically have several endings expressing separate grammatical categories. This may lead to longer inflected forms in agglutinating languages than inflecting ones. Another important feature bearing on language acquisition is that inflecting languages, in contrast to agglutinating ones, typically possess different inflectional patterns (partly depending on gender). There is a tendency for agglutinating languages to have a richer inventory of cases than even strongly inflecting ones. In the former language type, case endings tend to be unambiguous as far as the functions expressed are concerned, while in inflecting languages inflectional markers are often syncretic so that only their paradigmatic or syntagmatic context disambiguates their functions (e.g. Russian –u expresses ACC:SG in feminine nouns but DAT:SG in masculine ones). Another difference between

4

Ursula Stephany and Maria D. Voeikova

the languages is whether roots or bare stems may function as grammatical forms of nouns and whether noun forms typically carry an inflectional ending. In the former type, inflectional endings sometimes have to be added, whereas in the latter type they are typically substituted for each other (see Smoczy´nska 1985: 596–597). The presence of obligatory inflectional endings as opposed to their absence also influences the overall length of inflected forms.

2. The pre-/protomorphological approach to language acquisition The theoretical approach followed in most chapters of this book is the nonnativist and constructivist model of Pre- and Protomorphology applying basic parameters of Natural Morphology to the acquisition of inflection and derivation.3 Among these are transparency, uniformity, salience, and iconicity (Kilani-Schoch and Dressler 2005: 29–34; Dressler 2007: 5). One of the main tenets of the pre-/protomorphological approach is that inflection develops in three distinct stages, the premorphological stage, the protomorphological stage, and the stage of ‘morphology proper’.4 In the premorphological stage, word forms are rote-learned and usually occur in their base form (see below) with typically one form per lemma. If lexical and inflectional development of nouns is measured by the number of lexemes and grammatical types occurring at monthly intervals, the stage of premorphology will be characterized by a coincidence of these values, whereas in transition to protomorphology the number of forms will gradually exceed that of lemmas. The protomorphological stage manifests itself by the emergence of grammatical oppositions which develop into miniparadigms defined as follows: “A miniparadigm is a non-isolated set of minimally three phonologically unambiguous and distinct inflectional forms of the same lemma produced spontaneously in contrasting syntactic or situative contexts in the same month of recordings” (Bittner, Dressler, and Kilani-Schoch 2003: xxxix). Within the domain of miniparadigms only those functions which are formally distinguished are taken into consideration. During this stage children construct morphological patterns by analogy and evidence of first inflectional rules may be found in their speech (see Bittner, Dressler, and Kilani-Schoch 2003: xix). Generalizations based on early grammatical forms lead to morphological productivity (see below). Passing into the stage of morphology proper the children’s systems “approach qualitatively, if not quantitatively, the adult models” (Bittner, Dressler, and Kilani-Schoch 2003: xix).

Introduction

5

Insofar as it is functional and constructivist, the pre-/protomorphological approach is compatible with the usage-based approach to language acquisition (Tomasello 2003), in particular with Bybee’s network model of morphological organization and acquisition based on usage (Bybee 1988, 1991, 1995, 1998, 2001). A main difference between the pre-/protomorphological approach and the usage-based approach to morphology consists in the assumption of the development of symbolic rules in the former as opposed to schemas in the latter.

3. Percentage of base forms A useful measure for determining early inflectional development of nouns even before miniparadigms arise is the percentage of ‘base forms’ (PBF) occurring in the child’s utterances (Voeikova and Gagarina 2002). While oppositions and miniparadigms both belong to the paradigmatic axis of morphological structure, PBF measures the number of tokens of inflectional forms on the syntagmatic axis and determines the percentage of tokens representing unmarked inflectional forms of nouns on the basis of all forms of nouns. The base form of a noun is the form typically used by children as a default form of a given noun. Depending on the language this form is usually functionally unmarked coinciding with the citation form (e.g. nominative singular) in many languages and may have an inflectional marker or consist of the bare stem. While in Turkish the base form is a bare noun stem without an inflectional marker (see Ketrez and Aksu-Ko¸c, this volume), in Russian base forms may carry an inflectional marker, whereas some functionally ‘marked’ forms lack it (e.g. mam-a ‘mummy-NOM:SG’ vs. mam ‘mummy:GEN:PL’). While in Russian the nominative singular has to be considered as the base form of nouns across genders and declension classes, the situation is rather different in Greek, where the base form corresponds to the accusative singular. Although the latter differs from the citation form with masculine nouns, it is the formally least marked form across genders and declension classes and is therefore used as a default form by young children. In early language acquisition, base forms often serve multiple functions and differ from the standard language by underdifferentiation of grammatical categories (see Stephany 1997: 221). The percentage of base forms (PBF) determined at monthly intervals gradually diminishes during the course of development approaching the level of child-directed speech. While the use of (marked) plural forms of nouns depends on semantics or pragmatics, the use of other case forms besides the base form is syntactically determined. Since the plural is less frequent than (singular) case forms, PBF reflects the development of case rather than that of number.

6

Ursula Stephany and Maria D. Voeikova

4. Productivity During the stage of protomorphology, inflectional processes begin to become productive. Following Dressler (2003) productivity is “the ability to form new potential words” (either lexemes or grammatical forms) (quoted from Laaha et al. 2006: 279). A similar view is found in Bybee (2001: 12–13), where productivity is defined as “the extent to which a pattern is likely to apply to new forms (e.g., borrowed items or novel formations).” According to this author the productivity of a pattern is “largely, though not entirely, determined by its type frequency” (p. 13), i.e. by the number of lexical items that it applies to.5 Productivity forms a scale distinguishing several degrees (Dressler 2003; Laaha et al. 2006: 280–281). In the present volume, productivity is used to describe children’s creative use of grammatical forms. Furthermore, it is important to note that the child can only access productivity of inflectional affixes or inflectional patterns by the input received. Therefore, the productivity of the inflectional patterns of a given language may affect the order of acquisition of inflectional patterns. Productivity of grammatical forms in child speech is taken to demonstrate that inflected forms have reached the status of constructions for the child and are not simply repeated from rote memory as unanalyzed wholes. Overgeneralizations, although these are not usually plentiful, are the clearest evidence of inflectional creativity. Thus, the Russian form bl´ın-a for blin-´a ‘pancakeGEN:SG’ in the speech of Gvozdev’s son is used without the obligatory stress shift to the ending and follows the pattern stul/st´ul-a ‘chair:NOM:SG/-GEN:SG’ (Gvozdev [1949] 2007). Other criteria of productivity have been devised by Pizzuto and Caselli (1994: 156) and Gathercole, Sebasti´an, and Soto (1999: 144). These concern the number of distinct inflected forms of the same lemma or the number of lemmas used with the same inflection (see also Ketrez and Aksu-Ko¸c, this volume).

5. The early development of noun inflection in the different languages The chapters included in the present book will be presented in the order of position on the typological scale (see figure 1 above), starting with typically agglutinating Turkish and leading to French, which tends towards the isolating type. The chapter on Palestinian Arabic comes last. The chapter Early nominal morphology in Turkish: Emergence of case and number by F. N. Ketrez and A. Aksu-Ko¸c describes the early development of

Introduction

7

case, possessive, and number of the noun and pronoun in a monolingual girl between 1;3 and 2;0 years of age.The theoretical framework of the study is that of Natural Morphology and the pre- and protomorphological model of acquisition and the results give evidence for the three periods of inflectional development postulated in this model. The growing productivity of inflectional means in the child’s language is captured by a combination of different developmental measures such as the ratio of word forms per lemma, Percentage of Base Forms (PBF), and number and size of miniparadigms. There is clear empirical evidence that the category of case develops before number in Turkish nouns. The fact that the acquisition of pronominal inflection differs from that of the noun shows that grammatical development does not occur across-the-board in the nominal domain even in a language with a highly regular agglutinating morphology such as Turkish. The study of the Acquisiton of case and plural in Finnish by K. Laalo is mainly based on diary data from a girl and her younger brother, supplemented by tape-recordings from 1;7 to 2;10 for the girl and to 2;2 for the boy. Some rote-learned and adverbial-like first case forms occur extremely early in the speech of these children, namely before 1;0. When noun forms become more numerous, each of the children limits itself to one of the two patterns of Finnish case marking, either consonant gradation or suffixing, which may lead to a deviant combination of suffixes and stems. These individual strategies show children’s creativity as early as the middle of their second year. Both children pass through a ‘trochaic stage’ where they tend to shorten word forms to disyllabic ones. As a result of this, distorted stems may either occur with inflectional endings or undergo consonant gradation. A low PBF value, demonstrating a high percentage of diversely inflected forms in the children’s speech, and accordingly first oppositions and miniparadigms are found rather early. In her chapter on the The acquisition of numeral classifiers and optional plural marking in Yucatec Maya, the only transnumeral language studied in this volume, B. Pfeiler analyzes the data of two children acquiring Yucatec Maya as their native language: a girl and a boy observed from 1;11 to 4;9 and from 1;2 to 3;0 respectively. Since in Yucatec Maya plural marking of nouns is mainly optional and depends on the animacy hierarchy, the question arises whether plural marking develops in a different way from what is found in numeral languages, in which marking of plural reference is obligatory for most nouns. A theoretically important finding of the analysis of the development of animate and inanimate numeral classifiers and the plural suffix in possessive and nonpossessive constructions with animate or inanimate referents is that the children tend to mark number early and frequently, rather than to follow the infrequent singular and plural marking found in the input. This is especially true for the

8

Ursula Stephany and Maria D. Voeikova

girl, who, in contrast to the boy, even marks non-possessive inanimate nouns for plural, which is ungrammatical in the adult language. The author explains the fact that with the two Mayan children studied the plural develops in much the same way as with children acquiring numeral languages by drawing on the cultural environment in which Yucatec Mayan children grow up and by the indirect contact of preschool children with Spanish. The study of The early development of case and number in Estonian by R. Argus is mainly based on the data of a boy who was tape-recorded from 1;7 to 2;8 and whose development is compared to that of another boy and a girl observed from 1;8 to 2;5 and 1;10 to 2;1 respectively. As has been observed for Finnish, first rote-learned case forms emerge very early (at about 1;0) also in Estonian. As in Finnish, the children’s preference for marking case either by stem variation (gradation) or by suffixing varies individually. While one boy uses suffix marking from early on, the other one needs six months to proceed from stem variation to suffixation. Just as with Finnish children, a trochaic stage is also found with Estonian children. Productive case formation manifesting itself also in the occurrence of miniparadigms starts about six months after rote-learned case forms first emerge in all three children. The chapter The acquisition of case, number, and gender in Croatian by M. Kovaˇcevi´c, M. Palmovi´c, and G. Hrˇzica is primarily based on the data of a girl observed from 1;3 to 2;8, which is compared to that of another girl and ˇ a boy, all three of whom are acquiring the Zagreb variety of the Stokavian dialect. The three children show roughly the same order of emergence of case forms of nouns in the singular, with many of these early forms being anchored in restricted contexts of use. The authors observe a gradual growth of the morphological structure of nouns by an accumulation of item-based grammatical forms, which favors a usage-based explanation. Rather than being an obstacle to acquisition, the morphological richness and complexities of the Croatian case and number system serve as a trigger for the early development of nominal inflection. The different grammatical forms of nouns encountered by the child gradually develop into item-based grammatical schemas, some of which become productive in the course of development. There is a parallelism between child and adult language especially as far as the number of grammatical types of nouns is concerned. Some methodological proposals for the analysis of scarce data from the earliest periods of child development are discussed. The investigation of the Acquisition of case and number in Russian by N. Gagarina and M. D. Voeikova concentrates on ways of measuring morphological development in the nominal domain. As Russian child language has been an object of studies since the early forties of the last century, much is known on the development of case and number morphology. The main observations

Introduction

9

resulting from previous studies coincide with what is found by the authors in the speech of four Russian middle-class children (three boys and a girl) growing up in Saint Petersburg. Irrespective of their biological ages, their speech production is analyzed during the nine months following the appearance of number and case oppositions. The ages of emergence of these oppositions vary between 1;7 and 1;8 with the two early talkers and between 2;0 and 2;2 with the two late talkers. As found in previous studies, the earliest case opposition, which takes quite some time to develop, is that between nominative and accusative. By contrast, the other oblique cases develop rapidly and somehow appear ‘in a bunch’. Another famous earlier finding originally due to Gvozdev is children’s preference for salient inflectional endings when they construct innovative forms by selecting from the set of suffixes occurring in different declensional patterns, a phenomenon called ‘inflectional imperialism’ by Slobin. Also, Russian children develop systemic relations in the inflectional domain earlier than their normative constraints. The present chapter is concerned with the relation of MLU to Percentage of Base Forms (PBF), a measure introduced by these authors in an earlier study. They furthermore study the relation of the number of inflectional oppositions in nouns to the percentage of new lemmas in a critical period of development. In their chapter on The emergence of nominal inflection in child Greek U. Stephany and A. Christofidou study five middle-class children growing up in Athens from the second half of their second year to the last part of their third year. Much as in the acquisition of Russian, children acquiring Greek must construct different inflectional patterns of nouns depending on gender classes so that case and number distinctions develop locally within subclasses of nouns. The question whether the category of case or number develops first can therefore not be answered globally. Since inflectional systematicities are local rather than general (e.g. case distinctions with masculine nouns, but number distinctions with neuter ones), the authors claim that the development of Greek nominal inflection is more adequately captured by emergent generalizations or schemas than by across-the-board symbolic rules. There is evidence that in the acquisition of Greek, gender classes emerge from inflectional classes because inflectional patterns of nouns develop within gender classes (or even semantic subclasses within these) so that inflectional classes can be taken to serve as overt markers of gender classes. Greek child and input data (including a comparison between child- and adult-directed speech) are studied both quantitatively and qualitatively. Quantitative measures applied to the Greek data are mean size of paradigm, miniparadigms and inflectional contrasts as well as PBF. The study of The acquisition of number and case in Austrian German nouns by K. Korecky-Kr¨oll and W. U. Dressler is based on the observational data

10

Ursula Stephany and Maria D. Voeikova

of a boy acquiring Austrian German (1;3 to 2;7) and is cast in the pre- and protomophology approach to morphological development. While input token frequency is most relevant in the early phases, principles of Natural Morphology play an ever-increasing role in later development, providing a more adequate framework for explaining the acquisition of German case and number morphology than dual-route models focussing on the concept of default. According to the authors the compatibility of the results of this study with usage-based models of acquisition will depend on the possibility to accommodate the relatively early extraction of rule-like patterns found in the child’s development. In contrast to other German children, the boy studied in the present paper does not use quantifiers as forerunners of inflectional plural markers. The difference in the speed of development between number and case is explained by the more continuous, predominantly synthetic marking of plurals in contrast to the less continuous, essentially periphrastic marking of case. S. Noccetti’s chapter The emergence of nominal number in Italian compares the development of number in the acquisition of the Tuscan variety of Italian by a boy (observed from 2;0 to 3;6) and a girl (observed from 1;7 to 3;3) showing how they move on from the use of isolated noun lemmas to the formation of morphological patterns and functional categories. The data is divided into the four developmental periods of premorphology, protomorphology I and II, and modular morphology on the basis of morphological and syntactic changes in the children’s speech. The premorphological stage is characterized by extragrammatical operations. There is a tendency found in both children to refer to plurality by using numerals or other quantifiers. Although some plural forms do occur, their functional status is unclear. In protomorphology I, an intermediate stage between pre- and protomorphology characterized by trochaic word forms, the contrast of singular and plural forms emerges. In protomorphology II the most frequent or transparent patterns of plural formation become productive being overextended to other noun classes. An interesting individual variation is registered in the beginning of protomorphology I: while the boy first expresses the number contrast by numerals (‘one’ vs. ‘three’, ‘six’, ‘many’) and does not yet oppose singular and plural forms of nouns, the girl starts with producing inflectional contrasts of certain nouns. The chapter Number morphology in Spanish first language acquisition by V. Marrero and C. Aguirre is based on the analysis of the longitudinal corpora of two monolingual Spanish children, a boy and a girl, recorded in everyday situations from the beginning of speech production to the age of 2;6. It is cast in the pre- and protomorphology approach to language acquisition. The development of the plural is studied in the nominal domain (noun, pronoun, adjective, determiners) as well as the verb. While plurality is considered to be primarily

Introduction

11

associated with the noun on the semantic level, it is secondary regarding those parts of speech which obey the laws of grammatical agreement. Accordingly, plural paradigms first develop in the noun followed by the other nominal parts of speech and the verb. Since, with the exception of the masculine definite articles, all nominal categories have the same plural suffixes, plural marking quickly extends from the noun to the other nominal categories. While the plural is expressed by a single marker per utterance in the protomorphological stage, redundant plural marking (agreement) only develops in the morphological stage properly speaking. In contrast to previous findings there is evidence for the simultaneous rather than the sequential development of gender and number agreement. In her contribution Relations between the development of the category of nominal and verbal number in two French-speaking children M. Kilani-Schoch studies the typological differences between number marking on the noun and the verb, i.e. periphrastic number marking by determiners and (marginal) synthetic marking with the noun vs. synthetic marking by suffixes and periphrastic marking by clitic subject pronouns with the verb. The study is based on the corpora of two girls observed from 1;6 to 3;8 and 1;4 to 2;11 respectively. Special attention is paid to the relative emergence of fusional synthetic techniques for expressing number, namely prenominal liaison and portmanteau forms with nouns and preverbal liaison and suppletive or amplified stems with verbs. The author draws on functional relevance, contextual determination as well as input frequency for explaining the interesting finding that, counter to the crosslinguistic tendency according to which synthetic techniques precede periphrastic ones in morphological development, periphrastically marked noun plural emerges before synthetically marked verb plural in French. Within each of the domains of nominal and verbal plural marking, less ambiguous forms tend to appear earlier than more ambiguous ones. The chapter on Noun plurals in early Palestinian Arabic: A longitudinal case study by D. Ravid and R. Farah is one of the few studies which have been done on the acquisition of Arabic and is the first longitudinal case study tracing the route of an Arabic-speaking child during a period of one year (1;8 to 2;8) in learning about the noun plural. The interest of the study lies in the complexity of the Arabic inflectional system and the importance of studying its acquisition observationally. Given the complexity of Arabic pluralization patterns, the authors point out that their acquisition is better accounted for by a single-route than a dual-route model making a simple regular/irregular split as is found in English. One of the most interesting results of the study is that plural marking is not acquired across-the-board but according to certain pluralization patterns and gender classes. The findings do not support the dual-route model of

12

Ursula Stephany and Maria D. Voeikova

morphological acquisition because two equally ‘regular’Arabic patterns do not develop in unison and acquisition does seem to be subject to frequency effects. Comparing the results of their observational study to those of experimental ones of Arabic plural acquisition, the authors stress the importance of the former type of studies for coming to grips with the process of language acquisition.

Notes ∗

We would like to thank Katherine Maye-Saidi for correcting our English. 1. Since the observational studies of early language development are limited to one or, at most, five children per language, variability in rate of development is not a main concern in the studies included in the present book. 2. A comparable scale for verbs can be found in Bittner, Dressler, and Kilani-Schoch 2003: xv. 3. Regarding the model of Natural Morphology see Kilani-Schoch and Dressler (2005) and the literature quoted there. 4. In the pre-/protomorphological approach these stages are also sometimes called ‘phases’. On the difference between stage and phase models of development see Karmiloff-Smith (1992). 5. For a different view on the relation between productivity and frequency see KilaniSchoch and Dressler (2005: 129–132).

References Bittner, Dagmar, Wolfgang U. Dressler, and Marianne Kilani-Schoch 2003 Introduction. In Development of Verb Inflection in First Language Acquisition: A Cross-Linguistic Perspective, Dagmar Bittner, Wolfgang U. Dressler, and Marianne Kilani-Schoch (eds.), vii–xl. (Studies on Language Acquisition 21.) Berlin/New York: Mouton de Gruyter. Bybee, Joan 1988 Morphology as lexical organization. In Theoretical Morphology, Michael Hammond and Michael Noonan (eds.), 119–141. New York: Academic Press. 1991 Natural morphology: The organization of paradigms and language acquisition. In Crosscurrents in Second Language Acquisition and Linguistic Theories, Thom Huebner and Charles A. Ferguson (eds.), 67–91. (Language Acquisition and Language Disorders 2.) Amsterdam/Philadelphia: John Benjamins. 1995 Regular morphology and the lexicon. Language and Cognitive Processes 10: 425–455.

Introduction 1998 2001

13

The emergent lexicon. Chicago Linguistic Society 34: 421–435. Phonology and Language Use. Cambridge: Cambridge University Press.

Comrie, Bernard 1981 Language Universals and Linguistic Typology: Syntax and Morphology. Oxford: Basil Blackwell. Corbett, Greville G. 2000 Number. Cambridge: Cambridge University Press. Dressler, Wolfgang U. 2003 Degrees of grammatical productivity in inflectional morphology. Rivista di Linguistica 15: 31–62. 2007 Introduction. In Typological Perspectives on the Acquisition of Noun and Verb Morphology, Sabine Laaha and Steven Gillis (eds.), 3–9. (Antwerp Papers in Linguistics 112.)Antwerp: University ofAntwerp. Gathercole, Virginia C. Mueller, Eugenia Sebasti´an, and Pilar Soto 1999 The early acquisition of Spanish verbal morphology:Across-the-board or piecemeal knowledge? The International Journal of Bilingualism 3: 133–182. Gvozdev, Alexandr N. 2007 Reprint. Formirovanije u rebenka grammaticheskogo stroja russkogo jazyka [The construction of the grammatical system of Russian by the child]. InVoprosy izuˇcenija detskoj reˇci [Questions of the study of child speech], Alexandr N. Gvozdev, 149–459. Saint Petersburg/Moscow: Detstvo-Press, Tvorˇceskij centr Sfera. Original edition, Moscow: Akad. Pedag. Nauk RSFSR, 1949. Karmiloff-Smith, Annette 1992 Beyond Modularity: A Developmental Perspective on Cognitive Science. Cambridge, MA/London: MIT Press. Kilani-Schoch, Marianne, and Wolfgang U. Dressler 2005 Morphologie naturelle et flexion du verbe fran¸cais. (T¨ubinger Beitr¨age zur Linguistik 488.) T¨ubingen: Gunter Narr. Laaha, Sabine, and Steven Gillis (eds.) 2007 Typological Perspectives on the Acquisition of Noun and Verb Morphology. (Antwerp Papers in Linguistics 112.) Antwerp: University of Antwerp. Laaha, Sabine, Dorit Ravid, Katharina Korecky-Kr¨oll, Gregor Laaha, and Wolfgang U. Dressler 2006 Early noun plurals in German: Regularity, productivity or default? Journal of Child Language 33: 271–302. Pizzuto, Elena, and M. Christina Caselli 1994 The acquisition of Italian verb morphology in a cross-linguistic perspective. In Other Children, Other Languages: Issues in the Theory

14

Ursula Stephany and Maria D. Voeikova

of Language Acquisition, Yonata Levy (ed.), 137–187. Hillsdale, NJ: Lawrence Erlbaum. P¨ochtrager, Markus, Csan´ad Bod´o, Wolfgang U. Dressler, and Teresa Schweiger 1998 On some inflectional properties of the agglutinating type illustrated from Finnish, Hungarian and Turkish inflection. Wiener Linguistische Gazette 62/63: 57–92. Slobin, Dan I. 1997 The universal, the typological, and the particular in acquisition. In The Crosslinguistic Study of Language Acquisition, Dan I. Slobin (ed.), vol. 5, 1–39. Mahwah, NJ/London: Lawrence Erlbaum. Smoczy´nska, Magdalena 1985 The acquisition of Polish. InThe Crosslinguistic Study of LanguageAcquisition, Dan I. Slobin (ed.), vol. 1, 595–686. Hillsdale, NJ/London: Lawrence Erlbaum. Stephany, Ursula 1997 The acquisition of Greek. InThe Crosslinguistic Study of LanguageAcquisition, Dan I. Slobin (ed.), vol. 4, 183–333. Mahwah, NJ/London: Lawrence Erlbaum. 1998 A crosslinguistic perspective on the category of nominal number and its acquisition. In Studies in the Acquisition of Number and Diminutive Marking, Steven Gillis (ed.), 1–23. (Antwerp Papers in Linguistics 95.) Antwerp: University of Antwerp. 2002 Early development of grammatical number: A typological perspective. In Pre- and Protomorphology: Early Phases of Morphological Development in Nouns and Verbs, Maria D. Voeikova and Wolfgang U. Dressler (eds.), 7–23. Munich: Lincom Europa. Tomasello, Michael 2003 Constructing a Language: A Usage-Based Theory of Language Acquisition. Cambridge, MA/London: Harvard University Press. Voeikova, Maria D. 2002 The acquisition of case in typologically different languages. In Preand Protomorphology: Early Phases of Morphological Development in Nouns and Verbs, Maria D. Voeikova and Wolfgang U. Dressler (eds.), 25–44. Munich: Lincom Europa. Voeikova, Maria D., and Natalia Gagarina 2002 Early syntax, first lexicon and the acquisition of case forms by two Russian children. In Pre- and Protomorphology: Early Phases of Morphological Development in Nouns and Verbs, Maria D. Voeikova and Wolfgang U. Dressler, (eds.), 115–131. Munich: Lincom Europa.

Early nominal morphology in Turkish: Emergence of case and number∗ F. Nihan Ketrez and Ayhan Aksu-Ko¸c In this paper we trace the emergence of nominal inflectional morphology in the speech of a monolingual child acquiring Turkish between 1;3 and 2;0. Having determined three phases in the development of early nominal morphology, we observe that the core cases accusative and dative emerge first, followed by those that encode semantically salient relations. We also look at number and possessive morphology and find that case morphology is prior to number morphology in acquisition, whereas the possessive is marked almost at the same time as case. We note that morphological development proceeds simultaneously in the nominal and verbal domains, resulting in the differentiation of the categories of nouns vs. verbs. We also present data from the input corresponding to the points in development that we discuss.

1. Case and number in Turkish Turkish is an agglutinating language where nominals can receive three types of inflectional marking: Case, number and possessive. In predicate position they also get tense-aspect-modality and subject-verb agreement markers.1 There is no grammatical gender in Turkish. Case, number and possessive marking are not fused.2

1.1.

Case

Case marking in Turkish is synthetic and stem based. There are six major cases: Nominative, accusative, dative, genitive, locative, and ablative. In addition,Turkish possesses a comitative/instrumental morpheme that we analyze as a marginal case. Except for the nominative, which has no phonological realization, cases receive distinct morphological marking on nouns, question words, pronouns, and nominalized forms of the verb or the adjective. Double case marking is not allowed.3 The case system consists of a single paradigm that is fully productive and regular. Suffixes undergo vowel or/and consonant harmony as shown in (1).

16

(1)

F. Nihan Ketrez and Ayhan Aksu-Ko¸c

Lemma NOM4 ∅ ACC -(y)I [(y) i/ı/¨u/u] DAT -(y)A [(y) e/a] GEN -(n)In [(n)in/ın/¨un/un/im]6 LOC -DA [de/da/te/ta] ABL -DAn [den/dan/ten/tan] INSTR/COM -(y)lA [(y)la/le]

kedi ‘cat’ kedi kedi-yi kedi-ye kedi-nin kedi-de kedi-den kedi-yle

ben ‘I’ ben ben-i ban-a5 ben-im ben-de ben-den benim-le

Phonetically homophonous forms in the inflectional paradigm are rare, but exist. The accusative and the 3rd person singular possessive have the same phonological shape (-i) on nouns ending in consonants, for example. The word bebe˘g-i means either baby-ACC (the baby) or baby-POSS&3S (somebody’s baby). Each morpheme is syllabic and the typical stress pattern is word-final though with fairly even distribution across syllables. When a new suffix is attached to the stem, the word-final stress passes onto the final syllable. The word-final position of the suffixes reinforced by word-final stress is likely to render case markers perceptually quite salient. When a stem has all three nominal inflections, the order is stem-number-possessive-case. There are exceptions where case precedes number or possessive. These are limited to nominals that have ki (masa-da-kiler table-LOC-ki-PL ‘those on the table’ and Ayse-nin-ki-si Ayse-GEN-ki-POSS ‘the one that is Ay¸se’s’) and their use is not productive.7 Before going into the details of the acquisition pattern of this system, let us have a closer look at the case markers, their forms and functions. The nominative designates the subject and the accusative a specific direct object. The semantic roles typically realized are actor/agent and undergoer/patient, respectively. The dative marks oblique objects and may express direction, goal, or beneficiary of an action. The control of the unmarked nominative can be inferred from the use of agreement markers on the verb unless the person is third person singular, which does not have an overt agreement marker.8 The nominative is also signaled by its contrastive use with the accusative (and dative) in a transitive sentence frame. Caseless bare forms are also used as non-specific direct objects (kitap oku- ‘book-reading’) or subjects (su bas- ‘flood-, literally, water-cover-’), so it is difficult to distinguish nominative case from caseless bare forms especially in one-word utterances. Due to this difficulty, and also because the nominative case is not a phonologically realized morpheme whose emergence can be traced, we leave aside the discussion of nominative as a case in our discussion of case morphology. The accusative case marks the direct object and it is obligatory when the object is referential and specific (definite or indefinite). Objects marked for

Early nominal morphology in Turkish: Emergence of case and number

17

accusative can freely scramble preverbally, for purposes of focusing, and postverbally, to express afterthoughts (Erguvanlı 1984). When the accusative marking is dropped, the noun in the object position obtains a non-referential or nonspecific reading (such as book-reading as opposed to reading the/a particular book). In such cases the object obligatorily appears in the position adjacent to the verb, i.e., in the immediately preverbal position. In other words, the accusative case on nouns can be dropped under particular semantic/pragmatic conditions in adult grammar and it is often difficult to judge whether a child is performing an omission error or referring to an object in a non-specific or generic sense. Accusative case marking is always obligatory in pronouns and in those nouns that have possessive morphology. The dative case marks the direction or the goal of the action especially when it is used with intransitive verbs such as git- ‘go.’ Similarly, it marks the indirect object of ditransitive verbs, such as ver- ‘give’and koy- ‘put’. In addition, when a transitive verb is causativized (e.g., let/make somebody do something), the agent of the causativized verb (somebody) is marked with the dative case. Some verbs such as -(y)A inan- ‘believe,’ -(y)A g¨uven- ‘trust’ obligatorily take dative-marked complements. The dative case is also required by some adjectives (e.g., -(y)A a¸sık- ‘in love with’ -(y)A ba˘glı ‘dependent on’) and post-positions (e.g., -(y)A kadar ‘until’ -(y)A g¨ore ‘according to’). When compared to other case markers, its function is less transparent and predictable, but it has high frequency due to various obligatory conditions. The locative indicates static location, typically in a peripheral adverbial construction specifying where the action takes place (e.g., ev-de ‘at home’).9 The ablative marks the source of the action (ev-den ‘from home’), sometimes incorporating a ’partitive’ reading (kek-ten ye- ‘eat from the cake’). Its functions are more transparent when compared to the other case markers. Although the ablative can appear in unpredictable relations with a few verbs, such uses are not observed in our subject’s speech. Examples are -DAn ho¸slan ‘like’, -DAn vazge¸c ‘give up’ and -DAn sıkıl ‘be bored (about)’; here, the subject is an experiencer and the ablative marks the source of experience.10 The genitive appears in genitive-possessive constructions such as Deniz-in kalem-i (Deniz-GEN&3S pencil-POSS&3S) ‘Deniz’s pencil,’ where the possessor is marked with the genitive case and the possessed with a personal suffix that agrees with the possessor. Also, the referential/specific subject of an embedded clause is marked by the genitive case (e.g., Deniz-in gel-di˘g-i-ni duy-du-m (Deniz-GEN&3S come-NR-POSS&3S-ACC hear-PAST-1S) ‘I heard that Deniz was/is coming’). The nominal subject that bears the genitive case agrees with the verb or the possessed in person and number; agreement is realized through the possessive marker.

18

F. Nihan Ketrez and Ayhan Aksu-Ko¸c

1.2. The status of comitative and instrumental “cases” The comitative and instrumental morphemes are the fused forms of the postposition ile ‘with,’ which can be written and pronounced as an individual word as well. So the forms listed in (1) above can also be expressed by kedi ile ‘with cat’ and benim ile ‘with me’ (instead of kediyle and benimle). Evidence for the postpositional analysis of -(y)lA comes from the fact that it cannot be attached to a bare pronoun; rather, it requires a genitive case marking (o-nunla ‘it-GEN-INSTR’ versus *o-yla ‘it-INSTR’11). Other than the occurrence of -(y)lA in such examples, Turkish does not allow ‘double’ case marking, thus the status of -(y)lA as a case is questionable. Another reason for the postpositional analysis is that when -(y)lA is attached to a word having word final stress, the latter is not passed onto -(y)lA, while such a stress shift is found with other case markers, i.e, they are stressed when they constitute the final syllable of the word (kit´ap > kitapt´a/kitab´a/kitab´ın ‘book-LOC/DAT/GEN’ vs. kit´ap > kit´apla/*kitapl´a ‘book-INSTR’). In these respects, -(y)lA behaves like other postpositions such as i¸cin ‘for’ or kadar ‘until,’ which also require case-marked nominals as their complements but do not result in stress shift as they are not a part of the word. One property of -(y)lA that would be problematic for the postpositional analysis is that it is required with some adjectives such as -(y)lA ilgili/alakalı ‘interested in/related to’ -(y)lA sorumlu ‘responsible for,’ -(y)lA sınırlı ‘restricted to’. Because of its behavior with adjectives, it can be considered similar to other case markers. In the present study, we will include -(y)lA in our analysis as it is a part of nominal paradigms, is produced as a bound morpheme, and results in commission errors with other case morphemes. We discuss it as a marginal case, however, to stress the difference between -(y)lA and other cases, and leave a detailed discussion of post-positions versus case distinctions in Turkish for a further study.

1.3.

Semantic/pragmatic functions

Blake (1994: 1) defines case as follows: “. . . typically, case marks the relationship of a noun to a verb, at the clause level, or of a noun to a preposition, postposition or to another noun at the phrase level.” Observing these distinctions we refer to (1) cases that mark the relationship between a verb and its nominal arguments as relational cases, and to (2) cases that relate a noun to a postposition or to another noun at the phrase level as internal cases. Within this latter category, case assignment may be motivated (a) semantically (e.g. the locative is typically associated with the notion of ‘static location’), and/or (b) structurally, like

Early nominal morphology in Turkish: Emergence of case and number

19

the genitive called for by the genitive-possessive construction or by nominal complements in Turkish. The two relational cases, the nominative and the accusative, specify the roles of the core arguments of ‘actor’ and ‘undergoer’. With ditransitive verbs, the dative also functions as a relational case, marking the role of ‘experiencer/ theme/locative’ (goal, recipient, beneficiary, causee) (Haig 1997). It functions as an internal case to encode ‘locative-goal’ (Nilsson 1985). The locative and the ablative designate ‘static location’ and ‘locative-source’, respectively; they have a relational function only with a few verbs (e.g., kal- ‘remain’, kork- ‘be scared’). The comitative or instrumental are noun-internal cases the assignment of which is determined by the semantic role the NP occupies.

1.4.

Case marking in the input

Given the highly variable word-order conditioned by pragmatic factors (Erguvanlı 1984), position cannot serve as a cue for syntactic relations in Turkish so case marking is the most important syntactic mechanism. As noted above, the system is transparent, simple, regular, and consistent and presents the same form-meaning correspondences pervasively over all kinds of nominals. All these characteristics suggest that case markers would have very high frequency in the input. However, the language also allows for nominal ellipsis which would affect frequency negatively; given subject-verb agreement, subject eliding is the norm rather than the exception (Aksu-Ko¸c and Slobin 1985; K¨untay and Slobin 1996, 2001). A balance between these opposing forces of consistency, clarity and pervasiveness on the one hand, and relatively low frequency on the other should determine the ease and timing of acquisition. Our analysis of the input shows that, up to age 2;0, only about 54% of nominals carry an inflectional marker, while 46% are bare forms (Aksu-Ko¸c et al. 2005). Since these figures do not include pronouns, which are always inflected in object role, the frequency of inflected forms is presumably higher. It can therefore be assumed that Turkish CDS, at least as it is represented in the speech of our subject’s mother, does not present impoverished data in terms of the frequency of nominal inflections. The issue of frequency will be taken up in relation to different case markers and number marking in the following sections.

1.5.

Number

Number on nouns (including proper names), demonstrative pronouns, and question words is morphologically marked by the suffixation of the plural morpheme

20

F. Nihan Ketrez and Ayhan Aksu-Ko¸c

-lAr 12 after the stem and before the possessive and case markers, e.g., kedi ‘cat,’ kedi-ler (cat-PL) ‘cats,’ kedi-ler-de (cat-PL-LOC) ‘on/in cats,’ kedi-ler-im (cat-PL-POSS) ‘my cats,’ kedi-ler-im-de (cat-PL-POSS-LOC) ‘on my cats.’ -lAr does not appear in the presence of numerals and other plural quantifiers (e.g., iki kedi-∅ (two cat) ‘two cats’). In other words, there is no agreement relation in Turkish between quantifiers and the noun as far as the grammatical category of number is concerned. Although the plural morpheme is attached directly to the stem, preceding the case and possessive markers in the majority of structures, there are some exceptions to this order. In the word kedi-de-ki-ler (cat-LOC-ki-PL) ‘those that are on the cat’ where we have ki, the plural morpheme follows the locative case marker. The reason is that ki can be considered a pronominal marker so that the plural morpheme pluralizes the noun pronominalized with ki. In another example, the plural marker follows the possessive anne-m-ler-i (mother-POSSPL-ACC) ‘my mother and others.’ Such examples are restricted to some kinship terms. The meaning of the structure is different when the ordering of possessive and plural is reverse as in anne-ler-im-i ( mother-PL-POSS-ACC), which means ‘my mothers.’ In the last example, the scope of the plural marker is limited to anne ‘mother’ since it is immediately attached to the noun stem. In this section, we presented the adult nominal morphology system and showed that it is rule-based and salient. Exceptions to the rules are rather rare. In the following sections, we will present the pattern of the emerging nominal morphology in our subject and will subsequently trace the points in time and order of emergence of case markers in their specific syntactic and semantic functions.

2. Subject and method This study is based on the spontaneous speech samples of a monolingual Turkish child, Deniz, recorded longitudinally between the ages 1;3.3 and 2;0.4. Recordings were done approximately twice a month and each of them was about 20 minutes long. During the sessions, the child was engaged in various natural everyday activities. All the child utterances were transcribed and coded morphologically according to the CHAT conventions of CHILDES (MacWhinney 2000). All nouns, pronouns, and proper nouns that can potentially bear case and number morphology were targeted for analysis and analyzed both quantitatively and qualitatively. The details of the subject’s recordings and the amount of speech collected are presented in table 1. For the input, three sessions corresponding to three developmental phases that will be discussed below were sampled.

Early nominal morphology in Turkish: Emergence of case and number

21

Table 1. Age, MLU, total number of morphemes and utterances 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

age 1;3 1;4 1;5 1;6 1;7 1;8

1;9

1;10 1;11 2;0

1;3.3 1;3.12 1;3.27 1;5.9 1;5.28 1;6.9 1;7.3 1;7.8 1;7.23 1;8.11 1;8.14 1;8.27 1;9.1 1;9.19 1;10.3 1;10.19 1;11.10 1;11.23 2;0.4

MLU (in morphs) no. of morphs no. of utterances 1.26 117 98 1.34 101 75 1.24 96 77 1.20 105 87 1.58 217 137 1.73 192 111 2.53 639 252 1.95 317 162 2.74 548 200 2.93 838 286 3.03 570 188 3.42 938 274 3.28 627 191 2.01 517 148 3.81 1339 351 3.52 448 127 3.64 1414 388 3.17 1306 412 4.06 1076 265

In analyzing the child’s speech, productivity was evaluated for each morpheme by looking at its presence in obligatory contexts according to the following criteria (See Brown 1973; Allen 1996; Kilani-Schoch and Dressler 2002 for the use and discussion of similar criteria): (a) a given morpheme should occur with at least two different nouns, (b) the noun should occur in at least two contrasting forms (either marked vs. unmarked or with two different markings), (c) omission or commission errors should not exceed the number of grammatical productions of the morpheme, (d) it should occur in two successive samples. The definition of productivity above, though different from what is understood by adult-like productivity, provides a criterion for detecting the transition from use of uninflected or rote learned forms to use of productive inflectional morphology and for tracing the changes occurring in this early phase of development. This notion of productivity does not mean that the child has acquired adult-like productivity when use of a certain morphological form satisfies all of these requirements, where inflectional morphemes appear in all grammatically obligatory contexts, since, as Brown has noted, “the constraints that define obli-

22

F. Nihan Ketrez and Ayhan Aksu-Ko¸c

gation are themselves acquired over time” (1973: 257). However, it means that morphological processes are applied in a rule-governed way given the scope of lexical and grammatical structures the child has. In line with Bybee’s (2001: 12–13) definition of productivity as “the extent to which a pattern is likely to apply to new forms” we trace the productivity of an inflection “by its type frequency: The more items encompassed by a schema, the stronger it is, and the more available it is for application to new items.” Given that the Turkish inflectional system is almost without exceptions, our subject’s use of grammatical morphemes with new nominal items – excluding derived lexical classes or complex syntactic formations such as complement constructions – appears to meet this limited definition of productivity by 2;0 years of age.

3. Demarcation of phases The analysis was carried out in terms of the three phases of pre-morphology, proto-morphology and morphology proper within the theoretical framework of Natural Morphology (Dressler (ed.) 1997). We were able to find evidence for the three phases of morphological development predicted by this theory in the speech of our subject. The pre-morphological phase has been found to extend from 1;3, at the beginning of observation, through 1;5 (MLU:1.20–1.34). The proto-morphological phase starts at 1;6 and ends at 1;9 (MLU:1.58–3.42). After 1;10 (MLU: 3.17–4.06), our subject’s inflectional morphology of the noun reaches what can be called a productive level according to the above criteria, and derivational morphology begins to emerge. These demarcation points correspond approximately to those observed in the development of verb inflection in the speech of the same child (Aksu-Ko¸c and Ketrez 2003), providing evidence for the simultaneous development of the inflectional morphological system in the nominal and verbal domains. The demarcation of phases has been done on the basis of the increasing inflectional variety observed in our subject’s speech. Inflectional variety is determined by an analysis of (1) the number of lemmas in comparison to the number of types and tokens of word forms at monthly intervals, (2) the number and size of “mini-paradigms” (Bittner, Dressler, and Kilani-Schoch 2003), (3) the number of inflectional morphemes attached to a given nominal stem, and (4) the proportion of bare stem forms (PBF) versus inflected forms in the child’s speech. This analysis in terms of phases is not a claim for sharp transitions with sudden emergences. Rather, it allows for characterizing the child’s speech in terms of systematic changes – reflecting both discontinuities and continuities – that occur within a grammatical domain across time.

Early nominal morphology in Turkish: Emergence of case and number

3.1.

23

Lemmas and grammatical types

The number of nominal lemmas and the grammatical forms they occur in (types) are shown in fig. 1 and table 2 for each month of observation. In the figure, simple nouns, proper nouns and pronouns have been merged to present a more general picture. In the table, pronouns and nouns are reported separately while proper nouns are included in the category of nouns. As can be seen in the figure, the number of lemmas and types overlap until the age of 1;6, as there are just a few lemmas that appear in more than one form. From 1;6 on, the number of lemmas as well as that of grammatical types increase. What is more important is that, due to a rise in the variety of inflected forms, the number of grammatical types grows much more rapidly than that of lemmas. After 1;9, this rise of grammatical types ceases and the distance between types and lemmas remains stable. The decrease in both lemmas and grammatical types observed at 1;11 can be attributed to the lower amount of data recorded in comparison with the sample at 2;0 and is thus not developmentally significant (see table 1). During the whole period we analyze, the ratio of different inflected word-forms to different lemmas in the mother’s speech directed to Deniz is equal to 1.91, indicating that the child as yet falls short of the adult in terms of the mean size of her paradigms for nouns. However, a comparison of the development of grammatical variation in the noun and the pronoun reveals that variation increases more rapidly in the pronoun than in the noun. For the pronoun, the type/lemma ratio rises from 1.0 in the pre-morphological phase to 1.29 at 1;6. There is a notable increase between 1;7 and 1;9, ranging from 2.25 to 3.55 and stabilizing around 3.00–3.50 after this age. The early productivity achieved in case marking is thus at first supported by pronouns that have only a restricted number of lemma-types and that occur with high frequency.

lemma & type

Lemma Type 200 100 0 1;3

1;4

1;5

1;6

1;7

1;8 age

Figure 1. Lemmas and grammatical types

1;9

1;10 1;11 2;0

24

F. Nihan Ketrez and Ayhan Aksu-Ko¸c

Table 2. Lemmas, grammatical types, tokens and type/lemma ratios Phase Premorphology

Protomorphology

Morphology Proper

age 1;3 1;4 1;5 1;6 1;7 1;8 1;9 1;10 1;11 2;0

N/PRO: N/PRO: N/PRO: N/PRO: N/PRO: N/PRO: N/PRO: N/PRO: N/PRO: N/PRO:

lemma type 30/2 32/2 24/0 24/0 26/2 30/2 55/7 61/9 82/8 99/18 83/11 111/2 85/9 127/32 73/13 99/38 64/9 92/27 97/12 146/44

token type/lemma 161/2 1.06/1.00 70/0 1.0/0 73/2 1.15/1.00 174/23 1.11/1.29 278/62 1.21/2.25 287/321 1.34/2.91 239/284 1.49/3.55 301/278 1.36/2.92 172/235 1.44/3.00 325/357 1.50/3.67

The overview of the type/lemma ratio presented in table 2 points to three phases of morphological development: In the first phase, there is lack of variation; in the second phase, increasing variation is due to growing variability of grammatical types which results in instability; finally, in the third phase, variation is low again, reflecting relative stability. In the following sections we will look at the same picture from other angles but reach the same conclusion as far as the borderlines of morphological phases are concerned.

3.2.

Mini-paradigms

Details concerning the size of mini-paradigms are indicated in table 3. Until 1;6, there are only two 2-member mini-paradigms which, however, consist of a single example each of two different lemmas marked for accusative and dative and which do not reappear in the subsequent months. According to our criteria above, they do not constitute evidence for productivity. Nor does Deniz use any other inflectionally marked nouns during this period (see table 4). It is only at the age of 1;6 that we find evidence for the emergence of inflected forms and mini-paradigms. From then on, there is an increase not only in the overall number of multi-member paradigms, but also in the number of members of the individual paradigms. This number rises to nine members in one and the same paradigm and to 38 multi-member paradigms overall at the age of 1;9. We consider this to be the peak and the end of the proto-morphological period since all nominal inflections – number, possessive, and case – have emerged and gained some level of productivity by this age.13 The peak reached at 1;9 is not exceeded until 2;0, the end of the period studied. Table 5 presents some

Early nominal morphology in Turkish: Emergence of case and number

25

Table 3. Number of mini-paradigms and number of paradigm members Phase Premorphology

Protomorphology

Morphology Proper

age 1;3 1;4 1;5 1;6 1;7 1;8 1;9 1;10 1;11 2;0

Number of paradigm members 2 3 4 5 6 7–9 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 4 0 1 0 0 0 7 5 3 1 0 0 18 4 1 1 2 1 24 7 4 2 0 1 17 5 4 1 2 0 11 2 3 2 0 2 24 6 4 1 1 3

total 2 0 0 5 16 27 38 29 20 39

Table 4. The earliest mini-paradigms and failed attempts age 1;3

lemma a:bi ‘brother’ anne ‘mother’

1;5

bebek ‘baby/doll’ baba ‘father’ a:bi ‘brother’ baba ‘father’ abla ‘sister’

form a:bi a:bi-ye a:bi-*0ye anne anne-*0si-yi anne-*0si-*0ni bebek bebek-*0i-*0ni baba baba-*0sı-*0nı a:bi a:bi-*0nin baba baba-*0nın abla abla-*0nın

category N N-DAT N-*0DAT N N-*0POSS-ACC N-*0POSS-*0ACC N N-*0POSS-*0ACC N N-*0POSS-*0ACC N N-*0GEN N N-*0GEN N N-*0GEN

examples of two-member paradigms as well as the first four-member paradigm. It is observed that a given type of marking occurs with a number of different lemmas, an important indication for the development of productivity. Thus, we consider age 1;6 as the starting point of the proto-morphological period (at least as far as morphological marking in the nominal domain is concerned).

26

F. Nihan Ketrez and Ayhan Aksu-Ko¸c

Table 5. Two- and multi-member paradigms age 1;6 (2-member)

(4-member)

lemma bu ‘this’ o ‘it/she/he’ ev ‘house/home’ el ‘hand’ Omi15 ‘grandma’

1;7 (5-member)

bura ‘here’

1;8 (6-member)

ayakkabı ‘shoe’

(7-member)

bu ‘this’

1;9 (9-member)

bu ‘this’

2;0 (7-member)

anne ‘mother’

form bu bu-nu o-nu o-nun ev ev-i el-i el-*0im-e omi omi-m omi-n omi-m-e bura-dan bura-ya bura-da bura-da-yım bura-da-ymı¸s ayakkabı ayakkabı-*0sı-nı ayakkabı-yla ayakkabı-lar-ı-nı ayakkabı-sı-nı ayakkabı-sı bu bu-nu bu-na bu-ndan bu-nun-la bu-nlar bu-nlar-ı bu bu-nu bu-na bu-ndan bu-nun-la bu-nlar bu-nlar-ı bu-nun bu-nlar-dan anne anne-yi anne-yle anne-m anne-si-ne anne-si-nin anne-si

category PRO14 PRO-ACC PRO-ACC PRO-GEN N N-ACC N-POSS N-*0POSS&1S-ACC N N-POSS&1S N-POSS&2S N-POSS&1S-DAT PRO-ABL PRO-DAT PRO-LOC PRO-LOC-1S PRO-LOC-PERF N N-*0POSS&3S-ACC N-COM N-PL-POSS&3S-ACC N-POSS&3S-ACC N-POSS&3S PRO PRO-ACC PRO-DAT PRO-ABL PRO-GEN&3S-INSTR PRO-PL PRO-PL-ACC PRO PRO-ACC PRO-DAT PRO-ABL PRO-GEN&3S-INSTR PRO-PL PRO-PL-ACC PRO-GEN&3S PRO-PL-ABL N N-ACC N-COM N-POSS&1S N-POSS&3S-DAT N-POSS&3S-GEN N-POSS&3S

Early nominal morphology in Turkish: Emergence of case and number

27

Interestingly, the biggest paradigms (in terms of the number of members) are observed in pronouns rather than nouns. For example, the 9-member paradigm, which is the biggest paradigm we record, represents the inflexion of the demonstrative pronoun bu ‘this.’ The first 7-member paradigm is also observed with the same pronoun. Simple nouns may reach the size of 7-member paradigms, an example of which is the paradigm of anne ‘mother’ at 2;0 (see table 5). The difference in the productivity of pronouns and nouns can be attributed to the fact that pronouns replace nouns. Since they are used more frequently, there are more occasions for using them with different cases. Besides, as mentioned above, pronouns are obligatorily case-marked even in immediately preverbal position, where nouns can appear in bare form.

3.3.

PBF and inflections per word

Another criterion that we have used for the demarcation of phases is PBF (Percentage of Base Forms16) developed by Gagarina and Voeikova (2002). According to their observations in Russian data, when PBF reaches 50% in child speech it has reached the adult-level value since 50% is also the PBF recorded for childdirected speech. The PBF in Deniz’ mother’s CDS based on the samples of three different points in time that correspond to our three phases in development is 46% (318/704) and remains roughly the same throughout this period. In Deniz’s speech, PBF reaches 58.6% around the age of 1;9. Although this is the lowest it gets during the period analyzed, it is still higher than the adult figure. Table 6 shows the details of the counts for Deniz.

Table 6. PBF and inflections per word (lemmas, types, and tokens) Phase Premorphology

Protomorphology

Morphology Proper

age 1;3 1;4 1;5 1;6 1;7 1;8 1;9 1;10 1;11 2;0

PBF17 98.7 100 98.6 88.3 75.5 62.5 58.6 63.9 63.1 62.6

bare 161 70 74 174 257 380 307 370 257 427

+INFL 2 0 1 23 83 228 216 209 150 255

stem+1 2/2/2 0 1/1/1 10/13/22 23/39/79 37/57/207 46/68/160 39/59/190 37/58/131 45/73/201

stem+2 0 0 0 1/1/1 3/4/4 11/13/21 19/24/56 14/16/19 13/15/19 22/33/54

28

F. Nihan Ketrez and Ayhan Aksu-Ko¸c

Between the ages of 1;3.3 and 1;5.9, forms lacking overt inflectional markers are the norm. After 1;6, we see a gradual increase in the number of inflected forms and multi-member paradigms of two or three members and PBF drops to about 58.6%. This increase ceases at 1;9 and the morphological variation becomes stable. No decrease is observed during the next period. Compared to the corresponding figures in the input we can say that the child’s pattern of usage is gradually approaching the adult pattern but does not yet reach the adult level during the period analyzed.

3.4. What determines demarcation When the frequency and proportion of case, possessive and number morphemes within inflected nominal stems are compared it becomes apparent that variation of inflected forms is mostly due to case morphology (table 7). The second most common type of inflection is the possessive.The use of the plural morpheme does not exceed 2.7% of all inflected forms through the end of the period analyzed. This is exactly the pattern prevalent in the input: 65–80% of the mother’s inflected forms bear case, 46–68% possessive, 7–14% plural and 2–9% other morphology. In the following sections, we will study the individual case and number morphemes within the child’s emerging system of nominal morphology.

Table 7. Proportion of forms marked for case, plural, and possessive (Tkn = token, Total = total of inflected tokens) Phase

age

Premorphology

1;3 1;4 1;5 1;6 1;7 1;8 1;9 1;10 1;11 2;0

Protomorphology

Morphology Proper

Total

case18 Tkn (%)20 2 (100) 0 1 (100) 16 (69) 71 (86) 215 (94) 201 (93) 186 (89) 121 (81) 224 (88) 1037 (88)

POSS plural other19 total Tkn (%) Tkn (%) Tkn (%) 0 0 0 8 (35) 9 (11) 20 (88) 61(28) 42 (20) 42 (28) 52 (20)

0 0 0 0 0 4 (2) 6 (2.7) 1 (0.4) 2 (1.3) 3 (1.1)

0 0 0 0 4 (5) 3 (2) 2 (0.9) 0 1(0.6) 6 (2.3)

2 0 1 24 83 228 216 209 150 255

234 (20) 16 (1.4) 16 (1.4) 1168

Early nominal morphology in Turkish: Emergence of case and number

29

4. Case Case morphology has already emerged by the very first session recorded at 1;3.3, but, as argued above, there is no evidence for the beginning of productivity until 1;6. The proto-morphological stage is characterized by evidence of the child’s detection of the typologically relevant features of the morphological structure of the input and there is thus an increasingly productive use of nouns with appropriate inflectional endings. In the stage of morphology proper, all cases of the standard language are found and they are used productively and with less error. The major and the marginal cases in Deniz’s speech and the number of instances where they were omitted are presented in tables 8, 9, and 10 (percentages are given only for errors of omission).21 In the following section, the emergence and development of the individual cases will be studied, beginning with the syntactically most functional ones.

4.1. Accusative and dative Although Deniz uses accusative, dative, and locative case forms once each already during the pre-morphological period, there is no evidence yet for their Table 8. Major cases: Types (tokens) Phase Premorphology

age *0ACC 1;3 4 (4) 80% 1;4 – 1;5 – Protomorphology 1;6 – 1;7

Morphology Proper

3(3) 18% 1;8 1(1) 1% 1;9 3(7) 5% 1;10 3(5) 4% 1;11 4(4) 11% 2;0 3(5) 5%

ACC 1(1) – – 3(10) 5(14) 17(88) 14(53) 25(62) 17(33) 18(54)

*0DAT 2(2) 67% – – 3(3) 43% 2(3) 9% 2(2) 7% 2(2) 4% 2(2) 3% 1(2) 3% 1(2) 2%

DAT 1(1)

*0LOC LOC – –

ABL22 –

– – 1(1)

– – –

– 1(1) 1(1)

– – –

11(18)

–

5(8)

3(3)

14(28) 2(2) 3% 25(55) –

14(74) 4(4)

17(61)

10(21) 7(14)

11(31) 7(23)

14(27) –

8(30) 1(1)

24(47) –

19(72) 6(8)

30

F. Nihan Ketrez and Ayhan Aksu-Ko¸c

productive use. In the first session of the next phase, she produces four nominal forms marked for accusative, dative, locative, and genitive, respectively. As shown in table 8, the accusative case is used by 1;6 according to our definition of productivity: The child uses it with different lemmas (3 nouns and 2 pronouns, see table 5), all of which also occur as bare or nominative forms, and omission errors are less frequent than instances of correct case marking. The accusative also shows a rapid increase in the subsequent months, both in terms of types and tokens. In (2) the case-marked pronoun o-(n)u ‘this-ACC’ is used without a verb which may be inferred to be either one of the ditransitives koy- ‘put’ or ver- ‘give’. No omission of the accusative marking is observed at this age. (2)

1;5.28 o-nu o-nu diy [: de˘gil] o-nu! it-ACC it-ACC not it-ACC ‘it, it, not it.’ (doing a puzzle with her mother)

At 1;7.3, nouns marked for accusative appear with additional verbs such as yap- ’draw,’ boya- ’paint,’ and kapat- ’turn-off’ as in (3). Since the accusative (and the dative) are relational cases that are selected by the verb, their use with verbs of different types of argument structures reflect increased generality in productive use. However, there are also contexts where the accusative is omitted throughout the proto-morphological and even the morphological period. An example of omission is (4), where the doll being referred to is a specific doll and would therefore require an accusative marking. The omission would not be ungrammatical, however, if the intended meaning were the ‘activity of dollcovering’. As described in section 2 above, a direct object may appear in bare form in immediately preverbal position. Rather than referring to a specific object, it is interpreted as being incorporated in the verb having a non-referential and non-specific reading referring to the activity aimed at the given type of object in general. Interestingly, in the majority of omissions of accusative marking, objects do occur in preverbal position. There is also evidence in the girl’s speech that, in addition to the preverbal position, she can produce case-marked objects in post-verbal position, which may suggest a very early sensitivity to the constraints found in the adult language, namely that dislocated objects must bear case marking. (3)

1;7.3 te:b-i gabat-tı-m [: kapattım]. tape+recorder-ACC turn+off-PAST-1S ‘I turned off the tape recorder.’

Early nominal morphology in Turkish: Emergence of case and number

(4)

31

1;7.3 bebek o¨ :t-t¨u-m. doll-0ACC cover-PAST-1S ‘(I) doll-covered.’

Although, in the proto-morphological period, Deniz seems to have difficulty in producing both the arguments and the verb in one single utterance, she nevertheless uses appropriate case markings with the argument(s) expressed without the verb according to their thematic roles, as in the successive utterances of (5). (5)

1;7.8 MOT: bu-nu san-a mı tak-a-yım? this-ACC you-DAT Q attach-OPT-1S ‘shall I attach this on you?’ CHI: gat [:tak]. (. . . ) attach&IMPER ‘attach.’ CHI: bu-nu ban-a [=! cries]! (. . . ) this-ACC I-DAT ‘this to me.’ CHI: bu-nu gat! this-ACC attach ‘attach this.’

In summary, the accusative case is used productively and in a functionally appropriate way during the entire proto-morphological period. Given particular pragmatic conditions, some of the omissions of case marking which occur during this period are not necessarily errors, such as unmarked non-referential and nonspecific objects. As far as the dative is concerned, it is only used with a single noun at 1;6, but is omitted with three other nouns. In one example Deniz correctly marks the noun omi ‘grandma’ by the dative case in replying to her mother’s question containing a question word in the dative (kim-e git-ti-n? (who-DAT go-PAST-2S) ‘to whom did you go?’) by saying o:mi:-m-e (omi-POSS&1S-DAT) ‘to my omi’. In the same session, responding to the same question in another utterance, she fails to produce the dative case and uses the bare form A:zu: [: Hakan]. These examples show that Deniz does not yet mark indirect objects reliably by the dative at 1;6. As opposed to the accusative, the dative becomes productive more gradually during the proto-morphological period. At 1;7.3, it marks responses to nere-ye

32

F. Nihan Ketrez and Ayhan Aksu-Ko¸c

‘where-to’ and kim-e ‘who-to’ questions. It is more frequently produced as an internal inflection of the noun phrase than as a relational case with ditransitive verbs. By the end of the proto-morphological period, Deniz produces contrastive forms of a given noun with the dative being one of these (e.g., anne-ye ‘motherDAT’ vs. anne-nin ‘mother-GEN’; ben-i ‘I-ACC’ vs. bana ‘I-DAT’ vs. ben-im ‘I-GEN’; ora-ya ‘there-DAT’vs. ora-dan ‘there-ABL’). Such contrasts first occur on person names and pronouns. However, omissions of the dative in obligatory contexts continue to occur through the end of the proto-morphological stage. In contrast to the non-marking of the accusative, omission of the dative marking always results in ungrammatical forms.

4.2.

Locative and ablative

The first example of the locative case in Deniz’s speech occurs in the premorphological period at 1;5.9 in response to the question sen-in aya˘g-ın ner-de? (you-GEN foot-POSS&2S where-LOC) ‘where is your foot?’which she answers by bu:-da (here-LOC). Since the locative case is only once used with the demonstrative pronoun bura in a predicative function, it appears to be a frozen form. In the proto-morphological period, the locative case also appears with nouns stating the location where an action takes place (e.g. at 1;7.23, ba:tte-de (garden-LOC) ‘in the garden’ in reply to her mother’s question about the place where some children in a picture book are playing. Although the locative does not have a relational function and occurs much less frequently in the input, it emerges at the same time as the accusative and the dative, presumably because (a) it marks a conceptually important and accessible semantic notion, (b) it is a general, all-purpose locative marker that does not specify the particular relation between the object and its location (such as ‘in’, ‘on’, ‘under’ etc.), (c) nouns/pronouns marked for the locative may function predicatively and, since the copula receives zero marking in the present tense, the case marker is the word or sentence-final salient element, and finally (d) its use is obligatory even with indefinite nouns (as opposed to the accusative case marking). The ablative is used to indicate the locative source before indicating the material source or partial quantity. At 1;7.23, Deniz uses one and the same noun contrastively in the dative, the locative, and the ablative as in araba-ya ‘carDAT,’ araba-da ‘car-LOC’, and araba-dan ‘car-ABL’ (see table 5 for similar uses of the pronoun bu ‘this’). These observations suggest that during the protomorphological period Deniz begins to differentiate the semantics of space into source, goal and static location as dictated by the language. Throughout the

Early nominal morphology in Turkish: Emergence of case and number

33

proto-morphological period, the error-rate of omission of the locative (and the ablative) is lower than that of the dative and the accusative.

4.3.

Genitive (and possessive)

The genitive first emerges with nominal-internal scope. Like the other cases, it is initially used with pronouns such as o-nun ‘it-GEN&3S’. In Deniz’s speech, the genitive emerges at 1;7.3 and is missing in obligatory contexts before this age (see example (6)). Most of the later examples in which the genitive is correctly marked are responses to adult questions about specifying the possessor. Thus, at 1;7.3, when talking about the tape recorder, her mother asks ba¸ska kimin? (else who-GEN&3S) ‘whose else’s is it?’ and she responds baba+annenin (grandmother-GEN&3S) ‘grandmother’s,’ appropriately marking the noun for the genitive. The possessive morpheme marking agreement on the possessed is used from the beginning of proto-morphology. Since most contexts of early mother-child discourse are in the here-and-now, the possessor (usually either one of the interlocutors) needs not be mentioned unless for purposes of emphasis, and the possessive suffix attached to the nominal that refers to the possessed object is sufficient to express the relation. The possessive is therefore more frequent than the genitive in the input, and becomes gradually so in the child’s speech as well. (6)

1;6.9 CHI:

dede (/2) &de. [correct form: dede-nin] grandfather-*0GEN MOT: evet dede-nin teyb-i yes grandfather-GEN tape-recorder-POSS ‘yes, grandfather’s tape-recorder.’ CHI: ba:ba+anne. [correct form: ba:ba+anne-nin] grandmother-*0GEN. ‘grandmother’s.’ During the period analyzed in this paper, the genitive is exclusively used in possessive constructions, where it marks the possessor. Contexts with embedded subjects which would have to be marked by the genitive do not occur. There are also no examples where the genitive functions as a relational marker indicating the complement of a verb. Table 9 presents the type and token frequencies of the genitive case indicating the possessor together with the possessive marker specifying the possessed.

34

F. Nihan Ketrez and Ayhan Aksu-Ko¸c

Table 9. Genitive and possessive Phase Premorphology

Protomorphology

Morphology Proper

age 1;3 1;4 1;5 1;6 1;7 1;8 1;9 1;10 1;11 2;0

*0POSS 5(5) 100% – – 2(2) 20% 1(1) 10% 5(8) 29% – – 2(2) 5% –

POSS – – – 7(8) 8(9) 17(20) 34(61) 26(42) 30(42) 39(52)

*0GEN – 3(6) 100% – 3(4) 80% 2(2) 7% 1(1) 5% 2(2) 5% 1(1) 4% – –

GEN – – – 1(1) 7(28) 6(18) 16(34) 6(24) 10(25) 14(31)

As described above, case markers follow possessive markers when they are attached to the same stem. The only exceptions are forms with ki, since this is attached to the case-marked stem and followed by the possessive (and eventually the plural). The first example of a structure comprising case-ki-possessive is produced at 1;8,11. Although there are no errors concerning the order of these suffixes, we cannot conclude whether the child is aware of the ordering rules and their exceptions or merely using a rote-learned form since there are not enough examples of the ki structure in this period. (7)

1;8.11 ben-im-ki-ti bit-ti. I-GEN-ki-POSS&3S finish-PAST ‘mine has finished.’

4.4.

Marginal cases: Instrumental and comitative

Both the instrumental and the comitative are rare in the input throughout the period studied, amounting to only around 1% of tokens and indicating that the contexts for their use are very limited in comparison with the accusative and the dative. The two cases are also very rare in the speech of our subject, each of them missing in an obligatory context during the first month of the protomorphological period. After 1;8, they remain scarce, but are used without error in the limited number of contexts observed. In an utterance recorded at 1;7.23, Deniz says deyen-ne (train-INSTR) ‘with train’ in response to her mother’s question bur-da ne-yle oynu-yor-lar? (hereLOC what-INSTR play-PROG-3P) ‘what are they playing with here’? This

Early nominal morphology in Turkish: Emergence of case and number

35

Table 10. Comitative and instrumental Phase Premorphology Protomorphology

Morphology Proper

age 1;3–1;5 1;6 1;7 1;8 1;9 1;10 1;11 2;0

*0COM – 1(1) 100% – – – – – –

COM – – – 1(1) 3(5) 2(2) – 7(12)

*0INSTR – 1(1) 100% – – – – – –

INSTR – – – 1(2) – 1(2) 2(5) –

example illustrates the child’s appropriate use of the instrumental case in the late proto-morphological phase.

4.5.

Commission errors

Besides the omission errors discussed above, there are also commission errors to be observed in Deniz’s speech. While the former occur from the very beginning, the latter can be considered to be a property of the proto-morphological period. Commission errors consist in the substitution of one of the case markers for another. These errors are rather rare, however, and are only observed once or twice for each case. An example of a genitive case substituted by an ablative is shown in (8). Note that the function of the ablative case is to indicate the source; it is, therefore, not surprising that it substitutes the genitive case, which also, in a sense, expresses the source, the possessor of the tape-recorder in this example. An instance of a substitution of the accusative by the comitative case is given in (9). Here, Deniz immediately corrects her error in the next utterance, in which she reacts to her mother’s question comprising a model of the correct case. (8)

1;7.23 MOT: kim-in teyb-i-ydi bu? who-GEN tape+recorder-POSS-PAST this ‘whose tape-recorder was this?’ CHI: dede-den. GEN > ABL grandfather-*ABL ‘from grandfather.’

36

(9)

F. Nihan Ketrez and Ayhan Aksu-Ko¸c

1;10 CHI:

Lale-yle di-yo-lay. ACC > COM Lale-*COM say-PROG-3P ‘They are telling with Lale.’ (they are talking about Lale) MOT: kim-i? who-ACC CHI: Lale-yi. Lale-ACC . ‘Lale.’

Besides these examples, the other types of commission errors and the ages at which they occur are the following (each indication represents a single error unless noted otherwise): ACC > LOC (1;11), ACC > INSTR (1;11), DAT > ABL (1;9), DAT > COM (1;9), DAT > ACC (1;8) ABL > LOC (1;8, 2;0), GEN > ABL (1;7), INSTR > DAT (1;9). Interestingly, the locative and the comitative are never substituted by other cases. In all, the errors observed with the other cases are very infrequent and the correct use of cases is much more frequent than the incorrect one.

4.6.

Case in the period of morphology proper

In the morphological period, the beginning of which is observed in our data, the language-particular properties of both nominal and verbal morphology are further elaborated and generalized to new lemmas and new structures. As far as the nominal system is concerned, we observe that in the proto-morphological period, nouns, proper nouns, demonstrative pronouns, personal pronouns and wh-words are marked for case, plural and possession. At the beginning of the morphological period, additional types of words that receive marking are indefinite, reflexive and reciprocal pronouns, adjectives and cardinals. The accusative and the dative get to be used with nouns which function as the arguments of novel verbs. Multiple marking on the same form and the emergence of derivational morphology are further developmental achievements observed in this period. Correct use of case with verbs marked for voice alternations provides evidence for developments in the interface of morphology and syntax. Occasional errors are nevertheless observed when children use complex morphology but cannot perform the accompanying syntactic operation, such as marking the object noun with the accusative when they passivize a transitive verb instead of leaving it unmarked for the nominative (Aksu-Ko¸c and Ketrez 2003; Ketrez 1999, 2000a, 2000b). This is, then, the period when complex morpho-syntactic relations begin to be constructed.

Early nominal morphology in Turkish: Emergence of case and number

4.7.

37

Summary and discussion

As shown in table 8 above the accusative, dative and locative cases emerge first, but only the accusative starts to be used productively at 1;6. The locative and dative as well as the ablative and genitive do so only at 1;7. The marginal cases emerge one month later. Although dative and accusative emerge earlier and are used more frequently than the other cases, Deniz shows fewer errors with the locative and the ablative, so the latter two cases seem to be easier to acquire. We attribute this ease of acquisition, despite their relatively low frequency in the input (2–3% of all noun tokens versus 11–13% recorded for accusative and dative), to their semantic accessibility and predictability of use which derives from the fact that adverbial adjuncts do not belong to the kernel structure of the sentence. In other words, since they are not selected by the verb as arguments, they are not subject to various constraints that may arise from verb’s particular properties. Possessive marking emerges and becomes productive together with the case markers, whereas the plural morpheme, which will be presented below, emerges by 1;8, together with the marginal cases. Table 11 presents a summary of the emergence of inflectional structures at a given age within the different phases of development. While there are only two frozen examples of case marking in the pre-morphological period, in the next developmental phase, between 1;6 and 1;9, a variety of structures emerge. After 1;9, we do not find any new structures. Rather, further development observed between 1;10 and 2;0 consists in new formal markings or new combinations of those so that what has emerged before gets to be used with a greater number of lemmas or in more diverse constructions. This picture is consistent with

Table 11. Structures occurring in each phase (ages of emergence) Phase Premorphology Protomorphology

Morphology Proper

age 1;3–1;5 1;6

What is new in each phase? N/PRO N-POSS N-POSS-case 1;7 N/PRO-case-AGR N/PRO-case-tense/aspect 1;8 N/PRO-PL-POSS-case N/PRO-PL-case PRO-case-ki-POSS 1;9 N-PL-POSS PRO-case-ki-PL 1;10–2;0 –

case type ACC23 DAT, LOC ABL, GEN INSTR COM

38

F. Nihan Ketrez and Ayhan Aksu-Ko¸c

the conclusion reached earlier, that, by 1;9–1;10, Deniz marks all the formal categories enumerated in table 11 reliably (i.e. in almost 90% of obligatory contexts) and that no significant development in terms of the acquisition of new forms is observed after that age. It should, however, be noted that the emergence of the entire formal apparatus of case marking does not mean that mastery of all its functions has been achieved; this will necessarily have to wait until further developments in complex syntax and semantics take place in the following years (see Aksu-Ko¸c and Slobin 1985; Aksu-Ko¸c 1994; Ketrez 2005).

5. Number Plural marking is absent during the pre-morphological and the earlier phases of the proto-morphological period. As can be seen in table 12, there are no obligatory contexts up to this point where plural marking should be produced. This is not surprising since number is not strongly grammaticalized in Turkish. In fact, the frequency of plural marking in Deniz’s mother’s speech is quite low, averaging around 4–5% of all nouns in the sessions sampled. Plural marking emerges in the proto-morphological period when case marking gets consolidated; it is first used on nouns (10) and demonstrative pronouns (11). It is next extended to wh-words (12) and to nouns marked for possessive (13). Table 12. Types and tokens of plural marking with -lAr Phase

age

Premorphology Protomorphology

1;3–1;5 1;6, 1;7 1;8 1;9 1;10 1;11 2;0

Morphology Proper

(10)

*0PL

PL

– – – – – – –

– – 3(4) 3(6) 1(1) 2(2) 2(3)

1;7.23 du duy-lay [: mum-lar] bu:-da duy-lay [: dur-sun-lar]. that candle-PL here-LOC stay-*0OPT-3P ‘let these candles stay here.’

Early nominal morphology in Turkish: Emergence of case and number

(11)

39

1;7.23 bu-nnay-ı [: bu-nlar-ı] giy-me-di-n. this-PL-ACC put.on-NEG-PAST-2SG ‘you did not put these on.’ (referring to a pair of slippers)

(12)

1;8.11 de-yey [: ne-ler] vay buy-da? what-PL there.is here-LOC ‘what is in here?’

(13)

1;8.11 ayak-lar-ı-nı. foot-PL-POSS&3S-ACC ‘his feet.’ (Deniz wants to paint somebody’s feet in a book.)

In contrast to the earlier phases of the proto-morphological period during which nominals are marked either for case or possessive, a salient feature of the later phases of the proto-morphological and the morphological period is that different nominal inflections are suffixed to the same form, resulting in double (or multiple) marking such as plural + possessive + case. As noted earlier, about 31% of all inflected nouns in the child’s input display multiple marking. Plural morphology occurs in such multi-inflection combinations as soon as it emerges. No difficulty is observed in the production of plural morphology. At the age of 1;9.19, Deniz produces the first case-ki-plural structure, carrying three types of morphological markers (see example 14). As pointed out above, the special feature of this structure is that the plural morpheme follows the case marker, while it precedes case in nominal structures without ki. This example again suggests that the child knows the morpheme order rules, since she does not overgeneralize the order of plural < case, which is more common in the input and in adult speech in general. (14)

1;9.19 u¨ t-te-ki-ye: [: u¨ st-te-ki-ler]. top-LOC-ki-PL ‘the ones on the top.’ (wants her mother to take the legos on the top.)

In summary, synthetic number marking in Turkish not only emerges late as compared to case marking, but is also much less frequently used both type- and

40

F. Nihan Ketrez and Ayhan Aksu-Ko¸c

tokenwise. It is recorded in only 1.3% of all inflected forms within the entire period analyzed. Inflectional number marking in Turkish appears at the same time as the peripheral cases, after core cases have gained productivity. The development of core cases before number is not surprising since, in Turkish, core cases are necessary for specifying the syntactic roles of the constituents in the sentence, whereas the less strongly grammaticalized category of number is usually not grammatically indispensable. The ‘singular’ form of the noun is numerically neutral, denoting a category or an individual member of a category (Lewis 1967; Ketrez 2003). Hence it can be used after numerals and remains morphologically unmarked for plurality, and it can be used to denote a collection/plurality of objects, inferences regarding plurality being context-dependent. These factors may be among the reasons for the later emergence of number marking as compared to case marking on nouns. As far as number marking on verbs is concerned, Turkish requires subjectverb agreement and number marking on verbs is fusional. 1st person plural in the first paradigm that children learn is -uz, 2nd person -sunuz, and the 3rd person -lAr. Thus, the plural suffix -lAr is used with both nouns and verbs. However, plural marking for the 3rd person is subject to a number of constraints: It is ungrammatical with [– animate] subjects, optional in the case of [+ animate] subjects, and preferred only with [+ definite] ones. Hence, the child has to learn to differentiate between contexts where it should be marked and those where it should not (in fact, children are sensitive to these restrictions and over-marking is very rare). This system rather deviates from other typological features of Turkish that favor synthetic marking. However, the fusional character of number marking on verbs does not appear to pose a difficulty for children as they acquire the 1st person plural marking on verbs slightly before 3rd person plural marking on verbs and plural marking on nouns. This can be explained by the fact that personnumber marking on verbs is a syntactically obligatory mechanism in Turkish, which is a null-subject language. The 1st person plural, which is obligatory and more frequent in the input (14% of all verb forms), emerges both before the 3rd person, which is optional (1% of all verbs), and before nominal number (4.5% of all nouns). The later appearance of plural marking on nouns as compared to case marking and also on the verb for the 3rd person plural (both -lAr) can therefore also be explained by the low input frequencies reflecting their conditions of use in the language.

Early nominal morphology in Turkish: Emergence of case and number

41

6. Discussion and conclusion In this study, we have presented a detailed analysis of the emergence of nominal morphology in a Turkish child’s speech. The study concentrated on case and number morphology as well as the emergence of possessive marking. The developmental period studied was divided into three phases: Pre-morphology, proto-morphology and morphology proper. We showed that during the premorphological period, our subject lacks productive nominal morphology. From 1;6 onwards, there is a sharp rise in the amount and productivity of inflectional morphology. This increase continues until the age of 1;9. From 1;9 onwards, nominal development becomes stable in that there are no further changes in terms of emergence of new morphological forms. However, productivity continues to increase as the type and number of nominal constructions that are inflectionally marked multiply. The results of the present analysis allow a comparison of the development of nominal and verbal inflection in our subject (see Aksu-Ko¸c and Ketrez 2003 on the verb). Nominal morphology starts to develop at the same time as verbal morphology revealing a differentiation between the two lexical classes. As shown in Aksu-Ko¸c and Ketrez (2003), the child Deniz follows the same pattern of emergence of verbal morphology as the one found in the present paper for the emergence of nominal morphology. Namely, there is evidence that the same three phases can be distinguished for nominal as well as for verbal morphology at approximately the same ages. In this respect, the pattern of the acquisition of Turkish as a first language contrasts with its acquisition as a second language by adults and children. Haznedar (2003), who studies the acquisition of Turkish as a second language with particular emphasis on the use of case markers and nominal morphology, observes that her subjects cannot yet produce case morphology during a period when their verbal tense/aspect and agreement morphology is almost 100% correct. Her results are in line with earlier findings by G¨urel (2000). This contrasts with our finding that the demarcations of the developmental periods in the two inflectional domains more or less coincide, i.e., as long as Deniz lacks productive case morphology she is lacking productive verbal morphology as well. We find a parallelism especially in the acquisition of number agreement on verbs and plural marking on nouns, which both appear around the same time. There is another contrast between first and second language acquisition as far as errors of omission are concerned. While the omission rate of case morphemes in obligatory contexts is quite low in Deniz’s speech – between 1;3 and 2;0, she omits the accusative in less than 10% of obligatory contexts – this rate is 85.6% in Turkish second language acquisition (Haznedar 2003: 146).

42

F. Nihan Ketrez and Ayhan Aksu-Ko¸c

Relational as well as internal case morphology reveal development in syntax at the clause as well as the phrase level. Use of local cases and internal noun morphology provides evidence for developments in various semantic domains such as possession and location. We have shown that the major core cases accusative and dative emerge around the same time but that the accusative becomes productive earlier, probably because it is the major indicator of syntactic relations in this flexible word-order language. The dative, which can also designate peripheral constructions, is first observed in its relational function in our subject’s speech with ditransitive verbs. The dative has been observed to emerge early, even earlier than the accusative, in other Turkish children’s speech as well (Topba¸s, Mavi¸s, and Ba¸sal 1997).24 The locative is also among the first cases children acquire, emerging at the same time as the two core cases. Its early emergence may be attributed to its structural and semantic properties discussed in section 5.2 above. The acquisition of the other marginal cases follows. However, all cases or, to be more precise, all nominal inflectional morphology emerges during the proto-morphological period between the ages 1;6 and 1;9. Although errors are still observed until 2;0, the end of the period studied in this paper, the abundance of grammatical morphemes adequately produced shows that there is major progress in terms of productivity in the nominal system. When compared to children acquiring a fusional language such as Greek, where different cases develop differently in different genders and in the singular as compared to the plural (Stephany 1997; Christofidou 1998; Stephany and Christofidou, this volume), the task of the Turkish child is less demanding. The early emergence and the few errors observed in the nominal paradigms can be attributed to language-specific properties such as regularity and transparency of paradigms and saliency of individual morphemes.

Notes ∗

This study is based on the data collected for the project “A longitudinal study of the acquisition of Turkish” (Project no: 96S0017) that was supported by a grant to A. Aksu-Ko¸c, from the Bo˘gazi¸ci University Research Fund. We would like to thank Maria Voeikova and in particular Ursula Stephany for their critical editorial comments on the earlier versions of this paper. 1. We include these markers in our calculation of different types, but do not discuss their acquisition pattern. 2. There are exceptions such as plural personal pronouns (e.g., biz ‘we’) 3. But see the discussion of comitative and instrumental case markers that are attached to pronouns and ki-constructions discussed below.

Early nominal morphology in Turkish: Emergence of case and number

43

4. Abbreviations (adapted from MacWhinney 2000): ABL: ablative, ACC: accusative, AOR: aorist, CHI: child, COM: comitative, DAT: dative, DIM: diminutive, GEN: genitive, INSTR: instrumental, LOC: locative, MOT: mother, NEG: negation, NP: noun phrase, NR: nominalization marker, OPT: optative, past: past tense, PROG: progressive, POSS: possessive, PL: plural, 1/2/3P: 1/2/3 person plural, 1/2/3S: 1/2/3 person singular, Q: question particle, V: verb, N: noun, PRO: pronoun, (&) morphological fusion, (+) compound marker, (–) morpheme boundary, (>) commission, (0): omission of the following morpheme, (*0): ungrammatical omission. 5. Note the stem change. 6. -Im is an irregularity that occurs only with the first person singular pronoun ben. 7. ki is a particle borrowed from Persian. When attached to nouns, the noun complex has a function similar to that of a relative clause. 8. The command/optative form of the verb is the only exception (e.g., git-sin ‘let him/her/it go’). 9. The locative is also used with some more abstract verbs in less transparent locative relations such as ‘-DA anla¸s- ‘to have an agreement on.’ With the exception of a few examples such as these, which are not observed in our data, the function of the locative case is very clear and predictable. 10. In addition to verbs, adjectives and postpositions take complements marked by the ablative (e.g., -DAn beri ‘since,’ -DAn itibaren ‘from . . . onward,’ -DAn emin ‘sure/confident’). Such examples are not observed in our subject’s speech either. 11. In the first and second person pronouns (ben and sen), the genitive can be dropped in spoken Turkish (benle and senle), although it remains obligatory (benimle and seninle) in written Turkish. A similar kind of tendency is seen in the third person singular pronoun o, however, the short form is pronounced onla rather than *oyla. Note that the /n/ is a part of the genitive case suffix and does not belong to -yle. Therefore the pronunciation of the suffix as -nla may suggest that the omission of one of the syllables is a phonological omission due to ease of pronunciation, rather than a morphological omission of the genitive case. In other words, the two different pronunciations of the personal pronouns cannot be attributed to a morphological transition phase (from the status of a postposition to that of a case marker) even if there is such a phase. 12. -lAr can also be attached to verbal stems as the third person plural agreement marker when the subject is animate (Sezer 1978) and referential (Ketrez, 2003). The agreement marker can optionally be dropped when the subject is expressed (and not dropped due to pro-drop). With personal pronouns, number and plural are fused (ben ‘I’ vs. biz ‘we’), except in the case of the 3rd person plural (o ‘s/he, it’ vs. o-nlar ‘they’). When -lAr is attached to mass nouns such as milk, meat, coffee, rice, it assigns a countable meaning such as ‘a bottle of milk,’ ‘two pieces of meat’, or ‘two kinds of rice.’ When attached to demonstrative pronouns this, that and to the personal pronoun it, -lAr serves to derive the plural pronouns these, those and they. 13. The number of forms that can potentially occur in a single nominal paradigm of Standard Turkish, taking all possible combinations of zero, possessive, case, and

44

14.

15.

16. 17. 18. 19.

20.

21.

22.

F. Nihan Ketrez and Ayhan Aksu-Ko¸c number marking into consideration, has been calculated to amount to 98. However, in child-directed speech, the situation is far less complex. Inflectional diversity expressed in terms of the different combinations of suffixes an average noun occurs with is reported by K¨untay and Slobin (1996) to be 7.65 in the speech of the mother of a Turkish child 1;8–2;3 years old. Although in table 7 we observe that Deniz can produce mini-paradigms of 7 or 9 members at 1;9 or 2;0, these cannot be said to represent her average performance since the mean size of her noun paradigms is around 1.50 at that age, as discussed in section. 4.1. If a PRO or N appears in an unmarked form in the table, this means that it appeared as a bare form in the child’s speech. Its function could be either subject, or indefinite object (in the case of nouns). Although the maternal grandmother of the child is German she is fluent in Turkish and speaks in Turkish to Deniz like everyone else at home. The influence of the German language does not go beyond a few German kinship terms, such as omi ‘grandma’, tante ‘aunt’, and muti ‘mommy’ in Deniz’s speech. We define PBF as bare stems that do not carry any inflection in Turkish. Percentage of bare forms based on tokens. Comitative and instrumental are included, but they constitute less than 5% of all case-marked forms. The “other” category of the table includes tense/aspect/modality markers and agreement markers found on nominals occurring in predicate position as well as a number of words having ki (e.g., masa-da-ki (table-LOC-ki) ‘the one on the table’). Note that the individual percentages may exceed 100% because a given stem can bear more than one type of inflection. That is, a stem that has both a case marker and a possessive marker is included in the calculation of the possessive proportion as well as the case proportion. Only during the pre-morphological period, all uses add up to exactly 100%. As soon as the child enters the proto-morphological period, she starts using stems with more than one nominal suffix attached to them. For the same reason, the total frequencies given for Inflected Tokens do not equal the sums of the tokens indicated in the rows but the totals of inflected tokens given in table 8. See table 5 for the first such example that was recorded during the first month of the proto-morphology period and table 11 for a summary of multi-suffix forms emerging during proto-morphology. As has been noted in the introduction, the nominative can be assumed to be functional once the accusative and the dative are. However, it has been excluded from analysis and does not appear in these tables because (1) it does not have a phonological realization and therefore is not a morpheme whose emergence can be traced, and (2) caseless bare forms are also used as non-specific direct objects, making it difficult to distinguish nominative versus caseless bare forms especially in one-word utterances. The genitive is treated separately because it presents a specific structure where the possessed is marked to agree with the possessor. The instrumental/comitative are marginal cases. A *0ABL column is not included because there are no ungrammatical omissions.

Early nominal morphology in Turkish: Emergence of case and number

45

23. This column presents the emergence of case morphology only, thus the possessive morphology is not included. 24. The authors do not report whether it becomes productive before the accusative case.

References Aksu-Ko¸c, Ayhan 1994 Development of linguistic forms: Turkish. In Relating Events in Narrative: A Crosslinguistic Developmental Study, Ruth A. Berman and Dan I. Slobin, 329–385. Hillsdale, NJ: Lawrence Erlbaum Associates. Aksu-Ko¸c, Ayhan, and F. Nihan Ketrez 2003 Early verbal morphology in Turkish: Emergence of inflections. In Development of Verb Inflection in First Language Acquisition: A CrossLinguistic Perspective, Dagmar Bittner, Wolfgang U. Dressler, and Marianne Kilani-Schoch (eds.), 27–52. Berlin: Mouton de Gruyter. Aksu-Ko¸c, Ayhan, F. Nihan Ketrez, Klaus Laalo, and Barbara Pfeiler 2005 Emergence of agglutinative morphology: A comparative study of Turkish, Finnish and Yucatec Maya. Paper presented at the symposium Emergence of verbal and nominal morphology from a typological perspective, 10th International Congress for the Study of Child Language, Berlin, July 25–29, 2005. Aksu-Ko¸c, Ayhan, and Dan I. Slobin 1985 The acquisition of Turkish. In The Crosslinguistic Study of Language Acquisition, Dan I. Slobin (ed.), vol. 1, 839–878. Hillsdale, N.J.: Lawrence Erlbaum Associates. Allen, Shanley 1996 Assessing productivity in acquisition data from polysynthetic languages: An Inuktitut example. Paper presented at the 7th International Congress for the Study of Child language. Istanbul. Bittner, Dagmar, Wolfgang U. Dressler, and Marianne Kilani-Schoch 2003 Introduction. In Development of Verb Inflection in First Language Acquisition: A Cross-Linguistic Perspective, Dagmar Bittner, Wolfgang U. Dressler, and Marianne Kilani-Schoch (eds.), vii–xxxvii. Berlin: Mouton de Gruyter. Blake, Barry J. 1994 Case. Cambridge: Cambridge University Press. Brown, Roger 1973 A First Language: The Early Stages. Cambridge, MA: Harvard University Press. Bybee, Joan 2001 Phonology and Language Use. (Cambridge Studies in Linguistics 94.) Cambridge: Cambridge University Press.

46

F. Nihan Ketrez and Ayhan Aksu-Ko¸c

Christofidou, Anastasia 1998 Number or case first? Evidence from Modern Greek. In Perspectives on Language Acquisition. Selected Papers from the VIIth International Congress for the study of Child Language, Ayhan Aksu-Ko¸c, Eser ¨ Erguvanlı Taylan, A. Sumru Ozsoy, and Aylin K¨untay (eds.), 46–59. Istanbul: Bo˘gazi¸ci University. Dressler, Wolfgang U. 1997 Universals, typology and modularity in Natural Morphology. In Language History and Linguistic Modeling, Raymond Hickey and Stanislaw Puppel (eds.), 1399–1421. Berlin/New York: Mouton de Gruyter. Dressler, Wolfgang U. (ed.) ¨ 1997 Studies in Pre- and Protomorphology. Wien: Verlag der Osterreichischen Akademie der Wissenschaften. Erguvanlı, E. Eser 1984 The Function ofWord Order in Turkish Grammar. Berkeley: University of California Press. Gagarina, Natalia, and Maria Voeikova 2002 Early syntax, first lexicon and the acquisition of case forms by two Russian children. In Pre- and Protomorphology: Early Phases of Morphological Development in Nouns and Verbs, Maria Voeikova and Wolfgang U. Dressler (eds.), 115–133. Munich: Lincom Europa. G¨urel, Ay¸se 2000 Missing case inflection: Implications for second language acquisition. In Proceedings of the 24th Annual Boston University Conference on Language Development, Catherine Howell, Sarah A. Fish, and Thea Keith-Lucas (eds.), 379–390. Somerville, MA: Cascadilla Press. Haig, Geoffrey 1997 The dative as a default case in Turkish. In Proceedings of the 8th International Conference on Turkish Linguistics. August 7–9, 1997, 37–48. Ankara: Ankara University Press. Haznedar, Belma 2003 Missing surface inflection in adult and child L2 acquisition. In Proceedings of the 6 th Generative Approaches to Second Language Acquisition Conference (GASLA 2002), Juana M. Liceras, Helmut Zobl, and Helen Goodluck (eds.), 140–149. Somerville, MA: Cascadilla Proceedings Project. Ketrez, F. Nihan 1999 Early verbs and the acquisition of Turkish argument structure. M. A. thesis, Bo˘gaziçi University, Istanbul. 2000a What does a child have to acquire when acquiring passive? In Studies on Turkish and Turkic Languages, Aslı G¨oksel and Celia Kerslake (eds.), 283–289. Wiesbaden: Harassowitz Verlag.

Early nominal morphology in Turkish: Emergence of case and number 2000b

47

Eylem o¨ ge ˘ yapılarının ediniminde eden-¨ozne ili¸skisi [agent-subject relation in the acquistion of argument structure], In XIII. Ulusal Dil¨ bilimi Kurultay Bildirileri, A. Sumru Ozsoy and Eser E. Taylan (eds.), 189–199. Istanbul: Bo˘gazi¸ci University Press. 2003 -lAr-marked nominals and three types of plurality in Turkish. Proceedings of CLS-39 (Chicago Linguistics Society Annual Meeting 39.) Chicago: University of Chicago Press. 2005 Children’s scope of indefinite objects. Ph. D. diss., University of Southern California, Los Angeles. Kilani-Schoch, Marianne, and Wolfgang U. Dressler 2002 The emergence of inflectional paradigms in two French Corpora: An illustration of general problems of pre-and proto-morphology. In Preand Protomorphology: Early Phases of Morphological Development in Nouns and Verbs, Maria Voeikova and Wolfgang U. Dressler (eds.), 45–59. M¨unchen: Lincom Europa. K¨untay, Aylin, and Dan I. Slobin 1996 Listening to a Turkish mother: Some puzzles for acquisition. In Social Interaction, Social Context and Language: Essays in Honor of Susan Ervin-Tripp, Dan I. Slobin, Julie Gerhardt, Amy Kyratzis, and Jiansheng Guo (eds.), 265–286. Hillsdale, NJ: Lawrence Erlbaum Associates. 2001 Discourse behavior of lexical categories in Turkish child-directed speech: Nouns vs. Verbs. In Research on child language acquisition: Proceedings for the 8th Conference of the International Association for the Study of Child Language, Margareta Almgren,Andoni Barre˜na, Mar´ıa-Jos´e Ezeizabarrena, Itziar Idiazabal, and Brian MacWhinney (eds.), 928–946. Somerville, MA: Cascadilla Press. Lewis, Geoffrey 1967 Turkish Grammar. Oxford: Oxford University Press. MacWhinney, Brian 2000 The CHILDES Project: Tools for Analyzing Talk. 2 vols, 3rd ed. Mahwah, NJ: Lawrence Erlbaum Associates. Nilsson, Birgit 1985 Case marking semantics in Turkish. Ph. D. diss. University of Stockholm. Sezer, Engin 1978 Eylemlerin c¸ o˘gul o¨ znelere uyumu. (The agreement between plural subjects and the verb). Genel dilbilim Dergisi 1: 25–33, Ankara. Stephany, Ursula 1997 The acquisition of Greek. In The Crosslinguistic Study of Language Acquisition, Dan I. Slobin (ed.), vol. 4, 183–333. Mahwah, NJ/ London: Lawrence Erlbaum.

48

F. Nihan Ketrez and Ayhan Aksu-Ko¸c

Topba¸s, Seyhun, ˙Ilknur Mavi¸s, and Mine Ba¸sal 1997 Acquisition of bound morphemes: Nominal case morphology in Turkish. In Proceedings of the 8th International Conference on Turkish Linguistics. August 7–9, 1997, 127–137. Ankara: Ankara University Press.

Acquisition of case and plural in Finnish∗ Klaus Laalo 1. Aims After a structured overview of the Finnish case system, the acquisition of case markers and case functions as well as early plural forms are studied in the light of corpora including diary material and recordings of two Finnish-speaking children. The theoretical background is usage-based: it is assumed that the child constructs grammatical categories in a stepwise way; the forms which the child picks up from the input and uses in his/her own speech are slowly organized into patterns (schemas) so that the grammatical system of the language acquired will emerge. The important steps are regarded to be premorphology, when the first (rote-learned) forms are acquired and used in short utterances (typically consisting of only one word), and protomorphology, when the child detects morphological regularities and starts to produce analogical forms based on morphological contrasts. Two topics are dealt with in detail: the children’s innovative use of morphological means (especially types of analogies) and the first nominal oppositions (paradigmatic contrasts with two members) and miniparadigms (paradigmatic constrasts with at least three members; for a thorough presentation of the term miniparadigm, see Kilani-Schoch and Dressler 2002: 50–51). Both the analogies and the oppositions are a sign that the child actively processes the resources of the language.

2. The Finnish case system 2.1.

Nominative and oblique cases

The case system of Modern Finnish consists of the unmarked nominative case and three groups of oblique cases. The nominative singular is the basic form of nominals (nouns, adjectives, numerals and pronouns) and has no suffix. The nominative plural is formed with the suffix -t.A function of the nominative which is especially important in early language acquisition is naming. The subject of the sentence is usually in the nominative, but in existential sentences it may be in the partitive.

50

Klaus Laalo

The three groups of oblique cases are grammatical, local and marginal cases. The grammatical and local cases are a core part of Finnish morphosyntax. The main functions of the Finnish cases with special reference to child language are presented in detail in Toivainen (1997: 94–102); their most important functions in the present context will be dealt with in section 3.1 below. Finnish is an accusative type language: while the subject is unmarked the object is usually marked. It remains unmarked only when there is no subject, e.g. in imperative and passive (impersonal) sentences (cf. 2.2). Object marking in sentences with a subject is important, because the word order of Finnish is relatively free serving different textual functions and may also be used to express emphasis or contrast. Therefore the object may precede the verb for several reasons (e.g. thematic or emphatic). In Finnish, location is expressed by means of local cases rather than by prepositions. Some local cases also have other functions such as beneficial (allative) and instrumental (adessive). Morphophonological alternations, notably grade alternation and alternations of the stem-final vowel, often occur in the context of suffix attachment. These morphophonological alternations are so salient that sometimes children detect them before the suffixes and use them as a means of inflecting Finnish words (cf. 3.2.3). To illustrate the Finnish inflectional system, let us compare some different noun paradigms: (1)

Nominative Genitive Partititive Illative PL nominative PL partitive PL illative Gloss

pallo pallo-n pallo-a pallo-on pallo-t pallo-j-a pallo-i-hin ‘ball’

jalka jala-n jalka-a jalka-an jala-t jalko-j-a jalko-i-hin ‘foot’

vesi vede-n vet-t¨a vete-en vede-t ves-i-¨a ves-i-in ‘water’

A stem like pallo ‘ball’ does not have any morphonological alternations. In stems like jalka ‘foot’ there is grade alternation lk : l in certain forms and also a change in the stem-final vowel -a > -o before the plural suffix i/j. In inflectionally complex stems like vesi ‘water’ stem-final e is deleted in the partitive singular (this shortened stem is called the consonant stem), there is grade alternation t : d, and assibilation t > s occurs before i.

Acquisition of case and plural in Finnish

2.2.

51

Grammatical cases

The partitive suffix -TA has four allomorphs in Standard Finnish: ta, t¨a, a and a¨ . The long variants -ta and -t¨a are used with monosyllabic stems (e.g. maa-ta ‘land-PARTIT’), consonantal stems (e.g. vet-t¨a ‘water-PARTIT’) and after a long vowel or diphthong (e.g. tiistai-ta ‘Tuesday-PARTIT’). The short variants -a and -¨a are used after a short vowel with stems of at least two syllables. In colloquial Finnish, this ending -A is assimilated to the final vowel of the stem, which is lengthened (e.g. talo-a ‘house-PARTIT’ is assimilated to taloo and tytt¨o-¨a ‘girl-PARTIT’ to tytt¨oo¨ ). These assimilated forms also frequently occur in child language. The Finnish partitive has many grammatical functions, one of which is marking partial quantity. The ending of the genitive singular is -n; the genitive plural has many allomorphs, but they all have the final n in common. The genitive primarily expresses possession but it also has some grammatical functions, such as marking the head of a postposition (example 2). (2)

sinu-n kanssa-si you-GEN with-POSS.2S ‘with you’

The accusative, which is the most common case for marking the object in Finnish, does not have a stable ending: the accusative singular is sometimes morphologically unmarked (just like the nominative singular) and sometimes homophonous with the genitive. There are also some specific accusative forms of pronouns: all six personal pronouns as well as the interrogative pronoun ken ‘who’ form the accusative with the suffix -t, e.g. kene-t ‘who-ACC’, meid¨a-t ‘we-ACC’. Pronominal forms carrying the suffix -t are described as accusatives in practically all Finnish grammars, whereas the forms with an n-suffix are interpreted as genitives in some of them (e.g. by Hakulinen et al. 2004). The advantage of the latter interpretation is a certain simplification of the nominal paradigm. The disadvantage is that the functions of the genitive become more heterogeneous this way. In the plural, the accusative is always nominative-like and carries the suffix -t. In the singular, the nominative-like accusative marks the object when the subject is not overtly expressed, as in imperative and passive sentences. Otherwise, the genitive-like accusative is used for distinguishing the two central noun phrases of the sentence, the direct object (carrying the suffix -n) and the subject (in the nominative). Examples:

52

(3)

Klaus Laalo

a.

b.

c.

2.3.

imperative

lue kirja read book ‘read the book!’ passive kirja lue-taan book read-PASS ‘the book is read’ active indicative sin¨a lue-t kirja-n you read-2S book-ACC ‘you read the book’

Local cases

The Finnish local cases constitute a fairly coherent system of nine members: three general local cases which are grammaticalized to a great extent (partitive, essive and translative), three internal cases expressing IN-location (illative, inessive and elative) and three external local cases expressing ON- and AT-location (allative, adessive and ablative) and also certain other functions. In each of these three groups, there is one case for expressing motion toward (= TO-cases TRANSL, ILL and ALL), one for expressing the static location in or at/on (= INcases ESS, INESS and ADESS) and one for expressing going out or away from somewhere (= FROM-cases PARTIT, ELAT and ABL). Many local case forms have been fossilized and have become local adverbs which usually come in groups of three in the TO-, IN- and FROM-cases (e.g. koti-in ‘home-ILL’ (= home), koto-na ‘home-ESS’ (= at home), koto-a ‘homePARTIT’ (= from home).

2.3.1. The general local cases: partitive, essive and translative The general local cases are the oldest of the three subgroups of local cases. They have many other functions besides purely local ones: the partitive has different grammatical functions altogether and is nowadays used in local function only in certain adverbs; the role cases essive and translative most often express an abstract location or role. The case ending for translative is -ksi (-kse-), e.g. yst¨av¨a-ksi ‘friend-TRANSL’ (= to [become] a friend), yst¨av¨a-kse-ni ‘friend-TRANSL-POSS.1S’ (= to [become] my friend). The case suffix for essive is -nA, e.g. yst¨av¨a-n¨a ‘friend-ESS’ (= as a friend), lahja-na ‘present-ESS’ (= as a present). The TO-case translative expresses that something is becoming something, e.g. kasva-a suure-ksi ‘grow-3S big-TRANSL’ (= grows and becomes big), and the IN-case essive

Acquisition of case and plural in Finnish

53

often renders a role, e.g. opettaja-na ‘teacher-ESS’ (= as a teacher). The old local use of these three cases is still preserved in some frozen expressions, especially in such adverbs as taa-kse ‘behind-TRANSL’ (= behind-TO) : taka-na ‘behind-ESS’ (= behind-IN) : taka-a ‘behind-PARTIT’ (= from behind) formed from the stem taka- ‘behind’ or in the postposition luo-kse ‘to’, luo-na ‘at, by’, luo-ta ‘from’.

2.3.2. External local cases (AT-cases, ON-cases) The three external local cases are allative ‘to’ (suffix -lle), adessive ‘at, on’ (suffix -llA) and ablative ‘from’ (-ltA). These cases have other functions besides the purely local ones. The adessive is also used to express instrumental and possessive functions with inanimate and animate nouns, respectively: (4)

a.

b.

avaime-lla key-ADESS ‘with the key’ minu-lla me-ADESS ‘in my possession, at my disposal’

The beneficiary function of the allative corresponds to the dative in certain other languages: (5)

sinu-lle you-ALL ‘to you’

2.3.3. Internal local cases (IN-cases) The three internal local cases are illative ‘into’, inessive ‘in’ (-ssA), and elative ‘out of’(-stA). Of all groups of local cases, the internal ones have the purest local functions.The illative suffix has many allomorphs. The allomorphs are -hVn after diphthongs and with monosyllabic stems, e.g. tiistai-hin ‘Tuesday-ILL’ (= till Tuesday), talo-i-hin ‘house-PL-ILL’ (= into the houses), maa-han ‘ground-ILL’ (= into the ground), -Vn after a short vowel in all other syllables than the first, e.g. talo-on ‘house-ILL’ (= into the house), banaani-in ‘banana-ILL’ (= into the banana), and -seen (siin in the plural), e.g. estee-seen ‘obstacle-ILL’ (= into the obstacle), este-i-siin ‘obstacle-PL-INTO’ (= into the obstacles).

54

2.4.

Klaus Laalo

Marginal cases

Besides the grammatical and local cases, there are the following marginal cases: comitative instructive abessive

-ine ‘with’ -in ‘by means of’ -ttA ‘without’

There is the following evidence for the marginal status and the relative unproductiveness of the above cases: First, these cases do not occur as frequently as other cases. Second, there is some alternation between synthetic marking and prepositional as well as postpositional constructions, e.g. raha-tta ‘money-ABESS’ (= without money) ∼ ilman rahaa ‘postposition money-PARTIT’ (= without money), laps-ine-en ‘child-COM-poss.suff.’ (= with his/her children) ∼ lastensa kanssa ‘child-poss.suff’ (= with his/her children). Third, the case suffix of the comitative (-ine) is fused with the plural suffix, but the comitatives are used for both singular and plural referents. The comitative and the instructive are only productively used in the plural; in singular, the suffix of the instructive is identical with the suffix of the genitive and that is why the instructive is used only in frozen expressions in the singular, e.g. jala-n ‘foot-INSTRUC’ (= afoot). The most frequently occurring marginal cases are the instructive and the comitative, but children start to use them rather late. There are transition problems between the marginal cases and adverbs (P¨ochtrager et al. 1998: 77–78). If the marginal case forms are taken to be adverbs, their suffixes are derivational rather than inflectional.

3. The acquisition of the Finnish case system 3.1.

Forms and functions of the first case markers, a general overview

Finnish-speaking children use case forms from the very beginning. The reason is that nouns always occur in a certain case form in the input. Case forms occurring in the one-word stage are mostly suffixless nominatives, but a great number of nouns also occur in the partitive and in certain local case forms. Especially illatives such as koti-in ‘home’ and syli-in ‘into the lap’ are used early by many children. Grammatical cases emerge early in the speech of Finnish children. The two main reasons seem to be their high frequency in the input language and their overall shape (Gestalt), which is shorter and less complex than that of local (and

Acquisition of case and plural in Finnish

55

marginal) cases, because the case suffix does not constitute an additional syllable (except in some partitive forms, which are however often contracted in colloquial Finnish: -VA > -VV, e.g. maito-a ‘milk-PARTIT’ > maito-o, and -tA > -t, e.g. mon-ta ‘many-PARTIT’ > mon-t). The relevance of syllable structure for the order of emergence is supported by the fact that the first local case used by Finnishspeaking children is usually the illative (Toivainen 1980: 113), especially its short allomorph -Vn (lengthening of the final stem vowel + n). In early child language, this form may also replace the longer TO-case allative (see 4.2 below). The first oblique cases used by Finnish-speaking children fulfill the following functions (cf. Toivainen 1997: 94–102) and have the following inflectional markers: – partitive: object marking (partial quantity, e.g. food), the suffix variants -TA and -A ∼ -V (= vowel lengthening, assimilated from -A), e.g. vet-t¨a ‘waterPARTIT’ (= some water), maito-a ∼ maito-o ‘milk-PARTIT’ (= some milk), mehu-a ∼ mehu-u ‘juice-PARTIT’ (= some juice); – genitive: possession, marking the head of a postposition, suffix –n; – accusative: object marking (total object + resultative aspect), the suffix of accusative singular is mostly -n but see 2.2; – illative: motion to(ward) something, the most common suffix variant is -Vn = lengthening of the final stem vowel + n, e.g. syli-in ‘lap-ILL’ (= into the lap); – other local cases: besides location, also beneficiary (allative) and instrument + possessor (adessive). The functions of beneficiary of the allative (e.g. a¨ idille ‘mother-ALL’ (= to the mother)) and possessor (and instrument) of the adessive are so important in the use of these cases that when expressing concrete location, children occasionally use the internal local cases illative and inessive instead of the external local ones allative and adessive (cf. 4.2 below). The illative tends to be the first local case form. When it emerges, children usually start with the short allomorph -Vn lengthening the stem-final vowel and adding n. The reason may be that this allomorph is shorter than the others and also lighter in the sense that it does not constitute a syllable of its own but is attached to the last vowel of the stem forming the coda of the last syllable. In the Finnish child language, the illative can be favoured in early local expressions at the expense of the external local case allative, which is marked by a longer suffix and also carries other functions besides the local one. These first case forms used by the child constitute the nucleus of the Finnish case system. In the following section, we will present a detailed analysis of the acquisition of the case system by two Finnish-speaking children.

56

Klaus Laalo

3.2. The development of singular case forms in two Finnish-speaking children: diary data 3.2.1. Diary data and recordings The main subject of the present study is a girl, Tuulikki, who was born in June 1991. Diary data were gathered fairly intensively from her speech ever since her first words at the age of 1;0 for about five years; the latest diary data are from her school age. Recordings of about one hour per month have been made since the age of 1;7; there is no recording at 2;0 and only a short one (of about 5 minutes) at 2;7. Her brother Tuomas, born six years later, will be studied especially in the early stages of language acquisition. There are diary data from the boy’s first words and recordings of 30 to 60 minutes per month starting from the babbling period until school age. In the present paper, only the period from 1;7 to 2;2 has been analyzed. The recording situations are mostly free play situations at home besides a few recordings where picture books are looked at and discussed. The children interact mostly with their parents and only occasionally with each other.

3.2.2. Emergence of the first cases At the onset of speech production, Finnish-speaking children typically use nouns in just a single form. Count nouns preferably occur in the nominative, e.g. the parental names a¨ iti ‘mother’ and is¨a ‘father’, names of body parts, such as Tuulikki’s nen¨a ‘nose’ at 1;4 and mas(s)u ‘tummy’ at 1;8, or names of animals (hauva ‘bow-wow’, heppa ‘gee-gee’etc.) and common nouns referring to people such as Tuulikki’s vauva ‘baby’ and tytt¨o ‘girl’ at 1;3. Early nominatives of Tuomas are auto ‘car’ at 0;10, and the phonetically variable pa(a)ppa ∼ bappu ∼ b¨appy ‘ball’ at 1;0. Uncountable nouns first occur in the partitive. Tuulikki’s first partitives were rote-learned expressions connected with eating routines: puuvo-o ‘porridgePARTIT’ (= some porridge) at 0;11, pulla-a ‘bun-PARTIT’ (= some buns) at 1;0, and tet-t¨a (< vet-t¨a) ‘water-PARTIT’ (= some water) at 1;3. The first partitives of Tuomas were much alike: 0;8 tet-t¨a (< vet-t¨a) ‘water-PARTIT’ (= some water), and 0;10 puvvo-o ‘porridge-PARTIT’ (= some porridge). At the ages given, the partitive forms of these lemmas were not yet opposed to the nominative. The next cases to emerge in both children’s speech were the illative (especially forms with the allomorph -Vn) and a form suffixed by –n which can either function as a genitive or as an accusative. Typically, the first local case forms of the two subjects studied in this paper and of Finnish-speaking children in general

Acquisition of case and plural in Finnish

57

are adverblike (cf. 3.2.4), so that the children may not be aware of the fact that these forms are inflected for certain cases and that they function as frozen forms in their speech. In this early period, the children use inflected forms of nouns as unsegmented chunks or gestalts. The very first partitive and illative forms have a clear connection to psychosocial development since they serve children’s primary needs: with the partitive forms, they ask for a number of different things, for example food; with the illative forms, they express the wish to get somewhere (e.g. koti-in = koti ‘homeILL’ (= let’s go home)) or to get something some place (e.g. to-hon for Standard Finnish tuo-hon ‘that-ILL’ (= over there), meaning ‘put it over there’). The genitive-accusative is the first form to be systematically contrasted with the inflectionally unmarked basic form, the nominative. There is some variation as far as the early marking of the genitive-accusative in concerned: While some children concentrate on the suffix n and do not observe stem alternations (e.g. Tuomas 1;8 ukko ‘old man’ : ukko-n ‘old man-GEN’ vs. the standard genitive uko-n with quantitative gradation kk : k), other children may concentrate on stem alternation and omit the suffix. Thus, Tuulikki’s early genitives are marked with quantitative gradation at the age of 1;7, e.g. lappu ‘bib’ : lapu ‘bib:GEN’ vs. the standard genitive lapu-n ‘bib-GEN’. (Note that in Finnish there is quantitative gradation of stops, namely kk : k, pp : p, tt : t, besides different types of qualitative gradation, e.g. p : v as in leip¨a ‘bread’ : leiv¨a-n ‘bread-GEN’, t : d as in a¨ iti ‘mother’ : a¨ idi-n ‘mother-GEN’, k and loss as in maku ‘taste’ : mau-n ‘tasteGEN’ etc.).

3.2.3. The suffix variants of the first case forms: the trochaic tendency This section is based on the diary data of both children from the very beginning of speech production. The suffix variant of the first rote-learned partitives in Tuulikki’s speech is mostly vowel lengthening, which is a colloquial variant of the partitive suffix –A and is also frequently used by Tuulikki’s and Tuomas’s parents: (6)

a.

Tuulikki 1;5 mehu-u (< mehu-a) juice-PARTIT ‘some juice’

In stems ending in A, the standard suffix is identical with vowel lengthening, as in b.

Tuulikki 1;5 hyv¨a-¨a good-PARTIT ‘(this tastes) good’

58

Klaus Laalo

c.

d.

Tuulikki 1;6 kuuma-a hot-PARTIT ‘(this is) hot’ Tuulikki 1;6 leip¨a-¨a bread-PARTIT ‘some bread’

In monosyllabic stems and consonantal stems the suffix is –tA: e.

f.

g.

Tuulikki 1;3 t¨a-t¨a this-PARTIT ‘this kind of (food)’ Tuulikki 1;7 uut-ta new-PARTIT ‘something new’ Tuulikki 1;7 lien-t¨a soup-PARTIT ‘some soup’

When first using the partitives uutta and lient¨a at 1;7, Tuulikki repeated them several times together with the respective nominative forms (uusi ‘new’ : uut-ta and liemi ‘soup’ : lien-t¨a) as if pondering about how they might be related to each other. The first rote-learned partitives of Tuomas have the same variants as his sister’s, namely -A ∼ -V in two-syllabic nouns and -tA in monosyllabic nouns and consonantal stems. A phenomenon typical of early child language is the so-called trochaic stage, which restricts the concatentation of suffixes.This stage is not only observed with the two children studied in the present paper but can be observed with children acquiring Finnish more generally. In the trochaic stage, bisyllabic word forms are preferred (R¨ais¨anen 1975: 256; Laalo 1994, 2001; Savinainen-Makkonen 2000; cf. Wijnen, Krikhaar, and Den Os 1994; Gerken 1994).There are also some four-syllabic, prosodically compound-like forms such as elepantti ‘elephant’ (R¨ais¨anen 1975: 256) andTuulikki’s ala-antti < kameleontti ‘chameleon’(at 1;8) consisting of two consecutive trochee feet. The preference for trochaic patterns may lead to the shortening of standard word forms in the following way: The initial syllable, which bears the main stress and is accordingly prosodically salient, and the second syllable, which belongs to the same trochaic foot, are preserved while the other syllables are deleted (e.g. Tuulikki’s 1;5 ikkuna ‘window’ > ikku,

Acquisition of case and plural in Finnish

59

1;7 molemmat ‘both’ > mole, 1;7 patteri ‘radiator’ > patte). Sometimes the second and third syllables are merged (e.g. Tuulikki’s 1;8 puolukka ‘lingonberry’ > puukka). The trochaic tendency was strong in Tuulikki’s speech but weaker in Tuomas’s although he also used trochaic shortenings such as ikku (< ikkuna ‘window’) at 1;5 and api (< apina ‘ape’) at 1;8. Although using the salient nucleus of the word consisting of its first two syllables is quite effective for communicating lexical meaning, the preference for trochees imposes limitations on the development of inflection, and the use of morphological elements is accordingly restricted in this stage. By shortening word-forms to their relatively invariant initial parts, thereby leaving it to the extralinguistic and linguistic context to specify the attempted grammatical form, the trochaic tendency represents one way to cut the complexity of Finnish word forms down to their semantic essentials. This way the language becomes more manageable at a stage of development when the functions of the entire system of complex Finnish morphology with all its derivative and inflectional endings is only developing and far from having been mastered. Truncated forms of nouns lacking suffixes are not the only ones to occur in the children’s speech during the trochaic stage, however. Both Tuulikki and Tuomas also use a special type of shortened partitive forms consisting of a truncated noun stem with the appropriate partitive suffix (-A or vowel lengthening) attached to it. A similar partitive based on a shortened stem is also mentioned in Toivainen (1980: 125): Ville 1;11 bani-a (< banaani-a) ‘banana-PARTIT’. Such forms based on truncated stems represent another strategy used by young children in their struggle to come to grips with word forms exceeding a certain length by reducing them to more manageable entitities. Examples of trochaic truncations from Tuulikki’s data: (7)

a. b.

1;6 NOM bana (< banaani) ‘banana’ : PARTIT bana-a 1;8 NOM appe (< appelsiini) ‘orange’ : PARTIT appe-e

Examples of trochaic truncations from Tuomas’s data: (8)

a. b.

1;7 PARTIT pipa-a (< pipari-a) ‘bisquit-PARTIT’ 1;8 PARTIT puukka-a (< puolukka-a) ‘lingonberry-PARTIT’

Besides endingless forms and partitives, Tuulikki uses (bisyllabic) illatives from as early as 1;5 onwards. In the input, the most frequent variant of the Finnish illative is -Vn (= vowel lengthening + n), but also in the input the final n is often weakened or dropped; the longer illative variant -hVn is used in monosyllables, and this suffix may be shortened to -hV in a similar way. Examples:

60

(9)

Klaus Laalo

a.

b.

Tuulikki 1;6 suu-hun mouth-ILL ‘into the mouth’ Tuulikki 1;7 kuppi-i(n) cup-ILL ‘into the cup’

Tuomas’s first illatives have similar suffixes: (10)

a.

b.

Tuomas 1;5 t¨a-h¨a(n) this-ILL ‘here’ Tuomas 1;6 kaappi-i(n) cupboard-ILL ‘into the cupboard’

As with the early partitives, there are also often trochaic truncations to be found among the early illatives in Tuulikki’s speech: a truncated noun stem carries the case suffix (examples 11). This is evidence that the child has noticed that illatives should have a long final vowel: (11)

a.

b.

c.

< kattila-an saucepan-ILL ‘into the saucepan’ 1;8 keitti-i(n) < keitti¨o-¨on kitchen-ILL ‘into the kitchen’ 1;8 ty¨okk¨a-¨a(n) < ty¨okk¨ari-in [coll. shortening of ty¨ohuonee-seen] workroom-ILL ‘to the workroom’

1;7 katti-i(n)

Tuomas uses trochaic shortenings in the illative only occasionally, e.g. 1;7 kattiin < kattila-an ‘saucepan-ILL’ (= into the saucepan), but two months later he produces some illatives with the longer suffix -seen, e.g. 1;9 *kyypy-seen (∼ kyypy-yn < kylpy-yn) ‘bath-ILL’ (= into the bath) by analogy with ammeeseen ‘bathtub-ILL’ (= into the bathtub, nominative amme). The first accusatives and genitives in Tuulikki’s speech mostly contrast with the nominatives. In the beginning, Tuulikki uses forms functioning as genitiveaccusative both with the n-suffix and without it. Thus, her first forms used in genitive or accusative contexts at the age of 1;7 have the following character-

Acquisition of case and plural in Finnish

61

istics: (a) They can be suffixless variants such as lapu (< lapu-n ‘bib-ACC’), meko (< meko-n ‘dress-ACC’) distinguished from the nominatives lappu ‘bib’ and mekko ‘dress’ by consonant gradation or suffixless variants distinguished from the nominative byA ∼ U alternation (e.g. minu ‘I:GEN’vs. min¨a ‘I:NOM’). The final n of genitive forms is also often weakened in Toivainen’s data (1980: 141–142) so that the inflection is only signaled by stem alternation. (b) Forms functioning as accusatives and expressing the direct object can also carry the suffix -n, e.g. kaike-n Tuuti s¨oi ‘everything-ACC Tuuti ate’ (= Tuuti ate everything), tuti-n vie ‘the baby’s dummy-ACC takes’ (= takes the baby’s dummy). There are also genitives ending in -n constructed with a postposition: tori-n poikki ‘the market square-GEN across’ (= across the market square), leiv¨a-n p¨aa¨ -lle ‘bread-GEN head-ALL’ (= on the bread). In contrast to his sister, Tuomas does not omit the case ending -n in accusative and genitive forms but at first ignores most stem alternations simply adding the suffix -n to the nominative stem, e.g. 1;8 k¨asin (for k¨aden vs. NOM k¨asi ‘hand’) and 1;8 ukkon (for ukon, cf. NOM ukko ‘old man’), tytt o¨ n (for tyt o¨ n, cf. NOM tytt¨o ‘girl’). As can be seen from these examples, the two siblings use different strategies for producing nouns inflected for the genitive-accusative as opposed to the nominative. While Tuulikki pays attention to the overall differences between the two form classes, including stem alternation but sometimes ignoring the suffix –n, her brother produces the suffix, but does not observe gradation. The children studied in Toivainen (1980: 143–144) sometimes omit qualitative grade alternation, especially in the genitives of the kinship terms a¨ iti ‘mother’, set¨a ‘uncle’ and t¨ati ‘aunt’, but not quantitative grade alternation: papan ‘Grandpa’s’ (vs. NOM pappa), tyt¨on ‘the girl’s’ (vs. NOM tytt¨o). Tuulikki and Tuomas not only differ in the way they distinguish the genitiveaccusative from the nominative but also in the functions to which they put their early genitives. Tuulikki’s first genitives mainly link a noun to a postposition (e.g. tori-n poikki ‘market square-GEN across’ (= across the market square), whereas most of Tuomas’s early genitives are possessive. While, at the age of 1;7, Tuulikki uses accusatives and linking genitives besides nominatives, her first possessive genitives are found at the age of 1;8. She uses them to indicate, for example, the possessors of different servings of porridge, e.g. isi-n puuro ‘Daddy-GEN porridge’ (= Daddy’s porridge). Tuomas’s first accusatives and possessive genitives occur at 1;8 and his first linking genitives at 1;9. The difference between the two children’s development of the functions of the genitive can probably be explained by the fact that, being a second child, Tuomas has a greater need to express the possessive relation. This difference between first-born and younger siblings seems to be fairly common.

62

Klaus Laalo

3.2.4. From local adverbs to productive use of local case forms As far as the development of local expressions to be studied in this section is concerned, Finnish-speaking children at first use local adverbs marked for case as frozen forms. Thus, the first ablatives are typically local adverbs such as t¨aa¨ -lt¨a ‘from here’, tuo-lta ‘from over there’ and sie-lt¨a ‘from there’ and the first elatives belong to the adverbial type t¨a-st¨a ‘from here’, tuo-sta ‘from over there’, mi-st¨a ‘from where’ etc. The first inessives used by Tuulikki are mi-ss¨a ‘what-INESS’ (= where) at 1;5, t¨a-ss¨a ‘this-INESS’ (= here) at 1;7, and maa-ssa ‘ground-INESS’ (= on the ground) at 1;8. In the trochaic stage, trisyllabic forms are shortened so that the inflection is marked only by stem alternations, e.g. grade alternation in 1;8 vaipa < vaipa-ssa ‘nappy-INESS’ (= in the nappy), cf. NOM vaippa ‘nappy’, or a combination of different alternations, e.g. 1;9 k¨ale < k¨ade-ss¨a ‘hand-INESS’ (= in the hand), cf. NOM k¨asi ‘hand’). This type of unspecified oblique forms decrease from the age of 1;10 on, when Tuulikki starts to use trisyllabic inessives besides bisyllabic ones: k¨ade-ss¨a ‘hand-INESS’ (= in the hand, cf. NOM k¨asi ‘hand’), m¨oki-ss¨a ‘cottage-INESS’ (= in the cottage, cf. NOM m¨okki ‘cottage’), laatiko-ssa ‘box-INESS’ (= in the box, cf. NOM laatikko ‘box’) etc. As indicated above, Tuomas’s trochaic stage is less pronounced and shorter. Accordingly, he uses both di- and trisyllabic inessives from the very beginning, e.g. at 1;8 to-ssa ‘that-INESS’ (= over there), nen¨a-ss¨a ‘nose-INESS’ (= in the nose), puu-ssa ‘tree-INESS’ (= in the tree); at 1;9 auto-ssa ‘car-INESS’ (= in the car), kaapi-ssa ‘cupboard-INESS’ (= in the cupboard), kuva-ssa ‘pictureINESS’ (= in the picture). Since the consonant cluster of the elative suffix -stA is articulatorily demanding, it is often assimilated to –ttA in early child language. Because the inessive suffix -ssA is often changed to -ttA in early child language, the suffixes of the elative and the inessive may coincide. This may be one of the reasons why Tuulikki occasionally uses different strategies to express the elative function in an unambiguous way. One such strategy is the use of periphrastic constructions such as the adposition (also adverb) pois ‘away’ combined with a noun unmarked for case: (12)

Tuulikki 1;9 suikku pois (< suihku-sta) shower away (< shower-ELAT) ‘from the shower’

Another strategy for expressing the elative used by both children is the same as that already encountered with some early genitive-accusatives, namely a suf-

Acquisition of case and plural in Finnish

63

fixless stem form with stem alternation, either gradation (laatiko < laatiko-sta ‘box-ELAT’ (= from the box) vs. NOM laatikko) or alternation of the final stem vowel (kuole < kuore-sta ‘shell-ELAT’ (= from the shell) vs. NOM kuori). The underspecified expression of location by use of the suffixless oblique form of the noun stem is also found in Tuomas’s data, e.g. 1;8 polve for adessive polve-lla ‘knee-ADE’ (= on the knee) vs. NOM polvi. In such utterances, the specific locative relation must be inferred from the context. Sometimes the adposition pois does not replace synthetic case marking in the children’s speech but marks the elative function more clearly by expressing it both synthetically and periphrastically, just as in the redundant expressions of adult Finnish. The adposition may also clarify the case function when the case suffix is obscured by phonetic changes: (13)

a.

b.

c.

Tuulikki 1;10 pois suu-tta (< suu-sta) away mouth-ELAT ‘out of the mouth’ Tuulikki 1;10 pois tuoli-tta (< tuoli-sta) away chair-ELAT ‘off the chair’ Tuomas 1;9 konee-tta pois (< konee-sta) machine-ELAT away ‘away from the machine’

The functions of the external local cases are often first expressed by adverbs, e.g. Tuulikki 1;5 tuo-l(l)a (that-ADESS ‘over there’), 1;6 t¨aa¨ -ll¨a (this-ADESS ‘over here’). The onset /lt/ of the suffix of the external FROM-case ablative is articulatorily demanding and is often simplified to single t (14a, 14b) or geminate ll (14c). The adposition pois ‘away’ is occasionally used in constructions with the ablative function for similar reasons as with the elative. By simplification of the consonant cluster of the suffix the ablative may become homophonous with the partitive or the adessive in the children’s speech; in order to distinguish the ablative from the partitive (14a, 14b) or from the adessive (14c) or to clarify the case function of the bare stem (14d) the children may add the adposition pois: (14)

a.

Tuulikki 1;8 pois p¨aa¨ -t¨a (< p¨aa¨ -lt¨a, cf. PARTIT p¨aa¨ t¨a) away head-*ABL ‘take off’

64

Klaus Laalo

b.

c.

d.

Tuomas 1;8 pois tie-t¨a (< tie-lt¨a, cf. PARTIT tiet¨a) away way-*ABL ‘out of the way’ Tuomas 1;10 patta-lla (< patja-lta, cf. ADESS patjalla) pois mattress-*ABL away ‘off the mattress’ Tuulikki 1;9 pois p¨oyl¨a (< p¨oyd¨a-lt¨a, cf. NOM p¨oyt¨a) away table (< table-ABL) ‘from the table’

The order of development of the local case suffixes found in the data of Tuulikki and Tuomas is also attested in other Finnish-speaking children. Thus, Toivainen (1980: 160–163) presents the following serial order of acquisition of the local case forms based on data from 25 Finnish-speaking children aged 1–3 years and defined by the age of the median child: adessive adverb (e.g. t¨aa¨ -ll¨a ‘here’) illative nominal inessive adverb (e.g. mi-ss¨a ‘where’) allative adverb (cf. the "sublatives", e.g. mi-nne ‘where to’) adessive nominal inessive nominal allative nominal elative adverb (e.g. mi-st¨a ‘from where’) illative adverb (e.g. mi-hin ‘where to’) ablative adverb (e.g. sie-lt¨a ‘from there’) elative nominal ablative nominal It is interesting to note that with the exception of the illative local cases first emerge in deictic local adverbs. The different behaviour of the illative may be explained by the fact that many of the first illatives to occur in child speech do not correspond to established particles in the standard language so that there is some lexical variation here. Nevertheless, the early nominal illatives are used in much the same way as local adverbs, e.g. syli-in ‘into the lap’ (cf. NOM syli ‘lap’) and koti-in ‘home’ (cf. NOM koti ‘home’), which first occur as frozen forms of nouns in our data; there are also adverbs such as muka-an ‘along’ (cf.

Acquisition of case and plural in Finnish

65

Table 1. Tuulikki’s local case forms in the early diary data (types) Age 1;5 1;6 1;7 1;8 1;9 1;10 1;11

Illative 6 6 8 22 20+ 20+ 20+

Inessive 1 ADV 1 ADV 2 ADV 4 6 18 9

Elative – – 1 ADV 2 5 10 8

Adessive 1 ADV 1 ADV 1 ADV 2 ADV 3 16 12

Allative – – 2 ADV 2 ADV 3 14 12

Ablative – – – 1 ADV 3 3 5

muka-na ‘with’, a frozen essive form of the same stem) among the first illatives. There is no clear-cut boundary between the use of these early frozen forms and the productive nominal use of local cases: As the child acquires more word forms, the different suffixes become more familiar and the means of expressing location synthetically develop. The early development of Tuulikki’s local cases is presented in table 1. Illatives are already used in 6 lemmas at the age of 1;5, and at the age of 1;8 illatives occur in as many as 22 lemmas. After that, the diary notes about illatives are no longer comprehensive. The first inessives (mi-ss¨a ‘where’) and adessives (tuo-lla ‘over there’) which occur between 1;5 and 1;6 are adverbs and so are the first elatives (1;7), allatives (1;7) and ablatives (1;8). There is a spurt in the use of local case suffixes with different lexemes between 1;9 and 1;11. After that, the diary notes no longer focus on this topic. Nevertheless, the figures show that the early item-based categories develop into more general inflectional schemas: While the child first uses some adverbs carrying certain case endings, later on she starts to use these case suffixes with more and more lexical stems.

3.2.5. Marginal cases The marginal cases are the last case forms to emerge because of their low frequency in the input. There are only a few abessives, instructives and comitatives in the data. The use of these cases will be demonstrated by a number of selected examples from the diaries of the two children. In Tuulikki’s diary data, the first abessives are nominalized forms of verbs, namely forms of the mA-infinitive: (15)

Tuulikki 2;9 j¨at¨a-n se-n sy¨o-m¨a-tt¨a leave-1S it-ACC eat-INF-ABESS ‘I leave it uneaten’ (= I won’t eat it)

66

Klaus Laalo

Tuulikki’s first instructives are common Finnish frozen forms, e.g. 1;11 k¨as-i-n (hand-PL-INSTRUC) ‘by hand’, 3;7 kyns-i-n (nail-PL-INSTRUC) ‘by using nails’. Tuulikki also produced a few instructive forms of her own: (16)

Tuulikki 4;2 suur-i-n t¨omistyks-i-n big-PL-INSTRUC loud.step-PL-INSTRUC ‘with big, loud steps’

Tuomas was not as innovative as his sister in this respect, so that his instructives rather belong to conventional Finnish phraseology as in the following example based on the word voima ‘force’: (17)

Tuomas 2;9 yhteis.voim-i-n joint.force-PL-INSTRUC ‘with joint forces’

The comitative kenk-i-ne-en ‘with the shoes’ (cf. NOM SG kenk¨a ‘shoe’) used by Tuulikki demonstrates a feature typical of colloquial Finnish, namely substitution of the first person possessive suffix by the third person: (18)

Tuulikki 3;4 m¨a kulje-n kenk-i-ne-en ulko-na I walk-1S shoe-PL-COM-POSS3 out-ESS ‘I walk outside with the shoes on’

A Standard Finnish comitative form is ruotoineen in: (19)

Tuulikki 4;0 oikea-t hylkee-t sy¨o-v¨at ruoto-i-ne-en real-PL seal-PL eat-3P bone-PL-COM-POSS3 ‘real seals eat (fish) with its bones’

The first comitatives found in Tuomas’s diary data are found in a playing situation where he speaks about a toy train: (20)

Tuomas 3;2 tunnele-i-ne-en ja silto-i-ne-en tunnel-PL-COM-POSS3 and bridge-PL-COM-POSS3 ‘with its tunnels and bridges’

It is interesting to note that in Toivainen’s (1980) audiotaped data of 25 children between 1;0 and 3;0 there are no examples of instructive, comitative or abessive

Acquisition of case and plural in Finnish

67

forms. This is not surprising since the marginal cases are infrequent in early child speech so that quite intensive recordings would be required in order to document them; in my own recorded material there is only one instructive form (see section 5.3).

3.2.6. Summary of case development in the light of the diary data The very first occurrences of the case forms have been registered in the diary data.This data represents the overall development of the children quite faithfully: The first cases to emerge are the grammatical ones, followed by local cases and finally by marginal cases. The cases first emerge as item-based categories but gradually develop into more general inflectional schemas. Diary data enable the researcher to make interesting qualitative observations. If the material is collected regularly, the emergence of the very first inflectional forms, various special forms, different analogies and many infrequent forms can be registered (see section 4 for the examination of certain interesting phenomena). A well-known weakness of diary data is that the focus is on novelties while most of the established forms are not registered. Thus a quantitative analysis would be unrewarding. Accordingly, the use of the local case suffixes is documented quite exhaustively in Tuulikki’s diary data till the age of 1;10 but not after that age. In what follows, the analysis of diary material will be complemented by the analysis of the audiotaped data.

3.3. The development of singular case forms from 1;7 to 2;10: audiotaped data In this section, the analysis of the case development is supplemented by the quantitative analysis of the audiotaped data. The period of tape recordings partially overlaps with that of the diaries. While the diaries were kept from the very first words produced by the children with a gradual decrease of their intensiveness as time went on, the tape recordings started for Tuulikki at the age of 1;7 and for Tuomas a little earlier, although useful speech samples of the boy actually also date only from 1;7 on. An overview of singular case forms of nouns found in both children’s tape recordings between the ages of 1;7 and 2;10 for Tuulikki and between 1;7 and 2;2 for Tuomas are presented in tables (2a) and (2b), respectively (case forms based on pronominal stems have been excluded; for plural forms of nouns see section 5 below).

68

Klaus Laalo

Table 2a. Singular case forms of nouns in the early recordings of Tuulikki (types/tokens): grammatical and local cases Age Grammatical cases NOM PARTIT GEN 1;7 35/60 8/11 1/2 1;8 45/117 12/23 7/11 1;9 37/83 15/21 6/16 1;10 10/35 5/12 3/6 1;11 19/59 1/5 3/9 2;1 44/90 5/5 3/3 2;2 47/85 9/11 3/7 2;3 47/86 6/7 2/2 2;4 32/51 4/4 2/3 2;5 29/56 9/9 4/4 2;6 26/37 6/8 3/3 2;8 37/71 10/12 4/6 2;9 30/47 11/13 1/1 2;10 47/60 16/22 6/6

ACC 3/3 3/6 2/2 – 1/1 1/1 2/2 3/4 4/5 2/2 3/3 1/2 1/1 5/5

Local cases ILL INESS 10/18 – 14/29 – 6/11 – 3/6 – 2/2 1/1 6/9 6/12 2/3 1/1 2/2 1/1 2/4 1/1 2/2 9/11 5/6 1/2 7/12 1/1 7/16 6/7 10/14 4/4

ELAT – – – – – 1/1 – 1/1 1/1 – – 2/5 2/2 4/5

ALL – – – (1/2) 4/5 2/2 3/6 1/1 1/1 – 3/5 1/1 4/4 1/1

ADESS – – – (1/1) 1/1 7/8 – 10/17 8/11 – 1/1 4/4 1/1 1/1

ABL – – – – 1/3 – – – – – 1/1 1/1 1/1 –

Table 2b. Singular case forms of nouns in the early recordings of Tuomas (types/tokens) Age Grammatical cases NOM PARTIT GEN 1;7 33/201 – – 1;8 45/133 8/26 3/5 1;9 51/149 14/41 8/33 1;10 16/70 6/10 6/6 1;11 23/32 5/7 2/2 2;0 25/54 1/1 1/1 2;1 16/26 – 2/3 2;2 38/96 5/6 5/11

ACC – 1/2 2/2 – 3/3 1/1 2/2

Local cases ILL INESS 1/2 – 7/8 – 5/12 1/1 5/13 1/1 1/1 2/5 5/7 1/1 3/3 2/5 3/5 8/11

ELAT – 1/1 1/1 1/1 – 1/2 2/2 1/2

ALL – – 3/4 1/1 1/1 – 2/3 –

ADESS – – 2/5 – – 2/2 3/7 8/10

ABL – – 1/1 – – – – –

In the recordings, especially nominative and partitive occur very frequently, but genitive and accusative forms are also regularly used from the first recording on by Tuulikki; with Tuomas, the genitive is in constant use from 1;8 on. The illative is the only local case occurring since the first recording. In the early recordings, illatives are used frequently for different reasons: at 1;7, picture books are looked at and afterwards they are then stacked (pino-on ‘stack/pileILL’); in a pretend game played at 1;8, tea and coffee are poured into the cups of various real and imaginary persons (kuppi-in ‘cup-ILL’). During this early period, allatives are sometimes replaced by illatives (4 tokens at 1;8; cf. section

Acquisition of case and plural in Finnish

69

4.2). The very first allative form in Tuulikki’s recordings is p¨aa¨ -lle ‘head-ALL’ (= onto) and the first adessive is the postposition a-lla ‘under-ADESS’in tyyny-n a-lla ‘pillow-GEN [particlestem]-ADESS’ (= under the pillow), both occurring at 1;10. It must be pointed out, however, that such forms are fossilized forms in the adult language as well. The first allatives properly speaking emerge at 1;11 in Tuulikki’s data.When allatives emerge and start to function as TO-cases at the age of 1;10 and 1;11, the number of illatives decreases temporarily. In spite of the fact that other local cases occur less frequently in the recorded data, there is considerable lexical variation from one month to the next so that given case forms are not mere repetitions but occur with different lemmas: there are allatives such as 1;11 hylly-lle ‘shelf-ALL’ (= on the shelf, local function), 1;11 a¨ idi-lle ‘mother-ALL’ (= to the mother, beneficiary function), 2;3 isi-lle ‘father-ALL’ (= to the father, beneficiary), 2;4 ove-lle ‘door-ALL’ (= to the door, local function); elatives such as 2;1 peiko-sta ‘troll-ELAT’ (= from the troll), 2;3 talli-sta ‘stable-ELAT’ (= from the stable); inessives such as 1;11 alu-ssa ‘beginning-INESS’ (= at the beginning), 2;4 auto-ssa ‘car-INESS’ (= in the car) etc. There are also oppositions of local case forms of one and the same lemma. These and miniparadigms more generally will be analyzed in section 6 below. The general local cases translative and essive also show lexical variation. Between 2;1 and 2;9, there are many translatives but only two essives: 2;1 karhu-ksi ‘bear-TRANSL’, 2;5 isi-ksi ‘father-TRANSL’, kilti-ksi ‘good-naturedTRANSL’, 2;6 iso-ksi ‘big-TRANSL’, onne-ksi ‘luck-TRANSL’, 2;8 joululahjaksi ‘Christmas present-TRANSL’, suoja-ksi ‘protection-TRANSL’, 2;9 jatko-ksi ‘addition-TRANSL’; 2;4 paljaa-na ‘naked-ESS’ and 2;8 joululahja-na ‘Christmas present-ESS’. The only occurrences of the marginal cases abessive (at 2;10 huole-tta ‘worries-ABESS’ (= free from worries)) and instructive (at 2;10 k¨as-i-n ‘handsINSTRUC (= by hand)) are lexicalized expressions also in Standard Finnish. In addition to local case forms of noun stems there are also local adverbs based on pronominal stems (e.g. t¨a-ss¨a ‘this-INESS’ (= here), t¨aa¨ -lt¨a ‘this-ABL’ (= from here)). Due to their transparent form and high frequency of occurrence such adverbs may serve as important models for both the form and use of local cases. Turning to the tape-recorded data of Tuomas’s, the distribution of the boy’s first case forms is much the same as that of Tuulikki’s: The grammatical cases are well represented in the data. Although there are tokens of all local cases from 1;9 on, the earliest one to emerge and the most frequently used is the illative. The role cases essive and translative only appear at the age of 2;2. As has been found with Tuulikki, there is considerable lexical variation in the local cases also in Tuomas’s speech. Thus, the first elatives are kupi-sta ‘cup-

70

Klaus Laalo

ELAT’ (= from the cup) at 1;8, konee-sta ‘machine-ELAT’ (= from the machine) at 1;9, and ikkuna-sta ‘window-ELAT’ (= from the window) at 1;10; the first inessives are is¨a-ss¨a ‘father-INESS’(= in the father) at 1;9, kaapi-ssa ‘cupboardINESS’ (= in the cupboard) at 1;10, as well as tuoli-ssa ‘chair-INESS’ (= in the chair) and reuna-ssa ‘edge-INESS’ (= on the edge) at 1;11. This is evidence that these early elatives and inessives are unlikely to be frozen forms. More evidence comes from data demonstrating that the respective lemmas do not only occur in these case forms but also at least in one other form, see e.g. table 7 in section 6 for kuppi at 1;8. With both children, local cases not only express local relations, but especially the adessive also serves other functions such as use of a vehicle or an instrument (e.g. Tuulikki at 1;11 bussi-lla ‘bus-ADESS’ (= by bus), Tuomas at 2;1 laiva-lla ‘boat-ADESS’(= by boat) and at 2;2 auto-lla ‘car-ADESS’(= by car), avaime-lla ‘with a key’) or the possessor (e.g. Tuulikki at 2;1 isi-ll¨a ‘in father’s possession’, kaveri-lla ‘in the pal’s possession’). Besides a local meaning, the allative may also convey a beneficiary role (e.g. Tuulikki at 1;11 and Tuomas at 2;1 a¨ idi-lle ‘to the mother’, Tuomas at 1;9 nalle-lle ‘to the teddy bear’). The tape-recorded data of the two subjects studied in the present paper not only allow the study of frequency distribution of case forms but also to determine the decreasing percentage of base forms of nouns (PBF) in the course of development. The PBF measure (Percentage of Base Forms) established by Voeikova and Gagarina (2002) may thus be used as a measure of children’s inflectional development. In Finnish, inflectionally unmarked base forms of nouns are identical with the nominative singular. In adult-directed (spoken and written) adult Finnish, PBF amounts to a little more than 30% of noun tokens (Hakulinen et al. 2004: 1179). In Tuulikki’s input, PBF is 57% at 1;8 and drops to a nearly 35% at 2;9; the corresponding values for Tuomas are similar, namely 56% at 1;7 and 33% at 2;2. Thus, PBF is much higher when parents speak to children in the second half of their second year often participating in naming acts than when addressing themselves to them in their third year. Tuulikki’s PBF rates stay more or less close to the input rate characteristic of her youngest age. Although there is no evidence for a steady drop of PBF values between the beginning and the end of the observational period, PBF reduces to about 41% at 2;6 and stabilizes at this level during the last two months of observation (table 3a). Although Tuomas’s early PBF rates are much higher than Tuulikki’s, he reaches roughly the same level as his sister by 1;9. The first drop of PBF below 50% occurs five months earlier with the boy than with the girl (table 3b).

Acquisition of case and plural in Finnish

71

Table 3a. Percentage of base forms of nouns (PBF) in Tuulikki’s transcripts Age 1;7 1;8 1;9 1;10 1;11 2;1 2;2 2:3 2;4 2;5 2;6 2;8 2;9 2;10

Utterances 144 304 219 76 309 561 293 206 358 173 160 359 207 315

MLU 1.554 1.599 2.324 2.763 2.401 3.075 2.232 2.932 2.366 3.329 3.369 3.304 3.343 3.800

SG NOM 60 117 83 35 59 90 85 86 51 56 37 71 47 60

SG OBL 34 69 54 24 27 42 30 35 31 30 32 47 47 60

PL 9 3 4 2 3 20 11 26 16 26 21 16 20 26

All nouns 103 189 141 61 89 152 126 147 98 112 90 134 114 146

PBF 58.25 61.90 58.86 57.38 66.29 59.21 67.46 58.50 52.0 50.0 41.11 52.98 41.23 41.1

Table 3b. Percentage of base forms of nouns (PBF) in Tuomas’s transcripts Age 1;7 1;8 1;9 1;10 1;11 2;0 2;1 2;2

Utterances 215 403 186 167 124 120 135 435

MLU 1.535 1.355 1.548 1.754 1.774 2.325 2.541 3.278

SG NOM 201 133 149 70 32 54 26 96

SG OBL 2 42 100 32 16 17 24 53

PL – 1 13 1 11 27 5 43

All nouns 203 176 262 103 59 98 55 192

PBF 99 76 57 68 54 55 47 50

Summarizing the quantitative analysis of the singular case forms occurring in the tape-recorded data, the grammatical cases – and foremost the inflectionally unmarked nominative singular – are by far the most important ones as far as their frequency of occurrence is concerned. Tuulikki’s transcripts give evidence of a more steady use of the other grammatical cases besides the nominative singular than the boy’s. The illative is the only local case to be found in every recording of both children. As far as the other local cases are concerned, those occurring are often represented by a single token only. The role cases essive and translative emerge rather late with both children, namely at 2;1 and 2;2, and there are no more than a few tokens of each of them. The marginal cases instructive and abessive are limited to Tuulikki’s speech sample gathered at 2;10 and are only represented by a single token each.

72

Klaus Laalo

In order to demonstrate the development of the inflectional system of the noun more stringently, paradigm formation of noun lemmas documented by the tape-recorded data will be analyzed in section 6 below.

4. Creative use of singular case forms by the two children In Finnish child language, there are deviations from the norms of the standard language both in the morphological construction of certain case forms and their use. As will be demonstrated in this section, there are also some interesting deviations from the typological features of Finnish inflectional morphology to be noted in early child speech. One of the main tasks in the acquisition of Finnish is mastery of the complex interplay between stem forms and suffixes in the inflectional system due to the great number of inflectional classes and the rich allomorphy of suffixes. It is not surprising that in the early stages of the acquisition of Finnish inflection, children will use different strategies for simplifying this complex task. Thus, Tuomas first uses a suffix-based strategy concentrating on the endings of inflected forms while ignoring qualitative and quantitative stem alternations even when these alternations are quite salient. In contrast, Tuulikki first concentrates on stem alternations (either qualitative vowel alternations or grade alternation of consonants; see sections 3.2.3 and 3.2.4 above) marking unspecified oblique forms which lack specific case suffixes so that the hearer must refer to the context in order to disambiguate these vague forms. In this section, formal as well as functional deviations of cases from Standard Finnish occurring in early child speech will be studied. It will be seen that the early child system of case forms may be underdifferentiated as compared to that of the adult language.

4.1. Analogical partitives While the functions of the partitive are mastered quite well from the very beginning, its forms are not. There are different analogical tendencies to be found in children’s construction of partitives. One such characteristic is double case marking, e.g. piha ‘yard’ : correct partitive piha-a : double partitive *piha-a-ta, and vesi ‘water’ : correct partitive vet-t¨a ‘some water’ : double partitive *vett¨a-¨a. It must be noted, however, that doubly marked partitives also occur in adult speech. Some double partitives occur in negated forms of certain phrases which already include a partitive for other reasons and where the negation is emphasized by using a second, contrasting partitive, e.g. ei mon-ta-a ‘not many’

Acquisition of case and plural in Finnish

73

(cf. moni ‘many’, partitive mon-ta), ei kah-ta-a kerta-a ‘not twice’ (cf. kaksi ‘two’, partitive kah-ta; in this expression, the possible syntagmatic influence of the immediately following partitive kerta-a ‘time, occasion, turn’ must also be taken into consideration). Among the analogical child partitives there are also some transparent ones based on the final stem vowel /e/ of vocalic stems rather than the more opaque final consonant of consonantal stems, e.g. Tuulikki’s lume-e ‘some snow’ (Standard Finnish lun-ta, cf. NOM lumi ‘snow’, oblique stem lume-) at 1;7 and tois-ta [toitta] puole-e ‘the other side’ (Standard Finnish tois-ta puol-ta, cf. NOM puoli ‘side’, oblique stem puole-) at 1;9. This type of transparent partitives becomes rather frequent from 2;0 onwards.

4.2.

Forms and functions of the illative

In the early stages of language acquisition, Finnish-speaking children tend to replace the longer allomorphs -hVn, -seen, -siin of the illative suffix by its short allomorph -Vn, which also has the highest input frequency among these allomorphs. Another simplification found with children acquiring Finnish concerns shortening of the stem of certain (contracted) nouns both in the partitive and the illative (R¨ais¨anen 1975: 256–257, Niemi and Niemi 1985: 159–160, Laalo 1998: 374–375). An example from Tuulikki’s speech between 1;8–1;10 is the illative laste.huone-e(n) < lasten.huonee-seen ‘children’s room-ILL’ (= into the children’s room). Another tendency found in child language is the use of the illative suffix -Vn instead of the allative suffix -lle (see section 3.1 above). With Tuulikki, this tendency lasted for several months and was most pronounced between 1;8 and 1;10 (see (21)). (21)

a.

b.

1;8 nalle hypp¨a-¨a patja-an teddy jump-3S mattress-ILL ‘the teddy bear is jumping in the mattress’ (for allative patja-lle ‘on the mattress’) 1;9 p¨oyt¨a.niina-a(n) puto table.cloth-ILL *fall-PAST (truncated from putosi) ‘fell in the tablecloth’ (for allative p¨oyt¨a.liina-lle ‘on the tablecloth’)

74

Klaus Laalo

There are similar examples from Tuomas between 1;8 and 1;10: (22)

a.

b.

c.

1;8 puukka-a (< puolukka-a) lettu-un lingonberry-PARTIT pancake-ILL ‘put some lingonberry into the pancake’ (for allative letu-lle ‘on the pancake’) 1;9 toise-en rasva-a [vahvaa] another-ILL butter-PARTIT ‘[put some] butter to the other [slice of bread]’ (for allative toise-lle ‘on the other’) 1;10 puto-s lattia-a(n) fall-PAST floor-ILL ‘fell in the floor’ (for allative lattia-lle ‘on the floor’)

While the allative is often replaced by the illative in early child language, the opposite happens only exceptionally. Thus, Tuomas replaces the illative by the allative only once at 1;9. Possible explanations for children’s preference of the illative rather than the allative are both functional and formal. While the illative is a communicatively important purely local case (cf. 3.2.2), the allative has other functions besides the local one, the most important of them being the beneficiary. Illative forms with a suffix consisting of the lengthening of the last stem vowel + n fit the trochaic pattern favoured in early child Finnish (cf. section 3.2.3), whereas the allative violates this pattern by adding another syllable to the word. One more reason why the illative is preferred to the allative in early child Finnish is that from early on children often encounter expressions such as jalkaan ‘foot-ILL’(cf. jalka ‘foot:NOM’) or k¨ate-en ‘hand-ILL’(cf. k¨asi ‘hand:NOM’) during their daily dressing routines when something is to be put on the foot, leg or hand (rather than ‘into’ it). These forms occur in expressions such as puetaan housut ∼ keng¨at jalka-an ‘let’s put the trousers ∼ shoes on the leg/foot’ or puetaan lapaset k¨ate-en ‘let’s put the mittens on the hand’. As a result, both jalkaan and k¨ateen are also used early by the children themselves. Since the illative is used in the function of external rather than internal local TO-cases in such contexts, it is this phraseological use that may influence the children’s own use of the illative. In a cross-linguistic perspective, a comparable tendency to favour inner local cases at the expense of outer ones has been found in Hungarian child language. In the data studied by Pl´eh, Vinkler, and K´alm´an (1997) two thirds of all local case markers belong to inner local cases. Furthermore, according to Dasinger (1997: 40–42), the illative is one of the most salient local cases.

Acquisition of case and plural in Finnish

75

The development of the allative differs from that of the illative and the other local cases because the allative suffix is the only local case suffix which does not occur in adverbs, but is replaced by the suffix –nne instead. Consequently, in contrast to other local case forms, the children do not become trained by allative adverbs to use the allative suffix.

5. The plural 5.1. The Finnish plural In the nominative plural and the homophonous accusative plural form, plurality is marked by the suffix t, immediately attached to the singular stem, e.g. silm¨a ‘eye’ : silm¨a-t ‘eye-PL:NOM’, pallo ‘ball’ : pallo-t ‘ball-PL:NOM’. In the oblique cases, the case suffix follows the plural suffix, which is -i- or -j- (in intervocalic position), e.g. silm-i-¨a ‘eye-PL-PARTIT’, pallo-j-a ‘ball-PL-PARTIT’. In the plural, nominal stems are subject to certain morphophonological alternations. While rounded stem-final vowels (o, u, y or o¨ ) remain unchanged before the plural suffix i/j (e.g. pallo-j-a ‘ball-PL-PARTIT’, pallo-i-ssa ‘ballPL-INESS’) unrounded ones (e, i, a, a¨ ) either change (see below) or are deleted. The stem-final vowels e and a¨ are always deleted but stem-final a only deletes when the first stem vowel is rounded (see (23)), otherwise it changes (see below). (23)

e-stems: nimi ‘name’ : nime-t ‘name-PL:NOM’ : nim-i-¨a ‘name-PL-PARTIT’ a¨ -stems: silm¨a ‘eye’ : silm¨a-t ‘eye-PL:NOM’ : silm-i-¨a ‘eye-PL-PARTIT’ a-stems with a rounded first vowel: muna ‘egg’ : muna-t ‘egg-PL:NOM’ : mun-i-a ‘egg-PL-PARTIT’ (vs. jalka ‘leg’ : jalko-j-a ‘leg-PL-PARTIT’)

Stem-final vowel changes occurring in oblique plural forms are the following: Stem-final a dissimilates to o after an unrounded first stem vowel (e.g. jalka ‘leg’: jalko-j-a ‘leg-PL-PARTIT’). Stem-final i becomes e before the plural suffix i/j (e.g. greippi ‘grapefruit’ : greippe-j-¨a ‘grapefruit-PL-PARTIT’ : greippe-ihin ‘grapefruit-PL-ILL’). Stem-final long vowels and diphthongs are shortened (e.g. maa ‘land’ : ma-i-ta ‘land-PL-PARTIT’, vapaa ‘free’ : vapa-i-ta ‘free-PLPARTIT’, suo ‘swamp’ : so-i-ta ‘swamp-PL-PARTIT’). As will be shown below, the complex morphophonology of plural forms leads to analogical formations in child Finnish.

76

Klaus Laalo

While plural stem formation of certain noun types is complex, plural case formation is relatively simple, because most case suffixes are identical in the singular and the plural. As noted in section 2.1, the nominative singular can be considered as the basic form of most Finnish nominals representing the lemma and occurring most frequently in Finnish speech. Plural case forms, including the nominative, are not as frequent as the corresponding singular ones. The nominative plural is quite frequent with referents forming pairs, however, e.g. k¨asi ‘hand’: k¨ade-t ‘the hands’, jalka ‘foot’ : jala-t ‘the feet’, lasi ‘glass’ : silm¨a.lasi-t ‘the spectacles’ (literally: ‘eye.glasses’). While the nominative plural expresses definiteness, the plural partitive, which is accordingly also used very frequently, expresses open quantification, i.e. an indefinite amount.

5.2. The first plural case forms The first case forms to emerge in the plural are the same as those found in the singular, namely nominative and partitive. As in the adult language, the nominative plural typically refers to pairs of things while the partitive plural expresses open quantification. Only a few children use the nominative before the partitive plural (Toivainen 1980: 153). In colloquial Finnish, the short variant of partitive plural actually has no case marking and is thus identical to the plural stem in many inflectional classes (e.g. sana ‘word’, standard plural partitive sano-j-a ‘word-PL-PARTIT = some words’, coll. plural partitive sano-i ‘word-PL’). This may facilitate the children’s early use of the colloquial variant of the plural partitive. Once this form of the partitive has been mastered, the formation of the other oblique cases in the plural is easier on this basis: case suffixes are added to this caseless form carrying the plural allomorph typical of oblique cases. The first occurrences of local case forms in the plural are usually fossilized just as their singular counterparts (cf. 3.2.4). Thus, Tuulikki’s first plural illative at 1;6 is t¨o-i-hin (a lexicalized plural form, work-PL-ILL ‘to work’) used when her parents leave home for work, and her first adverb-like plural paradigm of inner local cases is t¨oihin ‘away’ : t¨oiss¨a ‘absent’ : t¨oist¨a ‘back home’. Actually, t¨o-i-ss¨a (work-PL-INESS) means ‘at work’ and t¨o-i-st¨a (work-PL-ELAT) ‘from work’, but the child’s perspective is different: when the parents are at work, they are away. The meanings ‘away’ and ‘absent’ are attested in different situations, but this 3-member paradigm is adverb-like (cf. pois ‘away’ and poissa ‘absent’), not actually a proper plural paradigm yet. According to Tuulikki’s diary data, the first plural (partitive and nominative) forms actually denoting plural referents (and having a singular counterpart in

Acquisition of case and plural in Finnish

77

the corpus) are used at the age of 1;7: the first occurrence of the plural partitive is the colloquial form kyn-i-i (< kyn-i-¨a ‘pen-PL-PARTIT’) noted at 1;7:27, which contrasts with the singular kyn¨a (1;7:10), and the first plural nominative is tossut ‘bootees, slippers’. At the age of 1;8, the plural partitive is used in rotelearned colloquial forms such as kymm-i-i (< kylm-i-¨a, ‘cold-PL-PARTIT’, coll. kylm-i-i, cf. NOM SG kylm¨a ‘cold’) and also in shortened forms which could be regarded as a kind of proto-plurals, e.g. the truncated form rusoi (< rusinoi-ta ‘raisin-PL-PARTIT’, cf. NOM SG rusina ‘raisin’). It must be remembered that, at this age, Tuulikki is in a rather strong trochaic stage and mainly uses bisyllabic word forms which are often truncated (cf. section 3.2.3). While some truncations, colloquial or analogical forms persist from 1;9 to 1;10, also many standard forms occur in the girl’s speech. As pointed out in section 5.1 above, in the formation of oblique plural case forms quite complex morphophonological processes apply to the stem and these make acquisition of such case forms rather difficult resulting in a number of analogical plural forms. After the trochaic stage, ending at 1;11, Tuulikki uses the following types of plural partitives: a) standard forms: siemen-i-¨a ‘seed-PL-PARTIT’, l¨aa¨ kke-i-t¨a ‘medicine-PLPARTIT’, possu-j-a ‘piggie-PL-PARTIT’, kakku-j-a ‘cake-PL-PARTIT’; b) coll. forms (-jA > -i): vaippo-i ‘napkin-PL:PARTIT’ (Standard Finnish vaippo-j-a), muru-i ‘crumb-PL:PARTIT’ (Standard Finnish muru-j-a); c) an analogical type ending in -ia instead of -oja: *rah-i-a ‘coin-PL-PARTIT’ (cf. Standard Finnish raho-j-a), *nauh-i-a ‘ribbon-PL-PARTIT’ (cf. Standard Finnish nauho-j-a); d) idiosyncratic analogical formations: *pilvei (Standard Finnish pilv-i-¨a ‘cloud-PL-PARTIT’), *m¨arkei (Standard Finnish m¨ark-i-¨a ‘wet-PLPARTIT’), *k¨ase-j-¨a (Standard Finnish k¨as-i-¨a ‘hand-PL-PARTIT’). The analogical formations of type (c) are constructed on the basis of the more common pattern of plural partitives of A-stems, e.g. silm¨a ‘eye’ : silm-i-¨a ‘eyePL-PARTIT’, muna ‘egg’ : mun-i-a ‘egg-PL-PARTIT’. The idiosyncratic analogical formations (d) are based on different stem types, for example on the productive noun type of i-stems; the colloquial plural partitive of i-stems ends in -ei, and their standard plural partitive in -ejA. The first plural-like forms occurring in Tuomas’s speech at 1;7 are the rotelearned nominatives potta(a) (= portaat ‘stairs’), tattaa (= rattaat ‘push-chair’), and (t)aappa(a) (= saappaat ‘boots’). Semantically, their plurality is uncertain, however; at least rattaat ‘push-chair’ clearly refers to a single entity. One month later, the boy uses several nominatives carrying the plural suffix -t with a clear

78

Klaus Laalo

plural meaning, e.g. autot ‘the cars’, ovet ‘the doors’, [s]ukat ‘the socks’, tossut ‘the slippers’. In the same month, the short colloquial plural partitives emerge: kyn-i-i (< kyn-i-¨a) ‘pen-PL-PARTIT’, kenk-i-i (< kenk-i-¨a) ‘shoe-PL-PARTIT’. As far as the local cases are concerned, they emerge in the plural as soon as they have reached productivity in the singular with both children. In Tuulikki’s diary data, there are several dozens of singular inessives, elatives, adessives and allatives at the age of 1;10 while the first corresponding plural forms occur at the age of 1;11. Illative plurals emerge earlier and ablative plurals much later than plural forms of the other local cases: illatives are first found in disyllabic types (e.g. k¨as-i-in ‘hand-PL-ILL’) at the age of 1;10 and in trisyllabic ones one month later (e.g. k¨arry-i-hin ‘carriage-PL-ILL) but ablative plurals only occur at the age of 2;3 (e.g. mu-i-lta ‘other-PL-ABL’, ‘from the others’). As for Tuomas’s development of plural forms of local cases: the first diary note is 1;9 ratta-i-sta pois ‘push-chair-PL-ELAT away’ (= away from the push-chair), the first clear plurals are from the age of 1;10: silm-i-st¨a vet-t¨a pois ‘eye-PL-ELAT water-PARTIT away’(= water, i.e. tears, away from the eyes), 1;10 silm-i-in paista-a valo-t ‘eyePL-ILL shine-3S light-PL’ (= the lights are shining into the eyes), 1;10 silm-i-ss¨a vet-t¨a ‘eye-PL-INESS water-PARTIT’ (= there is water, i.e. tears, in the eyes). To conclude, nominatives and partitives are the first plural forms to emerge. While plural nominatives are formed by simply adding the suffix -t to the singular stem, plural partitives have special stem-formation rules and serve as a basis on which other oblique plural forms can be built. The next section will be devoted to a quantitative analysis of the plural forms occurring in the tape-recorded data of both children.

5.3.

Plural case forms in the recordings

In the recorded material, the first plural forms occur at 1;7 in Tuulikki’s data and at 1;8 in those of Tuomas. The types and tokens of plural case forms found in the children’s tape-recorded data during the observational period are presented in tables (4a) and (4b). Although pronouns have not been included in the countings it should be mentioned that plural forms of pronouns are used especially for expressing possession (e.g. Tuulikki 2;2 me-i-ll¨a ‘we-PL-ADESS’, 2;4 n¨a-i-ll¨a ‘these-PL-ADESS’, 2;5 ni-i-ll¨a ‘those-PL-ADESS’, 2;8 me-i-lt¨a ‘we-PL-ABL’, 2;9 ni-i-lle ‘those-PL-ALL’). The predominance of nominatives and partitives found in the singular also holds in the plural. Local cases are relatively well represented, especially the external local cases in their different functions such as beneficiary, e.g. Tuulikki 2;1 kalo-i-lle (fish-PL-ALL) ‘to the fishes’, or instrumental, e.g. Tuulikki 2;6

Acquisition of case and plural in Finnish

79

Table 4a. Tuulikki’s plural case forms (types/tokens) Age 1;7 1;8 1;9 1;10 1;11 2;1 2;2 2;3 2;4 2;5 2;6 (2;7) 2;8 2;9 2;10

NOM – 1/1 2/2 1/1 1/1 6/7 6/9 10/13 6/6 10/12 9/14 – 7/8 9/11 7/9

PARTIT 2/2 – 2/2 – 2/2 6/11 1/1 10/12 4/5 9/12 5/6 2/2 6/8 1/1 3/3

GEN – – – – – 1/1 – – – – 1/1 – – 2/4 4/7

ILL 1/1 – – – – – – – 1/1 – – – – – 1/1

ALL – – – – – 1/1 – 1/1 – – – 1/1 – 1/1 1/2

ADESS – – – – – 1/1 – 1/1 2/3 – 1/1 – – 1/1 1/1

Other cases – INSTRUC 1/2 – – – – – – ELAT 1/1 INESS 2/2 – – – – ELAT 1/1

Table 4b. Tuomas’s plural case forms (types/tokens) Age 1;7 1;8 1;9 1;10 1;11 2;0 2;1 2;2

NOM – 1/1 7/11 – – 5/22 1/1 11/15

PARTIT – – 2/2 1/1 6/11 3/3 – 14/24

ILL – – – – – 1/2 – 1/1

INESS – – – – – – 1/4 –

ALL – – – – – – – 1/1

ADESS – – – – – – – 2/2

moottori.py¨or-i-ll¨a (motor.cycle-PL-ADESS) ‘with the motor-cycles’. The internal local cases are not as frequent as the external ones, nor are they in plural as frequent as in singular – especially there is a notable difference between illative singular and illative plural. The grammatical cases occurring in the plural are mostly nominatives and partitives, because genitive forms are not as common in the plural as in the singular (there are no plural genitives to be found in the recordings of Tuomas), and the plural accusatives are homophonous with the nominative. The distribution of the case forms in plural differs from the distribution in singular in the following respects: there are gaps in certain months even in the most frequent grammatical cases; internal local cases are represented only

80

Klaus Laalo

Table 5a. Plural case forms in Tuulikki’s father’s child-directed speech (types/tokens) Age 2;2 2;3 2;4 2;5 2;6 2;8 2;9 2;10

NOM 6/12 9/10 18/31 6/9 5/7 4/4 7/8 20/24

PARTIT 6/6 5/13 14/19 1/1 10/11 1/2 1/1 6/7

GEN 1/1 1/1 1/1 – 1/1 – – 4/4

ALL – – – – – 1/1 2/2 2/2

ADESS – 2/3 1/1 – 1/1 – – 2/2

Other cases ELAT 1/1 – INESS 1/1, TRANSL 1/1 TRANSL 1/1 ILL 1/1, ELAT 3/4 – – ELAT 1/1

Table 5b. Plural case forms in Tuomas’s father’s child-directed speech (types/tokens) Age 1;7 1;8 1;9 1;10 1:11 2;1 2;2

NOM 7/15 11/21 7/14 1/2 3/3 4/4 12/23

PARTIT 3/6 16/27 3/6 – 11/19 1/1 14/17

GEN – 4/8 – 1/2 – – 1/1

ILL – 2/2 – – 2/2 2/2 –

INESS – – – – – 2/2 –

ELAT 1/1 – – 1/1 – –

ALL – 3/6 – 1/1 – – 2/2

occasionally; external local cases are used quite regularly, but even they have mostly only one token in a certain recording, and ablative plurals are totally absent; there is only a single marginal case occurring in the plural, Tuulikki’s instructive sormin as early as at the age of 1;8. This form is lexicalized even in the adult language. In general, singular forms are used far more frequently than plural forms. The situation is much the same in parental speech: plural forms are not frequent (tables 5a and 5b).

6. Case and number contrasts and miniparadigms of nouns When different grammatical forms of a number of lemmas enter the child’s lexicon, these forms will begin to systematize: On the one hand, different grammatical forms of one and the same lemma enter into schematic relations forming inflectional (mini)paradigms and on the other, similar endings of different noun stems will relate, a process which may ultimately result in the segmentation and classification of suffixes. The use of different grammatical forms of a given lemma will lead to the construction of miniparadigms. A miniparadigm can

Acquisition of case and plural in Finnish

81

Table 6. Development of oppositions and miniparadigms in Tuulikki’s tape-recorded data Age 1;7 1;8 1;9 1;10 1;11 2;1 2;2 2;3 2;4 2;5 2;6

Case oppositions 1 9 8 3 2 6 7 8 5 11 6

Miniparadigms – 2 2 – – 3 1 1 1 2 2

Miniparadigms cumulatively – 2 4 4 4 7 8 9 10 12 14

be defined as a set of minimally three accurate and distinct inflectional forms of the same lemma which are produced spontaneously in contrasting contexts (Kilani-Schoch 2003: 279). This section will be mainly based on the recorded data of the children, starting with Tuulikki’s. Already in the first recordings of Tuulikki’s speech at age 1;7, there are some signs of emerging oppositions of grammatical forms of given nouns (table 1): the lemma kirja ‘book’ is used both in the partitive and nominative. The oppositions with two contrasting case forms and the miniparadigms (with at least three members) occurring in the transcripts of Tuulikki at 1;8 and 1;9 are presented in tables 7 and 8. If the nominative form does not occur in the recording, it is placed in brackets; the nominative is followed by a plus sign (+) if at least three different case forms of the lexeme are used in the same recording. The genitive-like accusatives are included in the genitive column. As can be seen in table 7, at 1;8 only two out of eleven nouns occurring in more than one case form are used with three cases, while the other nine lemmas occur in one marked case form besides the unmarked nominative singular. The frequency of use of given lexemes depends to a large extent on the topic of conversation. In Tuulikki’s recordings at 1;7 and 1;8, the nouns kirja ‘book’ and kuppi ‘cup’ are very much used because Tuulikki likes to show her picture books and talk about them and also to play with her toy teaset and serve coffee and tea to her parents and toy animals. Accordingly, each of these nouns occurs in three of the most frequent cases, kirja in the nominative, partitive and accusative, and kuppi in the nominative, accusative and illative. Although there are only two more true miniparadigms found in the taperecorded data at 1;9, table 8 demonstrates that simple contrasts of two grammatical forms of a given noun include two further categories, the inessive singular

82

Klaus Laalo

Table 7. Oppositions and miniparadigms, Tuulikki 1;8 Noun appe(lsiini) ‘orange’ kansi ‘cover’ kirja+ ‘book’ kuppi+ ‘cup’ linna ‘castle’ mansikka ‘strawberry’ nalle ‘teddy’ puuro ‘porridge’ p¨oyt¨a ‘table’ suu ‘mouth’ tutti ‘dummy’

NOM 1 1 1 9 2 1 7 1 1 1 1

PARTIT 1

GEN 1 1 5

5

ILL

6 1

1 1 6 1 1 1

Table 8. Oppositions and miniparadigms, Tuulikki 1;9 Noun/Adj ammu+ ‘moo-cow’ avain ‘key’ (hylly) ‘shelf’ isi ‘daddy’ kana ‘hen’ kello ‘clock’ (k¨asi) ‘hand’ pieni ‘little’ tipu+ ‘birdie’ valo ‘light’

NOM PARTIT 6 1 2 1 5 4 4

GEN 2

ILL

1 2

1

PL

1 1 6

1 6 3

INESS

1 PARTIT 1 PARTIT 1

1 1

and the partitive plural. As is to be expected, the topic of conversation again influences the frequency and consequently the diversity of forms with which lemmas are used. Thus, the frequent and many-sided use of ammu ‘moo-cow’ and tipu ‘birdie’ is due to Tuulikki’s visit to a farm, where she had seen real cows, which made a strong impression on her, and to Easter time, when she had the occupation of playing with toy birds being part of the Easter decoration at home. Even though the number of contrasts between forms of nouns grows slowly and case-number contrasts become more diverse between 1;9 and 2;1, there are no miniparadigms to be found in the recorded data between 1;9 and 1;11. Contrasts occurring at 1;10 in Tuulikki’s data are juusto ‘cheese’(nominative and

Acquisition of case and plural in Finnish

83

Table 9. Oppositions and miniparadigms, Tuulikki 2;1 Noun/Adj NOM PARTIT ILL ADESS (aasi) ‘donkey’ 1 auto+ ‘car’ 4 karhu+ ‘bear’ 11 (Kontula) 3 lelu+ ‘toy’ 1 1 m¨okki ‘cottage’ 1 1 peikko ‘troll’ 1 a¨ iti ‘mother’ 3 1

Other cases PL ALL 1 INESS 1 PARTIT 1 TRANSL 1 NOM 1 INESS 7 PARTIT 6, NOM 2 ELAT 1

partitive), p¨oyt¨a ‘table’ and a¨ iti ‘mother’ (nominative and genitive) and at 1;11, bussi ‘bus’ (adessive and ablative) and isi ‘daddy’ (nominative and genitive). It is only from 2;1 onward that the number of miniparadigms grows considerably (see table 6 above). As compared to 1;9 (table 8), the number of miniparadigms comprising plural forms has grown at 2;1 (table 9). At this point, also new case forms such as the inessive auto-ssa ‘in the car’, the translative karhu-ksi ‘to a bear’ and the plural nominatives karhu-t ‘bears’ and lelu-t ‘toys’ appear as members of miniparadigms. Summarizing the development of miniparadigms in Tuulikki’s tape-recorded data, most of her early miniparadigms include nominative, partitive and genitive singular forms. Besides, local cases appear as members of miniparadigms in every month from 1;8 on. The translative is found in a miniparadigm only once at 2;1 (karhu-ksi ‘(to become) a bear’). A plural form first appears in a miniparadigm at the age of 1;9 (tipu ‘birdie’ : tipu-a ‘birdie-SG-PARTIT’ : tipu-j-a ‘birdie-PL-PARTIT’). While plural forms in miniparadigms are mostly limited to nominatives and partitives, a plural adessive occurs at 2;3: hepo-i-lla ‘horse-PL-ADESS’ (= with the horses). The development of case contrasts and mini-paradigms is slow and steady, there is no developmental spurt. The tokens of case oppositions and miniparadigms found in Tuomas’s taperecorded data are summarized in table 10. Although the tape-recorded data do not give evidence of any case contrasts at 1;7 since they contain almost exclusively nominative singular forms and the only oblique case form, the illative koti-in ‘home-ILL’ (= home), is not accompanied by the nominative form koti ‘the home’, the diary material shows that, in this early month, Tuomas also uses several partitives (e.g. juusto-a [tuuttoo] ‘cheese-PARTIT’, kuumaa ‘hot-PARTIT’, leip¨a-¨a [peep¨aa¨ ] ‘bread-PARTIT’) and some illatives with a nominative counterpart (e.g. jalka-an [kaakaa] ‘foot-ILL’, kuppi-in ‘cup-ILL’). After the first miniparadigm documented at 1;8 in the tape-recorded data, there are three more of them one month later and an almost constant increase at

84

Klaus Laalo

Table 10. Development of oppositions and miniparadigms in Tuomas’s tape-recorded data Age 1;7 1;8 1;9 1;10 2;0 2;1 2;2

Case oppositions – 8 3 9 8 3 4

Miniparadigms – 1 4 1 3 3 4

Miniparadigms cumulatively – 1 5 6 9 12 16

Table 11. Oppositions and miniparadigms, Tuomas 2;1 Noun laiva ‘ship’ paita+ ‘shirt’ saari+ ‘island’ sisko ‘sister’ tuoli+ ‘chair’ vesi ‘water’

NOM 3 3 2 2 1 1

GEN

ALL

ADESS 3

ILL

INESS

ELAT

2

1

1

5

1 1

1 1 2

1 1

the age of 2;1 and 2;2 (table 10). One of the miniparadigms found at 2;1 even comprises 6 members (table 11). When pretending to travel to some island by boat, stay there for a while and then travel back, Tuomas uses the word saari ‘island’ in the nominative and in 5 local cases. While all of the boy’s case forms are in the singular until 2;1, one month later, the plural is used in the nominative and partitive with 4 different nouns and one of them, el¨ain ‘animal’, also occurs in two external local cases when Tuomas speaks about his toy animals. According to the theoretical framework of usage-based theory of language acquisition, grammatical systems are constructed from grammatical forms of specific lexical items and even abstract grammatical schemas are considered to remain tied to their lexical bases. It therefore seems important to show which types of nouns the two children studied in this chapter use to express different grammatical and semantic relations in their early stages of the acquisition of Finnish. Nouns belonging to certain semantic classes are frequently found in the tape-recordings of both Tuulikki and Tuomas. One of these are names for family members, such as a¨ iti ‘mother’and is¨a (and the diminutives isi and isk¨a) ‘father’; the little brother Tuomas also uses sisko ‘sister’. Another frequently used lexeme

Acquisition of case and plural in Finnish

85

Table 12. Oppositions and miniparadigms, Tuomas 2;2 Noun el¨ain+ ‘animal’ housut ‘trousers’ kaveri+ ‘pal’ sisko+ ‘sister’ tunneli ‘tunnel’ Tuomas+ vauva ‘baby’ a¨ iti ‘mother’

NOM GEN Other cases

1 17 2 14 1 5

5

ADESS 1, ESS 1, TRANSL 4 PARTIT 1

PL NOM 2, ALL 1, ADESS 1 NOM 1, PARTIT 2 PARTIT 3

NOM 1 3 1 1

ADESS 1

group are names of (toy) animals, e.g. ammu ‘moo-cow’, heppa ‘horse’, karhu ‘bear’ and tipu ‘birdie’ used by Tuulikki as well as ankka ‘duck’ and the more abstract el¨ain ‘animal’ used by Tuomas. Furthermore, both children use nalle ‘teddy’in the very same case forms, namely the nominative, genitive and allative (Tuomas at 1;9 and Tuulikki at 2;2). Other lexemes occurring in miniparadigms are typically connected with the respective playing situations during recordings (e.g. Tuulikki’s kirja ‘book’, kuppi ‘cup’, lelu ‘toy’, violetti ‘violet’, kaakao ‘cocoa’, kylm¨a ‘cold’, peite ‘blanket’ and ruoka ‘food’ and Tuomas’s kaveri ‘pal’, kuppi ‘cup’, paita ‘shirt’, tie ‘way’, tuoli ‘chair’ and ympyr¨a ‘circle’). Auto ‘car’ is used in different forms by both children when playing with toy cars. Only a small number of lexemes are used by the children in more than two or three inflectional forms. As mentioned above, there is one 6-member miniparadigm of the noun saari ‘island’ to be found in Tuomas’s tape-recorded data at 2;1. Besides, a 5-member miniparadigm occurs at 2;2, namely kaveri ‘pal’ used in the nominative and adessive singular, the plural partitive and in the singular forms of the role cases essive (kaveri-na ‘as a pal’) and translative (kaveri-ksi ‘to (become) a pal’). A remaining question is whether miniparadigms consisting of at least three members tend to develop from contrasts of two grammatical forms. Although this kind of development might be expected, it is only occasionally attested by the tape-recorded data. Thus, in Tuulikki’s speech, the noun nalle ‘teddy’ occurs in two grammatical forms at 1;8 and in a miniparadigm at 2;2. There is one similar example in Tuomas’s data, namely sisko ‘sister’ occurring in an opposition at 2;1 and in a miniparadigm one month later.

86

Klaus Laalo

7. Conclusion Summarizing the main findings of this paper on the acquisition of Finnish nominal inflection, in the premorphological period, Finnish-speaking children use nouns in just one form each. In a usage-based approach to language acquisition, it is important to keep in mind that these are case forms used in different contexts. For most nouns, the first form occurring in child speech is the inflectionally unmarked nominative singular, but some nouns referring to uncountables are first used in the partitive and certain other nouns in the illative. Whether the forms used by the child in this early period carry a suffix or not, all of them are rote-learned as unanalyzable chunks. Nevertheless, the child is getting at least tentatively familiarized with some of the suffix elements of the language and with the use of these inflected forms. The genitive and accusative forms are the first clearly inflectional forms contrasting systematically with the forms used earlier by the children. These two cases have identical suffixes (-n) in the singular in most instances (cf. 2.2), but their functions differ: the genitive expresses possession, the accusative marks the object. When the child starts to use genitive and accusative forms the first case oppositions emerge and the active processing of morphological elements of nouns begins (for verbs, see Laalo 2003). There are two reasons why the genitive and accusative forms are used relatively early; one is their pragmatic importance and consequently their high frequency in the input and the other is their formal simplicity, since the suffix -n is a short and superstable marker. In Finnish child language, local case endings first occur with adverbs and only gradually develop into the many-sided uses of local cases in their different functions. The ablative is restricted to local functions, it occurs less frequently in the input and is therefore acquired as the last member of the external local cases. The use of plural case forms begins with the partitive and nominative. The local cases are used in the plural soon after they have been established in singular. Clear evidence that morphological processing is under way and that development has proceeded from the premorphological well into the protomorphological period is the child’s creative use of morphological means and the production of analogical forms (see sections 4 and 5.2). The construction of true miniparadigms, i.e. paradigms consisting of at least three members, is another important sign of inflectional development. The most important members of early oppositions and miniparadigms are first of all the grammatical cases (nominative, partitive, genitive and accusative) and also the three local cases illative, adessive and allative (see section 6). Of these three local cases, the illative emerges very early. Both the adessive and the allative

Acquisition of case and plural in Finnish

87

have other important functions besides purely local ones. The other local cases, the inessive, elative and ablative, occur rather infrequently and are consequently only occasionally found in miniparadigms. As far as plural case forms are concerned, their occurrence in miniparadigms is mostly limited to nominatives and partitives, but there are also a few adessives and allatives to be found. To what degree miniparadigms occur in the respective recordings seems to largely depend on the topic of conversation: lexemes related to the topic tend to occur more frequently and in a number of different forms. Nevertheless, the child’s familiarity with different inflectional forms of frequently occurring lexemes can also be assumed to play an important role, since the child will feel more secure with such forms and use them in her own speech production more freely. Lexemes such as these include names of family members, toys and (toy) animals.

Notes ∗

I wish to thank the editors for their valuable suggestions concerning earlier versions of this paper, and both the editors and Virginia Mattila for brushing up my English. All remaining shortcomings are my own responsibility.

References Dasinger, Lisa 1997

Issues in the Acquisition of Estonian, Finnish, and Hungarian: A Crosslinguistic Comparison. In The Crosslinguistic Study of Language Acquisition, Dan Isaac Slobin (ed.), vol 4, 1–86. Mahwah, NJ: Lawrence Erlbaum.

Gerken, Louann 1994 A metrical template account of children’s weak syllable omissions from multisyllabic words. Journal of Child Language 21 (3): 565– 584. Hakulinen, Auli, Maria Vilkuna, Riitta Korhonen, Vesa Koivisto, Tarja Riitta Heinonen, and Irja Alho 2004 Iso suomen kielioppi [The large grammar of Finnish]. Helsinki: Suomalaisen Kirjallisuuden Seura. Kilani-Schoch, Marianne 2003 Early verb inflection in French:An investigation of two corpora. In Development of Verb Inflection in First Language Acquisition: A CrossLinguistic Perspective, Dagmar Bittner, Wolfgang U. Dressler, and

88

Klaus Laalo

Marianne Kilani-Schoch (eds.), 269–295. (Studies on Language Acquisition 21.) Berlin: Mouton de Gruyter. Kilani-Schoch, Marianne, and Wolfgang U. Dressler 2002 The emergence of inflectional paradigms in two French corpora: An illustration of general problems of pre- and protomorphology. In Preand Protomorphology: Early Phases of Morphological Development in Nouns and Verbs, Maria D. Voeykova and Wolfgang U. Dressler (eds.), 45–59. (Lincom Studies in Theoretical Linguistics 29.) Munich: Lincom Europa. Laalo, Klaus 1994 Kaksitavuvaihe lapsen kielen kehityksess¨a [The disyllabic stage in language acquisition]. Viritt¨aj¨a 98 (3): 430–448. 1998 V¨alikatsaus suomen kielen varhaismorfologiaan. Viritt¨aj¨a 102 (3): 361–385. [Summary: Interim review of protomorphology in Finnish child language, http://www.kotikielenseura.fi/virittaja/hakemistot/ jutut/vir98laalo.html] 2001 The tendency to trochaic word-forms in Finnish child language. Congressus Nonus Internationalis Fenno-Ugristarum 7.–13.8.2000 Tartu. Pars V: Dissertationes sectionum: Linguistica II, 209–214. 2003 Early verb inflection in Finnish: A preliminary approach to miniparadigms. In Development of Verb Inflection in First Language Acquisition: A Cross-Linguistic Perspective, Dagmar Bittner, Wolfgang U. Dressler, and Marianne Kilani-Schoch (eds.), 323–350. (Studies on Language Acquisition 21.) Berlin: Mouton de Gruyter. Niemi, Jussi, and Sinikka Niemi 1987 Acquisition of inflectional marking: A case study of Finnish. Nordic Journal of Linguistics 10 (1): 59–89. Pl´eh, Csaba, Zsuzsanna Vinkler, and L´aszl´o K´alm´an 1997 Early morphology of spatial expressions in Hungarian children: A CHILDES study. Acta Linguistica Hungarica 44: 249–260. Budapest: Akad´emiai Kiad´o. P¨ochtrager, Markus, Csan´ad Bod´o, Wolfgang U. Dressler, and Teresa Schweiger 1998 On some inflectional properties of the agglutinating type illustrated from Finnish, Hungarian and Turkish inflection. Wiener Linguistische Gazette 63: 57–92. R¨ais¨anen, Alpo 1975 Havaintoja lapsenkielest¨a [Observations from child language]. Viritt¨aj¨a 79 (3): 251–266. Savinainen-Makkonen, Tuula 2000 Learning long words – a typological perspective. Language and Speech 43 (2): 205–225.

Acquisition of case and plural in Finnish

89

Toivainen, Jorma 1980 Inflectional Affixes Used by Finnish-Speaking Children Aged 1–3 Years. Helsinki: Suomalaisen Kirjallisuuden Seura. 1997 The acquisition of Finnish. In The Crosslinguistic Study of Language Acquisition, Dan Isaac Slobin (ed.), vol. 4, 87–182. Mahwah, NJ: Lawrence Erlbaum. Voeikova, Maria, and Natalia Gagarina 2002 MLU, first lexicon and the acquisition of case forms by two Russian children. In Pre- and Protomorphology: Early Phases of Morphological Development in Nouns and Verbs, Maria D. Voeykova and Wolfgang U. Dressler (eds.), 115–131. (Lincom Studies in Theoretical Linguistics 29.) Munich: Lincom Europa. Wijnen, Frank, Evelien Krikhaar, and Els Den Os 1994 The (non)realization of unstressed elements in children’s utterances: Evidence for a rhythmic constraint. Journal of Child Language 21(1): 59–84.

The acquisition of numeral classifiers and optional plural marking in Yucatec Maya∗ Barbara Pfeiler Introduction Most studies about the acquisition of number to date have dealt with IndoEuropean languages, in which number is usually represented by the distinction between singular and plural, expressed by unmarked and marked inflectional forms respectively. Yucatec Maya,1 an indigenous language of Mexico, belongs to the class of transnumeral languages, “in which the basic pattern is ‘to disregard number’ (Lucy 1992: 55f.)” (Stephany 2002: 10). According to Lucy and Gaskins 2001: 260–261) “. . . all nouns in Yucatec are semantically unspecified as to quantificational unit” so that they require supplementary marking in the context of numeral modification, such as numeral classification.2 In classifier languages all nouns occur as complements to classifiers in counting contexts, which gives the impression that all nouns are mass nouns and are morphosyntactically neither singular nor plural. The classifiers have referential function and individualize nominal concepts (Senft 2000: 26). In Yucatec Maya the numeral ‘one’ in conjunction with the classifier [NUM+CLFR]3 is the basic way to indicate singular (Lucy 1996: 53). Though languages with numeral classification typically lack the category of nominal number (see Greenberg 1972), Yucatec Maya has a plural suffix that does not appear in a noun phrase containing a classifier phrase which refers to more than one item. The current study examines the acquisition of the numeral ‘one’ in conjunction with the classifier [NUM+CLFR] and noun plural inYucatec Maya-speaking children in Yucatan, Mexico. We are interested in the usage and function of the numeral classifiers for animate and inanimate entities. TheYucatec Maya pluralization involves one overt plural suffix and is limited to animate entities (Lucy 1996: 55; 1992) and to inanimate entities possessed by animate beings. There is no obligatory plural marking, but a preference according to the animacy hierarchy. This study provides an insight into which types of nouns are pluralized and the function of classifiers in early child Yucatec.

92

Barbara Pfeiler

1. Description of numeral classifiers and noun plural in Yucatec Maya The most common classifiers of modern colloquial Yucatec Maya are -p’´eel and -t´uul designating inanimate and animate referents respectively. Although this indicates that the classification of Yucatec Maya nouns is mainly based on animacy, in everyday usage the inanimate classifier -p’´eel sometimes replaces other classifiers, even the animate -t´uul, since many classifiers for form, arrangement and position have fallen into disuse and are now marked by the inanimate classifier (Schumann 1990: 54; Brice˜no 1993: 8–9). Noun phrase constructions containing a numeral classifier are composed of a numeral prefix, the classifier and a noun (example 1). (1)

hun-p’´eel(h) xanab one-INANI shoe ‘one shoe’

Depending on the context, the classifier can take the deictic form (example 2) or the anaphoric one (example 3). (2)

He’ hun-p’´eel-a’. ASS one-INANI-PROX ‘Here is this one.’

(3)

Ts’ah ten hun-p’´eeh(l)4. give:IMP PRO1 one-INANI ‘Give me one.’

The numeral is used in a predicate function that counts an anaphoric referent (Lehmann, in press) (example 4). (4)

t´uul-ul. Cinco5 u five 3POSS ANI-REL ‘There are five of them.’ (lit.: five (are) of them)

The fact that the classifiers are obligatory with numerals in Yucatec Maya, but not with demonstratives, indicates that the demonstratives also have an individuating force in that language. Furthermore, Lehmann (in press) highlights the structural function of the numeral classifiers, which “serve as props for the numerals, which are affixes” and concludes that the function of the numeral classifiers in Yucatec Maya is not to classify the nouns they combine with, but, after having undergone several grammaticalization processes, they have become nominalizers in the broadest sense.

The acquisition of numeral classifiers and optional plural marking inYucatec Maya

93

In comparison to the obligatory marking of the classifier phrase, the plural reference is optionally marked by suffixing -(o’)ob to the root or stem (example 5). (5)

m´aak ‘person’ person:TRANSNUM m´aak-o’ob ‘persons’ person-PL6

The suffix -(o’)ob7 marks plural not only in nouns, but also in adjectives8 (example 6) and verbs9 (example 7). (6)

le chichan-o’ob-o’ DEF small-PL-DIST ‘the small ones’

(7)

K-u haantik-o’ob. IPFV-3ERG eat-3ABS:PL ‘They are eating it.’

The plural form -(o’)ob is morphologically productive, which means that plural forms of new words can be formed (Dressler 2004), such as Spanish loan words. Sometimes the plural form of the Spanish noun has been imported into Yucatec Maya as a neutral form, e.g. flores-o’ob ‘flower-s’.

2. Acquisition studies on number in Yucatec Maya In previous investigations into the acquisition of number in Yucatec Maya (Pfeiler 1998; Blaha Pfeiler and Carrillo 2001) it was found that the numeral classifier -p’´eel is used with the meaning of ‘one’,10 and the suffix -o’ob marks multiplicity in nouns and verbs from age 1;11 on (see table 1). Furthermore, inflectional marking of plural generally occurs with animate nouns or with inanimate nouns used in possessive constructions with an animate possessor. In table 1 (based on Blaha Pfeiler and Carrillo 2001) the emergence of quantificational devices in the girl Sandi (SAN) is summarized. Table 1 shows that the first quantificational devices are found around age 2, namely the animate and inanimate numeral classifiers and the lexical quantifier u l´aak’ N ‘another N (more of them)’. The plural suffix occurs with nouns, demonstratives, and adjectives. From age 2;2 on the child produces examples of agreement marking between nouns and adjectives as well as between nouns and

94

Barbara Pfeiler

Table 1. Emergence of quantificational devices in SAN (1;11–3;11) Age

1;11

2;0

Numeral Classifiers NUM-p’´eel [-animate] (N) NUM-t´uul [+animate] (N) p’´eel-a’ ‘(one)-INANI-here’ p’´eel-o’ ‘(one)-INANI-there’ p’´eel-e’ ‘(one)-INANI-there’ hun-p’´eel N ‘a/one INANI:N’

Plural marker -o’ob

Nouns Verbs

2;1

2;2

2;3 2;5

2;6 2;7 2;11 3;2 3;4 3;11

t´uul-a’ ‘(one)-ANI here’ t´uul-o’ ‘(one)-ANI there’

Agreement Noun/Verb Adjective/Noun

u l´aak’ N 3POSS other N ‘another (equal) N’ ya’ab ‘much/many’

u l´aak’ ‘another (equal)’ ya’ab N ‘many N’ u l´aak’ hun-p’´eel-a’ ‘3POSS another one-INANI-here’ u heel ‘other (different)’

ka’-p’´eel-e’ ‘two-INANI-there’

hun-p’´eel N ‘one-INANI N’ hun-t´uul N ‘one-ANI N’ ka’-t´uul ‘2-ANI’

Lexical quantifiers

Agreement CLFR/Verb tul´aakal N ‘all N’ u l´aak’ u p’´eel u N another 3POSS.INANI 3POSS N ‘another (one) of N’

verbs. The lexical modifier u heel ‘others (more of those)’appears at age 2;6. Two months later the child produces contrasting expressions with the meanings of u l´aak’ and u heel ‘another/other’ of the same and a different kind. More complex quantificational modifiers, such as u l´aak’ hun-p’´eel-a’ ‘another one like this’, are found from 2;5 on. The quantifier tul´aakal ‘all’ is registered at age 3;4.

The acquisition of numeral classifiers and optional plural marking inYucatec Maya

95

We found interesting differences with respect to grammatical plural marking between the child’s speech and the input in our previous studies onYucatec Maya. While the plural tends to be marked on verbs in the input, from age 2 on the child shows a tendency to mark it on nouns in spite of the higher frequency of verbs in her speech. In contrast to adult usage, the girl not only marks animate nouns but also inanimate ones for plural. The question arises whether the child uses the plural marker because she is as yet unaware of the semantic difference between animate and inanimate nouns. While in our previous studies the focus was on the frequency of use of the plural suffix on nouns and verbs in the speech of one child, the present paper includes the analyses of the data of two children and their input with a focus on the types of nouns that are pluralized or used with classifiers. Therefore particular attention will be paid to the criteria that structure the classifying system in Yucatec Maya, such as the animacy hierarchy (see Lucy 1996; Lucy and Gaskins 2001).

3. The data The data of the present paper come from longitudinal studies of the first-born children (the girl Sandi and the boy Armando) of two monolingual families living in Yalcob´a, in the Eastern part of Yucatan. The child samples selected for the analysis are presented in table 2. We analyzed children’s entire adult input for this study. It is worth mentioning that across both longitudinal studies the children are mainly raised by their siblings in the Yucatec Maya community and adults do not usually engage in conversations with small children (see Pfeiler 2007). Therefore the input data analyzed is much smaller than the child data. The following analyses have been carried out: Emergence, development, and contexts of use of the classifiers hun-t´uul (‘one-ANI’) and hun-p’´eel (‘one-INANI’) in ARM’s and SAN’s samples as well as in their input; Type and frequency of pluralized nouns in the data of ARM and SAN at the same ages (from 2;0 to 3;0) and in the input. For the analysis of the animacy hierarchy a larger sample of SAN’s speech covering the age range from 1;11 to 4;9 and her input were selected. Table 2. The samples of analyzed material Child ARM SAN

Age 1;2–3;0 1;11–4;9

Number and duration of recordings 70 (49 hours) 88 (69 hours)

96

Barbara Pfeiler

4. Emergence and development of animate and inanimate numeral classifiers and types of nouns they refer to Numeral classifiers referring to a quantificational unit develop in both children first as a bare inanimate classifier (example 8a) or as a classifier followed by a deictic, such as p’´eel-a’ (CLFR:INANI-PROX) (example 8b) or p’´eel-o’ (CLFR:INANI-DIST). (8)

SAN (2;1.12) a. %sit: children are eating cherries SAN: He’ela’ p’´eel Mech. PROX INANI Mech ‘Here is one, Mech.’ MOT: Ha hantke’? ‘Will you eat it?’ b. SAN: P’´eel-a’. INANI-PROX ‘This one.’

It is only around age four that SAN (example 9) uses the proximal deictic form -a’ with the respective ostensive evidential h´e’(l) of the construction (h´e’(l). . . -a’) according to the speaker’s perception.11 (9)

SAN (3;11.11) He’ p’´eel-a’ chan p’´eel-a’. for u chan p’´eel-a’. he’ hun12 -p’´eel-a’ ASS one-INANI-PROX 3POSS small INANI-PROX ‘The one (thing) here, his/this small one.’

Since most of the children’s utterances with numeral classifiers do not contain the noun the classifier refers to, it is impossible to tell from the structure of the utterance whether the animacy criterion has been applied correctly. The only thing that can be done in this situation is to analyze the contexts in which the utterances are used and take the input into consideration. In this section we will analyze the samples from SAN and ARM at the ages from 2;0 to 3;0, classify the types of nominal referents and compare them with the input. Table 3 shows the number of tokens of the classifiers hun-t´uul and hun-p’´eel in contexts where the unexpressed noun either refers to animate or inanimate entities in the two

The acquisition of numeral classifiers and optional plural marking inYucatec Maya

97

children’s speech, the classifier phrases with overtly expressed nouns, the errors, and finally the number of contexts where the unexpressed nouns are impossible to identify (hun-p’´eel ?N). Table 4 shows the number of tokens of the classifiers hun-t´uul and hun-p’´eel in contexts where the unexpressed noun either refers to animate or inanimate entities in the input and the classifier phrases with overtly expressed nouns. Comparing the children’s speech to the input (table 4) it is found that the inanimate numeral classifier p’´eel is used much more often by the children (105 tokens) than by the adults (24 tokens). Although the classifier for animate referents -t´uul emerges quite early as well, it is less frequently used than its inanimate counterpart -p’´eel by the children as well as the adults in child-directed speech. The difference between the two classifiers takes a long time to be acquired: Errors persist with ARM through 2;7 and with SAN through 3;0. Besides classifier omission the main error consists in the overgeneralization of the INANI classifier -p’´eel to ANI nouns while -t´uul is correctly limited to ANI nouns (with two exceptions). It is, however, worth mentioning that in adult Yucatec Maya the construction combining inanimate classifiers with animate referents is possible in the sense of ‘whichever entity’. The inanimate classifier is even once used in reference to an animate entity (a cat) in the input at 2;0. As far as the children are concerned, the Spanish loan word cami´on ‘truck’ shows the trouble they have in choosing the correct classifier for this noun (table 3, age 2;7): While SAN simply omits the classifier, ARM uses the animate instead of the inanimate one. The fact that some animate nouns (‘dog’, ‘chicken’) are used with both classifiers shows that the children’s system is variable as far as the category of animacy is concerned. Surprisingly, at age 3;0 SAN uses totally ungrammatical forms choosing the inanimate classifier with humans like ‘child, children’ and the animate classifier with the inanimate noun ‘thing’. The input data suggest that the children have chosen the most frequent classifier occurring in the input (hun-p’´eel in the construction of hun-p’´eel and hay p’´eel ‘how much/many INANI’) as an all-purpose classifier and only slowly learn to distinguish between animate and inanimate classifiers. The use of the question forms hay p’´eel (5 out of 24) and hay t´uul (3 out of 10) in the input imply the context of counting and fosters the production of classifiers in the children’s speech. However, the rare use of complete nominal phrases with overtly expressed nouns [NUM+CLFR+N] in the input (2/2) contributes to the children’s difficulty of mastering the animacy hierarchy. It must also be noted that about 20% of the nouns (N = 134) to which classifiers refer are not identifiable so that animacy cannot be determined.

98

Barbara Pfeiler

Table 3. Frequency of animate and inanimate classifiers and noun lemmas they refer to (types/tokens), SAN and ARM Age 2;0

2;1 2;2 2;3

2;4 2;5

2;6

2;7

2;8 2;9

2;10

2;11

Numeral classifiers (ANI, INANI), number of noun lemmas SAN ARM ANI: 0/0 INANI: 3/3 ANI: 0/0 INANI: 0/0 INANI for ANI: p’´eel (chicken) ANI: 0/0 INANI: 0/0 ANI: 0/0 INANI: 1/2 ANI: 1/6 INANI: 2/11 ANI: 0/0 INANI: 0/0 ANI: 0/0 INANI: 2/5 ANI: 0/0 INANI: 3/4 INANI for ANI: cinco-p’´eel (cloth) p’´eel-e’ (chicken) Unidentified: 3 ANI: 1/1 INANI: 1/1 ANI: 0/0 INANI: 0/0 Unidentified: 3 ANI: 0/0 INANI: 2/2 ANI: 0/0 INANI: 5/9 INANI for ANI: [hun-p’´eel u k’´aan]NP one-INANI 3POSS hammock p’´eel-e’ (dog) ‘one of his/the hammocks’ hun-p’´eel-a’ (chicken) Unidentified: 1 Unidentified: 2 ANI: 0/0 INANI: 4/7 ANI: 0/0 INANI: 8/11 INANI for ANI: hun-p’´eel-o’ (bird) [u p’´eel ba’al-il]NP 3POSS INANI thing-REL ‘one of the things’ Unidentified: 2 ANI: 0/0 INANI: 1/3 ANI: 2/2 INANI: 2/2 CLFR omitted: ANI for INANI: hun-kamion-i’ [hun-t´uul u heel kamion]NP one-ANI 3POSS other truck one-truck-TOP ‘one other truck’ Unidentified: 1 ANI: 1/1 INANI: 5/5 ANI: 0/0 INANI: 0/0 ANI: 0/0 INANI: 4/8 ANI: 0/0 INANI: 1/1 Unidentified: 1 [hun-p’´eel-∅ ### in b´aaxal]NP one-INANI-3ABS 1POSS toy ‘there is one ### my toy’ ANI: 0/0 INANI: 3/4 ANI: 0/0 INANI: 6/10 [hun-p’´eel chan naylon]NP [hun-p’´eel chan t´uunich]NP one-INANI small stone one-INANI small nylon ANI: 3/3 INANI: 2/7 ANI: 1/1 INANI: 1/1 [hun-t´uul kaan-o’]NP [hun-t´uul chan peek’]NP one-ANI snake-DIST one-ANI small dog Unidentified: 3

The acquisition of numeral classifiers and optional plural marking inYucatec Maya

99

Table 3. Frequency of animate and inanimate classifiers and noun lemmas they refer to (types/tokens), SAN and ARM [continued] Age

Numeral classifiers (ANI, INANI), number of noun lemmas SAN ARM 3;0 ANI: 1/1 INANI: 2/3 ANI: 0/0 INANI: 2/6 INANI for ANI: Unidentified: 6 hun-p’´eel (child) [hun-p’´eel paalal-o’ob]NP one-INANI child-PL ANI for INANI: [hun-t´uul chan ba’al]NP one-ANI small thing Unidentified: 1 Total ANI: 7/12 INANI: 37/72 ANI: 3/3 INANI: 23/33 Noun phrases: 5/5 Noun phrases: 4/4 INANI for ANI: 4/4 INANI for ANI: 3/3 ANI for INANI: 1/1 ANI for INANI: 1/1 CLFR omitted: 1 Unidentified: 11 Unidentified: 12

Table 4. Input frequency of animate and inanimate classifiers and noun lemmas they refer to (types/tokens) Age 2;0

2;1 2;2 2;3 2;4 2;5 2;6 2;7 2;8 2;9 2;10 2;11 3;0

Numeral classifiers (ANI, INANI), number of noun lemmas INANI: 1/1 ANI:1/1 INANI for ANI: hay p’´eel (cat) how many-INANI [hun-t´uul chan xi’ipal]NP one-ANI little boy ANI: 0/0 INANI: 1/1 ANI: 2/2 INANI: 1/1 ANI: 0/0 INANI: 2/5 Unidentified: 1 ANI: 1/1 INANI: 2/3 ANI: 0/0 INANI: 0/0 ANI: 0/0 INANI: 1/1 ANI: 0/0 INANI: 2/3 ANI: 1/1 INANI: 1/1 ANI: 1/1 INANI: 2/4 ANI: 1/1 INANI: 1/3 INANI: 1/1 ANI: 2/3 [hun-t´uul kaax]NP one-ANI chicken ANI: 0/0 INANI: 0/0

100

Barbara Pfeiler

Table 4. Input frequency of animate and inanimate classifiers and noun lemmas they refer to (types/tokens) [continued] Age Total

Numeral classifiers (ANI, INANI), number of noun lemmas ANI: 9/12 INANI: 15/22 Noun phrases: 2/2 INANI for ANI: 1/1 Unidentified: 1

5. Type and frequency of pluralized nouns in the children’s data and the input Given that in Yucatec Maya grammatical number marking is optional and restricted to certain series of nouns, the nouns marked for plural by the suffix -o’ob in the children’s speech samples from 2;0 to 3;0 will be classified according to the categories of animacy and possession. SAN’s pluralized nouns will be compared to her input13 (table 5). Besides the number of pluralized nouns, the main difference between the two children’s speech is SAN’s overuse of the plural marker in inanimate nonpossessive constructions in pretend play, a phenomenon not found with ARM.As is to be expected, the ungrammatical pluralization of inanimate non-possessive nouns does not occur in child-directed speech. Otherwise there is a close lexical correspondence between pluralized animate nouns and inanimate possessive ones in SAN’s speech and her input. The inventory of nouns used by SAN and her mother is typical of Yucatec Maya mother-child interaction, where nouns referring to persons and animals are the most frequent ones. In contrast to this, Kaluli mother-child interaction especially focuses on proper nouns, kinship terms, and other relationship terms because of the cultural importance placed on learning such terms (Schieffelin 1985: 534).

5.1. The children’s use of the plural marker and its overgeneralization Half of the tokens of inanimate pluralized nouns not occurring in possessive constructions in SAN’s data are formed from the general word for things ba’al ‘thing, object’ (also used for referring to toys). In example (10) the child first answers the interviewer’s question by using the plural form of this general term. When the interviewer insists on more specific information on what it is that the child is picking up she interestingly uses the non-pluralized form of ba’al since number is unimportant in this context. She does not, however, comply with the

The acquisition of numeral classifiers and optional plural marking inYucatec Maya

101

Table 5. Animacy and possession of forms marked by -o’ob ‘PL’ in child speech (2;0– 3;0) and SAN’s input Noun classes [+animate]

Referents animals

ARM k’´eek’en ‘pig’ wakax ‘cow’

humans

m´aak ‘person’ paalal ‘infant’ ch’´uup ‘girl’

[–animate] [+possessive]

objects

ba’al ‘thing’ nook’ ‘cloth’ nah ‘house’ b´aaxal ‘toy’ l´apiz ‘pencil’

[–animate] [–possessive]

plants, fruits, objects (restricted use during play)

SAN s´ıinik ‘ant’ kaax ‘chicken’ miis ‘cat’ peek’ ‘dog’ ts´ıimin ‘horse’ uulum ´ ‘turkey’ k’an-s´ıinik ‘ant’ m´aak ‘person’ paal ‘infant’ ch’´uupal ‘girl’ chaampal ‘child’ ba’al ‘thing’ nook’ ‘cloth’ nah ‘house’14 k’ab ‘arm’ koh ‘tooth’ ook ‘leg’ o’och ‘food’ xanab ‘shoe’ ich ‘face, eye’ k’´aan ‘hammock’ biber´on ‘bottle’ he’ ‘egg’ b´aaxal ‘toy’ abal ‘plum’ iik’ ‘chili’ flor(es) ‘flower’ k’´aax ‘wood’ t´uunich ‘stone’ cubeta(s) ‘pail’ cubo(s) ‘bucket’ plato(s) ‘plate’ pote(s) ‘pot’

Input (SAN) s´ıinik ‘ant’ kaax ‘chicken’ miis ‘cat’ peek’ ‘dog’ ts´ıimin ‘horse’ u´ ulum ‘turkey’ ch’´ıich’ ‘bird’ sapito(s) ‘frog’ m´aak ‘person’ paal ‘infant’ nene’ ‘baby’ xi’ipal ‘boy’ Ba’al ‘thing’ nook’ ‘cloth’ nah ‘house’ k’ab ‘arm’ koh ‘tooth’ ook ‘leg’ o’och ‘food’ xanab ‘shoe’ bolador ‘flying object’ che’ ‘wood’ k’aaba’ ‘name’ kib ‘candle’ flor(es) ‘flower’

102

Barbara Pfeiler

request for more specific information, probably because the interviewer is well aware of what she is doing since she can see her. (10)

SAN (3;2) INT: Tu’ux yan-ech. where EXIST-2ABS ‘Where are you?’ SAN: T-in mool ba’al-o’ob. PROG-1ERG pick up thing-PL ‘I am picking up things (toys).’ INT: Ba’ax ba’al-o’ob-i’? what thing-PL-LOC ‘Which things there?’ SAN: Ba’al. ‘Stuff.’

The last overgeneralization of the plural marker to nouns with inanimate referents in SAN’s corpus is registered at age 4;1 (example 11). (11)

SAN (4;1) Cha’-∅ in lu’us-ik-∅ let-IMP 1ERG remove-INC-3ABS le nukuch kubos-o’ob-a’. DEF big bucket-PL-PROX ‘Let me move the big buckets.’

In contrast to SAN, the boy’s plural noun forms are restricted to animate nouns (either marked for possession or not) and to possessive inanimate nouns so that he already follows the rules of standard Yucatec Maya before the age of 3;0 (see table 5). The contexts in which SAN and ARM mark nouns for plural resemble what is typically found in numeral languages, i.e. nouns are pluralized when they refer to more than one referent, e.g. when the children are chasing several chickens. Adult speakers would pluralize such nouns only in situations in which they want to highlight the multiplicity of referents. The children thus do not hear as many pluralized lemmas from the adults as they themselves use.The children’s behavior can be explained by the following observations: Adults are generally not present when children play in the yard. Rather, the infants are taken care of by other children, like siblings, cousins, or children of the neighborhood. In everyday life SAN and ARM thus receive much more input from school children learning Spanish than from adults. As shown by an analysis of Yucatec

The acquisition of numeral classifiers and optional plural marking inYucatec Maya

103

Maya textbooks (Pfeiler 2001) Yucatec plural marking is heavily influenced by Spanish in these textbooks, since the plural marker -o’ob is suffixed to all nouns in the context of multiplicity, regardless of the animacy hierarchy. The dialogue of the two children quoted in example (12) demonstrates their use of the noun b´aaxal ‘toy’ in both its pluralized and non-pluralized forms. SAN, at age 3;2, contrasts the pluralized form with the unmarked form (examples 12a and 12b) in different utterances, while ARM, at age 2;6, still produces both forms in one and the same utterance (example 12c). SAN’s use of the lexical modifier ya’ab ‘many’ with a noun unmarked for plural in example (12a) corresponds to standard speech. (12)

SAN (3;2.15) and ARM (2;6.2) a. SAN: He’elo’ Armando, there Armando ya’ab b´aaxal. much/many toy:TRANSNUM ‘Armando, there are many toys.’ b. SAN: Ay he’el u baax(a)l-o’ob-a.’ oh here 3POSS toy-PL-PROX ‘Oh, here are his toys.’ c. ARM: He’el u b´aaxal u b´aax(a)l-o’ob-a’. here 3POSS toy 3POSS toy-PL-PROX ‘Here is his toy, his toys.’

In spite of the fact that number agreement is optional inYucatec Maya and rarely occurs in the input, some examples are found in the speech of both children. At age 2;2, SAN makes the attributive adjective agree with the noun (example 13, from Pfeiler 1998: 84). (13)

SAN (2;2.15) le mehen-tak-o’ob miis-o’ob-o’ DET little-PL-PL cat-PL-DIST ‘those little cats’

Later examples demonstrate plural agreement between a quantitative object pronoun and the verb (example 14c) and double marking of the plural on the head noun and the dependent noun of a possessive construction (example 15a). In the dialogue between SAN and her mother (example 14), the mother uses a transitive verb marked by -o’ob, which functions as a 3rd person plural object inflection. It is interesting to note that in example (14c) the child, while picking

104

Barbara Pfeiler

up the verb form marked for plural object but combining it with a different aspect, overgeneralizes the plural marker to the quantifier tul´aakal functioning as an object. In the standard form the quantifier tul´aakal is not pluralized. (14)

SAN (2;8.27) a. SAN: Tech ts’one m´aako’. for Teech ts’on-∅ le m´aak-o’. you shoot-IMP DEF person-DIST ‘You, shoot that man.’ b. MOT: S´aam in l´aah PTL 1ERG all ts’on(-eh)-o’ob. shoot(-TRANS/SUBJ)-3ABS.PL ‘I’ve already shot everybody.’ c. SAN: Tul´aak(a)l-o’ob t-a all-3ABS.PL PFV-2A ts’on(-ah)-o’ob. shoot(-TRANS/COMP)-3ABS.PL ‘You shot everybody.’

In example (15a) it remains unclear whether the plural suffix occurring in the form u k’an-o’ob refers to the possessed object (‘hammock’) or the possessor (‘cat’). This ambiguity also exists in corresponding adult expressions. Although in example (15b) the verb is not marked for its plural object, SAN’s utterance is acceptable in standard Yucatec Maya. (15)

SAN (4;1.19) a. La’ bix u k’an-o’ob look how 3POSS hamock-PL le miis-o’ob-o’. DEF cat-PL-DIST ‘Look, how the cats’ hammock(s) is/are.’ b. Ma’ inw e’es-ik. no 1ERG show-INC ‘I won’t show them.’

These examples show the use of the grammatical plural in contexts where its marking is optional and it generally does not occur in adult speech. Although the frequency of pluralized nouns in the input is low, the children hear the suffix -o’ob in the context of verbs more frequently. Adults generally mark plural

The acquisition of numeral classifiers and optional plural marking inYucatec Maya

105

number on verbs, and number marking is obligatory for the 1st and second person. The 3rd person marker is zero in the singular and -o’ob in the plural. The children’s exaggerated use of plural marking of nominals as compared to the adult input and in particular the examples of overgeneralization of the plural suffix to inanimate nouns in the very specific context of play can be explained by the above-mentioned change which Yucatec Maya is undergoing, as this is evidenced by the textbooks used in the intercultural bilingual school, where plural marking does not depend any more on the animacy hierarchy and is in the process of being extended to inanimate non-possessed contexts.

5.2. The animacy hierarchy in SAN’s speech (1;11 to 4;9) and her input In order to study the role of the animacy hierarchy in child speech, which plays a fundamental role in Standard Yucatec Maya number marking, SAN’s entire speech sample from 1;11 to 4;9 has been taken into consideration. The reason is that the children’s acquisition of the animacy hierarchy from age 2;0 to 3;0 is impossible to determine since pluralized nouns occur very infrequently in their speech during this period. Only 13 out of 158 recordings include pluralized nouns so that, in this period, less than 3.5% of noun types and tokens respectively are marked for plural in each of the two children’s speech and 2% of noun types and 1.6% of tokens in SAN’s input. Table 6 summarizes the use of pluralized nouns belonging to the categories [+/– animate], [+/–possession], [+/–alienable].15 While pluralized nouns are about equally divided between the two categories of animate and inanimate possessive nouns in SAN’s input, each of the categories of animate, inanimate possessive, and inanimate non-possessive nouns amounts to roughly one third of types and tokens in the child’s speech. In contrast to the raw numbers of tokens of the two classes of animate and inanimate, possessive pluralized nouns, the numbers of types are similar in SAN’s speech and her input. The reason why the

Table 6. Frequency of plural marking on nouns in SAN’s speech and her input according to animacy and possession (percent of types/tokens) Nouns [+animate] [–animate +possession +/–alienable] [–animate -possession] Total

SAN 29.3/38.0 (12/54) 36.6/35.2 (15/50) 34.2/26.8 (14/38) 41/142

Input 44.8/50.0 (13/24) 55.2/50.0 (16/24) – 29/48

106

Barbara Pfeiler

Table 7. The animacy hierarchy in child Yucatec Maya and the input SAN [+animate] [–animate +possessed +alienable] [-animate +possessed –alienable] [–possession –animate]

INPUT [+animate], [–animate +possessed +alienable] [–animate +possessed –alienable]

tokens of pluralized animate and inanimate possessive nouns are more numerous in the child’s speech than in the input is that the children pluralize nouns more frequently than verbs. The ranking of the use of noun tokens marked for plural in SAN’s speech and her input is presented in table 7 based on Lucy (1996). While animate nouns and inanimate possessed alienable ones are pluralized from early on in the input, the girl at first only pluralizes animate nouns. Although nouns characterized by the features [–animate +possessed –alienable], like body parts, do require plural marking in Yucatec Maya, nouns belonging to the category [–animate +possession +alienable] are more frequently marked for plural in both samples. The main difference between adult and child speech is the marking of nonpossessed inanimate nouns for plural by the child but not the adults.

6. Conclusion This study provides an insight into which types of nouns are pluralized and the function of classifiers in early child Yucatec Maya. The development of the inanimate classifier hun-p’´eel ‘one’, which marks the singular, and the nominal plural marker -o’ob is similar: 1) the inanimate numeral classifier and the plural suffix emerge together in the speech of both children (at age 1;11 with SAN and at 2;1 with ARM); 2) the early classifiers carry a demonstrative function; 3) constructions containing a numeral as well as a classifier combined with a noun ([NUM+CLFR+N]) emerge at age 2;7. A surprising finding is that, in spite of their rather infrequent occurrence in adult child-directed speech, the inanimate classifier -p’´eel and the nominal plural suffix -o’ob both emerge early, together with other inflectional markers of mainly verbal morphology. Unicity is not always marked, but when it is, the construction ‘one-CLFR’ is used in counting and for indefinite reference. Although in childdirected speech the classifier -p’´eel rarely occurs without a numeral, e.g. in a demonstrative construction such as le p’´eel-a’ ‘DEF INANI-PROX’ (= this

The acquisition of numeral classifiers and optional plural marking inYucatec Maya

107

one) or a possessive one like u p’eel u N(-il) ‘3POSS INANI 3POSS N-REL’ (= his/her one of N), the children use the forms p’´eel-a’ ‘INANI-PROX’ (= one here) and p’´eel-o’‘INANI-DIST’(= one there) still at age 3, often in the context of requests implying a unique referent with a demonstrative function. The use of the inanimate classifier with animate nouns indicates language change in progress in the classifying system ofYucatec Maya. Furthermore, the inanimate classifier has taken the meaning of ‘one’.This study also shows thatYucatec Maya employs its classifiers in most of the cases in deictic and anaphoric expressions and rarely in a complete noun phrase. This process of change is documented by Lehmann (in press) who argues that Yucatec Maya requires “individuation by a classifier in counting, but not in exophora”. In order to explain the development of plural marking in Yucatec Maya children it is important to take their cultural environment into consideration. Yucatec Maya children grow up in constant interaction with members of the extended family or neighbors in a yard where animals are raised. At the same time Mayan children are primarily in contact with other children and less so with adult family members like the mother or grandmother. This cultural environment offers them the opportunity of encountering frequent reference to a multiplicity of animate beings. Taking into consideration that in every culture some classes of nominals have a special status in adult-child interaction, nouns referring to people (e.g. paal(al) ‘child’) and animals (e.g. kaax ‘chicken’, miis ‘cat’, peek’ ‘dog’) can be said to play this role in Yucatec Maya. However, a profound analysis of the cultural background could explain why, as far as plural marking is concerned, Yucatec Maya children seem to behave as if they were acquiring a numeral language in which plural marking is obligatory rather than a transnumeral one in which it is optional. Our analysis clearly shows that both children attach importance to the use of the numeral classifier hunp’´eel referring to a single unit and to marking multiplicity of animate referents or inanimate possessed ones. In contrast to this, adults prefer to mark multiplicity on the verb rather than on the noun in their child-directed speech so that plural marking on nouns is rare in the children’s input. Moreover, the children even tend to use the plural suffix in agreement, where it is optional, and one of them overgeneralizes it to inanimate unpossessed nouns. As mentioned above, Mayan children grow up in extended families and interaction between children exceeds that between adults and children. It may therefore be assumed that a specific analysis of the peer group’s speech during play with SAN could provide an explanation for her frequent use of pluralized nouns used in non-possessive constructions referring to inanimate items. The question arises whether in the Yucatec Maya community infants learn the language spoken by adults or rather the variety spoken by older children. In contrast to the generations of their parents

108

Barbara Pfeiler

and grandparents, attending school is nowadays compulsory for children. There is a choice between monolingual Spanish schools and intercultural bilingual ones where both Yucatec Maya and Spanish are taught during the first three years of instruction. In both modalities the mother tongue of the Mayan children is likely to be influenced either by Spanish grammar or by certainYucatec Maya bilingual textbooks, where plural marking is obligatory and has been grammaticized to the point of also being used for agreement. From the results of our study we can conclude that a tendency towards language change is clearly reflected in the child data with respect to the role of the animacy hierarchy structuring the classifying system of Yucatec Maya. This is shown by the children’s use of the inanimate classifier on the one hand and by their plural marking of nouns irrespective of the category of animacy (and possession) on the other.

Notes ∗

1. 2. 3. 4. 5.

6. 7.

8. 9.

10. 11.

I would like to thank Ursula Stephany for her insightful comments. I am also grateful to Christian Lehmann for his comments on a previous version and to Carlos Carrillo Carre´on for discussing the child data. Thanks to Clifton Pye for checking my English. Yucatec Maya is the most homogeneous language in the Maya language family as far as dialectal variation is concerned. Nouns in classifier languages can be characterized as nouns with generic reference (Royen 1929: 775). In this construction the numeral can also function as an indefinite determiner. According to Lehmann (in press) phrases like hun-p’´eel ‘one (inanimate object)’, hun-t´uul ‘one (animate being)’ are typically used in anaphora. In modern Yucatec Maya numerals higher than ‘four’ are expressed in Spanish. In general Spanish numerals do not combine with a classifier. However, in child speech (ARM, at age 2;3) the following example is found: cinco-p’´eeh ‘five’ referring to clothes. The word for ‘person’ is not normally specified for number. If a noun ends in /’/, only –ob is suffixed to it in forming the plural, e.g. k’u’ ‘nest’, k’u’-ob ‘nests’. A few nouns form their plurals with –al instead of –(o)’ob, e.g. paal ‘infant, child’, paal-al ‘boys’ (Bricker, Po’ot Yah, and Dzul de Po’ot 1998: 358). Several adjectives, such as color terms take the plural suffix –tak. In verbs number and person are marked by cross-reference pronouns. The third person plural takes a plural suffix, which is identical to the nominal plural and is optional, just as with nouns. The inanimate classifier has the meaning ‘one’ also in modern Tzeltal (Polian 2006). According to Hanks (1990) the term h´e’e(l=). . . a’ indicates a referent accessible to touch and h´e’e(l=). . . o’ a visual one. The latter form is used anaphorically.

The acquisition of numeral classifiers and optional plural marking inYucatec Maya

109

12. The classifiers -p’´eel (INANI) and -p’´ıit (a little) in everyday language can occur without the numeral ‘one’. 13. No plural marking is found in ARM’s input. 14. In everyday speech the terms for ‘house’, ‘food’, ‘shoe’, and ‘toy’ are always used with possessive markers. 15. The distinction between alienable and inalienable possession is important with inanimate nouns. Inalienable possession requires obligatory marking in contrast to alienable possession.

References Blaha Pfeiler, Barbara, and Carlos Carrillo Carre´on 2001 La adquisici´on del maya yucateco: el n´umero. In La adquisici´on de la lengua materna: espa˜nol, lenguas mayas, euskera, Cecilia Rojas Nieto and Lourdes de Le´on (eds.), 75–97. Mexico: Centro de Investigaciones y Estudios Superiores en Antropolog´ıa Social/Universidad Nacional Aut´onoma de M´exico. Brice˜no Chel, Fidencio 1993 La cuantificaci´on en maya: el uso de clasificadores numerales y de mensurativos. Master’s thesis. M´erida: Universidad Aut´onoma de Yucat´an. Bricker, Victoria, Eleuterio Po’ot Yah, and Ofelia Dzul de Po’ot 1998 A Dictionary of the Maya Language as Spoken in Hocab´a, Yucat´an. Salt Lake City: The University of Utah Press. Dressler, Wolfgang U. 2004 Degrees of grammatical productivity in inflectional morphology. Italian Journal of Linguistics 15: 31–62. Greenberg, Joseph 1972 Numeral classifiers and substantival number: Problems in the genesis of a linguistic type. Working Papers in Language Universals 9: 1–39. Stanford University. Hanks, William F. 1990 Referential Practice: Language and Lived Space among the Maya. Chicago: University of Chicago Press. Lehmann, Christian in press On the function of numeral classifiers. In Essais de linguistique g´en´erale et de typologie linguistique, Floricic Franck (ed.), Louvain: ´ Presses de l’Ecole Normale Sup´erieure. Lucy, John A. 1992 Language Diversity and Thought: A Reformulation of the Linguistic Relativity Hypothesis. Cambridge: Cambridge University Press.

110

Barbara Pfeiler 1996

Grammatical Categories and Cognition:A Case Study of the Linguistic Relativity Hypothesis. Cambridge: Cambridge University Press. Lucy, John, and Suzanne Gaskins 2001 Grammatical categories and the development of classification preferences:A comparative approach. In Language Acquisition and Conceptual Development, Melissa Bowerman and Stephen Levinson (eds.), 257–283. Cambridge: Cambridge University Press. Pfeiler, Barbara 1998 Acquisition of number in Yucatec Maya. In Studies in the Acquisition of Number and Diminutive Marking, Steven Gillis (ed.), 77–95. (Antwerp Papers in Linguistics 95.) Universiteit Anwerpen. 2001 Acerca de la adquisici´on del n´umero en el maya yucateco. Paper presented at the I Seminario sobre Adquisici´on de Lengua Ind´ıgena, San Cristobal de las Casas, Chiapas, Mexico. June 27, 2001. 2007 “Lo oye, lo repite y lo piensa.” The contribution of prompting to the socialization and language acquisition in Yukatek Maya toddlers. In Learning Indigenous Languages: Child Language Acquisition in Mesoamerica, Barbara Pfeiler (ed.), 183–202. Berlin: Mouton de Gruyter. Polian, Gilles 2006 El´ements de grammaire du Tseltal. Paris: L’Harmattan. Royen, Gerlach 1929 Die nominalen Klassifikations-Systeme in den Sprachen der Erde. M¨odling bei Wien: Anthropos. Schieffelin, Bambi B. 1985 The acquisition of Kaluli. In The Crosslinguistic Study of Language Acquisition, Dan Isaac Slobin (ed.), Vol. 1, 525–593. Hillsdale, NJ: Lawrence Erlbaum Associates. Schumann, Otto 1990 Aproximaci´on a las lenguas Mayas. Cuaderno de Trabajo 6. M´exico: Instituto Nacional de Antropolog´ıa e Historia. Senft, Gunter 2000 What do we really know about nominal classification systems? In Systems of Nominal Classification, Gunter Senft (ed.), 11–50. Cambridge: Cambridge University Press.

The early development of case and number in Estonian∗ Reili Argus The goal of this paper is to describe the acquisition of case and number in Estonian within the pre- and protomorphological approach (e.g. Dressler and Karpf 1995; Dressler et al. 2002), determining which features characterize the pre- and protomorphological phases of language acquisition. Spontaneous data from three children aged 1;7–2;8 have been used to determine which morphophonological, morphological and morphosyntactic features play a role in the acquisition of Estonian. Since not only nouns but also pronouns and deictic adverbs are marked for case and number, these nominal categories are considered as well. The data were divided into one-month periods, where type and token frequencies as well as input frequencies for each session have been counted and analyzed.

1. Description of the target language: case and number in Estonian In this section, the case system of Standard Estonian will be presented. Estonian is a Finnic language of the Finno-Ugric language family. In Estonian, nominals, i.e. nouns, adjectives, numerals and pronouns, are all inflected for number and case. Estonian has 14 nominal cases, both in the singular and plural. Three of them (1–3 in table 1) are grammatical cases. There are 11 adverbial cases, sometimes also called semantic cases, including the 6 local cases (4–9 in table 1). The other adverbial cases are sometimes called marginal cases (10–14 in table 1). The Estonian case system is presented in table 1, with case suffixes, meanings of case forms and one example each. Adjectives and numerals used as attributes agree with their head nouns in case, and adjectives and ordinal numerals agree in number as well, except when a noun is in the translative, terminative, abessive, or comitative cases. In these latter cases the attribute is in the genitive. Predicative adjectives which function as complements of the copula-like verb olema and the subject are usually in the nominative.

112

Reili Argus

Table 1. The Estonian case system in the singular 1

Case Meaning Nominative What? Who?

Case suffixes –

2

Genitive

–

Example kuusik ‘fir-forest’ kuusiku

-t, (-tt), -d, ∅, -da -sse -de, -ha, -he, -hu, ∅ -s -st -le -l -lt -ks

kuusiku-t kuusiku-sse – kuusiku-s kuusiku-st kuusiku-le kuusiku-l kuusiku-lt kuusiku-ks

-ni -na -ta -ga

kuusiku-ni kuusiku-na kuusiku-ta kuusiku-ga

3 4

Partitive Illative-1 Illative-2 5 Inessive 6 Elative 7 Allative 8 Adessive 9 Ablative 10 Translative

Whose? (possession) What? (as an object) What? Who? (as an object) Where to? (into)

Where? (in) From where? (out of) Where to? (onto the top of) Where? (on top of) From where? (from the top of) (become) somebody, (get) into the state of being sth. or sb., what for 11 Terminative till, until, up to 12 Essive as (to be sth. or sb.) 13 Abessive without 14 Comitative with

1.1.

Morphophonemics and case formation

The Estonian language has developed historically from an agglutinating to a more fusional language type (Erelt 2003: 7) and, as is common in inflectingfusional languages, there is often no clear boundary between the stem and a grammatical formative (Erelt et al. 1995: 129). Only 10% of nouns have an unchangeable stem in all their inflectional forms (Erelt et al. 1995: 156). Therefore, phonological changes of the stem are very important in the inflection of Estonian nouns. Phonological changes in stems are principally of two kinds: gradation changes (affecting the root and medial sounds) and other changes (omission, addition and ordering changes of final phonemes). In addition, there are some suppletive stems. When the stem is subject to gradation it will occur in strong or weak forms in different grades; in the case of a change concerning the final phoneme there will be different final phonemes in different case forms (Erelt et al. 1995: 130). Gradation in Estonian includes on the one hand alternation in quantity, in which a phonetically stronger stem shape, the so-called strong grade (3rd degree of length or Q3), alternates with a phonetically weaker one, the weak grade

The early development of case and number in Estonian

113

(2nd degree of length or Q2), for example vanni ‘bath:GEN’ Q2 vs. ‘vanni1 ‘bath:ILLAT’ Q3. On the other hand, there is alternation in quality, which is mostly reflected in the change, assimilation or loss of a single onset obstruent of the second syllable in the weak grade, while the stem can be in 1st, 2nd or 3rd degrees of quantity and the weak grade stem is a stem without the obstruent in question (Viitso 2003: 26). A nominal paradigm may contain case forms consisting of the bare stem or suffixed stems. The genitive is a bare vocalic stem and all adverbial cases (except the short illative) are formed by adding a case suffix to the genitive stem, i.e. by an agglutinating technique, so that ten out of eleven adverbial cases of Estonian are formed by adding a case suffix to the genitive form. For the short illative the so-called illative stem is used with the choice of the allomorph depending on the morphophonemic character of the stem (see table 1). The illative and partitive can be formed in two ways: either by stem alternation or by adding a suffix. In short, stem change plays the most important role in the formation of the grammatical cases and the illative 1; suffixation is used to form the other cases. As pointed out by Viitso (2003: 38), “a noun can have from one to six stem allomorphs.” The interrelations between stem allomorphs in a paradigm (see table 2) also depend on whether the stem is subject to weakening or strengthening gradation (Viitso 2003: 38). “Case endings in the genitive plural, partitive singular and plural, illative-2 and allative singular depend largely on the stem or word type” (Viitso 2003: 38). Among the grammatical cases, the nominative and genitive singular are not marked by a suffix but are usually distinguished by stem change (see table 2). With some nouns, there is syncretism between two or three cases, usually the nominative, genitive and partitive singular (e.g. muna ‘egg:NOM/GEN/ PARTIT:SG’). The formation of the partitive singular is complex in Estonian. The partitive suffix has four variants depending on the inflectional class of the noun (-t (-tt), -d, ∅ and -da). The regular and most frequent suffix is -t (auto : auto-t ‘car-PARTIT’), the suffix -tt (occurring in 9 words) is different from -t only orthographically since there is no phonological difference between -t and -tt. -d occurs with monosyllabic vocalic stems (koi : koi-d ‘moth-PARTIT’) and some disyllabic stems (lumi : lun-d ‘snow-PARTIT’), -da is found with only a small group of words, such as pronouns ending in a short vowel (see : se-da ‘thisPARTIT’). According to Viitso (2003: 40) “the partitive singular has no overt case marker after a disyllabic vocalic stem of Q1” (ema : ema ‘mother:PARTIT’) or “after a disyllabic vocalic stem of Q3 in words with weakening gradation” (siil : ‘siili ‘hedgehog:PARTIT’) (for more details see Viitso 2003: 39–40).

114

Reili Argus

Table 2. Interrelations between stem allomorphs in noun paradigms (based on tables 19 and 20 in Viitso 2003: 38–39) Case Stem with weakening gradation Nominative ‘siil ‘hedgehog’ Genitive siili Partitive ‘siili Illative ‘siili Inessive siili-s Elative siili-st Allative siili-le Adessive siili-l Ablative siili-lt Translative siili-ks Terminative siili-ni Essive siili-na Abessive siili-ta Comitative siili-ga

Stem with strengthening gradation hammas ‘tooth’ hamba hammas-t hamba-sse hamba-s hamba-st hamba-le hamba-l hamba-lt hamba-ks hamba-ni hamba-na hamba-ta hamba-ga

The illative has a long and a short form. The long form, which can be used with all nouns, is an agglutinating formation with the ending -sse added to the genitive stem. The short illative, which is more frequent than the agglutinating one in standard Estonian (see Hasselblatt 2000: 803), has 5 allomorphs: ∅, stem, h+vowel, -de and -da. While the zero allomorph (kast : ‘kasti ‘box:ILL’) and the so-called illative stem (elu : ellu ‘life:ILL’) are part of the ‘active morphology’, the allomorphs with -hV (pea : p¨ahe ‘head:ILL’) and the shortened stem before the suffix -de (uus : uu-de ‘new-ILL’) are lexicalized; they are no longer productive and therefore belong to the so-called ‘passive morphology’ (Erelt et al. 1995: 200; see also Viitso 2003: 40–41). When both illative forms are possible, the choice between the two productive patterns depends mostly on stylistic factors, but also on some grammatical ones like verb government, the long illative being sometimes considered an archaism, characteristic of the written language. Although there are many nouns where both forms are used equally often, with the most frequent nouns the short illative is usually preferred. The meanings of the adverbial cases can also be rendered by postpositions. In this case location is expressed by a construction consisting of a noun in the genitive and a postposition (koti sisse ‘bag:GEN into’), while the local postpositions can have three inflected forms, e. g. sisse ‘into’, sees ‘in’ and seest ‘from’. Synthetic case forms and periphrastic expressions with postpositions are almost equally frequent; nevertheless, postpositions are somewhat more typical of the spoken language and may be used to emphasize physical location.

The early development of case and number in Estonian

1.2.

115

Inflectional classes of nouns

As described above, the choice between different case allomorphs depends mostly on the morphophonological features of the stem. According to these features Estonian nouns can be divided into inflectional classes or patterns, with nouns belonging to a given inflectional class inflecting in more or less the same way. Another possibility of describing Estonian noun inflection is to divide nouns according to so-called inflectional features. This will, however, lead to a huge amount of different patterns where nouns belonging to a given pattern inflect in exactly the same way. Usually, textbooks or school grammars distinguish as few as 7 inflectional classes based on the morphophonological features of the stem, while, in dictionaries, an approach based on inflectional features is chosen so that usually as many as 48 patterns can be found. Since the morphophonological structure of the word plays an important role in child speech as well as child-directed speech (see Argus 2004; Pajusalu 2001), a more strictly morphophonemically centered approach has been chosen in the present paper. In table 3 the nouns of Standard Estonian (derivations excluded) are classified as belonging to a particular inflectional class according to the phonological characteristics of their base form (nominative or genitive singular): on the first level, the division is based on the number of syllables, on the second level, on quantity or quality alternation of syllables, and on the third level on the word-final phoneme. According to these phonological features, all Estonian nouns can be divided into 12 subclasses (Erelt et al. 1995 II: 139) as shown in table 3. The most important inflectional class is the 3rd class with words pattering like siil, since many highly frequent nouns belong to this class. Almost all inflectional classes represented in table 3 have subtypes and can be divided into (a) words with productive morphology (in which the rules of inflection strictly derive from the phonological structure of the word) and (b) words with non-productive morphology (lexicalized forms, where the phonological structure by itself does not provide sufficient information for predicting the inflection of the word). About 95% of Estonian nominals have productive morphology (Erelt et al. 1995: 126), but some words with unproductive morphology are frequent and relevant in child-centered situations. Generally, inflectional subtype (a) is an open type, but in some cases subtype (b) can also be open. Thus, a word derived from a lexeme belonging to class (b) will also be inflected according to pattern (b).

116

Reili Argus

Table 3. Inflectional classes of nouns Inflectional Number class of syllables 1 1 2 1 3 1 4 2 5 2 6 2 7 2 8 2 9 2 10 2

I I II III I∼II III III

≥3 ≥3

11 12

1.3.

Gradation (where relevant)

Morphophonemics (nouns ending in)

Type

long vowel diphthong consonant short vowel -e another vowel vowel vowel consonant -el or -er a consonant other than r or l a vowel a consonant

maa / idee koi siil k˜one ema auto aasta kanal kringel album

Number of productive nouns2 90 180 5220 50 440 580 190 620 500 460

asendamatu 1400 seminar 500

Case functions and semantic roles

The most important semantic roles and grammatical functions expressed by the Estonian cases are summarized in tables 4a, 4b, and 4c. Table 4a. Main semantic roles of Estonian cases: Grammatical cases Case Main Roles and Functions Nominative agent (subject) neutral participant (predicative) state (predicative)

Genitive

total object (i.e. object of a telic, perfective, bounded predicate) possessor, (possessive attribute)

total object

Examples Koer jookse-b. ‘dog:NOM run-3S’ See on koer. ‘it:NOM be:3S:PRES dog:NOM’ Vend on loll ‘brother:NOM be:3S:PRES fool:NOM’ (The brother is a fool) Kirjuta kiri. ‘write:IMP letter:NOM’ See on venna auto. ‘it:NOM be:3S:PRES brother:GEN car:NOM’ (This is brother’s car) Poiss ehita-s silla. ‘boy:NOM build-PAST bridge:GEN’ (The boy built the bridge)

The early development of case and number in Estonian

117

Table 4a. Main semantic roles of Estonian cases: Grammatical cases [continued] Case Main Roles and Functions Partitive partial object (i.e. object of an atelic, imperfective, unbounded predicate)

Examples Joo-n ‘piima. ‘drink-1S:PRES milk:PARTIT’ (I drink milk)

Table 4b. Main semantic roles of Estonian cases: Local cases Case Illative

Main Roles and Functions goal-oriented (dynamic)

Inessive

location (stative)

Elative

source-oriented (separative, dynamic)

Allative

goal-oriented (directional)

Adessive location (stative)

Ablative

source-oriented (separative)

Examples Teg-i augu maa-sse. ‘make-PAST hole:GEN ground-ILL’ ((Somebody) made a hole into the ground) Ole-n kooli-s. ‘be-1S:PRES school-INESS’ (I am in the school) Tule-n kooli-st. ‘come-1S:PRES school-ELAT’ (I’m coming from the school) H¨uppa-s voodi-le. ‘jump-3S:PAST bed-ALL’ ((Somebody) jumped on the bed) On voodi-l. ‘be:3S:PRES bed-ADESS’ ((Somebody) is on the bed) V˜otsi-n selle laua-lt. ‘take-1S:PAST it:GEN table-ABL’ ((I) took it from the table)

Table 4c. Main semantic roles of Estonian cases: Marginal cases Case Main Roles and Functions Terminative local boundary (goal)

temporal right-hand boundary

Essive

status

Examples Jookse-b maja-ni. ‘run-3S:PRES house-TERM’ ((Somebody) runs as far as the house) Maga-b hommiku-ni. ‘sleep-3S:PRES morning-TERM’ ((Somebody) sleeps until morning) Mina ema-na luba-n seda. ‘I:NOM mother-ESS promise1S:PRES it:PARTIT’ (I as a mother promise this)

118

Reili Argus

Table 4c. Main semantic roles of Estonian cases: Marginal cases [continued] Case Abessive

Main Roles and Functions without companion

Comitative companion

instrument

1.4.

Examples Ta ei saa kaisukaru-ta maga-da. ‘he:NOM not can teddy.bear-ABESS sleep-INF’ (He can’t sleep without his teddy bear) M¨angi-b venna-ga. ‘play-3S:PRES brother-COM’ ((Somebody) is playing with his brother) Kirjuta-b pliiatsi-ga. ‘write-3S:PRES pencil-COM’ ((Somebody) is writing with a pencil)

Number in Estonian: plural case forms

Estonian distinguishes two numbers, singular and plural. While the singular is unmarked, the plural marker follows the stem and precedes the case suffix (e.g. koer-te-ga ‘dog-PL-COM’ (= with the dogs). Attributive adjectives and ordinal numerals agree with their head nouns in number, except in the terminative, essive, abessive and comitative cases, where the attribute is in the genitive case (e.g. ilusa-sse t¨udruku-sse ‘beautiful-INESS girl-INESS’, but ilusa-te t¨udrukute-ga ‘beautiful-PL:GEN girl-PL-COM’). There are two sets of plural markers: the agglutinating de-plural (kala-de-l ‘fish-PL-ADESS’) and the vowel plural, divided into the agglutinating i-plural (laulja-i-l ‘singer-PL-ADESS’) and the fusional stem plural (kalu ‘fish:PL’) in which the stem vowel is fused with plural suffix (for details see Erelt et al. 1995: 207). The agglutinating de-plural is common to all nominals (except some pronouns) and occurs in all cases except the partitive, which is always formed by the vowel plural (mune ‘egg:PL:PARTIT’) or with the suffix -sid (munasid ‘egg-PL:PARTIT’), in which the plural and case allomorphs are fused. The suffix -de has three allomorphs: -d, -de and -te (ema-d ‘mother-PL:NOM’, emade ‘mother-PL:GEN’, pois-te ‘boy-PL:GEN’). The suffix -d occurs only in the nominative; in all other cases plurality is expressed either by the oblique case plural suffix -de or -te (depending on the final phoneme of the stem) or by modification of the stem vowel (stem plural). In some patterns the vowel plural can occur in parallel with the de-plural, the choice between them depending on the stylistic value of the two forms. The plural suffix is added to the weak stem variant (the nominative stem) in the case of words with a strengthening stem (e.g. hammas-te ‘teeth-PL’), but to

The early development of case and number in Estonian

119

the strong-grade stem variant (the partitive stem) in the case of words with a weakening stem (e.g. jalga-de-ga ‘feet-PL-COM’).

2. The data The principal data for this study comes from the Andreas subcorpus of the CHILDES Estonian language database and from diary notes between the ages 1;0 and 1;7, both compiled by Maigi Vija. Those parts of the database to be used here are recordings from age 1;7.24 to 2;8, comprising about 28 hours of spontaneous speech altogether and diary notes. From the month-long period (age 2;0) during which the child’s speech was recorded every day I have selectively analyzed a third of the recordings, starting from the first recording, choosing one recording, skipping the next two, and so on. As additional material the following data from the CHILDES database was used: 3 hrs. and 40 min. of Hendrik’s recordings of spontaneous speech collected between the ages 1;8 and 2;5, plus diary notes between the ages 1;0 and 2;5 (compiled by Reili Argus), as well as 5 hrs. and 10 min. of Anna’s recordings (compiled by Kaja Kohler) from 1;10 to 2;1 (tables 5a, b, and c).

Table 5a. Data analyzed: Andreas Age Length of recordings

MLU Noun in tokens words

1;7 45 min 1.1 1;8 45 min 1.2 1;9 45 min 1.5 1;10 45 min 1.6 1;11 90 min 1.8 2;0 360 min 3.6 2;1 90 min 2.7 2;3 45 min 3.8 2;4 45 min 4.1 2;5 45 min 4.1 2;6 60 min 3.8 2;7 60 min 3.9 2;8 60 min 4.6

207 349 176 183 386 2107 975 200 341 183 487 559 435

Number Noun of imi- lextations emes

25 60 16 23 23 72 18 3 5 9 13 5 5

82 73 75 76 110 445 188 70 104 43 137 145 136

Nouns with more than one case form 4 14 10 8 13 89 57 16 35 13 40 45 41

Pronouns Local (types/ Adverbs tokens) (types/ tokens) 1/1 1/5 1/13 1/12 2/44 2/160 1/109 1/26 2/81 1/27 2/85 2/67 2/64

2/24 1/2 1/39 1/2 2/152 2/77 1/41 1/51 2/12 2/80 2/63 2/41

120

Reili Argus

Table 5b. Data analyzed: Anna Age

Length of recordings

MLU Noun in tokens words

1;10 90 min 1.2 1;11 105 min 1.4 2;0 90 min 1.5 2;1.0 25 min 1.5

69 106 101 74

Number Noun of imi- lextations emes

36 54 23 13

41 54 62 43

Nouns with more than one case form 2 4 3 4

2/2 3/5 3/20 1/6

Nouns with more than one case form 0 0 0 0 1 2 5 3 5 11

Pronouns Local (types/ Adverbs tokens) (types/ tokens) 0 0 0 0 0 0 1 0 1 4/7 2/8 3/9 3/10 4/9 4/22 2/9 3/56 1/12 7/196 4/28

Pronouns Local (types/ Adverbs tokens) (types/ tokens) 1/1 2/7 2/10 2/2

Table 5c. Data analyzed: Hendrik Age

Length of recordings 1;8 6 min 1;9 5 min 1;10 14 min 1;11 18 min 2;0 16 min 2;1 21 min 2;2 23 min 2;3 14 min 2;4 25 min 2;5 41 min

MLU Noun in tokens words

Number Noun of imi- lextations emes

1.6 1.6 1.5 1.4 1.5 1.6 1.7 1.8 2.0 2.8

0 1 2 2 0 0 1 0 1 1

9 26 53 87 72 84 106 79 116 386

8 9 10 25 23 30 33 26 45 84

Since the data from Hendrik and Anna is somewhat restricted, comparisons cannot be made on all points. Tables 5a, b, and c give an overview of the amount of recordings, types and tokens of nouns and pronouns as well as of local adverbs, which inflect for three local cases in Estonian. Instances of language play and imitations have been omitted from the analysis. It should be mentioned that the somewhat higher value of Hendrik’s MLU in the beginning of the observational period (age 1;8–1;9) is mainly due to the fact that the transcription of Hendrik’s speech is such that repetitions of words within utterances have unfortunately been taken into consideration by the MLU CLAN program.

The early development of case and number in Estonian

121

3. The acquisition of case in Estonian The case system is often regarded as one of the most difficult aspects of Estonian grammar: there are many cases, a single case often has several different allomorphs and the inflectional classes of nouns are not always simple or regular. In spite of the fact that the agglutinating case formation is not complicated in most cases historical developments have produced unpredictable stem changes (especially in the partitive and illative cases). This paper investigates how a child acquiring Estonian as his first language learns this difficult system, what helps or confuses him in the process and which preferences are found in different phases of the acquisition of inflectional morphology. Either words exhibiting a certain morphophonological structure may be preferably inflected or words which occur frequently in the input. The development of the Estonian case system will be analyzed from the appearance of the first rote-learned case forms in the premorphological phase to the appearance of 6- to 7-member miniparadigms. Tables 6a and 6b present the order of emergence of case forms in Andreas’ data by age indicating MLU values. In the first recording at 1;7 almost all nominals are in the nominative and a few of them in the partitive case; there is only one lexeme in the comitative. The first genitives, illatives and inessives are used at 1;8, the first allative and elative at 1;11. The order of appearance of different case forms in Hendrik’s speech is almost the same as in Andreas’. At the beginning of the observational period, almost all nouns are in the nominative and only few lexemes occur in the partitive and genitive. The next case forms to emerge in Hendrik’s speech are the illative and inessive (age 1;8–1;10); the number of nouns in the genitive and partitive increases at the age of 1;11. There are no instances of marginal cases in this boy’s speech either. In a 60-minute recording of Andreas’speech at the age of 2;0, the frequencies of the child’s singular case forms are quite similar to those found in the input. The most frequent case occurring in the input is the nominative (52 types/191 tokens). While the partitive (20/36), genitive (15/31) and illative (9/20) also occur quite frequently, the inessive, comitative, and allative are found less frequently, followed by rarely used elative and translative forms. Gagarina and Voeikova (2002: 128) found “a clear correspondence between MLU and PBF” (percentage of base forms of nouns) in Russian child speech with a minimal value of PBF reached at an MLU (measured in words per utterance) of about 1.7. Their study thus shows that case distinctions are tied to an MLU greater than 1.0. The base form in Estonian is the nominative singular. Since Andreas’ recordings only start at age 1;7, we cannot argue with absolute certainty that the category of case only emerges when sentences contain more than one word on average.

122

Reili Argus

Table 6a. The emergence of case forms from grammatical cases of nouns in Andreas’ speech (types/tokens) Age 1;7(MLU 1.1) 1;8 (MLU 1.2) 1;9 (MLU 1.5) 1;10 (MLU 1.6) 1;11 (MLU 1.8) 2;0 (MLU 3.6) 2;1 (MLU 2.7) 2;3 (MLU 3.8) 2;4 (MLU 4.1) 2;5 (MLU 4.1) 2;6 (MLU 3.9) 2;7 (MLU 3.9) 2;8 (MLU 4.6)

NOM 62/160 57/151 35/87 45/100 84/155 213/1155 88/626 37/95 64/249 33/136 67/235 95/323 87/238

GEN

PARTIT 8/22 9/12 14/46 5/18 17/66 99/266 40/90 7/22 15/51 16/24 25/66 19/36 32/52

5/10 4/7 9/13 20/50 64/230 47/145 19/31 18/43 7/18 18/32 35/73 22/37

Table 6b. The emergence of case forms from semantic cases of nouns in Andreas’speech (types/tokens) Age 1;7 (MLU 1.1) 1;8 (MLU 1.2) 1;9 (MLU 1.5) 1;10 (MLU 1.6) 1;11 (MLU 1.8) 2;0 (MLU 3.6) 2;1 (MLU 2.7) 2;3 (MLU 3.8) 2;4 (MLU 4.1) 2;5 (MLU 4.1) 2;6 (MLU 3.9) 2;7 (MLU 3.9) 2;8 (MLU 4.6)

COM ILL 1/5 2/6 4/4 1/2 2/8 7/15 22/32 16/71 10/18 15/34 4/5 2/2 3/8 7/14 3/4 6/9 4/7 11/24 12/15 6/6 10/16

INESS ALLAT ELAT ADESS ABL TERM TRANSL 1/2 1/1 1/2 2/3 14/25 10/19 3/3 11/13 2/5 7/10 2/4 9/10

1/1 20/48 7/17 1/1 6/9 3/4 3/5 8/26 16/33

1/1 7/17 10/16 2/2 6/7 1/1 4/8 2/3 4/4

5/11 5/20 2/3 6/8 2/3 7/18 7/10 4/9

1/1 1/1

1/1 1/2 2/2

1/1 1/1

3/9 2/5 1/3 1/1 1/1 1/1 5/8

At age 1;7, when Andreas’ MLU slightly exceeds 1.0 words per utterance, PBF, i.e. nominative singular forms, amounts to 85% (see fig. 1); however, there are already 11 nouns with more than one case form. PBF reaches its lowest point in Andreas’ speech at 2;3 (48%), but varies between 50% and 67% until the end of observation at 2;8. At 1;8, PBF drops to 70% and at 1;9 it almost reaches 55.9%, the value also found in a randomly chosen sample of the mother’s child-directed

The early development of case and number in Estonian

123

90 80 70 60 50 40 30 20 10 0 1;7

1;8

1;9

1;10

1;11

2;0

2;1

2;3

2;4

2;5

2;6

2;7

2;8

Age

Figure 1. PBF of nouns (tokens of nominative singular forms) in Andreas’ speech

speech at the child’s age of 2;0. PBF in child-directed speech is higher than that of adult-directed spoken Estonian, where it only amounts to 40% (Hennoste 2004: 18). The values of PBF are subject to intersubjective variation. While Hendrik’s PBF value is only 48% at age 2;3 (with an MLU of 1.8), it is still as high as 80% at the end of Anna’s observational period at 2;1. Evidence for the possibility of relating PBF to MLU counted in words comes from the fact that Hendrik’s and Anna’s PBF values are comparable at a point when their MLU is 1.5 (Hendrik: 75%, Anna: 80%). The Estonian data suggest that PBF values reach the level found in the input when the children’s MLU is approximately two words.

3.1. The premorphological phase: emergence of the first rote-learned case forms The premorphological phase is considered to be the phase in which children “access a limited number of lexically stored morphologically complex word forms” (Dressler et al. 2002: 392) so that certain inflected forms may be considered to emerge during this phase. Since the child’s utterances initially consist of a single word which often lacks the final syllable and thus the case ending, and since there isn’t any supporting syntactic context either, it is sometimes problematic to determine case forms. Although the extra-linguistic context as well as the parent’s reaction to the child’s utterance, which often consists in a repetition of the

124

Reili Argus

word in the correct case form, may help to determine the child’s intended case form, care must be taken not to over-interpret the first words occurring during the single-word period. Such forms are therefore more appropriately referred to as nominative- or partitive-like (e.g. when the word has a partitive stem but lacks a partitive suffix). Also, suffixless adverbial forms cannot be assigned to any particular case form. In order to distinguish between emergence and productive use of case inflection, we base ourselves on criteria set up by Pizzuto and Cazelli (1994: 156) and used by several authors, e.g. Gathercole et al. (1999) and for Estonian data by Vihman and Vija (2006). According to these criteria a given inflectional form is considered to be productive when either or both of the following conditions hold: “(a) the same verb root appear[s] in at least two distinct inflected forms, and (b) the same inflection [is] used with at least two different verbs” (Pizzuto and Cazelli 1994: 156), in this study we use the more rigorous interpretation: a given inflectional form is considered to be productive when both conditions hold.

3.1.1. First rote-learned nominative-, partitive-, illative- and inessive-like noun forms Children acquiring Estonian begin creating the case system with nominativelike noun forms: These appear at approximately age 1;0 in the diary data of both Andreas and Hendrik, while the first oblique forms emerge only about 3 months later in Andreas’ speech and 6 months later in Hendrik’s. Since the earliest audio recordings of all three Estonian children already comprise noun forms corresponding to several cases in the adult language, the period of nominativeshaped nouns shown by the diary notes from Hendrik and Andreas might be called a ‘caseless’ period in which there are only nominative forms used in different situations. However, it must not be forgotten that the frequent accurately used nominative-like forms of nouns do represent an important first building block of the Estonian case system. On the basis of tape-recorded data, the first oblique cases occurring in the speech of all three children may be held to be the partitive (Andreas 1;7.24: ‘kommi ‘candy:PARTIT’, ‘saia ‘bread:PARTIT’), the inessive (Andreas 1;8.10: ‘kotti ‘bag:INESS’) and the short illative (Hendrik 1;6: ‘˜oue ‘yard:ILL’). While the illative o˜ ue ‘into the yard, outdoors’ is frequent in the input, the nominative o˜ u is very rare in Standard Estonian and not available at all in the input. Rather, this word is mostly used in adverbial expressions and the form o˜ ue is to a great extent lexicalized. Although some adult-like forms of nouns carrying a suffix do occur already in the premorphological phase, most of the first rote-learned case forms do not

The early development of case and number in Estonian

125

carry a suffix. In such forms, the category of case is expressed by the stem variant corresponding to the case form and its function in Standard Estonian. The reason for the preference of early suffixless forms may be phonological rather than morphological and may consist in the child’s inability to pronounce closed second syllables. Forms lacking final consonants have been found to predominate during the prelinguistic or babbling period (Vihman and Boysson-Bardies 1991: 316). Also, the ‘trochaic stage’ (defined by Laalo 1994: 430) seems to play an important role: At this stage, only 1.2% of Andreas’ and Hendrik’s tokens exceed a length of two syllables. In Estonian, all longer words are shortened to two syllables by the two boys studied (Andreas 1;8.25: taaki for draakoni ‘dragon:GEN’) and some monosyllabic words may even be lengthened to obtain the preferred pattern (e.g. Hendrik 1;7: lutt > luti ‘pacifier’). The trochaic stage can be considered to end for Andreas at 1;10.03 when no more shortened words are to be found in the child’s speech and the percentage of words longer than two syllables has risen to 9.5%, thus approaching the value of approximately 10% characteristic of child-directed speech. Since morphological rules in Estonian derive to a great extent from the morphophonological pattern of the word, the preference for certain morphological patterns cannot readily be distinguished from morphophonological preferences. Still, the preferences apparent in the first rote-learned case forms are not purely morphophonological: it must not be forgotten that a choice of short (fused) over long (agglutinating) case forms only exists in the partitive and illative singular. The frequency of short illative and partitive forms is much higher than that of the long forms also in the input. Andreas’ first rote-learned partitives occurring between 1;8 and 1;11 are all formed with the zero allomorph based on words with a disyllabic strong grade vocalic stem: ‘nuppu ‘button:PARTIT’ (the weak stem of nupp ‘button:NOM’ is used in nupu ‘button:GEN’). The first partitive suffixes emerge at 1;11 and are marked by the allomorphs -da, -t or -d, the first of which occurs only with the pronoun see ‘it:NOM’ vs. se-da ‘it-PARTIT’. The second allomorph is used with trisyllabic stems (e.g. kohukes-t ‘cottage cheese-PARTIT’) while the third is added to monosyllabic vocalic stems and to loanwords ending in a diphthong (e.g. patarei-d ‘battery-PARTIT’) in the speech of Andreas. More generally, at 1;11, use of the partitive case ending is still lexically based and extremely rare as far as its use with different noun lemmas is concerned. Although the token frequency of the partitive form of the personal pronoun see ‘it’ is quite high (29 instances), there is only one token each of partitive forms of nouns carrying the suffixes -t and -d. Other partitive forms of 14 lexemes are expressed by the appropriate choice of the suffixless stem (32 tokens). At 2;0, when the suffix -t occurs more frequently (28/47 types/tokens of a total of 90/296 partitive

126

Reili Argus

types/tokens), the ending d remains limited to only two lexemes. An explanation for this is the different frequency of these suffixes in the input, the suffix -d occurring only with words belonging to the small and unproductive inflectional class 2 (see table 3). The short illative form appears inAndreas’speech at 1;8 (e.g. ‘kotti ‘bag:ILL’), while the first occurrence of the long illative with the ending -sse is only found at age 2;0 (e.g. tasku-sse ‘pocket-ILL’). Short illatives are usually used with high-frequency words and are highly irregular. Although long illatives can, in principle, be formed from these words (e.g. koti-sse ‘bag-ILL’) they are neither used in colloquial Estonian nor in child-directed speech, so that they are not found in the input. The child’s use of the short illative rather than the long one is therefore simply a reflection of input frequency. Both of the children Andreas and Hendrik also distinguish case forms with the help of the weak and strong grade (e.g. Andreas at 1;8: piti ‘picture:GEN’ vs. ‘pitti ‘picture:PARTIT’). Hendrik uses quantity contrasts of the stem to differentiate between local cases from age 1;8 on (e.g. ‘nanni ‘bath:ILL’ in Q3 vs. nanni ‘bath:INESS’ in Q2; linna ‘town:INESS’ in Q2 vs. ‘linna ‘town:ILL’ in Q3). The suffixless case forms in Q3 are often used to express the partitive or illative and many times occur in commands addressed to children. Since children are frequently exposed to commands (see also Dasinger 1997: 39), these forms may well affect the early acquisition of grade contrasts. With the three children, the first forms of nouns contrasting with the nominative are usually forms of other grammatical cases, namely the partitive and genitive; some instances of the nominative opposed to the illative or comitative forms of a given lexeme are also found. Pragmatic factors seem to play an important role in the differentiation of the nominative and oblique case forms: Lexemes demonstrating the first nominative and partitive contrasts in the speech of two children were food items, such as kook ‘cake:NOM’ vs. ‘kooki ‘cake:PARTIT’ (Andreas 1;7, Hendrik 2;0); m¨amm ‘food:NOM’ vs. ‘m¨ammi ‘food:PARTIT’ (Hendrik 2;1). Nominative forms are used in pointing to food and partitives in demanding food. Words occurring in other contrasts are all names of toys. Thus, the nominative and genitive torn ‘tower:NOM’ vs. torni ‘tower:GEN’ (Andreas 1;7) were used in a play situation, the nominative for pointing to a toy and the genitive for commenting on the child’s own activity (e.g. teen torni ‘(I) make a tower’). Among Andreas’ first rote-learned case forms at age 1;7–1;8, there are three comitatives of three different lexemes (e.g. k¨atu-ga ‘hand-COM’). Since the early appearance of the comitative in Estonian was also noticed by Vihman (1982: 151), who points out that this case ending “is a full syllable in length and always occurs word-finally, it is completely unchangeable in form (that is, there is no allomorphy), and it is not normally redundant.” It must, however, be

The early development of case and number in Estonian

127

mentioned that there are other adverbial cases which also have no allomorphy but do not occur as early. The role of frequency played by the comitative is rather dubious, as the comitative is not frequent in the input. What seems to be most important for the early emergence of this case is that it most often functions as an instrument in the input, which is also the first function to emerge in child speech. In the speech of the three children, the first rote-learned case forms express semantic roles such as neutral participant (nominative), object (partitive), destination (illative) and location (inessive). In addition to the importance of these semantic relations the nouns expressing them occupy a salient position in the utterance in the input: thus, the object, destination and location are usually in sentence-final position. This factor may have some bearing on the early acquisition of these forms and Slobin’s operating principle of attention (end of unit) (Slobin 1985: 1251) may be relevant also on clause level.

3.1.2. First genitives Use of the genitive is closely connected to the need for indicating possession. This case does not usually occur during the one-word period but appears only when the mean length of utterance exceeds 1.00 and possession can be clearly expressed by mentioning both possessor and possessed. Nevertheless, some genitive-like nouns may be found in earlier one-word utterances, where the possessive meaning of the child’s utterance can be determined from the situational context and the adult’s reaction. In spite of the fact that the noun emme in example (1) belongs to an inflectional class in which there is syncretism between genitive and nominative (inflectional class 6 in table 3), it becomes quite clear from the context that the form used by the child indicates the possessor. Andreas’ first clear use of genitive forms is found at age 1;8, when his MLU has clearly risen above 1.0. Hendrik’s first genitives appear at 1;6 (MLU 1.4), a bit earlier than Andreas’. There are no genitive forms to be found in all of Anna’s data, and she does not express the possessive relation by other linguistic means either. (1)

Andreas 1;7.24 MOT: a¨ ra se-da v˜ota, AUX see-PARTIT take:IMP See on masin, it:NOM be:3S machine:NOM diktofon, a¨ ra tee. dictating machine:NOM AUX do:IMP ‘Don’t take it, it is a machine, a dictating machine, don’t do it.’

128

Reili Argus

CHI:

emme. mummy:GEN/NOM:SG ‘mommy’ MOT: emme oma jaa. mummy:GEN own yes. ‘(it is) mommy’s yes.’ The early genitives in the two boys’ speech all come from nouns belonging to the inflectional class in which the nominative and genitive singular coincide. Examples from Hendrik at 1;6 are kutsu m¨amm ‘dog:GEN food:NOM’, venna auto ‘brother:GEN car:NOM’ and from Andreas at 1;8 emme k¨asi ‘mummy:GEN hand:NOM’. Input preferences seem to play a considerable role here: Twosyllable words in Q2 which do not formally distinguish between the nominative and genitive singular occur very frequently in Estonian child-directed speech (see also Pajusalu 2001: 85). Not only do these forms fit the trochaic phonological pattern, but the homophony of the two cases also provides the beginning learner with convenient multi-purpose forms. Diminutives occurring in child-directed speech also play a role in the appearance of first genitives. There are several ways of forming diminutives in Estonian, either by adding one of the derivational suffixes -u (k¨asi ‘hand:NOM’ > k¨at-u ‘hand-DIM’) and -ke(ne) (koer > koera-kene ‘dog-DIM’) or by shortening and modifying the stem simultaneously adding a suffix-like element – often -i – (ema > emme ‘mother:DIM’, isa > issi ‘father:DIM’, vend > venna ‘brother:DIM’). The first and last of these options are the ones mainly used in child-directed speech. With these two patterns of diminutive formation the stem loses its grade alternation and shifts to the inflectional pattern described above where nominative and genitive are homophonous and the partitive takes the regular case suffix -t. In early child speech as well as in child-directed speech diminutives mainly refer to family members. In Estonian, case marking also plays a role in expressing the contrast between perfective and imperfective aspect. While the nominative or genitive form of the noun accompanied by particles such as a¨ ra ‘away, off, out’ or otsa ‘to the end’ is used for the perfective meaning of an action exerted on a ‘total object’, the imperfective meaning applied to ‘partial objects’ is marked by the partitive case (Metslang 2001: 443). In the children’s speech, the function of total object is first expressed by a construction in which the genitive form of the noun is accompanied by a perfective particle. In example (2), the child uses the endingless genitive stem of the noun overtly marking perfectivity quite clearly by using two such particles.

The early development of case and number in Estonian

(2)

Andreas 2;1 Atsu s¨oo¨ -b koore Atsu:NOM eat-3S cream:GEN ‘Atsu will eat the cream.’ Atsu s˜o-i a¨ ra Atsu:NOM eat-PAST VERBAL.PTL otsa. VERBAL.PTL ‘Atsu will eat the cream up.’

129

a¨ ra. VERBAL.PTL koore cream:GEN

Since adverbial cases are formed by adding the case suffix to the genitive (vocalic) stem (see section 1.1), this stem form also plays an important role in the acquisition of local and other adverbial cases. Before expressing adverbial case functions by suffixed nouns, both Andreas and Hendrik use bare vocalic (genitive) stems as all-purpose forms for these functions as well as others (e.g. Andreas 1;8: musta instead of musta-ks ‘dirty-TRANSL’, poti instead of poti-le ‘potty-ALL’). The only exception is the comitative, which occurs twice with the appropriate case suffix in Andreas’ speech. Case forms marked by a suffix mostly appear a month or two later. Case distinctions of some marginal cases also develop later.

3.1.3. Synthetic vs. periphrastic expression of local relations The three possibilities for expressing location in Estonian (local case forms, adverbs or postpositions with the genitive form of the noun) develop in the same direction, from interior to exterior and from goal-oriented (directional) to source-oriented (separative) (see table 4b). Deixis develops from egocentric to exocentric. The premorphological phase is not only characterized by the emergence of case forms of nouns but also by deictic adverbs appearing almost at the same time as the first illative and inessive case forms of nouns. Adverbs such as siia, siin, siit ‘up to here, here, from here’ and sinna, seal, sealt ‘up to there, there, from there’ have three local case forms each and form triads including a directional, a stative and a separative member. During the premorphological period, local adverbs are more frequently used to express local relations than nouns in local case forms. The reason will most probably be that local adverbs are also much more frequent than local case forms of nouns in the input because, in this stage of the child’s development, mother-child interaction usually refers to the here-andnow. Thus, in the tape-recorded data, Andreas (aged 1;10) expresses location with 3 local adverbs 29 times and with 3 different nouns in a local case only

130

Reili Argus

Table 7. Development of local adverbs (Andreas) Deictic local adverbs Stative Dynamic goal-oriented (‘directional’) source-oriented (‘separative’)

Proximal siin ‘here’ 1;7 siia ‘up to here’ 1;7 siit ‘from here’ 1;10

Distal seal ‘there’ 1;11 sinna ‘up to there’ 1;8 sealt ‘from there’ 2;0

9 times. The more frequent use of local adverbs as compared to local case forms of nouns is typical of the other two children as well. Proximal deictic local adverbs emerge before distal ones in the speech of all three children. The goal-oriented dynamic and the stative proximal adverbs siia ‘up to here’ and siin ‘here’ appear two or three months before the source-oriented dynamic ‘separative’ adverb siit ‘from here’ in Andreas’ (1;7 vs. 1;10) and Hendrik’s speech (1;11 vs. 2;1). The distal source-oriented adverb sealt ‘from there’ is first found only during the protomorphological phase. Neither the proximal nor the distal dynamic source-oriented adverb is found in Anna’s data. Table 7 gives an overview of the emergence of local adverbs in Andreas’ speech. The development of local cases of nouns resembles that of local adverbs insofar as, within the sets of interior local cases, the directional illative as well as the stative inessive emerge in Andreas’ speech at 1;8, two or three months before the separative elative, which only appears during the protomorphological phase. While the elative does not occur at all in Anna’s and Hendrik’s data, this is case is restricted to a single token with Andreas occurring at 1;11 (jala-st ‘footELAT’ (= from foot)). The first allative is found with Andreas at 1;11 when he uses the pronominal form mu-lle ‘me-ALL’, which should be considered to have been learned as an unanalyzed form. One month later, the allative begins to occur with nouns such as emme-le ‘mummy-ALL’ (uttered when handing a piece of bread to mummy) and pissi-le ‘pee-ALL’ (indicating that he wants to go to pee). Since the stem emme occurs in two different case forms and the allative suffix is used with at least two different lexemes, this marks the beginning of the productive use of the exterior local case according to the above criterion (see sect. 3.1). From 2;0 on, Andreas uses the allative more frequently (both typeand tokenwise) than the illative and the inessive. In standard Estonian, either the genitive of the noun accompanied by a local postposition or local case forms may occur in the same contexts without any difference in meaning (Tauli 1980: 316), but postpositions are sometimes used to emphasize the physical state of being inside or on top of something. As far as Andreas is concerned, he first develops synthetic forms carrying a suffix for interior local cases (e.g. k¨oo¨ gi-s ‘kitchen-INESS’, 1;11). Some constructions consisting of the genitive form of the noun and a postposition precede synthetic

The early development of case and number in Estonian

131

forms for the expression of exterior local relations (e.g. bussi peale ‘to the bus’, 1;10). Other postpositions appear at the same time as locative case forms (e.g. auto peal ‘car:GEN on top of (= on the car)’ and emme-l ‘mommy-ADESS’ at 2;0). This kind of acquisition order can be explained by the usage of different possibilities for expressing local relations or more abstract roles as possessor or recipient: construction with the postposition peale can be used for the semantic role of location, while the corresponding case form can be used more likely for the role of recipient. This kind of different usage is not so relevant for interior local cases, where the child acquires first local case forms and not postpositions. Later, from the age of 2;0 onward, the number of postpositions abruptly increases in Andreas’ speech. At 2;0 the distribution of the three different possibilities for expressing location (case forms, local adverbs and genitive with local postposition) is as follows: 49% of all local constructions are expressed by local adverbs, 41% by local case forms and only 10% by the genitive and postpositions. Hendrik uses local adverbs even more frequently (67% of tokens) at the end of premorphological phase; 30% of all local constructions are expressed by local case forms and only 3% by the genitive and postpositions. The same tendency to prefer local adverbs is observable in Anna’s speech: at age 2;1, 80% of local constructions are expressed by local adverbs, 15% by local cases and 5% by postpositions. As far as the relation between the expression of local relations in the input and the order of development is concerned, it seems important to note that adverbs and local case forms of nouns occur with about equal frequency in a 60-minute input sample at one child’s age 2;0 (77 noun tokens vs. 79 adverbs). Postpositions used with nouns are only used a little more than half as often in locative expressions (35 tokens). One of Andreas’ first allative forms is used for expressing location, but the role of recipient is more frequent among allatives: thus, in the data recorded at 2;0, 33 of 46 tokens in the allative case express the recipient while only 13 of them express local relations. On the whole, our results on the early acquisition of Estonian local cases are similar to those of other researchers (Vihman, 1982: 148; Dasinger, 1997: 40). Contrary to what we found withAndreas, however, Vihman (1982: 148) observes that in the acquisition of Estonian local cases by a bilingual child growing up in the USA, postpositions such as juurde ‘for, to’, sisse ‘into’ appear first while the corresponding case suffixes are acquired later. The use of allative case in the function of recipient before its use for the locative function does not agree with what has been noted by Dasinger (1997: 43), in the speech of children studied in this paper the allative case forms emerged in functions of location and recipient at the same time.

132

Reili Argus

3.1.4. Main achievements in the acquisition of Estonian case during the premorpological phase The data analyzed in the present paper shows that the main achievement concerning the acquisition of the Estonian case system during the premorphological phase is the acquisition of stem changes of nouns which function as forms of grammatical cases in standard Estonian and may form the basis of other cases. During this early developmental period, the child uses frequent endingless allpurpose case forms of nouns almost correctly. This is also true of one or two forms each of very few lexemes carrying an inflectional suffix. At the beginning of the observational period at 1;7, there are errors with approximately 12% of noun tokens and 20% of noun types. Production errors concerning noun morphology which occur in the premorphological phase can be divided into two classes, functional ones as well as overuse of suffixless forms. Functional errors are very likely to occur when the inventory of grammatical forms is small so that forms are used in inappropriate contexts besides appropriate ones. Such functional errors occur with 5 noun types in Andreas’ speech, with the illative most frequently being used instead of the inessive. Overuse of the suffixless forms, usually the genitive, is found with 12% of all noun types and 7.3% of all noun tokens in local and marginal cases and concerns contexts in which a form carrying a suffix would have been appropriate. The number of last type of errors decreases to 8% at the end of the premorphological phase, with the amount of overuses of suffixless stems decreasing to 4.7%. As far as the development of the function of case forms is concerned, grammatical cases such as genitive and partitive are first acquired in their main semantic roles, namely possessor and object. The nominative, however, is not acquired first in its main role, the agent, but in the role of a neutral participant (predicative). As far as the input is concerned, the nominative usually encodes the subject (Blake 2001: 31), much less the predicative and only marginally the total object. From this it is to be expected that the latter functions of the nominative will emerge later than the subject function in the children’s speech. It can be supposed that the role of the subject is pragmatically not as important for the child as, for example, the role of the neutral participant in pointing sentences. Looking more closely at the child’s sentences containing a verb which requires an agent, it becomes apparent that at the premorphological phase the child uses verbs mostly for commenting on his own activities so that the appropriate verb form is marked for first person and there is thus no need for an overt subject. During the premorphological phase there are also many utterances which accompany a pointing gesture and in which the nominative functions as a neutral participant. Among the adverbial cases, the illative and inessive are used in their

The early development of case and number in Estonian

133

main semantic roles, destination (illative) and location (inessive). The allative is also first acquired in its most important semantic roles of location and recipient and only afterwards in the roles of beneficiary.

3.2. The protomorphological phase: the emergence of miniparadigms and overgeneralizations The premorphological phase is of utmost importance for the development of morphology because morphological rules can be detected when a certain amount of word forms have been memorized. It is in the protomorphological phase of language acquisition that the child starts to generalize over such rote-learned forms (see Dressler 2003: 10). The emergence of ‘true miniparadigms’ in children’s speech, which is characteristic of this second phase of morphological development, is evidence of the detection of morphology. A ‘true miniparadigm’ is defined by Bittner, Dressler and Kilani-Schoch (2003: xvi) “as corresponding to a non-isolated set of minimally three phonologically unambiguous and distinct inflectional forms of the same lemma produced spontaneously in contrasting syntactic or situative contexts in the same month of recordings”. Before the appearance of true miniparadigms, the tape-recorded data includes some ‘miniparadigm candidates’, usually consisting of imitations of one or two forms. The first three-member paradigm candidate occurs in Andreas’ data at 1;7.24 and consists of the nominative, partitive and comitative forms of the lexeme buss ‘bus’. It must be mentioned that the comitative is not productively used at this age. Rather, the early comitative form of buss has been rote-learned since, in the same dialogue, the child uses the suffixless genitive stem of other lexemes in the function of the comitative. At 1;8.10, Andreas uses the nominative, partitive and genitive of the lexeme emme ‘mummy’. Since this lexeme belongs to an inflectional class where the genitive and nominative singular are homonymous, it is problematic to speak of a true miniparadigm here. Although contrasts of two forms also occur in the recordings between 1;8.10 and 2;0, true miniparadigms only appear at the latter age (see table 8). At this point, 69% of lexemes occur in only one case form, 17% in two contrasting case forms and approximately 13% of lexemes form miniparadigms. In spite of the fact that more data were collected at 2;0 than during the following month (360 min. at 2;0 as opposed to 90 min. at 2;1; also see table 5a), the percentage of lemmas having one, two or more case forms is almost the same. The first half of the 26th month is a period of rapid inflectional development, with paradigms growing to a size of four or five members which include not only

134

Reili Argus

the grammatical cases and the comitative, but also exterior local cases, mostly marking destination and location. The first miniparadigms are clearly lexeme-based: there is a certain choice of case forms for each lexeme. Thus, the noun kast ‘box’ is used in the genitive, inessive and elative (Andreas at age 2;0) while other lexemes may be used with different cases. For some pragmatically important lexemes, such as issi ‘daddy’, emme ‘mummy’, piss ‘pee-pee’, see ‘this’, the child needs more forms while for others, three seem to suffice (e.g. m¨uts ‘hat’). Adverb-like nouns usually have only two local case forms, illative and inessive (e.g. ‘˜oue ‘(to go) outside:NOM’ and o˜ ue-s ‘(to be) outside-INESS’); ‘tuppa ‘room:ILL’(= (to go) inside) vs. toas ‘room-INESS’ (= (to be) inside) Andreas 2;1). Table 8 represents some of Andreas’ first miniparadigms at age 2;0. As can be seen, his miniparadigms mostly consist of grammatical cases. Some of the local cases, like the illative and ablative, and marginal cases such as the terminative, abessive and essive, do not occur in miniparadigms at this age. Nominative plural and partitive plural constitute a miniparadigm together with singular forms of grammatical cases.

Table 8. Andreas’ miniparadigms at age 2;0 NOM lill ‘flower’ klots ‘box’ Atsu ‘child’s name’ issi ‘father’ piss ‘peepee’ kiisu ‘cat’

GEN lille

PARTIT INESS ELAT ALL ‘lille

ADESS TRANSL COM

NOM: PL PARTIT:PL lilli

klotsi

‘klotse

Atsu

Atsu-t

Atsu-le Atsu-l

issi

issi-t

issi-le

pissi

pissi-le

kiisu-t

kiisu-le kiisu-l

kasti ‘box’ kutsa ‘doggy’ m¨uts m¨utsi ‘m¨utsi ‘hat’ kilpkonn kil-konna [kilkonn] ‘tortoise’ see selle se-da ‘this’

Atsu-ga

pissi-ks

Atsu-d

pissi-ga

kiisu-d

kasti-s kasti-st kutsa-le

kutsa-ga

kutsa-d

kil-konna-d

selles

selle-le

selle-ga

need

The early development of case and number in Estonian

135

Andreas’ first words forming miniparadigms at 2;0 belong to one and the same regular and productive inflectional class with weakening gradation (see the entry siil in tables 2 and 3): poeg ‘son’, lill ‘flower’, klots ‘toy block’. At this point of the boy’s development, the input frequency of words of this class is quite high: 35% of all nominal lexemes are monosyllabic, with 83% of these belonging to the above-mentioned inflectional class, where the nominative is monosyllabic and the genitive fits to the trochaic phonological structure. The input frequency of these words does not change significantly during the observational period, so that approximately 40% of all nouns belong to the above mentioned inflectional class also when the boy has reached age 2;8. The above described inflectional class in the most productive one in Estonian, for example, a lot of new loanwords tend to inflect according to this class. There are not many trisyllabic words in Estonian. Most longer words are compounds only the head (2nd constituent) of which inflects for case. Andreas’ first miniparadigm of a trisyllabic word occurs at age 2;0 (lusikas ‘spoon:NOM’, lusika-st ‘spoon-ELAT’, lusika-ga ‘spoon-COM’). The other longer words in his miniparadigms are all compounds with a monosyllabic inflecting head belonging to the inflectional class with weakening gradation (see table 3, entry siil). At 2;0, Andreas’ miniparadigms are thus formed either from monosyllabic words of the type lill ‘flower’ or disyllabic words of the type kiisu ‘kitty’, so that there appear to be two prototypical inflectional classes, one for monosyllabic words with grade alternation and another one for disyllabic words without grade alternation. These two inflectional classes of nominals are the most productive ones in Standard Estonian. Although there are several contrasts of two case forms to be found in the speech of Hendrik and Anna (e. g. asi ‘thing:NOM’ vs. ‘asja ‘thing:PARTIT’ Hendrik 2;5), there are few miniparadigms in their data, probably because of the restricted amount of data available for these children. All of their miniparadigms are presented in tables 9 and 10. Two of Hendrik’s four miniparadigms found at 2;5 as well as Anna’s single miniparadigm at 2;1 consist of grammatical cases, while Hendrik’s other two miniparadigms include one local case, the illative. It is certainly no coincidence that miniparadigms of the suppletive pronoun see ‘this, it’ are the first to be found with both Hendrik and Anna, since a miniparadigm of this pronoun also occurs early (at 2;0) in the larger corpus of Andreas. This pronoun is the most frequent nominal occurring in the input to all three children and is used in the nominative, partitive, genitive, comitative, allative and elative (with decreasing token frequency) by their caretakers. Since see (as well as some other demonstratives) follows an idiosyncratic suppletive inflection, its case forms must be acquired one by one. The miniparadigms of this pronoun therefore seem to

136

Reili Argus

Table 9. Hendrik’s first miniparadigms Age 2;1 2;5

NOM see ‘this’ nann [vann] ‘bath’ auk ‘hole’ uks ‘door’ koll ‘monster’

GEN PARTIT ILL COM NOM:PL selle seda selle-ga nanni ‘nanni augu ‘auku ukse ukse-d kolli kolli-d

Table 10. Anna’s first miniparadigms Age 2;0 2;1

NOM see ‘it’ liblikas ‘butterfly’

PARTIT seda liblika-t

NOM:PL need liblika-d

support the observation that the first miniparadigms are formed of lexemes frequently occurring in child-directed speech and that more cases are distinguished in pronouns than in nouns at the beginning of the protomorphological phase. The choice of case forms appearing in miniparadigms seems to depend on the semantic category of the noun. There are three main kinds of miniparadigms to be found in Andreas’ data at 2;6: miniparadigms of nouns for food containing three grammatical cases (e.g. p¨ahkli ‘nut:GEN’, p¨ahkli-d ‘nut-NOM:PL’, p¨ahklei-d ‘nut-PARTIT:PL’), miniparadigms of animate nouns and baby-talk words containing grammatical cases and the allative (e.g. issi ‘daddy:NOM’: issi ‘daddy:GEN’ : issi-t ‘daddy-PARTIT’ : issi-le ‘daddy-ALL’) and finally miniparadigms of words expressing location consisting of two or more local cases (e.g. sahtli-sse ‘drawer-ILL’, sahtli-s ‘drawer-INESS’, sahtli-st ‘drawer-ELAT’). By the end of the observational period at age 2;8, Andreas’ miniparadigms mostly include three or four members, often with both singular and one or two plural case forms. The seven-member miniparadigm of the pronoun see ‘this’ constitutes an exception. The percentage of nouns occurring in miniparadigms (11%–13%) is approximately the same at 2;8 as in the beginning of the protomorphological phase at age 2;0. Table 11 gives an overview of the development of the number and size ofAndreas’miniparadigms.At age 2;0, when the amount of data analyzed is four times as large as a month later, the number of miniparadigms is also exactly four times bigger. While several 4-member miniparadigms occur in each month with the exception of 2;3, the number of miniparadigms containing more than four forms is still very restricted by 2;8. Paradigms found in the input (recorded data from child’s age 2;0–2;8, one hour per month) consist of maximally 6 members, while there are only 1–3 four-member and 3–5 three-member

The early development of case and number in Estonian

137

Table 11. The development of miniparadigms in Andreas’ speech Age

1;7 1;8 1;9 1;10 1;11 2;0 2;1 2;3 2;4 2;5 2;6 2;7 2;8

MLU 2-member 3-member 4-member oppos. miniminiparad. parad. 1.1 1 1 1.2 3 1.5 6 1.6 5 1.8 18 3.6 78 32 6 2.7 42 8 5 3.8 13 3 4.1 21 8 4 4.1 4 5 3 3.8 23 8 4 3.9 28 12 2 4.6 21 14 3

5-member 6-member 7-member miniminiminiparad. parad. parad.

4 3

1 1 1

3 1 1 1

1

1 1 1

miniparadigms in these data. Also, the number of simple contrasts of forms is not much larger in the input than in the child’s speech. Thus, there are 25 such contrasts in the child’s data at age 2;7 and 32 oppositions at child’s age 2;7 in the input. The semantic roles expressed by the first rote-learned case forms in the premorphological phase and by nouns partly occurring in case contrasts or miniparadigms in the protomorphological phase are largely the same, namely neutral participant, agent, state, partial object, location, recipient, companion and instrument. Tables 12–14 present an overview of semantic roles expressed by nouns in Andreas’ speech during the protomorphological phase from 2;0 to 2;8. While the nominative is used as a predicative in the premorphological phase, the function of agent appears later, at age 1;11 (e.g. Adu kirjutab ‘Andreas is writing’). The use of the construction consisting of child’s name and a verb in the 3rd person for commenting his own activities increases at Andreas’ age 1;11 (see also Vija 2007: 376). Hence, the a phase when children refer to themselves by the 3rd person often using their name is connected with the emergence of the nominative case forms in the semantic role of the subject. Andreas mainly uses nouns in the genitive to express a total object or goaloriented local relations; in the latter case, the noun is accompanied by a locative postposition. The genitive is also used within noun phrases for expressing an attribute, namely the possessor. These functions of the genitive do not all emerge

138

Reili Argus

Table 12. Semantic roles of case forms of grammatical cases in Andreas’ speech in the protomorphological phase Case Semantic roles Nominative neutral participant agent

state

Genitive

possessive attribute

total object

Partitive

partial object

Examples See on buss ‘it:NOM be:3S:PRES bus:NOM’ 2;0 Atsu v˜ota-b Atsu:NOM take-3S ‘Atsu is taking’ 2;0 Ma ole-n haige I:NOM be-1S sick:NOM ‘I am sick’ 2;8 issi s¨ulle father:GEN lap:ILL ‘(into) daddy’s lap’ 2;0 v˜ot-si-n klaveri kaasa take-PAST-1S piano:GEN POSTPOS ‘(I) took (the) piano with (me)’ 2;8 (N¨aen) bussi bus:PARTIT ‘(I see) the bus’ 2;0

simultaneously: while the functions of possessor and goal-oriented local relations appear early, the function of the total object does not emerge until age 2;8. The partitive almost always signals a partial object. Other frequently occurring semantic roles are the possessor expressed by the adessive case (e.g. kiisu-l ei ole konti ‘cat-ADESS NEG.PTL COP:PRES bone:PARTIT’ [= cat has no bone], 2;0) and the instrument expressed by the comitative (e.g. Andreas kopsi-b haamri-ga ‘Andeas:NOM tap-3S hammerCOM’ [=Andreas is tapping with a hammer]). The functions described above are also the ones found most frequently in the input. While the interior local cases, inessive and elative, are used by Andreas only with their main functions, the exterior local cases also fulfill some marginal ones, such as beneficiary (at age 2;1) and recipient (2;0). Of the marginal cases, the translative occurs at the end of Andreas’ observational period expressing a state. Semantic roles develop from the prototypical ones, such as partial object, location and neutral participant, already found in the premorphological phase, to more marginal ones such as total object and beneficiary at the end of the observational period at 2;8. The only exception is the role of the agent, which is not expressed during the premorphological phase.

The early development of case and number in Estonian

139

Table 13. Semantic roles of case forms of local cases in Andreas’ speech in the protomorphological phase Case Illative

Semantic roles location (goal, dynamic)

Inessive

location (stative)

Elative

location (source, separative)

Allative

location (dynamic, goal-oriented) recipient

Adessive location (stative) possessor

Ablative

location (source, separative)

Examples tule-me koju come-1P home:ILL ‘(we) come home’ 2;8 Ei ole kasti-s NEG.PTL COP:PRES box-INESS ‘(It) is not in the box’ 2;0 (V˜ota) kasti-st a¨ ra box-ELAT VERBAL.PTL ‘(Take) out of the box’ 2;0 (L¨ahen) pissi-le pee-ALL ‘(I go) to pee’ 2;0 vanaema and-is mul-le asja-d grandmother:NOM give-PAST me-ALL thingPL:NOM ‘Grandmother gave me things’ 2;8 ole-me maa-l be-1S countryside-ADESS ‘(we) are in the countryside’ 2;8 mu-l see hea vesi ‘I-ADESS it good:NOM water:NOM ‘I have the good water’ 2;8 v˜ota-me teise-lt poolt take-1P another-ABL side ‘Let’s take from the other side’ 2;8

Table 14. Semantic roles of case forms of marginal cases in Andreas’ speech in the protomorphological phase Case Semantic roles Translative state (result)

Comitative companion

instrument

Examples Ma tee-n ta k˜ovema-ks. I:NOM make-1S her stronger-TRANSL ‘I’ll make her stronger’ 2;8 S˜oida-me issi-ga m¨aguv¨aljaku-le. drive-1P daddy-COM playground-ALL ‘We’ll drive with daddy to the playground’ 2;8 Lusika-ga sega-n. spoon-COM mix-1S ‘(I) mix with a spoon’ 2;8

140

Reili Argus

3.2.1. Overgeneralizations As sketched in section 1 above, Estonian has many inflectional classes and a quite complex morphophonology. Children use different strategies to simplify this complicated inflectional system. One such strategy is the overextension of a regular inflectional pattern to irregular nouns. At 2;0, Andreas forms the partitive of the irregular noun juhe ‘wire:NOM’ based on the genitive stem juhtme, as is usual with regularly inflecting nouns without grade alternation, producing juhtme-t instead of juhe-t ‘wire-PARTIT’ without quality alternation. Between 2;1 and 2;4, there are also examples of overgeneralization of a regular and more productive inflectional class marking a given grammatical category overtly (Slobin 1985: 1222) and thus avoiding grammatical homonymy. Such a preference is observable in Andreas’ speech. There is syncretism of the suffixless forms of the nominative, genitive and partitive with nouns belonging to the ema class (see table 3). However, Andreas inflects words like muna ‘egg’ and saba ‘tail’ of this inflectional class according to the pattern of another inflectional class (see auto in table 3), overtly marking the partitive by the suffix -t (e.g. muna-t ‘egg-PARTIT’). Interestingly, although Andreas’ parents do not use such forms in their child-directed speech, a similar development is noticeable in colloquial Estonian, where the same inflectional class shift has taken place so that the word ema ‘mother’ is often inflected like words belonging to the inflectional class where the partitive takes the suffix -t (see also Keevallik 2003: 362). So-called class shift can also take place between two equally productive inflectional classes. At 2;0, Andreas uses the partitive suffix -t also with words belonging to the inflectional class of siil (see table 3) (e.g. kausi-t ‘bowl-PARTIT’ instead of kaussi Q3). Overgeneralization of the regular suffix -sse can be observed in Andreas’ speech at 2;4, when, after acquiring the illative suffix -sse (age 2;2), he starts to use it also with words for which the short illative is normally used: t¨uhja-sse ‘empty-ILL’ instead of ‘t¨uhja. These examples of the overextension of overt case marking by suffixes seems to be evidence that, in the protomorphological phase, when the child starts to construct the case system, he prefers more transparent case suffixes like -t or -sse to less transparent case formation by grade alternation or zero marking resulting in homonymous forms. The examples of overgeneralizations of the partitive and the illative suffix presented here support Slobin’s Operating Principle “Extension”, in which it is claimed that the child will mark a given grammatical notion he has discovered “on every member of the word class” so that it will also be overextended to zero marking in the standard language (Slobin 1985: 1222). It must be noted, however, that the percentage of such overgeneralizations is very low; thus, only 0.6% of

The early development of case and number in Estonian

141

all partitive singular and 8% of all partitive plural forms are overgeneralized. This seems to show that the child relies more on memory with the more frequent singular forms but less so with the less frequent plural forms.

3.2.2. Main achievements in the acquisition of Estonian case during the protomorphological phase Since some case contrasts and even true miniparadigms emerge early, the acquisition of Estonian noun inflection cannot be strictly separated into two clear-cut stages of pre- and protomorphology, so that the transitional period to protomorphology may already start when most case forms are still rote-learned and some two-form contrasts emerge. During the protomorphological period itself, i.e. from 2;0 to 2;8, the percentage of nouns or pronouns occurring in true miniparadigms (13%), two-form oppositions (18%) or in a single case form (70%) does not change much. As far as Andreas is concerned, the main development taking place during the protomorphological phase concerns the growing diversity of cases included in miniparadigms. While in the beginning of this phase mainly grammatical cases and a few local and marginal ones tend to occur in miniparadigms, by the end of the observational period at 2;8, the number of local cases has grown considerably so that 80% of miniparadigms contain one or several local case forms. Also, instances of deviant use of case forms decrease to a considerable degree during the protomorphological phase dropping to only 1.2%. It has been argued by Dressler (1994: 91) that the protomorphological phase is over when “the subsystems of inflectional, derivational, and compounding morphology start to develop”. Since not all case formations have become productive and there is also no sign of the acquisition of derivational morphology at 2;8, the protomorphological phase of the acquisition of morphology cannot be considered to have been completed for Andreas by the end of the observational period.

4. The development of the grammatical category of number Besides the category of case, number constitutes the other grammaticized and thus obligatory category of nouns in Estonian. In this section, the development of number in the pre- and protomorphological phases will be investigated, studying the devices used by Estonian children for referring to plurality as well as the emergence of different plural case forms. As concluded by Stephany (2002: 20) “transparent number marking such as the Turkish agglutinating and the

142

Reili Argus

French analytic technique, seems to be mastered much more easily and become productive more quickly than the nontransparent systems of the typical IndoEuropean languages such as Lithuanian, Greek, or Russian with their complex morphophonemics.” Since Estonian represents both of these techniques, the agglutinating one with most plural case forms and the inflecting-fusional one with complex morphophonemic processes in the plural partitive, it is interesting to investigate which of these techniques is preferred by Estonian children. The first device used to refer to plurality by the children Andreas and Hendrik in the premorphological phase is the bare plural stem (homophonous with the genitive singular) instead of the correct forms carrying a plural suffix. As described above, use of the genitive stem as a polyfunctional device in oblique cases is characteristic of the premorphological phase of Estonian language acquisition. Both children use bare plural stems lacking the plural suffix -d (e.g. erne- for herned ‘pea-’, Andreas 1;7, ratta- for rattad ‘wheel-’, Hendrik 1;8). The first properly suffixed plural forms occur in Andreas’ speech only four months later, at 1;11. Besides suffixless plural stems, rote-learned correct plural partitive and genitive case forms are also among the first devices found for expressing plurality (e.g. lill-i ‘flower-PARTIT:PL’, pois-te ‘boy-GEN:PL’, Andreas at 1;8). In contrast to the developmental sequence found in the singular, the first correct plural partitive forms emerge early while correct plural nominatives appear much later. Not only are the morphophonemic alternations of the partitive plural extremely complex, but often two different plural formations are equally possible with one and the same lexeme. This is why the partitive plural is often considered to be the most difficult case to learn. Besides, there is some inconsistency in the use of different plural partitive suffixes in colloquial Estonian (Keevallik 2003: 362). The early emergence of the partitive plural cannot be explained by input frequency since plural nominatives occur 10 times more frequently in the input. One reason is certainly to be sought in the importance of the semantic functions of the partitive case quite generally. The plural partitive not only fulfills the object function, but it is also important for pragmatic reasons in children’s speech, since it is used for asking for e.g. toys or food. Besides the form lill-i ‘flowerPARTIT:PL’ mentioned above, Andreas’ first rote-learned partitive plural forms are kartule-i-d ‘potato-PL-PARTIT’ and two forms carrying the partitive plural suffix -sid (e.g. maja-sid ‘house-PARTIT:PL’) at 2;0. At 1;11 the proper plural nominative suffix is used in half of all plural forms in Andreas’ speech while, a month later, this percentage rises to 94%. Finally, by 2;1, no more suffixless nominative plural forms are to be found in the boy’s speech. The only two singular-plural contrasts of one and the same lexeme are to be found at 1;11, thus in the premorphological phase of

The early development of case and number in Estonian

143

Andreas’ development, namely plural stem without plural suffix, t¨ahe- (for t¨ahed) ‘star-’ vs. t¨aht ‘star:NOM:SG’ and kotsi- (for klotsid) ‘block-’ vs. ‘klotsi ‘block:PARTIT’, when plural forms do not have any suffix yet, but the child has chosen a correct stem variant for plural forms and uses it in the proper context. Nominative plural is productively used by Andreas at age 1;11, when there are several lexemes occurring in nominative plural, and the same lexemes are also used in other case forms. Partitive plural is acquired a month later, at age 2;0. There are two instances of the stem plural with the formative -e (e.g. klots-e ‘block-PARTIT:PL’) at age 2;0. The agglutinating plural partitive emerges later in Andreas’ speech (e. g. kartule-i-d ‘potato-PL-PARTIT’ 2;0), and agglutinating i-plural forms were productively used only at age 2;4. The early emergence of fusive plural partitives can be influenced by the high frequency on fusive partitive plural forms in the input: 60% of all plural partitive forms in the input were formed by using fusive technique. When both techniques can be possible, parents use only the most common fusional technique. At 2;6, when the number of partitive plural forms increases, two kinds of incorrect uses occur: 1) overgeneralization of the more frequent partitive plural formative -e to the pattern with standard -i (silm-e ‘eye’ instead of silm-i; king-e ‘shoe’ instead of king-i) and 2) contamination of the two partitive plural formatives -eid and -sid (onu-seid ‘uncle-PARTIT:PL’ instead of onu-sid; t¨adi-seid ‘aunt-PARTIT:PL’ instead of t¨adi-sid). No overgeneralizations are found in the formation of the agglutinating i-plural forms. In Hendrik’s data at 2;5 all plural forms are suffixless: e.g. krooni- ‘crown-’ (for krooni-d), silma- ‘eye-’ (for silma-d), numbri- (for numbrid) ‘number-’. All of the nouns in these examples have also occurred in some singular case form earlier in the boy’s development, but these suffixless plural stems cannot be taken to indicate the acquisition of plural by the end of the observational period at age 2;5. Anna’s development of the nominative plural differs from that of the boys in two respects: not only does she use plural forms carrying a suffix from the very beginning (e.g. hamba-d ‘tooth-NOM:PL’, k˜orva-d ‘earNOM:PL’, 1;10), but the number of nominative plural forms occurring in her data is much larger than in the other two children’s speech. However, in Anna’s data, almost all plural nominatives represent the only forms of the respective lexemes and are supposed to be rote-learned, while in the speech of Andreas plural forms also occur in miniparadigms of nouns. In the main child’s data, plural case forms are mainly limited to the nominative and the partitive until the end of the observational period (table 15), while singular case forms cover most of the other Estonian cases, at least from 2;0 onward (see tables 6a and 6b above).

144

Reili Argus

As is to be expected, singular case forms also occur much more frequently than plural ones. At 2;0, when nominative plural forms are very frequently used, the other plural case forms, namely the partitive, genitive, allative and comitative are still rare and rote learned. Emergence and frequency of plural case forms in child speech correspond to input frequencies to a great extent. In the entire recordings of Andreas’ input nominative plural forms occur most frequently both type- and tokenwise, followed by plural partitives, genitives, comitatives and allatives in descending order of frequency. Nominative and partitive plural are also the first forms to emerge in the child’s speech. All plural case forms missing from the input (such as local cases and most of the marginal cases) aren’t found in the child’s speech either. The main developmental steps of the plural of nouns from the premorphological to the protomorphological phase are the following: In the premorphological phase the genitive stem is used as a prototypical oblique case stem instead of the suffixed nominative plural form and furthermore rote-learned plural case forms like plural partitives are to be found. In the protomorphological phase the suffixed plural nominative is the most frequent plural case form becoming productive at 1;11. There are no overgeneralizations to be found with the agglutinating technique of the nominative plural and the partitive plural, but only with the fusional technique of the partitive plural so that it can be assumed that the agglutinat-

Table 15. The emergence of plural case forms of nouns in Andreas’ speech (types/ tokens) Age 1;7 (MLU 1.1) 1;8 (MLU 1.2) 1;9 (MLU 1.5) 1;10 (MLU 1.6) 1;11 (MLU 1.8) 2;0 (MLU 3.6) 2;1 (MLU 2.7) 2;3 (MLU 3.8) 2;4 (MLU 4.1) 2;5 (MLU 4.1) 2;6 (MLU 3.8) 2;7 (MLU 3.9) 2;8 (MLU 4.6)

NOM:PL PARTIT:PL ALL:PL GEN:PL COM:PL 2/2

1/1

1/1 3/3 54/100 17/31 11/16 19/26 6/10 27/42 8/12 11/17

6/6 1/1 2/6 1/1 6/8 6/8 7/7

1/1

1/1

1/1 2/2 2/2

The early development of case and number in Estonian

145

ing technique is mastered more easily than the fusional, although the fusional technique of the partitive plural is used more often by the child. The order of emergence of semantic roles expressed by plural case forms is not exactly the same as with singular case forms: the first plural nominatives express either the subject or total object, the first partitives the partial object and the comitatives an instrument

5. Summary and discussion The acquisition of Estonian noun inflection constitutes a considerable task so that it may take until the second half of the third year to even acquire the very base of the Estonian case-number system: stem alternations and the most important grammatical cases. Only 12 of 14 cases have emerged in the singular and 11 of them have become productive by 2;8. In the plural, only 5 cases have emerged by 2;8 and only 2 of them are productive. At the end of Andreas’ observational period at 2;8 the two marginal cases essive and abessive are still missing from the boy’s speech, but these cases are also rare in the input. The agglutinating part of Estonian inflectional techniques is mastered more rapidly than the fusional part: thus, agglutinating partitive plural forms are acquired before fusional ones, which are not yet productive at age 2;8. Although the agglutinating formation of Estonian case forms is quite regular and is therefore not difficult to master, there are many irregular fused forms of grammatical cases mostly occurring with very frequent lexemes, which must be learned in an item-based way (e.g. the partitive plural). Item-based learning of noun inflection dominates for quite a long period, and many case forms only occur in some lexically specific constructions. Hence it can be argued that the results presented here also agree with a usage-based model of language acquisition (see Tomasello 2003). Our findings that quantity alternations are acquired early in Estonian coincide with Dasinger’s (1997: 31) and are theoretically interesting because they confirm the hypothesis that the “first typological differences emerge already in premorphology” (Dressler 2003: 10). Since case forms may be differentiated solely by grade alternation and the choice between weak and strong grade stems may also form the basis for attaching an appropriate suffix to the proper stem, the ability to distinguish between weak and strong grade stems is very important for mastering Estonian nominal inflection. For this reason, acquisition of the stem variants of nouns must be considered as the most important grammatical achievement of the premorphological phase of Estonian language acquisition. In the premorphological phase of Estonian language acquisition, lexemes are not limited to a single form; quite often a given noun occurs with both its stem

146

Reili Argus

variants, although the child’s choice between different stems is not always correct in a given context. Therefore, as far as the development of noun morphology is concerned, the premorphological phase of Estonian language acquisition is not characterized by limitation to single rote-learned forms of nouns and cannot be sharply distinguished from the protomorphological phase. Rather, the dividing line between these two developmental stages is not clear-cut. The early case forms occurring in the premorphological period with all three children fulfill quite a number of semantic functions such as neutral participant, partial and total object, destination, instrument and companion, possessor and location. Differences between the children are small. These semantic roles, as well as the lexemes and case forms by which they are expressed, are also frequently encountered in the children’s input so that input frequency may be said to play a major role in this developmental period. This is also true of the preference for short suffixless case forms attested in the input as well as the child data in instances where both a short and a suffixed form are possible in Standard Estonian. Local cases emerge in a particular order in the children’s data studied in this chapter: the cases indicating goal-directed movement, like illative and allative, are the first to appear and become productively used. While the source-oriented elative emerges later than the goal-directed cases on the one hand and the stative locative cases inessive and adessive on the other, it nevertheless becomes productive at the same time as the other locative cases. The order of emergence of local cases found in the present study are consistent with the results reached by Vihman (1982) and Dasinger (1997). Following Osterreich (1977: 83), Dasinger (1997: 40–42) explains the later emergence of the elative indicating “Source” as compared to cases expressing “Locative State” and “Goal”, which is characteristic not only of Estonian but also of Finnish and Hungarian, by the fact that as soon as an object has moved away from its source it is no longer the source which is salient for the child but rather the object (or person) moving. Dasinger adds to this that “attention to an object in motion leads to the attendant awareness of the cessation of the movement, and hence the object’s final resting place, or the movement of the object outside of the visual field, and hence its general direction or goal” (p. 42). Aside from pragmatic salience the input frequency and the different numbers of semantic roles of local cases have an impact on their order of emergence: cases indicating goal-directed movement are more frequent in the input, and goal-oriented local cases can express other things besides physical location. Thus, the allative is used for recipient and the adessive for possessor. The emergence of goal-oriented and static local cases is typical for the premorphological period, while source-oriented local cases only emerge in the protomorphological period. Although the three children’s

The early development of case and number in Estonian

147

development of noun inflection studied in the present paper is similar, there is some intersubjective variation with respect to the duration of the premorphological period, emergence of case suffixes and, to some extent, the development of local cases (see sect. 3.1.3). The preference for synthetic vs. periphrastic ways of expressing local cases is not clear-cut: while with certain local relations postpositions precede synthetic forms (exterior local relations), with others it is the other way round. The most relevant explanation can be the child’s intention to differentiate semantic roles of the recipient or possessor expressed by case forms from psychical local relations expressed by case forms. Plural case forms emerge later than singular ones in the children’s speech and their order of emergence differs. Although rote-learned plural partitives emerge earlier than plural nominatives, the latter become productive first. In contrast to what we found in the singular, the genitive plural is not among the first frequently used case forms. While in the premorphological phase the nominative plural is expressed by bare plural stems instead of the appropriate forms carrying a suffix, the nominative plural suffix -d emerges in the protomorphological phase. By the end of the observational period at 2;8, many plural case forms have not yet emerged and the nominative and partitive plural are the only plural case forms to be productively used. The first nouns forming miniparadigms in child Estonian belong to the two main regular and productive inflectional classes (one with weakening stem and the other one without grade alternation), the only exception being the suppletive pronoun see ‘it’. Since these nouns are also frequent in the input, their regularity cannot be the only reason why they preferentially occur in Andreas’ first miniparadigms. Productivity of the case system starts to develop in the protomorphological phase when the child seems to become aware of the functional importance of inflection and morphophonemics. When the child first starts to construct the case system of his native language he prefers more transparent case suffixes to less transparent ways of case formation by grade alternation or to signaling case by syncretism. It is at this point that class shifts are to be observed. While, in the beginning of his inflectional career in the premorphological period, the child uses homophonous forms for the nominative, genitive and partitive, in the protomorphological period he starts to inflect some of these nouns according to the pattern where nominative and genitive are homophonous but the partitive has a regular transparent case suffix. In the protomorphological phase the child thus decreases the number of inflectional classes to be learned by overextending more regular classes to irregularly inflected words. According to Bittner et al. (2003: xix) the protomorphological phase ends when the child’s inflectional system of a given part of speech approaches the adult one so that “basic language-specific properties of target morphology” have

148

Reili Argus

been acquired and the subsystems of inflectional, derivational and compounding morphology are starting to develop. The results of the present study of the acquisition of case and number in Estonian suggest that, at the end of Andreas’ observational period at 2;8, this child’s case system is qualitatively approaching the adult system, since basic language-specific properties of the target morphology like grade alternation and the core of Estonian cases in their principal semantic roles have been acquired. Furthermore, the rate of overgeneralizations of different inflectional patterns amounts to less than 10% of all tokens. The child’s miniparadigms resemble those found in the input as far as their average size and range is concerned. Miniparadigms occurring in Andreas’ speech and the input have 2 to 3 members on average and range from 2 to 7 members (Andreas) vs. 2 to 6 members (input). The only exception are pronouns, which exhibit 6–7 member miniparadigms in Andreas’ data at 2;8. Since not all Estonian cases have developed by this age, the subsystems of derivational and compounding morphology have not yet started to develop and the amount of overgeneralizations does not drop significantly at child’s age 2;8, it can be assumed that the child has not yet entered the phase of morphology proper.

Notes ∗

I would like to thank the editors of the volume, Ursula Stephany and Maria Voeikova, for insightful and helpful comments on earlier versions of this paper. I am grateful to M. M. Vihman, V. Vihman and U. Stephany for improving my English. All remaining inadequacies are my own responsibility. 1. An apostrophe marks strong grade, in cases where the alternation of Q2 and Q3 is not reflected in conventional orthography. 2. The number of productive nouns has been calculated on the basis of the Dictionary of Estonian Language (1976) and demonstrates proportions of each class rather than indicating the exact number of nouns contained in the class.

References Argus, Reili 2004

Eesti keele k¨aandes¨ ¨ usteemi omandamine: esimestest s˜onadest miniparadigmadeni. [Acquisition of case in Estonian: From first words to miniparadigms]. Emakeele Seltsi aastaraamat [Yearbook of the Estonian Language Society] 49: 23–48.

The early development of case and number in Estonian

149

Bittner, Dagmar, Wolfgang U. Dressler, and Marianne Kilani-Schoch 2003 Introduction. In Development of Verb Inflection in First Language Acquisition: A Cross-Linguistic Perspective. Dagmar Bittner, Wolfgang U. Dressler, and Marianne Kilani-Schoch (eds.), vii–xxxvii. (Studies on Language Acquisition 21.) Berlin/New York: Mouton de Gruyter. Blake, Barry J. 2001 Case. Cambridge: Cambridge University Press. Dasinger, Lisa 1997 Issues in the acquisition of Estonian, Finnish, and Hungarian: A crosslinguistic comparison. In The Crosslinguistic Study of Language Acquisition, Vol 4, Dan Isaac Slobin (ed.), 1–86. Mahwah, N.J.: Lawrence Erlbaum Associates. Dressler, Wolfgang U. 1994 Evidence of the first stages of morphology acquisition for linguistic theory: Extragrammatic morphology and diminutives. Acta Linguistica Hafniensia 27 (1): 91–108. 2003 Morphological typology and first language acquisition: Some mutual challenges. In On-line Proceedings of the Fourth Mediterranean Morphology Meeting. Catania. Dressler, Wolfgang U., and Annemarie Karpf 1995 The theoretical relevance of pre- and protomorphology in language acquisition. In Mechanisms of Morphological Change (Yearbook of Morphology 1994). Geert Booij and Jaap van Marle (eds.), 99–122. Dordrecht: Kluwer. Dressler, Wolfgang U., Marianne Kilani-Schoch, and Sabine Klampfer 2002 How does a child detect morphology? Evidence from production. Trends in Linguistics. Studies and Monographs 151: 391–425. Berlin: Mouton de Gruyter. Erelt, Mati (ed.) 2003 Estonian Language. (Linguistica Uralica, Supplementary Series, 1.) Tallinn: Estonian Academy Publishers. Erelt, Mati, Reet Kasik, Helle Metslang, Henno Rajandi, Kristiina Ross, Henn Saari, Kaja Tael, and Silvi Vare 1995 Eesti keele grammatika [Estonian Grammar]. Tallinn: Eesti Teaduste Akadeemia Eesti Keele Instituut. Gagarina, Natalia, and Maria Voeikova 2002 MLU, first lexicon and the early stages in the acquisition of case forms by two Russian children. In Pre- and Protomorphology: Early Phases of Morphological Development, Maria Voeikova and Wolfgang U. Dressler (eds.), 115–131. M¨unchen: Lincom.

150

Reili Argus

Gathercole, Virginia, C. Mueller, Eugenia Sebasti´an, and Pilar Soto 1999 The early acquisition of Spanish verbal morphology:Across-the-board or piecemeal knowledge? The International Journal of Bilingualism 3: 133–182. Hasselblatt, Cornelius 2000 Eesti keele ainsuse sisse¨utlev on l¨uhike [Illative in Estonian language is short]. Keel ja Kirjandus 11: 796–803. Hennoste, Tiit 2004 M˜onede k¨aa¨ nete sagedus ja lauseliikmelisus suulises k˜ones [Frequency of some case forms in spoken Estonian]. In Lauseliikmeist ¨ eesti keeles, Liina Lindstr¨om (ed.), 16–25. Tartu: Tartu Ulikool. Keevallik, Leelo 2003 Colloquial Estonian. In Estonian Language, Mati Erelt (ed.), 343– 379. (Linguistica Uralica, Supplementary Series, 1.) Tallinn: Estonian Academy Publishers. Laalo, Klaus 1994 Kaksitavuvaihe lapsen kielen kehityksessa [Trochaic stage in language acquisition]. Viritt¨aj¨a 3: 430–448. Metslang, Helle 2001 On the development of the Estonian aspect: The verbal particle a¨ ra. In ¨ Circum-Baltic Languages, Osten Dahl and Maria Koptevskaja-Tamm (eds.), vol. 2, 443–479. Amsterdam: John Benjamins Publishing Company. Osterreich, Helgi 1977 The acquisition of Estonian locative cases. Ph. D. diss., Department of Linguistics, University of New Mexico. Pizzuto, Elena, and Maria Cristina Caselli 1994 The acquisition of Italian verb morphology in a cross-linguistic perspective. In Other Children, Other Languages: Issues in the Theory of Language Acquisition, Yonata Levy (ed.), 137–187. Hillsdale, NJ: Lawrence Erlbaum Associates. Pajusalu, Karl 2001 Baby talk as a sophisticated register: A phonological analysis of South Estonian. Psychology of Language and Communication 5 (2): 81–92. Slobin, Dan Isaac 1985 Crosslinguistic evidence for the language-making capacity. In The Crosslinguistic Study of Language Acquisition. Dan I. Slobin (ed.), Vol. 2: Theoretical Issues, 1157–1256. Hillsdale, NJ: Lawrence Erlbaum Associates. Stephany, Ursula 2002 Early development of grammatical number: A typological perspective. In Pre- and Protomorphology: Early Phases of Morphological

The early development of case and number in Estonian

151

Development in Nouns and Verbs, Maria D. Voeikova and Wolfgang U. Dressler (eds.), 7–23. M¨unchen: Lincom Europa. Tauli, Valter 1980

Eesti keele grammatika II. Lause˜opetus [Estonian Grammar II. Syntax]. Uppsala: Finsk-ugriska institutionen. Tomasello, Michael 2003 Constructing a Language. A Usage-Based Theory of Language Acquisition. Cambridge, MA: Harvard University Press. Vihman, Marilyn May 1982 The acquisition of morphology by a bilingual child: A whole-word approach. Applied Psycholinguistics 3: 141–160. Vihman, Marilyn May, and Benedicte Boysson-Bardies 1991 Adaptation to language: Evidence from babbling and first words in four languages. Language 67 (2): 297–319. Vihman, Marilyn May, and Maigi Vija 2006 The acquisition of verbal inflection in Estonian. In The Acquisition of Verbs and their Grammar: The Effect of Particular Languages, Natalia Gagarina and Insa G¨ulzow (eds.), 269–295. (Studies in Theoretical Psycholinguistics 33.) Dordrecht: Springer. Vija, Maigi 2007 Pronoomenid lapsekeeles: m˜onda mina ja sina omandamisest [Pronouns in child language: some fragments of a study of the acquisition of pronouns mina and sina by Estonian children]. Rakenduslingvistika aastaraamat [Yearbook of Applied Linguistics], 373–384. Viitso, Rein 2003 Structure of Estonian language. Phonology, morphology and word formation. In Estonian Language, Mati Erelt (ed.), 9–129. (Linguistica Uralica, Supplementary Series, 1.) Tallinn: Estonian Academy Publishers.

The acquisition of case, number, and gender in Croatian∗ Melita Kovaˇcevi´c, Marijan Palmovi´c, and Gordana Hrˇzica 1. Introduction Croatian belongs to the family of South-Slavic languages along with Slovenian, Serbian, Macedonian, and Bulgarian. It is spoken from, roughly, the river Drava to the Adriatic Sea (excluding parts of Bosnia & Hercegovina) and from the river Sutla on the Croatian-Slovenian border to the Danube in Slavonia, on the ˇ Croatian-Serbian border. The three major dialects, Kajkavian, Cakavian and ˇ Stokavian, are classified on the basis of the interrogative pronoun ‘what’. Croatian is a highly inflected language in which grammatical morphemes usually express three grammatical categories (e.g. case, number, and gender for nouns and adjectives and person, number and tense for verbs). The canonical word order is SVO, but overall word order is relatively free. Declensional classes are based on three gender classes. In addition there are different declensional classes across genders. Verbs are subdivided into six conjugational classes obeying different paradigmatic patterns. In addition, masculine paradigms differ for animate and inanimate nouns. Masculine nouns also distinguish between plural and paucal number (two + N, three + N, four + N) while in the feminine and neuter genders these differences are blurred by syncretism. Derivational morphology plays an important role in the Croatian lexicon and grammar, enhancing the transparency of lemmas as well as their formal and semantic relationships. Thus, the root bol ‘pain’ serves as the base of about a dozen of derived lexemes in Croatian (e.g. bolnica ’hospital’, bolest ‘illness’, boljeti ‘hurt’, bolniˇcarka ‘nurse’, etc.), while their English equivalents show no derivational connection. This derivational transparency could certainly affect the process of the acquisition of Croatian lexicon and the morphological structure of this language. In this paper we will focus on the development of case, number, and gender in early Croatian child language. The core of our arguments will focus on the question in how far the morphological richness of Croatian may promote the early development of inflection. A language with a complex morphological system, such as Croatian will impose different demands on the young child

154

Melita Kovaˇcevi´c, Marijan Palmovi´c, and Gordana Hrˇzica

than less highly inflected languages such as German or languages with a regular inflection such as Turkish.

2. The inflectional system of the Croatian noun 2.1.

Case

The inflectional system of the Croatian noun has two numbers, singular and plural, and seven cases in each number: Nominative (Nom.), Accusative (Acc.), Genitive (Gen.), Dative (Dat.), Locative (Loc.), Instrumental (Instr.), and Vocative (Voc.).

2.1.1. Case functions Although in traditional Croatian grammars a large number of various meanings is given for most of the seven cases (e.g. Bari´c et al. 1997; Raguˇz 1997; Teˇzak and Babi´c 2000), each of them has one or a few prototypical functions. Thus, the Nominative is the basic case form used for labeling. All other cases except the Vocative express different relationships within sentence structure. The prototypical meanings of cases which express syntactic-semantic relationships within the structure of sentences are the following: the Nominative marks the subject or the predicative noun, the Accusative signals the direct object and is also governed by certain prepositions in phrases functioning as the goal of an action. The Dative expresses the indirect object of intransitive verbs and may function as recipient or goal. The Locative is used in prepositional phrases functioning as a complement of a verb. The Instrumental signals a device or company. The Genitive is the case with the greatest number of roles. The most important of these are possessor, expression of an unavailable entity (missing object as in e.g. ‘there is no . . . ’), and the part-whole relation (genitivus partitivus). In their prototypical meanings, the Accusative represents object and goal (only with certain prepositions), the Dative represents recipient and goal (usually without prepositions), the Locative is used to denote topic and location and is always governed by a preposition, and the Instrumental represents means and company. The Vocative case is used for calling or addressing someone. This form is disappearing from the Croatian morphological system, being substituted by the Nominative with most nouns. Its usage is mostly limited to fossilized forms of certain nouns and some proper names (e.g. gospoda ‘madame’ → gospodo!, profesor ‘professor’ → profesore!).

The acquisition of case, number, and gender in Croatian

155

2.1.2. Prototypical declensions Each of the basic gender classes has its own prototypical declension. Thus, the prototypical declension of feminine nouns ending in –a is known as the e-declension (traditional grammars name the declensions after the Genitive singular ending). The prototypical masculine declension is the a-declension, which differentiates between animate nouns (Acc. singular ending –a, homophonous to the Genitive) and inanimate ones (Acc. singular = Nom. singular). Neuter nouns are declined in the same way as masculine inanimate nouns.

2.1.3. Other declension patterns Although the majority of the most frequent nouns belong to the prototypical declension class, some frequent nouns do not. Masculine nouns ending in a vowel do not always take the prototypical declension pattern.Those ending in –a follow the e-declension. Some nouns ending in other vowels also follow the e-declension paradigm, whereas the others stick to the prototypical a-declension: proper hypocoristics (some of them serve as proper names, like Ante or Ivo) follow the e-declension (e.g. sluga ‘servant’ or given names such as Borna, Matija). Some masculine nouns ending in other vowels also belong to the e-declension such as hypocoristics of proper names. Most masculine proper names and foreign words follow the a-declension. The rather small group of feminine nouns ending in a consonant (e.g. no´c ‘night’, ljubav ‘love’) belongs to distinct class, the i-declension.

2.1.4. Additional rules in e-declension Some feminine nouns ending in a vowel -a plus a velar stop k, g, or the glottal fricative h palatalize the consonant to the dental affricate of fricatives c, z, and s, respectively, before the ending –i of the Dative and Locative singular. This morphophonological change is not limited to feminine -a nouns but occurs in some other nouns as well. Feminine nouns ending in –ica form the Vocative by –e instead of –o. Although, as stated above, the Vocative is gradually being lost, it is more often used with this group of nouns than with other feminine nouns. There are also some rules affecting the Genitive plural. If the noun has two consonants before final –a, a mobile vowel –a– is interpolated (e.g. sestra ‘sister’ – nemam sestara ‘I have no sisters’), or the usual ending –a is replaced with the alternative –i (e. g. sekunda ‘second’ – 5 sekundi ‘five seconds’).

156

Melita Kovaˇcevi´c, Marijan Palmovi´c, and Gordana Hrˇzica

2.1.5. Additional regularities of the masculine a-declension In masculine nouns with the palatal final consonant (ˇc, c´ , sˇ, zˇ, d, dˇz, lj, nj, j; sometimes also r and c), the Instrumental ending –om is changed to –em, and the Vocative ending –e to –u.

2.1.6. Additional rules in the neuter a-declension If the neuter noun includes a palatal (polje ‘field’), a palatal group (igraliˇste ‘playground’), or a consonant that shares some morphophonological features of palatals (r, c – e.g. sunce ‘sun’, more ‘sea’), their endings in the Nominative and Instrumental are –e and –em rather than the default endings –o and –om. There is a group of nouns that do not follow this rule. They end in –e, but without a palatal preceding it. Instead, they have an interfix in front of the case ending in the Genitive, Dative, Locative, and Instrumental singular and in all cases in plural (e.g. ime ‘name’, djeˇcak bez im-en-a ‘a boy without a name’). Some other neuter nouns that follow this rule also have interfixed forms. Possible interfixes are –en–, –et– (mostly for the young of humans or animals, e.g. tel-e ‘calf’ – nema tel-et-a ‘there is no calf’) and –es– (doublets – nouns that can be declined with or without this interfix). A mobile vowel –a– is inserted in the Genitive plural in neuter nouns with two consonants in front of the final –a– (e.g. pism-o ‘letter’ – nema pis-a-ma ‘there are no letters’).

2.2.

Number

There are two categories of number in Croatian, singular and plural. The plural is formed on the basis of the prototypical classes and declensions described above. Feminine –a nouns take the suffix –e (djevojˇcica ‘girl’ → djevojˇcice ‘girls’), masculine –C nouns take –i (djeˇcak ‘boy’→ djeˇcaci ‘boys’), while neuter nouns take –a (vino ‘vine’ → vina ‘vines’).

2.2.1. Short and long plural Some masculine nouns in Croatian can have an interfix when forming plural. This is known as the ‘long’ plural form by contrast to the ‘short’ plural. The long plural is formed with the interfix –ov–/ –ev– placed in front of the plural suffix. The choice of the interfix is phonologically conditioned (-ev- is used after a palatal). Monosyllabic nouns usually have the ‘long’ plural although there are

The acquisition of case, number, and gender in Croatian

157

a few very frequent monosyllabic ones with a ‘short’ plural form. Examples of long and short plural form are the following: slon ‘elephant’ → slon-ov-i pingvin maˇc ‘sword’ → maˇc-ev-i

‘penguin’ → pingvin-i

2.2.2. Other plural forms and collective nouns Some Croatian count nouns may form collectives, thereby expressing uncountable quantities, e.g. cvijet ‘flower’→ cvjetovi ‘flowers (count)’→ cvije´ce ‘flowers (coll.)’.

2.2.3. Pluralia tantum Croatian also possesses pluralia tantum which, although being semantically singular, have a plural declension (e.g. usta ‘mouth’, hlaˇce ‘trousers’).

2.3.

Gender

Since the suffixes of Croatian nouns are gender dependent, nominal gender is predictable. There are three genders in Croatian: feminine, masculine, and neuter. In the majority of cases, the declensional pattern is determined by gender. The three genders are not evenly distributed: masculine and feminine nouns taken together make up 86% of all Croatian nouns (43.3% masculine; 42.9% feminine), while neuter accounts for only 13.7%. These statistics are based on the analysis of the 4000 most frequent noun lemmas in the adult spoken language compiled from one million tokens and a total of 114.717 noun lemmas. (Vuleti´c 1991).

3. Data The analysis of the present paper is predominantly based on the speech of the Croatian girl Antonija between the ages 1;3 and 2;8. Data from the other two children of the Croatian child language corpus (http://childes.psy.cmu.edu/data/ slavic/), which consists of the transcripts of three children, Antonija, Vjeran, and Marina, whose speech was recorded between their first and third birthdays, has also been included in the statistics for additional analysis and comparisons. The entire data has been transcribed according to the CHILDES Project (MacWhinney 2000).

158

Melita Kovaˇcevi´c, Marijan Palmovi´c, and Gordana Hrˇzica

Table 1. An outline of Croatian noun inflection a. General categorization e-declension feminine –a djevojˇcica ‘girl’ masculine –a sluga ‘servant’ –a (names) Matija, Luka –e/-o (hyp.) Ivo (from Ivan) neuter

i-decl. –C no´c ‘night’

a-declension

-C djeˇcak‘boy’ –o, –e (names) Mario, Hrvoje –e/–i (foreign words) ataˇse ‘attach´e’, taksi ‘taxi’ –o (stem-final l) pepeo ‘ashes’ –o/–e nebo ‘sky’, more ‘sea’

b. Additional rules e-declension feminine k, g, h + i → c, z, s ending –ica → Voc. –ice

a-declension masculine neuter animate or inanimate infixes -et-, -en-, -esalternative Instr. and Voc. alternative Instr. and Nom. endings of stems ending in endings of stems ending in a palatal a palatal mobile a sometimes appear- k, g, h + e → cˇ , zˇ , sˇ ing in Gen.Pl. alternative ending i somek, g, h + i → c, z, s times appearing in Gen.Pl. mobile -a- sometimes appearing in Gen.Pl.

Target child. Antonija’s parents are upper middle class urban dwellers; her mother is a university teacher and her father an engineer. Since both of them ˇ were born and raised in Zagreb, they speak the Zagreb Stokavian dialect which ˇ is a sort of koine of Stokavian and Kajkavian. A somewhat stronger influence of Kajkavian (spoken in the environment of Zagreb) can be noticed especially when Antonija talks with her grandparents. It neglects many of the morphophonological processes imposed by the standard language (which are, however, observed by the speakers of the urban language in Zagreb). It is therefore often difficult to decide whether the occurrence of such processes or their absence in the child’s language are merely reflections of the input language or they rather stem from the child’s own construction of the language. The other children. The other two children have similar family backgrounds. Vjeran’s mother is an English teacher, and his father an electrical engineer.

The acquisition of case, number, and gender in Croatian

159

Table 2. Antonija’s data Age No. of utterances (Antonija) 1;3 44 1;4 85 1;5 42 1;6 64 1;7 273 1;9 197 1;10 303 1;11 150 2;0 207 2;1 334 2;2 483 2;3 584 2;4 586 2;5 644 2;6 362 2;7 428 2;8 371

No. of utterances (input) 163 238 97 98 547 304 529 212 327 508 726 811 964 980 665 701 691

No. of word tokens (Antonija) 60 87 50 91 386 481 646 354 561 892 1360 1578 1586 1737 954 1151 904

No. of word tokens (input) 555 1039 492 341 1941 1198 2120 1016 1501 2367 3142 3520 3886 3826 2654 2674 2750

MLU in words (Antonija) 1.36 1.02 1.19 1.42 1.41 2.44 2.13 2.36 2.71 2.67 2.81 2.70 2.70 2.69 2.63 2.68 2.44

ˇ They also both speak the Zagreb Stokavian dialect. Since the boy’s babysitter is ˇ from the district of Lika (and speaks a Stokavian dialect) Vjeran is less exposed to the Kajkavian dialect than Antonija. Marina’s parents also live in Zagreb. Marina’s mother is a university teacher and her father is an actor. Since her father is an actor, his speech is rather close to ˇ the standard, in spite of the fact that he grew up in Zadar, a Cakavian-speaking area. Analyzed material and method. Antonija’s data are transcripts of audio-recordings collected from 1994 to 1996 in Zagreb. The child was recorded in her home in spontaneous interactions with her parents or grandparents. Each recording session lasted about 45 minutes; there were usually three sessions per month which amounts to 42 recording sessions altogether. For the purposes of the present study, the recordings of each month have been grouped into monthly sets (see table 2). Due to personal reasons, no data could be gathered at the age of 1;8. The speech addressed to the child is considered to constitute the input language and has been collapsed across all adults participating in the recordings. Conversations were usually related to everyday situations, such as eating, play-

160

Melita Kovaˇcevi´c, Marijan Palmovi´c, and Gordana Hrˇzica

ing with toys, commenting on people in Antonija’s environment, etc. Formulaic expressions extracted from songs or books, as well as immediate imitations, have been excluded from the analyses. The data were analyzed in a computer-assisted way within the CHILDES project (MacWhinney 2000).

4. Main objectives The analysis was based on the growth of noun vocabulary as well as morphological development. Parents’ data are used for comparison. The analysis was constrained to the inflectional morphology. The child’s (or children’s) data were compared with the input language. Seven cases, three genders, two numbers, three noun classes, and a large number of different paradigms can be viewed as a tricky acquisition task for a child acquiring Croatian (Jelaska et al. 2002). The main aim was to gain better insight into the impact of language’s morphological complexity on the time course of the emergence of particular categories, namely, case, number, and gender within the noun system. The main objectives of our study may be defined by following four questions: (a) Is morphological richness in language a trigger or an obstacle for early acquisition? (b) Could morphological richness be an argument for morphological preferences in the course of early language acquisition (c) Are there functional preferences in case acquisition? (d) How much does the topic shared between child and adult result in shared lexicon and morphological features?

5. Results and discussion 5.1.

Gender

Since the inflection of the Croatian noun is largely based on gender, it is important to consider the distribution of the three genders in Antonija’s speech. The distribution of Antonija’s early masculine, feminine, and neuter reflects the distribution of Croatian nouns in child-directed speech quite well (table 3). There is a larger number of feminine than masculine nouns (58% feminine, 36% masculine, figure 1) in the child’s speech which can partially be explained in terms of the context of conversation (e.g. talk about Antonija’s numerous dolls). However, another possible explanation is that the transparency of the feminine noun paradigm (see below) may influence Antonija’s readiness to acquire such nouns. The absolute number of neuter nouns is relatively small throughout the

The acquisition of case, number, and gender in Croatian

161

Table 3. The distribution of gender classes in child speech and input (number of tokens) Age

MASC Antonija 1;3 6 1;4 11 1;5 4 1;6 4 1;7 37 1;9 26 1;10 51 1;11 35 2;0 23 2;1 31 2;2 67 2;3 79 2;4 67 2;5 65 2;6 48 2;7 47 2;8 26 Total 627

FEM Antonija 17 18 6 6 65 20 39 83 32 61 140 123 65 135 63 107 52 1032

NEUT Antonija 1 8 0 0 1 1 11 2 3 13 13 12 19 9 7 11 1 112

MASC Input 16 56 7 7 113 88 111 76 56 78 151 197 215 117 129 124 79 1620

FEM Input 85 135 34 34 222 64 200 141 124 242 311 314 277 385 263 275 230 3336

NEUT Input 4 25 4 4 4 8 36 14 11 24 33 34 46 16 16 27 20 326

whole period in both input and child language and corresponds to their small frequency in language in general. Moreover, the majority of them cannot be considered to belong to child language. Neuter nouns occur only in the context of food and drink and are restricted to only a few lemmas. In addition they are frequently diminutivized, but instead of standard suffixes, dialectal suffixes are used by Antonija (Palmovi´c 2007).

Figure 1. Distribution of gender in Antonija’s speech (left) and input (right) (% of tokens)

162

Melita Kovaˇcevi´c, Marijan Palmovi´c, and Gordana Hrˇzica

Before turning to the details of the development of case-number inflection of nouns of the three genders, the child’s overall development of declension may be described by determining the relation between the number of noun lemmas and grammatical form types of nouns. The development of Antonija’s case system can be viewed from the relation between the number of noun lemmas and noun types (this will be discussed in the next section). While the number of different lemmas reflects Antonija’s vocabulary, the number of different types points to the existence of the same lemmas in different cases. Figure 2 shows the increase in the number of noun lemmas, but an even faster increase in the number of noun types. This corresponds to an increasing number of noun lemmas in oppositions and mini-paradigms. A comparison of the lemma curves and the type curves of the child data with those of the input depicted in figure 2 is quite instructive since it shows that the child’s speech and that of the adults are tuned to each other. Although the raw numbers in the input exceed those in the child data, the general development of the curves is similar, with the exception of the initial period, where the two child curves overlap. Such a phase is absent from the input data so that the lexical and morphological complexity of the latter can be said to precede the child’s developmental phases. The overlapping lines in the first months of data collecting in Antonija’s data (fig. 2, left) clearly indicate premorphological stage. In the same period the input data (right) indicate somewhat simplified child-directed speech.

Figure 2. Relation between noun lemmas and grammatical form types. Left: Antonija, right: input

The acquisition of case, number, and gender in Croatian

5.2.

163

Oppositions, mini-paradigms, and paradigms

Oppositions, mini-paradigms and paradigms in child’s language may be considered as a reliable indicator for the acquisition of noun inflection. They can also be viewed as the evidence for the productive use of morphological markers and outline the developmental pattern. The first oppositions appear as early as at 1;6. As shown in table 4, by the end of the recording period, five-member paradigms are present in her speech. The first oppositions of inflected forms contrast the Nominative with the Accusative and are found in feminine nouns belonging to -e class (e.g. lopt-a ‘ballNomSg’ lopt-u ‘ball-AccSg’). Since the masculine nouns occurring all end in a consonant (e.g. loptaˇc ‘ball’, nos ‘nose’, gol ‘goal’ sˇtapi´c ‘stick’ (diminutive)), there is syncretism between the Nominative and the Accusative and, as the child is in the two-word-utterance phase, it is still difficult to grasp the function from the context. Accordingly, for masculine nouns in this early period the Nominative and the Accusative are hard to distinguish. Case markers of feminine nouns are therefore easier for a child to detect due to more transparent paradigms with a low degree of syncretism (different forms for different functions). Therefore, even before Antonija’s second birthday several mostly feminine nouns can be found in three- and four-member mini-paradigms that consist of Nom. ∼ Gen. ∼ Acc. ∼ Instr. and Nom. ∼ Dat. ∼ Acc. ∼ Instr. as in the following example (Antonija, 1;10): Table 4. Cumulative number of nouns in newly produced oppositions and miniparadigms (Antonija) Age 1;6 1;7 1;9 1;10 1;11 2;0 2;1 2;2 2;3 2;4 2;5 2;6 2;7 2;8

MLU (words) 1.42 1.41 2.44 2.13 2.36 2.71 2.67 2.81 2.70 2.70 2.69 2.63 2.68 2.44

2 members 3 9 13 22 32 35 43 47 60 69 78 81 90 91

3 members

4 members

5 members

2 2 3 4 6 6 7 9 10 12 14

1 2 2 2 3 3 3

1 6 9 12 13 17 21 25 33 37 45 47

164

Melita Kovaˇcevi´c, Marijan Palmovi´c, and Gordana Hrˇzica

bak-a bak-e bak-u bak-om ‘granny-Nom’ ‘granny-Gen’ ‘granny-Acc’ ‘granny-Instr’ (e.g. ide baka (Nom.) ‘granny comes’, nema bake (Gen.) ‘there is no granny’, zovi baku! (Acc.) ‘call granny!’, idem s bakom (Instr.) ‘I go with granny’). The only masculine nouns with such early mini-paradigms are the e-class masculine nouns tata ‘daddy’ and deda ’grandfather’, which actually have the same paradigm as feminine nouns. Some nouns occur only in oblique cases, i.e there is no opposition between the Nominative and Accusative, but, e.g. Genitive and Accusative. These nouns are usually used in limited contexts; thus, the partitive Genitive is often used while talking about eating and drinking (e.g. Hocu sok-a (Gen.) ‘I want juice’, Nema cˇ aj-a (Gen.) ‘There is no tee.’), while locations (e.g. du´can ‘shop’ and ku´ca ‘house’) occur in theAccusative, Genitive and Locative (e.g. Idemo u du´can (Acc. = Nom.) ‘Let’s go to the shop’, but Tata je u du´can-u (Loc.) ‘Daddy is in the shop’ or kraj ku´c-e (Gen.) ‘near the house’) depending on the preposition. The question is how to interpret this finding. In the absence of the child’s production of all possible case forms of a given lemma, we cannot be sure that she is in command of entire paradigms. In the theoretical framework of usagebased theories of language and language acquisition (Bybee 2001; Tomasello 2003), a likely interpretation of our data is that full paradigms and more abstract schemas gradually develop from individual grammatical types of lemmas in the course of language development. Since schemas are based on specific lexical items it is not surprising that we do indeed find a single lemma used in all possible case forms (of the singular) at 2;2. Starting at 1;10, we find all different case forms distributed across different lemmas, which means that, in the early phases of inflectional development, different lemmas fulfill different functions in syntactic contexts. Since we are here concerned with the earliest stages of the acquisition of Croatian, the development of complete paradigms and more abstract schemas will continue way beyond 2;8, when our observation stops.

5.3.

Distribution and functions of case-number forms

As it can easily be predicted, the percentage of Nominative decreases with the child’s age and a more diverse distribution emerges. Figures 3 and 4 provide more details of Antonija’s early case system with two time points (1;7 and 2;5) being taken as two language profile representatives. Before 1;7 the number of noun tokens per month was relatively small (between 10 and 30). Antonija produced her first words shortly before 1;3, and during the first months of recording there was a lot of babbling still present in her

The acquisition of case, number, and gender in Croatian

165

Figure 3. Distribution of case-number forms – Antonija 1;7

Figure 4. Distribution of case-number forms – Antonija 2;5

speech. Naming and calling were the first functions of her (one-word) utterances. It is at 1;7 that we find a number of different case-number forms of nominal lemmas (figure 3). These are not evenly distributed, with the Nominative-Singular amounting to 64% of tokens. However, as stated in the previous paragraph, part of the production of Vocative forms is preserved only in the functional sense (established by the context) and not marked by morphological markers. The Accusative (14%) is used when referring to objects or putting something onto an object, e.g.: na stol on table:Acc.Sg. ‘onto the table’ For many of Antonija’s early nouns these three cases are homophonous, and only the context tells us whether a noun is in the Accusative, Nominative, or

166

Melita Kovaˇcevi´c, Marijan Palmovi´c, and Gordana Hrˇzica

Vocative case. The Genitive is restricted to a few common nouns: cˇ aj ‘tea’, sok ‘juice’, mlijeko ‘milk’, and a few proper names including tata ‘daddy’, deda ‘grandfather’, or baka ‘grandmother’, always occurring with the two verbs piti ‘drink’ or nema ‘not have, not be present’ as in the following examples: piti cˇ aj-a drink tea-Gen.Sg. ‘drink tea’

nema tat-e not have daddy-Gen.Sg ‘daddy is not here’

Since the Genitive is restricted to a few words; its frequency reaches only 1%. If we subtract the percentage of Genitives from figure 3 and bear in mind that Nominative, Accusative and Vocative forms are often homophonous, the majority of the first noun lemmas occur in only one form. This implies that at 1;7 Antonija is still in the pre-morphological phase. However, it should be noted that even at that early point, when the child is still producing predominantly one-word utterances, there are 9 oppositions which clearly confirms very early morphological marking. Emergence of early oppositions could be certainly attributed to the typological features of Croatian, namely, its morphological richness. Their early presence in the child language still does not indicate the full acquisition of case morphology, but rather points to the process of the gradual construction of the grammatical system of the language being acquired. If we choose a much later point in the data, e.g. 2;5, when the child is producing predominantly three-word utterances, we can see that the distribution of nouns looks different (figure 4). This distribution of cases is partially caused by the production of the first syntactic structures. While the percentage of the Vocative drops, the percentage of the Nominative decreases to 47% (from 64% at 1;7) and the percentage of the Accusative increases (30% from 14% at 1;7). The percentage of the Genitive increases due to the fact that it has assumed the function of expressing a possessive relation besides being governed by certain verbs. Although it is strictly speaking ungrammatical (but occurs in the input), the construction od + Gen. ‘from + Gen.’is the first way of expressing possession in Croatian child language (Kuvaˇc, Palmovi´c 2002). In adult language, Genitive can be used for expressing possession only in certain syntactic contexts in which the possessor precedes the possessed noun acting as a modifier (Matasovi´c 2000): Mark-ov-a ku´c-a Marko-Adj-Nom.Sg. house-Nom.Sg. ‘Marko’s house’

The acquisition of case, number, and gender in Croatian

167

ku´c-a kralj-a Mark-a house-Nom.Sg. king-Gen.Sg. Marko-Gen.Sg. ‘the house of king Marko’ However, the Genitive seems to be a “default” way of expressing possession since children stick to the ungrammatical od + Gen. phrase even at school (note that this phrase is not colloquial Croatian). The percentage of Accusatives has increased at 2;5 as compared to 1;7 due to the greater number of sentences with transitive verbs. Since Croatian is a null-subject (pro-drop) language and the subject is accordingly often omitted, the number of Accusative tokens (and that of other case forms) is relatively high as compared to languages where the Nominative plays an important role in the core structure of the sentence. By the end of the first half of her third year, Antonija uses the Accusative with a number of different verbs to express a variety of situations. Another important function of the Accusative is to express direction, i.e. dynamic locative relations. At 2;5 Antonija commonly constructs phrases with the prepositions na ‘at’ and u ‘in’ with Accusative, but not yet with Locative. The explanation is thatAntonija’s early utterances relate to actions (“events”) (Nelson 1985) so that she expresses direction rather than location. The expression of goal-directed, dynamic locative expressions thus precede static indications of location (such as the location of an object in a box or on a table, for example) in Antonija’s development; the latter occur later. An overall comparison of the distribution of cases in input and child language (figures 3, 4, 5, and 6) shows that the distribution of cases in the input language remains similar at 1;7 and 2;5 while the emergence of the cases in child language increases and the distribution of cases becomes more comparable to the input language. While in the input language the Accusative, Genitive and Vocative appear with a similar frequency at both points, the appearance of the Nominative slightly decreases (53% vs. 46%). Indeed, in child language the Accusative appears twice as much at 2;5 than at 1;7, there is a great increase of the Genitive (1% vs. 8%) while the percentage of Vocative case decreases noticeably (21% vs. 3%). The input language seems to be to a certain degree intuitively tuned to the linguistic abilities of the child at the respective points of development. First, due to numerous instances of pointing and naming various objects in the child’s environment which document the use of a referential style on the part of the adults (Nelson 1973), the Nominative occurs more frequently at 1;7 than at 2;5. Second, the prevalence of same cases in both input and child language is evident (Nominative, Accusative, Genitive, Vocative).

168

Melita Kovaˇcevi´c, Marijan Palmovi´c, and Gordana Hrˇzica

Figure 5. Distribution of cases – input 1;7

Figure 6. Distribution of cases – input 2;5

On the other hand, the occurrence of certain case forms in child language at an early age does not mean that all of their functions have been acquired. On the contrary, usage is restricted to only some of the functions, or even more often, restricted by the situational and linguistic context (as would be predicted by Usage-Based Theories).

The acquisition of case, number, and gender in Croatian

169

Figure 7. Order of emergence of singular case forms in Antonija’s corpus

5.4.

Order of emergence of case forms

The seven cases of Croatian do not emerge all at once in child Croatian; rather, their appearance demonstrates a certain order, not only in our main subject Antonija, but also in the other two children studied in this respect (figures 8 and 9).The first case forms to occur are the Nominative and the Vocative. When used in early one-word utterances, neither of them expresses syntactic relations. Since the Vocative is most often homophonous with the Nominative the addressing function of the former and the labeling function of the latter must be told apart by context (see above). The first nouns marked for case emerge at 1;4. The order of their emergence is the Accusative (1;4) followed by the Genitive (1;5), the Dative (1;9), the Locative (1;9) and finally the Instrumental (1;10). Thus, all seven Croatian cases have emerged by 1;10 in Antonija’s speech (see figure 7). Note that the figures 7, 8, 9 represent cumulative data, i.e. that the earliest time point in which more than one lemma occurs in a certain case is taken for the time point for the emergence of that case; this case is then represented in all time points (months) that follow. For example, although the Instrumental is missing from the data in 2;5 (see figure 4), it emerges at 1;10 and remains on figure 7 in the months following this point. When the order of emergence of cases in Antonija’s data is compared to that of the other two children from our corpus, Marina and Vjeran, it can be seen that the order is mainly the same for the three children. With Marina and Vjeran, some of the core cases first occur in one and the same month (figures 8 and

170

Melita Kovaˇcevi´c, Marijan Palmovi´c, and Gordana Hrˇzica

Figure 8. Order of emergence of singular case forms in Marina’s corpus

Figure 9. Order of emergence of singular case forms in Vjeran’s corpus

9). This leads us to assume that there is a preferred order of case acquisition that is determined by typological features of the target language. However, in order to draw firmer conclusions in this respect additional analysis of syntactic structures would be needed. The overall course of the acquisition of the case system, as described above, does not imply that all nouns will follow the same order. Case forms which

The acquisition of case, number, and gender in Croatian

171

are pragmatically important will develop earlier. The question which cases are communicatively more important than others largely depends on the semantic classes to which nouns belong. That means that some nouns will appear in all seven cases while others (used only in certain situations) can appear in less, even in only one case. For example, the noun godina ‘year’ is found in Antonija’s corpus only in the Genitive singular (godine) and Genitive plural (godina) (from age 2;3). The reason is that its most common use is with numbers, which in Croatian are constructed with nouns in the Genitive. The Genitive singular is used with the numbers 2, 3, 4, for example: Imam tri godin-e I have three year-Gen.Sg. ‘I’m three years old’ The Genitive plural is used for other numbers, as in Ima deset godin-a she has ten year-Gen.Pl. ‘she’s ten years old’ There are several other examples of nouns appearing only in the Genitive, all of them denoting food and drinks. Nouns that denote places, such as ku´ca ‘house’, du´can ‘store’, and plac ‘street market’, appear in Antonija’s corpus only in cases expressing place and goal, that is in the Locative (place), Dative (goal), and Accusative (goal, with certain prepositions). The Nominative is thus not the first case form to be acquired of all nouns.

5.5.

Plural

The Plural is rare in Antonija’s speech and only accounts for about 8% of all noun tokens (147/1780) of a small number of lemmas in the entire corpus. Most of these, particularly those found at the earlier ages, commonly occur in the plural rather than singular in the input, as well (e.g. cipele ‘shoes’, suze ‘tears’, noge ‘legs’, sˇlapice ‘slippers’, smoki´ci ‘chips’, pletenice ‘braids’). Later on, nouns commonly used with both numbers also occur in the plural (e.g. lutke ‘dolls’, bojice ‘crayons’, kolaˇci ‘cookies’), but, again, they denote objects in child’s surroundings that are usually grouped together (e.g. a jar of cookies, a box of crayons). The first opposition between singular and plural forms occurs at 2;1: sˇlapic-a sˇlapic-e slipper-Nom.Sg. slipper-Acc.Pl.

172

Melita Kovaˇcevi´c, Marijan Palmovi´c, and Gordana Hrˇzica

More oppositions between singular and plural forms can be found at 2;3 (4 nouns) and 2;5 (3 nouns). Because of the low frequency of plural case forms, it is hard to determine the development of case distinctions in the plural. As previously stated, while case forms in the singular are regularly used after they have first emerged, this does not apply to cases in the plural in Antonija’s data. Also, Plural cases do not follow the same order of emergence as singular cases. Taken together with the scarcity of data, this does not allow us to come to any justified conclusions as to whether the appearance of a certain plural form is accidental or represents an instance of a slowly emerging paradigm. However, the data indicate that there are more regular appearances of plural forms after 2;1 while the Nominative and the Accusative are the cases that appear most frequently. The variety of different case forms in the plural is blurred by an even greater degree of syncretism than in the singular. Some cases even do not appear in the plural at all, such as the Dative in Antonija’s and Marina’s corpuses, the Instrumental in Antonija’s corpus, and the Vocative in Marina’s and Vjeran’s corpuses (figures 10, 11, 12); therefore, the data in the plural are not represented cumulatively. Given the scarcity of data on plural case forms and the many gaps in their use in Antonija’s speech, it is impossible to demonstrate that any of them is productively used so that emerging schemas will remain strictly tied to a few lexical items in the second half of her third year.

Figure 10. Order of emergence of cases in the singular and the plural – Antonija

The acquisition of case, number, and gender in Croatian

173

Figure 11. Order of emergence of cases in the singular and the plural – Marina

Figure 12. Order of emergence of cases in the singular and the plural – Vjeran

5.6.

Error analysis

Although errors and overgeneralizations are not only regarded as a challenging topic (e.g. Klampfer et al. 2001), but also as a good source of evidence in a process of acquisition of morphology, Croatian data are not very informative in this respect. Taking morphological complexity as a starting point which

174

Melita Kovaˇcevi´c, Marijan Palmovi´c, and Gordana Hrˇzica

can be viewed as a fertile ground for morphological errors we note that there are surprisingly few inflectional errors produced by the child studied in this paper. In Antonija’s entire corpus, only twenty errors related to the inflexion of the noun occur, i.e. there are 20 tokens of erroneous forms in relation to 1,780 noun tokens. Their analysis, compared to correctly inflected forms, allows making the following general observations: A) The child always produces a morphological marker, regardless of its correctness. In other words, morphological marking is preferred to non-marking.This could be explained by the effect of morphological complexity. The child is aware of frequent presence of morphological markers and accordingly tries to fill morphological slots by producing markers. B) When a noun requiring case marking is constructed with a preposition, the child more frequently selects the adequate preposition rather than the correct morphological marking. This could be explained by the fact that a preposition carries greater weight than the inflectional marker which is determined by the case. For example, there are 7 errors concerning prepositions u ‘in’ and na ‘on’, such as the wrong choice of the case marker for the masculine noun park ‘park’: *u park-i instead of u park-u in park-Loc.Sg.Fem. in park-Loc.Sg.Masc. The child may not choose the right case marking with a preposition that governs two possible cases (Accusative and Locative) which could also imply that the child has not yet acquired relevant syntactic parameters (there are no such errors with the Accusative). There are some errors related to pluralia tantum, which the child tries to inflect in the singular (five tokens); for example: *ust-ek-u instead of ust-ek-ima mouth-Dim-Loc.Sg. mouth-Dim-Loc.Pl. *kolic-om stroller-Instr.Sg.

instead of kolic-ima stroller-Instr.Pl.

6. Conclusions The analysis of the speech of a Croatian child, Antonija, recorded between the ages of 1;3 and 2;8, has provided evidence for answering the questions raised at the beginning of this paper. Rather than being an obstacle morphological complexity proves to be a trigger for the early development of inflection. The first oppositions of case forms appear at the age of 1;6 so that the first grammatical

The acquisition of case, number, and gender in Croatian

175

markers are already used in the one-word stage. Such a course of morphological development differs significantly from that of other morphologically less complex languages. Being confronted with morphological richness in the target language (Kovaˇcevi´c et al. 1998), the child behaves as an ‘efficient’ learner from the very beginning balancing between the acquisition of vocabulary and morphology. After a critical vocabulary mass has been reached, the see-saw starts shifting to the other side. This is nicely documented in figure 2 by a sudden increase in grammatical types while the number of lemmas increases at a slower pace. In other words, there is no ‘cutting-point’ between an initially merely lexical acquisition and a later morphological development, but rather gradual growth of the morphological structure by accumulating items of grammatical forms. In a morphologically rich language such as Croatian a child will soon encounter more than one grammatical form of many nouns and these will gradually develop into item-based grammatical schemas, some of which will become productive in the course of development. The ‘efficient’ learner stretches further, making a complex system simpler. Contrary to the verbal system, in which this is achieved mostly by overgeneralizations (Anąel et al. 2000), simplicity is achieved in noun morphology by identifying the ‘proper’ paradigm, one that assures the most efficient communication. In this respect the feminine e-class paradigm appears as the paradigm with the highest level of transparency in Croatian. The order of emergence of cases is relatively stable across all three children in the Croatian corpus. The order of their appearance is defined by case function and it is determined by context. The Nominative and the Accusative emerge first expressing core sentence relations. However, since early vocabulary is highly contextualized, certain nouns appear only in restricted contexts, i.e. in particular collocations which require particular cases. These cases are not necessarily the first cases to appear in the course of acquisition, i.e. the Nominative, Accusative and Vocative. On the other side, because of this context-determined noun usage it is not very likely that the whole paradigm will appear on one target lemma. The early appearance of oppositions and mini-paradigms on different lemmas, especially in e-class nouns, can therefore serve as a sign of a productive use of case forms. Moreover, the plural in Croatian takes long time to acquire. School age children still experience difficulties in forming proper plural forms, and even adults sometimes produce ungrammatical plurals. However, although with low frequency, the plural appears relatively early, but it is restricted to prototypically paired objects such as hands, gloves, shoes, or feet. There is a certain degree of parallelism between child and adult language. Figure 2 can be informative in this respect, as well: it shows that the number of types increases not only in child language due to the acquisition processes, but also in the input data. Of course, the absolute number of occurrences of any

176

Melita Kovaˇcevi´c, Marijan Palmovi´c, and Gordana Hrˇzica

category is always in favor of the input language; however, speaking in relative terms, the appearance of all categories is similar. This paper has focused on nouns to the exclusion of pronouns and adjectives, which are also marked for case and number in Croatian. A full account of the acquisition of the inflexion of nominals of the Croatian language will have to include these parts of speech and also take later stages of language acquisition into consideration.

Notes ∗

Acknowledgement:The research done for this paper is a part of the projectAcquisition of Croatian in Crosslinguistic Perspective (No. 0013002) supported by a grant from the Croatian Ministry of Science and Education.

References Anąel, Maja, Sabine Klampfer, Marianne Kilani-Schoch, Wolfgang U. Dressler, and Melita Kovaˇcevi´c 2000 Acquisition of verbs in Croatian, French and Austrian German: An outline of a comparative analysis. Suvremena lingvistika, Vol. 1-2, No. 49-50: 5–27. Bari´c, Emilija, Mijo Lonˇcari´c, Dragica Mali´c, Slavko Paveˇsi´c, Mirko Peti, Vesna Zeˇcevi´c, and Marija Zinka ˇ 1997 Hrvatska gramatika [Croatian Grammar]. Zagreb: Skolska knjiga. Bybee, Joan 2001 Phonology and Language Use. Cambridge: Cambridge University Press. Jelaska, Zrinka, Melita Kovaˇcevi´c, and Maja Anąel 2002 Morphology and semantics: The basis of Croatian case. In Pre- and Protomorphology: Early Phases of Morphological Development in Nouns and Verbs, Maria D. Voeikova and Wolfgang U. Dressler (eds.), 177–189. M¨unchen: LINCOM Europa. Klampfer, Sabine, and Katharina Korecky-Kr¨oll 2001 Morphological potentiality in children’s overgeneralization patterns: Evidence from Austrian German noun plurals. Wiener Linguistische Gazette 67–69: 25–44. Kovaˇcevi´c, Melita, Zrinka Jelaska, and Blazenka Brozovi´c 1998 Comparing lexical and grammatical development in morphologically different languages. In Perspectives on Language Acquisition: Selected Papers from the VIIth International Congress for the Study of

The acquisition of case, number, and gender in Croatian

177

Child Language, Ayhan Aksu-Ko¸c, Eser Ergunvanlı Taylan, A. Sumru ¨ Ozsoy, and Aylin K¨untay (eds.), 368–383. Bebek/Istanbul: Bo˘gazi¸ci University Printhouse. Kuvaˇc, Jelena, and Marijan Palmovi´c 2002 The course of acquisition of possessive pronouns in Croatian. Paper read at the IX. International Congress for the Study of Child Language, Madison, WI, 16–21 July, 2002. MacWhinney, Brian 2000 The CHILDES Project: Tools for Analyzing Talk. Hillsdale, New Jersey: Lawrence Erlbaum. Matasovi´c, Ranko 2000 The possessive and adjective phrases in Croatian. Suvremena lingvistika. Vol. 1-2, No. 49-50: 99–110. Nelson, Katharine 1973 Structure and strategy in learning to talk. Monographs of the Society for Research in Child Development, 38 (No. 149). 1985 Making Sense: The Acquisition of Shared Meaning. New York: Academic Press. Palmovi´c, Marijan 2007 The acquisition of diminutives in Croatian. In The Acquisition of Diminutives: A Cross-Linguistic Perspective, Ineta Savickienë and Wolfgang U. Dressler (eds.), 73–88. Amsterdam/Philadelphia: John Benjamins. Raguˇz, Dragutin 1997 Praktiˇcna hrvatska gramatika [Practical Croatian Grammar]. Zagreb: Medicinska naklada. Teˇzak, Stjepan, and Stjepan Babi´c 2000 Gramatika hrvatskoga jezika: Priruˇcnik za osnovno jeziˇcno obrazovanje [Croatian Language Grammar: Basic Language Handbook]. Zaˇ greb: Skolska knjiga. Tomasello, Michael 2003 Constructing a Language: A Usage-Based Theory of Langauge Acquisition. Cambridge, MA: Harvard University Press. Vuleti´c, Duˇsanka 1991 Istraˇzivanje govora [Speech Research]. Zagreb: Fakultet za defektologiju.

Acquisition of case and number in Russian∗ Natalia Gagarina and Maria Voeikova Introduction This paper examines the acquisition of number and case in four monolingual Russian-speaking children on an extended amount of longitudinal data. Our threefold aim is (a) to trace the emergence and development of case and number oppositions from the onset of noun production, (b) to establish developmental phases and to define the criteria of the transition between them, and (c) to account for the mechanisms of case and number acquisition. The study follows the theoretical assumptions of the pre- and protomorphological project (Dressler and Karpf 1995; Dressler (ed.) 1997), which agree with the constructivist (functionalist) view on language acquisition (Slobin 1985; Tomasello 2003, among others). This view presupposes a decisive role of children’s input and linguistic experience as well as an important role of the communicative function of language in the acquisition process. Proponents of these approaches agree that linguistic categories are not innate but are consequently formed by means of selective elaboration of input by children. This can be seen from the premorphological phase of language acquisition in which the early forms used by children are yet not contrasted and, thus, may not be considered as an exhibition of language-specific morphological features. This initial phase is characterized by the use of rote-learnt forms corresponding to adult inflected word forms without having this status in child speech as long as contrasts between different grammatical forms are lacking; by uninflected (truncated) forms and by extragrammatical operations – first attempts of children to construct complex linguistic signs as being scrutinized in (Dressler and Karpf 1995; Dressler (ed.) 1997; Voeikova 1997 for Russian). The subsequent protomorphological phase is characterised by the emergence and development of productive (with the various degree of productivity) morphological patterns. The acquisition of these patterns by Russian children was shown for verbal inflectional categories in Kiebzak-Mandera (2000) and Gagarina (2003, 2008) and for noun categories in Voeikova and Gagarina (2002). It was demonstrated that, for example, the emergence (and productive use) of verb inflection takes place within a rather restricted period of time – about three to four months from the onset of verb production. The acquisition of nominal inflectional endings starts even earlier: in the case of one of the examined

180

Natalia Gagarina and Maria Voeikova

children, Liza, inflectional endings of all six case forms in the singular and first plural nominative forms emerged within one month – at 1;9. However, some of the important case functions still lack even at the age of 3;0. The fact that at least some of the first forms are creatively constructed by children (however, with the lowest degree of productiveness) may be proven by their errors; first of all, by the misuse of secondary markers, such as the wrong position of stress, non-standard palatalization, and the absence of suppletive forms. In contact with the input, children selectively choose patterns they operate with. These patterns are not chosen randomly, but on the basis of a group of interacting factors, such as type and token frequency, structural transparency, saliency, inter alia (cf. Dressler 2001, 2007). Through learning the grammatical patterns and their constraints, children acquire the grammatical categories of their language, among them grammatical case and number – the subjects of the present article.

1. Acquisition of Russian case and number – state of the art Studies on the acquisition of case and number in Russian go back to the pioneering work of Gvozdev who documented the speech development of his son Zhenja “from first words to the first grade” in diary notes (Gvozdev 1981) and analyzed this rich data collection later on (Gvozdev [1949] 2007). The earliest period observed by Gvozdev was called by him “sentences consisting of amorphous words-routes from 1 year to 1 year 10 months” (2007: 162–165). This stage closely corresponds to the premorphological stage as sketched above. The early number and case oppositions occur with a slight lag after the onset of word production, namely between the age of 1;9 and 1;11 (2007: 171–175). Gvozdev gives numerous examples of the first non-target number oppositions documented in his son’s speech before 2;0 (e.g. *ugoli for ugli ‘carbon:NOM:PL’ or *prasjonˇciki for porosjata ‘pig:NOM:PL’), which he calls analogy formations (Gvozdev 2007: 171). These and similar examples from his data can be treated as overgeneralizations – a clear sign of the early application of a productive pattern of number formation. Case differentiation at Zhenja starts with the opposition between NOM and ACC. Gvozdev’s example (2007: 169) daj *gal’andu pisat’ for daj karandaˇs pisat’ ‘give:IMP pencil:ACC:SG write:INF’ for ‘give me a pencil for writing’ (1;11.7) exemplifies the use of the ACC inflection of the feminine DC1 instead of the correct masculine DC2 (see table 2). The opposition between NOM and ACC is the first productive opposition to emerge, also documented in other studies on Russian and some other accusative languages (see Bittner 2002

Acquisition of case and number in Russian

181

for German; Clahsen, Eisenbeiss, and Vainikka 1994; Eisenbeiss 2003; Eisenbeiss, Bartke, and Clahsen 2006 for an overview, Eisenbeiss, Narasimhan, and Voeikova 2008). E.g. in Croatian children, the NOM-ACC opposition also occurs at the earliest (Kovaˇcevi´c, Palmovi´c, and Hrˇzica, this volume). In Finnish, however, partitive and genitive forms are opposed to the nominative as early as the accusative (Laalo, this volume), in Estonian, in which the accusative is absent and its functions are carried out by the genitive for countable nouns, or by the partitive for the uncountable nouns, this opposition is not as clear (Argus, this volume). In Greek, the NOM-ACC distinction is at first limited to the small class of masculine nouns, while the inflectionally expressed number distinction develops very early with neuter nouns, but not with masculine ones (see Stephany 2006; Stephany and Christofidou, this volume). Thus, the earliest contrasts may be explained in terms of thematic roles (Agent vs. Undergoer), or rather with the help of the notion of “manipulating activity scene” introduced in Slobin (1985). Slobin argues that children first mark prototypical animate agents opposed to inanimate objects and that only later on they start to use case morphemes grammatically correctly, irrespective of the semantic class of nouns involved. Thus, the early use of the first case markers is language-specific, cognitively anchored and, most probably, usage-based in a sense that neither all the nouns, nor all the syntactic objects obey the rules of assigning the inflectional ending. In general, Gvozdev’s descriptive study reveals the rapid development of case and number inflection, which leads to the creative use of these inflections after the crucial period of approximately two months from the onset of noun production. His study shows the high number of overgeneralizations, a phenomenon that Slobin later called ‘inflectional imperialism’ (Slobin 1966). Slobin (1966) and Gvozdev (2007) both stress the dispreference for a certain declension class as a whole; rather, for every case, Russian children show a predilection for the most salient inflectional marker, which is then substituted for the less salient ones. Ceitlin (2000) and Lepskaja (1988) continued the descriptive tradition of Gvozdev. Ceitlin (1987, 1988, 2000) based her investigations on more recent data than that of Gvozdev, both diary and longitudinal, but mostly on the sporadic observations of the use of morphological forms by children. Her results corroborate the preferential status of NOM-ACC contrast. The analysis of the acquisition of grammatical number and case in Ceitlin (1987, 1988) shows another important tendency: children first acquire productive inflectional markers and overgeneralise them to the unproductive forms, creating new forms and filling in the lacunae in the language system. They produce non-standard plural forms of uncountable nouns, e.g. *kapust-y ‘cabbage-NOM:PL’, *vozdux-i

182

Natalia Gagarina and Maria Voeikova

‘air-NOM:PL’, *ˇserst-i ‘wool-NOM:PL’ or form non-existing singulatives like *salatink-a ‘salat-SING:NOM:SG’ “using potential capacities of the language system” (Ceitlin 1988: 35–36). In general, Ceitlin’s findings may be summarized as follows: 1) systemic morphological relations are acquired earlier than the normative constraints, 2) unproductive patterns are replaced by the productive ones. Occasional morphological forms examined by Ceitlin demonstrating the creative way of form-building are characteristic of the transition from protomorphology to the adult-like morphology. Lepskaja (1988) concentrated her research on the order of emergence of case oppositions and showed that after starting with the NOM-ACC opposition children use several oblique case forms together “in one bunch” (Lepskaja 1988: 51). The genitive, dative, and locative case contrasts to the nominative have equal chances to emerge after the NOM-ACC opposition. Also in our data, children all start with this first contrast, however, the second oblique case to emerge is different in all four subjects: e.g. Filipp starts with the locative from nouns denoting transport poezde ‘train:PREP:SG’ and mashine ‘car:PREP:SG’ without the required preposition, whereas Vanja even uses the instrumental tetej ‘aunt:INSTR:SG’ in the comitative function ‘with the aunt’. The omission of prepositions in prepositional phrases alongside with correctly used inflectional endings is typical in child Russian. Gvozdev points to the fact that “in general, prepositions usually are omitted in all case forms and only at the end of this period (2;2–2;4) some of them occur, usually in the frequently used expressions” (Gvozdev 2007: 214). The formal approaches adhering to the postulates of Generative Grammar assume that the acquisition of case is interconnected with the development of syntactic structure and the emergence of thematic roles. This point of view is represented in (Babyonyshev 1993; Clahsen, Eisenbeiss, and Vainikka 1994) and explains the order of emergence of case forms by the growing complexity of sentence structure. As will be shown below, our data do not corroborate these findings and provide evidence that at least some children already exhibit case marking oppositions before their MLU exceeds 2.0 (see section 4.4. about the development of indirect case forms compared to the MLU). Our evidence questions a syntactic explanation of the acquisition process of inflection and serves as a base for functional accounts of children’s preferences in the acquisition of morphology. Furthermore, it suggests a strong learner-specific variation in the acquisition of nominal morphology. Lexically (semantically) oriented approaches bind the emerging cases in Russian children to certain semantic classes of nouns (Ionova 2007). Within this framework it is shown that certain case forms may only emerge with a limited set of nouns depending on their meaning, e.g. agent nouns will be used first in the nominative, and later on in the genitive with the meaning of possessor,

Acquisition of case and number in Russian

183

or in the dative as beneficiary. In contrast to this, in early child Greek, even in the absence of case marking, the syntactic function of nouns was clear because mostly animate nouns were used as subjects whereas inanimate ones functioned as direct objects (Stephany 1997). Thus, two contradictory tendencies may be observed: children either first mark nouns for case in prototypical contexts, or omit case markers that are redundant taken the semantic co-relation between e.g. animate and inanimate nouns. In our data inanimate nouns occur predominantly in the accusative, whereas nouns denoting instruments are used in the instrumental and the names of containers are first marked for the locative. The lexically-based approach, as well as the usage-based theory, receives support from the fact that different case forms are in a way predestined for the use in different speech acts since each of them has its own pragmatic values. Thus, in child Finnish, early partitive, and illative forms are typical for request situations in which children “indicate the wish for food, . . . or the wish to get somewhere” (Laalo 2002: 90).An important weakness of pragmatic theories is, however, that they do not explain the language-specific impact on the order of acquisition of case inflections. Expressing their needs is typical for all children, irrespective of their native language; however, in some languages unmarked or truncated forms may serve this function, whereas in other languages (e.g. Finnish) the first requests get the correct morphological marking. Since many such correctly marked forms are not formulaic and were creatively constructed by children, the question arises how do they know which inflectional ending should be used. Thus, Zhenja Gvozdev uses the GEN –a ending to mark objects of request from early on: daj * tulka ‘give chair:GEN:SG’(1;10), daj *tankana ‘give glass:GEN:SG’ (1;11) even with the countable nouns that should be put into the accusative (Gvozdev 2007: 173). One may presuppose that children somehow grasp the general meaning of the case forms and try to apply them basing on it. The idea of general meanings of Russian cases formulated by Jakobson (1936, 1958) presupposes that the different functions of every case are semantically related to each other. Thus, all functions of the GEN are taken to be based on partiality. However, we found no evidence that children younger than 3 years are able to predict case marking deductively, basing themselves on the general meaning of cases. E.g. the examples from Zhenja Gvozdev cited above contradict the general idea of partiality. Rather, children inductively generalize the situations they are accustomed to and produce new utterances containing the analogies of case forms which they have already acquired. The forms *tulka ‘chair:GEN:SG’ and *tankana ‘glass:GEN:SG’ were constructed by Zhenja by analogy with the forms *maka ‘milk:GEN:SG’ *is’ka ‘egg:GEN:SG’ registered in his speech a month earlier (1;10) (Gvozdev 2007: 172).

184

Natalia Gagarina and Maria Voeikova

Children first acquire case forms in their principal functions. Important factors that may trigger the acquisition of forms and semantic roles are: cognitive availability, pragmatic importance, frequency in the input, transparency of expression, and level of syntactic development (Ionova 2007: 115–122). However, none of these factors may play a decisive role in the acquisition, rather, the combination of them. For example, according to Lepskaja (1988), the instrumental and the locative are acquired by Russian children later than the other cases because of their low frequency, minor pragmatic value and because of the fact that speaking about location and instrument demands a high level of syntactic development. In our data, indeed, the locative and instrumental forms occur at the latest. The same order of development is registered in children learning Croatian (see Kovaˇcevi´c, Palmovi´c, and Hrˇzica, this volume). This coincides with the idea that, unlike the core syntactic cases, semantic cases are later acquired by children (Eisenbeiss, Bartke, and Clahsen 2006, Eisenbeiss, Narasimhan, and Voeikova 2008: 373). To express location, Russian children under observation use frozen phrases with nouns in the locative case having an adverbial status, like na ulice ‘in street:PREP:SG’ (Filipp 1;10), na krovatke ‘on bed:PREP:SG’ (Liza 1;11). Location and direction in Russian are marked by the same prepositions but by different cases, PREP and ACC, accordingly. It is known that in languages with distinct case inflections for locative state vs. locative goal (e.g. German: dative vs. accusative), children typically confuse the two forms (Eisenbeiss, Bartke, and Clahsen 2006). According to the earlier observations made by Mills, German children overgeneralize the accusative for both (locative state and locative goal) functions. Mills (1985: 192) notes that, in German, “the directional use (accusative) was always correct except where the article was omitted.” A little later she notes the following: “Far more errors occur when dative case must be used with stative meaning” and that “the early use seems to indicate overgeneralization of accusative and therefore no indication of marking a distinction in the two meanings” (p. 192).1 In Russian, the accusative of goal occurs early, sometimes in adverbial-like constructions. The Gvozdev’s examples give evidence of such early use, e.g. nalej vody kruˇzku, cˇ aˇsku ‘pour water:GEN:SG mug:ACC:SG, cup:ACC:SG’, gus’ki kantinku *kladil ‘toys:NOM/ACC:SG basket:ACC:SG put:PAST: MASC:SG’, jaiˇcko letelo dunduk ‘egg:NOM:SG fly:PAST:NEUT/FEM:SG box:NOM/ACC:SG’ (1;11) (Gvozdev 2007: 174). The unusual contexts, selfcorrections (cup vs. mug) and erroneous verb forms (kladil vs. klal) show that all these utterances were constructed by Zhenja. In all of them he uses the correct inflectional ending of the accusative, however, the prepositions are omitted. The locative forms in his speech stole ‘(at) table:PREP:SG’, stul’ki sidit ‘chair:DIM:PREP:SG sit’ occur at the same period (1;11), but Gvozdev stresses

Acquisition of case and number in Russian

185

that there is less evidence of their productive use: “only the form tol’ki2 for the adult na stul’ˇcike ‘at (the) chair’ is a clear innovation, also the unvoiced pronunciation komote instead of komode is a sign of productive building from the nominative stem” (Gvozdev 2007: 195). The more creative use of the accusative may also be due to the fact that it has a highly reliable marker –u for feminine nouns, whereas the locative forms are usually homonymous with the dative and/or genitive ones depending on the declension class. In general, the newly acquired forms are first used in one prototypical function (compare the development of the genitive in the Greek boy Christos investigated by Christofidou and Stephany (1997: 138), who used it first only to denote possessor).The same dominance of only one prototypical function per case was also registered in speech of Lithuanian children (Savickien˙e 2003: 64–90). Limitation of the emergence of grammatical distinctions with nouns may also relate to gender classes that are clearly language-specific (see Stephany 2006; Stephany and Christofidou, this volume). In contrast to lexically driven approaches, the pre- and protomorphological approach to the acquisition of case forms by children, which we adopt here, considers the structural properties of language signs related to the function they fulfil and scrutinises the various types of inflectional classes differing by their transparency, saliency and iconicity of case and number forms. Previous results on the acquisition of the Russian systems of verb and noun inflection gained by the present authors within the pre-/protomorphological approach have been integrated into the present study (see Voeikova and Dressler (eds.) 2002; Bittner, Dressler, and Kilani-Schoch 2003) and the main definitions of key terms offered by Bittner, Dressler, and Kilani-Schoch (2003: xxxix–xl) in the introduction of the work dedicated to the development of verbal inflection in first language acquisition will be followed in the present paper. We will further develop our idea of measuring progress made in the acquisition of grammatical morphology with the help of the notion Percentage of Base Forms (PBF) (Voeikova and Gagarina 2002) and consider the interrelation between the growth of the children’s lexicon and the enrichment of the morphological complexity of their speech. The paper is structured as follows: After dealing with the state of the art of the acquisition of Russian case and number (section 1), the different parts of the nominal case and number system of the target language will be presented (sect. 2). Section 3 reports on the informants and the coding/tagging procedure. In the main section of the paper (section 4), the results of our analyses will be presented, namely, (a) the emergence of the oblique case forms and of number oppositions and the development of miniparadigms with regard to the

186

Natalia Gagarina and Maria Voeikova

productivity of the inflectional classes to which the nouns belong, (b) interrelations between the lexical and inflectional spurt and c) the percentage of base forms in child speech related to the MLU. The paper will end with a discussion of our results and some general conclusions.

2. The Russian case and number system as an object of acquisition Russian has an accusative case marking system (Blake 2001: 119) with a fusion of case, number and gender marking. Case marking is mostly stem-based and synthetic. There is case syncretism but, in adult Russian, adjectival inflections help to distinguish between homonymous case forms. Russian is a free wordorder language making case marking necessary for understanding the meaning of utterances.

2.1.

Case and semantic roles

The nominal paradigm consists of six cases in the singular and plural: nominative (NOM); genitive (GEN), dative (DAT), accusative (ACC), instrumental (INSTR), locative, or prepositional (PREP). NOM serves as a base form and is used by children instead of other case forms at the early stages. This case may also mark the agent in active constructions as well as the patient in passive constructions, compare Pap-a risuet maˇsin-u ‘father-NOM paints car-ACC’ and Maˇsin-a narisovana pap-oj ‘car-NOM (was) painted (by) father-INSTR’. However, passive constructions are very rare and mostly lexicalized; therefore they do not really influence the process of acquisition at the early stages. ACC typically marks an object, e.g. mam-a cˇ itaet knig-u ‘mommy-NOM:SG reads book-ACC:SG’. The grammatical object is marked differently with inanimate and animate nouns, namely by the NOM and GEN, respectively. The formal distinction between animate and inanimate nouns is characteristic of masculine nouns in the singular and of all nouns in the plural. ACC may also express direction (na ulic-u ‘to the street-ACC:SG’) or duration (celyj den’ ‘(for a) whole day:ACC:SG’). GEN is used to express partial and negated objects, as well as the possessor, e.g. u mam-y net konfet-y ‘at mommy-GEN no candy-GEN’ (= mommy has no candy). In impersonal sentences, the experiencer and the addressee are marked by DAT, e.g. bol’no miˇsk-e ‘(it) hurts the bear-DAT’, spoj zajk-e pesenk-u ‘sing hare-DAT song-ACC’. INSTR has many semantic functions expressing

Acquisition of case and number in Russian

187

Table 1. Principal semantic roles of Russian cases Agent Object Experiencer Possessor Addressee, Beneficiary Partial object Negated object Instrument Theme Location Direction Source Point in time Temporal boundary Duration

NOM × × ×

GEN

× × ×

× ×

DAT

×

ACC ×

INSTR ×

PREP

× × ×

×

× ×

× ×

instrument, agent (in passive sentences), temporal adverbial, path, and accompanying person, compare: Olja risuet karandaˇs-om ‘Olja is drawing (with a) pencil-INSTR’, zajˇcik byl pojman lis-oj ‘a hare was caught (by the) fox-INSTR’, poedem na daˇcu let-om ‘(we) will go to the dacha (in) summer-INSTR’, sˇel ja les-om ‘went I (through the) forest-INSTR’. The case PREP may only be used after prepositions and expresses location or theme of conversation, e.g. na ulic-e ‘on (the) street-PREP’, ob ulic-e ‘about (the) street-PREP’. With respect to the distinction between semantic and syntactic cases, INSTR and PREP are predominantly semantic cases in Russian, whereas the other cases are syntactic in most uses. Every case has a prototypical semantic function, as well as some marginal functions expressed by a combination of the case form with prepositions or other means. A summary of the principal semantic roles is given in table 1. Core functions of cases are marked in bold type. Whereas most case forms have a prototypical function, for GEN, it is difficult to decide, whether denoting possessor, or partial and negated object is more essential for this form. Therefore, for GEN we did not define any prototypical function.

188

Natalia Gagarina and Maria Voeikova

2.2.

Inflectional classes and case marking

Like most inflecting-fusional languages with synthetic case marking, Russian has declension classes. Three principal types of declension, depending on the gender and phonological type of the stem (Zaliznjak 1977)3 are relevant for the first stages of language acquisition. The 1st (feminine and masculine nouns ending in -a) and 2nd (masculine nouns ending in a consonant and neuter nouns) declension classes4 (henceforth DC1 and DC2) are of the most important ones in child-centred situations and may be considered as macroclasses in the sense of Dressler et al. (1995–1996; 2006), whereas the 3rd declension class (henceforth DC3), constituted by feminine nouns with a final consonant, has a low frequency in the input and is acquired later. Children usually erroneously recognise these nouns as masculine and decline them as belonging to DC2; compare the example from Gvozdev (2007: 178): daj sol-a* (DC2) instead of daj sol-i (DC3) ‘give salt-GEN:SG’. Usually the child uses the word sol’ ‘salt’ in the nominative. Gvozdev supposes that the form sol’a* (DC2) is not a morphological opposition to the NOM but a small-scale analogy with maka (DC2) (= moloka) ‘milk:GEN:SG’ (diary data from 1;9.28). Both factors may impact the production of this form: children try to ‘get rid of’ the unproductive DC3 by adding the non-standard feminine ending –a to such nouns, e.g. *veˇs’-a instead of veˇs’ ‘thing’ (Ceitlin 1988: 44) or by using diminutives (Olmsted 1994; Kempe, Brooks, and Pirott 2001). These facts are additional indicators of the productivity of DC1 and DC2 in the early speech production of children. Also with regard to phonology, it makes sense to assume the two principal declension types DC1 and DC2. Since nouns belonging to DC1 carry the inflectional ending –a in the NOM, children have to learn the operation of changing inflectional endings rather than adding them to unmarked base forms (Smoczy´nska 1985). Both productive DCs have similar inflectional endings in the plural (except for GEN (and ACC of animate nouns)), which allows competitive inflectional variants. The plural inflections of NOM, DAT, and ACC of inanimate nouns are considered superstable markers. However, oblique case forms are acquired significantly later in the plural than in the singular. Usually case distinctions are expressed by inflectional suffixes and by accompanying palatalization of consonants before the inflectional endings containing front vowels. As shown by the data, palatalization is misused by children partly due to pure phonological reasons, since they tend to conserve the same phonological form of the stem throughout the whole paradigm. Changes of derivational suffixes and suppletive marking are relatively rare. Russian stress placement is not constant and stress shift is neither phonologically nor morphologically determined. According to Ceitlin (1988: 39–43), children tend to avoid stress shifts

Acquisition of case and number in Russian

189

and stem alternations, conserving one and the same phonological stem accented on one and the same syllable for all inflectional forms. Almost 86% of all nouns in the input have unstressed inflectional suffixes. These differ from stressed suffixes by the fact that the mid vowels /o/ and /e/ do not occur in unstressed position and are reduced to [a] and [i] respectively resulting in a three-vowel system in unstressed syllables. Thus, stress shift influences the number of distinct case-number forms occurring in a paradigm. Smoczy´nska (1985) refers to this phonological rule and its implications for case forms in Russian in contrast to Polish, which does not possess such a rule. Compare the following example taken from Gvozdev: din’ *klEsu* ‘move (the) armchair:ACC:SG’ (DC1) instead of the correct sdvin’ krEslo (DC2) (Gvozdev 2007: 168). The system of inflectional endings is shown in table 2. As pointed out above, the choice of vowels in inflectional endings in unstressed position is restricted to [i, u, a] or zero so that the hearer has to rely on the context for discovering the grammatical function of syncretistic forms. Another problem for the learner in his endeavour to tease declensional classes apart is the high degree of homophony between inflectional endings marking different case-number categories in the two principal DCs. For instance, -i marks NOM in neuter nouns and GEN and DAT in feminine nouns; -u is a marker of Table 2. Stressed and unstressed inflectional endings in DC1 and DC25

NOM

DC1: FEM (and some MASC) ending in -a -a mAma ‘mommy’

GEN DAT ACC

-i (y) -e -u

mAmy mAme [i] mAmu

INSTR

-oj

mAmoj [aj]

PREP

-e

o mAme [i] ‘about mommy’

NOM:PL -i (y)

mAmy

DC2: MASC ending in a consonant (and NEUT ending in -o [a] , -e [i]) ∅/-o/-e dom ‘house’, FilIpp ‘Philip’, stol ‘table’, mAl’ˇcik ‘boy’, krEslo [a] ‘armchair’, pOle [i] ‘field’ -a dOma, FilIppa, krEsla, pOlja -u dOmu, krEslu, pOlju ∅/-a/ dom, FilIppa, krEslo [a], -o/-e pOle [i] -om/-em dOmom [am], FilIpom [am], krEslom [am], pOlem [im] -e o dOme [i], o FilIppe [i], o krEsle [i], o pOle [i] ‘about the house’ (etc.) -A/-a domA, poljA, krEsla -i (y) stolY, mAl’ˇciki

190

Natalia Gagarina and Maria Voeikova

ACC for feminine nouns but of DAT for masculine and neuter ones; -a marks the NOM for feminine and GEN (and ACC for animate) masculine and neuter nouns. Syncretic forms occur in the paradigm of every noun, thus, the unstressed inflectional ending in tjoti of the lexeme tjotja ‘aunt’ may be GEN:SG, DAT:SG, PREP:SG or NOM:PL.6 In Russian, one can only recognise the function of the inflectional ending of a noun if it is compared to the base form (Voeikova 2002: 29). Also, the morphological number is fused with case. As shown in table 2, for NOM:PL only one inflectional ending is used with feminine nouns (DC1 and unproductive DC3), whereas in DC2 (masculine and neuter nouns) there are two concurrent markers: stressed –a and both stressed and unstressed –i (-y). The plural cases besides NOM almost do not occur in our data, therefore table 2 contains only case forms in the singular.

3. Data and method Longitudinal data from four monolingual children, one girl and three boys, were analyzed for this study. All children grew up in middle-class families in St. Petersburg, where the standard version of colloquial Russian is spoken. The three boys, Filipp, Vanja, and Vitja are single children, while Liza is the second child (her brother is ten years older). All children were systematically recorded and/or video-taped for two to five hours a month in every-day ‘routine’situations at their homes in St. Petersburg and in the summer country houses during the summer. The children differ both by the pace of language acquisition and by the strategies used. Filipp and Liza are so-called early talkers, whereas Vanja and Vitja started to talk later. While Filipp is rather an ‘expressive’ child, Liza is a ‘referential’ child (Nelson 1981). Filipp demonstrates an imitative strategy in language acquisition. From 1;7 to 1;8, the number of noun lemmas in his recordings grows considerably (from 24 to 87). He does not use nouns much for making predications, but they rather occur in naming situations (e.g. ball for ‘this is (called) a ball’ rather than ‘take the ball/give me the ball/mummy’s ball’ etc.). Liza, instead of naming things, utters short predications about them, or, to put it another way, her utterances consist of a focus (rhema) only. Consequently, at this early stage, Liza’s utterance length is limited to one or two morphologically marked words. She is rather careful in phonetically distinguishing different inflectional endings although her pronunciation cannot be said to be “accurate”. She often preserves the syllable structure of the word while changing its phonetic structure, for example, igigiki (1;8) – for ogurˇciki ‘cucumbers:DIM:PL’ (cf.

Acquisition of case and number in Russian

191

Stephany 1997 for Greek). Liza’s speech is also characterised by a number of so-called family-specific words registered during the whole period of recordings. Interestingly, these words are declined developing multiple case oppositions and may also serve as base forms for derivation. An example of such a word is the name of Liza’s brother Aljoˇsa – Apka (family specific). Both Vitja and Vanja are late talkers and are usually behind in their language development in comparison with the two other children. However, their progress both in MLU and in the number of case/number oppositions is rapid compared to the early talkers Filipp and Liza. Babytalk words, i.e. words that do not exist in target Russian and child-directed speech are stably used and are not easily superseded by their counterparts from the adult language, for example bizinja instead of maˇsina for ‘car’ (Vanja from 2;1 to 2;3). For the purpose of the present study we analysed nine subsequent months of speech of the four children starting from the emergence of number and case oppositions. In the preceding premorphological period, as expected from the previous investigations (“one-word sentences with amorphous words-routes” as described by Gvozdev) children used either truncated, or base (NOM) forms of nouns in all situations. The starting point corresponds to the end of premorphological phase which covers the period ‘before the detection of the grammatical morphology’ (Bittner, Dressler, and Kilani-Schoch 2003: xviii). During the chosen interval of nine subsequent months children develop marking of case and number and acquire case and number oppositions, i.e. this interval covers the period when ‘children detect and reconstruct or construct creatively morphological patterns of analogies or of first rules’ (Bittner, Dressler, and Kilani-Schoch 2003: xix). Since, as it has been shown in the development of verb morphology (cf. Gagarina 2000, 2003), the phases do not abruptly replace one another, we consider the transitional period between the two phases in the development of case and number morphology. Thus, this time interval corresponds to the final phase of premorphology, the transition to protomorphology, and the development of the latter. The 1st month of recordings taken as a starting point for all children was chosen due to the occurrence of first oppositions. For example, two months before the conventional oppositions occurred in her speech, Liza uttered gag-a ‘(child-specific) goose-NOM:SG’ vs. gag-i - ‘goose-NOM:PL’ in a proper context with one goose vs. several geese. It may be that she had heard both forms and connected them to different situations (one bird vs. several birds). However, we did not take this example as a true productive singular/plural opposition because 1) her vocabulary at that time was limited to 8 words per recording, 2) until one month later she still could not produce such oppositions with other lemmas. The onset of analyses differs across children, since they started using case oppositions at a different age (see table 3). Filipp and Liza, the early talkers,

192

Natalia Gagarina and Maria Voeikova

Table 3. Onset of analyses and number of produced nouns (lemmas) Onset of case oppositions 1;7 1;8 2;0 2;2

Filipp Liza Vanja Vitja

Number of noun lemmas 24 33 39 29

start at 1;7 and 1;8 respectively; while Vanja and Vitja did so at 2;0 and 2;2, respectively. The number of noun lemmas in all children does not differ considerably before case oppositions occur, see table 3 (for a detailed description of data, see appendix 1). All data were transcribed in CHAT format according to the CHILDES conventions (MacWhinney 2000). For the grammatical coding of speech, the morphological program MORCOMM, specially developed for the coding of the spontaneous data of Russian-speaking adults and children (Voeikova 2000; Gagarina, Voeikova, and Gruzincev 2003) was used.

4. Results Results will be presented in four steps reflecting the different parts and aspects of our analysis. We will begin by an overview of the onset of the development of case oppositions and the development of miniparadigms and full paradigms discussing the development of productiveness (section 4.1). After describing the development of nominal number (section 4.2), the interrelation of inflectional and lexical development will be studied in the nominal domain (section 4.3). Section 4 will be concluded by a study of the relation between the percentage of base forms and MLU (section 4.4).

4.1.

Case oppositions, miniparadigms and the development of productivity

The first contrasted case forms in Filipp’s data at 1;7 are ded ‘grandfather:NOM’ and deda ‘grandfather:VOC’.These forms represent single instances, are contextdependent and can only be treated as rote-learnt. As stated above, our analyses start when the child data give evidence of the onset of number or case oppositions. At this point, the children use base forms (nominative) of nouns or other rote-learnt case forms; no signs of analogies, morphological patterns of case

Acquisition of case and number in Russian

193

40 30 20 10

9

8

7

6

5

4

3

2

1s t

m

on

th

0

VIT (1=2;2, 2=2;3, 3=2;4, ...)

LIZ (1=1;8, 2=1;9, 3=1;10, ...)

FIL (1=1;7, 2=1;8, 3=1;9, ...)

VAN (1=2;0, 2=2;1, 3=2;2, ...)

Figure 1. Emergence and development of case oppositions

and number morphology are attested, therefore all children can be said to be within the premorphological phase. Figure 1 demonstrates the proportion of noun lemmas with oppositions and/or miniparadigms out of all noun lemmas nine months from the onset of case oppositions. The first month, the emergence of oppositions, indicated in the diagram corresponds to the different biological age of children, e.g. the 1st month for Filipp corresponds to age 1;7, for Liza to 1;8, for Vanja to 2;0 and for Vitja to 2;2. All children start with a similar proportion of about 5% of nouns exhibiting oppositions. With three of the four children, Liza, Vanja and Vitja, the development of oppositions proceeds in a parallel way in the two subsequent months reaching 20% to 24% of nouns with contrastive forms. Only Filipp’s development differs from that of the other children in that the percentage of nouns exhibiting oppositions remains at the same level between the second and the fourth months from the onset point. This may be due to his general imitative strategy. Three children (Filipp, Liza, and Vanja) display a kind of a morphological spurt with various lags after the onset of oppositions and when the percentage of lemmas with oppositions stay between 6% and 10%. We define a morphological spurt as a steep increase (minimum 10%) of the amount of case oppositions within one month. For Vanja and Liza it starts one month after the first oppositions occur (at the age of 2;1 and 1;9, respectively). Vitja’s data show a steady increase from the very beginning up to the fourth month (from 2;2 to 2;6) with

194

Natalia Gagarina and Maria Voeikova

a small regress afterwards. This steady increase amounts in almost 10% after the first month and 6% within the two subsequent months. Filipp gives evidence of a morphological spurt only four months after the onset of oppositions when oppositions in his speech increase from 7% to 19%. This morphological spurt is followed by a period during which the percentage of oppositions in his speech fluctuates between 15% and 20% and does show neither the stable progressive development nor regressive tendencies (we call such periods with no clear developmental tendencies ‘plateau periods’). We consider the morphological spurt in four children as indicating a transitional period leading from pre- to protomorphology. During this period children noticeably increase lemmas with oppositions. Shortly afterwards no considerable increases are attested, but rather qualitative changes take place. The ‘reorganization’ of the number of oppositions within paradigms in this transitional period is manifested by the occurrence and development of paradigms with three forms, so-called miniparadigms, and paradigms with more case-number forms which are stably produced and do not disappear in the subsequent months. Despite the increasing number of lemmas with more case-number forms children rarely produce miniparadigms. This is caused by several factors. First, the number of members depends on the declension class. Thus, the 1st and 2nd productive classes exhibit a maximum of five forms in the singular, whereas the 3rd unproductive class has only 3 forms. Second, recordings only selectively represent the grammatical forms children produce so that a steady progress may not adequately be reflected in the data analyzed. These tendencies are at best reflected in Liza’s and Filipp’s data; the children start early but progress more slowly, at least, as far as the percentage of oppositions and the size of miniparadigms is concerned. Additionally, a rapid growth of vocabulary is documented for these two children. Liza starts with 33 noun lemmas and reaches a maximum of 176 lemmas at 2;2, during this very month a slow decrease in the percentage of case and number oppositions is found. The number of nouns in Filipp’s recordings reaches a maximum of 188 at 2;3, when the percentage of oppositions decreases (this remarkable relationship between vocabulary growth and morphological development will be discussed in section 4.3. below). The development of oppositions in our data starts with the productive 1st and nd 2 DC. This is predictable from the input because the percentage of feminine nouns of the unproductive 3rd DC occurring in the input does not exceed 3% of all noun lemmas.The effects of frequency and productivity coincide and together minimize the number of oppositions in the 3rd unproductive inflectional class. Thus, feminine nouns of the type myˇs’ ‘mouse’ occur later and never become frequent. Unproductive masculine nouns of the 1st class (type papa ‘daddy’) as well as neuter nouns of the 2nd class (type moloko ‘milk’) are also infrequent

Acquisition of case and number in Russian

195

in child-directed speech and rare in the early recordings (see also Ceitlin 1988: 44–46). While the number of oppositions in nouns belonging to the productive declension classes slowly increases in the course of development, unproductive nouns stay at the same level. The investigation of the emergence and development of case and number oppositions and the distinction between rote-learnt and productive forms provides an essential basis for the establishment of developmental phases. The first instances of miniparadigms were most often attested with frequent word forms, like children’s names, the words for mama and papa and their diminutive forms. It is not accidental that in Filipp’s recordings these first miniparadigms occur in the 2nd productive class, whereas they are found in the 1st productive class in Liza’s speech, the two DCs to which the children’s names belong. Even the first oppositions may show (a low degree of) productiveness as in the following example: at the age of 1;8, Filipp uses the genitive of ‘milk’ in the request ‘much milk’, whereas the nominative of ‘milk’ occurs in other contexts. Both forms are correctly used, but neither is registered in the mother’s speech in the previous recordings. Therefore, this opposition may be considered to have been productively created, but it has the lowest degree of productiveness in comparison with later forms. The transition from premorphology to protomorphology is characterized by the following features: 1) the rapid increase and stable use of miniparadigms of frequent nouns (with the growing degree of productivity), 2) the emergence of miniparadigms with less frequent as well as infrequent nouns, and 3) the growing size of paradigms, i.e. an increasing number of paradigm members and parallel increase of lemmas with paradigms (and further increase of productivity degree). The transition from pre- to protomorphology is not abrupt but must rather be considered as a continuum during which oppositions of both productive 1st and 2nd classes increase and oppositions in the unproductive 3rd class emerge. The set of case forms developing during this transitional phase is restricted to oppositions of grammatical cases, namely NOM-ACC, NOM-GEN and NOM-DAT. Not all factors characterizing this transitional period are quantitative, however. There are also important qualitative changes during the periods of the children’s development which appear as ‘plateau periods’, i.e. periods in which no considerable changes in the amount of lemmas with paradigms was observed (see figure 1): for Liza – starting from the 4th month (age 1;11), for Vanja – starting from the 5th month (age 2;4) and for Vitja – starting from the 4th month (age 2;5). At the end of the transitional stage, i.e. a month after the registration of the first (low degree) productive miniparadigms, the semantic cases locative and instrumental begin to enter nominal paradigms. First (contrastive) locative forms of nouns are registered in certain contexts occurring with a limited set of nouns

196

Natalia Gagarina and Maria Voeikova

such as vehicles, rooms or containers, e.g. na kuxne ‘in kitchen:PREP:SG’, na velosipede ‘by bicycle:PREP:SG’. (1)

Filipp (2;0) MOT: a chem ty ego kormil? and what:INSTR you him-ACC feed:PAST:MASC? ‘And with what did you feed him?’ FIL: *kolbaskoj dala. sausage:DIM:INSTR:SG give:PAST:FEM ‘(I) gave him sausage.’

Example (1) shows that the child picks up the instrumental case from his mother’s question, but uses it with the ditransitive verb ‘to give’ requiring the accusative for the direct object (kolbasku ‘sausage:DIM:ACC’). The subsequent inflectional development of the children is characterized by a slight increase in the number of 3-member miniparadigms and/or the occurrence of 4-member miniparadigms of the most frequent nouns (‘mummy’ or proper names), by numerous combinations of inflectional endings with simple nouns and derived ones (ˇsapk-a ‘cap-NOM:SG’, sˇapk-i ‘cap-NOM:PL’, sˇapoˇck-a ‘cap:DIM-NOM:SG’, sobak-a ‘dog-NOM:SG’, sobak-i ‘dog-NOM:PL’, sobak-u ‘dog-ACC:SG’, sobaˇck-u ‘dog:DIM-DAT:SG’, sobaˇck-i ‘dog:DIM-GEN:SG’, sobaˇck-a ‘dog: DIM-NOM: SG’), and by the emergence of the opposition of oblique cases in the plural, mostly limited to grammatical cases (e.g. ptiˇcek ‘bird:DIM:GEN:PL’, loˇsadok ‘horse:DIM:ACC:PL’). As a general rule, the addition of new members to miniparadigms is first observed with frequent nouns following clearly productive inflectional patterns and only later with frequent nonproductive nouns such as krylyˇsko ‘wing:NEUT:DIM:NOM:SG’, which also occurs in the GEN:SG and GEN:PL. Example (2) demonstrates Filipp’s difficulties at 2;2 of getting the correct case marking in different constructions and the ways he solves this problem partially relying on the adult model. Since the use of case and number in Russian constructions containing a numeral is formally motivated (GEN:SG is used after ‘two’, ‘three’ and ‘four’, whereas GEN:PL occurs with numerals from ‘five’ up), Filipp seems to interpret nouns used in such constructions as base forms. (2)

FIL: skol’ko *petuˇski? how many cock:DIM:NOM:PL ‘How many cocks?’ (correct: petuˇs-kov ‘cock:DIM-GEN:PL’)

Acquisition of case and number in Russian

197

MOT: a skol’ko voron? and how many raven:GEN:PL? ‘And how many ravens?’ FIL: a gde vtoraja *voron? and where another:NOM:SG:FEM raven:GEN:PL ‘Where is the second raven?’ (correct: voron-a ‘raven-NOM:SG’) Five-member miniparadigms are very rare in our data. One of the reasons is that it is improbable that a child would use one and the same word in so many different contexts within one and the same recording or several recordings within one month. For the same reason, such paradigms are also rare in the input (in the three boys’ recordings it occurs only with the word maˇsina ‘car’). The number and types of forms to be expected in nominal paradigms is not only language specific but also limited by the topic of conversation, its length, and the different activities in which a child may be involved. At the end of the observational period, nine months from the onset of case oppositions, children more frequently produce miniparadigms of nouns belonging to unproductive 3rd DCs (e.g. dver’ ‘door:NOM:SG’ which is used in all possible inflectional forms, namely dver-i ‘door-NOM:PL’ and dverj-u ‘doorINSTR:SG’; also compare the unproductive plural porosj-ata ‘pig-NOM:PL’ from porosj-onok ‘pig-NOM:SG’ with a suppletive plural suffix). To sum up, the first oppositions to be observed in the transitional phase between pre- and protomorphology are characterized by a low degree of productivity, i.e. children create first contrastive forms, but these oppositions are not yet integrated into a system of case paradigms. Generally, early case oppositions develop in two directions: first, after the emergence of miniparadigms, the number of lemmas with miniparadigms steadily increases and the inflectional spurts are observed; second, lemmas with four and more members in the paradigms increase. Four to five months after the onset of case oppositions, the growth of lemmas with case oppositions slows down or stops (the ‘plateau periods’) and the qualitative development is activated. The qualitative changes embrace the process of entering of semantic cases into the paradigm and the emergence of new meanings of the cases.

4.2.

Number oppositions

Number oppositions emerge together with the first case oppositions in Filipp, Liza and Vitja. Liza uses the first number oppositions in the nominative with

198

Natalia Gagarina and Maria Voeikova

the nouns jagody ‘berries’ and griby ‘mushrooms’ at 1;9. In Vitja, the first number opposition occurs at 2;2 with the noun kamuˇsk-i ‘stone:DIM-NOM:PL’. The number of lemmas exhibiting number oppositions of a given case form is not high in general (22 lemmas with Liza, 23 lemmas with Vanja, 9 lemmas with Vitja, 16 lemmas with Filipp), but they remain stable during the whole observational period. Also, all children correctly use singular-plural oppositions and exhibit a sharp spurt in the development of number oppositions one or two months after they first occur. The first plural forms typically occur with nouns which are usually used in the plural because they refer to a pair or a multitude of things (e.g. glaz-a ‘eyeNOM:PL’, jagod-y ‘berry-NOM:PL’). Thus, Filipp uses the memorized form grib-y ‘mushroom-NOM:PL’ at 1;8. In the same recording Filipp produces the plural form af-y ‘dog-NOM:PL’ of the onomatopoetic word af in an appropriate context (seeing two dogs in a picture book). This use may be considered as evidence for the cognitive and linguistic maturation that enables the boy such productive plural formation. There is more evidence for the productive use of number formation in the next recording from the same month where Filipp contrasts the singular form miˇsk-a ‘bear:DIM-NOM:SG’ with the plural miˇsk-i ‘bear:DIM-NOM:PL’ referring to one and two teddy bears respectively. Furthermore, there is a kind of plural agreement found at 1;9 when Filipp says miˇsk-i bjak-i ‘bear:DIM-NOM:PL bad-NOM:PL’. Acquisition of the plural is probably facilitated by the fact that 1) the category of number is semantically more transparent than case and 2) that the nominative plural is expressed more regularly than case, since the nominative plural suffix –i occurs in stressed as well as in unstressed syllables. In the next recording at 1;10 one more piece of evidence for the productive formation of the plural occurs when Filipp says *kafik-i instead of oduvanchik-i ‘dandelion-NOM:PL’ adding the plural suffix to the truncated stem used in the singular. At the same age he correctly produces the unproductive form derevj-a ‘tree-NOM:PL’ with the less frequent inflectional ending -a and the additional secondary plural suffix -j-. The frequency and regularity of the plural ending –i(y) has an impact on acquisition; there are several overgeneralizations of this ending as in the example *glaz-y instead of glaz-a ‘eye-NOM:PL’, whereas the –a suffix is never overgeneralized in spite of its productivity in colloquial Russian. As shown by the analysis of more than 10,000 parental reports to the Early Intervention Institute (St. Petersburg), girls start to use singular forms of oblique cases at the mean age of 27.8 months but plural forms only at about 29.0. The corresponding mean ages for boys are 31.5 months for case and 31.9 for number (Shapiro and Chistovich 2000). However, the standard deviation for these mean ages is eight to nine months, which points to a great variation in biological age

Acquisition of case and number in Russian

199

for both processes. As far as the reliability of parental reports is concerned, it is important to take into consideration that parents will not take notice of given case or number forms as soon as they emerge in their children’s speech but only when their usage of grammatical oppositions has reached a much more advanced level. We may therefore interpret the findings of the Saint-Petersburg study as pointing out the age limits of a protomorphological phase. According to this study, girls express quantity lexically at an average of 21.3 months and boys at 24.8 by the equivalents of ‘more’ or ‘one more’ etc. before marking number inflectionally, i.e. before contrasting singular and plural forms of nouns. Another problem in the acquisition of number in the nominal domain is the form of noun phrases containing a numeral, where the noun is in the GEN:SG or GEN:PL depending on the numeral used (see example 2 above). Errors in this domain are frequent with all Russian children and are found until the age of five years, e.g. tri *grib-ov ‘mushroom-GEN:PL’ instead of tri grib-a ‘three mushroom-GEN:SG’ and sem’ *grib-a ‘mushroom-GEN:SG’ instead of sem’ grib-ov ‘seven mushroom-GEN:PL’. As mentioned in section 3, our data show marked learner-specific strategies that are manifested in the correlation between the acquisition of cases and the development of other components of grammar and the lexicon. While Liza constructs shorter utterances with a finite verb form or a noun with varying inflections, Vanja produces longer utterances comprising infinitives and base forms (nominative). The different strategies used by these two children show that either morphology or syntax is elaborated at the cost of the other domain so that a less complex syntactic structure is ‘compensated’ by a richer morphology or vice versa. The reason for this is that the young child is simply unable to construct both morphologically and syntactically complex utterances. Also, the rich inflectional morphology of Standard Russian comprises grammatically required but sometimes semantically redundant marking (e.g. in agreement). In their syntactically less complex utterances children will get along with less of this. Furthermore, ‘argument poverty’ may facilitate case marking so that the acquisition of case is accelerated. Since little children basically communicate about the here and now with members of their families so that these can also rely on pragmatic factors for interpreting children’s utterances, missing syntactic complexity will usually not affect comprehensibility.

200

Natalia Gagarina and Maria Voeikova

4.3.

Case-number oppositions in relation to the acquisition of new noun lemmas

There is an interesting discrepancy to be noticed between the development of the lexical and morphological levels of the children’s language studied in this paper. Our observations show that the number of new lemmas learned by a child in a certain period of time and the number of grammatical oppositions of nouns used are related by a kind of inverse proportion. In order to reveal the nature of this relation we determined the percentage of new noun lemmas in the entire set of nouns occurring in the monthly data of the children and compared it with the percentage of nouns used in grammatical oppositions. Figures 2 and 3 represent the results of this analysis for one pair of children each from the onset of speech production to the end of the observational period. Liza and Filipp produce their first words several months before the first grammatical oppositions occur and at the point when the percentage of new lemmas reaches 90% of all nouns (see figure 2). The detailed diary of Liza shows that her first four words appear at age 1;0 followed by an intensive increase of the lexicon in the following six months. Filipp starts with three words at the age of 1;4. By the onset of contrastive forms, the lexicon of each of the children contains

100 90 80 70 60 50 40 30 20 10 0 1;7

1;8

1;9

L-new lem

1;10

1;11 L-opp

2;0

2;1

2;2

FIL-new lem

Figure 2. New lemmas and oppositions in Liza and Filipp

2;3

2;4 F-opp

Acquisition of case and number in Russian

201

more than two dozen nouns, with 90% of them constituting new lemmas having been acquired in that very month. As the children’s lexicons grow it is natural that the percentages of new noun lemmas decrease reaching about 30% by the end of the observational period in the first half of the second year. While the percentages of nouns exhibiting grammatical oppositions develop similarly in both children between ages of 1;11 and 2;3, important differences are observed in the earlier months from 1;8 to 1;10 (see figure 2). While nouns used with contrasting forms reach 20% of lemmas by 1;10 in Liza’s speech, they slightly rise above 5% in Filipp’s so that almost no morphological development is registered in the first four months after the emergence of grammatical oppositions at 1;7. After these early months, from 1;10 to 2;2, Filipp’s data exhibit an inverse relation between the percentage of new lemmas and lemmas used in contrasting forms with the curve depicting morphological development rising and the one representing lexical development descending. Although this tendency can also be observed in Liza’s data, it is not as pronounced. 100 90 80 70 60 50 40 30 20 10 0 1;10 1;11

2;0

2;1

V-new lem

2;2

2;3 V-opp

2;4

2;5

2;6

2;7

Vi-new lem

2;8

2;9

2;10

Vi-opp

Figure 3. New lemmas and oppositions in Vanja and Vitja

The late talkers Vanja and Vitja show similar tendencies concerning the relation between the lexical and inflectional development of nouns, although their production of contrastive forms starts only at the ages of 2;0 and 2;2 respectively when more than 70% (Vanja) and 60% (Vitja) of all noun lemmas are new ones (see figure 3). The percentage of lemmas exhibiting oppositions develops in a more pronounced way within the first two months after their emergence and

202

Natalia Gagarina and Maria Voeikova

progresses somewhat more slowly later on, with a decrease in the last month of observation for both children. As far as the size of paradigms is concerned, twoand three-member ones are the most frequent. Lemmas with larger paradigms of four to six members represent sparse instances of irregular nouns occurring several months after the first two-member miniparadigms. We consider twomember paradigms to be productive as soon as children use at least 30% of noun lemmas in oppositions (the same percentage as in the input), all grammatical cases occur in their speech and contrastive forms are registered with lemmas of various declensional classes. All four children exhibit a period (of several months) when the number of case-number oppositions and of new lemmas shows a reverse proportion/relation: the strong increase in the number of case-number oppositions is accompanied by a decrease in new lemmas, thus suggesting ‘a shift of acquisitional attention’ between lexicon and morphology.

4.4.

Percentage of Base Forms (PBF)

In the preceding sections of this paper, where we were concerned with grammatical oppositions, types rather than tokens of inflectional forms were counted. In order to study the use and syntagmatic role of inflectionally marked forms of nouns in child speech we determined the Percentage of Base Forms (henceforth PBF) measured in tokens as proposed by Voeikova and Gagarina (2002). This measure describes the proportion of noun forms marked for case and number in a way which contrasts with the respective inflectional ‘base form’ of nouns and therefore demonstrates the degree of inflectional diversity to be found in child speech. PBF can be taken to measure the development of case-number distinctions and therefore the inflectional development of nouns. The base form corresponds to the least marked, default form of nouns in standard speech and is often used and overused by children in the premorphological and protomorphological phases. Base forms must be defined for each language separately. In many languages, they will correspond to the nominative singular. This is the case for Russian, where the nominative singular is unmarked (has a zero inflectional ending) in most masculine and some feminine nouns. The base form of most feminine nouns, however, ends in a salient –a (NOM:SG). In the premorphological phase, children use the nominative singular throughout as in the following example: mam-a kniˇzk-a cˇ itat’‘mommy-NOM:SG book:DIM-NOM:SG read:INF’ (Gvozdev 2007: 165). In calculating PBF measures for child speech only forms distinctly marked for case and number, both correct and deviant ones, were counted as differing

Acquisition of case and number in Russian

203

from the base form. Inflectional forms homonymous with the nominative singular (e.g. ACC:SG of inanimate masculine nouns) were also counted as base forms. Reduced forms whose case could not be recognized were excluded. When children exclusively use the base form of nouns PBF will reach the maximum of 100%. Since some forms are reduced PBF only amounts to about 90% in the early recordings. It is important to note that depending on the language and language type PBF values may differ for different declension classes (see Stephany and Christofidou, this volume). In DCs in which most case forms are differently marked, PBF will be lower than in those that contain more forms that are homophonous with the nominative singular. In Russian both DC1 and DC2 have 5 different forms for 6 cases in the singular, whereas in DC3 (feminine nouns ending with a consonant) only 3 such forms are distinguished (dver’‘door:NOM/ACC’, dver-i ‘door-GEN/DAT/PREP’, dver-ju ‘door-INSTR’). Taking a hypothetical example of ten tokens each of two nouns belonging to DC1 and DC3 respectively, with case functions evenly distributed between NOM and ACC, PBF for DC1 noun tokens will be 50% as compared to 100% for DC3 tokens. Furthermore, a crucial difference between DC1 and DC2 consists in the fact that inanimate nouns belonging to DC2 do not distinguish the NOM from the ACC, whereas in DC1 this difference is consistently made. However, for the purposes of this study we did not calculate PBF separately for different DCs. While a PBF value of 30% is typical of adult-directed Russian, parents tend to use the base form of nouns in 50% of noun tokens occurring in their speech addressed to young children.This is partly due to the simplified register used with children and partly to the fact that in colloquial adult-directed Russian NOM is used instead of oblique case forms under a variety of conditions described by Krasil’nikova (1990: 35–53). In view of these considerations we would expect that, in the course of development, child language should approximate the PBF input value of 50% and gradually drop to the value of 30% typical of standard Russian. Figure 4 illustrates the values of PBF of the four children during a period of nine months from the onset of form production. In the beginning the maximum PBF value is about 90% forVitja (89%), Filipp (91%) and Vanja (95%) but only 73% for Liza dropping to less than 50% by the end of the observational period.The low levels of both the PFB maximum of 73% and its minimal value of less than 50% in the girl’s speech can be explained in several ways. Liza is one of those children who restrict their early utterances to one or two correctly inflected nouns and starts to use all oblique case markers within a single one month (at about 1;9) when her MLU (words/utterance) is about 1.1, e.g. a didik-i (instead of ot deduˇsk-i) ‘from grandfather:DIM-GEN:SG’ (1;10),

204

Natalia Gagarina and Maria Voeikova PBF

100 90 80 70 60 50 40 30 20 10

Filip

Liza

Vanja

9

8

7

6

5

4

3

2

1

0

Vitja

Figure 4. Percentage of base forms in the speech of the four children

a Liziˇck-i (instead of dlja Lizoˇck-i) ‘for Liza:DIM-GEN:SG’ (1;10), a majak-am (instead of za molok-om) ‘for milk-INSTR:SG’ (1;10), i kipinet’k-am (instead of po stupeneˇck-am) ‘up the step:DIM-DAT:PL’ (1;11). It is only a month later, at the age of 2;0, that the syntactic domain of her language is enriched developing from holophrastic to multiword utterances; e.g. zajˇcik beˇzit po doroˇzk-e ‘ a/the hare:DIM:NOM:SG runs along path:DIM-DAT:SG’. As far as the boys’ development of PBF is concerned, the age difference between Vanja and Vitja on the one hand and Filipp on the other must be taken into consideration. Thus, the decreasing PBF curve is steeper in the two older boys dropping from about 90% to a final 50% already after three or four months, while Filipp’s curve slopes much more gently reaching a final value above 60%. Although the three boys start with very high PBF values in comparison to Liza, Vitja’ and Filipp’s development differs from that of Vanja. Both Vitja’s and Filipp’s early utterances contain one or more base forms of nouns, e.g. ig-a na ‘book-NOM:SG take’ (= take the book) (Filipp, 1;7, handing a book to his mother); papa sok dzz’ ‘father:NOM:SG juice:NOM:SG dzz’:ONO’ (= father (brought) juice) (Vitja, 2;2). Vanja’s early speech of the same developmental period looks different. Most of his utterances containing a noun are holophrastic with the noun in the base form. Inflected forms of nouns are either plural (far-y ‘headlight-NOM:PL’, 2;1), or instrumental singular case forms referring to accompanying persons (mam-ij instead of mam-oj) ‘mommy-INSTR:SG’, pap-ij (instead of pap-oj) ‘daddy-INSTR:SG’ 2;2). These forms not only lack the obligatory prepositions, but also have an incorrect ending with an ensuing

Acquisition of case and number in Russian

205

automatic palatalization of the final stem consonant. With the three other children the instrumental only appears after some of the other oblique cases or at least the accusative have emerged so that the order of emergence of cases in Vanja’s speech is unusual. For some reason he has a habit of picking up instrumental forms from the input using them to answer the imaginary question ‘with whom?’ Unlike Liza’s referential style and the expressive style of both Filipp and Vitja, Vanja enriches his morphological inventory by a small number of rotelearnt case forms that slowly become productive. This shows that children may differ in their styles or strategies of using early case marking and that these differences are not necessarily connected with an earlier or later onset of speech production. One such strategy consists in using correctly marked syntactic arguments in correspondingly short sentences (Liza) and another one to form syntactically more complex sentences containing several inflectionally unmarked base forms of nouns (Filipp and Vitja). These two strategies of grammatical development may be called morphological and syntactic respectively. A third way to approach an adult-like percentage of inflectionally marked and unmarked forms consists in using only certain inflectional markers as a reaction to the frequent questions of adult interlocutors or in playing situations (Vanja). In the previous case study of two children (Voeikova and Gagarina 2002) we found that Filipp and Vanja reached their minimal PBF values at different age levels but at approximately the same MLU level of 1.8 words/utterance. Extending this analysis to the data of the other two children of the present paper we found that this correlation does not hold generally. While Vitja reaches his lowest PBF level at an MLU of 2.0, Liza already attains it when her MLU is only about 1.2. Rather than being directly related to PBF, MLU values reflect children’s differing strategies in the acquisition of grammar in their own right (see figure 5). Considering her ‘morphological strategy’ of language development, it is to be expected that in the beginning Liza has a low MLU level. After little syntactic development in the first four months Liza progresses considerably in the next four months, leaving both Vanja and Filipp behind and approaching Vitja’s MLU in the eighth month. According to his expressive style, this boy has the highest MLU value of the four children, especially so from the fifth month onward. As is to be expected, Liza’s and Vitja’s preferences for morphological vs. syntactic marking not only influence their respective PBF values but their MLU as well. Filipp’s expressive style on the one hand and the fact that he is an early talker on the other may account for his intermediate MLU values. His young age and corresponding cognitive immaturity may explain that his MLU curve is the most sloping one, giving evidence of more syntactic progress only from the sixth month on. Although Vanja’s MLU curve rises from low values to 2.0 already by the third month, his ensuing syntactic progress is slow. Interestingly,

206

Natalia Gagarina and Maria Voeikova

3 500 3 000 2 500 2 000 1 500 1 000 500 0 1

2

3 Filip

4

5 Liza

6 Vanja

7

8

9

Vitja

Figure 5. MLU (words/utterance) of the four children

his PBF value abruptly drops from about 90% to less than 50% during the same period (see figure 4), so that this child is elaborating the syntactic complexity of utterances and morphological marking of arguments at the same time. However, this period of intensive grammatical development is rather short and is followed by a long plateau period. An important result of our analyses presented in this section is that PBF and MLU values relate not statistically but in quite complex ways demonstrating children’s differing strategies of grammatical development.

5. Discussion and conclusion To sum up, the general aim of this study was to trace the emergence and development of case and number oppositions in the longitudinal data of four children and to establish the criteria of their qualitative development, i.e. the establishment of developmental phases and the transitions between them. Our results were obtained by three sets of analysis. After examining the development of case and number oppositions, the relation between the emergence of new lemmas and oppositions was scrutinised, turning finally to the development of PBF and MLU and their interrelation. Findings show the generally similar pace of the devel-

Acquisition of case and number in Russian

207

opment of oppositions for the four children in spite of the fact that they follow different strategies and go through the same stages in different biological ages. Number oppositions in the nominative either occur together with or subsequently to the first case oppositions in the singular. Thus, the question whether case or number is first acquired by Russian children remains unclear. Similarly to Zhenja Gvozdev, the children under observation start to use the first case and number forms almost simultaneously. Their first plural forms are limited to the nominative, whereas plural forms of oblique cases only occur two to four months later. We found that the acquisition of case oppositions corresponds to inflectional classes of nouns. Consistent case marking and oppositions of case forms signals that the use of inflectional endings has become (fully) productive. In order to gain a deeper understanding of the rather long and heterogeneous process leading to this point, we will briefly retrace its initial phases studied in the present paper. Although the entire spectrum of the six Russian cases emerges in the singular in a short period of time, their productive use evolves to different degrees so that several degrees of productiveness may be theoretically distinguished. At the lowest level – the lowest degree of productiveness – the child correctly uses a given lemma in two inflected forms in different contexts. At the next level – the next degree of productiveness – a given case form occurs with at least three lemmas of various declensional classes. This level corresponds to the qualitative changes in the number of lemmas with oppositions, e.g. an inflectional spurt. The highest level of productiveness has been attained when the child overregularizes certain case forms and when various types of morphophonemic markings of a given case form occur. Further, various types of morphophonemic markings and the contrastive forms construct a system of cases that approaches the target Russian one. Apart from scrutinizing the development of productiveness it seems necessary to establish solid criteria for the transition from the pre- to the protomorphological phase for gaining deeper insights into the process of inflectional development. These criteria must take into consideration the following factors: the number of lemmas exhibiting oppositions in child speech and the inflectional classes they belong to as well as the correspondence of the size and number of declensional paradigms in children compared to the input. A most appealing result of the present study concerns the interrelations among the lexical, morphological, and syntactic levels of language development. Rather than developing linguistic skills embracing all three levels simultaneously, children show a kind of ‘compensatory tendency’ or ‘shift of acquisitional attention’ demonstrating a spurt in one domain (e.g. in the acquisition of new lemmas) accompanied or followed by a plateau period or even a slight regress in other spheres (e.g. in the emergence of the oppositions). A strong

208

Natalia Gagarina and Maria Voeikova

interrelation between the accumulation of new lemmas and morphological oppositions was found in all four children, whereas the relation between syntactic and morphological development showed individual variation between the three boys and the girl. While the girl attained the minimal level of PBF (about 50%) at a time when her MLU was only about 1.2, the three boys reached this PBF value only at an MLU value of 1.8 to 2.0. More research is needed in order to finally solve issues concerning the elaboration of degrees of productiveness and, above all, to determine what is universal rather than language- and learner-specific in the paths leading to the mastery of languages.

Appendix 1 Table 1. Data of Filipp Age 1;7 1;8 1;9 1;10 1;11 2;0 2;1 2;2 2;3 Total

Analyzed Utterances 690 880 761 205 189 320 671 685 567 4968

MLU 1.293 1.629 1.392 1.498 1.767 1.978 2.037 2.276 2.305

Nouns Lemmas 24 28 36 42 36 29 61 50 63 291

Tokens 87 303 363 104 85 204 451 406 420 2423

Nouns Lemmas 33 63 109 106 110 142 158 64 88 873

Tokens 66 153 232 191 202 279 314 111 245 1793

Table 2. Data of Liza Age 1;8 1;9 1;10 1;11 2;0 2;1 2;2 2;3 2;4 Total

Analyzed Utterances 117 279 377 306 268 281 310 120 304 2362

MLU 1.009 1.047 1.202 1.199 1.679 2.249 2.590 2.833 2.638

Acquisition of case and number in Russian

209

Table 3. Data of Vanja Age 2;0 2;1 2;2 2;3 2;4 2;5 2;6 2;7 2;8 Total

Analyzed Utterances 410 471 442 636 625 607 539 439 686 4855

MLU 1.261 1.263 2.138 2.283 2.203 2.054 2.174 2.346 2.633

Nouns Lemmas 39 59 83 120 110 137 121 80 121 870

Tokens 188 237 311 556 434 389 411 214 388 3128

Nouns Lemmas 29 67 62 64 73 67 123 104 114 703

Tokens 165 320 190 201 219 214 608 447 386 2750

Table 4. Data of Vitja Age 2;2 2;3 2;4 2;5 2;6 2;7 2;8 2;9 2;10 Total

Analyzed Utterances 272 387 296 359 315 261 640 373 397 3300

MLU 1.669 2.052 1.885 2.111 2.873 3.222 3.255 3.284 3.486

Notes ∗

We want to express our deep gratitude to Ursula Stephany for her helpful comments on an earlier version of this paper, to John Tammena for improving our English and to Katherina Becker for her help in formatting the chapter. All remaining errors are our own. 1. We would like to thank Ursula Stephany for drawing our attention to these important observations in her comments to a previous version of this paper. 2. Gvozdev used a special kind of graphical transcription; therefore Zhenja’s words do not look like adult Russian forms. 3. Compare the similar description of Russian declensional types by Corbett (1982: 197–211). In his system neuter nouns are regarded as forming a special declension

210

Natalia Gagarina and Maria Voeikova

class 4 (DC4). Corbett (1991, 2003) demonstrates that, in Russian, gender is not phonologically but morphologically determined, depending on declensional patterns. However, Russian child-directed speech exemplifies rather a simplified and more logical relation between genders and DCs. All nouns ending with –a belong to the feminine gender excluding hypocoristics and some kinship terms like papa ‘daddy’, Nouns ending with a consonant are all masculine, due to the fact that feminine nouns ending with a consonant are mostly used as diminutives and, thus, go to the DC1 (Olmsted 1994, Kempe, Brooks, and Pirott 2001). Therefore, we adopt here a classical scheme of Russian declension that is suitable for child and child-directed speech. 4. Numbers of declension classes in our description differ from those of Zaliznjak (1977) and Russkaja grammatika (1980). In numbering the declension classes we follow the school grammatical tradition according to which feminine nouns ending with –a belong to DC1, whereas masculine nouns ending in a consonant form DC2. In recent grammatical descriptions it is vice versa for purely technical reasons. 5. Stressed syllables are marked by a capital vowel. The pronunciation of unstressed inflectional endings is indicated in square brackets if it differs from the stressed endings. Many case markings are opaque due to the different pronunciation of the inflectional endings –e and –o in stressed and unstressed position. This may be another reason for the later acquisition of these forms by children. 6. We only consider the phonological case forms, since preschool children do not know the written variants in which DAT:SG and PREP:SG are represented differently from GEN:SG and NOM:PL.

References Babyonyshev, Maria 1993 The acquisition of Russian case. In Papers on Case and Agreement II, Colin Phillips (ed.), 1–44. (MIT Working Papers in Linguistics 19.) Cambridge, MA: MIT. Bittner, Dagmar 2002 Emergence of grammatical complexity and markedness in the acquisition of verb and noun phrases in German. In Future Challenges for Natural Linguistics, Katarzyna Dziubalska-Kołaczyk and Jarek Weckwerth (eds.), 25–56. M¨unchen: Lincom. Bittner, Dagmar, Wolfgang U. Dressler, and Marianne Kilani-Schoch 2003 Introduction. In Development of Verb Inflection in First Language Acquisition: A Cross-linguistic Perspective, Dagmar Bittner, Wolfgang U. Dressler, and Marianne Kilani-Schoch (eds.), vii–xxxix. Berlin/New York: Mouton de Gruyter.

Acquisition of case and number in Russian

211

Bittner, Dagmar, Wolfgang U. Dressler, and Marianne Kilani-Schoch (eds.) 2003 Development ofVerb Inflection in First LanguageAcquisition:A Crosslinguistic Perspective. Berlin/New York: Mouton de Gruyter. Blake, Barry J. 2001 Case. Cambridge: Cambridge University Press. Ceitlin, Stella N. 1987 Kategorija cˇ isla v ontogenese [Category of case in ontogenesis]. In Detskaja reˇc’ kak predmet lingvistiˇceskogo izuˇcenija [Child speech as an object of linguistic inquiry], Stella N. Ceitlin (ed.), 34–41. Leningrad: Leningrad Herzen State Pedagogical Institute Press. 1988 Okkazional’nyje morfologiˇceskije formy v detskoj reˇci [Occasional morphological forms in child speech]. Leningrad: Leningrad Herzen State Pedagogical Institute Press. 2000 Jazyk i rebjonok: Lingvistika detskoj reˇci. [Language and Child: Linguistics of Child Speech]. Moscow: VLADOS. Christofidou, Anastasia, and Ursula Stephany 1997 The early development of case forms in the speech of a Greek boy: A preliminary investigation. In Papers and Studies in Contrastive Linguistics, Katarzyna Dziubalska-Kołaczyk (ed.), vol. 33: 127–140. Pozna´n. Clahsen, Harald, Sonja Eisenbeiss, and Anne Vainikka 1994 The seeds of structure: A syntactic analysis of the acquisition of case marking. In Language Acquisition Studies in Generative Grammar, Teun Hoekstra and Bonnie D. Schwarz (eds.), 85–118. Amsterdam/Philadelphia: Benjamins. Corbett, Greville 1982 Gender in Russian: An account of gender specification and its relationship to declension. Russian Linguistics 6(2): 197–232. 1991 Gender. Cambridge: Cambridge University Press. 2003 Types of typology, illustrated from gender systems. In Noun Phrase Structure in the Languages of Europe, Frans Plank (ed.), 289–334. (Empirical Approaches to Language Typology, EUROTYP 20-7.) Berlin/New York: Mouton de Gruyter. Dressler, Wolfgang U. 2001 The emergence of morphology – a constructivist approach. Croatian Review of Rehabilitation Research 37: 22–36. 2007 Introduction. In Typological perspectives on the acquisition of noun and verb morphology, Sabine Laaha and Steven Gillis (eds.), 3–10. (Antwerp Papers in Linguistics 112.)Antwerp: University ofAntwerp. Dressler, Wolfgang U. (ed.) ¨ 1997 Studies in Pre- and Protomorphology. Wien: Verlag der Osterreichischen Akademie der Wissenschaften.

212

Natalia Gagarina and Maria Voeikova

Dressler, Wolfgang U., Danuta Dra˙zyk, Rychard Dra˙zyk, Katarzyna DziubalskaKołaczyk, and Evelina Jagla 1995–96 On the earliest stages of the acquisition of Polish declension. Wiener Linguistische Gazette 53-54: 1–21. Dressler, Wolfgang U., and Annemarie Karpf 1995 The theoretical relevance of pre- and protomorphology in language acquisition. Yearbook of Morphology 1994, 99–122. Dressler, Wolfgang U., Marianne Kilani-Schoch, Natalia Gagarina, Lina Pestal, and Markus P¨ochtrager 2006 On the typology of inflection class systems. Folia Linguistica XL/1-2: 51–74. Eisenbeiss, Sonja 2003 Merkmalsgesteuerter Grammatikerwerb: Eine Untersuchung zum Erwerb der Struktur und Flexion der Nominalphrase. Ph. D. diss. University of D¨usseldorf. (http://deposit.ddb.de/cgi-bin/dokserv?idn= 97646330x&dok var=d1&dok ext=pdf&filename=97646330x.pdf) Eisenbeiss, Sonja, Susanna Bartke, and Harald Clahsen 2006 Structural and lexical case in child German: Evidence from languageimpaired and typically-developing children. Language Acquisition 13: 3–32. Eisenbeiss, Sonja, Bhuvana Narasimhan, and Maria D. Voeikova 2007 The acquisition of case. In The Oxford Handbook of Case, Andrej Malchukov and Andrew Spencer (eds.), 369–383. Oxford: Oxford Univ. Press. Gagarina, Natalia 2000 Early verb development in one Russian-speaking child. In First verbs: on the way to mini-paradigms, Dagmar Bittner, Wolfgang U. Dressler, and Marianne Kilani-Schoch (eds.), 143–162. (ZAS Papers in Linguistics 18.) Berlin. 2002 The early verb development and demarcation of stages in three Russian-speaking children. In Development of Verb Inflection in First Language Acquisition: A Cross-Linguistic Perspective, Dagmar Bittner, Wolfgang U. Dressler, and Marianne Kilani-Schoch (eds.), 131– 169. Berlin/New York: Mouton de Gruyter. 2008 Stanovlenije grammatiˇceskih kategorij russkogo glagola v detskoj reˇci. [First language acquisition of verb categories in Russian]. Saint Petersburg: Nauka. Gagarina, Natalia, Maria D. Voeikova, and Sergej Gruzincev 2003 New version of morphological coding for the speech production of Russian children. In Investigations into Formal Slavic Linguistics, Peter Kosta, Joanna Blaszczak, Jens Frasek, Ljudmila Geist, and Marzena Zygis (eds.), 243–258. Frankfurt etc.: Peter Lang.

Acquisition of case and number in Russian

213

Gvozdev, Alexandr N. 2007 Reprint. Formirovanije u rebenka grammatiˇceskogo stroja russkogo jazyka [Construction of the grammatical system of Russian by the child]. InVoprosy izuˇcenija detskoj reˇci [Questions of the study of child speech], Alexandr N. Gvozdev, 149–459. Saint Petersburg/Moscow: Detstvo-Press, Tvorˇceskij centr Sfera. Original edition, Moscow: Akad. Pedag. Nauk RSFSR, 1949. 1981 Ot pervyx slov do pervogo klassa: Dnevnik nauˇcnyx nabljudenij. [From the first words to the first grade: Scientific diary]. Saratov: Izdatel’stvo Saratovskogo universiteta. Ionova, Natalia V. 2007 Semantiˇceskie funkcii padeˇznyx form I predloˇzno-padeˇznyx konstrukcij imeni suˇscˇ estvitel’nogo v reˇci detej doˇskol’nogo vozrasta [Semantic functions of case forms and prepositional phrases in the speech of preschool children]. Ph. D. diss. Chair of Child Language, ˇ State University of Cerepovec. Jakobson, Roman 1936 Beitrag zur allgemeinen Kasuslehre: Gesamtbedeutungen der russischen Kasus. TCLP IV: 240–288. 1958 Morfologiˇceskie nabljudenija nad slavjanskim skloneniem (Morphological observations on Slavic declension). Paper read at the IVth International Congress of Slavists, Moscow. In: Roman O. Jakobson, Izbrannye trudy (Selected works), 133–238. Moscow: Progress, 1985. Kempe, Vera, Patricia J. Brooks, and Lora Pirott 2001 How can child-directed speech facilitate the acquisition of morphology? In Research on Child Language Acquisition: Proceedings of the 8th Conference of the International Association for the Study of Child Language, Margareta Algren, Andoni Barrena, Mar´ıa-Jos´e Ezeizabarrena, Itziar Idiazabal, and Brian MacWhinney (eds.), 1234–1244. Somerville, MA: Cascadilla Press. Kiebzak-Mandera, Dorota 2000 Formation of the verb system in Russian children. Psychology of Language and Communication 4: 27–46. Krasil’nikova, Elena V. 1990 Imja suˇscˇ estvitel’noe v russkoj razgvornoj reˇci [The noun in spoken Russian]. Moscow: Nauka. Laalo, Klaus 2002 Acquisition of case in Finnish: A preliminary overview. In Pre- and Protomorphology: Early Phases of Morphological Development in Nouns and Verbs. Maria D. Voeikova and Wolfgang U. Dressler (eds.), 83–105. (LINCOM Studies in Theoretical Linguistics 29.) M¨unchen: LINCOM Europa.

214

Natalia Gagarina and Maria Voeikova

Lepskaja, Natalia I. 1988 Osvoenie det’mi kategorii padeˇza [Acquisition of the category of case by children]. In Semantika v reˇcevoj dejatel’nosti na materiale ontogeneza [Semantics in the speech activity on the material of ontogeˇ nesis], Alexandr M. Saxnaroviˇ c (ed.), 48–58. Moscow: AN SSSR, Institute of Linguistics. MacWhinney, Brian 2000 The CHILDES Project: Tools for Analyzing Talk. Vol. 1: Transcription, Format and Programs. 3rd ed. Mahwah, NJ: Lawrence Erlbaum. Mills, Ann E. 1985 The acquisition of German. In The Crosslinguistic Study of Language Acquisition, Dan I. Slobin (ed.), vol. 1, 141–254. Hillsdale, NJ: Lawrence Erlbaum. Nelson, Katherine 1981 Individual differences in language development: Implications for development and language. Developmental Psychology 17: 170–187. Olmsted, Hugh 1994 Diminutive morphology of Russian children: A simplified subset of nominal declension in language acquisition. In Alexander Lipson: In Memoriam. Charles E. Gribble, RobertA. Rothstein, Edythe C. Haber, Hugh M. Olmsted, Robert Szulkin, and Charles E. Townsend (eds.), 165–207. Columbus, OH: Slavica Inc. Savickien˙e, Ineta 2003 The Acquisition of Lithuanian Noun Morphology. Wien: Verlag der ¨ Osterreichischen Akademie der Wissenschaften. Slobin, Dan I. 1966 Acquisition of Russian as a native language. In The Genesis of Language: A Psycholinguistic Approach. Proceedings of a Conference on “Language Development in Children”, Frank Smith and George A. Miller (eds.), 129–148. Cambridge, MA/London: MIT Press. 1985 Crosslinguistic evidence for the language-making capacity. In The Crosslinguistic Study of Language Acquisition, Dan I. Slobin (ed.), Vol. 2, 1157–1256. Hillsdale, NJ: Lawrence Erlbaum. ˇ ˇ Sapiro, Jakov N., and Inna A. Cistoviˇ c 2000 Rukovodstvo po ocenke urovnja razvitija detej ot 1 goda 2 mesjacev do 3 let 6 mesjacev po rusificirovannoj sˇ kale RCDI-2000 [A Handbook for the Estimation of the Development of Children from 1 year 2 months to 3 years 6 months with the RCDI-2000 Scale adapted to Russian]. Saint-Petersburg: Institute of Early Intervention. Smoczy´nska, Magdalena 1985 The acquisition of Polish. In The Crosslinguistic Study of Language Acquisition. Dan I. Slobin (ed.), vol. 1, The data, 595–686. Hillsdale, NJ: Lawrence Erlbaum.

Acquisition of case and number in Russian

215

Stephany, Ursula 1997 The acquisition of Greek. InThe Crosslinguistic Study of LanguageAcquisition, Dan I. Slobin (ed.), vol. 4, 183–333. Mahwah, NJ/London: Lawrence Erlbaum. 2006 A cognitive-functional approach to the development of number and case in early child Greek: Lexically based constructions vs. abstract ‘symbolic’ rules. Paper read at the Second Biennial Conference on Cognitive Science, June 9–13, 2006, Saint Petersburg, Russia. Tomasello, Michael 2003 Constructing a Language: A Usage-Based Theory of Language Acquisition. Cambridge, MA/London: Harvard University Press. Voeikova, Maria D. 1997 Extragrammatical operations in the speech of a Russian child. In Studies in Pre- and Protomorphology, Wolfgang U. Dressler (ed.), 37–45. (Ver¨offentlichungen der Kommission f¨ur Linguistik und Kommunika¨ tionsforschung 26.) Wien: Verlag der Osterreichischen Akademie der Wissenschaften. 2000 Russian Existential Sentences. M¨unchen: LINCOM Europa. 2002 The acquisition of case in typologically different languages. In Preand Protomorphology: Early Phases of Morphological Development in Nouns and Verbs, Maria D. Voeikova and Wolfgang U. Dressler (eds.), 25–45. (LINCOM Studies in Theoretical Linguistics 29.) M¨unchen: LINCOM Europa. Voeikova, Maria D., and Wolfgang U. Dressler (eds.) 2002 Pre- and Protomorphology: Early Phases of Morphological development in Nouns andVerbs. (LINCOM Studies in Theoretical Linguistics 29.) M¨unchen: LINCOM Europa. Voeikova, Maria D., and Natalia Gagarina 2002 Early syntax, first lexicon and the acquisition of case forms by two Russian children. In Pre- and Protomorphology: Early Phases of Morphological Development in Nouns and Verbs. Maria D. Voeikova and Wolfgang U. Dressler (eds.), 115–133. (LINCOM Studies in Theoretical Linguistics 29.) M¨unchen: LINCOM Europa. Zaliznjak, Andrej A. 1977 Grammatiˇceskij slovar’ russkogo jazyka [Grammatical dictionary of the Russian language]. Moscow: Russkij jazyk.

The emergence of nominal inflection in Greek∗ Ursula Stephany and Anastasia Christofidou 1. Introduction Children are sensitive to the typological properties of the language they acquire from early on (Slobin 1997; Laaha and Gillis (eds.) 2007). Those growing up with highly inflected languages such as Modern Greek will frequently hear different grammatical forms of a given lexeme used in different grammatical and semantic-pragmatic contexts. In spite of the fact that the Greek noun is not as highly inflected as the verb, acquisition of nominal inflection of this inflecting-fusional language is quite complex, comprising the three categories of case, number, and gender. As is usual in this type of language, the formation of case-number forms obeys different patterns that apply to largely arbitrary classes of nominal lexemes partially based on gender. Further, frequency of the occurrence of the three gender classes and case-number forms of nouns greatly differs in spoken Greek, regarding both the types and tokens. In contrast to a child acquiring a typically agglutinating language such as Turkish, where nouns form a single inflectional class, a child learning an inflecting-fusional language like Greek must construct different inflectional patterns depending not only on parts of speech but also on subclasses within a given part of speech, such as gender classes of nouns and inflectional classes within or (exceptionally) across genders. It is therefore to be expected that the early development of case and number distinctions will apply to specific nouns and subclasses of nouns rather than the totality of Greek nouns. The two main theoretical approaches of morphological development that will be discussed in the present paper are the usage-based approach and the pre- and protomorphology approach.

1.1. The usage-based approach to language acquisition In order to capture the process of inflectional development of the noun by the analysis of observational data of an inflecting-fusional language like Greek, the grammatical forms and functions of the individual nouns found in each child’s speech must be studied in detail so that the classes of nouns sharing inflectional properties can be established and patterns of use emerge. Usage-based cognitive-functional theory of language and language acquisition according to

218

Ursula Stephany and Anastasia Christofidou

which “language structure emerges from language use” (Tomasello 2003: 5; see also Bybee 2001), seems particularly suitable for capturing the development of nominal inflection in an inflecting-fusional language exhibiting many distinct inflectional patterns. Another advantage of this approach for describing the structure and acquisition of languages of this morphological type is that affixes are taken to emerge from associations between inflected word forms so that “the stems and affixes . . . are never extracted from the word in which they occur” (Bybee 2001: 24). Since grammatical generalizations are based on particular forms they “can only be emergent patterns, not explicit rules” and are best described by schemas, which are “organizational patterns in the lexicon” (Bybee 2001: 26). Schemas are “highly affected by the number of participant items” so that “productivity is gradient”; they are furthermore “highly affected by the particulars of existing types” so that they apply to subclasses rather than to entire classes of parts of speech (Bybee 2001: 27). As will be shown in the present paper, the development of Greek nominal inflection is more adequately captured by emergent generalizations or schemas than by across-the-board algebraic or symbolic rules (see also Christofidou 2004; Stephany 2006). In a highly inflected language like Greek, inflectional patterns exhibit different degrees of productivity including highly productive and totally unproductive ones. A morphological model such as the dual mechanism or dual process approach to language acquisition of generative grammar (e.g. Pinker 1991; Clahsen 1999), establishing a mere dichotomy between regular and irregular forms, where the former are considered to be generated by rules while the latter are listed in the lexicon, is thus inadequate for capturing morphological storage and processing in a language of this type. As far as frequency of occurrence is concerned, token frequency must be distinguished from type frequency. While the token frequency of grammatical word forms favors rote learning and thus entrenchment, type frequency, “the number of different word forms that are instances of a particular schema” (Croft and Cruse 2004: 296), favors pattern construction (“rule formation”) in language acquisition (Tomasello 2003: 235). Since “the primary factor determining the existence of a schema . . . is a (relatively) high type frequency” (Croft and Cruse 2004: 302), high type frequency is of prime importance for the development of inflectional patterns by the child.

1.2. The pre- and protomorphology approach Most of the empirical work carried out within the “Crosslinguistic Project on Pre- and Protomorphology in Language Acquisition” (coordinated by Wolf-

The emergence of nominal inflection in Greek

219

gang U. Dressler, Austrian Academy of Sciences, Vienna) is based on or at least compatible with the model of Natural Morphology (e.g. Dressler (ed.) 1997; Voeikova and Dressler (eds.) 2002; Bittner, Dressler, and Kilani-Schoch (eds.) 2003). This non-nativist, functional approach pays special attention to the influence of language typology on language acquisition. It takes acquisition of morphology in children studied from the onset of speech through the end of their third year to proceed in three subsequent stages (also called ‘phases’),1 namely premorphology followed by protomorphology and finally ‘morphology proper’. So far, research has focused upon the first two of these stages. Although, as in most other theoretical approaches, language acquisition is considered to start with item-based learning (Bittner, Dressler, and Kilani-Schoch 2003: vii), ‘morphological operations’ and later on grammatical rules are stressed. Thus, the “premorphological phase” is defined “as the phase where only few morphological operations occur – both extragrammatical (or ‘expressive’) ones, such as reduplications . . . and precursors of later grammatical rules” (Voeikova and Dressler 2002: 3). In the “protomorphological phase”, “the system of morphological grammar and its subsystems start to develop” so that morphological patterns begin to be creatively constructed (Voeikova and Dressler 2002: 3–4). One type of evidence that a child has entered this second stage of morphological development is the emergence of ‘miniparadigms’ (see sect. 4 below). The emphasis on morphological operations and grammatical rules partially contrasts with the usage-based approach described above. While the pre- and protomorphological approach postulates different stages of morphological development finally leading to a morphological component of grammar governed by symbolic rules, an approach following the basic tenets of usage-based theory considers morphological acquisition to consist in the construction of schemas of different degrees of generality and abstractness. The latter approach therefore does not only allow researchers to keep track of the specific linguistic forms from which schemas emerge and upon which they rely, but also makes it unnecessary to assume ‘turning points’ in linguistic development leading to a qualitatively different stage. Instead, the process of language acquisition is envisaged as consisting of “smoothly gliding developmental phases” (Christofidou and Stephany 2003: 117).

1.3.

Goals and overview of the chapter

In the present chapter, we will trace the early development of inflection in the noun (with some outlooks on the definite article) in five monolingual Greek children, who were studied in the period from the second half of their second to

220

Ursula Stephany and Anastasia Christofidou

the last part of their third year. It will be shown that the question whether case or number distinctions of the Greek noun develop first (Christofidou 1998) is not a fruitful one, since these categories partially depend on gender and declensional classes of nouns and thus develop locally. Our study will therefore also contribute evidence to the much-debated question of whether, in Modern Greek, gender determines inflection class or vice versa (see sect. 2). Moreover, the role played by gender in the early development of case distinctions, which was studied in a Greek boy during the period between the end of his second through the first half of his third year by Christofidou and Stephany (1997) and Christofidou (2003), will be further examined in this subject as well as four other ones. After a sketch of Modern Greek nominal inflection (sect. 2), the data analyzed in the present study will be presented (sect. 3). The first part of the analysis (sect. 4) will consist in quantitative approaches to the development of case, number, and gender of the Greek noun in child and input data (with a comparison between child-directed and adult-directed speech). After measuring the onset and development of noun inflection in a global way by determining the number of grammatical types per lexeme (mean size of paradigm) (sect. 4.1), this development will be described in more detail by tracing the emergence of grammatical contrasts and the formation of paradigms in different classes of nouns (sect. 4.2). The role played by the tokens of inflectionally marked forms of nouns in child and child-directed speech will be determined by applying the parameter of Percentage of Base Forms (PBF) to our data (sect. 4.3). The quantitative study will be completed by a functional analysis of case and number forms of nouns and definite noun phrases in the children’s speech (sect. 5). Finally, some theoretical conclusions will be drawn from our results (sect. 6).

2. Greek nominal inflection The Greek nominal system comprises the three grammatical categories of gender, case, and number. It distinguishes three genders (masculine, feminine, and neuter), four cases (nominative, genitive, accusative, and vocative), and two numbers (singular and plural). Two main proposals for the description of the Greek declensional system coexist until the present day: one is primarily based on gender (e.g. Triantafyllidis [1941] 1978; Seiler 1958; Christofidou 2003) and the other one on the distribution of inflectional endings (e.g. Kourmoulis 1964; Mackridge 1985; Klairis and Babiniotis 2005). The gender-driven approach is supported by diachronic arguments (Seiler 1958)2 as well as synchronic ones (Christofidou 2003). The main synchronic criteria advanced by Christofidou (2003) are inflectional productivity

The emergence of nominal inflection in Greek

221

and the morphological integration of loan words or neologisms (see Dressler 1997). The latter show that inflection is determined by gender assignment rather than the phonological structure of noun endings. As will be shown by acquisition evidence, inflectional endings and declensional patterns of nouns have to be considered as overt markers of gender classes, since the development of nominal inflection largely proceeds within gender classes. Gender assignment may depend on semantic or formal information (Corbett 2003). As is common in Indo-European languages, the Greek gender system comprises a more or less small core of semantic gender assignment based on animacy and sex distinctions (especially of human nouns) with an ensuing interdependence of case and gender (Lyons 1968: 293–294; Corbett 2003: 311). As noted by Stephany (1997a), the interdependence of case and animacy is “especially noticeable in early child Greek” (p. 220) so that the syntactic functions of arguments are mostly unambiguous even in the absence of subject and object marking (pp. 220–221). In Standard Greek, large numbers of nouns are not covered by semantic assignment rules but by formal information (Setatos 1998; Ralli 2002; Christofidou 2003: 117–124). Modern Greek gender is not phonologically determined, however, since it cannot be assigned on the basis of a single form of the noun, but is based on more than one inflected form (on a similar situation in Russian see Corbett 2003: 312–316). Examples such as x´eri ‘hand:NEUT’ vs. m´ıti ‘nose:FEM’ demonstrate that Greek gender assignment is not based on the final vowel of the inflectionally unmarked singular case form. Neither is it determined by the final stem vowel as shown by examples such as pap´u ‘grandfather:MASC:OBL:SG’ (pap´us ‘NOM:SG’) vs. maim´u ‘monkey:FEM:NOM/ ACC:SG’. What is relevant instead is the inflectional type of the noun. While pap´us is inflected according to the pattern of ‘diptota’ masculine nouns (pap´us ‘MASC:NOM:SG’ vs. pap´u ‘MASC:OBL:SG’) (see below), the noun maim´u follows the pattern of feminine nouns (maim´u ‘FEM:NOM/ACC:SG’vs. maim´us ‘FEM:GEN:SG’), so that pap´us is masculine whereas maim´u is feminine. Although a given Greek gender class may comprise more than one declensional type (e.g. ‘diptota’ and ‘triptota’ masculine nouns, see below), the reverse does not hold since, with the exception of the regressive class of feminine nouns ending in –os (e.g. i ps´ıfos ‘the:FEM:NOM:SG vote:FEM:NOM:SG’), no declensional type comprises more than one gender (see Christofidou 2003: 114–115). Because of the strong interrelation between the declensional type of the noun and its gender, the grammatical category of gender emerges interdependently with the particular case-number forms of nouns in the acquisition of Greek. The distribution of the three genders in Greek texts is by no means uniform. In a representative corpus of oral and written text analyzed by Kavoukopoulos

222

Ursula Stephany and Anastasia Christofidou

(1996: 10) gender distribution in (97% of) nouns is the following: 42% feminine, 32% neuter, and 23% masculine. According to a much smaller corpus of 600 randomly selected nouns (see Mackridge 1985: 52 following Mirambel 1959: 84) the percentage of neuter nouns exceeds that of feminine ones, but masculine nouns are again the least frequent ones. As will be shown in section (4.2), neuter nouns are the most frequent ones in both early child Greek and child-directed speech followed by feminine and finally masculine ones. The nominative may be considered as basic in Greek since it represents the citation form of nouns and is used to express the grammatical subject as well as predicative nouns referring to the subject. It also occurs in verbless phrases like (na) o Gi´orgos ‘(there is) George’, in signatures, inscriptions etc. (see Mackridge 1985: 55; Tzartzanos [1946] 1991: 82). In contrast to the nominative, which is marked by final –s in the nominative singular of masculine nouns (e.g. o a´ nthropos ‘the:MASC:NOM:SG man:MASC:NOM:SG’), the accusative is unmarked in the singular of all three genders so that nouns end in the thematic vowel (e.g. ton/tin/to a´ nthropo/kir´ıa/pedh´ı ‘the:MASC/FEM/NEUT:ACC:SG man:MASC/lady:FEM/child:NEUT:SG’). Due to its unmarked character the accusative singular represents the ‘base form’ of nouns in early child Greek (see sect. 4.3). The accusative expresses the direct object of transitive verbs (e.g. vl´epo to mathit´ı ‘I see the pupil:MASC:ACC:SG’) and is used in complements to prepositions (e.g. ap´o tin p´orta ‘from the:FEM:ACC:SG door:FEM:ACC:SG’) and adjectives (e.g. jem´ato lul´udhja ‘full (of) flowers:NEUT:ACC:PL’) as well as in nominal adverbials denoting time, place etc. (e.g. ti n´ıxta ‘(at) the:FEM:ACC: SG night:FEM:ACC:SG’) (see Mackridge 1985: 58; Tzartzanos [1946] 1991: 84). Depending on gender and inflectional classes of nouns, the GEN:SG is marked by final –s, vowel change or vowel addition, or may remain unmarked (see below). It is used more rarely than the accusative and the nominative (Chatzisavas 1992: 76), its main functions being possessive and partitive. The genitive can be considered to be regressive in colloquial Modern Greek (Kavoukopoulos 1989: 265–284; Anastasiadi-Simeonidi 2003: 29), since it is often substituted by prepositional phrases, mostly with inanimate nouns (e.g. to p´odhi ap´o to trap´ezi ‘the leg of the:NEUT:ACC:SG table:NEUT:ACC:SG’ instead of to p´odhi tu trapezj´u ‘the leg the:NEUT:GEN:SG table:NEUT:GEN:SG’ (= the leg of the table)). Moreover, neuter diminutives ending in –aki (e.g. pedha´ ki ‘child-DIM:NEUT:NOM/ACC:SG’) as well as some other nouns do not form the genitive case (Triantafyllidis 1963; Thomadaki 2008). It is interesting to note that, according to the corpus analyzed by Kavoukopoulos (1989: 279), in written texts the genitive is almost as frequent as the nominative and even more frequent than the accusative, while in spoken Greek the nominative

The emergence of nominal inflection in Greek

223

Table 1. The main declensional system of the Greek noun

SINGULAR NOM GEN ACC VOC PLURAL NOM GEN ACC VOC gloss

Masculine Subclass I

Subclass II

Feminine

Neuter

o pat´eras tu pat´era ton pat´era pat´era

o a´ nthropos tu anthr´opu ton a´ nthropo a´ nthrope

i mit´era tis mit´eras tin mit´era mit´era

to mor´o tu mor´u to mor´o mor´o

i pat´eres ton pat´eron tus pat´eres pat´eres ‘father’

i a´ nthropi ton anthr´opon tus anthr´opus a´ nthropi ‘human being’

i mit´eres ton mit´eron tis mit´eres mit´eres ‘mother’

ta mor´a ton mor´on ta mor´a mor´a ‘baby’

occurs four times as frequently and the accusative twice as frequently as the genitive. The vocative plays a minor syntactic and functional role and is formally distinguished from the other cases only in some animate masculine nouns ending in –os (e.g. Al´eksandhros/Al´eksandhre ‘Alexander:MASC:NOM/VOC:SG’). Masculine nouns ending in –os and a few such feminine nouns (e.g. o a´ nthropos ‘the:MASC:NOM:SG man:MASC:NOM:SG’, i odh´os ‘the:FEM: NOM:SG street:FEM:NOM:SG’) distinguish three forms each in the singular and plural, namely nominative, accusative, and genitive (see table 1). Most Greek nouns, however, namely feminine, neuter, and masculine not ending in –os, contrast only two forms each in the singular and plural, either the genitive and the nominative/accusative (feminine and neuter) or the nominative and an oblique case (masculine nouns not ending in –os). Nouns belonging to these two declensional types are referred to as ‘three-case nouns’(‘triptota’) and ‘two-case nouns’ (‘diptota’) respectively. The only nouns which formally distinguish all four Greek cases in the singular belong to an animate subset of masculines ending in –os (see table 1). Since there is considerable syncretism of cases in Greek nouns, the definite article (and to a certain extent also the indefinite article and adjectives) has an important role to play in case distinction. It differentiates the accusative from the nominative with feminine and masculine nouns in both numbers (i/tin ‘the:FEM:NOM/ACC:SG’, o/ton ‘the:MASC:NOM/ACC:SG’, i/tis ‘the:FEM:NOM/ACC:PL’, i/tus ‘the:MASC:NOM/ACC: PL’).

224

Ursula Stephany and Anastasia Christofidou

3. Data The Greek child and input data analyzed in the present paper come from the longitudinal observation of the boy Christos in the period from the second half of his second year (1;7.11) through the last third of his third year (2;8.25) (corpus Christofidou) and the four children Spiros, Mairi, Janna, and Maria observed at one, two, or three different points in time during approximately the same age range (corpus Stephany) (see table 2).3 The audiotapes of the boy Christos, who was the only child of an upper middle-class family during the period of observation, growing up monolingually inAthens, consist of (almost) weekly samples of (approximately) 20 min. each of (semi-)spontaneous speech and were recorded during play time or while looking at picture books. The boy may be characterized as an analytic (‘referential’) child (see Peters 1977), since formulaic and frozen forms are rare in his speech. Christos’data have been divided into three periods: Period I (1;7–1;10) can be described as premorphological for nouns, since these are not yet contrastively marked for case or number (Christofidou 2004; see also Kilani-Schoch et al. 1997: 18–20). In Period II (1;11–2;4) the contrastive use of noun forms emerges (see also Kilani-Schoch et al. 1997: 22–26). In Period III (2;5–2;8) more complex morphological patterns like the dissociation of declension classes within the same gender develop (see also Kilani-Schoch et al. 1997). Month-long samples of Christos’ input at 1;7, 2;0, and 2;5 have also been analyzed for the purpose of the present study. The corpus collected by Stephany and published in the database of the CHILDES Project (MacWhinney 2000), represents the computerized transcriptions of the audiotaped speech of four children gathered in natural speech situations such as playing or eating, in the children’s homes or a day nursery (Janna). With the exception of Janna, who grew up in an upper middle-class family, the other children come from lower middle-class families. All of them were only children during the period of observation, typically interacting with their mothers (Mairi, Spiros), grandmother and mother (Maria), nurses and parents (Janna). Mairi, the linguistically most advanced child of the four children studied by Stephany may be characterized as a ‘referential’child, much like Christos. When she was first observed at the age of 1;9, the development of inflection was well under way with verb inflection being much more advanced than nominal inflection (on the development of verb inflection see Stephany 1985, 1997a; Christofidou and Stephany 2003). In the present paper, we will study the development of nominal inflection in the tape-recorded data of Christos and the four children observed by Stephany and compare their acquisition of the grammatical categories of the Greek noun,

The emergence of nominal inflection in Greek

225

Table 2. The child data Subject

Age

Words/ Utterances Christos 1;7–1;10 1,425/1,153 Spiros 1;8–1;9 738/446 Mairi 1;9–1;10 5,761/1,796 Janna 1;10–1;11 836/547 Maria – –

Age

Words/ Utterances 1;11–2;4 12,116/5,984 – – 2;3 2,912/1,175 2;5 956/374 2;3 1,221/476

Age

Words/ Utterances 2;5–2;8 13,153/4,953 – – 2;9 3,179/1,148 2;11 1,304/422 2;9 1,475/470

namely case, number, and gender, with special emphasis on Christos and Mairi and their input. Christos’collection starts slightly after the onset of speech and he was continuously observed from the age of 1;7 to 2;8. Mairi was not observed from the onset of speech and her audiotaping took place at three temporally separated periods (see table 2).

4. Quantitative approaches to the developmental course of Greek nominal inflection In this section, the early development of nominal inflection will be traced through several quantitative measures applied to the grammatical forms of nouns. As stated above, we will more specifically compute the ratio of grammatical types per noun (sect. 4.1), types of marked inflectional forms and size of nominal paradigms (sect. 4.2), and finally PBF (Percentage of Base Forms), a quantitative study of tokens of unmarked forms (sect. 4.3). Child speech will be compared to child-directed speech (henceforth CDS) and – when made possible by relevant data – to adult-directed speech (ADS).

4.1.

Grammatical types per noun

Inflection consists in the usage of different grammatical forms of a given lexeme. Therefore, the growing ratio of grammatical forms per lexeme (or mean size of paradigm; see Xanthos and Laaha 2007: 13) at different points in time may be considered as a measure of inflectional development. In the premorphological stage of language acquisition, lexical items ideally occur in a single form each so that the ratio of grammatical forms and lemmas will theoretically equal 1. The three children first observed by Stephany (1997a) at 1;9 (Spiros, Mairi) and 1;11 (Janna) have already started to use more than one form per noun on average

226

Ursula Stephany and Anastasia Christofidou

Grammatical Types per Lemma

Types per lemma in Mairi's speech and CDS 1.25 1.20 MOT MAI

1.15

MAI

1.10 1.05 1;9

2;3

2;9

Age

Figure 1a. Grammatical types per lemma in Mairi’s speech and CDS

since their ratios of grammatical form types per lemma exceed 1.00, albeit only slightly. The ratio of the linguistically least advanced child Janna is below 1.05, which means that she uses less than every 20th lemma of her 30 nouns in more than one grammatical form. Although the most advanced child Mairi does so with nearly every 7th of her 94 noun lemmas already at 1;9, her development only rises to less than the use of every 5th noun in more than one form at 2;9 (fig. 1a). The ratio of grammatical forms and noun lemmas of Mairi’s mother’s CDS at 1;9 does not considerably exceed the child’s ratio at 2;9. The mother’s ratio only surpasses the child’s at 1;9, while both ratios more or less coincide during the child’s third year of life. This means that Mairi’s frequency of contrasting grammatical forms of nouns reaches the input norm already in the first half of her third year. There is individual variation of the type/lemma ratio among the three children observed longitudinally by Stephany. Both Janna and Maria remain considerably below the ratio reached by Mairi in the last period of observation. Turning to Christos, of whom observation began somewhat earlier, in period I (1;7–1;10) the type/lemma ratio nearly corresponds to the ideal value of premorphology (fig. 1b). In the next six months, the boy uses almost every 5th noun in more than one form on average. The reason why the child’s ratios in periods II (1;11–2;4) and III (2;5–2;8) seem to exceed those of the input is probably due to the restricted input samples analyzed, which cover only one month of each of the three periods. Nevertheless, these results show that the use of contrasting grammatical forms of nouns in Christos’ speech at least reaches the input norm in the first half of his third year, as is the case with Mairi. For a language like Greek, in which the categories of case and number are differently expressed in noun classes essentially based on gender, the average

The emergence of nominal inflection in Greek

227

Grammatical Types per Lemma

Types per lemma in Christos' speech and CDS 1.25

MOT CHR

1.20

CHR

1.15 1.10 1.05 1.00

I

II

III

Figure 1b. Grammatical types per lemma in Christos’ speech and CDS

number of grammatical types per noun occurring at different points of development is too global a measure of inflectional development. In order to gain a deeper insight into the process of inflectional development, paradigm formation of nouns must be studied in detail.

4.2.

Paradigm formation of Greek nouns

Development of inflection sets in as soon as lexemes occur in more than a single grammatical form. The notion of ‘miniparadigm’ (Kilani-Schoch and Dressler 2002: 50–52; Bittner, Dressler, and Kilani-Schoch 2003) covers the early phases of paradigm construction, distinguishing between mere occurrence of more than one grammatical form of a lexeme and “morphological relatedness” between distinct grammatical forms (Kilani-Schoch and Dressler 2002: 50). A “true miniparadigm” is defined “as corresponding to a non-isolated set of minimally three phonologically unambiguous and distinct inflectional forms of the same lemma produced spontaneously in contrasting syntactic or situative contexts in the same month of recordings” (Bittner, Dressler, and Kilani-Schoch 2003: xvi). In contrast to the development of Greek verbal inflection (see Christofidou and Stephany 2003: 107–113), the notion of miniparadigm is of limited value for the description of the early development of nominal inflection. The reason is that it is only applicable to case distinctions in the SG of MASC ‘triptota’ nouns (ending in –os), which may distinguish four case forms. However, in child as well as child-directed and adult-directed speech, these nouns occur less frequently than NEUT and FEM ones, with regard to both types and tokens (see below). All other Greek nouns are ‘diptota’ nouns, distinguishing merely two

228

Ursula Stephany and Anastasia Christofidou

Table 3a. Grammatical forms and nominal paradigms in Mairi’s data Age 1;9

2;3

2;9

Noun lemmas NEUT 55 FEM 27 MASC 12 Total 94 NEUT 65 FEM 49 MASC 18 Total 132 NEUT 69 FEM 45 MASC 14 Total 128

1 form 48 (87%) 26 (96%) 8 (67%) 82 (87%) 59 (91%) 44 (90%) 14 (78%) 117 (89%) 59 (85.5%) 38 (84%) 10 (71%) 107 (84%)

2 forms 7 (13%) 1 (4%) 4 (33%) 12 (13%) 6 (9%) 5 (10%) 4 (22%) 15 (11%) 10 (14.5%) 7 (16%) 4 (29%) 21 (16%)

case forms in the SG (see sect. 2 above). For the majority of Greek nouns, “true miniparadigms” would thus have to include a singular case contrast besides at least one PL form. It is therefore to be expected that early Greek nominal paradigms will mainly be limited to two forms, either two SG case forms or one SG and one PL form. For this reason, the description of paradigm formation in Greek language acquisition cannot be limited to miniparadigms but must include contrasts of two grammatical forms of a given lexeme, be these case or number contrasts. Since the development of specific case or number distinctions of Greek nouns is bound to gender, which mainly determines their inflectional class (see sect. 2), paradigm formation of NEUT, FEM, and MASC nouns will be described separately. As hypothesized, there are no true miniparadigms to be found in Mairi’s speech (table 3a), but, interestingly, they only marginally occur also in her input. During the entire period in which she was observed 84% to 89% of nouns are limited to one grammatical form. The percentage of lexemes occurring in two grammatical forms is somewhat higher with MASC nouns than those belonging to the other two genders, since MASC nouns are the only ones which mark the NOM:SG and distinguish it from the unmarked ACC:SG (see table 4 below). Roughly the same picture emerges from the child’s input. As far as Christos is concerned, a single true three-member miniparadigm occurs in periods II (1;11–2;4) and III (2;5–2;8) (table 3b). This miniparadigm consists of the three case forms of the boy’s name used almost exclusively instead of the respective pronouns for expressing self-reference and possession (o Chr´ıstos/tu Chr´ıstu/ton Chr´ısto ‘the:NOM/GEN/ACC:SG Christos:NOM/GEN/ ACC:SG’). This miniparadigm also occurs in his mother’s babytalk in reference

The emergence of nominal inflection in Greek

229

Table 3b. Grammatical forms and nominal paradigms in Christos’ data Age 1;7–1;10

1;11–2;4

2;5–2;8

Noun lemmas NEUT 32 FEM 14 MASC 7 Total 53 NEUT 113 FEM 88 MASC 55 Total 256 NEUT 122 FEM 105 MASC 47 Total 271

1 form 31 (97%) 14 (100%) 7 (100%) 52 (98%) 94 (83%) 79 (90%) 33 (60%) 202 (79%) 96 (79%) 92 (88%) 21 (45%) 206 (76%)

2 forms 1 (3%) – – 1 (2%) 19 (17%) 9 (10%) 21 (38%) 53 (20.6%) 26 (21%) 13 (12%) 24 (51%) 63 (23%)

3 forms – – – – – – 1 (2%) 1 (0.4%) – – 2 (4%) 2 (1%)

to her son. In contrast to what is found in both Mairi’s speech and her input, a relatively high percentage of MASC nouns are used in two grammatical forms in Christos’ speech from age 1;11 on and to a certain extent also in his input. It may be hypothesized that the frequent use of the three different grammatical forms of the MASC proper noun Chr´ıstos in the boy’s conversations with his mother and his ensuing familiarity with this paradigm will facilitate the use of other MASC nouns in more than one grammatical form. As assumed above, types of marked case-number forms are unevenly distributed across gender classes in both Mairi’s and Christos’ speech (figs. 2a and 2b; for raw numbers of noun lemmas see tables 3a and 3b). While Mairi uses 50% of MASC nouns in the NOM:SG already at 1;9, the proportion of MASC nouns occurring in this marked form rises to 64% by 2;9. Only one MASC lemma each is found in the marked forms GEN:SG and PL:ACC at 2;3 and 2;9. In contrast to MASC nouns, the much larger number of NEUT lemmas only shows a number contrast, with little change in the percentage of lemmas used in this marked form in the course of development. FEM nouns are the only ones to provide evidence of both a number and case contrast from 1;9 on. The number of FEM nouns occurring in the PL increases more rapidly than lemmas used in the GEN:SG. Christos’development of grammatical types is comparable to Mairi’s: Marked forms of MASC nouns are mainly limited to the NOM:SG and the marked GEN:SG (ending in –u) only occurs with the lemma Chr´ıstos in period II. Marked NOM/ACC:PL forms of NEUT nouns are found in the three periods,4 while FEM nouns show both a number and case contrast from 1;11 onward (fig. 2b). Only a single PL form of a MASC noun occurs in period III (jeran´ı

230

Ursula Stephany and Anastasia Christofidou Marked inflectional forms of nouns: Mairi (types)

70% 60% 50%

1;9

40%

2;3

30% 20%

2;9

10% 0% MASC:NOM

NEUT:PL

FEM:GEN

FEM:PL

Figure 2a. Development of marked inflectional forms of nouns in Mairi’s data

Marked inflectional forms of nouns: Christos (types) 70% 60% 50% 1;7–1;10

40% 30%

1;11–2;4

20%

2;5–2;8

10% 0% MASC:NOM MASC:GEN MASC:PL NEUT:PL FEM:GEN FEM:PL

Figure 2b. Development of marked inflectional forms of nouns in Christos’ data

‘crane:MASC:NOM:PL’). In spite of the fact that Christos’ percentages of MASC nouns marked for the NOM:SG are much lower than Mairi’s, he uses about twice as many nouns of this gender in this form (see tables 3a and 3b). While NOM:SG forms of MASC nouns increase type-wise in Christos’ speech, they decrease token-wise from 55% in period II to 37.5% in period III. The reason is that the boy gradually abandons the strategy of referring to himself by his name. Regarding the other three children, nominal inflection develops in roughly the same way as in the case of Mairi and Christos. While MASC nouns are limited to the NOM-ACC contrast, NEUT ones only exhibit a number contrast. Spiros, at 1;9, only distinguishes number of FEM nouns. Janna develops the

The emergence of nominal inflection in Greek

231

Table 4. Emergence of case-number contrasts of nouns in Stephany’s and Christofidou’s corpora Masculine

Neuter Feminine

NOM:SG –s ACC/OBL:SG -V (V = thematic vowel) GEN:SG -u NOM/ACC:SG NOM/ACC:PL NOM/ACC:SG NOM/ACC:PL GEN:SG

Mairi 1;9, Spiros 1;9, Janna 1;11, Christos 1;11, Maria 2;9 Spiros 1;9, Mairi 2;3, Christos 2;4 Mairi 1;9, Spiros 1;9, Christos 1;9, Janna 1;11, Maria 2;3 Mairi 1;9, Spiros 1;9, Janna 1;11, Christos 1;11, Maria 2;3 Mairi 1;9, Christos 1;11, Maria 2;9, Janna 2;11

GEN:SG of FEM nouns before the number contrast (at 2;5 vs. 2;11). Regarding Maria, a late talker, the marked NOM:SG form of MASC nouns only emerges at 2;9. Her NEUT and FEM nouns are limited to a number contrast at 2;3 and there is a single instance of the GEN:SG found at 2;9. These findings clearly demonstrate that the development of inflectional types and paradigm formation proceeds within gender classes of nouns rather than across-the-board. Thus, the number contrast primarily develops in the NOM/ ACC form of NEUT nouns and to a smaller extent of FEM ones. Since the GEN is extremely rare or even impossible with NEUT nouns (see sect. 2), the NOM/ACC is the only case form occurring in child Greek and CDS. In contrast to NEUT nouns, MASC nouns at first only develop a case distinction, mainly the NOM-ACC:SG contrast. As mentioned above, MASC nouns are the only ones to mark the NOM:SG contrasting it with the ACC:SG, which is an unmarked form in all three genders. The marked GEN:SG form of ‘triptota’ subclass II MASC nouns ending in –os is much less used than the NOM:SG and may develop much later. With FEM nouns, both a number contrast develops in the NOM/ACC form and a case contrast between the NOM/ACC:SG and GEN:SG. With the exception of the GEN of NEUT nouns, the children studied in the present paper learned to exploit the possibilities offered by the case system of nouns in the SG by the end of the observational period (2;8, 2;9, 2;11) (see table 4; see also Stephany and Christofidou 2008). A comparison of the children’s use of nouns in their marked case-number forms with the mothers’ CDS will help to explain the above findings. With the exception of the marked GEN:SG of MASC subclass II nouns, the categories of marked inflectional forms of nouns in Mairi’s speech coincide with those most strongly represented in the input (see figs. 2a and 3a). The other types of marked inflectional forms of Greek nouns, such as the VOC:SG of MASC subclass II

232

Ursula Stephany and Anastasia Christofidou

Marked inflectional forms of nouns in Mairi's input (types) 70% 60% 50% 40% 30% 20% 10% 0%

1;9 2;3 2;9

MASC:NOM MASC:GEN

NEUT:PL

FEM:GEN

FEM:PL

Figure 3a. Marked inflectional forms of nouns in Mairi’s input

Marked inflectional forms of nouns in Christos' input (types) 70% 60% 50% 40% 30% 20% 10% 0%

1;7 2;0 2;5

MASC:NOM MASC:GEN

NEUT:PL

FEM:PL

FEM:GEN

Figure 3b. Marked inflectional forms of nouns in Christos’ input

nouns, NOM:PL and ACC:PL forms of MASC nouns, and the GEN of NEUT nouns in the SG and PL, occur with even less lemmas than the GEN:SG and NOM/ACC:PL of FEM nouns in Mairi’s input from 1;9 to 2;9 and are also very infrequent token-wise. Use of nouns in marked inflectional forms is also similar in Christos’ and his mother’s speech (see figs. 2b and 3b). Furthermore, the selection and ranking of marked forms of nouns in the three genders coincides in both mother-child dyads. The fact that NOM:PL or ACC:PL forms of MASC nouns, which rarely occur in Christos’ speech between 2;5 and 2;8, are not documented in his input is probably due to the small sample analyzed for the present study. As far as marked case-number forms of nouns are concerned, our results show that caretakers and children mainly rely on the communicatively most important ones and that the choice of specific inflectional forms differs according to gender class.

The emergence of nominal inflection in Greek

233

Although Chatzisavas (1992) counted tokens of inflectional forms of nouns rather than form types, a comparison of the ranking order of marked casenumber forms of each gender occurring in adult-directed spoken Greek (ADS) with the percentages of lemmas used in marked inflected forms by Christos’ and Mairi’s mothers in their CDS is illuminating (see figs. 3a and 3b). Marked SG forms of MASC nouns are strongly biased toward the NOM with regard to both type in CDS and token in ADS. In contrast to this, the marked forms of NEUT nouns are predominantly NOM/ACC:PL, while both the GEN:SG and NOM/ACC:PL of FEM nouns are used to an almost equal degree, but less frequently than the marked forms of the MASC and NEUT gender just mentioned. In their child-directed register the mothers follow the distribution of grammatical forms reported for adult-directed spoken Greek in an even more pronounced way in some respects (infrequent use of the GEN as compared to the NOM or NOM/ACC). In other respects, however, they deviate from this adult-directed pattern by using a more limited inventory of marked forms (SG forms of MASC nouns and NOM/ACC forms of NEUT nouns). After having studied the type-wise development of marked inflectional forms of nouns, we will now turn to the token-wise distribution of marked and unmarked case-number forms.

4.3.

Percentage of Base Forms (PBF)

Percentage of Base Forms (PBF) is a parameter devised by Voeikova and Gagarina (2002: 123) for studying the interdependence of the acquisition of Russian nominal inflection and syntax on the one hand and that of nominal inflection and the lexicon on the other. It simultaneously serves as a measure of the increasing role played by case and number distinctions of nouns in the course of language acquisition tracing the use of marked vs. unmarked inflected forms in children’s speech. In articleless languages such as Russian, noun forms are the only locus of expression of the grammatical categories of the noun. In Greek, the distinction of certain grammatical forms of the noun exhibiting syncretism is achieved with the help of determiners, especially the definite article. Therefore the development of the article has an important role to play in certain distinctions of case-number forms of nouns. In the present section we will focus on the use of grammatical forms of the noun itself, not only for the sake of the comparability of our study with other languages, especially articleless ones, but also because the distinction of grammatical forms by periphrastic means plays a secondary role as compared to synthetic techniques (see sect. 5 below; also see Stephany 1997a).

234

Ursula Stephany and Anastasia Christofidou

PBF has been defined for Russian as “the percentage of noun tokens in the nominative” (Voeikova and Gagarina 2002: 123). While PBF amounts to 30% in adult-directed Russian speech and reaches a minimum of about 42% in one child’s input, the corresponding value in two children’s speech studied until the age of 2;3 is roughly 50% (Voeikova and Gagarina 2002: 123). The most suitable candidate for Greek base forms of nouns is the inflectionally unmarked SG form ending in the thematic vowel. Depending on the inflectional class of the noun, this is the NOM/ACC:SG of FEM and NEUT nouns and the ACC:SG of MASC ones. These inflectionally unmarked forms of nouns are the first ones to emerge in Greek language acquisition (Stephany 1997a: 200; Christofidou and Stephany 1997: 129; see also sect. 5). In a language with a completely regular, agglutinating morphology such as Turkish, which lacks declensional subclasses of nouns, a single PBF value calculated for nouns quite generally is sufficient to compare the child’s development of noun inflection to that of the input (on Turkish see Ketrez and Aksu-Ko¸c, this volume). By contrast, in languages of the inflecting-fusional type such as Greek, a general PBF value established across genders and inflectional classes can only give a rough indication of the development of nominal inflection, while the more specific properties of the developmental process must be worked out by distinguishing between inflectional and gender classes (see below). Calculated across gender and inflectional classes, PBF amounts to roughly 54% in adult-directed spoken Greek (based on Chatzisavas 1992). In Mairi’s mother’s CDS, PBF values approach 80% and thus exceed the value characteristic of adult-directed speech (ADS) (see fig. 4a). There is individual variation of PBF values in the input. While this value remains high in Mairi’s mother’s CDS during the child’s development from 1;9 to 2;9, the corresponding value of Christos’ input drops from an initial 79% to 63% from 2;0 on, approaching the PBF value of ADS (see fig. 4b). In contrast, PBF values of two other mothers’ CDS of Stephany’s corpus at their children’s ages of 1;9 and 1;11 even exceed those of Mairi’s mother, amounting to 87% and 81% respectively. Comparing PBF in CDS to that found in ADS more generally it can be noted that the characteristics of ADS are emphasized in CDS so that the percentage of tokens of marked inflectional forms young Greek children have to cope with is minimized. In addition to the reduction of the inventory of inflected forms found in section (4.2.), this is another way in which Greek mothers simplify the input in the domain of inflectional forms of nouns. As is to be expected, PBF values gradually decrease over the course of the children’s development (figs. 4a and 4b). The most drastic drops occur regarding Janna and Christos after 1;11. Overall, the children’s PBF values are closer to CDS than to ADS.

The emergence of nominal inflection in Greek

235

PBF in Greek CS, CDS, and ADS 120%

PBF

100% 80%

1;10

60%

2;4

40%

2;10

20% 0% Spiros

Mairi

Janna

Maria

CDS Mairi

ADS

CS, CDS, ADS

Figure 4a. PBF in Greek child speech (Corpus Stephany), child-directed speech, and adult-directed speech

PBF in Greek CS, CDS and ADS 120%

PBF

100% 80%

I

60%

II

40%

III

20% 0%

Christos

CDS Christos

ADS

CS CHR, CDS CHR, ADS

Figure 4b. PBF in Christos’ CS and CDS and in ADS

In order to gain a deeper insight into the distribution of the respective PBF values in Greek ADS and CDS as well as into the children’s development of nominal inflection, we have calculated PBF separately for each gender class. A comparison of child speech (CS) with CDS and ADS shows the same ranking order in the three types of speech ranging from FEM nouns with the highest PBF values to MASC ones with the lowest (see tables 5a and 5b). The only exception are MASC nouns, which are exclusively used in their unmarked form by Christos before 1;11 and by Maria at 2;3 (PBF 100%). Otherwise, MASC nouns have considerably lower PBF values than the other two gender classes.

236

Ursula Stephany and Anastasia Christofidou

Table 5a. PBF values of nouns in CS, CDS, and ADS according to gender (Corpus Stephany) Gender class CS

Feminine Neuter Masculine

1;9, 1;11 Mi Ja 98 100 76 97 69 91

CDS (MOT Mi)

Sp 86 85 30

2;3, 2;5 2;9, 2;11 Mi Ja Ma Mi Ja Ma 94 95 96 86 94 92 75 80 70 71 64 65 62 43 100 47 33 65

ADS (based on Chatzisavas 1992)

1;9 2;3 2;9 95 93 90 65 77 71 76 57 36 46 21 38

Table 5b. Christos’ PBF values of nouns in CS and CDS according to gender Gender class Feminine Neuter Masculine

CS (Christos) 1;7–1;10 1;11–2;4 100 95 82 83 100 42

2;5–2;8 89 71 55

CDS (MOT Christos) 1;7 2;0 2;5 99 92 77 80 88 68 56 24 43

ADS 65 57 38

As with the general PBF values mentioned above, gender-sensitive PBF values in CDS are much higher than in ADS, with the exception of MASC nouns in Mairi’s input and in Christos’ from 2;0 on. This shows that inflectionally marked forms and their contrast with unmarked ones plays a greater role with MASC nouns in the mothers’ discourse than with the other two genders. This seems to have an influence on the children’s development, since they make use of the NOM-ACC:SG contrast with a high percentage of MASC nouns (see figs. 2a and 2b above). Studying the role of marked vs. unmarked forms of nouns in Mairi’s and Christos’ speech in more detail, it becomes clear that there is a considerable drop of inflectionally unmarked forms in both children by the end of observation only in the case of MASC nouns (see tables 5a and 5b). Although the distribution of marked and unmarked forms of nouns per gender in the children’s speech roughly corresponds to the input, both mothers offered many more instances of inflectionally marked MASC nouns at each developmental stage than their children produced. This contributes to the entrenchment of such forms in the children’s memory. Another distributional characteristic of Greek ADS, CDS, and CS is the unequal frequency of nouns of the three genders occurring in discourse (for

The emergence of nominal inflection in Greek

237

ADS see Chatzisavas 1992 and Kavoukopoulos 1996). MASC nouns are much less frequent than those of the other two genders both type- and token-wise. Again, this tendency is more pronounced in CDS than in ADS and accordingly also in CS (tokens of MASC nouns: ADS: 23%; CDS: 7%–17%; CS: 7%–19%, with the exception of 25% found in the case of one child at 2;9). Although FEM nouns occur more frequently than NEUT ones in ADS, the latter tend to exceed the former in CDS as well as CS. This may be attributed to the high number of NEUT diminutives ending in –aki typical of child-centered speech situations (Stephany 1997b; Thomadaki and Stephany 2007). While the children’s early inflectional development of NEUT nouns can be explained by the fact that nouns of this gender are very frequent in the input both type- and token-wise, the same cannot be said of FEM nouns and even less of MASC ones, which occur much less frequently. The reasons for the emergence of particular types of inflectionally marked forms must therefore be sought in their functional load in discourse on the one hand and in the type of formal marking on the other. The only relevant (and sometimes the only possible) inflectional distinction with most NEUT nouns is the number contrast, uniformly expressed by final –a and therefore quite salient; many NEUT nouns refer to objects or (toy) animals and are attributed the NEUT gender by derivation. In contrast, many FEM and MASC nouns are proper nouns, kinship terms, or animal names referring to animate beings. With these classes of nouns, the SG is much more commonly used than the PL. Although a distinction between NOM and ACC seems to be functionally important in the case of such nouns irrespective of their gender class, the reason for the earlier development of this case contrast with MASC nouns is that it is marked on the noun itself while FEM nouns can only distinguish between the two cases by the form of the DEF.ART (see sect. 5.1.2 below). Thus, the token-wise most infrequent class of MASC nouns is the first to develop a functionally important case contrast, while the priority of the functionally important number contrast of NEUT nouns may be partly due to their high frequency. As shown in sect. 4.2, the NOM-ACC contrast of MASC nouns is, however, even more strongly exploited type-wise than the number contrast of NEUT nouns in the children’s input. The results obtained from the study of PBF lend further support to our finding that the different grammatical categories of Greek nominal inflection develop locally with certain subclasses of nouns rather than across-the-board.

238

Ursula Stephany and Anastasia Christofidou

5. Development of case and number distinctions in Greek nouns In the first part of this section, we will trace the development of forms and functions of case in the SG in MASC and FEM nouns in the observational periods of the children. In the second part, the development of case-number forms in the PL will be described. Since NEUT nouns do not develop case distinctions in our data, they will only be analyzed in section (5.2). The DEF.ART plays an important role in the distinction of syncretistic inflectional forms such as the NOM/ACC:SG of FEM nouns, with the result that its development may offer important insights into the role of synthetic vs. periphrastic marking of case-number distinctions in Greek language acquisition.

5.1.

Development of singular case distinctions in Greek nouns

While the most important syntactic function of the NOM-ACC distinction is the expression of subject vs. object, the GEN primarily expresses the possessive. In Stephany’s data it also fulfills the benefactive function by 2;9/2;11 (Stephany 1997a: 205). In addition to the functions of subject, object, and possessive, another function emerging early is the locative, which is expressed by a diversity of linguistic means, namely locative adverbs and prepositional phrases (for details see Stephany 1997a: 205, 282–287). Comitative and instrumental meanings of prepositional phrases introduced by me ‘with’ have emerged by 2;3/2;5 in Stephany’s data. Temporal and manner adverbials play a less important role than locative ones and are expressed by adverbs (e.g. t´ora ‘now’, met´a ‘later’, e´ tsi ‘this way’). For a detailed analysis of the argument structure of sentences found in Stephany’s corpus between 1;9/1;11 and 2;9/2;11 see Stephany (1997a: 272–288).

5.1.1. Development of singular case distinctions of masculine nouns 5.1.1.1.

Emergence of the nominative-accusative contrast

There is evidence in the four children’s data from Stephany’s corpus, as well as in Christos’ speech, that the inflectional development of MASC nouns starts with their unmarked SG form ending in the thematic vowel (e.g. pap´u ‘grandfather:OBL:SG’). This ‘all-purpose’ unmarked form splits into a new marked NOM:SG form ending in –s and the old unmarked form, which then specializes to become an OBL case form. The –V vs. –Vs pattern is the most important inflectional development of MASC nouns (see sect. 2). The OBL case form

The emergence of nominal inflection in Greek

239

of MASC subclass II nouns (ending in –os) is limited to ACC contexts and the VOC with some nouns (but see below on Christos’development), while the OBL form of other MASC nouns is used in VOC, GEN, and ACC contexts. As shown in sections (4.2) and (4.3), the communicative importance of the NOM:SG of MASC nouns is reflected by their distribution in the children’s discourse. Thus, in Mairi’s speech at 1;9, 65% of MASC noun tokens function as NOM and only 35% as ACC (n = 102). In the case of both Spiros and Mairi, synthetic as well as periphrastic marking of the NOM-ACC:SG contrast of MASC nouns sets in before the end of the second year (by 1;9), but both techniques are differently weighted in the two children’s speech: While Spiros relies more heavily on synthetic marking, in Mairi’s speech periphrastic marking also plays an important role (see also Stephany 1997a: 226–229). In Christos’ speech there is clear evidence for these techniques in the second observational period (1;11–2;4). Regarding Maria and Janna, the NOM-ACC contrast also emerges in the first part of the third year. In order to gain a deeper insight into the development of the NOM-ACC contrast of MASC nouns Christos’ and Mairi’s data will be analyzed in more detail.5 Within a month after the first occurrence of the NOM:SG marker –s in pap´us ‘grandfather:MASC:NOM:SG’ at 1;11.0, Christos uses the marked NOM:SG forms of 88% of MASC noun tokens in contexts where these are required, while “only 2% of forms marked by –s are misused with an oblique function (n = 97)” (Christofidou and Stephany 1997: 134; see also Christofidou 2004). Since one third of unmarked forms of MASC nouns used in NOM contexts are accompanied by the MASC:NOM:SG form o of the DEF.ART, only 9.7% of MASC:SG NPs functioning as NOM are unmarked for case in Christos’period II data (1;11–2;4).6 The DEF.ART still, however, plays a minor disambiguating role in Christos’ speech as compared to the synthetic marking of the NOM:SG on the noun. This is also shown by the observation that he starts to overgeneralize –s marking to foreign names like Pl´uto, which are uninflected in the input, less than two weeks after the first occurrence of the first form marked for NOM:SG (see Christofidou and Stephany 1997: 130–134). Thus, from early on, the NOMACC contrast becomes productive with MASC nouns (referring to male beings). The -V/-s contrast of MASC nouns is never overgeneralized to feminine nouns. The first contrast of a NOM:SG and ACC:SG form of one and the same MASC noun occurs at 1;11.13, two weeks after the appearance of the marked NOM (pap´us/pap´u ‘grandfather:NOM/ACC:SG’) (Christofidou and Stephany 1997: 134). In the beginning, NOM:SG forms of MASC nouns are used to identify animate beings, but from 2;2.14 on, they also occur with inanimate nouns (e.g. jeran´os ‘crane’) (Christofidou and Stephany 1997: 132). There is further evidence that Christos marks the MASC:NOM:SG systematically. One is

240

Ursula Stephany and Anastasia Christofidou

single instead of double case marking of appositional constructions such as N´ıko pap´us for pap´us N´ıkos ‘grandfather:MASC:NOM:SG Nikos:MASC:NOM:SG’ (1;11.13).7 The other one is the correct application of the contrastive marked/ unmarked pattern to most MASC forms of pronouns and adjectives from 1;11 on. Until the age of 2;4, 76% of NOM pronoun tokens (n = 21) and 80% of the respective adjective tokens (n = 15) are correctly inflected. Most of these MASC:NOM:SG tokens consist of the pronoun aft´os ‘this’ and the adjectives kal´os ‘good’ or kak´os ‘bad’; all of these are contrasted with other case and gender forms. Summarizing our findings so far, it can be said that the basic declensional pattern -Vs vs. -V of Greek nouns (see sect. 2) is well established in the MASC gender in Christos’ speech by 2;4. In the next four months, this patterns gains in productivity, since, by 2;8, it has risen from 28% to 37.5% of MASC types (see fig. 2b in sect. 4.2 above). In Mairi’s data there is evidence for the NOM:SG-ACC:SG contrast of MASC nouns already by 1;9; this contrast is still far from being generally established, however. Mairi uses the unmarked form of MASC nouns ending in the thematic vowel in many nominative as well as oblique contexts so that the original system lacking case distinctions still determines more than half of her tokens of these nouns occurring in NOM contexts (45% marked MASC:NOM tokens, n = 66). The fact that, with a single exception, marked MASC:NOM:SG forms are limited to NOM contexts (30 tokens) while, in ACC contexts, only unmarked forms occur (31 tokens), is further evidence of the development of the NOM-ACC contrast. In addition, the MASC:NOM:SG form o of the DEF.ART is found in the case of 75% of unmarked MASC noun tokens occurring in NOM contexts (n = 36), while o is only once wrongly used in an ACC context. Taking periphrastic case marking into consideration, tokens of MASC NPs marked for NOM:SG in the appropriate contexts amount to 87% (n = 68) in Mairi’s speech already at 1;9. While variable synthetic marking or use of the unmarked form in NOM:SG contexts continues to occur in the girl’s speech at 2;3 and 2;9, the NOM-ACC:SG contrast o/to(n) of the DEF.ART becomes more firmly established over the course of her third year so that, by 2;9, the NOM:SG is no longer overused in ACC contexts neither with regard to MASC nouns nor the DEF.ART. Comparing Christos’ development of the NOM-ACC contrast of MASC nouns to Mairi’s, it can be stated that this boy, like Spiros, relies more heavily on synthetic marking than the girl. It must not be forgotten that, in contrast to Mairi, both Christos and Spiros may be expected to hear many examples of the NOM-ACC:SG contrast of their respective first names, which belong to the class of MASC nouns. Although in a language like Greek, which does not rely on word order for distinguishing between subject and direct object, acquisition of the NOM-ACC

The emergence of nominal inflection in Greek

241

contrast seems to be of major importance, Stephany (1997a) found that there is no relationship between the emergence of case marking and transitive constructions and that “what is relevant for the distinction of subject and object in child Greek is the semantic category of animacy rather than the grammatical category of case” (p. 220). 5.1.1.2. The vocative and genitive Besides the NOM and ACC the other two grammatical cases which may be formally distinguished regarding MASC nouns are the VOC and GEN, both of which are much less frequently used in spoken Greek than the NOM and ACC (see sect. 2). With the exception of certain MASC subclass II nouns ending in –os, the unmarked form of the noun ending in the thematic vowel is used as an OBL form for the VOC, ACC, and GEN (e.g. j´oka ‘little.son:DIM:MASC:VOC: SG’ from j´okas NOM:SG). Only certain MASC nouns ending in –os distinguish a special VOC form ending in –e (e.g. Al´eksandhre ‘Alexander!’ (NOM Al´eksandhros), but Chr´ısto ‘Christos!’, Sp´ıro ‘Spiros!’). There is a single token of the marked VOC of subclass II MASC nouns in Stephany’s corpus (Mairi, 1;9, p´ıthike ‘monkey:MASC:VOC:SG’) and Christos only uses the marked VOC with a single lexeme (l´ıke ‘wolf:MASC:VOC:SG’) in the third period (2;5–2;8). Most of the altogether infrequent instances of the VOC are correctly expressed by the unmarked form of subclass I MASC nouns (e.g. Christos, 1;11–2;4, pap´u ‘Grandpa:MASC:VOC:SG’). Subclass II ‘triptota’ MASC nouns also distinguish the GEN:SG from the other cases by the ending –u (e.g. l´ıku ‘wolf:MASC:GEN:SG’), while ‘diptota’ MASC nouns (subclass I) use a common oblique case form for GEN, ACC, and VOC (see sect. 2) so that with this latter class the difference between the ACC and GEN:SG is only expressed by the form of the article (e.g. ton/tu ‘the:MASC:ACC/GEN:SG’, e´ nan/en´os ‘a:MASC:ACC/GEN:SG’). In Christos’data, the GEN:SG of ‘triptota’MASC nouns is the second marked noun form to appear at 2;3.18 with the sporadically used GEN of his own name expressing possession (Chr´ıstu ‘Christos:MASC:GEN:SG’). However, the MASC:GEN:SG form tu of the DEF.ART emerges earlier (in period II, 1;11– 2;4) and is used with the unmarked form of both subclasses of MASC nouns (21/3 tokens of Standard Greek ‘diptota’/‘triptota’) (see examples 1 answering the question ‘whose is this?’). Christos thus first constructs a nonstandard unitary ‘diptota’ class of MASC nouns merely distinguishing the NOM:SG form ending in –s from an unmarked OBL:SG one ending in the thematic vowel, thereby following the more frequent inflectional pattern of subclass I MASC nouns (see Christofidou and Stephany 1997: 136; Christofidou 2004: 6–9).

242

(1)

Ursula Stephany and Anastasia Christofidou

Christos, 2;3 a. tu Mim´ıti (= Dhim´ıtri) the:MASC:GEN:SG Dimitris:MASC:OBL:SG ‘Dimitris’.’ b. tu F´ıto instead of of.the Christos:MASC:OBL:SG tu Chr´ıstu of.the Christos:MASC:GEN:SG ‘Christos’.’

Within the month following its emergence (at 2;3.18), the -os/-u pattern becomes more frequent but remains limited to the lexical item Chr´ıstos. In the third period ´ (2;5–2;8) it is extended to two other MASC proper nouns ending in –os (Angelu and J´orghu ‘GEN’). All 31 GEN tokens of the three lexemes are properly used and 30 of them are accompanied by the correct MASC:GEN:SG form tu of the DEF.ART. At the end of period II (2;4), Christos thus starts to distinguish between the two declensional patterns of MASC nouns marking animate nouns ending in –os (‘triptota’, subclass II) by final –u in the GEN:SG while using the unmarked general OBL form of other MASC nouns (‘diptota’, subclass I) for expressing the possessive or benefactive functions (for a detailed analysis see Christofidou 2004: 6–11). In contrast to Christos’ development, there is no evidence in Stephany’s data for an initial overextension of the unmarked form of nouns ending in –os to the GEN. In this corpus the GEN:SG occurs very infrequently with MASC nouns so that there are just a few marked GEN forms to be found (e.g. Spiros 1;9, P´ıu for tu Sp´ıru ‘the:MASC:GEN:SG Spiros:MASC:GEN:SG’; Mairi 2;3, tu bab´a ‘the:MASC:GEN:SG Daddy:MASC:OBL:SG’). MASC nouns are not found in GEN contexts in Janna’s or Maria’s corpora. The experimental studies referred to by Stephany (1997a: 218) demonstrate that consistent marking of the GEN:SG by –u on ‘triptota’ MASC nouns (subclass II, ending in –os) is only achieved by 4;10.

5.1.2. Development of singular case distinctions of feminine nouns The only case distinction marked on FEM:SG nouns is that between the GEN:SG ending in –s and the unmarked NOM/ACC form ending in the thematic vowel (see sect. 2). The NOM:SG and ACC:SG are distinguished by the definite article (i/ti(n) ‘the:FEM:NOM/ACC:SG’) while the GEN:SG is marked both on the noun and by the article (tis/m´ıas ‘DEF/INDEF.ART:FEM:GEN:SG’). As mentioned in section (2), the GEN is much more infrequently used in adult-

The emergence of nominal inflection in Greek

243

directed spoken Greek than the NOM and ACC. The same is true of CDS and child speech so that FEM nouns by far most frequently occur in their unmarked NOM/ACC form (see figs. 2a, 2b, 3a, and 3b in sect. 4.2). As far as Christos’ development of case distinctions with FEM nouns is concerned, in period I (1;7–1;10) evidence for the NOM-ACC contrast is limited to 7 tokens of a single lemma (e.g. i jaj´a ‘the:FEM:NOM:SG grandmother’ vs. s(t)i jaj´a ‘to.the:FEM:ACC:SG grandmother’). In the next period (1;11–2;4) the number of such tokens increases considerably (139 NOM vs. 87 ACC tokens). Since the DEF.ART is correctly formed in 86% of NOM and 83% of ACC tokens occurring in NOM and ACC contexts respectively, the NOM-ACC contrast with FEM NPs has developed to a considerable degree by 2;4, in spite of the fact that the OBL form ti of the DEF.ART used by the child does not distinguish between ACC and GEN (see below). The first marked GEN:SG form of a FEM noun is found at 1;11 (mam´as ‘Mummy:FEM-GEN:SG’) in Christos’ speech. Since this form remains the only one between 1;11 and 2;1, it may be taken to be rote-learned (see also Christofidou and Stephany 1997: 135). Between 2;1 and 2;4, GEN:SG forms of 5 lexemes occur in 20 tokens, varying with the unmarked NOM/ACC form (5 tokens) (e.g. 2;2.14, El´eni-s ‘Helen:FEM-GEN:SG’; 2;2.26 tsi mam´a for tis mam´a-s ‘the:FEM:GEN:SG mummy:FEM-GEN:SG’). In period II (1;11–2;4) Christos begins to use a child OBL form ti of the FEM DEF.ART which does not distinguish the GEN from the ACC. 39% of FEM noun tokens occurring in GEN contexts (n = 41) are accompanied by ti (or rarely tsi) and one is constructed with the correct form tis. Three of the 17 noun tokens accompanied by the OBL form ti of the DEF.ART are unmarked for the GEN:SG so that GEN and ACC are not distinguished. Taking both synthetic and periphrastic marking of the GEN:SG of FEM nouns into consideration, Christos succeeds in marking 66% of these nouns carrying a possessive function in period II. The FEM:GEN:SG form tis of the DEF.ART emerges at 2;4 resulting in a potentially unambiguous three-case distinction of NOM, ACC, and GEN with FEM nouns. In the third period (2;5–2;8), the boy distinguishes the three cases of FEM NPs in 56% of tokens (n = 463) by employing both synthetic and periphrastic techniques. The development of case contrasts with FEM nouns in Mairi’s speech is still at a beginning stage at 1;9. Not only is the DEF.ART missing in the case of 54% of FEM nouns (n = 140), but, as with Christos, a common FEM:OBL:SG child form ti covering both the ACC and the GEN of the DEF.ART is found along with the standard ACC form tin (used with vowel-initial nouns in ACC contexts). Further, the most entrenched NOM:SG form i also occurs in one GEN and one ACC context. The NOM is functionally much more important regarding FEM nouns than the ACC, since 41% of FEM noun tokens accompanied by the

244

Ursula Stephany and Anastasia Christofidou

DEF.ART occur in the NOM and only 5% in the ACC. The occurrence of the GEN is marginal. The only example of a synthetically marked GEN:SG is the form M´eri-s combined with the NOM form of the DEF.ART instead of the GEN (example 2). The other instance of a FEM noun used in a possessive function consists of the child OBL form of the article and the unmarked form of the noun (ti P´egi for tis P´egi-s ‘the:FEM:GEN:SG Peggy:FEM-GEN:SG’ (= (It’s) Peggy’s)). (2)

Mairi, 1;9 i the:FEM:NOM:SG tis the:FEM:GEN:SG ‘It’s Mairi’s.’

M´eri-s Mairi:FEM-GEN:SG M´eri-s Mairi:FEM-GEN:SG

´ıne instead of is ´ıne is

By 2;3, the NOM-ACC contrast of FEM NPs is more firmly established, since both cases are marked by the respective forms of the DEF.ART with equal frequency (34 tokens each of i vs. ti/tin), although FEM nouns are more often unaccompanied by the DEF.ART in ACC than in NOM contexts (zero article in 15 NOM vs. 22 ACC contexts). Evidence for the synthetic GEN with FEM nouns remains scarce, since only five of the girl’s 49 FEM nouns are used in GEN contexts, with three of them marked by final –s or varying between their unmarked and marked forms (e.g. ti* M´eris/M´eri* for tis M´eri-s ‘the:FEM:GEN:SG Mairi:FEM-GEN:SG’). As the FEM DEF.ART continues to be limited to the distinction between the NOM:SG i and the child OBL form ti, only a simple contrast between NOM and OBL is accomplished by periphrastic case marking, while the GEN is distinguished from the other two cases by synthetic marking. The most important achievement at 2;9 is the emergence of the GEN:SG form tis of the DEF.ART and the more systematic marking of the GEN on FEM nouns, finally establishing the three-way contrast between NOM, ACC, and GEN in the SG. However, standard and non-standard forms continue to vary in GEN contexts (tis/ti*/∅ N-s, ti*/i* N-V*). Case distinctions of FEM nouns only begin to develop in the third year regarding the other children (for details see Stephany and Christofidou 2008).

5.1.3. Summary of case development Although SG case distinctions marked on MASC and FEM nouns both oppose the unmarked form of the noun ending in the thematic vowel to a marked form ending in –s, the marked forms play different functional roles in the two genders,

The emergence of nominal inflection in Greek

245

expressing agent (MASC) vs. possessor (FEM). Both of these functions are communicatively important from early on, especially so regarding animate referents. However, at least in Mairi’s early speech (on Christos see below), the possessive relation is by far most frequently expressed pronominally (e.g. Mairi 1;9, to pir´uni mu ‘the:NEUT:NOM/ACC:SG fork:NEUT:NOM/ACC:SG’me:GEN’ (= my fork); see Stephany 1997a: 240) rather than by the marked GEN form of FEM nouns (86 PRO tokens vs. 1 N token) (Stephany 1997a: 217). In contrast to this, the marked NOM:SG form of MASC nouns is used for pointing and identifying or expresses the grammatical subject. It is therefore to be expected that this form develops earlier and more quickly than the marked GEN:SG of FEM nouns, in spite of the fact that both are signaled by final –s following the thematic vowel. This is indeed what we have found with the five children studied in the present chapter, irrespective of the age at which the NOM-ACC contrast of MASC nouns and the GEN-NOM/ACC opposition of FEM nouns appear. Once both contrasts have begun to develop, the functional distinctions and distributional differences between the -V/-s patterns of MASC and FEM nouns contribute to a gender distinction of Greek nouns (see Stephany 1997a: 220). Although the Greek DEF.ART is strongly grammaticized and frequently occurs in the input, children tend to omit it at first in a substantial number of tokens (Stephany 1997a: 226; see also Marinis 2003: 125–130). However, we found that the distinction between NOM and ACC forms of FEM nouns (i/ti(n) N-V) emerged before the end of the second year in Christos’ and Mairi’s speech, but only in the first half of the third year in Janna’s and Maria’s. The periphrastic distinction of GEN and ACC of subclass I (‘diptota’) MASC nouns (tu/to(n) N-V) emerges in the period from 1;11 to 2;4 in Christos’ speech, but only at 2;3 in Mairi’s and even after 2;9 and 2;11 in Maria’s and Janna’s. While the synthetically marked GEN of MASC subclass II (‘triptota’) nouns is emerging in Spiros’ speech at 1;9 and in Mairi’s at 2;3, Janna and Maria do not provide evidence of its appearance until the end of observation. The difference between the emergence of the NOM-ACC contrast with FEM nouns and the GEN case in the MASC gender may be attributed to the low frequency of MASC nouns. Christos’ early expression of the possessive relation with MASC nouns by the DEF.ART is due to the fact that, in contrast to Mairi, he denotes possession by using names, including his own, (mostly in the GEN) rather than pronouns. Comparing the respective roles of periphrastic and synthetic case marking in the early development of Greek noun inflection, we have shown that children do not rely on periphrastic means for marking case distinctions, not even with regard to syncretistic case forms such as the ACC:SG and GEN:SG of subclass I MASC nouns and the NOM:SG and ACC:SG of FEM nouns. As has been

246

Ursula Stephany and Anastasia Christofidou

demonstrated above, the DEF.ART nevertheless contributes to unambiguously marking case even if it is not yet systematically used. As soon as article use becomes more reliable, the number of case distinctions becomes more adultlike and the case system is more firmly established (see also Stephany 1997a: 226– 234). Christos is the only child who, in the beginning, inflects all MASC nouns according to the pattern found in the more frequently used subclass I nouns, which distinguish the marked NOM from a general unmarked OBL form used in ACC as well as GEN contexts so that the latter are only distinguished by the form of the article. This results in single inflectional paradigms for MASC and FEM nouns respectively (Christofidou and Stephany 1997: 136; Christofidou 2004) and a corresponding biunique relation between inflectional pattern and gender (regarding NEUT nouns see sect. 5.2.3 below). It is only at 2;3.18 that he starts to distinguish between two declensional patterns of MASC nouns marking animate nouns ending in –os (subclass II) by final –u in the GEN:SG while using the unmarked general OBL form of other MASC nouns (‘diptota’, subclass I) accompanied by the GEN:SG form tu of the DEF.ART for expressing the possessive or benefactive functions. An explanation for this finding could be that case marking by vowel substitution (GEN:SG of subclass II nouns) is less transparent than case marking by adding –s (NOM:SG of both subclasses of MASC nouns). But most importantly, the GEN:SG of MASC nouns also occurs much less frequently than the NOM:SG. Regarding the other children studied in the present paper, there is no evidence for the construction of a unitary twocase distinction of MASC nouns preceding the distinction of two declensional subclasses, although this may be due to the fact that they were not continuously observed.8 Since instances of overgeneralizations of inflectional patterns remain few in the speech of the children studied in the present paper and do not occur in each of the corpora, “there is no ‘stage of overgeneralizations’ to be found in the development of MG nominal (or verbal) inflection” (Stephany 1997a: 225) and thus no evidence “for an overall U-shaped learning curve for the acquisition of word forms” (Stephany 1997a: 323).9 Overgeneralizations are rather analogical formations occurring from early on. NEUT nouns do not occur in the GEN in our data, the only exception being a single, probably rote-learned token found in Janna’s data at 2;11 (tu sxol´ıu ‘the:NEUT:GEN:SG school:NEUT-GEN:SG’). This state of affairs may be explained by the fact that the frequently occurring NEUT diminutives lack the GEN (see sect. 2) and also that many neuters are inanimate and are therefore not used with a possessive function (Christofidou and Stephany 1997: 136; see also Stephany 1997a: 217). There is thus no case distinction to be found with

The emergence of nominal inflection in Greek

247

regard to NEUT nouns in our data so that their inflectional pattern is limited to a number distinction in the NOM/ACC.

5.2.

Development of the plural in Greek nouns

In the speech of the five children studied in the present chapter, the formal distinction of SG and PL is marked early and frequently on the NOM/ACC form of NEUT nouns although it also emerges in FEM nouns before the end of the second year. PL forms of MASC appear in the last part of the third year in the case of only two children and are limited to one or two tokens.

5.2.1. Neuter nouns In Christos’ speech 29% of all NEUT nouns are used in the PL at 2;8 (see fig. 2b, sect. 4.2). When we only take the forms of NEUT nouns into consideration, the plural emerges early and is richly documented both type- and token-wise in Christos’ speech. In the first period (1;7–1;10), NEUT:PL forms are even the only marked forms of nouns Christos uses (9 PL types each ending in –ja (14 tokens) or –a (15 tokens)). In spite of a considerable number of types and tokens of NEUT:PL forms, it remains uncertain whether Christos uses both numbers contrastively in this early period. Of the three PL types already found in the first recording, one belongs to the class of pluralia tantum, a second one only occurs in the PL throughout the entire first period (pedhj´a ‘child:NEUT:NOM/ACC:PL’), and the third one (ghlik´a ‘cake:NEUT:NOM/ACC:PL’) is an imitation.10 The drop of NEUT:PL types in comparison to other inflectionally marked forms, which can be observed in period II (1;11–2;4) (see fig. 2b), indicates that many early plural forms are used as lexical rather than inflected forms. Although in period II types and tokens of NEUT:PL forms increase considerably (27 types and 206 tokens), there is as yet no clear contrastive use of SG and PL forms to be found (for details see Christofidou 1998: 51–52). The first number contrasts are observed after 2;0 (e.g. at 2;3.5, dzip´aki/dzip´akja ‘jeep:DIM:NEUT:NOM/ACC:SG/PL’), but some PL forms continue to be used in both SG and PL contexts (e.g. kak´alja for portok´alja ‘orange:PL’ and portok´ali ‘orange:SG’); also, SG and PL forms may vary in successive utterances (Stephany 2002: 15; for details see Christofidou 1998: 52–53). This state of affairs points to the fact that Christos has not yet fully grasped the contrast between SG and PL forms of nouns, an interpretation also corroborated by informal tests conducted by his mother, where the boy alternates between NEUT PL and SG forms in reference to one or several entitites of the same kind (Christofidou

248

Ursula Stephany and Anastasia Christofidou

1998: 53–54). Finally, at 2;4, the first clear instances of a contrastive use of both number forms are found (example 3). In the third period (2;5–2;8) the use of PL forms of NEUT nouns rises further both type- and token-wise and there is clear evidence that both numbers are distinguished both formally and functionally (see Christofidou 1998). (3)

Christos, 2;4.12 MOT: aft-´a ti ´ıne edh´o? this-NEUT:NOM/ACC:PL what are here ‘What are these over here?’ CHR: pap´akja. duck:DIM:NEUT:NOM/ACC:PL ‘Ducklings.’ MOT: pap´akja. (agreeing) CHR: dhen ´ıne pap´aki not is duck:DIM:NEUT:NOM/ACC:SG a(f)t-´o. this-NEUT:NOM/ACC:SG ‘This (one) is not a duckling.’ MOT: dhen ´ıne pap´aki aft´o? ´ıne kokor´aki. ‘This one is not a duckling? It’s a little cock.’

In Christos’ development synthetic PL marking of NEUT nouns clearly precedes analytic marking. In period I (1;7–1;10), NEUT:PL nouns are not yet accompanied by the DEF.ART, although it must be mentioned that article use is not syntactically obligatory in 28 of the 29 instances. In period II (1;11–2;4) 84% of tokens continue to be bare nouns (n = 207). Also, the DEF.ART does not reliably mark the PL since the forms to ‘the:NEUT:NOM/ACC:SG’ and (rarely) tu ‘the:NEUT:GEN:SG’ vary with correct ta ‘the:NEUT:NOM/ACC:PL’. In the last observational period (2;5–2;8), Christos uses 45% of his plural forms of NEUT nouns with the correct form of the DEF.ART. PL forms of NEUT nouns are well documented in Mairi’s and Spiros’ speech before the end of the second year and in Janna’s and Maria’s in the first half of the third year (for Mairi see fig. 2a, sect. 4.2). Since, with the exception of Janna at 1;11, the absolute number of NEUT nouns by far exceeds that of the other two genders in Stephany’s child data, PL forms of NEUT nouns are accordingly the best documented overtly inflected forms of nouns, both type- and token-wise. In Mairi’s corpus, there is a total of 53 NEUT nouns used in the plural, 17 types at 1;9, 17 at 2;3, and 19 at 2;9 (72, 33, and 50 tokens respectively, 155 tokens overall). In comparison to this, only 22 MASC and 22 FEM nouns occur in the

The emergence of nominal inflection in Greek

249

marked forms of the NOM:SG (73 tokens) and the GEN:SG or NOM/ACC:PL (28 tokens) respectively in her entire corpus. Mairi expresses the PL of NEUT nouns synthetically as well as periphrastically already at 1;9, when 63% of the 84 NOM/ACC:PL tokens are accompanied by the correct form of the DEF.ART (ta) or (rarely) some other determiner or modifier agreeing with the noun in gender, number, and case. At 2;3 there is one instance of an error consisting in a combination of the SG form to of the DEF.ART with a PL form of the noun. This indicates that the synthetic technique of signaling number is more reliable than the periphrastic one in Mairi’s speech. At 2;9, all plural forms of neuter nouns are correctly formed both synthetically and periphrastically. While many PL forms of NEUT nouns occurring in Mairi’s data at 1;9 refer to pairs or common multitudes of entities (e.g. m´atja ‘eyes’, dh´odja ‘teeth’), in the course of development a growing number of NEUT:PL forms belong to countable nouns not generally used in the plural (e.g. pedh-´akja ‘childDIM:NEUT:NOM/ACC:PL’). Although she contrasts the SG and PL forms of nouns of both of these classes formally as well as semantically/pragmatically already by 1;9 (examples 4), the SG and PL of nouns belonging to the second class are opposed more frequently at 2;3 and 2;9. While the percentage of PL forms occurring with NEUT nouns remains more or less stable from 1;9 to 2;9, more NEUT nouns occurring in the PL are contrasted with their SG form at 2;9 than at 1;9 (58% vs. 41%). (4)

Mairi, 1;9 a.

b.

na (min) k(r)i´osi to MOD.PTL (not) catch.a.cold:3S the:SG mor´o mu. baby:NEUT:NOM/ACC:SG of.me ‘So that my baby will not catch a cold.’ MOT: pu p´ai o p´ıthikos? where goes the monkey ‘Where is the monkey going?’ MAI: (s)ta mor´a. (to) the:PL baby:NEUT:NOM/ACC:PL ‘To the babies.’

With the exception of a PL formation of the uncountable babytalk word mim´ı ‘boo-boo’ in e´ xo pol´a mim´ıka* ‘have:1S many:NEUT:NOM/ACC:PL boo-boo: NEUT:PL*’ (= I have many wounds) produced by Mairi at 2;3, almost all of her PL tokens of NEUT nouns are correctly expressed by forms ending in –a

250

Ursula Stephany and Anastasia Christofidou

from 1;9 on. These PL forms either belong to the inflectional pattern -i/-ja (m´ati/m´atja ‘eye/eyes’) with the PL ending –a added to the SG form of the noun11 or they follow the pattern -o/-a (z´oo/z´oa ‘animal/animals’), in which the final thematic vowel is substituted by /a/. In spite of the fact that the -i/-ja pattern is much more frequent than the -o/-a pattern, both of them are well represented in all three stages of Mairi’s development with no overgeneralizations to be observed. In contrast to this, only a single rote-learned form of the rather infrequent imparisyllabic pattern -a/-ta occurs at 2;9 in the girl’s speech (pr´ama-ta ‘thing:NEUT-NOM/ACC:PL’). Given that, at 1;9, 14 NEUT nouns following the -i/-ja pattern occur in the PL (66 tokens) and that SG and PL forms are contrasted with five of them, it may be affirmed that this pattern has become productive in Mairi’s speech before the end of the second year. This claim is supported by several current measures of productivity. Brown and Cazden’s by now classic criterion of appearance of a grammatical morpheme in 90 percent of obligatory contexts in three successive speech samples (see Brown 1973: 271) is fulfilled in Mairi’s eight samples gathered between the ages of 1;9.18 and 1;10.3. As far as paradigmatic relations between forms are concerned, the first of Pizzuto and Caselli’s (1994: 156) two criteria of productivity, namely appearance of the same root in at least two distinct inflected forms (see Gathercole, Sebasti´an, and Soto 1999: 144), is satisfied by those NEUT nouns which appear in both numbers, while their second criterion, which requires that the same inflection is used in the case of at least two different nouns, is by far exceeded by NEUT nouns belonging to the -i/-ja pattern. Besides these criteria, the occurrence of overextensions of inflectional patterns is taken as evidence of analogy or productivity. Mairi produces the first such form at 2;3. The productivity measure of miniparadigms is inapplicable to NEUT nouns in our data, since due to lacking case distinctions they do not occur in more than two distinct inflected forms (see table 4, sect. 4.2). According to the usage-based approach to language, “the productivity of a pattern, expressed in a schema, is largely, though not entirely, determined by its type frequency” (Bybee 2001: 13), which makes a pattern more familiar to the child (Slobin 1985: 1165–1166). The -i/-ja pattern, which is used with many more nouns than the -o/-a pattern (partially due to the high frequency of NEUT diminutives ending in -´aki/-´akja), can be claimed to be more productive in the three stages of Mairi’s language acquisition and will therefore have formed a stronger schema (see also Stephany 1997a: 325). Another explanation of the higher productivity of the -i/-ja pattern observed in Mairi’s as well as Christos’ speech may be its diagrammatic iconicity, which the -o/-a pattern lacks (see Jakobson 1965; Dressler 1987: 102–103; Kilani-Schoch and Dressler 2005).

The emergence of nominal inflection in Greek

251

The -i/-ja pattern found in Christos’ speech between 1;11 and 2;4 also meets both of Pizzuto and Caselli’s criteria for productivity (1994: 156), although it must be recognized that these criteria represent a minimal requirement for inflectional development (Gathercole, Sebasti´an, and Soto 1999: 144; see also Christofidou and Stephany 2003: 108). Since obligatory contexts of PL forms cannot always be determined in Christos’ early data, Brown’s criterion of use in 90% of obligatory contexts is inapplicable. The first overextension of a PL pattern is found after 2;8 in the boy’s speech (Christofidou 1998). Just as with Christos, synthetic marking of the PL of NEUT nouns precedes periphrastic marking also regarding the other three children observed by Stephany so that Mairi’s early periphrastic marking may be an exception. At 1;9, Spiros distinguishes the SG and PL of two NEUT nouns not only formally but also semantically (see Stephany 2002: 16). Maria’s and Janna’s speech provides evidence of a semantic-pragmatic distinction of SG and PL of NEUT nouns in the first half of their second year (for details see Stephany and Christofidou 2008).

5.2.2. Feminine and masculine nouns In contrast to NEUT nouns, which only develop a number distinction in early child Greek, MASC ones are at first limited to a case distinction in the SG. Although FEM nouns develop both a case distinction in the SG and a number distinction, the latter plays a more important role than the former (for Mairi and Christos see figs. 2a and 2b, sect. 4.2). While PL forms of NEUT nouns occur in Christos’ earliest data (1;7–1;10), the first synthetically marked FEM PLs are only found in period II (1;11–2;4), where 10 FEM nouns occur in their PL forms ending in –es (31 tokens). Except for f´okja ‘seal:FEM:NOM/ACC:SG’ (PL f´okjes), there is no evidence for spontaneous contrastive use of both numbers for any of them. The FEM PL develops rapidly between 2;5 and 2;8 documented by 46 PL tokens, 36 of which are contrasted with their SG forms. Although the DEF.ART does not distinguish the NOM:SG and NOM:PL of FEM nouns (i ‘the:FEM:NOM:SG/PL’), 46% of FEM PL nouns are constructed with the DEF.ART in Christos’ period III (n = 46), as compared to only 16% in period II (n = 31). As has been found in the case of NEUT nouns, synthetic number marking is also preponderant in the case of FEM nouns. It is interesting to note that in period II there is one instance in which the ACC:PL of a FEM noun is distinguished from the NOM:PL by the form of the DEF.ART (tis ‘the:FEM:ACC:PL’ vs. i ‘the:FEM:NOM:PL’). This distinction becomes more frequent in period III (12 tokens NOM:PL vs. 9 tokens ACC:PL). In the third

252

Ursula Stephany and Anastasia Christofidou

period, the undeclinable numeral dh´ıo ‘two’ and the acronym BMW (make of car) are once inflected for PL (dh´ıes beb´es for dh´ıo beemv´e ‘two BMW’). Regarding Mairi and Spiros, the PL of FEM nouns has emerged before the end of the second year, but it occurs quite rarely. In Janna’s and Maria’s data it is found in the first half of the third year. As in the case of Christos it is synthetically marked (for details see Stephany and Christofidou 2008). At 1;9, Mairi uses only two of her 27 FEM nouns in their PL form. Both of these nouns usually occur in the plural and follow the -a/-es pattern (e.g. karam´eles ‘candy:FEM-NOM/ACC:PL’). There is no remarkable change in the use of FEM:PL forms in Mairi’s speech at 2;3; three of 49 FEM nouns occur in the PL, two of them following the -a/-es pattern and one being an instance of the less frequent imparisyllabic PL ending in -dhes (maim´u-dhes ‘monkey:FEMNOM/ACC:PL’ from maim´u ‘monkey:FEM:NOM/ACC:SG’). All PL forms are adequately used to refer to more than one referent, although Mairi only contrasts SG and PL with one of the three nouns. She uses the underdifferentiated child form ti of the DEF.ART referring to the ACC of both numbers (Standard Greek ti(n)/tis ‘the:FEM:ACC:SG/PL’) with one SG and one PL form of a FEM noun. This shows that the child relies on synthetic marking for expressing the number distinction. By 2;9, 8 of 45 FEM nouns occur in the PL, with 7 of them following the -a/-es pattern.The ending -es (9 tokens) varies with nonstandard -e (3 tokens). Again, SG and PL are only contrasted with a single noun. Regarding MASC nouns, the number distinction is only found in one (possibly imitated) form in period III (2;5–2;8) of Christos’data and with the exception of a single instance in Mairi’s speech at 2;9, PL forms of MASC nouns do not occur in the data of the children observed by Stephany (see however Stephany 1997a: 237). Since the GEN:PL has not yet emerged in our data, synthetically marked PL case forms are limited to the NOM/ACC with NEUT and FEM nouns and to two isolated MASC:NOM:PL and MASC:ACC:PL forms in Mairi’s speech and one MASC:NOM:PL form in Christos’. There is a slight indication of the emergence of the periphrastically marked NOM-ACC contrast of PL forms of FEM nouns in Christos’ and Janna’s third year.

5.2.3. Summary of number development with a note on gender Inflectional marking of the PL expressed by synthetic means is the earliest inflectional domain to develop with a substantial number of nouns in the NEUT gender and to gain productivity already by 1;9 in the case of Mairi and between 1;11 and 2;4 in the case of Christos. Although the PL is also expressed by synthetic means with FEM nouns from the very beginning, it occurs much

The emergence of nominal inflection in Greek

253

less frequently. There is only slight evidence for the emergence of the PL with MASC nouns in our data and almost none for the NOM-ACC contrast in the PL. Regarding FEM nouns the emergence of the SG-PL distinction precedes case distinctions in the SG in some children and may develop more rapidly than the latter (Christos and Mairi). There is some evidence of the periphrastically expressed NOM-ACC:PL contrast of FEM nouns in Christos’ speech between 2;5 and 2;8 and in Janna’s at 2;11. In contrast to the role played by the DEF.ART in case distinction (see sect. 5.1), “determiner use is in general irrelevant for distinguishing singular and plural” in Greek child speech (Stephany 1997a: 215). Although NEUT nouns marked for PL are frequently accompanied by the NEUT:PL form of the DEF.ART agreeing with the noun in the three grammatical categories of the Greek nominal system,12 “there is no doubt that synthetic marking of the category of number on the noun stem is more basic in Greek language development than its marking by the use of a plural determiner” (Stephany 1997a: 216). The main reason is that the PL is only frequently used with NEUT nouns, where it is clearly marked on the noun. In contrast to learner varieties of second languages lacking inflectional morphology, lexical expression of number by numerals or other quantifiers constructed with nouns unmarked for number is not typical of Greek first language acquisition.13 Since number serves a more strongly semantic-pragmatic function than case, the former may be considered as a less prototypically inflectional category than the latter, sharing properties of derivation “because it is more relevant to the meaning of the noun” (Bybee 1985: 34; see also Christofidou, Doleschal, and Dressler 1991: 76; Christofidou 1998: 56–57; Corbett 2000: 263).14 Still, number is grammaticized in Greek as it is in many languages, participating in the agreement system of verbs and nouns. Although, in early child Greek, use of PL forms of nouns depends on semantic-pragmatic factors, input frequency, or the preceding context, there is evidence for its grammatical function from early on at least with Christos and Mairi, since adjectives or pronouns may agree with nouns in number (e.g. Mairi, 1;9, megh´al-a dh´odja ‘big-NEUT:NOM/ACC:PL tooth:NEUT:NOM/ACC:PL’). As pointed out in section (5.1.3), case distinctions of MASC and FEM nouns result in a biunique relation between inflectional pattern and gender in Christos’ speech and also mainly so in Stephany’s corpus (-s vs. -V MASC:NOM:SG vs. MASC:ACC:SG but -V vs. -s FEM:NOM/ACC:SG vs. FEM:GEN:SG). Number distinction in NEUT nouns follows another pattern (-V vs. -(j)a NEUT:NOM/ ACC:SG vs. NEUT:NOM/ACC:PL) and completes the picture by adding the third gender of the Greek language. Since inflectional patterns of nouns have been found to develop within gender classes (or even semantic subclasses within

254

Ursula Stephany and Anastasia Christofidou

these), the former can be considered as overt markers of the latter. Furthermore, the strong grammaticization of the DEF.ART and its consequently frequent occurrence in the input contribute to the entrenchment of gender distinctions, especially so with undeclinable loanwords (e.g. o Donald ‘the:MASC:NOM:SG Donald’).

6. Conclusions One of the main results of the present study on the development of nominal inflection in Greek is that children start to inflect nouns from early on. In the four children observed before the end of their second year, the ratio of grammatical types per noun exceeds 1.00, albeit only slightly as far as Christos (1;8) and Janna (1;11) are concerned. While these two children nearly conform to the ratio of 1.00 of the ideal premorphological stage (Christos 1.01, Janna 1.03), the other two children reach ratios of 1.08 (Spiros) and even 1.14 (Mairi) at a comparable age (1;9), although it must be taken into consideration that none of the children of Stephany’s corpus were observed from the onset of speech. Interestingly, Mairi’s value is nearly as high as that of Christos’ CDS at 1;7, which is 1.16 (Mairi’s mother’s ratio at 1;9 is 1.20). These results are confirmed by the PBF values found at the same ages. Overall PBF values of nouns are highest for Christos (92%) and Janna (99%) and considerably lower for Spiros (77%) and Mairi (81%). PBF values differ with inflectional (gender) classes of nouns, which shows that inflectional development does not set in simultaneously with all nouns and develops at a different pace in each class. While PBF is highest for FEM nouns in three children’s speech, it is much lower for MASC nouns in Spiros’ and Mairi’s data (Spiros: FEM 86%, MASC 30%; Mairi: FEM 98%, MASC 69%) but only a little lower in Janna’s (FEM 100%, MASC 91%). Only in the case of Christos, does PBF reach 100% in both FEM and MASC nouns, while in NEUT ones it amounts to 82%. Mairi follows the model of the input (FEM 95%, MASC 36%) more closely than Christos (FEM 99%, MASC 56%). An additional parameter indicating that, in spite of low form/lemma and high PBF ratios, nominal inflection has begun to develop, is the number and type of inflectional contrasts. In two of the four children studied before the end of their second year (Mairi and Spiros at 1;9), three types of inflectional contrasts of nouns have emerged, namely the NOM-ACC contrast with MASC nouns and the number contrast with NEUT and FEM ones. In addition, there is evidence of the NOM-GEN contrast of FEM nouns in Mairi’s speech. Regarding Christos, both NEUT noun number forms emerge at 1;7, although there is as yet no evidence

The emergence of nominal inflection in Greek

255

for a semantic-pragmatic distinction between them. The first case contrast to appear in Christos’ speech at 1;11 is between NOM and ACC of MASC nouns. Although the children studied in the present paper were not found to rely much on periphrastic marking for distinguishing between grammatical categories, there is evidence for the emergence of the contrast between NOM:SG and ACC:SG of FEM nouns expressed by the form of the DEF.ART already before the end of the second year, at least in the case of Christos and Mairi. Especially in Mairi’s speech at 1;9 (and in Christos’ between 1;11 and 2;4), unmarked forms of MASC nouns are often accompanied by the MASC:NOM:SG form of the DEF.ART, which increases the number of tokens where the NOM-ACC contrast is expressed. Regarding Mairi the -i/-ja pattern of plural formation of NEUT nouns has become highly productive already by 1;9 according to several criteria. Productivity of this pattern is only achieved by Christos between 1;11 and 2;4 by the weaker criteria of Pizzuto and Caselli (1994). The question is whether there is evidence for a premorphological stage in our data. A positive answer is totally excluded for Mairi since, by 1;9, her speech not only provides evidence of the development of several inflectional categories in the nominal domain, but her mean size of paradigm (form/lemma ratio) of verbs amounting to 2.3 by far exceeds that of nouns (1.14). At this age, she uses 14 different grammatical types of verb forms in an adequate way (Christofidou and Stephany 2003: 96, 101; see also Stephany 1985). Unfortunately, we do not know anything about her earlier inflectional development, since she was first observed from 1;9. Christos’ inflectional development of nouns is much less advanced between 1;7 and 1;11 than Mairi’s at 1;9. Even if we accept that his noun morphology looks as if he were in a premorphological stage (Christofidou 2004), we cannot deny the fact that, between 1;8 and 1;11, he already uses two to three different grammatical types of verb forms fulfilling different communicative functions (Christofidou and Stephany 2003: 96, 100). This indicates that inflection develops locally in domains where it is functionally important for the child. Even Janna, who at 1;11 almost achieves the lowest measures of noun inflection, provides evidence that inflectional development has set in not only in the nominal, but especially so in the verbal domain. At 1;11, she uses three moods, two aspects, two tenses, and both the first and third person SG and PL of verbs (Stephany 1985, 1997a). Even if we played down her achievements in noun inflection we would hardly want to postulate that she is simultaneously in the premorphological and the protomorphological stage of development. The advantage of phase models is that changes in development do not “involve fundamental changes across the entire cognitive domain” such as language acquisition (Karmiloff-Smith 1992: 18) but are postulated to occur “at different

256

Ursula Stephany and Anastasia Christofidou

times across different micro-domains and repeatedly within each domain” (p. 6) such as nominal and verbal inflection and their sub-domains (inflectional/gender classes of nouns, aktionsart classes of verbs). In such models the development of case distinctions regarding MASC nouns, but of number distinctions regarding NEUT ones can easily be accommodated since neither of these grammatical categories develops across-the-board in Greek nouns. Rather, development is local, being at first based on particular lexical items and limited to certain subclasses within larger classes (see also Stephany 1997a: 325). This is at least the state of affairs which emanates from our production data. As long as children only oppose SG and PL of NEUT, but not MASC nouns in their speech, we do not have evidence to affirm that they have generalized the number contrast to all nouns. As pointed out above, the development of nominal inflection in Greek starts out from the inflectionally unmarked forms of nouns ending in their thematic vowel (see also Stephany 1997a). In adult- as well as child-directed speech, these forms represent the most frequently occurring inflected forms. Given that in the earliest phases of Greek language development inflectional contrasts of noun forms are largely or completely missing, must these forms be taken to lack inflection? In order to answer this question it is important to note that, in spite of being inflectionally unmarked, these forms fulfill certain functions not only in adult discourse but also in child speech, since they are learned by the child together with their occurrence in certain contexts and are mostly adequately used from the very beginning. As soon as the child picks up different grammatical forms from the input, with each of them possibly occurring with a different lexical item, and uses them in different contexts, inflectional distinctions can be assumed to have emerged in principle. However, rather than being realized with one and the same noun (e.g. l´ıkos/l´ıko ‘wolf:MASC:NOM/ACC:SG’) they may be distributed over several lexemes in the child’s speech (e.g. Spiros, 1;9, l´ıkos ‘wolf:MASC:NOM:SG’ vs. kathr´efti ‘mirror:MASC:ACC:SG’). Once inflectional distinctions develop with one and the same lexeme so that its forms are contrastively used, we may assume that the children have grasped “the inflectional principle” of Greek (Stephany 1989: 159; see also Stephany 1997a: 322–323). When case contrasts emerge with Christos, they immediately occur with one and the same lexeme (1;11.13, pap´us/pap´u ‘grandfather:MASC:NOM/ACC:SG’). After marked grammatical forms have emerged, the old forms of nouns ending in the thematic vowel, which are underdifferentiated for case in the child’s system (Stephany 1997a: 323), will specialize to fulfill the functions of a specific case form (ACC or OBL) thereby developing into grammatical forms of the adult language (see also Katis 1984; Stephany 1985, 1992). Over the course of time, grammatical distinctions are

The emergence of nominal inflection in Greek

257

generalized and spread out within specific subclasses of lexemes following the same inflectional pattern and beyond. In this view, children are considered to gradually construct the inflectional system of their language, based on the way they hear their caretakers use it. It therefore seems preferable not to assume two strictly distinct stages of inflectional development, a premorphological stage qualitatively different from the following (proto-)morphological one but rather to view inflectional structure as ‘emergent’. Our findings are readily interpretable in the framework of usage-based models of language acquisition. Such theories focus “on concrete examples of specific grammatical structures rather than on abstract, generalizable rules” and “young children’s linguistic skills (and perhaps even adults’) are much less abstract than previously believed” (Karmiloff and Karmiloff-Smith 2001: 139– 140). Also, linguistic structure (grammar) is “not an overarching set of abstract principles, but more a question of spreading of systematicity from individual words, phrases, and small sets” (Hopper 1987: 144). As pointed out by Stephany (1992: 290), in this view “it is not necessary to be able to handle large amounts of data simultaneously in order to learn the grammar of a language.” Further, variation in language use of individual speakers (see Bybee and Hopper (eds.) 2001) and in the construction of grammar by individual children is to be expected in such a theoretical framework. Although we have been able to show in this paper that nominal inflection begins to develop early, i.e. at least in the second half of the second year, it remains an open question how Greek language acquisition starts out in the first half of the second year with early talkers.

Notes ∗

1. 2.

3. 4. 5.

We would like to thank Demetra Katis, Evangelia Thomadaki, and Maria D. Voeikova for thoughtful comments on an earlier version of this paper and Katherine Maye-Saidi for correcting our English. All remaining shortcomings are our own responsibility. On the difference between stage and phase models of development see KarmiloffSmith (1992). Seiler (1958) shows that although declension was based on the phonological structure of noun endings in Ancient Greek, the Greek declensional system changed to a gender-driven one after the Hellenistic period. Computer-assisted coding and analysis of both corpora were effected within the CHILDES Project (MacWhinney 2000). But see sect. 5.2.1. for details. For details on the other three children’s development see Stephany and Christofidou (2008).

258

Ursula Stephany and Anastasia Christofidou

6. For a detailed analysis of definite NPs in Christos’ data see Marinis (2003). 7. For a more detailed analysis of appositional constructions in Greek child speech see Marinis (2003: 191–212). 8. In contrast to our analysis Marinis (2003: 124) affirms “that case syncretism does not seem to influence the acquisition of case marking” arguing that Christos masters marked case forms of Greek nouns of the three genders basically at the same age, especially those of the two subclasses of MASC nouns. In taking only those contexts into consideration which require marked forms of nouns, Marinis leaves aside an essential aspect of the Greek declensional system, namely the functional contrast between marked and unmarked forms, and assumes that marked case forms are acquired across gender classes. The reason why Marinis did not find a significiant difference between the acquisition of marked forms of ‘diptota’ and ‘triptota’ MASC nouns is most probably due to the fact that the GEN occurs much less frequently than the NOM so that the number of tokens of the marked GEN form of ‘triptota’ nouns does not have an important role to play. However, in order to understand the role of case syncretism in acquisition, it is necessary to study the emergence and early contrastive use of case forms rather than their mastery in a detailed manner. 9. On the development of the inflection of the verb see Stephany (1985, 1989, and 1997a) and Christofidou and Stephany (2003). 10. Marinis (2003: 119) postulates that PL marking sets in already at 1;11 in Christos’ speech counting nouns that are attested in both the singular and the plural. In contrast to this, Christofidou (1998) considers Christos’ early use of (mostly neuter) plurals to mainly constitute rote-learned standard reactions to standard pictures in a book or to be imitations, at least until 2;0. 11. Final /i/ becomes a palatal glide before the plural suffix. 12. Agreement errors concerning the category of number also occur. On grammatical agreement in early child Greek see Stephany (1997a: 204–205, 225–226, 251–252, 275 et passim). 13. As pointed out by Stephany (2002: 17–18), the lexical technique of marking plurality by numerals or quantifiers used pronominally or in loose connection with nouns may precede inflectional number marking in first language acquisition (e.g. Spiros, at 1;9, pol-´a ‘many-NEUT:NOM/ACC:PL’). 14. On the cognitive base of number see Wiese (1997) and Stephany (2002: 8–10).

References Anastasiadi-Simeonidi, Anna 2003 I morfologiki domi tis Neas Ellinikis ke i didaktiki tis [Morphological structure and teaching of Modern Greek]. Glossologia 15: 25–34. Bittner, Dagmar, Wolfgang U. Dressler, and Marianne Kilani-Schoch 2003 Introduction. In Development of Verb Inflection in First Language Acquisition: A Cross-Linguistic Perspective, Dagmar Bittner, Wolfgang

The emergence of nominal inflection in Greek

259

U. Dressler, and Marianne Kilani-Schoch (eds.), vii–xxxvii. (Studies on Language Acquisition 21.) Berlin/New York: Mouton de Gruyter. Bittner, Dagmar, Wolfgang U. Dressler, and Marianne Kilani-Schoch (eds.) 2003 Development of Verb Inflection in First Language Acquisition: A Cross-Linguistic Perspective. (Studies on Language Acquisition 21.) Berlin/New York: Mouton de Gruyter. Brown, Roger 1973 A First Language: The Early Stages. Cambridge, MA: Harvard University Press. Bybee, Joan 1985 Morphology: A Study of the Relation between Meaning and Form (Typological Studies in Language 9.) Amsterdam/Philadelphia: John Benjamins. 2001 Phonology and Language Use. Cambridge: Cambridge University Press. Bybee, Joan, and Paul Hopper (eds.) 2001 Frequency and the Emergence of Linguistic Structure. (Typological Studies in Language 45.) Amsterdam/Philadelphia: John Benjamins. Chatzisavas, Andreas 1992 To vasiko leksilogio tis ellinikis [The basic vocabulary of Greek]. In Praktika tou dhiethnous symposiou ja ti sinchroni elliniki glossa [Proceedings of the international symposium of the contemporary Greek language], Sorbonne, 14–15 Sept. 1992, Christos Klairis (ed.), 71–87. Athens: OEDB. Christofidou, Anastasia 1998 Number or case first? Evidence from Modern Greek. In Perspectives on Language Acquisition: Selected Papers from theVII th International Congress for the Study of Child Language, Ayhan Aksu-Ko¸c, Eser ¨ Erguvanlı Taylan, A. Sumru Ozsoy, and Aylin K¨untay (eds.), 46–59. Istanbul: Bo˘gazi¸ci University. 2003 Genos kai klisi tou ellinikou onomatos: Mia fisiki prosengiisi [Gender and declension of the Greek noun: A natural approach]. In To genos [Gender], Anna Anastasiadi-Symeonidi, Angela Ralli, and Despina Chila-Markopoulou (eds.), 100–131. Athens: Patakis. 2004 Provlimata kataktisis tis onomastikis klisis kai apotelesmata mias erevnas periptosis [Problems of the construction of noun inflection and results of a case study]. In Proceedings of the 6 th International Congress for Greek Linguistics, CD ROM edition, University of Crete. Christofidou, Anastasia, Ursula Doleschal, and Wolfgang U. Dressler 1991 Gender agreement via derivational morphology in Greek. Glossologia 9-10: 69–79.

260

Ursula Stephany and Anastasia Christofidou

Christofidou, Anastasia, and Ursula Stephany 1997 The early development of case forms in the speech of a Greek boy: A preliminary investigation. In Pre- and Protomorphology in Language Acquisition, Katarzyna Dziubalska-Kołaczyk (ed.), 127–139. (Papers and Studies in Contrastive Linguistics 33.) Poznan: Adam Mickiewicz University. 2003 Early phases in the development of Greek verb inflection. In Development of Verb Inflection in First Language Acquisition: A CrossLinguistic Perspective, Dagmar Bittner, Wolfgang U. Dressler, and Marianne Kilani-Schoch (eds.), 89–129. (Studies on Language Acquisition 21.) Berlin/New York: Mouton de Gruyter. Clahsen, Harald 1999 Lexical entries and rules of language: A multidisciplinary study of German inflection. Behavioral and Brain Sciences 22: 980–999. Corbett, Greville G. 2000 Number. Cambridge: Cambridge University Press. 2003 Types of typology, illustrated from gender systems. In Noun Phrase Structure in the Languages of Europe, Frans Plank (ed.), 289–334. (Empirical Approaches to Language Typology, EUROTYP 20-7.) Berlin/New York: Mouton de Gruyter. Croft, William, and D. Alan Cruse 2004 Cognitive Linguistics. Cambridge: Cambridge University Press. Dressler, Wolfgang U. 1987 Word formation (WF) as part of natural morphology. In Leitmotifs in Natural Morphology, Wolfgang U. Dressler (ed.), 99–126. Amsterdam: Benjamins. 1997 Productivity and potentiality in inflectional morphology. Clasnet Working Papers 7: 3–22. Dressler, Wolfgang U. (ed.) 1997 Studies in Pre- and Protomorphology. Wien: Verlag der ¨ Osterreichischen Akademie der Wissenschaften. Gathercole, Virginia C. Mueller, Eugenia Sebasti´an, and Pilar Soto 1999 The early acquisition of Spanish verbal morphology:Across-the-board or piecemeal knowledge? The International Journal of Bilingualism 3: 133–182. Hopper, Paul 1987 Emergent grammar. Annual Meeting of the Berkeley Linguistics Society 13: 139–157. Jakobson, Roman 1965 Quest for the essence of language. Diogenes 57: 22–38. Karmiloff-Smith, Annette 1992 Beyond Modularity: A Developmental Perspective on Cognitive Science. Cambridge, MA/London: MIT Press.

The emergence of nominal inflection in Greek

261

Karmiloff, Kyra, and Annette Karmiloff-Smith 2001 Pathways to Language: From Fetus to Adolescent. Cambridge, MA/London: Harvard University Press. Katis, Demetra 1984 The acquisition of the Modern Greek verb: With special reference to the imperfective past and perfect classes. Ph. D. diss. University of Reading, England. Kavoukopoulos, Fotis A. 1989 I dynamiki tis genikis sti neoelliniki [The dynamics of the genitive in Modern Greek]. Studies in Greek Linguistics. Proceedings of the 10th Annual Meeting of the Department of Linguistics, Faculty of Philosophy, Aristotle University of Thessaloniki, 265–284. Thessaloniki: Kiriakidi. 1996 Ousiastika, epitheta kai rimata: Statistikes kai alles episimansis [Nouns, adjectives and verbs: Statistics and other remarks]. In Zitimata neoellinikis glossas: didaktiki prosengisi [Questions of the Modern Greek language: A didactic approach], Georgia Katsimali and Fotis A. Kavoukopoulos (eds.), 7–16. University of Crete: Rethymnon. Kilani-Schoch, Marianne, Anna de Marco, Anastasia Christofidou, Maria Vassilakou, Ralf Vollmann, and Wolfgang U. Dressler 1997 On the demarcation of phases in early morphology acquisition in four languages. In Pre- and Protomorphology in Language Acquisition, Katarzyna Dziubalska-Kołaczyk (ed.), 15–32. (Papers and Studies in Contrastive Linguistics 33.) Pozna´n: Adam Mickiewicz University. Kilani-Schoch, Marianne, and Wolfgang U. Dressler 2002 The emergence of inflectional paradigms in two French corpora: An illustration of general problems of pre- and protomorphology. In Preand Protomorphology: Early Phases of Morphological Development in Nouns and Verbs, Maria D. Voeikova and Wolfgang U. Dressler (eds.), 45–59. (Lincom Studies in Theoretical Linguistics 29.) Munich: Lincom Europa. 2005 Morphologie naturelle et flexion du verbe fran¸cais. (T¨ubinger Beitr¨age zur Lingusitik 488.) T¨ubingen: Gunter Narr Verlag. Klairis, Christos, and Georgios Babiniotis 2005 Grammatiki tis Neas Ellinikis: Domoliturgiki-epikinoniaki. [Modern Greek Grammar: Structural-functional and communicative]. Athens: Ellinika Grammata. Kourmoulis, Georgios 1964 Morfologikai ekseliksis tis ellinikis glossis [Morphological changes of the Greek language]. Epistimoniki Epetirida Filosofikis Scholis Panepistimiou Athinon (EEFSPA) 15: 9–22.

262

Ursula Stephany and Anastasia Christofidou

Laaha, Sabine, and Steven Gillis (eds.) 2007 Typological Perspectives on the Acquisition of Noun and Verb Morphology. (Antwerp Papers in Linguistics 112.) Antwerp: University of Antwerp. Lyons, John 1968 Introduction to Theoretical Linguistics. Cambridge: Cambridge University Press. Mackridge, Peter 1985 The Modern Greek Language: A Descriptive Analysis of Standard Modern Greek. London: Clarendon Press. MacWhinney, Brian 2000 The CHILDES Project: Tools for Analyzing Talk. 2 vols. 3rd ed. Mahwah, NJ/London: Lawrence Erlbaum. Marinis, Theodoros 2003 The Acquisition of the DP in Modern Greek. Amsterdam/Philadelphia: John Benjamins. Mirambel, Andr´e 1959 La langue grecque moderne: Description et analyse. Paris: Klincksieck. Peters, Ann 1977 Language learning strategies. Language 53: 56–73. Pinker, Steven 1991 Rules of language. Science 253: 530–535. Pizzuto, Elena, and M. Christina Caselli 1994 The acquisition of Italian verb morphology in a cross-linguistic perspective. In Other children, Other Languages: Issues in the Theory of Language Acquisition, Yonata Levy (ed.), 137–187. Hillsdale, NJ: Lawrence Erlbaum. Ralli, Angela 2002 The role of morphology in gender determination: Evidence from Modern Greek. Linguistics 40: 519–551. Seiler, Hansjakob 1958 Zur Systematik und Entwicklungsgeschichte der griechischen Nominaldeklination. Glotta 37: 41–67. Setatos, Michail 1998 Grammatiko ke fisiko genos stin Koini Neoelliniki [Grammatical and natural gender in the Modern Greek Koine]. Epistimoniki Epetirida Filosofikis Scholis Thessalonikis 7: 117–136. Slobin, Dan I. 1985 Crosslinguistic evidence for the language-making capacity. In The Crosslinguistic Study of Language Acquisition, Dan I. Slobin (ed.), vol. 2, 1157–1256. Hillsdale, NJ: Lawrence Erlbaum.

The emergence of nominal inflection in Greek 1997

263

The universal, the typological, and the particular in acquisition. In The Crosslinguistic Study of Language Acquisition, Dan I. Slobin (ed.), vol. 5, 1–39. Mahwah, NJ/London: Lawrence Erlbaum.

Stephany, Ursula 1985 Aspekt, Tempus und Modalit¨at: Zur Entwicklung der Verbalgrammatik in der neugriechischen Kindersprache. (Language Universals Series 4.) T¨ubingen: Gunter Narr. 1989 The acquisition of inflectional morphology in English and Greek: A comparison. In Proceedings of the Second Symposium on English and Modern Greek, University of Thessaloniki, 28–30 March, 1988, 141–163. School of English, Department of Theoretical and Applied Linguistics, Aristotle University of Thessaloniki, Thessaloniki. 1992 Grammaticalization in first language acquisition. Zeitschrift f¨ur Phonetik, Sprachwissenschaft und Kommunikationsforschung 45: 289–303. 1997a The acquisition of Greek. InThe Crosslinguistic Study of LanguageAcquisition, Dan I. Slobin (ed.), vol. 4, 183–333. Mahwah, NJ/London: Lawrence Erlbaum. 1997b Diminutives in early child Greek. In Studies in Pre- and Protomorphology, Wolfgang U. Dressler (ed.), 147–156. (Ver¨offentlichungen der Kommission f¨ur Linguistik und Kommunikationsforschung 26.) ¨ Wien: Verlag der Osterreichischen Akademie der Wissenschaften. 2002 Early development of grammatical number: A typological perspective. In Pre- and Protomorphology: Early Phases of Morphological Development in Nouns and Verbs, Maria D. Voeikova and Wolfgang U. Dressler (eds.), 7–23. (Lincom Studies in Theoretical Linguistics 29.) Munich: Lincom Europa. 2006 A cognitive-functional approach to the development of number and case in early child Greek: Lexically based constructions vs. abstract ‘symbolic’ rules. Paper read at the Second Biennial Conference on Cognitive Science, June 9–13, 2006, Saint Petersburg, Russia. Stephany, Ursula, and Anastasia Christofidou 2008 The Acquisition of Greek Case, Number, and Gender: A Usage-based Approach. (Arbeitspapier 55, Neue Folge.) University of Cologne: Institut f¨ur Linguistik, Allgemeine Sprachwissenschaft. Thomadaki, Evangelia 2008 Elliptika klitika paradigmata ke sichnotita: I periptosi ton ipokoristikon [Defective declensional paradigms and frequency: The case of diminutives]. In Glossis charin [For the sake of language], Festschrift for Georgios Babiniotis, Amalia Moser, Ekaterini Bakakou-Orfanou, Christoforos Charalambakis, and Despina Chila-Markopoulou (eds.), 129–140. Athens: Ellinika Grammata.

264

Ursula Stephany and Anastasia Christofidou

Thomadaki, Evangelia, and Ursula Stephany 2007 Diminutives in Greek child language. In The Acquisition of Diminutives: A Cross-Linguistic Perspective, Ineta Savickien´e and Wolfgang U. Dressler (eds.), 89–123. (Language Acquisition and Language Disorders 43.) Amsterdam/Philadelphia: John Benjamins. Tomasello, Michael 2003 Constructing a Language: A Usage-Based Theory of Language Acquisition. Cambridge, MA/London: Harvard University Press. Triantafyllidis, Manolis 1963 I Geniki ton ipokoristikon se –aki kai to neoelliniko klitiko sistima [The genitive of hypocoristics in –aki and the Modern Greek declensional system]. In Apanta ManoliTriantafyllidi, vol. 2, 141–171. Thessaloniki: Aristotle University. 1978 Neoelliniki grammatiki (tis dimotikis) [Modern Greek (demotic) grammar]. Thessaloniki: Aristotle University. Original edition, 1941. Tzartzanos, Achilleas 1991 Neoelliniki sintaksis [Modern Greek syntax]. Thessaloniki: Kiriakidi. Original edition, 1946. Voeikova, Maria, and Wolfgang U. Dressler 2002 Introduction. In Pre- and Protomorphology: Early Phases of Morphological Development in Nouns and Verbs, Maria Voeikova and Wolfgang U. Dressler (eds.), 3–5. (Lincom Studies in Theoretical Linguistics 29.) Munich: Lincom Europa. Voeikova, Maria, and Wolfgang U. Dressler (eds.) 2002 Pre- and Protomorphology: Early Phases of Morphological Development in Nouns and Verbs. (Lincom Studies in Theoretical Linguistics 29.) Munich: Lincom Europa. Voeikova, Maria, and Natalia Gagarina 2002 Early syntax, first lexicon and the acquisition of case forms. In Pre- and Protomorphology: Early Phases of Morphological Development in Nouns and Verbs, Maria D. Voeikova and Wolfgang U. Dressler (eds.), 115–131. (Lincom Studies in Theoretical Linguistics 29.) Munich: Lincom Europa. Wiese, Heike 1997 Zahl und Numerale: Eine Untersuchung zur Korrelation konzeptueller und sprachlicher Strukturen. (Studia Grammatica 44.) Berlin: Akademie Verlag. Xanthos, Aris, and Sabine Laaha 2007 Method. In Typological Perspectives on the Acquisition of Noun and Verb Morphology, Sabine Laaha and Steven Gillis (eds.), 11–20. (Antwerp Papers in Linguistics 112.)Antwerp: University ofAntwerp.

The acquisition of number and case in Austrian German nouns Katharina Korecky-Kr¨oll and Wolfgang U. Dressler 1. Introduction German is a weakly inflecting-fusional language with some properties of the isolating type. Plural is marked by suffixation and/or stem vowel change (umlaut or metaphony) as well as by periphrastic means (articles, attributive adjectives). Case marking relies more on periphrastic means than on nominal case endings. This paper intends to show how the categories of number and case emerge and develop in a child acquiring Austrian German. We will argue that input token frequency is most relevant in the early phases, but that principles of Natural Morphology (i.e. productivity, transparency, and iconicity) play an ever-increasing role later on (cf. Seifert 1985).

2. German noun plurals 2.1.

German noun plural formation

The system of noun pluralization in German consists of a number of phonologically unrelated plural allomorphs with no clearly dominant form. German noun plurals are formed by four different suffixes (-s, -(e)n, -e, -er) or by a zero morpheme. All but the first plural markers may combine with umlaut, thus rendering the plural markings presented in 2.3. The assignment of the plural markers is largely bound to lexical, phonological and (sometimes) semantic characteristics of the nouns (K¨opcke 1993, Wurzel 1994, Wegener 1999). In the plural, articles differ from the singular (and are identical in 3 of 4 case forms with the feminine singular, see table 1).

2.2.

Previous research on the acquisition of German plural morphology

Investigations on the acquisition of German plural morphology (cf. Laaha et al. 2006) focussed on the debate between supporters of dual-route models (e.g.

266

Katharina Korecky-Kr¨oll and Wolfgang U. Dressler

Clahsen et al. 1992, 1996, Clahsen 1999, Pinker 1999, Marcus 2000) and singleroute models (e.g. Rumelhart and McClelland 1986, Plunkett and Marchman 1991, 1993, Daugherty and Seidenberg 1994, Behrens 2001, Tomasello 2003: 235–239). In contrast to English (and German) past tense, where the ‘regular’ weak verb class characterized by the suffix -ed, which represents the default in verb inflection as well as the variant with the highest type frequency, is the only transparent and productive class, these properties do not coincide in any of the German plural classes (see § 2.3 below). This is widely acknowledged in linguistic studies (e.g. Wegener 2002, cf. Dressler 1999) but often neglected or over-simplified in psycholinguistic research. Thus, from a linguistic point of view, -s plurals cannot be the default variant, if not a very weak default or an emergency default (cf. Wegener 2002, 2004 and references quoted). Evidence on the acquisition of German noun plurals cited as support for the dual-route model comes primarily from experiments with nonce words and from studies on overgeneralization errors committed by German-speaking children. Following an acceptability judgment experiment by Marcus et al. (1995) on adult German speakers, Bartke et al. (1995) and Bartke (1998) studied Germanspeaking children (aged 3;1–8;11) in rating plurals of nonce nouns. Results indicated that ‘irregular plural forms’ (i.e., all plural markers except -s) were judged as better when rhyming with actual German nouns than in a non-rhyme condition, whereas -s plurals were judged as better in the non-rhyme than the rhyme condition. For Bartke et al. (1995), Bartke (1998), Clahsen (1999), Niedeggen-Bartke (1999) and Marcus (2000), these findings were evidence for the view (cf. also Pinker 1999) that children (like adults) possess two distinct mechanisms for German plural inflection, a symbol-manipulating rule system for ‘regular’ -s plurals and direct lexical access for ‘irregulars’ (all other plural markers). When children fail to access a lexically stored irregular form (which is the case for unusual-sounding words such as non-rhymes and for non-canonical bases such as proper names), they apply the regular -s plural rule. Due to this qualitatively different (i.e. default) status of -s as opposed to all other German plural markers, they are predicted to generalize -s widely, despite its low type frequency (Marcus 2000: 163). However, so far, nearly all empirical studies on overgeneralization errors have found that German-speaking children overgeneralize a variety of different plural markers, i.e. the -s plural is never the only or the most frequent plural marker (cf. most recently Behrens, 2001, 2002; Bittner and K¨opcke, 2001; Schaner-Wolles 2001; Szagun 2001; Elsen 2002).

The acquisition of number and case in Austrian German nouns

267

Based on results from word detection experiments, Sonnenstuhl-Henning (2003) proposes an extended and refined dual-route model for German plurals, which operates on two different levels of lexical processing. In this model, the dual distinction between storage and computation is not only valid for an inflectional system as a whole, but also within subsystems, e.g. among -(e)n plurals (cf. also Clahsen 2006). In contrast, other researchers regard the diversity of plural error patterns observed in German-speaking children as evidence for frequency- and analogybased generalizations best explained within a single-route model: K¨opcke (1998) adopts an output-oriented, schema-based approach (cf. also Bybee’s 2001 schema model). Elaborating on data from Mugdan’s (1977) nonce-word task with German-speaking children (aged 6;0-7;0) and on longitudinal spontaneous speech data of 7 German children (aged 1;11–2;9), he argues that cue strength of plural markers (i.e. salience, type frequency, cue validity and iconicity) plays a key role in the acquisition of German plural morphology (cf. also Ewers 1999). Overgeneralization errors are claimed to be characterized by a drift from less reliable towards more reliable plural schemas (e.g. -e is not only a plural marker but also a frequent ending of base forms in the singular, thus it is an unreliable plural marker). Bittner and K¨opcke (2001) extend this analysis by proposing a theoretical integration with naturalness criteria (e.g. transparency, uniformity and constructional iconicity) elaborated within Natural Morphology (Mayerthaler 1981, Dressler et al. 1987, Wurzel 1984). In her work on dense longitudinal data of one German boy (aged 1;11–4;0), Behrens (2001) shows that, contrary to default assumptions, -s overgeneralization errors are limited to specific phonological environments in which -s occurs in the input language. Szagun (2001, summarized in Szagun 2004), in her analysis of spontaneous speech corpora of 22 German-speaking children (aged 1;4–3;8), shows a correspondence between children’s growth rates of type frequencies per different plural marker and frequencies observed in the children’s input and the relevance of phonological information for error patterns. The importance of frequency in terms of critical mass effects is stressed in the longitudinal diary study by Elsen (2002 with further literature), which is compatible both with Bybee’s and K¨opcke’s schema approach and with connectionist network models. Previous research of our own group, based on small corpora of longitudinal studies (Vollmann et al. 1997, Sedlak et al. 1998, Tonelli et al. 1998, Klampfer and Korecky-Kr¨oll 2002), has established that the earliest plural forms to emerge and overgeneralize inAustrian German are either only the -(e)n plural or the -(e)n and the -e plural (without umlaut). Moreover the emergence of form oppositions between singular and plural of the same noun appears to mark the transition from

268

Katharina Korecky-Kr¨oll and Wolfgang U. Dressler

the premorphological to the protomorphological phase (see § 5). Further work was based on experimental production data (cf. Klampfer, Korecky-Kr¨oll, and Dressler 2001, Laaha et al. 2006) and has focussed on the role of morphological productivity in determining pace of acquisition.

2.3. The Vienna productivity account In contrast to supporters of dual-route models (see 2.1), who focus on the concept of default, we argue that the acquisition of German noun plurals can be explained by principles of Natural Morphology (cf. Klampfer, Korecky-Kr¨oll, and Dressler 2001).1 The first subtheory of Natural Morphology, the subtheory of universal preferences, predicts that morphological patterns which are highly iconic and morphotactically transparent and accommodate biuniqueness are universally preferred and thus should be acquired earlier than less iconic and less transparent ones. The second subtheory of Natural Morphology, the subtheory of typological adequacy, states in which respects German noun inflection approaches the inflecting-fusional and the isolating type, the first e.g. in the cumulative expression of number, case and gender by the same marker, the latter notably by the use of articles and the frequent lack of inflectional markers on the noun, especially in the expression of case. Thus acquisition of articles is predicted to be decisive for acquisition of case. For reasons of cross-linguistic comparison, the present chapter focuses on synthetic inflectional techniques, but will also treat some quantitative aspects of the development of the article system. Within the third subtheory of Natural Morphology, the subtheory of languagespecific system adequacy, the most important concept is that of morphological productivity which holds for both inflection and derivation (cf. Dressler 2004, Dressler and Lad´anyi 2000, Bauer 2001). We define morphological productivity in inflection (cf. most recently Dressler 2004: 33–34, Kilani-Schoch and Dressler 2005: 128–139) as the capacity of producing and processing freely and automatically new inflectional forms from a base according to a given morphological pattern (or rule) with limitations lying only in the structural description of this abstract pattern. The most important criteria for productivity of the target system are the application to new loan words, to abbreviations, to conversions, and class shift from a less to a more productive pattern. Morphological productivity is understood as a primitive property on the level of the potential system of language (productive plural types can form potential variants to existing plurals whenever their structural description is met), from which type frequency on the level of language as institutional

The acquisition of number and case in Austrian German nouns

269

norm and token frequency on the level of performance are derived (together with intervening pragmatic variables, cf. Dressler and Lad´anyi 2000, Dressler 2004). For performance we adopt (cf. Dressler 2004: 51–55) the race model of Baayen and Schreuder (1991) and Frauenfelder and Schreuder (1992), whereby unproductive forms are nearly always directly accessed, productive infrequent forms computed by rule, whereas there is rivalry between both mechanisms for productive frequent forms. A child needs some time to detect conditions of patterns. This is why – especially at the beginning – input token frequency is very important. As argued in Dressler et al. (1996), morphological productivity can be detected by the child according to several factors: high type frequency coupled with often low token frequency (cf. Bybee 2001), higher morphosemantic transparency, and the often perceptually salient production of novel forms by the caretakers. There are eight main plural markers in German which can be divided into subclasses and differentiated according to their degrees of productivity, transparency, iconicity, type and token frequency: – PL 1: -s plurals (das Auto, die Auto-s ‘cars’) are productive, transparent, and iconic, but not frequent. They are obligatory with nouns ending in unstressed full vowels and excluded from nouns ending in schwa or a sibilant. They are the preferred plurals for unassimilated loan words (cf. Wegener 1999, 2002, 2004). – PL 2: -en plurals (die Frau, die Frau-en ‘women’) are productive and frequent for feminine nouns and weak masculine nouns but unproductive and rare for neuters. They are transparent and iconic. They are obligatory for feminines and (animate) masculines ending in schwa (where they do not increase the number of syllables, in parallel to the -s plural). They are the default plural of feminines. – PL 3: -e plurals (der Fisch, die Fisch-e ‘fish’) are productive, transparent, and iconic. They appear almost only in masculine and neuter nouns, where they are frequent and represent a weak default (default after word-final sibilants, most frequent variant after well-integrated consonant-final nouns). – PL 4: -e + umlaut plurals (die Bank, die B¨ank-e ‘benches’) are productive for masculines, but unproductive for feminines and neuters. Umlaut opacifies. Nouns of this class have a medium type and token frequency. – PL 5: zero plurals (der Lehrer, die Lehrer ‘teachers’) appear almost only in masculines and neuters ending in schwa plus sonorant, they are productive and transparent but not iconic. They have a high type and token frequency. – PL 6: umlaut plurals (derVater, dieV¨ater ‘fathers’) are unproductive, opaque, weakly iconic (because of the modified stem vowel), and rare.

270

Katharina Korecky-Kr¨oll and Wolfgang U. Dressler

– PL 7: -er plurals (das Kind, die Kind-er ‘children’)2 are unproductive, transparent, and iconic and appear only in masculines and neuters with low type and medium token frequency. – PL 8: -er + umlaut plurals (Buch, B¨uch-er ‘books’) differ from PL 7 in being opaque. According to our above-mentioned theoretical account we predict that: PL 1, 2, 3 should be acquired before PL 4 and PL 7 before 8 (iconicity, transparency) and PL 1–4 before 6–8 (productivity). Due to higher type and token frequency PL 2, 3 should emerge before PL 1. PL 5 is non-iconic, but productive, transparent, and frequent and thus should be acquired rather early. However, this prediction is hard to test because of the difficulty of distinguishing between zero plurals and non-application of plurality.

3. German case 3.1.

German case inflection

As in most languages, the nominative is the unmarked case. In weak masculines, the nominative singular is the only slot in the paradigm which has no -(e)n ending, e.g. der Mensch ‘the human being’, des/dem/den/die/der/den/die Mensch-en. Otherwise, inflectional case endings on the noun are only relatively frequent in dative plurals which all take an -n ending (except after -s and -n plurals). Here the expression of number and case is gender-independent and not cumulative (PL Kind-er ‘children’, PL:DAT Kind-er-n). This accommodates the universal preference for a biunique relation between meaning and form (cf. the operating principle of unifunctionality of Slobin 1985: 1227–1228). The other case ending, namely SG:GEN -s, is taken by strong masculine and neuter nouns as well as by proper names and hypocoristics. But genitives are often replaced with a prepositional phrase using von ‘of’ + dative: das F¨ußchen des Hahns → das F¨ußchen vom Hahn ‘the foot-DIM of the cock’ (example taken from the input of the present study). Also the gender-independent expression of the possessive genitive by -s in names and name-like items such as Papa-s Telefon ‘daddy’s telephone’ (another example from the input) partially accommodates biuniqueness. Periphrastic case marking with the help of articles and other modifiers is essential in German, although there are many instances of syncretism and homophony which contradict the principle of biuniqueness. Tables 1 and 2 illustrate the complexity of this system.

The acquisition of number and case in Austrian German nouns

271

Table 1. German case (singular) Case

Feminines

Weak masculines

Strong masculines

Nominativedefinite article

die liebe Katze ‘the dear cat’

der wilde L¨owe der rote Ball ‘the wild lion’ ‘the red ball’

Neuters das kleine Haus ‘the small house’

Nominative- eine liebe Katze ein wilder indefinite arL¨owe ticle

ein roter Ball

ein kleines Haus

Genitive

der/einer lieben Katze

des/eines wilden L¨owen

des/eines roten Balls

des/eines kleinen Hauses

Dative

der/einer lieben Katze

dem/einem wilden L¨owen

dem/einem roten Ball

dem/einem kleinen Haus

Accusative

die/eine liebe Katze

den/einen wilden L¨owen

den/einen roten Ball

das kleine/ ein kleines Haus

Table 2. German case (plural) Case

Feminines

Weak masculines

Strong masculines

Neuters

Nominative die lieben =Accusative- Katzen definite article

die wilden L¨owen

die roten B¨alle

die kleinen H¨auser

Nominative =Accusativeindefinite (no article)

0 liebe Katzen

0 wilde L¨owen

0 rote B¨alle

0 kleine H¨auser

Genitivedefinite article

der lieben Katzen

der wilden L¨owen

der roten B¨alle

der kleinen H¨auser

Genitiveindefinite (no article)

0 lieber Katzen 0 wilder L¨owen 0 roter B¨alle

0 kleiner H¨auser

Dative definite/ indefinite

den/0 lieben Katzen

den/0 kleinen H¨ausern

den/0 wilden L¨owen

den/0 roten B¨allen

272

Katharina Korecky-Kr¨oll and Wolfgang U. Dressler

We expect that this system, which is well-known to be difficult to master for foreigners learning German (cf. MacWhinney 1996) will also present difficulties to a child acquiring morphology (cf. Clark 1998: 379; Slobin 1973: 199, universal D4). The most important determiner is the definite article. In early child language, when case distinctions have not yet emerged or are still marginal, the singular– plural distinction (in the nominative) may be achieved in one of the following ways: 1. only by the article in case of zero plurals (only masculine and neuter as in das Fenster, PL die Fenster ‘the window-s’) 2. only by plural markers on the noun in feminines (die Katze, PL die Katze-n ‘the cat-s’) 3. elsewhere by both the article and a plural marker on the noun (das Auto, PL die Auto-s ‘the car-s’)

3.2.

Previous research on the acquisition of German case

Research on the acquisition of German case has become more specific only in the 1980es. In their longitudinal studies on 3 to 6 children, Clahsen (1984) and Tracy (1984) found a development of case marking in articles, which are the main German case markers, from protoforms such as [då] (cf. Szagun 2004) to overgeneralization of the indefinite ein to accusative marking and of accusative form articles to dative marking before adult-like use. They divide the acquisition of case-marking into four phases: 1) no marking, 2) case-neutral marking, 3) casemarking, 4) accusative forms, 5) dative forms. Basing herself on previous empirical studies Tracy (1986) draws several conclusions within a modular approach to language acquisition, such as the following: MASC:ACC -n develops with the definite before the indefinite article; the accusative develops before the dative in object marking, while in prepositional phrases the reverse order occurs. In a study of two bilingual (German- and French-learning) children, Meisel (1986) found differences as compared to monolingual children, notably as far as the distinction of discrete phases of article development is concerned. He reports that datives (although they are rare) are also overgeneralized to accusative contexts and that datives are marked more clearly and therefore acquired earlier. This is supported by Eisenbeiss, Bartke and Clahsen (2006), who point out that, in adult language, the dative is also more often marked than the accusative case. Moreover they generalize that structural (i.e. syntactically conditioned) case marking (cf. also Wittek and Tomasello 2005) is acquired before lexical case marking.

The acquisition of number and case in Austrian German nouns

273

In a survey article based on previous studies as well as the investigation of one longitudinal corpus, Clahsen, Eisenbeiss, and Vainikka (1994) and Eisenbeiss, Bartke, and Clahsen (2006) suggest the following more specific order of acquisition of German case marking: Genitive marking by -s develops first, followed by the accusative and finally the dative. Both of the latter two cases first develop with pronouns. In commission errors, the dative is substituted by accusative forms, also in prepositional phrases (rare counterexamples have been found by Stenzel 1994). As they work within a nativist approach, they assume inborn syntactic structures and a general category of case from the very beginning. This is of course problematic from a constructivist point of view: We rather assume that the child in the beginning uses undifferentiated all-purpose forms which formally correspond to the most frequent unmarked standard forms, namely the nominative singular. Since in verb-governed positions, the accusative is more central and frequent than the dative, the accusative should precede the dative in these constructions. The noun-dependent genitive has a very different structural position from the verb-dependent accusative. Thus we do not venture any hypothesis on the order of acquisition between accusative and genitive. Omission errors are preponderant with oblique suffixes of the indefinite article ein. In the most thorough study of the development of definite articles, Bittner (2006) reaches the conclusion that at first der and die are overgeneralized and that case distinctions are acquired before gender distinctions. There has been no study so far on the acquisition of the Austrian German case system, which shows signs of less productivity than the standard German one: The genitive almost only appears with proper names and hypocoristics. There is a less clear difference between strong and weak masculine nouns than in Standard German: In Austrian colloquial German, weak masculine nouns which do not end in schwa (e.g. der B¨ar ‘the bear’), often do not take case endings in dative and accusative (dem/den B¨ar vs. Standard dem/den B¨aren). As elsewhere in colloquial usage, the genitive is often replaced by the dative in prepositional phrases and in possessive constructions (see § 3.1). On the other hand datives are often replaced by accusatives, in masculine singulars (where the articles dem/einem are replaced by the accusatives den/einen) as well as in plurals in prepositional phrases (e.g. mit den Kindern ‘with the-PL:DAT children-PL:DAT’ → mit die Kinder ‘with the-PL:ACC children-PL:ACC’). Moreover, even proper names (in other cases but the genitive) normally take an article (Eroms 1989).

274

Katharina Korecky-Kr¨oll and Wolfgang U. Dressler

4. The data This paper is based on the longitudinal study of the Austrian boy Jan3 who was recorded in everyday situations in interaction with his mother. Transcription and morphological coding of the recordings were done according to the CHILDES guidelines (cf. MacWhinney 2000). The CLAN programs were used for analyses. Jan is an early talker who starts to speak around 1;3. In this study, we examined the entire data until the age of 2;0 and part of the data from 2;1 to 2;7. Table 3. Jan’s data Age

Number of sessions 1;3 1 1;4 1 1;5 1 1;6 1 1;7 1 1;8 4 1;9 4 1;10 3 1;11 4 2;0 4 2;1 3 2;2 3 2;3 3 2;4 3 2;5 3 2;6 3 2;7 3

Total min. 45 30 30 30 30 240 240 180 240 240 90 90 90 90 90 90 90

Number of utterances 220 176 173 199 191 1461 1616 1173 1418 1979 759 674 460 445 541 636 560

MLU (in words) 1.027 1.023 1.139 1.111 1.089 1.181 1.169 1.398 1.836 2.035 2.304 2.758 2.572 3.337 3.065 3.030 2.443

SD 0.189 0.149 0.473 0.297 0.285 0.451 0.436 0.617 0.904 0.137 1.336 1.637 1.658 2.001 2.107 1.962 1.859

Noun lemmas4 7 12 15 15 18 122 125 132 149 226 109 95 107 100 92 95 67

Noun Noun types tokens 7 20 12 50 15 42 15 76 18 36 128 610 136 573 141 544 166 683 260 1150 122 421 104 384 114 274 109 243 105 263 102 236 72 115

5. Demarcation of phases We divide the development of morphology in a child into three phases: premorphology, protomorphology, and morphology proper (cf. Dressler and Karpf 1995, Dressler 1997, Bittner, Dressler, and Kilani-Schoch 2003). In premorphology, the only morphological operations are extragrammatical ones or rote-learnt precursors of later grammatical rules (in the sense of MacWhinney 1978, see Dressler and Karpf 1995, Kilani-Schoch and Dressler 2002). In protomorphology, the system of morphological grammar starts to develop in nouns and verbs, but does not yet show a variety comparable to the adult system. Protomorphology is also characterized by a spurt of syntactic develop-

The acquisition of number and case in Austrian German nouns

275

ment, by an increase of articles and pronouns. It is the phase when first miniparadigms emerge, i.e. incomplete paradigms which show that the child masters part of the target paradigm of the adult language (Kilani-Schoch and Dressler 2002): a ‘true’ mini-paradigm is a non-isolated set of minimally 3 accurate and distinct inflectional forms of the same [. . . ] lexeme produced spontaneously in contrasting contexts. An extended mini-paradigm includes also prefixed forms in the verb and diminutives in the noun. The next phase is morphology proper, which shows a qualitatively adult-like system of morphology (including a largely correct use of periphrastic morphology). In Jan’s corpus, we date the beginning of the protomorphological phase at 1;8, when first oppositions of singular vs. plural, of simplex vs. compound vs. diminutive forms in nouns and of first verb forms emerge. Jan’s further progress in morphology takes place earlier with nouns than with verbs (from 1;11 onwards, cf. Klampfer and Korecky-Kr¨oll 2002). Nevertheless, it is still difficult to determine the start of morphology proper in Jan (tentatively 2;7 for nouns, see § 7.2 and § 8) without studying his further development of verb and adjective morphology (which cannot be done in this chapter).

6. The acquisition of noun plurals 6.1.

Quantitative overview

In Jan’s data, there are a total of 634 noun lemmas and 100 plural word types (5,720 noun tokens, 456 plural tokens). Thus 15.77% of his noun lemmas have a plural form, and 7.97% of his noun tokens are plurals (see table 4). The distribution in the input6 is 1,513 noun lemmas, 320 plural word types (13,211 noun tokens, 1,432 plural tokens) i.e. 21.15% of the noun lemmas are plural word types (10.84% of the noun tokens are plural tokens). From 2;2 onwards, Jan has approximately reached his mother’s plural token frequency.

6.2.

Distribution and emergence of plural classes in Jan’s data

6.2.1. Distribution of different plural classes in Jan’s and his mother’s speech Table 5 demonstrates the quantitative distribution of correct plurals according to their plural markers in Jan’s data and in the input. Jan shows a similar distribution to that of his mother: most plurals are -n plurals (in type and token frequency)

276

Katharina Korecky-Kr¨oll and Wolfgang U. Dressler

Table 4. Jan’s spontaneous plural types and tokens in relation to noun lemmas and noun tokens5 Age

1;3 1;4 1;5 1;6 1;7 1;8 1;9 1;10 1;11 2;0 2;1 2;2 2;3 2;4 2;5 2;6 2;7

Noun lemmas (absolute) 7 12 15 15 18 122 125 132 149 226 109 95 107 100 92 95 67

Noun tokens (absolute) 20 50 42 76 36 610 573 544 683 1150 421 384 274 243 263 236 115

Plural types (absolute) 0 2 0 2 0 12 10 10 17 29 19 14 13 14 21 10 12

Plural types (% of noun lemmas) 0.00% 16.67% 0.00% 13.33% 0.00% 9.84% 8.00% 7.58% 11.41% 12.83% 17.43% 14.74% 12.15% 14.00% 22.83% 10.53% 17.91%

Plural tokens (absolute) 0 4 0 5 0 34 30 17 67 93 29 39 31 20 44 28 15

Plural tokens (% of noun tokens) 0.00% 8.00% 0.00% 6.58% 0.00% 5.57% 5.24% 3.13% 9.81% 8.09% 6.89% 10.16% 11.31% 8.23% 16.73% 11.86% 13.04%

and the smallest number of plurals is found in the unproductive umlaut plural class 6. Zero plurals are more frequent in the mother’s data because the number of Jan’s ambiguous forms is considerable and has been excluded. -e and -e+umlaut plurals have a relatively high type and token frequency both in Jan’s and mother’s data; -s plurals, -er and -er+umlaut plurals have a low type but a relatively high token frequency for the unproductive plurals, whereas high token frequency of -s plurals is due to the form Auto-s and its compounds (69 tokens) as well as to Teletubbie-s (40 tokens), which together amount to 93.97% of Jan’s -s plural tokens. A chi-square test7 of the similarity between plural class distribution in the child’s output and his input shows a significant dissimilarity for the tokens (p < 0.001; df = 7). Thus token frequency distribution within Jan’s output cannot be directly derived from token frequency distribution within the maternal input. However, there appeared no significant dissimilarity in type frequency (p = 0.352; df = 5) as far as the presence vs. absence of plural items in input and output is concerned. A qualitative comparison of plural types shows that the following plurals appear only in Jan’s output (from most to least frequent): Schnellbahn-en ‘high-speed railways’, Tixo-s ‘sticky tapes’, Hose-n ‘trousers’

The acquisition of number and case in Austrian German nouns

277

Table 5. Distribution of plural markers in Jan’s data and the input (only correct and unambiguous forms) 1;3–2;7 Absolute numbers Jan Types Mother Types Jan Tokens Mother Tokens

s

% Jan Types Mother Types Jan Tokens Mother Tokens

s

7 23 116 200

7.53 7.19 27.68 13.97

en 39 106 131 388

e 16 59 67 292

e+U 0 9 11 26 68 30 26 103 213

U 2 7 2 25

4 15 19 100

er+U Total 5 93 16 320 28 419 111 1432

en 41.94 33.13 31.26 27.09

e 17.20 18.44 15.99 20.39

e+U 9.68 8.13 7.16 7.19

U 2.15 2.19 0.48 1.75

er 4.30 4.69 4.53 6.98

er+U 5.38 5.00 6.68 7.75

0 11.83 21.25 6.21 14.87

er

100.00 100.00 100.00 100.00

(mother uses only compounds), Ribisel-n ‘red currant-s’, Regenschirm-e ‘umbrellas’, Bauerntier-e ‘farmer’s animals’ (mother uses only the simplicia), Reh-e ‘roedeer’, Babym¨aus-e ‘baby mice’, Mamam¨aus-e ‘mummy mice’ (mother uses only the simplicia), H¨aschen-0 ‘bunnies-DIM’, Packerl-n ‘parcels-DIM’, Tropferl-n ‘drops-DIM’, W¨urferl-n ‘cubes-DIM’, Wuzerl-n ‘little pieces of dirt’ (context not completely clear). With the exception of both Maus compounds, which may be triggered by the simplex, all noun plurals only used by Jan belong to productive (sub)classes. Frequent plural tokens of the productive classes may be even more frequent in Jan’s speech but concern only very few lexemes (26.01% of his plural tokens vs. only 11.17% of the input plural tokens are Auto-s and Teletubbie-s). On the other hand, even frequent nouns of the unproductive classes (-er, -er+U, U ) are less frequent8 in Jan’s than in his mother’s speech (e.g. mother: 70 tokens of Kind-er ‘children’, Jan: 12 tokens i.e. 4.89% of input vs. 2.86% of output plural tokens; mother: 47 tokens of B¨uch-er ‘books’, Jan: 3 tokens i.e. 3.28% of input vs. 0.76% of output plural tokens).

6.2.2. Emergence and development of plural forms in Jan’s data (types/tokens) The distribution of plural markers will now be examined in the course of their emergence. Table 6 shows that from the beginning of protomorphology (1;8), Jan uses several plural markers correctly. Our criteria for productive use of a plural marker are 1) contrasting use of singular/plural form oppositions of the same lemma in one month, 2) several plural types of one plural class in the same month and 3) (not necessarily, but as an additional indicator) overgeneraliza-

278

Katharina Korecky-Kr¨oll and Wolfgang U. Dressler

Table 6. The distribution of plural forms per month (only correct and unambiguous forms, types/tokens) Age

1;3 1;4 1;5 1;6 1;7 1;8 1;9 1;10 1;11 2;0 2;1 2;2 2;3 2;4 2;5 2;6 2;7

-s

2/7 2/8 2/26 5/16 2/4 2/14 3/11 1/4 1/6 2/17 1/3

-(e)n -(e)n after schwa otherwise (f, m, n) (f, m, n; -l ending)

3/7,–,– 4/6,–, 1/9 3/6,–,1/5 3/4,–,1/6 4/8,–, 1/1 3/6,–,– 1/2,–, – 1/1,1/1,– 6/14,1/1,– 3/3,1/1,– 1/1,–,–

1/3,1/1,–;– –,–,1/2;– –,1/1,1/1;– –,–,1/10; 2/6 1/3,–,1/7; 1/1 1/1,–,1/1; 1/1 –,1/1,1/1;– 1/1,–,–;– –,–,–;1/2 1/2-,–;1/1 1/1,–,–;– –,–,– ;2/2

-e

-e+U (m, f)

1/1

1/3,–

1/4

1/1,–

3/7 2/6 1/1 – 8/28 2/3 – 1/2 3/5 4/4 1/3 3/3

2/4,– 1/1,– 1/1,1/5 –,1/1 – 1/1,3/9 –,1/1 –,1/2 – 1/1,–

zero U

-er

1/6 1/1 2/2 1/2 2/4 3/4 1/1 – 2/2 2/3 1/1

1/1 1/2 3/6 3/6 – 1/3 – 1/1 – – –

1/1 1/1 – – –

-er +U

1/1 1/9 1/1 1/6 1/2 1/1 2/5 – 2/3

tions of the respective marker. Feminines ending in -en and -s plurals are thus used productively from 1;8 onwards, when first contrasting form oppositions of singular and plural forms appear; -er plurals emerge earlier than less transparent -er+U plurals, whereas -e and -e+U plurals both emerge at 1;4 (the unproductive feminines only at 2;0), but are productively used only from 1;8 (-e plurals) and 2;0 (-e+U plurals) onwards. -e plurals also increase earlier in type and token frequency than their corresponding opaque -e+U counterparts. Unambiguous zero plural forms are used from 1;9 onwards, when the very first determiners emerge. The unproductive, opaque, and weakly iconic umlaut plural which has also low input frequency emerges only at 2;3 and is extremely rare. The first plurals to emerge at the age of 1;4 belong to the productive classes 3 and 4 (-e and -e+umlaut plurals): Keks-e ‘cookies’, Z¨ahn-e ‘teeth’ (cf. Klampfer and Korecky-Kr¨oll 2002).They are clearly rote-learned, cf. at 1;6 plural-dominant Schuhe ‘shoes’. The first singular/plural contrasts consist of imitated -e and -en plurals, e.g. 1;5 Fisch – Fische ‘fish’, Ohr – Ohren ‘ears’. At 1;8, the first spontaneous -en plurals (U-Bahn-en ‘underground railways’) and -s plurals (Auto-s ‘cars’) appear, even with contrasting singular forms. In the same month, there are further -s, -en, -e and -e+umlaut plurals without usage of

The acquisition of number and case in Austrian German nouns

279

Table 7. Emergence of plural markers in Jan’s data Age 1;4 1;8 1;8

1;9 1;11 2;0 2;3

Plural -e, -e+U (no contrasting singular) -s, -en (consonant-final feminine): both with contrasting singulars -en (weak masculine with schwa ending): without contrasting singular -en (ambiguous form: illegal –en plural or correct dative plural) -er : without contrasting singular -er+U : without contrasting singular 1st extended mini-paradigm: -e+U pure umlaut: with earlier contrasting singular

Example Keks-e, Z¨ahne Auto-s, U-Bahnen Elefant-en Pferd-en Kind-er Bl¨att-er Hand – H¨and-e –Hand-i V¨ogel

the singular: e.g. Foto-s ‘photos’, Blume-n ‘flowers’, Sache-n ‘things, matters’, Seifenblase-n ‘soap-bubbles’, Elefant-en ‘elephants’(a weak masculine), Stein-e ‘stones’, Haar-e ‘hair’, F¨uß-e ‘feet’. At 1;9, the first unproductive -er plural emerges: Kind-er ,children’. The corresponding singular Kind only appears as an imitated form, but there are several compounds (Kinderzimmer with the corresponding simple noun Zimmer, Kindergarten without the corresponding simple noun) which may strengthen the plural form but weaken its decomposability (see also Klampfer and KoreckyKr¨oll 2002: 65–67, Dressler, Kilani-Schoch, and Klampfer 2004: 401–402). At 1;11, a first -er+umlaut plural appears: Bl¨att-er ‘leaves’. At 2;0, the first and (until 2;7 only) extended miniparadigm (see § 5) can be found: FEM Hand (SG) – H¨ande (PL) – Handi (DIM) ‘hand’. Jan’s first correct use of the unproductive umlaut plural class 6 only appears at 2;3 (e.g. V¨ogel ‘birds’). From 2;3 to 2;7, when all plural markers have emerged, there are still some commission errors, but from 1;11 onwards Jan makes considerable progress in periphrastic plural marking due to the increase of the use of determiners (see also § 7.2 on periphrastic case marking).

6.2.3. Development of agreement Plural agreement in verbs emerges two months later than nominal plural subjects9 . Until 1;9, there are many infinitive and stem forms as well as ambiguous verb forms in Jan’s data which could be interpreted as either infinitives or 1st or 3rd person plural forms. A first unambiguous use of the 3rd person plural appears at 1;10: Leute trinken ‘people drink-3P’. During this period, Jan also sometimes uses the amalgam daisie [: da ist sie] ‘there is she’ as a universal verbal agreement marker, but there are many

280

Katharina Korecky-Kr¨oll and Wolfgang U. Dressler

instances where either gender or number of the predicate noun does not agree with the pronoun: daisie M¨ullauto ‘there is she garbage car:NEUT’ (1;9), daisie *Zugen ‘there is she train-PL’ (1;10). At 1;10, there is also a first example of an attributive adjective agreeing with its nominal head: male [: normal-e] Auto-s ‘normal-PL car-PL’. Although both the agreement of the verb with its subject in person and number and that of the attributive adjective with its nominal head in gender, case and number emerge at 1;10, verbal agreement is acquired earlier and shows only few errors after age 2;2. This can be explained by the greater transparency and uniformity of verbal agreement as compared to that of the attributive adjective. Agreement within the noun phrase is not transparent at all: correct choice of the form of the adjective depends on number, case, and gender of the noun on the one hand and on definiteness of the article on the other. In addition, there is a high degree of syncretism. Thus, errors are relatively frequent until the age of 2;6 (see also § 7.2).

6.3.

Omission and commission errors

In German, the zero plural poses a problem because it is homophonous with the singular form of the noun. Furthermore, in dialectal Austrian German, zero plurals may be used instead of -s plurals, although such a form is found only once in Jan’s input (viele Bussi-0 ‘many kisses’). At 2;5, Jan’s plural Pyjama-0 instead of Pyjama-s) ‘pyjamas’ (in a plural context) may be due to such dialectal influence. Moreover, word-final schwa deletion may lead to phonetic zero plurals (or pure umlaut plurals). There are few examples in the input (3 types and 6 tokens all concerning -e plurals e.g. die Keks-0 for standard die Keks-e ‘the cookies’). Jan’s pure umlaut plural masc.*H¨ut for standard H¨ut-e ‘hats’ at 2;0 may be a case in point. These potential forms are not counted as overgeneralizations in the following list: (1)

Jan’s overgeneralizations of plurals (1;10–2;2) Error Singular Target plural Gloss 1 Jeepen Jeep Jeeps jeep 1 Zugen Zug Z¨uge train 1 Tischen Tisch Tische table 1 Tubbytoasten Tubbytoast Tubbytoasts tubby toast 1 Vogeln Vogel V¨ogel bird 2 V¨ogeln Vogel V¨ogel bird 2 Hute Hute H¨ute hat

The acquisition of number and case in Austrian German nouns

281

At the age of 1;10, Jan uses the incorrect -en plurals *Jeep-en instead of Jeep-s, and Zug-en instead of Z¨ug-e) ‘trains’, which supports our hypothesis that small children tend to overgeneralize -en plurals due to their particularly high type and relatively high token frequency in the input. Note that these -en plurals are not potential adult plural variants since these nouns are strong masculines. At 2;0, first overgeneralized -e plurals appear: *Hut-e instead of H¨ute ‘hats’, *Station-e (for Station-en) ‘stops, stations’. -e plurals have the second highest token frequency in the input, but are never used with feminines such as Station, a class of feminine nouns with the latinate suffix -tion, which always take an -en ending. In contrast, Hute is a potential and more transparent plural form. At 2;1, non-potential substitutions of -en plurals still occur in strong masculine nouns: *Tubbytoast-en (for Tubbytoasts taken from Teletubbies, a popular TV series for children), *Tisch-en (for Tische) ‘tables’ (both with corresponding singular forms). There is also one plural form (eine Tubbybett-en ‘a tubbybeds’) combined with a singular article. From 2;2 onwards, Jan replaces unsuffixed plural forms with more iconic suffixed forms. Thus, instead of the unproductive, opaque and non-iconic masculine plural V¨ogel ‘birds’, the more productive, transparent and iconic -en plural *Vogel-n is used. A second commission error *V¨ogel-n makes this plural illegal but still more iconic (there are no combinations of umlaut and -n in German). Nevertheless, the child may have heard this form because it is a correct dative plural. Jan’s productions up to 2;2 are largely compatible with Behrens’(2002) results who also reports a high amount of -n overgeneralizations and several -e overgeneralizations in the early phases. Nevertheless, Jan does not overgeneralize -s,-e+U, -er, -er+U plurals at this early age (in contrast to Behrens’ child who produces all these kinds of overgeneralizations). (2)

Jan’s overgeneralizations of plurals (2;3–2;7) Error Singular Target plural Gloss 1 Fensters Fenster Fenster window 1 H¨unde Hund Hunde dog 2 Leuten – Leute people 1 Gespenste Gespenst Gespenster ghost 1 B¨uche Buch B¨ucher book 3 M¨ausen Maus M¨ause mouse

At 2;3, the iconic plural *Fenster-s replaces the zero neuter plural Fenster ‘windows’. At 2;4, the potential, but more opaque plural form *H¨und-e is found instead of the transparent standard form Hund-e ‘dogs’. The unproductive -er plural Gespenst-er ,ghosts’ is replaced with the productive *Gespenst-e which is a potential plural form (also *B¨uch-e for neuter B¨uch-er ‘books’ at 2;5). The

282

Katharina Korecky-Kr¨oll and Wolfgang U. Dressler

default –n plural of feminine nouns is used in *M¨aus-en (for M¨aus-e ,mice’, which is once used as a singular by Jan in the same recording at 2;5 and thus may work for him like a regular feminine ending in schwa like Blume – Blume-n ‘flower-s’). The -n plural is also found with the plurale tantum *Leuten (for Leute) ‘people’, in which gender is neutralized. At 2;6, the boy gives evidence of a mini-paradigm of this lemma, namely *Leut – Leute – Leute-n (singular – plural – dative plural) ‘people’. Apart from classical overgeneralization errors, Jan pluralizes Lego ‘lego’ by saying Lego-s in order to name his lego bricks at the age of 2;7. Some German-speaking adults also use the word as a pluralizable noun but not Jan’s family who treat it as a mass noun. Comparing Jan’s overgeneralization errors with his correct productions, it can be seen that he often uses the same plural ending both correctly and incorrectly in one and the same month (e.g. 1;10: *Zug-en ‘trains’, *Jeep-en ‘jeeps’ vs. B¨ar-en ‘bears’, Ohr-en ‘ears’). At 2;0, when he seems to discover –e plurals (10 correct form types, 28 correct tokens, form opposition Berg – Berge ‘mountains’), he uses both *Hut-e ‘hats’ and *Station-e ‘stations’. At 2;1, there is correct Tubbybett-en ‘tubby beds’ and incorrect *Tubbytoast-en ‘tubby toasts’. Interpreting our findings of Jan’s development of plural marking in the light of other findings it is at first necessary to point out that there is neither evidence for either an -s default plural (as claimed by Marcus et al. 1995, Bartke et al. 1995, Clahsen 1999, Pinker 1999) nor for an early -n default plural posited by Clahsen et al. (1992: 253), who found that both dysphasic and normal “children initially miscategorize -n as the default plural” before correcting themselves later. Although there are several -n overgeneralizations to be found in Jan’s data, -e analogies and later on -e+U analogies occur as well. The only -s analogy is *Fensters (for Fenster) ‘windows’ (zero marking). But in contrast to Behrens’ (2002, 2001: 24) finding that the child she studied “quickly identifies the full set of German plural markers and overgeneralizes all of them”, Jan does not overgeneralize unproductive -er, -er+U and U plurals. Behrens further states that “the error patterns observed are compatible with [...] learning accounts that predict that children will generalize based on the distributional properties of the target system”. When Jan starts to overgeneralize, he does not yet seem to recognize these distributional properties (e.g. Zugen and Tubbytoasten are inanimate masculine nouns and thus very unlikely to form an -en plural), but starting with 2;3, his overgeneralizations correspond to potential adult plurals, i.e. he does not produce anymore illegal plurals. By the end of the period analyzed so far (1;3–2;7), Jan uses plural formations creatively, seemingly taking into consideration both productivity and iconicity, although he does not yet, of course, master all aspects of the complex German plural marking system.

The acquisition of number and case in Austrian German nouns

283

7. The acquisition of case 7.1. Acquisition of case endings in nouns Genitive singular forms of nouns expressing the possessive are used by Jan from 1;10 until 2;7 (7 types/32 tokens). At 2;0 and 2;1, they have their highest token frequency (4 types/12 tokens, 3 types/9 tokens, respectively) in picture book sessions on Teletubbies (e.g. FEM Laala-s Ball ‘Laalaa’s ball’, MASC Dipsy-s Hut ‘Dipsy’s hat’), where these forms are elicited by pictures. In contrast to genitive, dative marking first appears in the plural. Ambiguous dative plurals (which might also be taken as incorrect -en plurals, e.g. Pferden ‘horse-PL2*’ or ‘horse-PL-DAT’) first appear at 1;8 and decrease at 2;1. They occur in clear plural contexts, but the case remains unclear because most of the utterances are one-word phrases. Datives are mainly used in prepositional phrases, first with prepositions omitted. The earliest relatively unambiguous dative plural in an intended prepositional phrase is found at 1;10 (ex. 3). The interpretation of this example as a dative plural rather than an overgeneralized plural form is crucially based on the assumption that the preposition zu ‘to’ has been omitted. This is quite probable since there is only one preposition (mit ‘with’) at age 1;10: (3)

*MOT: und wo willst du hin? ‘and where do you want to (go) to?’ *JAN: stein-e-n. stone-PL-DAT *MOT: ins kinderzimmer ‘into the children’s room’ *MOT: zu den legostein-e-n? to the lego stone-PL-DAT (= bricks)

At 2;0, the first contrasts of nominative/accusative and dative forms of one and the same lemma appear (baustein-e vs. baustein-e-n in ex. 4). (4)

a.

*MOT: nein, das kusch(e)lauto (=stressed) no the cuddly car:ACC kamma [: kann man] auch nicht ess(e)n. can one:NOM also not eat ‘No, you cannot eat the cuddly car either.’ *JAN: baustein-e (= stressed) auch nicht ess(e)n. building brick-PL:ACC also not eat. ‘Not eat building bricks either.’

284

Katharina Korecky-Kr¨oll and Wolfgang U. Dressler

b.

*JAN:

baustein-e-n pieln [: spielen]. building brick-PL-DAT play. *MOT: du moechtest you want mit den baustein-e-n spiel(e)n? with the building brick-PL-DAT play ‘You would like to play with the building bricks?’

A clear dative plural with a correct article and the correct ending -n is only found at 2;5: in den Berg-e-n ‘in the mountain-PL-DAT’. Thus we find the following order of acquisition of inflectional case markings: 1. SG:GEN, 2. PL-DAT. Weak masculines (e.g. L¨owe ‘lion’) are not marked for case in the singular until the age of 2;7.

7.2. Acquisition of periphrastic case markers As mentioned in § 3.2, determiners and attributive adjectives serve as the main loci for case marking in German. While both determiners and attributive adjectives are absent from the earliest recordings (up to 1;8), their use ‘explodes’ around 2;0 and especially so at 2;2 (tables 8 and 10).10 In contrast to many other children, Jan uses determiners from relatively early on (see also table 10). Indefinite articles emerge before definite ones (1;9 vs.

Table 8. Definite and indefinite determiners in relation to noun tokens Age

Def. det Indef. det Noun tokens Def. det/ Noun tokens 1;8 0 0 610 0 1;9 0 1 573 0 1;10 1 3 544 0.18% 1;11 5 11 683 0.73% 2;0 40 30 1150 3.48% 2;1 39 61 421 9.26% 2;2 167 66 384 43.49% 2;3 96 67 274 35.04% 2;4 116 52 243 47.74% 2;5 115 50 263 43.73% 2;6 85 111 236 36.02% 2;7 39 53 115 33.91% Input (1;3–2;7) 5901 3548 13 211 44.67%

Indef. det/ Noun tokens 0 0.17% 0.55% 1.61% 2.61% 14.49% 17.19% 24.45% 21.40% 19.01% 47.03% 46.09% 26.86%

The acquisition of number and case in Austrian German nouns

285

1;10). Reasons coming to mind are that in contrast to definite articles indefinite ones are often stressed so that the former are less salient. Also, all inflectional forms of the indefinite article contain the base ein so that they are completely transparent (ein, eine, einer, einem, einen, einer, eines), something that is not the case with the definite article (der, die, das, des, dem, den). Between 2;0 and 2;5 (with the exception of 2;1), there are more tokens of the definite articles than of the indefinite one, which corresponds to the input (44.67% definite vs. 26.86% indefinite article tokens). From 2;6 onwards, definite articles are again overtaken by indefinite ones in Jan’s data. Jan’s first determiner at 1;9 is the form ein ‘a’ of the indefinite article, which is the most frequent determiner form in the input and which also corresponds to the numeral ‘one’. The child first uses the form ein irrespective of case or gender distinctions (e.g. *ein Socken [MASC:SG:NOM or NEUT:SG:NOM/ACC] or for ein-en Socken [MASC:SG:ACC)] ‘a sock’ in an accusative context). It is of course impossible to tell whether gender (neuter instead of masculine) or case (nominative instead of accusative) is incorrect (see also footnote 16). Tables 9a, 9b and 10 show the development of agreement of the determiner and/or adjective with the noun: The boy’s first correct agreement forms are indefinite feminine articles with feminine nouns e.g. noch eine Suppe ‘anotherFEM:SG soup’ (1;10). From 2;1 onwards, when articles become more frequent, two or three periphrastically marked datives are to be found each month. First there are only feminines such as auf der Straße ‘in the-FEM:SG:DAT street’, but from 2;2 onwards also masculines and neuters with the marked article dem occur: mit dem Schiff ‘with the-NEUT:SG:DAT ship’, Bussi dem Kasperl ‘kiss (to) the-MASC:SG:DAT Punch’. In contrast to der, which can either be MASC:SG:NOM, FEM:SG:GEN, FEM:SG:DAT or PL:GEN, dem can only be a dative. Inflected forms of attributive adjectives emerge at 1;9 (hoher Turm ‘high tower’, which seems to be rote-learned) and become more frequent and diverse from 1;10 onwards. In accordance with Mills’ (1986: 65) report, Jan has a slight preference for unmarked forms ending in –e (e.g. gross-eAuto ‘big car’), but uses and also overgeneralizes other endings as well (except dative –m in articleless constructions like mit warm-em Wasser ‘with warm water’ which is also very rare in the input). The emergence of Jan’s correctly marked forms of the German case paradigm is illustrated in table 9a and b. The genitive has not been considered because periphrastically expressed genitives do not occur in the input or output. Homophonous nominatives and accusatives are not distinguished because their syntactic function is not always clear. Thus, noch eine Suppe may either mean ‘(I want) another soup’ or ‘(there is) another soup (left)’).

286

Katharina Korecky-Kr¨oll and Wolfgang U. Dressler

Table 9a. Emergence of Jan’s case paradigm: Indefinite and definite articles + N INDEF.ART+N NOM MASC SG 1;11: ein Apfel ‘an apple’ NEUT SG 1;11: ein Auto ‘a car’ FEM SG 1;10: noch eine Suppe ‘another soup’ PL 1;4: Kekse

DAT – – – 2;3: mit seiner Tasche ‘with his bag’

ACC=NOM

ACC=NOM

DEF.ART+N MASC SG

NOM 1;10: der Mist ‘the garbage’

2;0: [mit] Bausteinen spielen ‘play [with] building bricks’ DAT 2;2: dem Kasperl ‘(to) the Punch’

NEUT SG

1;11: das M¨ullauto ‘the garbage car’ 2;0: die Laalaa ‘the Laalaa’ 2;2: die Patschen ‘the slippers’

2;2 mit dem Schiff ‘with the ship’ 2;1: auf der Straße ‘on the road’ 2;5: in den Bergen ‘in the mountains’

FEM SG PL

ACC 2;1: einen Fernseher ‘a TV’ ACC=NOM

ACC 2;0: in den Topf ‘into the pot’ 2;2 den Mann ‘the man’ ACC=NOM ACC=NOM ACC=NOM

The most important combinations of determiner and/or adjective and noun have emerged by the age of 2;7: 3/4 of all slots (43 out of 57) are filled. Two of the empty slots are not filled with the exact constructions but with similar ones (marked in italics) which nevertheless show that Jan uses determiner inflections correctly: mit sein-er Tasche ‘with his-FEM:DAT bag’ (possessive adjectives inflect the same way as indefinite articles, e.g. ein-er ‘one-FEM:DAT’ vs. sein-er ‘his-FEM:DAT), dies-en gr¨un-en [Elefanten] ‘this-MASC:ACC greenMASC:ACC [elephant]’ (demonstrative adjectives inflect the same way as definite articles, e.g. d-en ‘the-ACC:MASC’ vs. dies-en ‘this-ACC:MASC’). As shown in table 10, omission of obligatory articles (column “Omission errors”) decreases considerably especially between 2;0 and 2;2. Commission errors (i.e. articles, amalgams or adjectives which are marked for gender, number and/or case but which do not agree with the respective nouns; examples in § 6.2.3, § 7.3, § 7.4) are rare at the beginning of development and most frequent between 2;3 and 2;6. From 2;5 onwards, they are more numerous than omission errors. At 2;7, Jan seems to have acquired agreement rather well (90.43% of correct

The acquisition of number and case in Austrian German nouns

287

Table 9b. Emergence of Jan’s case paradigm: Adjectives + N ADJ+N11 MASC SG NEUT SG FEM SG DEF.ART+ADJ+N MASC SG

NEUT SG

FEM SG

PL

NOM 1;9: hoher Turm ‘high tower’ 1;11: kleines Haus ‘small house’ 1;10: alte Zeitung ‘old newspaper’ NOM 2;1: der zweite Patschen ‘the second slipper’

DAT –

2;3: das schwere Buch ‘the heavy book’ 2;2: die große Straße ‘the big street’

2;7: mit dem roten [Maxerl] ‘with the red [piece]’ 2;4: aus der großen ACC=NOM Flasche ‘from the big bottle’

2;6: die anderen Autos ‘the other cars’ INDEF.ART+ADJ+ N NOM MASC SG 2;2: ein großer Mann ‘a tall man’ NEUT SG 2;1: ein sportliches Auto ‘a sporty car’ FEM SG 2;2: eine kleine Katze ‘a small cat’ PL 1;10: (nor)male Autos ‘normal cars’

–

ACC 1;11: anderen Topf ‘other pot’ ACC=NOM

–

ACC=NOM

DAT –

ACC – 2;7: diesen gr¨unen [Elefanten] ‘this green [elephant]’ ACC=NOM

–

ACC=NOM

DAT –

ACC –

–

ACC=NOM

–

ACC=NOM

–

ACC=NOM

agreement and correct non-use of determiners fulfils also Cazden’s (1968: 435) criterion of acquisition, i.e. the use of a given grammatical form in 90% of obligatory contexts, but a turning point appears to be already at 2;2, when correct agreement (60.94%) exceeds omission errors (15.36%).12

288

Katharina Korecky-Kr¨oll and Wolfgang U. Dressler

Table 10. Correct and incorrect determiner/adjective – noun agreement (absolute numbers and % of noun tokens) Age Correct agreement 1;9 2113 1;10 6 1;11 45 2;0 109 2;1 108 2;2 234 2;3 146 2;4 162 2;5 159 2;6 139 2;7 83

Correct non-use of determiners

Omission errors

Commission Unclear errors examples

3.66% 104 18.15% 428 74.69% 14 2.44% 1.10% 84 15.44% 434 79.78% 4 0.74% 6.58% 107 15.64% 481 70.32% 30 4.39% 9.48% 167 14.52% 788 68.52% 42 3.65% 25.65% 86 20.43% 177 42.04% 28 6.65% 60.94% 55 14.32% 59 15.36% 25 6.51% 53.28% 48 17.52% 38 13.87% 25 9.12% 66.67% 33 13.58% 22 9.05% 19 7.82% 60.46% 54 20.53% 19 7.22% 23 8.75% 58.90% 53 22.46% 9 3.81% 25 10.59% 72.17% 21 18.26% 2 1.74% 5 4.35%

6 16 21 44 23 11 17 7 8 10 4

Noun tokens

1.05% 573 2.94% 544 3.07% 684 3.83% 1150 5.23% 421 2.86% 384 6.20% 274 2.88% 243 3.04% 263 4.24% 236 3.48% 115

Commission errors mostly concern definite articles (94 tokens), followed by adjectives and indefinite articles (42 and 38 tokens, respectively). Amalgams of verb and pronoun or article (e.g. daisie M¨ullauto ‘there-is-she garbage car’) are frequent until 2;0 (also reported by Mills 1986: 63) but disappear from 2;1 onwards. Fillers used instead of determiners are relatively rare (17 tokens from 1;9 to 2;5), but most frequent at 2;0 (7 tokens), when Jan starts to realize that some element should be in the determiner position but when he has not yet grasped the distribution of determiners. A closer look at Jan’s commission errors (240 tokens) makes it clear that almost half of the errors cannot be clearly determined because of homophony inherent in the forms of the target language. Those gender and case errors which can be distinguished are listed in table 11. At most points in time, there are many more gender errors (which cannot be treated here in detail) than case errors.

Table 11. Gender and case errors Age Gender Case Total

1;9 2

1;10 1

2

1

1;11 16 2 18

2;0 29 29

2;1 11 1 12

2;2 7 4 11

2;3 13 3 16

2;4 11 1 12

2;5 11 4 15

2;6 15 4 19

2;7 1 1 2

The acquisition of number and case in Austrian German nouns

7.3.

289

Case errors

The first clear case errors appear at 1;11 and concern nominatives and accusatives only, e.g.: (5)

a. *JAN: dr¨ucken der lautsprecher. press the-NOM:SG:MASC* loudspeaker. b. *JAN: da is(t) ein weiss-en there is a white-ACC:SG:MASC* hiegel [: h¨ugel]. hill.

In example 5a, the masculine definite article is used in his nominative form der ‘the’ instead of the accusative form den, and in example b, the determiner is correct but the adjective weiss ‘white’ has the accusative masculine ending –en. Nominatives replacing accusatives are reported frequently in the literature (Clahsen 1984: 9, Mills 1985: 178, Tracy 1986: 54-55), but not the other way round. The only two cases of accusative overgeneralization to nominatives are probably due to the fact that Jan (as an early talker and a “commission” child) realizes quite early the importance of case-marking (1;11), but has not yet grasped its conditions, whereas German children reported in the literature rarely distinguish nominative from accusative marking before age 3;0 (Mills 1985: 155). From 2;2 to 2;7, accusatives are still, but rarely replaced with nominatives (4 tokens), as at 2;3: (6)

*JAN: der ball (= stressed) hab(e) ich auch the-NOM:SG:MASC* ball have I also oros [: orange] gemacht. orange made ‘I have also made the ball orange.’

The last two accusative – nominative errors (2;6, 2;7) are elliptic constructions in answers to questions (e.g. at 2;6 der Dipsy ‘the-NOM:SG:MASC Dipsy’ in reply to the question wen? ‘whom-ACC’?). In both examples, two utterances are in between the question and Jan’s answer. These errors may therefore be rather due to working memory than to a missing nominative – accusative distinction, for which we do not find any other evidence at 2;6 and 2;7 because of a majority of correct examples. From 2;2 onwards, Jan uses accusatives in dative contexts. These are his most frequent case errors (13 of 20 tokens from 2;2 to 2;6). Although such optional

290

Katharina Korecky-Kr¨oll and Wolfgang U. Dressler

replacement is a well-known phenomenon in Austrian German dialects (e.g. dial. mit die Kind-er ‘with the:ACC child-PL’ for stand. mit den Kind-er-n ‘with the:ACC child-PL-DAT’), it is not characteristic of Jan’s input. There are several possible explanations for the occurrence of this phenomenon in Jan’s speech in spite of his input: 1. From the point of view of Natural Morphology (Dressler et al. 1987, cf. already Jakobson 1971), the dative is more marked than the accusative and therefore acquired later. Although inflectionally marked dative plurals emerge in the same month (1;11) as periphrastically marked accusatives, dative marking remains problematic: Still at age 2;5 (and even if there are many correct uses of datives by that time), Jan uses the trivalent verb geben ‘give’ in a sentence with two accusative objects instead of a dative and an accusative object: (7)

*JAN: die Heidi # gibt die [self-correction] the-NOM:SG:FEM Heidi gives the-ACC:SG:FEM:* den Peter die den-ACC:SG:MASC* Peter the-ACC:SG:FEM Snucki [: Schnucki], ihre ziege. Schnucki her goat. ‘Heidi gives Peter Schnucki, her goat.’

Furthermore, nominatives and accusatives are never replaced by datives in Jan’s data, which fits to their marked character (cf. also Clahsen et al. 1994 with nativist explanations; Mills 1985 reports a single substitution). 2. From a phonological point of view, we must consider that one third of Jan’s accusative – dative errors are replacements of the masculine or neuter dative definite article dem by the masculine accusative definite article den, which are phonologically similar (Mills 1985: 224). This could possibly be a phonological assimilation of place of articulation, supported by dialectal replacements (cf. § 3.2). 3. Den is not only a marker for accusatives but also for dative plurals and thus perceived by children in dative contexts. 4. Several case errors occurring from 2;2 to 2;4 (5 out of 12) show that Jan might not yet master the difference between goal-oriented (directional, expressed by a preposition with accusative) and nondirectional locatives (expressed by a preposition with dative): Besides several correct uses of nondirectional locatives (e.g.

The acquisition of number and case in Austrian German nouns

291

auf der Straße ‘in the-FEM:DAT street’), we find examples like ich war auch in die Donau ‘I also was in the-FEM:ACC* Danube’ (both at 2;2). Otherwise, datives are more frequent with prepositions (e.g. mit ‘with’, von ‘of, from’, bei ‘at’) than as indirect objects (example 8) and therefore used by Jan earlier in an appropriate way (mostly correct from 2;3 onwards), but not in the case of nondirectional locatives. Further potential case errors are more likely to be gender errors. Since there is no NOM/ACC distinction in the neuter gender, shifting a masculine noun to the neuter class will at the same time neutralize this case distinction, e.g. at 2;0: (8)

*JAN: moechte ein stift.14 want(s) a-NEUT:SG:NOM/ACC*ˆMASC:SG:NOM* pencil

8. Conclusion German is a typologically rather heterogeneous language (cf. Lang and Zifonun 1996, Wegener 1998) where plural and case are marked both on the noun and by periphrastic means. Plural is more distinctly marked on the noun, whereas case marking is mostly achieved by periphrastic means. A large number of syncretisms (zero plurals; case forms; see tables 1 and 2) make acquisition relatively difficult. Gender plays an important role in the choice of the correct noun plural and case forms but German gender attribution is not transparent either (cf. das M¨adchen ‘the-NEUT girl’). For these reasons, it will take children a long time to find out the correct patterns of determiner, adjective, and noun stem endings in the appropriate syntactic contexts. On the other hand, the iconic, transparent, and superstable possessive genitive of proper nouns is correctly used from the very beginning. In contrast to theAustrian girl Katharina (Sedlak and Dressler 1997, Stephany 2002: 16), Jan does not produce numerals or other quantifiers such as zwei ‘two’, alle ‘all’ as forerunners of plural markers. Rather, he starts plural marking with rote-learned plural-dominant forms (Kekse ‘cookies’, Z¨ahne ‘teeth’), followed by the productive use of forms of high input frequency (e.g. –en and –e plurals). In Jan’s acquisition of noun plurals, we did not find any trace of an –s plural default (cf. also Behrens 2001, 2002, Bittner and K¨opcke 2001, Schaner-Wolles 2001, Szagun 2001, Elsen 2002, Stephany 2002: 17, note 11). From the very beginning, the relatively high token frequency of -s plurals is exclusively due to the lemma Auto-s ‘car-s’. It is important to note that they do not serve as a default since there is only one overgeneralized –s ending (*Fenster-s ‘window-s’) in contrast to several overgeneralizations of –en or –e. Until the age of 2;2, choice

292

Katharina Korecky-Kr¨oll and Wolfgang U. Dressler

of plural endings appears to be due to input frequency only. From 2;2 onwards, when Jan starts to add suffixes to suffixless target pure umlaut plurals and zero plurals (*Vogeln instead of *V¨ogel ‘birds’, *Fenster-s instead of *Fenster ‘windows’), iconicity seems to play a certain role (cf. Seifert 1985). Potential plurals which can be accounted for by productive adult rules occur from 2;3 onwards (e.g. *H¨und-e instead of Hund-e ‘dog-s’). Illegal plurals stop to be produced at 2;3 (cf. Ravid et al. 2008 for analyses after 2;7). Unproductive –er plurals (first without opacifying umlaut) emerge later than these productive plural types. The latest plural class to appear is the least productive, least iconic and most opaque pure umlaut plural. Our results fit well into the framework of Natural Morphology, where productivity, transparency and iconicity are crucial. On the one hand, our findings are hardly compatible with a dual-route model which posits -s plurals as the only regular or as the default plural. Nor are there any other indicators for a strictly complementary subdivision between a rule component and a storage component. What we found is that all productive plural rules of the adult target language may lead to overgeneralizations. The lack of overgeneralizations of the unproductive –er and pure umlaut plurals may simply be due to scarcity of data. On the other hand, a radical single-route model which postulates a clear dependency on input frequency seems to be fully predictive only for the beginning of the process of morphological acquisition. Thus our findings are compatible with usage-based models of acquisition (cf. Bybee 2001; Tomasello 2003), provided that they allow for a relatively early extraction of rule-like patterns. Later on, as also evidenced by experimental data from older children acquiring Austrian German (Klampfer, Korecky-Kr¨oll, and Dressler 2001; Laaha et al. 2006), degrees of productivity seem to play a role in the course of acquisition (in the time-course of emergence as well as in overgeneralization errors). The same holds for two other preference parameters of Natural Morphology, iconicity and transparency. Both disfavor the late emerging pure umlaut plurals as well as umlaut plurals in general. In contrast to expectations derivable from claims by e.g. Marcus et al. (1995) and Bartke (1998) (cf. § 2.2), who assume that the –s plural is the regular default plural and that therefore, there should be only –s overgeneralizations or analogies to rhyming plurals, Jan’s overgeneralizations (cf. §6.3) cannot be explained by rhyming plurals of the input (or frequent ones of the target system): the single error H¨und-e instead of Hund-e (21 tokens in the input) can hardly be explained by the single token of Abgr¨und-e ‘abysses’. Furthermore, there are no plausible rhymes to e.g. *Gespenst-e ‘ghosts’ and *Tisch-en ‘tables’, whereas the input includes the plurals Fenster-0 ‘windows’ and Fisch-e ‘fish’, which rhyme with the correct input plurals Gespenst-er and Tisch-e.

The acquisition of number and case in Austrian German nouns

293

As far as the development of case inflection is concerned, the iconic and transparent genitive singular ending in –s (occurring in both article and noun inflection in the standard language) is by far the first to be acquired (cf. §7), albeit only with proper nouns (including hypocoristics; cf. also Mills 1985: 185) where it is superstable and fully productive, whereas the much less productive –s genitive occurring with common nouns does not emerge at all in the observational period (it does not, however, occur in the input either). At 1;11, only nominally marked datives emerge (B¨aum-e-n), only periphrastically marked ones follow at 2;1, and only from 2;5 onwards, the dative may be correctly marked both synthetically and periphrastcally. Periphrastic case marking starts at 1;9 with combinations of determiner or adjective and noun in the nominative/accusative such as ein Socken [MASC] ‘a sock’, eine Suppe [FEM] ‘a soup’ or hoher Turm [MASC] ‘high tower’. Jan has only a slight preference for unmarked –e forms of attributive adjectives, but also overgeneralizes marked forms. From 2;2 onwards (when correct and incorrect agreement between nouns and determiners overtakes omission errors), Jan becomes more and more a “commission” rather than an “omission” child; e.g. he uses complex constructions (e.g. determiner–adjective–noun combinations), plurals, case endings but there are many gender and case errors which cannot always be distinguished from each other due to syncretism. In those cases where they can be differentiated (see table 11), the relatively high number of gender errors shows that case distinctions seem to be acquired more easily and earlier than gender distinctions (see also Bittner 2006). This period of periphrastic commission errors is followed by the stabilisation of the article system (including gender and case) at 2;7. Thus, 2;7 is a candidate for assigning the onset of morphology proper. However, there is an earlier turning point at 2;2, when both the determiner system ‘explodes’(see table 10) and commission errors in plural formation become rare. At 2;2, Jan produces the last plural (*V¨ogel-n ‘birds’) which is not a potential plural form within the German system. By 2;3, all plural markers have emerged (see table 7). The difference in speed of development between number and case might be interpreted as another instance of the acquisitional sequence ‘number before case’ (cf. Voeikova 2002) and, at least in German, also be explained by the predominantely continuous, synthetic marking of plurals in contrast to the less continuous, essentially periphrastic marking of case.

294

Katharina Korecky-Kr¨oll and Wolfgang U. Dressler

Notes 1. For a detailed approach to German noun inflection within the framework of Natural Morphology see Bittner (1991/2003). For Natural Morphology in general see Dressler (2000, 2004), Dressler et al. (1987), Kilani-Schoch and Dressler (2005). 2. Sometimes, -er plurals and -er+umlaut plurals are treated as one class because umlaut is obligatory with umlautable stem vowels. For details on class 7 and 8 see Wegener (2003). 3. We thank the Austrian Academy of Sciences and the FWF (Fonds zur F¨orderung der wissenschaftlichen Forschung) for supporting this study, which was part of the project P17276-G03. We also thank Sabine Laaha for helpful comments and for the morphological coding of Jan’s data. 4. Numbers of noun lemmas, noun types and noun tokens include incorrect and ambiguous forms, but exclude imitations and citations. Number of utterances, MLU and standard deviation are calculated from all utterances (including imitations and citations). Compounds are treated as separate lemmas (Auto ‘car’ and M¨ullauto ‘garbage car’ are two different noun types). 5. Plural types and tokens in this table include incorrect and ambiguous plural forms if they can be recognized as clear plurals. Thus, numbers of plural types and tokens differ from numbers in the following tables where ambigous and incorrect forms have been excluded. 6. The input was taken from the same period as Jan’s data (1;3 to 2;7). 7. We thank Gregor Laaha (BOKU Vienna, Department of Applied Statistics) for doing this analysis for us. 8. with one exception: plural dominant Ei-er ‘eggs’ is used 5 times by Jan and 4 times by the mother. 9. For details on the acquisition of verbs by Jan cf. Klampfer 2003, Laaha 2004. 10. Other determiners, mostly quantifiers such as viele ‘many’, appear from 1;11 onwards but these never function as substitutes for noun plurals. 11. Noun phrases comprising an adjective may be used without an article only in certain contexts such as forms of address. Although Jan did not always use articleless constructions correctly, correct adjective – noun agreement is nevertheless a further sign of the acquisition of periphrastic case marking. 12. An error was counted as an omission error if there was no obligatory periphrastic gender or case marking at all. If the article was missing, but if the adjective was marked for gender, it was either counted as correct or as an incorrect commission error. 13. These are 21 tokens of the clearly rote-learned form hoher Turm ‘high tower’. 14. It remains mostly unclear if the n of ein was prounounced as a syllabic nasal and therefore an allegro form of einen because there is no a phonological coding. This is another reason why these unclear cases were excluded from analysis as case errors.

The acquisition of number and case in Austrian German nouns

295

References Baayen, R. Harald, and Robert Schreuder 1991 War and peace: Morphemes and full forms in a non-interactive activation parallel route model. Brain and Language 68: 27–32. Bartke, Susanne 1998 Experimentelle Studien zur Flexion und Wortbildung. T¨ubingen: Niemeyer. Bartke, Susanne, Gary Marcus, and Harald Clahsen 1995 Acquiring German noun plurals. In: Proceedings of the 19th Annual Boston University Conference on Language Development, Dawn MacLaughlin and Susan McEwen (eds.), 60–69. Boston: Cascadilla Press. Bauer, Laurie 2001 Morphological Productivity. Cambridge: Cambridge University Press. Behrens, Heike 2001 How to learn a minority default: The acquisition of the German -s plural. Manuscript, Leipzig: Max-Planck-Institute for Evolutionary Anthropology. 2002 Learning multiple regularities: Evidence from overgeneralization errors in the German plural. In Proceedings of the 26th Annual Boston University Conference on Language Development, Barbora Skarabela, Sarah Fish, and Anna H.-J. Do (eds.), 72–83. Somerville: Cascadilla Press. Bittner, Dagmar 2003 Von starken Feminina und schwachen Maskulina. Die neuhochdeutsche Substantivflexion – Eine Systemanalyse im Rahmen der nat¨urlichen Morphologie. (ZAS Papers in Linguistics 31.) Berlin: Zentrum f¨urAllgemeine Sprachwissenschaft. Original edition, Ph. D. diss. Jena, 1991. 2006 Case before gender in the acquisition of German. Folia Linguistica 40: 115–134. Bittner, Dagmar, Wolfgang U. Dressler, and Marianne Kilani-Schoch (eds.) 2003 Development of Verb Inflection in First Language Acquisition: A Cross-Linguistic Perspective. Berlin: Mouton de Gruyter. Bittner, Dagmar, and Klaus-Michael K¨opcke 2001 On the acquisition of the German plural markings. In Papers for Ewald Lang, Ewald Lang, Marzena Rochon, Kerstin Schwabe, and Oliver Teuber (eds.), 21–32. (ZAS Papers in Linguistics 21.) Berlin: Zentrum f¨ur Allgemeine Sprachwissenschaft.

296

Katharina Korecky-Kr¨oll and Wolfgang U. Dressler

Bybee, Joan 2001

Phonology and Language Use. Cambridge: Cambridge University Press. Cazden, Courtney B. 1968 The acquisition of noun and verb inflections. Child Development 39: 433-448. Clahsen, Harald 1984 Der Erwerb von Kasusmarkierungen in der deutschen Kindersprache. Linguistische Berichte 89: 1–31. 1999 Lexical entries and rules of language: A multidisciplinary study of German inflection. Behavioral and Brain Sciences 22: 991–1060. 2006 Linguistic perspectives on morphological processing. In Advances in the Theory of the Lexicon, Dieter Wunderlich (ed.), 355–388. Berlin: Mouton de Gruyter. Clahsen, Harald, Sonja Eisenbeiss, and Anne Vainikka 1994 The seeds of the structure: A syntactic analysis of case marking. In Language Acquisition in Generative Grammar, Teun Hoekstra and Bonnie T. Schwartz (eds.), 85–118. Amsterdam: Benjamins. Clahsen, Harald, Gary Marcus, Susanne Bartke, and Richard Wiese 1996 Compounding and inflection in German child language. Yearbook of Morphology 1995, 115–42. Clahsen, Harald, Monika Rothweiler, Andreas Woest, and Gary F. Marcus 1992 Regular and irregular inflection in the acquisition of German noun plurals. Cognition 45: 225–255. Clark, Eve 1998 Morphology in language acquisition. In The Handbook of Morphology, Andrew Spencer and Arnold M. Zwicky (eds.), 374–389. Oxford: Blackwell. Daugherty, Kim G., and Mark S. Seidenberg 1994 Beyond rules and exceptions: A connectionist approach to inflectional morphology. In The Reality of Linguistic Rules, Susan D. Lima, Roberta L. Corrigan, and Gregory K. Iverson (eds.), 353–388. Amsterdam: Benjamins. Dressler, Wolfgang U. 1999 Why collapse morphological concepts? Behavioral and Brain Sciences 22: 1021. 2000 Naturalness. In: Morphologie/Morphology I, Geert Booij, Christian Lehmann, and Joachim Mugdan (eds.), 288–296. Berlin: Mouton de Gruyter. 2004 Degrees of grammatical productivity in inflectional morphology. Italian Journal of Linguistics 15: 31–62.

The acquisition of number and case in Austrian German nouns

297

Dressler, Wolfgang U. (ed.) ¨ 1997 Studies in Pre- and Protomorphology. Wien: Verlag der Osterreichischen Akademie der Wissenschaften. Dressler, Wolfgang U., Robert and Danuta Dra˛z˙ yk, Katarzyna Dziubalska-Kołaczyk, and Ewelina Jagła 1996 On the earliest stages of acquisition of Polish declension. In Proceedings of the Groningen Assembly on Language Acquisition, Charlotte Koster and Frank Wijnen (eds.), 185–195. Groningen: Centre for Language and Cognition. Dressler, Wolfgang U., and Annemarie Karpf 1995 The theoretical relevance of pre- and protomorphology in language acquisition. Yearbook of Morphology 1994: 99–122. Dressler, Wolfgang U., Marianne Kilani-Schoch, and Sabine Klampfer 2004 How does a child detect morphology? Evidence from production. In Morphological Structure in Language Processing, R. Harald Baayen and Robert Schreuder (eds.), 391–425. Berlin: Mouton de Gruyter. Dressler, Wolfgang U., and M´aria Lad´anyi 2000 Productivity in word formation (WF):A morphological approach.Acta Linguistica Hungarica 47: 103–144. Dressler, Wolfgang U., Willi Mayerthaler, Oswald Panagl, and Wolfgang U. Wurzel 1987 Leitmotifs in Natural Morphology. Amsterdam: Benjamins. Eisenbeiss, Sonja, Susanne Bartke, and Harald Clahsen 2006 Structural and lexical case in child German: Evidence from languageimpaired and typically developing children. Language Acquisition 13: 3-32. Elsen, Hilke 2002 The acquisition of German plurals. In Morphology 2000. Selected Papers from the 9th Morphology Meeting, Vienna, 24–28 February 2000, Sabrina Bendjaballah, Wolfgang U. Dressler, Oskar Pfeiffer, and Maria D. Voeikova (eds.), 117–127. Amsterdam: Benjamins. Eroms, Hans-Werner 1989 Regionalsprachliche Artikelparadigmen und die grammatikalische Behandlung der Artikel im Deutschen. In Probleme regionaler Sprachen, Hans-Werner Eroms (ed.), 103–123. Hamburg: Buske. Ewers, Heidi 1999 Schemata im mentalen Lexikon: Empirische Untersuchungen zum Erwerb der deutschen Pluralbildung. In Das Lexikon im Spracherwerb, J¨org Meibauer and Monika Rothweiler (eds.), 106–127. T¨ubingen: Francke. Frauenfelder, Ulrich H., and Robert Schreuder 1992 Constraining psycholinguistic models of morphological processing and representation: The role of productivity. Yearbook of Morphology 1991: 165–83.

298

Katharina Korecky-Kr¨oll and Wolfgang U. Dressler

Jakobson, Roman 1971 Beitrag zur allgemeinen Kasuslehre: Gesamtbedeutungen der russischen Kasus. In Selected Writings II, 23–71.The Hague: Mouton. Kilani-Schoch, Marianne, and Wolfgang U. Dressler 2002 Emergence of inflectional paradigms in two French corpora: An illustration of general problems of pre- and protomorphology. In Pre- and Protomorphology: Early Phases of Morphological Development in Nouns and Verbs, Maria D. Voeikova and Wolfgang U. Dressler (eds.), 45–59. M¨unchen: Lincom Europa. 2005 Morphologie naturelle et flexion du verbe fran¸cais. T¨ubingen: Narr. Klampfer, Sabine 2003 Emergence of verb paradigms in one Austrian child. In Development of Verb Inflection in First Language Acquisition: A Cross-Linguistic Perspective, Dagmar Bittner, Wolfgang U. Dressler, and Marianne Kilani-Schoch (eds.), 297–321. Berlin: Mouton de Gruyter. Klampfer, Sabine, and Katharina Korecky-Kr¨oll 2002 Nouns and verbs at the transition from pre- to protomorphology: A longitudinal case study on Austrian German. In Pre- and Protomorphology: Early Phases of Morphological Development in Nouns and Verbs, Maria D. Voeikova and Wolfgang U. Dressler (eds.), 61–73. M¨unchen: Lincom Europa. Klampfer, Sabine, Katharina Korecky-Kr¨oll, and Wolfgang U. Dressler 2001 Morphological potentiality in children’s overgeneralization patterns: Evidence from Austrian German noun plurals. Wiener Linguistische Gazette 67-69: 25–43. K¨opcke, Klaus-Michael 1993 Schemata bei der Pluralbildung im Deutschen: Versuch einer kognitiven Morphologie. T¨ubingen: Narr. 1998 The acquisition of plural marking in English and German revisited: Schemata versus rules. Journal of Child Language 25: 293–319. Laaha, Sabine (n´ee Klampfer) 2004 D´eveloppement pr´ecoce de la morphologie verbale: Une e´ tude comparative sur l’acquisition de l’allemand autrichien et du fran¸cais. Ph. D. diss. Universit¨at Wien/Universit´e Paris 5. Laaha, Sabine, Dorit Ravid, Katharina Korecky-Kr¨oll, Gregor Laaha, and Wolfgang U. Dressler 2006 Early noun plurals in German: Regularity, productivity or default? Journal of Child Language 33: 271–302. Lang, Ewald, and Gisela Zifonun (eds.) 1996 Deutsch - typologisch. Berlin: Walter de Gruyter.

The acquisition of number and case in Austrian German nouns

299

MacWhinney, Brian 1978 The Acquisition of Morphophonology. (Monographs of the Society for Research in Child Development 43,1.) Chicago: University of Chicago Press. 1996 Language specific prediction in foreign language learning. Are some languages harder to learn than others and why? Language Testing 12: 292–320. 2000 The CHILDES Project: Tools for Analyzing Talk. Vol. I: Transcription, Format and Programs. 3d ed. Mahwah, NJ: Lawrence Erlbaum. Marcus, Gary F. 2000 Children’s overregularization and its implications for cognition. In Models of Language Acquisition: Inductive and Deductive Approaches, Peter Broeder and Jaap Murre (eds.), 154–176. Oxford: Oxford University Press. Marcus, Gary F., Ursula Brinkmann, Harald Clahsen, Richard Wiese, and Steven Pinker 1995 German inflection: The exception that proves the rule. Cognitive Psychology 29: 189–256. Mayerthaler, Willi 1981 Morphologische Nat¨urlichkeit. Wiesbaden: Athenaion. Meisel, J¨urgen 1986 Word order and case marking in early child language. Evidence from simultaneous acquisition of two first languages: French and German. Linguistics 24: 123–183. Mills, Ann E. 1985 The acquisition of German. In The Crosslinguistic Study of Language Acquisition: The Data, Dan I. Slobin (ed.), Vol. 1: 141–254. Hillsdale, NJ: Lawrence Erlbaum. 1986 The Acquisition of Gender. A Study of English and German. Berlin: Springer. Mugdan, Joachim 1977 Flexionsmorphologie und Psycholinguistik. T¨ubingen: Narr. Niedeggen-Bartke, Susanne 1999 Flexion und Wortbildung im Spracherwerb. In Das Lexikon im Spracherwerb, J¨org Meibauer and Monika Rothweiler (eds.), 208–228. T¨ubingen: Francke. Pinker, Steven 1999 Words and Rules: The Ingredients of Language. London: Weidenfeld & Nicolson. Plunkett, Kim, and Virginia Marchman 1991 U-shaped learning and frequency effects in a multilayered perceptron: Implications for child language acquisition. Cognition 38: 1–60. 1993 From rote learning to system building: Acquiring verb morphology in children and connectionist nets. Cognition 48: 21–69.

300

Katharina Korecky-Kr¨oll and Wolfgang U. Dressler

Ravid, Dorit, Wolfgang U. Dressler, Bracha Nir-Sagiv, Katharina Korecky-Kr¨oll, Agnita Souman, Katja Rehfeldt, Sabine Laaha, Johannes Bertl, Hans Basbøll, and Steven Gillis 2008 Core morphology in child directed speech: Crosslinguistic corpus analyses of noun plurals. In Corpora in Language Acquisition Research: Finding Structure in Data, Heike Behrens (ed.), 25–60 (Trends in Language Acquisition Research 6.) Amsterdam: Benjamins. Rumelhart, David E., and James L. McClelland 1986 On learning the past tense of English verbs. In Parallel Distributed Processing: Explorations in the Microstructure of Cognition. James L. McClelland, David E. Rumelhart, and the PDP Research Group (eds.), vol. 2, 216–271. Cambridge: MIT Press. Schaner-Wolles, Chris 2001 On the acquisition of noun plurals in German. In Naturally! Linguistic Studies in Honour of Wolfgang Ulrich Dressler Presented on the Occasion of his 60th Birthday, Chris Schaner-Wolles, John Rennison, and Friedrich Neubarth (eds), 451–460. Torino: Rosenberg & Sellier. Sedlak, Maria, and Wolfgang U. Dressler 1997 The acquisition of number in German. Unpublished manuscript, Institut f¨ur Sprachwissenschaft, Universit¨at Wien Sedlak, Maria, Sabine Klampfer, Brigitta M¨uller, and Wolfgang U. Dressler 1998 The acquisition of number in Austrian German: A case study on the early stages. In Studies in the Acquisition of Number and Diminutive Marking, Steven Gillis (ed.), 51–76 (Antwerp Papers in Linguistics 95.) Antwerp: Universiteit Antwerpen. Seifert, Katharina 1985 Ikonizit¨at von Pluralformen. Ph. D. diss. Universit¨at Wien. Slobin, Dan I. 1973 Cognitive prerequisites for the development of grammar. In Studies of Child Language Development, Charles A. Ferguson and Dan I. Slobin (eds.), 175–208. New York: Holt, Rinehart and Winston. Slobin, Dan I. (ed.) 1985 The Crosslinguistic Study of Language Acquisition. Vols. 1–2. Hillsdale, NJ: Lawrence Erlbaum. Sonnenstuhl-Henning, Ingrid 2003 Deutsche Plurale im mentalen Lexikon: Experimentelle Untersuchungen zum Verh¨altnis von Speicherung und Dekomposition. T¨ubingen: Niemeyer. Stenzel, Achim 1994 Case assignment and functional categories in bilingual children: Routes of development and implications for linguistic theory. In Bilingual First Language Acquisition: French and German Grammatical Development, J¨urgen M. Meisel (ed.), 161–208. Amsterdam: Benjamins.

The acquisition of number and case in Austrian German nouns

301

Stephany, Ursula 2002 Early development of grammatical number – a typological perspective. In Pre- and Protomorphology: Early Phases of Morphological Development in Nouns and Verbs, Maria D. Voeikova and Wolfgang U. Dressler (eds.), 7–23. M¨unchen: Lincom Europa. Szagun, Gisela 2001 Learning different regularities: The acquisition of noun plurals by German-speaking children. First Language 21: 109–141. 2004 Learning by ear: On the acquisition of case and gender marking by German-speaking children with normal hearing and with cochlear implants. Journal of Child Language 31: 1–30. Tomasello, Michael 2003 Constructing a Language: A Usage-Based Theory of Language Acquisition. Cambridge, MA/London: Harvard University Press. Tonelli, Livia, Anna de Marco, Ralf Vollmann, and Wolfgang U. Dressler 1998 Le prime fasi dell’acquisizione della morfologia: Un confronto tra l’italiano et il tedesco. In Parallela 6: Italiano e tedesco in contatto e a confronto/Italienisch und Deutsch im Kontakt und im Vergleich, Patrizia Cordin, Maria Iliescu, and Heidi Siller-Runggaldier (eds.), 281–301. Trento: Dipartimento di Scienze Filologiche e Storiche. Tracy, Rosemarie ¨ 1984 Fallstudien: Uberlegungen zum Erwerb von Kasuskategorie und Kasusmarkierung. In Syntaktische Struktur und Kasusrelationen, Hartmut Czepluch and Hero Janßen (eds.), 271–313. T¨ubingen: Narr. 1986 The acquisition of case morphology in German. Linguistics 24: 47–87. Voeikova, Maria D. 2002 The acquisition of case in typologically different languages. In Preand Protomorphology: Early Phases of Morphological Development in Nouns and Verbs, Maria D. Voeikova and Wolfgang U. Dressler (eds.), 25–44. M¨unchen: Lincom Europa. Vollmann, Ralf, Maria Sedlak, Brigitta M¨uller, and Maria Vassilakou 1997 Early verb inflection and noun plural formation in four Austrian children. Papers and Studies in Contrastive Linguistics 33: 59–78. Wegener, Heide 1998 Deutsch kontrastiv. Typologisch-vergleichende Untersuchungen zur Grammatik des Deutschen. (Language Development 24.) T¨ubingen: Narr. 1999 Die Pluralbildung im Deutschen – ein Versuch im Rahmen der Optimalit¨atstheorie. Linguistik online 4, 3/1999. http://www.euv-frankfurt-o.de/∼wjournal/3 99/wegener.html. 2002 Aufbau von markierten Pluralklassen im Deutschen – eine Herausforderung f¨ur die Markiertheitstheorie. Folia Linguistica 36: 261–295.

302

Katharina Korecky-Kr¨oll and Wolfgang U. Dressler 2004

Pizzas und Pizzen – die Pluralformen (un)assimilierter Fremdw¨orter im Deutschen. Zeitschrift f¨ur Sprachwissenschaft 23: 47–112. Wegener, Heide (ed.) 1998 Deutsch kontrastiv: Typologisch vergleichende Untersuchungen zur deutschen Grammatik. T¨ubingen: Stauffenburg. Wittek, Angelika, and Michael Tomasello 2005 German-speaking children’s productivity with syntactic constructions and case morphology: Local cues act locally. First Language 25: 103– 125. Wurzel, Wolfgang U. 1984 Flexionsmorphologie und Nat¨urlichkeit: Ein Beitrag zur morphologischenTheoriebildung, (Studia grammatica XXI.) Berlin:AkademieVerlag. 1994 Gibt es im Deutschen noch eine einheitliche Substantivflexion? Oder: Wie ist die deutsche Substantivflexion m¨oglichst angemessen zu erfassen? In Funktionale Untersuchungen zur deutschen Nominal- und Verbalmorphologie, Klaus-Michael K¨opcke (ed.), 29–44. (Linguistische Arbeiten 319.) T¨ubingen: Niemeyer.

The emergence of nominal number in Italian∗ Sabrina Noccetti 1. Introduction The concept of number is a fascinating field of research both from a cognitive and a linguistic perspective. According to Wiese (2003), language enables us to cross the boundaries of early pre-numerical concepts already existing before the faculty of language. A systematic concept of number, which includes the ability to count and perform arithmetic operations, develops from a primitive cognitive skill which allows to grasp small sets of one, two or three items. This primitive numerical cognitive system permits to discriminate different sets of one, two or three items without counting (‘subitising’), to estimate the size of larger sets (‘noisy magnitude’) and to grasp the concept of sequential order. It also implies the cognitive ability to understand the notion of classification of items as belonging to a single class (Wiese 2003: 103–104). The infant, then, has the ability to assess different numerical properties of objects: quantitative (i.e., cardinality of sets), ordinal (i.e., the rank of the objects within a series) and nominal aspects (i.e., identity of objects within a set). According to Wiese, these early skills, which have a different mental representation from counting, are part of our neurological heritage and are central to the survival of our species. Indeed, children usually know that numbers are used for counting before they learn how to count (Stephany 1998: 3) and develop a concept of number before they can speak. It is therefore interesting to examine how children learn to translate the concept of plurality into grammatical categories and investigate the functionality of emergent grammars. The acquisition of number marking in Italian noun morphology has not been given much attention until recently and only concise outlines can be found in Cipriani et al. (1993), and Pizzuto and Caselli (1993). Both studies let infer that the emergence of number marking in child speech precedes its acquisition of several months. Actually, Pizzuto and Caselli (1993) observe that many grammatical morphemes appear at an early age but that the period of mastery is quite protracted especially for verb inflection, articles and clitics. Supporting usage-based theories of language acquisition, they maintain that the factors of input frequency, regularity of the morphological forms as well as phonological salience are responsible for the emergence and acquisition of morphemes.

304

Sabrina Noccetti

Cipriani et al. (1993) compare the morphosyntactic development of six Italian children with a language-impaired child.1 They identify four different periods in syntactic development on the basis of length of utterance (MLU), syntactic complexity and use of articles and prepositions. Since their examination of nominal morphology is only secondary and related to the expansion of syntax, there is neither a description of the different steps in the acquisition of the grammatical means to express nominal number nor an explicit account on how the development of plural markers occurs. However, they observe that the children occasionally realize gender and number agreement in nouns and modifiers only at the end of the period they call pre-syntactic (from about 1;7 to 2;2). Actually, since early child speech is characterized by utterances mainly consisting of a single element, agreement is not realized within the child’s utterances, but rather there are expressions of plurality and gender in the child’s reactions to the adult’s questions. In the successive period (from about 2;2 to 2;5), the growing number of two-element sentences permits to observe the realization of noun-adjective agreement. Cipriani et al. (1993: 52) broadly notice that there is an expansion of noun morphology and report some examples of class shift errors (usually from -e/-i inflectional noun classes to -o/-i) which would deserve further investigation to understand the limits of noun morphology expansion in the children analyzed. The third period (from about 2;5 to 2;9) and the fourth period (from 2;9 to 2;11) show number and gender agreement in nouns and adjectives, and rare errors between articles and nouns, which seem to involve irregular noun morphology. Conversely, a great deal of interest in the acquisition of nominal number marking has been shown by the project on pre- and protomorphology in child language. This is a developmental model based on the interactionist-constructivist approach, as it has been developed by Karpf (1991) and Dressler and Karpf (1995), where empirical data are analyzed and interpreted within the framework of Natural Morphology (Dressler 1985; 1995; Dressler and Karpf 1995; Gillis 1998; Voeikova and Dressler 2002). As far as the concept of number is concerned, one of the aims of the project of pre- and protomorphology is to detect the early linguistic expressions of plurality in children of typologically and genetically different native languages and how these expressions develop towards their target languages. The first results for fusional-inflectional languages, including languages with a complex morphophonemics, such as Lithuanian, Russian and Greek, reported in Stephany (2002), show that plural nouns emerge first as rote-learned forms and are initially used both for singular and plural reference. Instead, from the very beginning, plurality is expressed by means of lexical items, which precede synthetic number marking of nouns and are opposed to singular reference. In French (KilaniSchoch 1998: 45), in German (Sedlak et al. 1998: 66) and in Italian (Noccetti

The emergence of nominal number in Italian

305

2002: 65–66), for instance, the children first use numerals, e.g. deux ‘two’, zwei ‘two’, sei ‘six’, and quantifiers alle ‘all’, tanti ‘many’ to indicate a generic plural meaning. Then, there is a slow development of the grammatical means to express plurality, but evidence for morphological processing of numbers, generally revealed by overgeneralization errors and self-corrections, appears only later, some time after their emergence in child speech (Kilani-Schoch 1998; Kilani-Schoch and Dressler 2002; Sedlak et al. 1998). Tonelli et al. (1998), analyzing Italian longitudinal data in the framework of pre- and protomorphology (Dressler 1995; Dressler and Karpf 1995; Voeikova and Dressler 2002), observe that in premorphology singular and plural forms occur with different lemmas. Later, in protomorphology, both the singular and the plural form of nouns belonging to the regular and productive morphological classes emerge as the first step in the development of the morphological module. The unproductive classes are rare and usually occur in one form, either singular or plural. The results of Tonelli et al. (1998) are similar to those of Calleri et al. (2003), who found that the semantic concept of number is at first lexically expressed, but that the means to express number morphologically emerge shortly afterwards with the regular classes (i.e., MASC -o/-i, FEM -a/-e). Noccetti (2002) studies the acquisition of morphology in a child, Camillo, one of the children investigated in the present paper, also within the theoretical framework of pre- and protomorphology. The data show that plural marking on nouns emerges early in the child’s speech and that the child soon grasps the meaning of plurality cognitively. The acquisition of the linguistic means to express this category is, however, gradual, confirming the results of the previous studies mentioned. The first plural suffixes employed by Camillo to refer to plural referents belong to productive regular morphology. Evidence for their productivity only emerges in protomorphology, where there are examples of overgeneralizations of the FEM:PL suffix -e to irregular feminine nouns (Noccetti 2002: 68–69). The other unproductive classes are sporadically used either in the singular or the plural, i.e. there is no evidence of the acquisition of the irregular classes. In the present paper, the data of the child Camillo will be compared with the data of another child, Rosa. The paper focuses on the development of number in nominal inflection during the acquisition process of Italian as a first language within the theoretical framework of Natural Morphology. Its main objectives are: a) the description of the emergence and development of nominal number marking in the early grammars and b) to highlight how the children move on from the use and understanding of separate lexical items to the first formation of lexical and functional categories.

306

Sabrina Noccetti

2. The expression of number in Italian nominal inflection In Indo-European languages the category of number is obligatorily marked with grammatical resources. Italian has a two-value number system contrasting singular and plural (Corbet 2000). In Italian, a weakly inflecting-fusional language, the category of number is grammaticized and obligatorily expressed by inflectional morphology with nouns, adjectives, pronouns, determiners and verbs. There is a nominal inflection different from the verbal one. Nouns, adjectives, determiners and possessive pronouns are inflected for number and some adjectives, articles and pronouns are in addition inflected for gender, viz. masculine and feminine, whilst gender is intrinsic in nouns (see § 2.1., 2.2.).2 Even though there is no strict difference between numeral and transnumeral languages, since numeral languages also possess certain nouns which do not distinguish between singular and plural, namely mass nouns, Italian is a ‘numeral’ language because the majority of nouns occur in the plural form when they denote more than one referent (Wiese 2003; Corbet 2000). The strongly grammaticized character of the two categories of number and gender in Italian is shown by the fact that gender and number marking is obligatory with determiners and adjectives constructed with nouns (e.g. le bell-e ragazz-e ‘the:FEM:PL beautiful-FEM:PL girl-FEM:PL’) and with nouns constructed with quantifiers, which clearly indicate a plural context (e.g. quattro ragazz-e ‘four girl-PL’). Within the noun phrase, gender and number are controlled by the noun. The subject noun also governs the number of the verb, although with collective nouns an ad sensum agreement may occur, e.g. una folla di operai si avvicinarono alla porta della fabbrica ‘a crowd:SG of workers approach:3PL the door of the factory’ (Dardano and Trifone 1992: 182). Gender is established conventionally for inanimate nouns and follows the biological gender of almost all humans and animate nouns (Dardano and Trifone 1990, 1992; Marcantonio and Pretto 1991), with some exceptions, as for example the name of non-domestic animals such as FEM la zebra ‘the zebra’, MASC il pesce ‘the fish’ and MASC l’ippopotamo ‘the hippopotamus’, which refer both to the male and the female. Inflection is compulsory in Italian since the root does not exist as a free morpheme, e.g. *libr-, but must necessarily take a suffix, which is operationally substitutive, i.e. SG libr-o, PL libr-i ‘book, books’.3 Plural formation is not iconic as far as the length of the word is concerned because Italian words have the same length in both numbers (Thornton 2005: 165); however, it is generally morphotactically transparent in regular nouns, adjectives and some articles (see §§ 2.1, 2.2) because the morphs can be easily identified and processed. Still, Italian nouns and adjectives range from transparent to non-transparent plural

The emergence of nominal number in Italian

307

marking, e.g. plurals with a morphonological rule of palatalization (SG amico ‘friend’, PL amici) and irregular plural (SG uomo ‘man’, PL uomini) (see below).

2.1.

Nouns

In Italian nouns the categories of gender and number are fused in the same suffixes. Dressler and Thornton (1996) divide the Italian nominal system into macro- and microclasses.4 They assign the Italian nouns to three different inflectional microclasses, I, II, III (see table 1). Microclass I comprises two classes which share gender and morphological form: the nucleus of this microclass, I.1 (MASC nouns, -o/-i ‘SG/PL’), and I.2 (MASC nouns ending in -[k]-o and -[g]-o which palatalize the velar consonant in the PL, i.e. -[Ù]-i; -[Ã]-i respectively). The other two classes of microclass I are I.3 and I.4. They only partially share the features of I.1, but are not totally different to form two independent classes. In fact, I.3 includes nouns which are MASC in SG but FEM in PL (PL -a) and 1.4 consists of MASC nouns with PL in -i but SG in -a. Microclass II comprises one class, II.1 (FEM nouns, -a/-e ‘SG/PL’). Microclass III (-e/-i ‘SG/PL’), includes the classes III.1 (MASC nouns) and III.2 (FEM nouns). Moreover, there are isolated satellite paradigms of both MASC and FEM nouns which comprise a small number of nouns (some examples are given in table 1). The microclasses I.1 and II.1 are frequent and fairly productive (Dressler and Thornton 1996; Tonelli et al. 1998). Here, the criterion of productivity of a rule or a class means that new words, such as indigenous neologisms, foreign words and nouns of the other classes, are inflected according to this rule or integrated into this class (Dressler and Thornton 1996; Kilani-Schoch et al. 1997; Tonelli et al. 1998; Bittner et al. 2003; Dressler 2003). However, the productivity of I.1 and II.1 seems to be reduced on the basis of recent loans which persist in their original forms without any morphological integration into the Italian system and accordingly Dressler (2003: 39) casts doubts on the full productivity of these two classes. Class I.2 is less transparent than microclass I.1 because of the morphonological rule of palatalization in the PL (cf. examples in table 1). Thus, the opposition between SG and PL in this declension class is morphotactically opaque and unproductive (Dressler 2005: 13; Savickien¨e and Dressler 2007: 4). Since palatalization is an intrinsically phonological process, it is easy to understand that its morphologization will cause fluctuation in the formation of the plural; in fact, Rohlfs (1968: 44-45) attests two plural forms for Italian nouns ending in -[k]-o (e.g., gre[k]-i and gre[Ù]-i ‘Greeks’) and frequent class shift from microclass I.2 to I.1, which does not exhibit palatalization, e.g. Tuscan ami[k]-i (Rohlfs

308

Sabrina Noccetti

Table 1. Italian nominal inflection Inflectional classes I I.1 (MASC) I.2 (MASC)

II III e.g. of isolated paradigms

I.3 (MASC:SG/ FEM:PL) I.4 (MASC) II.1 (FEM) III. 1 (MASC) III.2 (FEM) (MASC)

(FEM)

SG/PL -o/-i -[k]-o/ [Ù]-i -[g]-o/ [Ã]-i -o/-a

Examples libr-o/libr-i ami[k]-o/ ami[Ù]-i biolo[g]o/ biolo[Ã]-i dit-o/dit-a

Glosses book/s friend/s

-a/-i -a/-e -e/-i -e/-i

poet-a/poet-i cas-a/cas-e leon-e/leon-i lepr-e/lepr-i uomo/uomini dio/dei mano/mani ala/ali

poet/s house/s lion/s hare/s man/men god/s hand/s wing/s

biologist/s finger/s

1968: 21–22). The plural of nouns ending in -[g]-o and -[k]-o also depends on the position of the stressed syllable within the word and on the semantic class to which the word belongs,5 but in spoken Italian there are frequent mistakes and Rohlfs (1968: 45) admits that the plural formation of these nouns is complex and confused (see also Dardano and Trifone 1992: 191). More recently, also Celata and Bertinetto (2005: 295) admit that the process of palatalization is unpredictable and extremely unstable in spoken Italian. Nouns belonging to macroclass III are rather frequent. As shown by their shift to the productive classes I.1 or II.1 in some dialects, including the Tuscan variety, this class is not considered stable (Rohlfs 1968: 14–15). By the criteria of stability and capacity to attract new words, this inflectional macroclass is not considered to be productive, although in substandard Italian some loan words ending in a consonant are integrated into class III, e.g. English film > Tuscan SG il film-e PL i film-i, English stop > Tuscan SG lo stopp-e PL gli stopp-i. However, the co-occurring plural forms in -e, e.g. PL i film-e, gli stopp-e, casts doubt on whether they are not phonologically rather than morphologically integrated, since the addition of a final -e adapts them to Italian word prosody by making them end in an open syllable. The following classes of nouns do not mark the plural: (a) nouns ending in a tonic vowel (e.g. citt´a ‘town’), (b) some monosyllabic nouns (e.g. gru ‘crane’), (c) certain loans (e.g. gorilla). Abstract nouns are usually singularia tantum, e.g. coraggio ‘courage’, pazienza ‘patience’.

The emergence of nominal number in Italian

309

2.2. Adjectives The adjective is inflected for gender and number in both predicative and attributive position, e.g. i ragazz-i educat-i sono poch-i ‘the:MASC:PL boy-MASC:PL polite-MASC:PL are:3P few-MASC:PL’. There are two groups of adjectives, (a) those which preserve gender distinction and are inflected like class I.1 nouns when MASC (e.g. il ragazzo alt-o ‘the boy tall-MASC:SG’) and like class II.1 when FEM (e.g. la ragazza alt-a ‘the girl tall-FEM:SG’); and (b) those inflected like the nouns of microclass III, i.e. SG -e and PL -i without gender distinction (e.g. il grand-e giardino ‘the:MASC:SG big-SG garden:MASC:SG’ and la grand-e casa ‘the:FEM:SG big-SG house:FEM:SG’). In standard Italian, there is a group of color adjectives ending in -one as well as the color names blu ‘blue’, rosa ‘pink’, viola ‘violet’ not inflected for number and gender (e.g. scarp-e marrone ‘shoe-PL brown’). In everyday spoken language though, the final vowel of color names ending in -one is reanalyzed as a suffix so that these words take the plural suffix -i (e.g. scarp-e marron-i ‘shoe-PL brown-PL’). When an adjective is constructed both with a FEM and a MASC noun, gender distinction is usually neutralized in favor of the unmarked MASC, e.g. quest-i bambin-i e quest-e bambin-e sono buon-i ‘this-MASC:PL boy-MASC:PL and this-FEM:PL girl-FEM:PL are:3P good-MASC:PL’ (Dardano and Trifone 1992: 218; Marcantonio and Pretto 1991: 322). Numerals are not inflected for gender, with the exception of uno ‘one’, which coincides with the indefinite article, MASC un-o, FEM un-a. Ordinal numbers are always inflected for gender and number following the most productive noun classes -o/-i and -a/-e, e.g. prim-o/prim-i ‘first-MASC:SG/PL’, prim-a/prim-e ‘first-FEM:SG/PL’. Possessive adjectives inflect for both gender and number in agreement with the possessed and follow the declensional pattern of class I.1 or II.1 nouns (see table 2). They agree with the possessed in gender and number (e.g. i mie-i libr-i ‘the:MASC:PL my-MASC:PL book-MASC:PL’). Table 2. Possessive adjectives PERSON 1S 2S 3S 1P 2P 3P

MASC:SG mi-o tu-o su-o nostr-o vostr-o loro (suo)

MASC:PL mi-e-i tu-o-i su-o-i nostr-i vostr-i loro (suoi)

FEM:SG mi-a tu-a su-a nostr-a vostr-a loro (sua)

FEM:PL mi-e tu-e su-e nostr-e vostr-e loro (sue)

310

Sabrina Noccetti

Sometimes, in the Tuscan variety, the input language of the children investigated, the possessive adjective loses its suffix, canceling gender and number distinctions, which are only expressed by the article and the noun (e.g. i mi’ fratelli). Possessive pronouns, which are formally identical to adjectives, cannot be truncated, e.g. il libro e` (il) su-o ‘the book is his/hers’. In the spoken language, there is neutralization of number in the third person plural (suo/sua) where the feminine or masculine singular form is normally used instead of loro (see table 2). Neutralization also applies when the possessive adjective refers to both masculine and feminine nouns (e.g. i miei fratelli e sorelle ‘the:MASC:PL my:MASC:PL brother:MASC:PL and sister:FEM:PL’).

3. Methodology The present paper is based on the data of two children, Camillo (CAM) and Rosa (ROS). They were born and grew up in central Italy and their input language is the Tuscan variety.6 Camillo’s data were collected from the age of 2;0 to 3;6 and Rosa’s from the age of 1;7 to 3;3.7 The children were audio-recorded twice a month and Rosa was also video-recorded. The recording sessions of Camillo, a late speaker, lasted from 15 to 30 minutes and those of Rosa about 30 minutes each. They mainly consisted of daily routines such as playing or book reading during spontaneous interactions with their mothers and sometimes other caretakers. Transcripts were computerized and analyzed within the CHILDES Project (MacWhinney 2000). In table 3 the ages and length of recordings have been summarized for each child. Sometimes two recordings have been merged when they took place in the same month. The data have been divided into three different periods representing different evolutionary steps in language development: premorphology, protomorphology, 1st and 2nd period (see table 3). The period of Modular Morphology (presumably from 2;8 for CAM and 2;11 for ROS) will not be analyzed in the present paper, because there are not enough data to describe the further development of noun morphology. The distinction of periods is here mainly related to changes in nominal morphology, to the emergence of syntactic agreement in the noun phrase and to syntactic development more generally. These criteria appear to be the most appropriate when studying the emergence of nominal number marking. However, they may vary when other features of child language are taken into account, i.e. the child’s development may not be parallel for all grammatical categories. Of course, greater developmental differences between grammatical categories are assumed to occur in the protomorphological period, while the transition from

The emergence of nominal number in Italian

311

Table 3. Data analyzed Periods of development Premorphology

Protomorphology I Protomorphology II

Age 2;0 2;1 2;2 2;3 2;3.15 2;4 2;5 2;6 2;7 2;8

CAM 30 60 60 30 15 40 50 25 60 60

Age 1;7 1;9 1;10

ROS 30 30 30

1;11 2;0 2;1 2;2 2;4 2;5 2;7 2;8 2;9 2;11

35 30 60 40 80 30 30 30 70 30

pre- to protomorphology is expected to involve grammar at large (Bittner et al. 2003: xxv). In accordance with previous studies (Dressler and Karpf 1995; Kilani-Schoch et al. 1997; Kilani-Schoch and Dressler 2002; Bittner et al. 2003; Dressler 2005), premorphology is the period in which child speech shows a small number of lexically stored inflectional forms and extragrammatical operations. The second period of protomorphology is the time in which the child begins to decompose form and meaning word-internally. In this period productivity develops, i.e. recurrence of a word-form or category in different types of words or lexemes (Kilani-Schoch and Dressler 2002: 45) and paradigms emerge. The criteria for establishing the onset of paradigms have been developed by KilaniSchoch and Dressler (2000: 50–52). They propose that the forms a) should be spontaneously produced by the child and not in imitation; b) should not be phonetically ambiguous or doubtful; c) should not be context-bound but used in different and, possibly, contrasting contexts; d) should not be frozen or formulaic; e) should not appear in isolation, i.e. at least three different grammatical forms should occur within a month. This criterion cannot be met by Italian nouns, and it seems preferable to rely on contrasting grammatical forms such as SG and PL. These criteria will be here followed to identify a true paradigm. Protomorphology is preceded by a transition phase, here called protomorphology I, in which features both of the preceding and the following period can be observed.

312

Sabrina Noccetti

In protomorphology I, children start to differentiate between morphological forms as shown by their use of inflected word forms. These inflected forms, which co-exist with some manifest remnants of premorphological extragrammatical operations (e.g. syllable suppression, reduplication), lead the child to formulate his/her first hypotheses on the input language and to develop morphological patterns which are further elaborated in the second period of protomorphology. The principal feature of this period is the growing number of paradigms, which will gradually lead the children to build the inflectional system of their language. This means that contrasts between inflectional forms occur with new lexemes, e.g. SG/PL in nouns, MASC/FEM and SG/PL in ADJ and DET, PERS/TNS in verbs (Bittner et al. 2003: xxv). The selection of the morphological patterns is carried out on the basis of type and token frequency and universal preferences for morphotactic and morphosemantic transparency, biuniqueness, and iconicity.

4. Overview of the structural characteristics of the children’s utterances, with special attention to number and gender agreement The aim of this section is the description of the development of the noun phrase with respect to plural and gender marking in nouns and adjectives during the period of language acquisition studied in the present paper. The data of CAM and ROS will first be analyzed independently and afterwards a comparison of the two children will be attempted.

4.1.

Camillo’s development

4.1.1. A developmental sketch of CAM’s sentence structure In premorphology, CAM mainly uses one-word utterances. As is common in early child speech, the development of two-word utterances is preceded by sequences of two or three one-word utterances expressing a single predication and united by a single intonational curve, but separated by a short pause (here indicated by a comma). These one-word sequences may be composed of elements belonging to different parts of speech and fulfilling different functions, such as a demonstrative pronoun and a locative adverb (e.g. at 2;0, questo, dentro ‘this one, inside’), a vocative noun and an imperative verb form (e.g. at 2;0, (vi)eni,

The emergence of nominal number in Italian

313

mamma ‘come, mummy’), or a vocative, a noun and an imperative verb form (e.g. at 2;2, babbo, oco, veni < babbo, gioco, vieni ‘daddy, game, come’). Although there are still one-word utterances to be found in protomorphology I beginning at 2;3.15, two-word utterances become more frequent. They are composed of an adjective and a noun (e.g. uguale ossacchiotto < uguale orsacchiotto ‘similar bear:DIM:SG’), an imperative verb form and an adverb (e.g. tai bono < stai buono ‘be good’), a noun and a past participle (e.g. at 2;4 occhi pati < occhi scappati ‘eyes escaped’), or an auxiliary verb and past participle (e.g. sono pati < sono scappati ‘they have escaped’). At the age of 2;5, in protomorphology II, the first object pronoun emerges in a two-word utterance (la pettino ‘(I) comb her’, see 4.1.4) and at 2;6, a four-word sentence composed of a noun and a verb phrase occurs, with prepositions and articles missing (a babbo Litta panno pappa < al babbo di Elisa piace la pappa ‘Elisa’s father likes food’). While prepositions and articles are sometimes replaced by the vowel e (e.g. etto bimbo viene e zia < questo bimbo viene da zia ‘this boy goes to aunt’s (home)’), at other times they are produced (e.g. in the prepositional phrase co a linga foli < con la lingua fuori ‘with the tongue outside’). In protomorphology II, also the first negative sentences occur (e.g. un lo so ‘I don’t know’, un c’`e niente nel cassetto ‘there is not anything in the drawer’, where the adverb un is the Tuscan form for standard Italian non ‘not’). Complex sentences such as the juxtaposition of two main sentences (e.g. questo e` il cane e questo e` il lilla ‘this is the dog and this is the gorilla’) and the first causative construction (pecch´e no toccagli il folno pecch´e blucia ‘because do not touch the oven because it burns) also emerge in this period.

4.1.2. Nouns and adjectives During the premorphological period, nouns as well as adjectives generally occur in a single form. The only two exceptions are MASC brutt-o (3 tokens) and FEM brutt-a (1 token) at 2;3 (see below), and giallo ‘yellow’ which occurs both in MASC and FEM:SG (giall-o 8 tokens, giall-a 1 token) and FEM:PL (giall-e 1 token). The MASC:SG and FEM:PL forms of an adjective are opposed only once in the following dialogue: (1)

MOT: c’ ha i pappagallini? ‘has he got the:MASC:PL parrots:MASC:PL?’ MOT: quanti ce n’ ha? ‘how many has he got?’ MOT: tanti? ‘many:MASC:PL?’

314

Sabrina Noccetti

CAM: giallo. ‘yellow:MASC:SG’ MOT: di che colore sono? ‘what color are they?’ CAM: gialle. ‘yellow:FEM:PL’ If the adjective giallo (MASC:SG) refers to pappagall-in-i (DIM-MASC:PL) the standard Italian form should be giall-i (MASC:PL). The second form, FEM:PL gialle, is an answer to the mother’s question which picks up the child’s idea of referring to the color rather than the quantity of the referents of pappagallin-i MASC:PL. Once more, the most adequate form here would have been giall-i MASC:PL. The MASC:SG form giall-o is the adjective form mainly occurring in the child’s speech (8 tokens vs. 1 FEM:SG token giall-a) and always refers to nouns of class I.1. Before interpreting these forms, it is necessary to observe the child’s input. In table (4) (and table 5 for ROS), noun endings occurring in the input have been calculated without taking their inflectional class into consideration so that, for instance, all nouns ending in -i and belonging to classes I.1, I.2, I.4, III.1 or III.2 plus all isolated paradigms such as man-i ‘hands’have been collapsed. Still, inflectional classes are indicated in parentheses. Nouns ending in a consonant (C) have also been included in the table. SG and PL have been distinguished because it is reasonable to assume that the child is confronted with different referents and contexts from which he may extract information on the use of different forms in environments where the referents differ numerically. While it is taken for granted that numerical distinction is self-evident, the same cannot be said for gender, which is arbitrary for inanimate nouns and adjectives.

Table 4. Tokens (%) of nouns ending in -i, -e, -o, -a or C in CAM’s input

-i (PL) -e (SG) -e (PL) -o (SG) -a (SG) -a (PL) C

-i PL (I.1/2/4+III.1/2+isol); -e SG (III.1/2); -e PL (II.1); -o SG (I.1/2/3/isol); -a SG (II.1+I.4+isol); -a PL (I.3) Premorphology Protomorphlogy I Protomorphology II 16 (12.4) 6 (9.2) 62 (10.7) 27 (20.9) 21 (32.3) 85 (14.7) 8 (6.2) 4 (6.1) 24 (4.1) 45 (34.9) 17 (26.2 210 (36.3) 32 (24.8) 17 (26.2) 183 (31.6) 0 0 5 (0.9) 1 (0.8) 0 10 (1.7)

The emergence of nominal number in Italian

315

Table 4 shows that, in premorphology, -o is the most frequent SG nounending in CAM’s input, with -a and -e coming second and third, while -i is much less frequent. The following combinations of N and ADJ endings will therefore presumably be most entrenched in the child: N-o ADJ-o, N-a ADJ-a, and N-e ADJ-e/ADJ-o, while the combination N-i ADJ-i is less frequent. In fact, in his output, the child utters only one context-bound ADJ in -i, tutti ‘all’. In addition, CAM uses only two ADJ in -e, arancion-e ‘orange’ and verd-e ‘green’, which are uttered in contexts where the mother asks for the color-e ‘color’ (N-e) of two painted objects. Thus, the child does indeed produce ADJ-e in combination with a N-e, although infrequently. In contrast to this, in the first recordings, nouns of inflectional classes III are never used by the child in combination with an adjective. Consequently, the combinations for these classes, N-e ADJ-o, and the PL N-i ADJ-i, never appear. We may hypothesize from this that the child has constructed a kind of parallelism between noun and adjective endings, prompted by their symmetrical pattern as well as their input frequency. Consequently he models the adjective form giall-e (FEM:PL in standard Italian) in example (1) on the ending of the noun color-e (MASC:SG) thereby creating a surface analogy. In premorphology, utterance sequences consisting of a MASC:SG or FEM:SG noun and the corresponding form of the possessive or some other adjective constructed with it are rare in Camillo (e.g. at 2;2.14, ett-a, mi-a, mi-a < racchetta ‘racket, mine-FEM:SG, mine-FEM:SG’; 2;3, lup-o, butt-o < brutto ‘wolf-MASC:SG, ugly-MASC:SG’, papp-a, butt-a < talpa brutta ‘mole-FEM: SG, ugly-FEM:SG’) and words are sometimes subject to vowel harmony so that no specific grammatical status can be determined (e.g. 2;3, lupu, buttu < lup-o brutt-o ‘wolf-MASC:SG, ugly-MASC:SG’). By comparing the two occurrences of the adjective brutto ‘ugly’ in lup-o, butt-o and lup-u, butt-u it appears that the child adapts the ADJ ending to that of the noun. Although buttu could result from applying vowel harmony to itself, the fact that it is realized as butt-a when combined with a FEM:SG noun shows that it is dependent on the shape of the noun. This again supports the hypothesis that the child is initially guided by surface analogies in constructing noun and adjective pairs (see also example 1). This will, however, result in a blind alley as soon as the child meets nouns and adjectives of the microclasses of inflectional class III as well as other irregular nouns. Through the period of protomorphology I, words may be put into inadequate gender classes (e.g. 2;3.15, gata - seemingly FEM:SG - for gatt-o ‘catMASC:SG’), but there are no more examples resulting from vowel harmony after 2;3.15. In protomorphology II, there are several other examples of adjectives presenting gender and number distinctions (e.g. questo/questi/questa ‘this-MASC:SG/ MASC:PL/FEM:SG’, mio/mia ‘my:MASC:SG/FEM:SG’). While giving evi-

316

Sabrina Noccetti

dence of the same N/ADJ combinations already found in premorphology, he now starts to use the MASC:PL combination N-i ADJ-i, e.g. ricc-i giall-i ‘hedgehogs yellow’. There is one III.1 class ADJ (-e/-i), ugual-e ‘equal’, which is used in the singular referring to a MASC noun in -o, and a N in -e combined with an ADJ in -o (latt-e ross-o ‘milk red’) so that new combinations occur, N-o ADJ-e, N-e ADJ-o (see § 4.1.1). As he meets other combinations of N and ADJ the boy seems to abandon the surface analogy between noun and adjective endings. It is in the second period of protomorphology that various adjectives of classes I.1 and II.1 appear more consistently in two- to three-term paradigms (e.g. quell-o/quell-i ‘that-MASC:SG/PL’; altr-o/altr-i ‘other-MASC:SG/PL’; rosso/ross-i/ross-e ‘red-MASC:SG/MASC:PL/FEM:PL’), which is a sign of lexical dispersion of the inflectional patterns I.1 and II.1. Class III adjectives (-e/-i) continue to occur in a single form. There is an interesting example of the inadequate use of the singular of the class III adjective verd-e ‘green’ at 2;5 (example 2). (2)

MOT: le calze di che colore sono? ‘the sock:FEM:PL, what color are they? CAM: verd-e. ‘green-SG’ MOT: No, ro..? ‘no, re..?’ CAM: ross-e. ‘red-FEM:PL’

The form is old since it is already used in premorphology when CAM (at 2;0) is asked to name colors in a book. It is also the only form of this adjective occurring in the input (2 tokens). The question is how to interpret the form verde in example (2). As in the example 1 above, it might either be a surface analogy based on the PL noun calz-e or the SG noun color-e. However, a more suitable hypothesis for this period of development is to take it as an overgeneralization of the inflectional pattern -a/-e (to which the other adjective ross-e ‘red-FEM:PL’ occurring in the same situation belongs) so that verde shifts from the inflectional microclass III to II.1 and -e is accordingly a PL marker. This hypothesis is supported by different types of observation: 1) The child now uses the appropriate ADJ form in agreement with the noun when he is asked the color of objects, so that a surface analogy based on the noun ending of color-e can be excluded. 2) II.1 nouns (-a/-e) are the second most frequent class (see table 9) with 40 lemmas and 79 tokens in the SG and 10 lemmas and 11 tokens in the PL, and class II.1 ADJ (-a/-e) are more frequent than class III ADJ

The emergence of nominal number in Italian

317

(-e/-i), 3 lemmas and 4 tokens in the PL vs. 1 lemma and 1 token of class III, (occurring three months later at 2;8) and 8 lemmas and 15 tokens in the SG vs. 3 lemmas and 7 tokens of class III. Thus, the child produces SG -a and PL -e nouns more frequently in combination with -a and -e ADJ than with -e/-i ADJ. 3) While one paradigm of class II.1 ADJ occurs in the child’s speech (ross-a/ross-e ‘red-FEM:SG/PL), no paradigms of class III adjectives are present. 4) There are two class II.1 noun paradigms to be found in the child’s speech (Elis-a/-e and barc-a (8 tokens)/barch-in-e ‘boat-FEM:SG/DIM-FEM:PL’ (1 token)) but none of III.2.8 5) There are 6 noun paradigms in the input belonging to class II.1 vs. 2 of class III.1. Since there are no examples of FEM class III.2 nouns, the ending -e is preferentially associated with PL contexts rather than SG ones. 6) The child uses the inflectional pattern II.1 creatively with the FEM proper noun Elis-a/Elis-e (2 tokens each), an uncommon though not impossible Italian form not present in CAM’s input (see § 5.1.3). For these reasons and given the fact that adjectives are mainly inflected according to the productive microclass II.1 and the child never opposes SG -e to PL -i with adjectives, verd-e in example (2) can be considered to be an overgeneralization of a productive inflectional paradigm. After 2;5 and all through protomorphology adjectives are never used inappropriately. Class III adjectives continue to be only used in the singular. At 2;8 an example is found where the class III adjective grande agrees in number with a class II.1 noun: bocc-a grand-e ‘mouth-FEM:SG big-SG’. It is only at the age of 3;0, in the transition to modular morphology, that the first spontaneous plural form of verde occurs in the noun phrase quell-e verd-i ‘those-FEM:PL green-PL’ with the ADJ correctly matching the plural of the noun. In protomorphology II, there are instances of gender and number distinction for class I.1 and II.1 adjectives found in two-word utterances (noun plus adjective), both in the SG and the PL (e.g. (papera) bell-a mi-a ‘(duck:FEM:SG) beautiful-FEM:SG my-FEM:SG’, leon-e plima piccin-o poi gloss-o ‘lionMASC:SG first small-MASC:SG then big-MASC:SG’, stival-in-i giall-i ‘bootDIM-MASC:PL yellow-MASC:PL’ at 2;5 and facci-a ross-a ‘face-FEM:SG red-FEM:SG’ at 2;8). In premorphology and protomorphology I, the child appears to be guided by both type and token frequency in the input and by extra-grammatical rules, as shown by the examples of surface analogy. In premorphology there are neither number nor gender paradigms giving evidence of an opposition of different grammatical forms of the same lexeme nor is there overt morpho-syntactic agreement since the child’s utterances are composed of a single element. Adjectives appear either in the FEM or MASC gender, in SG or PL and the context does do not unambiguously reveal their grammatical status.

318

Sabrina Noccetti

Unlike premorphology, during protomorphology I some two-word sentences appear which show an appropriate use of gender and number marking.Adjectives begin to be used in more than one form, but they seem to be combined with nouns on the basis of extra-grammatical rules. However, expansion of the child’s speech to two-element utterances and the presence of different forms of the same word (see also § 5.1.2) are more consistent in this period, and this is why it is reasonable to distinguish it from the previous one; but it is still a transitional stage before the child enters the period in which he moves from the use of separate lexemes to building morphological patterns. In protomorphology II, more instances of two- and three-word utterances demonstrate the adequate use of number and gender marking on nouns and adjectives (example 2 being an exception). This can first be noticed with inflectional microclasses I.1 and II.1, which show lexical diffusion soon after their emergence as they are extended to other noun and adjective lemmas. Entire paradigms of lemmas belonging to these microclasses occur within the same recording (see also 4.1.4. below) and there is one overgeneralization of the productive class II.1 pattern involving a non-productive class III adjective. Adjectives of III class do not occur in more than one grammatical form in the developmental periods here analyzed.

4.2.

Rosa’s development

4.2.1. A developmental sketch of ROS’s sentence structure In premorphology (1;7–1;10), the child uses onomatopoetic words to refer e.g. to cars (bum [bum]) and animals (e.g. cococo [kokok´o] for chickens). As also noticed by Cipriani et al. (1993: 44), ROS shows a massive use of fillers indiscriminately preceding both nouns and verbs. They consist of one of the single vowels [e], [a] or [i], (a/e eni < vieni ‘come’, a/i dada < tata),9 the nasals [n] or [m] (e.g. m/n babbo ‘dad’) or a vowel followed by a nasal (e.g. en miao ‘cat’, em endee < vendere ‘to sell’). Although ROS’ speech mainly consists of one-element utterances, there are also some two-element ones expressing a single predication but with their constituents separated by a short pause. As with CAM, these utterances are composed of different elements, a verb and a noun, (e.g. at 1;7, voo, oppa < voglio, orso ‘want bear’), a demonstrative and a verb (e.g. at 1;9, etto, tiei < questo, tieni ‘this, take’) or a demonstrative and an adverb (e.g. at 1;9 questo, dentro ‘this one, inside’). In protomorphology I, at 1;11, two- and three-word utterances without an internal pause begin to appear, although one-element utterances, fillers, and

The emergence of nominal number in Italian

319

onomatopoetic words continue to be used. Two- and three-word utterances may be composed of a demonstrative and a noun (e.g. ette bee < queste pecore ‘these sheep’), an adverb and a noun (e.g. dentro chicco ‘inside seed’), a noun and a past participle (e.g. notto e occhi < rotti gli occhi ‘broken:MASC:SG the eyes:MASC:PL’) a negative adverb and an infinitive verb form (e.g. at 1;11, dere no < sedere no ‘sit no’), an adjective preceded by a filler and an infinitive (e.g. enatto evare < un altro levare ‘another remove’). At 2;0 even a four-element utterance is found (voio unatta penna < voglio un’altra penna ‘want another pen’). One month later, at 2;1, fillers begin to be differentiated. The vowels [e], [a], [i] and the syllables en/em occupy the place of the definite and indefinite article, respectively, um is found in place of the clitic indirect object pronoun ti (e.g. at 2;2, um do lotta < ti do una botta ‘to you I’ll give a punch’) while un replaces the negative adverb non (e.g. at 2;2, un leve < non levare ‘do not take away’). Fillers and rare omissions of free morphemes compete with their replacement by standard forms, so that new elements appear in the child’s utterances. Actually, the first clitic pronouns emerge in two-word utterances (gli urla ‘she screams to him’) or negations (no(l)o so ‘I don’t know it’). Some prepositions also appear (e.g. at 2;1, numero di babbo ‘number of dad’). Sentence structure begins to expand, although subordinate clauses are generally not combined with the main clause, but are rather expansions of previous conversational turns as in example (3). (3)

ROS:

fuori ‘out’ MOT: perch´e? ‘why?’ ROS: perch´e i(l) gatto Birilla fuori dorme ‘because the cat Birilla outside sleeps’

It is only at 2;5 that an example is found in which the main clause follows the causal one (e.g. perch´e bimbo vuole sua mamma, ciange ‘because baby wants his mother, he cries’). Relative clauses also emerge in this period, but the relative pronoun is replaced by a filler (e.g. quell’omino ?m pala ‘that man who shoots’). Furthermore, infinitives depending on the modal verb volere ‘want’ are beginning to appear (e.g. at 2;5, voglio vedere quello ‘(I) want to see that one’).

4.2.2. Nouns and adjectives In premorphology, nouns occur in a single form, either SG or PL.There are 28 SG noun lemmas and only 3 PL ones (see table 11). The latter belong to the class

320

Sabrina Noccetti

of Italian plural-dominant nouns and are also PL in the input (scarpe ‘shoes’, biscotti ‘biscuits’, fiorellini ‘flower:DIM:PL’). At 1;7 there are two examples of vowel harmony in the noun tuttu < ciuccio ‘pacifier’ and at 1;9 in the name of ROS’s brother Nini < Nino. As is the case with nouns, adjectives are also used in a single form with the exception of natto < un altr-o ‘an other-MASC:SG’ (19 tokens) and natta < un’altr-a ‘an other-FEM:SG’ (2 tokens), where the initial n is probably a filler replacing the undetected article un/una. These ADJ forms refer to MASC or FEM nouns and are used by the child when she wants something, so that their meaning really is ‘I want another X’. The MASC form is the most frequent in premorphology and is also used to refer to FEM nouns, e.g. mosc-a fly-FEM:SG’, at 1;10. Seemingly, natto is a rote-learned form, as shown by its inadequate use through protomorphology I. Rosa’s data give evidence of inappropriate use of this and other adjectives within the noun phrase as far as gender and number are concerned (e.g. at 1;11, mollitt-e un att-o < mollette altr-e ‘pin-FEM:PL other-MASC:SG’, quest-o televisione ‘this-MASC:SG television:FEM:SG’). In protomorphology II, there are no more errors concerning adjectives following the inflection of the noun microclasses I.1 (-o/-i) and II.1 (-a/-e). As far as nouns are concerned there are some class shifts assigning class III nouns to the most productive classes I.1 and II.1: (2;2) volp-a ‘fox’ for volp-e, (2;5) la font-a ‘the:FEM fountain’ for font-e and (2;10) un cuor-o ‘a:MASC heart’ for cuor-e. In such examples the child ROS seems to assign gender and number of the noun on the basis of the articles anticipated by her mother in e.g. questa e` la? ‘this:FEM:SG is the:FEM:SG’). Table 5 shows that SG -a and -o are the most frequent noun endings in ROS’ maternal input during the periods of premorphology through protomorphology II (see § 4.1.2.).

Table 5. Tokens (%) of nouns ending in -i, -e, -o, -a or C in Rosa’s input

-i (PL) -e (SG) -e (PL) -o (SG) -a (SG) -a (PL) C

-i PL (I.1/2/4+III.1/2+isol); -e SG (III.1/2); -e PL (II.1) -o SG (I.1/2/3/isol); -a SG (II.1+I.4+isol); -a PL (I.3) Premorphology Protomorphology I Protomorphology II 12 (4.3) 37 (10.6) 178 (9.4) 30 (10.7) 24 (6.8) 171 (9.0) 17 (6.1) 25 (7.1) 111 (5.8) 105 (37.5) 109 (31.1) 728 (38.3) 115 (41.1) 147 (42) 703 (36.9) 1 (0.3) 4 (1.2) 3 (0.1) 0 4 (1.2) 9 (0.5)

The emergence of nominal number in Italian

321

Comparing these input frequencies with the child’s speech it is found that she produces more nouns belonging to inflectional classes I.1 and II.1 than to other classes. Also, the first contrasts of SG and PL forms of given lemmas already occurring at 1;11 in protomorphology I come from these classes (see § 5.2.2.). The most frequent article-noun combinations, in both input and output, are presumably la N-a and il N-o. Class shifts of this period are overgeneralizations of classes I.1. and II.1, which are interpretable as giving evidence of productivity. ROS makes productive use of the regular classes I.1 and II.1 all through protomorphology II. Productivity is not only revealed by class shift but also by the number of paradigms flourishing in this period (13 paradigms of II.1 and 11 of I.1), which even exceeds that of her mother (see table 15). In contrast to this, paradigms of class III.1 only emerge at 2;4 with the word giornale ‘newspaper’ which is first realized as SG nan-o (2 tokens) and subsequently as SG onnal-e and PL onnal-i (1 token each). The first occurrence of this noun undergoes a class shift to inflectional class I.1 showing that class III.1 is quite unstable. Devescovi and D’Amico (2001: 132) maintain that microclass III nouns are problematic and that gender errors with this class are also observed in older children and even in adults (Rohlfs 1968: 45; Celata and Bertinetto 2005: 295). Class III.1 reappears with SG and PL forms four months later (2;8, maiale/maial-in-i ‘pig-SG/pig-DIM-PL’; 2;9, giornal-e/giornal-i ‘newspaper/s’; 2;10, bicchier-e/bicchier-i ‘glass/es’, cuor-e/cuor-i ‘heart/s’. The lexical expression of plurality first emerges with nouns belonging to the productive inflectional microclasses I.1 and II.1. At first, it is observed between nouns and plural referents of the extra-linguistic context, and later between nouns and adjectives within the noun phrase. At 2;1, in protomorphology II, number marking of nouns and adjectives of classes I.1 and II.1 is always adequate, e.g. quest-o bimb-o ‘this-MASC:SG baby-MASC:SG’, chicche chine < gallin-e piccin-e ‘chicken-FEM:PL little-FEM:PL’. This may be explained by the fact that for class I.1 and II.2 nouns and adjectives, the task is facilitated by the redundancy of word-endings of the noun and the adjective. By contrast, the SG forms of class III nouns and adjectives only show appropriate number marking at 2;8 (e.g. quest-o can-e ‘this-MASC:SG dogMASC:SG’, la font-e e` chius-a ‘the:FEM:SG fountain-FEM:SG is closed-FEM: SG’). However, in this same period, the child produces maial-e piccin-in-e, where the class III.1 noun maial-e ‘pig’ is characterized by the form piccin-in-e ‘small-DIM-FEM:PL’ instead of the adequate MASC:SG piccin-in-o. The noun maiale is apparently interpreted as a FEM:PL form of a class II.1 noun. The other tokens of class III nouns are realized as diminutivized forms, which change the microclass of the noun to class I.1 (i.e. maial-in-o/maial-in-i ‘pig-DIM-SG/PL’).

322

Sabrina Noccetti

Although inadequacies in the use of class III.1 and III.2 nouns might result from the lower frequency of -e/-i nouns in the input (see table 5), they might also depend on their opacity with respect to gender, which cannot be derived from the ending of the noun but is only apparent from agreement with articles and adjectives. The child seems to process these nouns item by item. This is revealed by the co-occurrence of adequate and inadequate combinations with adjectives as well as instances of class shift transferring class III nouns to the most transparent classes I.1 and II.1 (see also § 5.2.2). To summarize, while nouns and adjectives occurring in premorphology are rote-learned, the variety of forms increases in protomorphology I, but it is in protomorphology II that there are signs of the productivity concerning the productive word classes. At the same time infrequent nouns belonging to the unproductive classes III are often used inappropriately as far as gender and number agreement between nouns and adjectives is concerned.

4.3.

CAM and ROS: A comparison

In premorphology there is little difference in grammatical development to be found between the two children since the prevailing one-element utterances are mainly evidence of rote-learned forms. Extra-grammatical operations, examples of vowel harmony occur with both children. CAM seems to rely on surface analogies, while ROS uses onomatopoeia and fillers and seems to pay more attention than CAM to prosody (see Peters 1983). In the transition from pre- to protomorphology I, when two-element utterances become more frequent both children do, however, preserve features of the previous period. It is at this point that some individual differences begin to appear. While ROS produces negative adverbs, infinitives, past participles and some chunks where the article and the adjective form an unanalyzed amalgam and which are sometimes used inadequately, CAM uses imperative verb forms and adjectives together with nouns and adverbs, but also two-element utterances with past participles. In protomorphology II utterances expand and both children begin to use nouns and adjectives in more than one grammatical form, namely both numbers (nouns and adjectives) and genders (adjectives). It is at this point of their development that the differences between CAM and ROS are most pronounced. CAM seems to reconstruct the morphological patterns of premorphology initially based on surface analogies and begins to extend the most transparent and frequent inflectional classes of nouns and adjectives to other lexemes even

The emergence of nominal number in Italian

323

giving evidence of an instance of overgeneralization. In contrast to this, ROS’ inadequate productions of protomorphology I do not seem to be based on surface analogy, but rather result from the combination of two or three rote-learned forms. Accordingly, it is only in protomorphology II that she begins to build morphological patterns overextending the inflectional classes I.1 and II.1 to nouns of the less transparent and less frequent class III.1.

5. Development of number marking in the noun In this section we will be concerned with how synthetic number marking in the noun develops with CAM and ROS in pre- and protomorphology. Special attention will be given to the development of the function of PL to express plurality of referents.

5.1.

Camillo

5.1.1. Premorphological period Nouns are at first mainly attested during the activity of naming objects of the surrounding environment in early mother-child interaction. The child first uses each noun in a single context and only gradually extends its use to other situations. In premorphology, CAM produces 323 words only 63 of which are noun lemmas (19.5% of words). Since there are altogether 64 grammatical types, it follows that noun lemmas generally occur in a single form, either SG or PL (table 6) and the ratio between types and lemmas is only 1.04 (table 7). During this period most nouns occur in the singular, which also prevails in the input (table 6). An analysis of the interaction contexts in which SG or PL nouns occur shows that, in premorphology, SG contexts are much more numerous and that the child almost always reacts adequately (table 8). SG and PL nouns are mainly uttered in question-answer pairs and refer to CAM’s toys, animals and objects depicted in his favorite book as well as to the characters of his favorite animated cartoon. All nouns are thus produced in contexts familiar to the child representing well-known schemas in which the child’s responses have been often tested (see also § 4.1.2). The few cases in which CAM does not answer or answers inadequately either concern cases where he ignores the name of the requested object (example 4) or the mother deviates from a well-known question-answer sequence (example 5 vs. 6).

324

Sabrina Noccetti

Table 6. Noun lemmas, types and tokens (%), SG/ PL and uncountable nouns (U) types/tokens (%) in CAM and MOT

CAM MOT

CAM MOT

CAM MOT

Lemmas 63 (19.5) 73 (18.5)

Types 64 (19.6) 81 (16.4)

Lemmas 10 (12.5) 36 (16.2)

Types 10 (12.1) 39 (13.9)

Lemmas 131 (22.8) 243 (31.8)

Types 149 (23.1) 295 (27.6)

Premorphology Tokens SG 162 53/144 (18.4) (82.8/88.9) 129 59/96 (7.5) (72/74.4) Protomorphology I Tokens SG 14 5/8 (8.9) (50/57.1) 65 31/54 (8.3) (79.5/83.1) Protomorphology II Tokens SG 267 109/213 (10.7) (72.7/79.8) 579 201/456 (9.5) (68.1/78.8)

PL 7/14 (10.9/8.6) 16/24 (20/18.6)

U 4/4 (6.3/2.5) 6/9 (8/7)

PL 4/5 (40/35.7) 7/10 (17.9/15.4)

U 1/1 (10/7.1) 1/1 (2.6/1.5)

PL 35/47 (23.3/17.6) 73/90 (24.8/15.5)

U 6/7 (4/2.6) 21/33 (7.1/5.7)

Table 7. Ratio of grammatical types and lemmas in CAM Period premorphology protomorphology I protomorphology II

CAM 1.04 1.00 1.13

MOT 1.10 1.08 1.21

Table 8. CAM’s reactions to singular and plural contexts contexts SG

PL

CAM’s reactions Premorphology Protomorph. I 118 13 adequate 108 12 inadequate 2 0 none 8 1 12 6 adequate 7 6 inadequate 2 0 none 3 0

Protomorph. II 89 86 0 3 37 32 2 3

The emergence of nominal number in Italian

(4)

MOT: questo cos’`e? ‘what is this-SG?’ CAM: boh! ‘boh (I don’t know)’ MOT: non lo sai? ‘don’t you know it?’

(5)

MOT: le caramelle chi ce le ha mangiate? ‘the candies, who ate them?’ CAM: gatto ‘cat’ MOT: e te sei rimasto senza? ‘and you remained without?’ CAM s`ı ‘yes’ MOT: ascolta, ma quante volte ce l’ha mangiate? ‘listen, how many times did he eat them?’ CAM: (se lo) chiappo! ‘(if him) catch:1S!’

(6)

MOT: le caramelle chi ce le ha mangiate? ‘the candies, who ate them?’ CAM: gatto ‘cat’ MOT: e te sei rimasto senza? ‘and you remained without?’ CAM s`ı ‘yes’ MOT: cosa gli fai al gatto se lo chiappi? ‘what will you do to the cat if you catch him?’ CAM: (yelling) gatto!! ‘cat!’

325

Nouns occurring in the PL in the child’s speech during premorphology are either plural dominant nouns in the adult language, e.g. spalle ‘shoulders’, giostre ‘carrousels’, calzettoni ‘socks’, or refer to several referents in the contexts familiar to the child, e.g. elefanti designating a group of elephants in an animated film. Since the contrastive use of both numbers is not yet observed during this period, it remains unclear whether the plural forms do indeed refer to a plurality of referents. Therefore, it is difficult to determine the status of inappropriately

326

Sabrina Noccetti

used forms such as banti < elefanti ‘elephants’ in example (7), which may represent the stored form of this lemma or a phonetic variant of elefante in the child’s speech. (7)

MOT: cos’`e quest-a? ‘what is this-FEM:SG?’ CAM: bant-i ‘(ele)phant-MASC:PL’ MOT: l’elefant-e ‘the elephant-MASC:SG’

Since the correct use of SG or PL noun forms seems to depend on the form in which they have been stored in their respective contexts rather than on their grammatical status, the child seems to consider them as unanalyzed wholes so that noun endings have not yet assumed their morphosyntactic functions. This interpretation is also supported by the examples of surface analogies and vowel harmony discussed in section 4.1.2. Although, in the premorphological period, CAM does not construct SG noun forms with numerals to express plurality, the child does express quantity by lexical means, numerals as well as adjectives. Numerals such as sei ‘six’ at 2;1 and tre ‘three’ at 2;3.15 in protomorphology I do not refer to a specific number of items but more generally mean ‘more than one’ (example 8). The adjective tanti ‘many’ is used in much the same way (example 9). (8)

MOT: quante papere hai comprato? ‘how many ducklings have you bought?’ CAM: tre ‘three’ MOT: no, due! ‘no, two’ CAM: tre ‘three’ MOT: tre, e` sempre tre! ‘three, it is always three!’

(9)

MOT: quanti limoni ha nonno laggi`u? ‘how many lemons has grandfather got there?’ CAM: tanti ‘many’

The emergence of nominal number in Italian

327

By these examples the child demonstrates that he has developed a concept of quantity and uses numerals or quantifiers to express plurality, but that he is not yet able to count (see also Stephany 1998: 3). Although mother-child interaction mainly follows fixed schemas in which nouns occur in the SG or PL in their respective contexts, it is from these contexts that the child eventually learns how to use and distinguish plural from singular forms and extend both to new contexts. Although there are no paradigms in the strict sense (see § 3) to be observed during the premorphological period, the forms cavallo/cavalli ‘horse:SG/PL’, which occur in different recordings and are context-bound, are nevertheless to be considered as precursors of the paradigmatic opposition SG/PL. As a matter of fact, the noun cavallo belongs to one of the two most frequently occurring noun classes (I.1) and both -o and -i are superstable markers expressing SG and PL, respectively.

5.1.2. Protomorphological period (I) Recordings in protomorphology I (2;3.15–2;4) only cover 55 min. as compared to 180 min. in premorphology and do not give evidence of the naming activity but mostly concern mother-child interactions during the holidays. Accordingly, they contain a much smaller number of noun lemmas and tokens (see table 6). As stated above (see §§ 4.1.1, 4.1.2) some syntactic and morphological development takes place during the period of protomorphology I such as an increasing number of two-element sentences, inflection of the verb and gender-number distinctions of adjectives and the co-occurrence of a diminutive with its base form within the same recording (e.g. orso ‘bear’ and ors-acchiotto ‘bear-DIM’ referring to the same object). If diminutives are considered as forms of their bases this may be counted as an instance of departure from the biunique relationship between lemmas and types of nouns. Otherwise, the use of nouns in a single form, either SG or PL, continues during protomorphology I so that no development with respect to number marking can be observed (see table 7).

5.1.3. Protomorphological period (II) At the age of 2;5, the onset of protomorphology II, important changes in CAM’s linguistic development are to be observed. The number of self-initiated utterances increases, sentences are more generally composed of two or three elements (see § 4.1.1) and the number of noun lemmas as well as their types and tokens increase substantially (see table 6). Most importantly for the development of inflection, the ratio of types and lemmas rises from 1.00 in protomorphology I to 1.13 (see table 7) showing

328

Sabrina Noccetti

that lemmas and grammatical types diverge. As far as the morphology of the noun is concerned, the first paradigms comprising the SG and PL forms of the same lemma emerge with the most productive inflectional classes, which occur most frequently in the input, namely class I.1 and II.1 (e.g. at 2;5 FEM Elisa/Elis-e, at 2;8 MASC giorn-o/giorn-i ‘day’). Also oppositions between base forms and DIM of nouns belonging to different inflectional classes become more frequent (e.g. treno/tren-ino ‘train:MASC:SG/-DIM:MASC:SG’, scarpe/scarpine ‘shoe:FEM:PL/-DIM:FEM:PL’, pesce/pesc-i-ol-ini ‘fish:SG/DIM:PL’)10 and a DIM adjective appears (bello/bell-ino ‘beautiful:MASC:SG/-DIM:MASC: SG’). In the context of the present paper the development of diminutives is important because they demonstrate CAM’s ability to segment the suffix from the noun base, at least as far as diminutives ending in -ino are concerned, since these have a default status in the child’s speech and his input (Noccetti et al. 2007: 142). Detection of the base of inflected forms is essential for the abstraction of generalizations on word components and their meaning. In protomorphology II the number of plural contexts increases as compared to the previous periods and the child’s reactions to both SG and PL contexts are generally appropriate (see table 8). At 2;5, the child unambiguously and creatively uses the PL form of a proper noun to refer to two pictures of a girl called Elisa (due Elitte < Elise ‘two Eliza:FEM:PL’). During the same month there is an example which shows that the child is more tuned to the noun than the article for determining gender and number of the noun phrase. Reacting to his mother’s FEM NP le rondin-i ‘the:FEM:PL swallow-FEM:PL’ he uses a MASC:SG expression, un-o piccin-o e un-o gross-o ‘one-MASC:SG smallMASC:SG and one-MASC:SG big-MASC:SG’, thus disregarding the FEM gender of the article and interpreting the noun as a MASC of class I.1 (-o/-i) rather than III.2 (-e/-i). As shown in table (9), the ending -i occurs much more frequently as a PL marker of class I.1 or III.1 nouns than of class III.2 in both the child’s speech and the input. More generally, the SG and PL forms of nouns belonging to the inflectional classes I.1 and II.1 occur much more frequently in the input so that they can be considered to be more strongly entrenched than those belonging to classes III.1 and III.2 (see table 9). A comparison of classes II.1 (-a/-e) and I.1 (-o/-i) shows that, in the input, class II.1 exceeds class I.1 typewise although class I.1 nouns are more frequent tokenwise in both the SG and PL. By and large, there is a tendency for frequent input noun classes to be still more frequent in the child’s output and for infrequent input classes to be more infrequent in the child’s speech. It is therefore not surprising that the child develops the first systematic relations of SG and PL forms with the classes showing high type frequency, namely classes I.1 and II.1 (see table 10). The reason why CAM develops more SG/PL contrasts

The emergence of nominal number in Italian

329

Table 9. Productive and unproductive noun classes in CAM and MOT (types /tokens) Inflectional class I.1 (-o/-i) II.1 (-a/-e) III.1 (-e/-i) III.2 (-e/-i) I.2 ([k]/[Ù]i) I.3 (-o/-a) I.4 (-a/-i) Irregular Uncountable

SG PL SG PL SG PL SG PL SG PL SG PL SG PL SG PL

Premorphology CAM MOT

Protomorph. I CAM MOT

Protomorph. II CAM MOT

20/54 1/1 19/44 4/11 9/39 2/2 4/6 0 1/1 0 0 0 0 0 0 0 4/6

4/6 1/1 1/2 1/1 0 2/3 0 0 0 0 0 0 0 0 0 0 1/1

49/87 16/24 40/79 10/11 16/34 5/8 2/4 1/1 1/1 0 0 1/2 0 0 1/8 1/1 7/8

27/45 8/11 21/32 4/8 9/15 4/5 2/4 0 0 0 0 0 0 0 0 0 6/9

12/14 4/6 13/17 3/4 3/20 0 0 0 0 0 1/1 0 0 0 2/2 0 1/1

75/182 20/37 83/175 22/24 27/73 13/17 8/11 2/3 1/4 1/1 2/5 3/5 0 0 3/6 3/4 20/32

Table 10. Number of paradigms in MOT and CAM per developmental period

I.1 I.1 I.2 II.1 III.1 III.2 Isol.

MOT CAM SG/PL SG/DIM:PL PL/DIM:SG SG/PL SG/DIM:PL PL/DIM:SG Premorphology 2 0 1 0 0 0 Protomorphology II 5 4 1 5 1 0 1 0 0 0 0 0 6 0 0 1 1 0 2 0 0 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0

with class I.1 than II.1 nouns in spite of the fact that both occur frequently in the input (see table 10) and II.1 nouns are even more frequent typewise (see table 9), may be explained by the fact that the gender-number markers -o/-i ‘MASC:SG/PL’ are more reliable than -a/-e since the latter signal ‘FEM:SG/PL’ in class II.1 but -e is also a SG marker with class III.1 and III.2 nouns and -a a

330

Sabrina Noccetti

PL marker with class I.3 nouns. Altogether, the months between 2;5 and 2;8 give evidence of the construction of a primitive morphological pattern, including the most frequent and transparent morphological patterns of adult grammar while excluding the less frequent, more opaque ones.

5.2.

Rosa

5.2.1. Premorphological period Similarly to CAM’s premorphological period, in ROS’ data of the same developmental stage, nouns occur in either the SG or PL. Consequently, the ratio between noun types and lemmas equals 1.00 (see table 12). The child’s plural nouns are plural dominant in adult speech and only occur in reactions to the adult’s questions anticipating gender, number and the first syllable of the requested noun (e.g. MOT: quest-i sono bi-? ‘this-MASC:PL are bi-?’ ROS: cotti ‘biscuit:MASC:PL’). Noun lemmas are rote-learned in a single default form and the child is not yet able to adapt their use to the requirement of the context so that SG nouns may occur in PL contexts and vice versa (e.g. INV: questi, come si chiamano? ‘this:MASC:PL, how are they called?’ ROS: nano < naso ‘nose:MASC:SG’). Although number is not yet marked inflectionally, ROS is able to express plurality by quantifiers (e.g. MOT: quante? ‘how many?’ ROS: ate < tante ‘many’) or numerals (due ‘two’, tre ‘three’) roughly meaning ‘more than one item’. Also, the two numerals due and tre are never used in front of SG nouns. In view of the later development of nominal inflection it is interesting to note that the most frequent nouns occurring in the child’s speech as well as in the input belong to the productive microclasses I.1 and II.1. While there are no SG/PL paradigms in the child’s speech yet, these are also rare in the input (two paradigms of class I.1 and one of II.1).

5.2.2. Protomorphological period (I) The age of 1;11 represents a landmark in Rosa’s linguistic development marking the beginning of the protomorphological period I (1;11–2;0). First of all, the nominal lexicon expands and the number of noun types and tokens grows considerably (see table 11). Although syllable reduction and some other phonetic inaccuracies still occur, noun endings are generally preserved and ROS begins to differentiate between SG and PL noun forms so that the ratio of grammatical types and lemmas rises to 1.13 (see table 12). The most frequent inflectional classes in the input (a/e and

The emergence of nominal number in Italian

331

Table 11. Noun lemmas, types and tokens (%), SG, PL and uncountable nouns (U) types/tokens (%) in ROS and MOT

ROS MOT

ROS MOT

ROS MOT

Lemmas 34 (14.4) 76 (24.8)

Types 34 (14.4) 93 (22.6)

Lemmas 43 (15.8) 89 (22.5)

Types 49 (17.5) 108 (21.2)

Lemmas 149 (11.8) 329 (32.7)

Types 208 (13.9) 449 (29.6)

Premorphology Tokens SG 122 28/95 (14.1) (82.4/77.9) 280 72/223 (13.7) (76.6/79.7) Protomorphology I Tokens SG 174 32/107 (21.2) (65.3/61.5) 350 70/240 (11.7) (64.8/68.6) Protomorphology II Tokens SG 1165 176/965 (12.3) (70.7/82.8) 1903 319/1503 (9.8) (71/79)

PL 3/4 (8.8/3.3) 16/30 (17/10.7)

U 3/23 (8.8/18.8) 6/27 (6.4/9.6)

PL 12/27 (24.5/15.5) 25/59 (23.1/16.8)

U 5/40 (10.2/23) 13/51 (12.1/14.6)

PL 55/149 (22.1/12.8) 92/277 (20.5/14.5)

U 18/51 (7.2/4.4) 38/123 (8.5/6.5)

Table 12. Ratio of grammatical types and lemmas in ROS Period Premorphology Protomorphology I Protomorphology II

ROS 1.00 1.13 1.39

MOT 1.22 1.21 1.36

o/i) tend to be also the most frequent in the child’s speech and the first to present a variation of forms. At 1;11, ROS utters her first SG/PL paradigms of classes I.1 and II.1 within the same recording, namely bimb-o/bimb-i ‘child-MASC:SG/PL’ and (ca)prett-a/ (ca)prett-e ‘goat-FEM:SG/PL’.11 Although there is some doubt whether these SG/PL forms represent true paradigms according to the criteria mentioned in § 3 (see also Kilani-Schoch and Dressler 2002: 50), the alternation of such noun forms may lead the child towards the detection of base and suffix and to the genuine opposition of SG and PL forms emerging in the second part of the protomorphological period. However, in protomorphology I, ROS more often reacts adequately to PL contexts than in the premorphological period (see table 13).

332

Sabrina Noccetti

Table 13. ROS’ reactions to SG and PL contexts contexts SG

Rosa’s reactions adequate inadequate none

PL adequate inadequate none

Premorph. 32 32 0 0 18 7 4 7

Protomorph. I 47 45 0 2 17 14 3

Protomorph. II 121 119 1 1 67 60 4 3

5.2.3. Protomorphological period (II) At the age of 2;1, the beginning of protomorphology II, the noun phrase expands with the emergence of some articles, a preposition and a clitic pronoun. Not only the nominal lexicon grows in the course of the child’s third year, but also almost one third of the nouns occur in two forms so that the ratio of grammatical types and lemmas rises to 1.39, even exceeding the input ratio (see table 12). This demonstrates the child’s greater independence of the input. While some PL noun forms found in this period continue to belong to the category of PL-dominant nouns, others belong to nouns which appeared in the SG in the earlier periods of development (e.g. penn-e ‘pen-FEM:PL’, bicchier-i ‘glass-MASC:PL’). Also, some PL-dominant nouns previously occurring in the PL now appear in the SG (e.g. gamb-a ‘leg-FEM:SG’, pied-e ‘foot-MASC:SG’). As in the preceding periods of development, the productive and transparent noun microclasses I.1 and II.1 continue to be the most frequent ones both typeand tokenwise (see table 14). While microclass III.1 ranks third, its nouns mainly occur in the SG. As far as these microclasses are concerned, the child’s ranking reflects her mother’s. Less frequent input classes such as I.2, I.3 and III.2 are either inexistent in the output or much less frequent. As is to be expected from the frequency of nouns belonging to microclasses I.1 and II.1 in the child’s data, their number of paradigms also noticeably augments during protomorphology II (see table 15). There are also various combinations of noun bases and diminutivized forms to be noted. As pointed out by Noccetti et al. (2007) diminutivized forms transferring nouns to one of the productive and transparent classes are preferred in case of opaque and unproductive noun bases (e.g. pesci-olin-o/pesci-olin-i ‘fish-DIM-MASC:SG/PL’ migrated to class I.1 from class III.1 pesc-e/pesc-i ‘fish-MASC:SG/PL’). This preference for diminutivization, which is manifest in the number of tokens from protomorphol-

The emergence of nominal number in Italian

333

Table 14. Productive and unproductive noun classes in ROS and MOT Inflectional class I.1 (-o/-i) II.1 (-a/-e) III.1 (-e/-i) III.2 (-e/-i) I.2 ([k]/[Ù]i) I.3 (-o/-a) I.4 (-a/-i) Irregular

SG PL SG PL SG PL SG PL SG PL SG PL SG PL SG PL

Uncountable

Premorphology ROS MOT 12/30 28/90 1/1 5/9 12/58 23/114 1/1 5/15 0 8/17 1/2 1/1 0 1/2 0 1/1 0 0 0 0 0 0 0 1/1 0 0 0 0 2/4 2/3 0 0 4/25 6/27

Protomorph. I ROS MOT 13/42 26/85 4/9 8/25 14/66 28/131 4/10 8/19 2/6 6/13 1/1 2/6 0 2/3 0 0 0 0 0 0 2/2 1/2 1/1 0 0 0 0 0 2/2 2/6 2/6 2/9 5/40 13/51

Protomorph. II ROS MOT 72/303 118/610 18/58 32/102 62/393 103/661 18/64 32/109 21/126 35/169 8/14 9/46 4/24 7/15 0 0 0 1/1 0 0 2/2 2/5 1/1 1/3 0 0 0 0 2/27 3/54 1/12 2/17 18/51 38/123

Table 15. Number of paradigms in MOT and ROS per developmental period MOT SG/ SG/ PL DIM:PL

PL/ DIM:SG

I.1 II.1 III.1 Isol.

2 1 0 0

0 1 0 0

0 0 0 1

I.1 II.1 Isol.

3 1 1

0 0 0

0 0 0

I.1 II.1 III.1 Isol.

9 11 3 1

2 7 4 0

1 0 0 1

ROS DIM:SG/ SG/ SG/ DIM:PL PL DIM:PL Premorphology 0 0 0 0 0 0 1 0 0 0 0 0 Protomorphology I 0 1 0 1 2 0 1 0 0 Protomorphology II 9 11 3 6 13 0 2 2 2 1 1 0

PL/ DIM:SG

DIM:SG/ DIM:PL

0 0 0 0

0 0 0 0

0 0 1

0 0 0

1 2 0 0

2 3 1 1

334

Sabrina Noccetti

ogy II onwards, indicates that ROS selects the most stable grammatical rules trying to reach maximal application (Noccetti et al. 2007:146–149). As with CAM, the number of plural contexts also increases in ROS’ data during protomorphology II as compared with the previous periods (see table 13). One reason for this is that ROS is frequently asked by her mother to count different items, which makes her produce NPs such as cocchi due ‘cock:MASC:PL two’ at 2;4. While in the previous period the main objective of mother-child interaction appears to be naming, the mother now seems to press the child towards the use of PL forms. On the other hand, as she grows, ROS produces more self-initiated utterances and the mother fine-tunes her speech to the cognitive capacities of her child (see Noccetti et al. 2007: 149). In fact, there are examples which show that by now the child is requested to produce grammatically correct utterances (example 10 from 2;5). (10)

MOT: cos’`e quell-a? ‘what is this-FEM:SG?’ ROS: caprett-e ‘goat-FEM:PL’ MOT: mh! (sound of disappointment) ROS: caprett-a piccin-a ‘goat-FEM:SG small-FEM:SG’

Another indication that the child adapts her answers to the contexts proposed by the adult and that she is able to contrast the SG and PL forms of a given lemma are her self-corrections. Thus, all inadequate productions presented in table (15) are followed by instances of self-corrections, which can be taken to show that the child is developing some form of metalinguistic awareness. Considering the numerous paradigms of nouns belonging to the inflectional microclasses I.1 and II.1 and given the examples of class shift of nouns of class III to the former classes as well as the child’s self-corrections, it can be affirmed that, in contrast to class III.1 nouns, use of I.1 and II.1 nouns becomes productive during protomorphology II.

6. Concluding remarks The main objective of the present paper has been the description of the emergence and development of nominal number marking in the early speech of two Italian children, Camillo and Rosa, showing how they move on from the use

The emergence of nominal number in Italian

335

of isolated noun lemmas to the formation of morphological patterns and functional categories. On the basis of morphological and syntactic changes in the children’s speech, their data have been divided into three developmental periods, premorphology, protomorphology I, protomorphology II. Since the category of number is grammaticized in Italian, bare roots of nominal forms do not occur in the input so that children produce inflected forms from the onset of speech production. Although the first rote-learned forms appearing in the children’s data (nouns and adjectives) either correspond to SG or PL forms of Standard Italian, their grammatical status in child language remains unclear. In the premorphological period, ROS’ and CAM’s language is characterized by extra-grammatical morphological operations, such as reduplications, vowel harmony and surface analogies. Although the children’s use of SG and PL noun and adjective forms is mainly correct, they sometimes react to the context in an inadequate way with respect to number marking. There is a tendency in both children to refer to plurality by using numerals or other quantifiers. These do not express an exact number but carry a general meaning of ‘more than one’ (or perhaps ‘a lot’). This means that children do express the concept of quantity before being able to mark it grammatically by suffixes. When, in protomorphology I, the number of two-element utterances augments, words begin to occur in more than one grammatical form. There are some differences between the two children in this period. While CAM uses the numerals ‘one’ and ‘three’, ‘six’, ‘many’ contrastively, he does not yet contrast SG and PL forms of one and the same noun, in spite of the fact that he enriches his lexical inventory with new nouns. ROS, on the other hand, already contrasts SG and PL forms of nouns belonging to the inflectional microclasses I.1 and II.1. These oppositions cannot yet be considered to form paradigms as they violate some of the necessary conditions of true paradigms (see Kilani-Schoch and Dressler 2002: 50–51). Still, they are important precursors to the grammatical contrasts emerging in the next period. In point of fact, they form patterns from which the child may detect bases and suffixes which can be extended to other lexemes. In fact, in protomorphology II, ROS even frequently operates class shifts from less frequent, more opaque noun classes to the frequent and transparent classes I.1 and II.1. Her numerous noun paradigms of these classes as well as her self-corrections are signs of productivity. CAM’s data, instead, show a gradual increase of types and tokens only in protomorphology II when he produces the first paradigms of I.1 and II.1 microclasses. Thus, despite some differences in the period of emergence, both children start their inflectional careers with the most transparent and frequent inflectional classes, overextending them to other noun classes. Overextension especially

336

Sabrina Noccetti

involves class III.1 nouns which, although less transparent, are quite frequent in both the input and output. Thus, the first examples of productivity, which emerge in protomorphology II, pertain to noun classes I.1 and II.1. While there are sparse paradigms of class III.1 and irregular nouns in ROS’ speech already in protomorphology II, III.1 noun paradigms emerge in CAM only later, during modular morphology. To conclude, both children oppose SG and PL forms of a given lemma in protomorphology II, but they can be taken to assign the status of a grammatical category to SG and PL suffixes only later in their development.

Notes ∗

1. 2. 3.

4.

5.

6. 7. 8.

I would like to thank Ursula Stephany and Pier Marco Bertinetto for commenting on earlier versions of this paper and making valuable suggestions for change. All remaining errors are, of course, my own. Rosa, the child whose data are analyzed in the present paper, is one of the normally developing children of Cipriani et al.’s study. Verb as well as article and pronoun inflection will not be examined in this paper by reason of limited space. For a different perspective where the form ‘libro’ is considered both as an inflected and an abstract word composed of the root libr- and the thematic vowel -o, which assigns the word to the class of masculines with plural in -i, see Scalise (1994: 67). A macroclass is defined as “the highest, most general type of inflectional class, which comprises several classes or (sub)classes and microclasses and whose nucleus is prototypically a productive microclass.” “A microclass is a set of those paradigms which share exactly the same morphological and morphonological generalizations” (Bittner et al. 2003: xxxix). If the word is a plurisyllabic proparoxytone its pre-final velar stop is palatalized in the plural, e.g. SG medi[k]-o ‘doctor’, PL medi[Ù]-i, otherwise the velar consonant is kept, e.g. SG fi[k]-o ‘fig’, PL fi[k]-i and SG ri[g]-o ‘line’, PL ri[g]-i. Words ending in -logo undergo palatalization in the plural only if they belong to the semantic class of animate rather than inanimate nouns, cf. astrolo[Ã]-i ‘astrologers’ vs. dialo[g]-i ‘dialogues’. The differences between Tuscan and standard Italian will be discussed in the text when necessary. Camillo’s data were collected and transcribed by Noccetti, while Rosa’s data are available in the CHILDES database. Note that diminutive suffixes are immediately attached to the root. Number and gender inflectional suffixes are external and assign the diminutivized nouns and adjectives to microclasses I.1 or II.1 (MASC in -o/-i and FEM in -a/-e). Also nouns and adjectives belonging to unproductive inflectional microclasses change the class

The emergence of nominal number in Italian

337

of the base to the most stable and most productive microclasses (e.g. MASC elefante/elefant-i ‘elephant’ > elefant-in-o/elefant-in-i ‘elephant-DIM-MASC:SG/PL’; FEM lepr-e/lepr-i ‘hare’ > lepr-in-a/lepr-in-e ‘hare-FEM:SG/PL’). For this reason, when trying to detect the development of plural marking, DIM words have been considered to form paradigms of I.1 or II.1 in the following cases: a) when they occur both in the SG and PL, independently of the class of the simplex; b) when they occur with their simplex, but only for I.1 and II.1 microclasses (e.g. barc-a/barch-in-e ‘boat-SG/boat-DIM:PL’ but not in cases such as lepr-e/leprin-e). 9. Tata refers both to sisters and adult females who are not close relatives. It is used by both children and adults. 10. Insertion of the vowel i in pesc-i-olino/pesc-i-olini is a spelling rule indicating the palatalization of sc [S], otherwise [sk], before the vowels o, a, u. 11. The noun capr-ett-a is the DIM of capr-a, but the base neither occurs in the input data nor in the child’s speech.

References Bitter, Dagmar, Wolfgang U. Dressler, and Marianne Kilani-Schoch (eds.) 2003 Development of Verb Inflection in First Language Acquisition: A Cross-Linguistic Perspective. Berlin/New York: Mouton de Gruyter. Calleri, Daniela, Marina Chini, Patrizia Cordin, and Stefania Ferraris 2003 Confronti tra l’acquisizione di italiano L1 e l’acquisizione di italiano L2. In Verso L’Italiano, Anna Giacalone Ramat (a cura di), 220–253. Roma: Carocci. Celata, Chiara, and Pier Marco Bertinetto 2005 Lexical access in Italian: Words with and without palatalization. Lingue e Linguaggio 4 (2): 293–318. Cipriani, Paola, Anna Maria Chilosi, Piero Bottari, and Lucia Pfanner 1993 L’acquisizione della morfosintassi in Italiano: Fasi e processi. Padova: Unipress. Corbett, Greville G. 2000 Number. Cambridge: Cambridge University Press. Dardano, Maurizio, and Pietro Trifone 1990 La lingua italiana. Bologna: Zanichelli. 1992 Grammatica italiana. Bologna: Zanichelli. Devescovi, Antonella, and Simonetta D’Amico 2001 Lo sviluppo della morfosintassi. In Psicologia dello sviluppo del linguaggio, Luigia Camaioni (ed.), 117–152. Bologna: Il Mulino. Dressler,Wolfgang U. 1985 Sur le statut de la suppl´eance dans la morphologie naturelle. Langages 78: 41–56.

338

Sabrina Noccetti 2003

Degrees of grammatical productivity in inflectional morphology. Rivista di Linguistica 15 (1): 31–62. 2005 Morphological typology and first language acquisition: Some mutual challenges. In Morphology and Linguistic Typology: On-line Proceedings of the Fourth Mediterranean Morphology Meeting (MMM4) Catania 21–23, September 2003, Geert Booij, Emiliano Guevara, Angela Ralli, Salvatore Sgroi, and Sergio Scalise (eds.), 7–20. University of Bologna. URL. http://morbo.lingue.unibo.it/mmm/. Dressler, Wolfgang U. (ed.) 1995 Studies on Pre- and Protomorphology. Vienna: Austrian Academy of Sciences. Dressler,Wolfgang U., and Annemarie Karpf 1995 The theoretical relevance of pre- and protomorphology in language acquisition. Yearbook of Morphology 1994, 99–122. Dressler, Wolfgang U., and Anna M. Thornton 1996 Italian nominal inflection. Wiener Linguistische Gazette 55-57: 1–24. Gillis, Steven (ed.) 1998 Studies in the Acquisition of Number and Diminutive Marking. (Antwerp Papers in Linguistics 95.) Antwerp: Universiteit Antwerpen. Karpf, Annemarie 1991 Universal grammar needs organization. Folia Linguistica 25 (3-4): 339–360. Kilani-Schoch, Marianne 1998 The acquisition of number in French. In Studies in the Acquisition of Number and Diminutive Marking, Steven Gillis (ed.), 25–49. (Antwerp Papers in Linguistics 95.)Antwerp: UniversiteitAntwerpen. Kilani-Schoch, Marianne, Anna De Marco, Anastasia Christofidou, Maria Vassilakou, Ralf Vollmann, and Wolfgang U. Dressler 1997 On the demarcation of phases in early morphology acquisition in four languages. Papers and Studies in Contrastive Linguistics 33: 15–32. Kilani-Schoch, Marianne, and Wolfgang U. Dressler 2002 The emergence of inflectional paradigms in two French corpora: An illustration of general problems of pre- and protomorphology. In Preand Protomorphology: Early Phases of Morphological Development in Nouns and Verbs, Maria D. Voeikova and Wolfgang U. Dressler (eds.), 45–59. M¨unchen: Lincom Europa. MacWhinney, Brian 2000 The CHILDES Project: Tools for Analyzing Talk. 3rd ed. Mahwah, NJ: Lawrence Erlbaum. Marcantonio, Angela, and Annamaria Pretto 1991 Il nome. In Grande Grammatica Italiana di Consultazione, Lorenzo Renzi (ed.), vol. 1, 315–503. Bologna: Il Mulino.

The emergence of nominal number in Italian

339

Noccetti, Sabrina 2002 Pre- and Protomorphology in Language Acquisition: An Italian Case Study. Pisa: Edizioni Plus Universit`a di Pisa. Noccetti, Sabrina, Anna De Marco, Livia Tonelli, and Wolfgang U. Dressler 2007 The role of diminutives in the acquisition of Italian morphology. In The Acquisition of Diminutives. A Cross-linguistic Perspective, Ineta Savickien¨e and Wolfgang U. Dressler (eds.), 125–153. Amsterdam: John Benjamins. Peters, Ann M. 1983 The Units of Language Acquisition. New York: Cambridge University Press. Pizzuto, Elena, and Maria Cristina Caselli 1993 L’acquisizione della morfologia flessiva nel linguaggio spontaneo: Evidenza per modelli innatisti o cognitivisti. In Ricerche sull’Acquisizione dell’Italiano, Emanuela Cresti and Massimo Moneglia (eds.), 165–187. Roma: Bulzoni. Rohlfs, Gerhard 1968 Grammatica storica della lingua italiana e dei suoi dialetti. Morfologia. Torino: Einaudi. Savickien˙e, Ineta, and Wolfgang U. Dressler 2007 Introduction. In The Acquisition of Diminutives. A Cross-linguistic Perspective, Ineta Savickien¨e and Wolfgang U. Dressler (eds.), 1–12. Amsterdam: John Benjamins. Scalise, Sergio 1994 Morfologia. Bologna: Il Mulino. Sedlak, Maria, Sabine Klampfer, Brigitta M¨uller, and Wolfgang U. Dressler 1998 The acquisition of number in Austrian German. In Studies in the Acquisition of Number and Diminutive Marking, Steven Gillis (ed.), 51–76. (Antwerp Papers in Linguistics 95.)Antwerp: UniversiteitAntwerpen. Stephany, Ursula 1998 A crosslinguistic perspective on the category of nominal number and its acquisition. In Studies in the Acquisition of Number and Diminutive Marking, Steven Gillis (ed.), 1–23. (Antwerp Papers in Linguistics 95.) Antwerp: Universiteit Antwerpen. 2002 Early development of grammatical number: A typological perspective. In Pre- and Protomorphology: Early Phases of Morphological Development in Nouns and Verbs, Maria D. Voeikova and Wolfgang U. Dressler (eds.), 7–23. M¨unchen: Lincom Europa. Thornton, Anna 2005 Morfologia. Roma: Carocci Editore. Tonelli, Livia, Anna De Marco, Ralf Vollmann, and Wolfgang U. Dressler 1998 Le prime fasi dell’acquisizione della morfologia: Un confronto tra l’italiano e il tedesco. In Parallela 6. Italiano e Tedesco a Contatto

340

Sabrina Noccetti

e Confronto. Atti dell’Incontro Italo-Austriaco dei Linguisti, Patrizia Cordin, Maria Iliescu, and Heidi Siller-Runggaldier (eds.), 281–301. Trento: Universit`a di Trento. Voeikova, Maria D., and Wolfgang U. Dressler (eds.) 2002 Pre- and Protomorphology: Early Phases of Morphological Development in Nouns and Verbs. M¨unchen: Lincom Europa. Wiese, Heike 2003 Numbers, Language, and the Human Mind. Cambridge: Cambridge University Press.

Number morphology in Spanish first language acquisition Carmen Aguirre and Victoria Marrero 1. Introduction In a previous paper (Marrero and Aguirre 2003), we tackled the emergence of the plural in nominal and verbal phrases in Spanish. We then focused our qualitative analysis on the early stages of development (up to 2;0). In this paper, we have extended our analyses to 2;6, with a more in-depth study of the latter stages of the process in the nominal domain (noun, pronoun, adjective, determiners), in verbal forms and their agreement. Our general theoretical position will be framed within the account of pre- and protomorphology. This approach, based on the model of Natural Morphology (Dressler et al. 1987) and self-organization of developing systems (Dressler and Karpf 1995), defends the influence of morphological richness and language typology in the morphology acquisition process. Dressler and Karpf (1995) divide morphological development into three main phases: premorphology, protomorphology, and morphology proper. In the premorphological phase, no system of grammatical morphology has begun to develop and only rote-learned morphological forms appear. In the protomorphological phase, the child detects morphological patterns and the paradigms begin to develop. In the last phase, morphology proper, the child’s system approaches the adult model. In our study, three major phases have been established: scarce plural forms with no contrastive plural use in premorphology; first plurals used in a contrastive way, with a single marker in the entire utterance and missing agreement, in the protomorphological stage; and the extension of plural marks, with paradigms development and emergence of agreement, in the morphological stage properly speaking. We consider that the concept of plurality is semantically prototypically associated with the noun. Only the noun referent can appear as “more than one”, causing the addition of the plural markers. We therefore assume that plural markers in nouns are primary, whilst plural markers in other elements such as adjectives or verbs are secondary, and normally appear as a consequence of agreement. For this reason, if only one element in the sentence has a plural marker, we expect it to be the noun. The determiner could also be a prime can-

342

Carmen Aguirre and Victoria Marrero

didate to mark plurality because its role is to present and indicate the scope of reference of the noun. Other elements besides the noun can mark plurality in the sentence, but the reason is always syntactic – agreement – rather than semantic. Based on these considerations, we predict that the child will begin to identify and produce plural suffixes in nouns and determiners first, and only later in adjectives and the verb. At a later stage the child is stimulated – by the input received, by his further analysis of input data (see White 1982 and Newport 1990) and by his or her own quest for accuracy – to produce more than one marker to express plurality. Our hypothesis is that the use of these redundant and semantically unnecessary markers forces the emergence of agreement as a syntactic mechanism for constituent identification. The child begins to use plural morphemes that are not strictly necessary to establish plurality meaning, provoking a search for another “reason”, a grammatical one. Agreement is present to identify and establish constituents.

2. The grammatical category of number in Spanish The category of number is marked on Spanish nouns, adjectives, determiners, pronouns and verbs, in a highly redundant and regular way, establishing a crossreferential network of agreement, both nominal and verbal. There are two numbers,1 singular and plural. The plural is overtly marked on nouns by one suffix with two allomorphs: –s or –es. This morphophonemic variation is phonologically conditioned: The paradigm ∅2 / –s is used when the noun ends in a vowel: casa / casas3 ‘house/s’. The paradigm ∅ / –es is used when it ends in a consonant: flor / flores ‘flower/s’. Determiners and adjectives have the same plural suffixes (–s or –es), with the exception of the masculine articles that have special forms with stem alterations: el / los ‘the: SG/PL’ - un / unos ‘a/some’. There are also some determiners which contain inherent lexical plurality, such as numerals (zero and one are exceptions), and some indefinite quantifiers (mucho ‘much’, m´as ‘more’, etc.). Spanish verbs have three persons in the singular and three in the plural. Number is fused with person in verbal suffixes (cant-o, canta-s, canta-∅, cantamos, cant´a-is, canta-n, ‘sing:PRES’).

Number morphology in Spanish first language acquisition

343

Gender-number agreement is obligatory in Spanish in the nominal phrase, between the head noun and the determiners and adjectives (ex. 1). Gendernumber agreement is also required between the noun and predicative adjective. Number agreement is a required between subject and verb (ex. 1 and 2). (1)

Unas ni˜nas preciosas juegan some:FEM:PL girl:FEM:PL beautiful:FEM:PL play:3P ‘Some beautiful girls are playing.’

(2)

Esas ni˜nas son preciosas these:FEM:PL girl:FEM:PL are:3P beautiful:FEM:PL ‘These girls are beautiful’

3. Data The present paper is based on the analysis of spontaneous speech drawn from the longitudinal corpora of two monolingual Spanish children recorded in everyday situations: the girl Mar´ıa (L´opez Ornat 1994) and the boy Mag´ın (Aguirre 1995), both natives of Madrid. Mag´ın’s corpus begins at age 1;7 and Mar´ıa’s at 1;8. Both corpora have been analyzed through the age of 2;6. The data are transcribed and encoded in CHAT and can be found in the CHILDES database (MacWhinney 2000). The main interlocutors in both corpora are the parents together with the children. The input data has been taken into consideration in the analysis. Table 1. Speech samples Age Mar´ıa’s MLU Number of Mar´ıa’s utterances analyzed 1;7 – 1;8 1.9 280 1;9 2.0 601 1;10 1.8 745 1;11 2.0 322 2;0 2.4 351 2;1 2.4 630 2;2 2.7 569 2;3 3.6 454 2;4 3.4 293 2;5 4.2 436 2;6 3.7 272

Mag´ın’s MLU Number of Mag´ın’s utterances analyzed 1.2 183 1.4 623 1.7 890 2.0 1736 2.2 755 2.3 711 2.2 396 2.4 1163 2.5 1013 2.7 1084 2.8 599 3.0 1155

344

Carmen Aguirre and Victoria Marrero

4. Stages of plural development 4.1. The premorphological stage At the outset the two children use the singular in contexts where the target is plural. The following example shows how Mar´ıa is using the singular form oso ‘bear’, and the filler a instead of the appropriate plural form los when the context requires plural. (3)

Mar´ıa 1;8 MOT: d´onde est´an los osos? ‘Where are the:MASC:PL bear:MASC:PL?’ MAR: a soso o t´a [: no est´a] FILLER bear not is ‘The bear isn’t here.’

We find some plural forms in nouns, in Mar´ıa, and articles, mostly in Mag´ın, (see table 2 in 4.2.1.), however they are scarce and restricted to a few specific words that usually appear in the plural form. These forms are unanalyzed, in the sense that there is no real singular/plural contrast. The use of these non-contrastive plurals is highlighted by the lack of agreement (even in gender) with the singular determiners, or the copula. These forms are nouns in Mar´ıa: “obibis” for the Spanish word mu˜necos ‘dolls’, botas ‘boots’, and apatos (zapatos ‘shoes’), all of them with the plural s marker. (4)

Mar´ıa 1;8 MAR: a obibis [: mu˜necos]; a bibis. ‘FILLER dolls’ MAR: a bibi. ‘FILLER doll’

In example number 4, the girl is asking for the dolls in her room. She uses the singular or plural form no matter what the context is, as two variants of the same word. The only plural form in Mag´ın at this stage is the definite plural masculine article los, expressed by an o, and used always in plural-required contexts. Mag´ın uses this plural masculine determiner los, but with phonetic variation ([po], [no], [o]) and restricted to two nouns: zapato ‘shoe’ and nene ‘boy’. Phonological harmony in consonants explain the selected form: po used many times with zapato (pronounced papapo) and no with nene.

Number morphology in Spanish first language acquisition

(5)

Mag´ın 1;8 po [: los] papapo [: zapato]. ‘the:PL shoe’

(6)

Mag´ın 1;8 no [: los] nene. ‘the:PL boy’

345

In most cases the plural forms are nominal, although Mag´ın presents some verbal examples: (7)

Mag´ın 1;8 no (es)t´an. ‘They’re not there.’

He uses no tan, instead of the adult form no estan (no tan e pipi ‘The bird aren’t there’) and be van, instead of the adult form se van (be van, avi´on ‘they’re leaving, plane’). These two forms are fixed structures that the child uses to express non-existence. Nearly all these occurrences of third person plural are agreement errors. These agreement mistakes are understood in the sense that they present formally a verb in the plural and a subject in the singular, but we cannot speak properly about agreement in this period, since this plural form does not have a plural intention. These utterances seem to be pivot constructions (in the sense of Braine 1963) with not´an and bevan being the pivot and e pipi or avi´on being the fillers of the frame created by the pivot. By this time the child also uses the singular forms no est´a and se va, but the choice of one or the other is still unmotivated. They seem to be phonological variants.

4.2. The protomorphological stage The most outstanding feature of this stage is that the child marks the plural in only one place out of all possible points in the utterance (usually two or three items: determiner, noun and verb). As we have already seen, the plural marker is a highly redundant morpheme, from a semantic point of view. It not only appears in nouns, but also in determiners, adjectives and verbs due to agreement constraints, although the semantic meaning could be conveyed by a single marker. Initially, children express plurality by marking just one word with the plural suffix. They only seem to aim at communicating the plural meaning and need only one word in the sentence for this purpose. Agreement is a syntactic requirement that the child has not acquired at this stage so that morphology has a semantic value and not yet a syntactic one at this stage. Mag´ın and Mar´ıa begin to contrast the plural to the singular form of nouns with the value of “more than one” at 1;9 (see the following examples and table 2).

346

Carmen Aguirre and Victoria Marrero

(8)

MAR 1;9 a. caca a [: en] manos ‘dirt on hands’ b. o est´a [: d´onde est´a] mano? ‘where is hand?’

(9)

MAR 1;9 a. Este poqu´ı [: zapato] no. ‘not this shoe’ b. E(s)te s´ı poqu´ıs [: zapatos]. ‘this:MASC:SG shoes:MASC:PL OK’

(10)

MAG: 1;9 a. MAG: b. MAG: c. MAG: d. MAG: e. MAG: f. MAG:

lo(s) nene. ‘the:MASC:PL boy:MASC:SG’ na nena . ‘the girl’ lo(s) bobo [: globo]. ‘the: MASC:PL balloon:MASC:SG’ un bobo [: globo]. ‘a:MASC:SG balloon:MASC:SG’ mira pece(s). ‘look at fish:MASC:PL’ e(l) pe(z). ‘the:MASC:SG fish:MASC:SG’

At the onset of plural acquisition, Mar´ıa uses the plural marker in nouns with the most common form in adult speech: ∅/–s. The following alternations between singular and plural to convey singular vs. plural meaning occur at 1;9 and 1;10: a mano / a manos ‘hand/s’; e mimo / mimos ‘caress/es’; ete poqu´ı / ete poqu´ıs ‘shoe/s’. Mag´ın’s use of plural markers in nouns, however, is restricted to the ∅/–es paradigm.4 This is more salient than the more common ∅/–s form and children like Mag´ın who usually elide /s/ in final syllable position5 may use /e/ as a plural morpheme. His first clear plural in nouns is pece ‘fish-pl’ at 1;9, followed by luce ‘lights’ and balone ‘balls’, at 1;10, and calcetine ‘socks’ and azule ‘blue:PL’, at 1;11. All of them are used in a productive manner, together with their respective singular forms: pez, luz, bal´on, calcet´ın and azul. A highly productive mechanism is marking plurality by means of the article:6 e(l), vs. (l)o(s): e huevo / o huevo ‘egg’. (11)

MAR 1;9 caca o pes [: en los pies] ‘dirt on the:PL feet’7

We understand productivity in the sense that the child is able to use the e(l)SG vs. (l)o(s)- PL to convey singular vs. plural meaning with any noun. The e/o contrast is phonetically easiest to perceive and to produce (in comparison with the ∅/–s contrast). However, it has some serious limitations: it cannot be applied to feminine referents, where the articles are la/las; nor does it work in contexts where the noun needs no determiner.

Number morphology in Spanish first language acquisition

347

We do not find any clear type of numeral determiners as plural predecessors or alternative plural devices used to refer to a plurality of entities, as reported by Kilani-Schoch (1998) for French and Sedlak et al. (1998) for Austrian German.8 Mag´ın uses the determiner otro ‘another’ to express “recurrence”: (12)

MAG 1;9 la rama. ‘the:FEM branch’ ot(r)a la rama. ‘another:FEM the:FEM branch’

In certain situations its meaning approaches the concept of plurality, it is a kind of plurality predecessor, when the child, looking at a set of elements, points to them one-by-one placing the determiner otro in front of every noun. Nevertheless, the pronounced use of otro (78 items at 1;9) coincides with the first plural nouns and articles, as we have already seen. (13)

MAG 1;9 MOT: mira, Mag´ın, ha visto estos caracoles peque˜nitos? ‘Have you seen these little snails?’ MAG: tat´on [: caracol]. ‘Snail’ MAG: ot(r)o tat´on [: caracol]. ‘Another snail’

4.2.1. The development of the plural in nominal categories The development of plural paradigms within each inflectional category invariably follows a similar pattern. The plural form emerges when the singular has already been acquired and is used quite often. There are some nouns, like zapatos ‘shoes’, that are used very often in the input in its plural form, but the child is not able to produce the -s plural marker and use this noun in its singular form: zapato /pap´apo/. In the following table we can see the first plural forms that appear in nominal categories in the protomorphological stage (1;9–1;11). The onset of the plural in nominal categories begins with the noun, followed very closely by the article. As we have seen, the child is prone to use only one single plural marker at this stage (mostly in the noun or in the article, and sometimes in the verb), but we also notice the first agreements (1;9–1;10): (14)

MAR 1;10: (l)os [: los] dientes. ‘the:MASC:PL teeth’

(15)

MAR 1;10: (l)as u˜nas. ‘the:FEM:PL nails’

(16)

MAG 1;11: mam´a, los azule(s). ‘mummy, the:MASC:PL blue:PL’

(17)

MAG 1;11: las f(l)ores. ‘the:FEM:PL flowers’

348

Carmen Aguirre and Victoria Marrero

Table 2. Singular /plural in nominal group (tokens), up to 1;11 Mag´ın Age

Mar´ıa

1;7 1;8 1;9 1;10 1;11 Age 1;8 1;9 1;10 1;11

Nouns sg pl 63 0 126 0 452 1 881 11 301 4 Nouns sg pl 130 710 315 37 280 25 125 16

Adjectives Articles sg pl sg pl 0 0 42 0 3 0 25 15 21 0 135 11 55 3 285 5 107 9 1009 1 Adjectives Articles sg pl sg pl 4 0 27 1 30 1 56 2 36 3 90 4 17 0 45 3

Dem/Poss/Indef sg pl 23 0 20 0 39 0 91 0 64 5 Dem/Poss/Indef sg pl 28 1 57 3 119 5 61 0

Pers. Pron. sg pl 0 0 2 0 3 0 44 0 55 4 Pers. Pron. sg pl 0 0 8 0 21 3 17 0

The first plural forms of adjectives appear at this period (1;9–1;11) following the same pattern as nouns, but they are still scarce usually, as we can see in table 2. (18)

MAG 1;10: qu´e sucios. ‘how dirty:MASC:PL’ (speaking about her feet)

(19)

MAR 1;9 MAR: mam´a tonto no. ‘Mummy dumb:MASC:SG not’ MOT: qui´en? ‘Who?’ MAR: mam´a tontos. ‘Mummy dumb:MASC:PL’

At the end of this protomorphological stage the first plural determiners (demonstratives and possessives) and pronouns appear. (20)

MAG 1;11 MAG: estos no. ‘not these:MASC:PL’ MAG: no los quiero. ‘not them:MASC:PL I want’

We can conclude that the acquisition order of plural paradigms in nominal categories is: nouns, articles, adjectives, pronouns and other determiners besides articles. Each category closely follows the previous one. As soon as the singular/plural morphological mechanism is acquired, it is easily developed in all the nominal categories. The fact that all these categories have the same plural suffixes, with the exception of the masculine articles, facilitates this task.

Number morphology in Spanish first language acquisition

349

4.2.2. Gender agreement versus number agreement In spite of the fact that the development of gender distinctions is not one of our main concerns in the present paper, our data reveal that gender emerges together with number, to which it is closely tied in the structure of Spanish. At 1;8, Mar´ıa uses a few feminine animate nouns (no more than ten), person or animal nouns with possible contrast with the masculine. Nevertheless, we do not find any contrast. Around 1;9 for Mag´ın and for Mar´ıa, we find the first gender contrasts in nouns and adjectives appearing sometimes in the same lemma.11 (21)

MAG 1;8 MAG: a nena. ‘the girl’ MAG: no nene. ‘the:PL boy:MASC:SG’

(22)

MAR 1;9 MAR: esta, esta flol [: flor]. this:FEM flower’ MAR: e(s)te pato? this:MASC duck’

(23)

MAR 1;9 MAR: mam´a m´a [: mala]. ‘Mom bad’ MAR: malo, malo (approaching the dog) ‘bad:MASC’

At the protomorphological stage (1;9–1;11), we find some adjectives used in both genders, masculine and feminine, for the same lemma. Mag´ın uses at 1;9: bonito / bonita ‘beautiful:MASC / beautiful: FEM’ and fr´ıo / fr´ıa ‘cold:MASC / cold: FEM’. At 1;10 he adds pobrecito / pobrecita ‘poor:MASC:DIM / poor:FEM: DIM’ and tonto / tonta ‘silly:MASC / silly: FEM’. Nevertheless, the child produces many errors, even when the adult tries to provoke a self-correction, as in the following sequences: (24)

MAR 1;9 FAT: y pap´a c´omo est´a? ‘And daddy, what’s he like?’ MAR: p´aapa [: guapa]. ‘good-looking:FEM’ FAT: pap´a guapo tambi´en. ‘Dad (is) good-looking:MASC too.’ MAR: no. ‘No.’ FAT: c´omo est´a pap´a? ‘What’s daddy like?’ MAR: p´aapa [: guapa]. ‘good-looking:FEM’ FAT: pap´a guapa? ‘Dad good-looking:FEM?’ MAR: noo. ‘No.’

350

Carmen Aguirre and Victoria Marrero

(25)

MAG 1;9 MAG: la nena. ‘the:SG girl’ MOT: esos son los nenes. ‘Those are the boys.’ MOT: las nenas no. ‘Not the girls.’

(26)

MAG 1;10 MAG: aqu´ı no. ‘not here’ MAG: tonta. ‘silly:FEM’ (addressing his father)

Sometimes the child uses adjectives with the same noun and in the same context both in the masculine and feminine form, showing his lack of confidence in gender agreement: (27)

MAG 1;9 MAG: ot(r)o flor. ‘another:MASC flower:FEM’ MOT: otra flor. ‘another:FEM flower:FEM’ MAG: (bo)nita. ‘beautiful:FEM’ MOT: anda, ven. ‘Come.’ MAG: (b)o(n)ito. ‘beautiful:MASC’ MOT: qu´e dices? ‘What do you say?’ MAG: bonita. ‘beautiful:FEM’

(28)

MAG 1;9 MAG: (es)t´a fr´ıa. ‘it is cold:FEM’ MAG: fr´ıa. ‘cold:FEM’ MAG: fr´ıo! ‘cold:MASC’ MAG: fr´ıa. ‘cold:FEM’

Over the subsequent months, the number of gender contrasts increases, but some errors still remain. Comparing the development of gender with number, we do not find any clear evidence for gender development preceding that of number.

4.2.3. The onset of plural verb forms Plural acquisition in verbs shows some delay relative to nouns and nominal categories (see table 3 in comparison with table 2). At the protomorphological stage, only Mag´ın shows a certain degree of productivity at 1;10. At this age he uses the first plural occurrences with singular/plural alternations. These forms begin to be productive in the 3rd person plural (son and est´an ‘are-3p PL’, queman ‘burn-3p PL’, mojan ‘wet-3p PL’) and in the 1st person plural (guardamos ‘keep1p PL’, abrimos ‘open-1p PL’, arropamos ‘wrap up-1p PL’, cogemos ‘take-

Number morphology in Spanish first language acquisition

351

Table 3. Singular/plural in verbs from 1;7 to 1;11 (number of tokens per month, without repetitions and frozen forms).13 Mar´ıa

Age

1;8 1;9 1;10 1;11 Mag´ın Age

1;7 1;8 1;9 1;10 1;11

Present Singular 1st 2nd 1 9 2 9 9 12 1 Present Singular 1st 2nd 1 4 27 45 1 23 6

3rd 21 77 41 28

3rd 25 62 113 299 109

Plural 1st 2nd

3rd 1

Past Singular 1st 2nd

9 2 Plural 1st 2nd

Plural 1st 2nd

3rd

Plural 1st 2nd

3rd

6 1

3rd

Past Singular 1st 2nd

3rd

2 3

1 22 18 46

1 6 8

3rd

9 3

3

1P’, tiramos ‘throw-1P’). The child uses the 1st person plural mostly with an exhortative function. He makes a semantic analogy with the much used vamos12 ‘go-1p PL’ (equivalent to let’s go) that has always an imperative meaning, not opposed to any singular form.

4.3. The morphological stage Around 2;0, Mag´ın and Mar´ıa show a dramatic increase in the use of plural morphology. The morphological singular/plural mechanism is largely mastered in nouns and articles and begins to be consistently applied in adjectives, demonstratives (estos ‘these’), quantifiers (muchos/as ‘many’), possessives (m´ıos ‘my/mine’, tuyos ‘your/yours’) and, one month later, even numerals (tres ‘three’). The three latter categories can appear as determiners or pronouns. The much more complex system of Spanish emphatic and clitic personal pronouns is more difficult to master. Pronouns distinguish not only between three persons in the singular and plural but also between different case forms. The following table shows the development of the plural in all nominal categories, from 2;0 until 2;6. – In general, the number of singular contexts is eight times more frequent than plural ones (88%–12%), following the common pattern of Spanish,

352

Carmen Aguirre and Victoria Marrero

Table 4. Singular/plural in nominals (tokens) from 2;0 to 2;6 Mag´ın Age

Mar´ıa

2;0 2;1 2;2 2;3 2;4 2;5 2;6 Age 2;0 2;1 2;2 2;3 2;4 2;5 2;6

Nouns Adjectives Articles sg pl sg pl sg pl 313 38 28 2 189 36 73 12 17 0 50 6 335 50 55 4 206 35 289 85 44 0 184 34 338 65 91 11 195 29 130 19 41 5 116 11 347 43 54 7 267 19 Nouns Adjectives Articles sg pl sg pl sg pl 146 31 28 1 51 10 188 33 39 11 83 14 190 38 20 9 103 13 198 33 26 5 121 25 107 29 25 4 65 14 245 45 36 7 137 33 141 20 19 0 80 11

Dem/Poss/Indf sg pl 20 5 15 0 65 4 62 15 65 16 26 1 57 4 Dem/Poss/Indf sg pl 54 17 52 15 62 10 92 14 45 4 53 20 92 15

Pers. Pron. sg pl 56 3 50 1 118 1 158 1 207 3 79 2 239 2 Pers. Pron. sg pl 27 0 143 0 123 2 120 3 68 0 113 8 65 0

with singular as the unmarked form for number (for considerations on the relationship between adult input and children’s speech on this respect, cf. Marrero and Aguirre 2001). – Personal pronouns show a very different situation from the other nominal categories: with less than 2% use of plurals. Our sample offers very few examples, mostly limited to the third person of the clitic pronouns los ‘themMASC-PL/las ‘them-FEM-PL’/les ’them-NEUT-PL’. At the end of sample (2;6) they are not productive at all. – With the other nominal categories the ranking of the use of plural forms is as follows: demonstratives, possessives, indefinites 15.5%, nouns 15.1%, articles 13.6%, adjectives 11.2%. Along the seven months, oscillations in frequency can be appreciated, allied to communicative characteristics of the sample, once the morphological device is consistently used; that is, in several types and many tokens of different word classes, with a clear contrastive value with the singular. – Considering inter-subject variation, Mar´ıa uses more plural nominal forms than Magin in almost all categories, but especially in adjectives (8%–16%). In verbal morphology the situation is quite different. The predominance of singular forms is larger than in adult Spanish: 92.1%. In fact, until the end of the

Number morphology in Spanish first language acquisition

353

Nominal categories. María 50 40 30 20 10 0 1

2

3

4

Nouns

Adjectives

Dem/Poss/indf

Pers. Pron.

5

6

7

Articles

Figure 1. Nominal categories (2;0–2;7). Mar´ıa

Nominal categories. Magín 100 80 60 40 20 0 1

2

3

4

5

6

7

Figure 2. Nominal categories (2;0–2;7). Mag´ın

second year they are almost the only ones used. The detailed distribution can be observed in table 5. From the first moment we find some uses of plurals in present tense in the third person, many of the plural utterances correspond to copula forms of ser and estar (‘to be’) in contrastive use with singulars. We find also specific uses of past tense, in contrasts like these, where the form of plural does not correspond to a plural referent: (29)

MAR 2;0 se cay´o e(l) patito, fall down:PAST:3S the duck:DIM:MASC:SG se rompieron. (humming) break:PAST:3P ‘The little duck fell down, they are broken.’

354

Carmen Aguirre and Victoria Marrero

Table 5. Singular/plural in verbs from 2;0 to 2;6 Mar´ıa

Age

2;0 2;1 2;2 2;3 2;4 2;5 2;6 Mag´ın Age

2;0 2;1 2;2 2;3 2;4 2;5 2;6

Present Singular 2nd 1st 33 16 68 23 109 11 107 27 47 11 105 33 38 17 Present Singular 2nd 1st 72 6 42 8 75 40 65 31 115 26 56 14 138 35

3rd 65 63 55 101 58 88 91

3rd 140 68 220 216 197 170 257

Plural 1st 2nd

1 2 7 4 6

3rd 5 5 14 17 8 16 18

Plural 1st 2nd 5 – 10 2 11 1 17

3rd 20 7 14 25 29 15 24

Past Singular 1st 2nd

1

3rd 3 7 7 9 6 12

1 Past Singular 1st 2nd 1 3 2 1 6 1 12 3 11 7 1 2 18 3

3rd 9 17 44 45 59 56 67

1 1 1

Plural 1st 2nd

Plural 1st 2nd

3rd 1 3

3rd 1

1

2 2 6

5

2

After 2;2 the use of the third person plural in the present tense increases (around 6% of verbal utterances, but 75–80% of plural verbal forms), and, to a lesser extent, the first person also (1.5–2% of the whole). The second person plural is not used. (30)

MAR 2;2 a.

b.

no (l)e gusta e(l) not it please:PRES:3S the:MASC:SG Pipo. Pipo ‘Pipo doesn’t like the school.’ a m´ı me gu(s)tan las to me me please:PRES:3P the:FEM:PL pirujas. wacky:FEM:PL ‘I like wacky witches.’

cole a school:MASC to

b(r)ujas witch:FEM:PL

First person (nosotros/as ‘we’) emerges next,14 also in the present tense, with contrasts such as the following:

Number morphology in Spanish first language acquisition

(31)

MAR 2;5 a. mam´a, ven aqu´ı, que te hago una pel´ıcula! ‘Mom, come here, so I can make:1S a film of you’ b. y hasemos [: hacemos] una pel´ıcula, a que s´ı? ‘And we make:1P a film, right?

(32)

MAR 2;6 a. vamos a coger uno, uno m´as g(r)ande. go:PRES:1P to take one, one more big ‘We’re going to take one, a bigger one.’ b. voy a sacar-los. go:PRES:1S to take out-them ‘I’m going to take them out.’

355

Both children show very similar data in use frequency of plural verbal forms: 7,8% (the boy) – 8% (the girl). But Mag´ın generalizes plural in past tense more than Mar´ıa. To conclude, we consider that children are in the morphological stage at this moment, because the plural is frequently used creatively and contrastively in many different contexts and consistently across various word classes. But the ongoing developmental process is at a crossroads, with person-number (in verbs and pronouns) posing problems.

4.3.1. The use of multiple plural markers and the development of agreement The primary characteristic of this stage is the development of agreement, possibly as grammatical means to structure a message that is becoming increasingly complex.15 There is a relationship between the increase in usage of the plural, the appearance of multiple inflectional markers in the utterance, and, as a consequence, the emergence of agreement. This connection can be visualized in the following charts, which represent the increase in nominal plural usage16 and the emergence of multiple markers for Mar´ıa (fig. 3) and Mag´ın (fig. 4). In previous stages (1;9–1;11), the scarce examples of pre-agreement are all limited to the noun phrase, i.e. to agreement of the determiner with the noun: (33)

a.

MAR 1;9 caca (l)o(s) p(i)es [: en los pies]. mess the:MASC:PL foot:MASC:PL

356

Carmen Aguirre and Victoria Marrero

80 70 60 50 40 30 20 10 0 María

1;9

1;10

1;11

2;0

Total of plural utterances

2;1

2;2

Multiple markers

2;3

2;4

2;5

2;6

Multiple markers %

Figure 3. Rate of multiple markers. Mar´ıa 100 90 80 70 60 50 40 30 20 10 0 Magín

1;9

1;10

1;11

2;0

Total of plural utterances

2;1

2;2

Multiple markers

2;3

2;4

2;5

2;6

Multiple markers %

Figure 4. Rate of multiple markers. Mag´ın

b.

MAR 1;11 se vaya(n) (l)as mojcas [: las moscas]. go the:FEM:PL flies Voy a sacar-los. go:1S to take.away-them:MASC:PL ‘The flies go. I’m going to take them away.’

With consistent use of the plural (cf. above) more than one marker per utterance appears and easily reaches percentages of 60%17 (range: 40%–70%).

Number morphology in Spanish first language acquisition

357

At 2;0, agreement not only extends to noun phrases containing a noun besides a determiner (34a, b, c), but also to subject-verb agreement in structures containing a copula (35a, b). (34)

MAR 2;0 a. las tutas [: piernas], las tulas [: piernas], the:FEM:PL legs, the:FEM:PL legs, las pernas. the:FEM:PL legs ‘the legs’ b. (l)as moscas feiyas [: feas] aqu´ı no the:FEM:PL flies:FEM:PL ugly:FEM:PL here not est´an. be:PRES:3P ‘The ugly flies are not here.’ c. (l)as otas [: otras] galletas. the:FEM:PL other:FEM:PL cookie:FEM:PL ‘the other cookies’

(35)

MAG 2;0 a. d´onde est´an los caramelos? where be:PRES:3P the:MASC:PL candy:MASC:PL ‘Where are the candies?’ b. son m´ıos. be:PRES:3P mine:MASC:PL ‘They are mine.’

At the turn of the second year, correctly and erroneously expressed agreement occurs in equal quantities in both children. One month later, at 2;1, accurate instances of agreement are twice as frequent as erroneous ones, especially so within the NP (cf. table 6 and figure 5): (36)

MAR 2;1 ya palitos, yo cojo lo now stick:DIM:MASC:PL I take the:MASC:PL gapalitos [: ?] m´ıos. gapalitos?:MASC:PL mine:MASC:PL ‘I now take the little sticks.’

(37)

MAG 2;1 mucha. en las piernas on the:FEM:PL leg:FEM:PL a lot:FEM

358

Carmen Aguirre and Victoria Marrero

Table 6. Agreement in Mag´ın and Mar´ıa Age Mag´ın Utter. with plurals 2;0 74 2;1 13 2;2 64 2;3 87 2;4 86 2;5 26 2;6 45

Correct agreement 22 5 37 57 49 13 24

Incorrect agreement 20 1 5 17 4 3 1

Mar´ıa Utter. with plurals

Correct agreement

Incorrect agreement

No need of agreement

53 54 55 36 70 45

18 21 36 19 43 15

8 7 1 3 3 1

27 26 18 14 24 29

Magín 100 50 0 1;9

1;10

1;11

2;0

2;1

2;2

2;3

2;4

2;5

2;6

María 100 80 60 40 20 0 1;9 1;10 1;11 2;0

2;1

2;2 2;3

Correct agreement Percentage of errors

2;4 2;5

2;6

Incorrect agreement

Figure 5. Rate of agreement errors

At 2;2, only a third of Maria’s and a sixth of Mag´ın’s utterances contain agreement errors. From 2;3–2;4 onward, fluctuations between multiple and single markers follow the adult model: they are the same in child speech as in the input, contextually conditioned. The absence of required markers in the two children becomes increasingly infrequent, even if isolated mistakes are still found in subject-verb agreement.

Number morphology in Spanish first language acquisition

359

Table 7. Mar´ıa. Errors in number agreement Age Nominal 2;1 qu´ıta-lo estos remove:IMP-it:MASC:SG this:MASC:PL tres sardina three sardine:SG eso zapato this shoe a manos the:FEM:SG hand:PL mucha cosa (3 times) ‘many thing:SG’ 2;2 E juguetes (2 times) the:MASC:SG toy:PL mi guopas [: ropas] my:SG clothes:PL los leotardo the:PL panty:SG una medias estos a panties these 2;3 2;4

2;5

2;6

Verbal no es tuyos not be:PRES:3S yours:PL ‘They isn’t yours.’

ta malos be:PRES:3S bad:MASC:PL gusta las galletas like:PRES:3S the:FEM:PL cookies los ot(r)os est´a the:MASC:PL others be:PRES:3S

la bocas (2 times) the:FEM:SG mouth:PL las ni˜na the:FEM:PL girl:SG la me(d)isinas the:FEM:SG medicine:PL viene la polic´ıas come:PRES:3S the:FEM:SG police:FEM:PL no puedo con el flores. not can:PRES:1S with the:MASC:SG flower:FEM:PL Pa(ra) que pase e´ stos so that pass:PRES:SUBJ:3S these:MASC:PL ‘so that these may pass’

Once the morphological stage has been reached, agreement errors drop dramatically even though sentences become increasingly complex, with the possibility of errors increasing as more elements need to be considered. As we have seen, many of the late errors in Mag´ın consist in constructing a singular verb form with a plural subject noun phrase. Mar´ıa, however, shows isolated cases of disagreement between articles and nouns until the end of observation. In summary, the two Spanish children begin to mark the plural using only one position out of the various possible markers throughout the utterance. When input received and the child’s own drive for communicative accuracy push him or her to introduce more plural markers in the utterance, agreement emerges, not

360

Carmen Aguirre and Victoria Marrero

Table 8. Mag´ın. Errors in number agreement Age Nominal 2;0 los ojo the:MASC:PL eye:SG (2 times) los tuyo ah´ı the:MASC:PL yours:SG there (2 times) los zapato, los tuyo. the:MASC:PL shoe:SG, the:MASC:PL yours:SG d´onde est´an los caramelo? where are the:PL cookie:SG los caramelo, mam´a the:PL cookie:SG, mum la llaves the:FEM:SG key:PL (5 times) los pendiente a poner the:MASC:PL earring:SG to put tiene la gotas has the:FEM:SG drop:PL esto m´ıos this:NEUT:SG mine:PL la moscas the:FEM:SG fly:PL 2;1 el otro la pompas the:MASC:SG other:MASC:SG the:FEM:SG bubble:PL 2;2 los otros no the:MASC:PL other:MASC:PL not una pilas a:FEM:SG battery:FEM:PL la pelotillas the:FEM:SG ball:DIM:PL (2 times) 2;3 esos me lo pones these:MASC:PL me it:SG put:PRES:3S mete los pie put:IMP the:MASC:PL foot:SG una veces a:FEM:SG time:PL (3 times) en el trabajos in the:MASC:SG jobPL (2 times) mi piezas my:PL piece:PL (4 times) son mi patos be:PRES:3P my:SG duck:PL tu pies your:SG foot:PL son la campanadas be:PRES:3P the:FEM:SG bell:FEM:PL esta campanadas this:FEM:SG bell:PL 2;4 una tijeras a:FEM:SG scissors:PL se ha dado con el los zapatos ‘have:PRES:3S hurt:PP with the:MASC:SG the:MASC:PL shoe:PL

Verbal se ha ca´ı(d)o lo(s) tuyo have:PRES:3S fall.down:PP the:MASC:PL yours:SG los caramelo est´a ah´ı the:MASC:PL cookie:SG be:PRES:3S there pasa las moscas pass:PRES:3S the:FEM:PL fly:FEM:PL son (l)o(s) hipop´otamo be:PRES:3P the:MASC:PL hippopotamus:SG

son mis pinto be:PRES:3P my:PL pencil:SG

estos me hace pupan this:MASC:PL me make:PRES:3S hurt los otros me hace pupa the:MASC:PL other:MASC:PL me make:PRES:3S hurt (2 times)

vacacione(s) se termina holiday(s) finish:PRES:3S (2 times)

Number morphology in Spanish first language acquisition

361

Table 8. Mag´ın. Errors in number agreement [continued] Age Nominal 2;5 los helicoptero the:MASC:PL helicopter:SG lo pinto lo(s) zapatos it:SG paint:PRES:1S the:MASC:PL shoe:PL los pez the:MASC:PL fish:SG 2;6 son malo que los cangrejos be:PRES:3P bad:MASC:SG than the:MASC:PL crab:PL

Verbal

only as a consequence of increasing communicative needs, but also, possibly, as a syntactic device. The process of acquisition is comparatively rapid, although some agreement errors remain for an extended period.

5. Discussion 5.1.

Is morphological number a relatively late category in language development?

Previous studies in Spanish acquisition report that number appears after gender in nominal categories (Hern´andez Pina 1984; P´erez Pereira 1991; Aparici, D´ıaz, and Cort´es 1996) and after person in verbs and pronouns (for Spanish see Fern´andez 1994; Ezeizabarrena 1997; Bel 1998; Grinstead 2000; for German Poeppel and Wexler 1993; for Italian Pizzuto and Caselli 1992, 1993; Guasti 1993).18 Gender, like number, has a powerful semantic motivation, related to sex differences, but only for animate nouns19 . In Spanish, though, there are fewer possibilities for masculine/feminine than for singular/plural variation, because only the few animate nouns occur with either gender. Most nouns have their gender assigned, for historical and genetic reasons, in a fixed way. Only articles, determiners, and especially adjectives can offer a true demonstration of the contrastive use of masculine/feminine. Even if our aim in this paper is number, not gender, some general observations can be made (cf. 4.2.3.). Our data does not show a clear occurrence of gender distinctions prior to those of number. Both morphemes seem to undergo similar processes: in the beginning the children produce items without morphological

362

Carmen Aguirre and Victoria Marrero

(gender or number) variation or with unmotivated variation: gender and number inherent in the input and the child’s own utterances have not yet been analyzed. Later, at the protomorphological stage, the first gender and number contrasts emerge (at 1;10 in both Mag´ın and Mar´ıa). As far as the acquisition of gender is concerned, early contrasts emerge in high frequency animate nouns (nene-a ‘boy - girl’, t´ıo-a ‘uncle - aunt’, etc.) and attributive adjectives agree with their noun especially so when grammatical gender is motivated by natural gender of animate referents (guapo-a, ‘good-looking’; feo-a, ‘ugly’; malo-a, ‘bad’; buenoa, ‘good’, etc.). For two to three months we observe hesitation, error-making and self-correction (consider, for instance the examples 21–28), but the proportion of correct utterances increases rapidly. In conclusion, we find no arguments to reinforce the hypothesis of a late acquisition of number compared with the other inflectional nominal morphemes in the nominal inflectional categories.

5.2.

Do children avoid plurals during the early stages of language development?

Hoekstra and Hyams (1998) maintain that children follow a strategy of avoidance of plurals for a prolonged period of their language development, since they are not competent enough to produce them. The plural is, then, optional: occasionally it appears, but at other times it is absent, even where it is necessary. To use the authors’ term, utterances may be ‘underspecified’ for number. As discussed in section 4.1., our data shows that at the premorphological stage (when children are around one and a half years of age) plurals are not productive in the children’s output, and they can be absent even in contexts where the adult input is plural and demands a plural answer:20 (38)

MAR 1;7 a. MOT: quieres tapar los pies? want:2S cover the:MASC:PL feet? ‘Do you want to cover your feet?’ MAR: (t)apa(r) e(l) p(i)e. cover the:SG foot b. MAR: mam´a a poqu´ıs [: zapatos] Mom the:SG shoes

Number morphology in Spanish first language acquisition

363

MOT: voy, primero, primero el pijama come:1S first first the pyjama y luego los poqu´ıs [: zapatos]. and then the:MASC:PL shoes ‘I’m coming, first, first the pyjama and then the shoes.’ c.

MOT: qu´e haces con los nenes, cu´entale a Pap´a. ‘What do you do with the boys, tell daddy.’ MAR: nene no. ‘Boy not.’

Nonetheless, once the child has entered the protomorphological stage, and generalizes the use of plural morphology, we find no traces of plural avoidance, merely the absence of some plural markers in sentences, i.e. agreement errors. In all of these situations, though, the semantic concept of plurality is grammatically demonstrated via one or more morphological elements.

5.3.

Plural in nominal categories versus verbal plural

In spite of some individual differences, the plural in verbs emerges later and occurs less frequently than in nouns, articles, and other determiners. One of the reasons for this difference may be that the expression of plurality in the Spanish verbal system is more complex than in the noun. On one side, because a single suffix expresses both number and person, and sometimes may be fused with other grammatical categories of the verb. On the other, we have also a suffix thematic vowel (a-e-i) linked to the three Spanish verb classes and a tense-mood-aspect suffix, used in the past tense and subjunctive or conditional mood. Under these conditions the child cannot easily isolate a regular plural suffix – something he or she is easily able to do for nominal categories. Plural usage in children does not emerge with the same frequency in all tenses, nor in all persons. As our data shows, the third person of the present tense initiates the plural verb system, at 1;10 for Mag´ın and 2;2 for Mar´ıa. In conclusion, the development of plural number in nominal categories is easier and occurs earlier than verb plural. On one side, nouns are more frequent than verbs in the early lexicon of Spanish speaking children. Considering data on tables 2, 3, 4 and 5, we can see that the whole amount of nominal phrases is 12,679, while verbs are 5,643 (73%–27%). So, children are more familiar with members of nominal parts of speech than with verb forms. On the other side,

364

Carmen Aguirre and Victoria Marrero

whilst in nouns, adjectives, and most determiners children have only to contrast ∅ with –s or –es morphemes, the Spanish verbal system offers three different choices for the first, second and third person. The same concept of plurality is represented by three different morphemes depending on person, which is opposed to the biuniqueness principle (“one meaning – one form”) in Natural Morphology (Dressler et al. 1987). In fact, our data shows a picture of increasing morphological complexity, in front of which the child is more and more able to establish new relationships, with many new elements interacting with each other. In the next paragraphs we will try to summarize the main lines of this process.

6. Conclusions In brief: 1. Children show three steps in the development of plural morphology: a premorphological stage with no contrastive plural use, a protomorphological stage where we begin to find productivity in nominal and verbal plural morphology, and a morphological stage, characterized by the consistent use of plural morphology and the development of agreement. 2. During protomorphology, the child has only a semantic communicative goal, and expresses plurality by marking only one word with the plural suffix. In the morphological stage, occurrence of plural markers increases progressively in plural utterances, and agreement possibly emerges, not only as an accumulative memoristic process, but also as a syntactic device, a means to establish grammatical relationships and to identify constituents. 3. Plural morphology is initially controlled in nominal categories, appearing later in verbs. In nominal categories, the acquisition order is first nouns and articles, later followed by determiners and adjectives. In verbal categories the plural suffixes appear first in the present tense (third person and then first person) and later on in the past tense. 4. Between 2;2 and 2;3 plural agreement begins to be dominant and error rates in plural utterances drop to about 10%. 5. Our data do not show any relevant delay in plural morphology acquisition relative to other inflectional morphemes of nominal parts of speech such as gender morphemes. Nevertheless, our claim is that the diminutive is slightly earlier and more rapidly acquired, and acts as a bootstrapping element for the discovery of morphological rules. It should therefore facilitate plural development.

Number morphology in Spanish first language acquisition

365

Notes 1. Remnants of dual number are to be found in some grammatical adjectives, such as ambos ‘both’ or sendos ‘one each’. 2. We won’t deal here with the question of regarding the absence of a morpheme as a zero morpheme. 3. If the noun ends in a stressed vowel, the plural marker may vary between –s or –es: esqu´ı / esqu´ıs – esqu´ıes (‘ski/s’). This fact, combined with the higher saliency of the ∅/-es paradigm, may be the reason for its overuse reported in many Spanishspeaking regions of the Americas: cafeses, papases, pieses, sofases instead of caf´es ‘coffees’, pap´as ‘daddies’, pies ‘feet’, sof´as ‘sofas’ etc. (Vaquero, 1998: 15). 4. The boy, following the pattern e/o (el / los SG vs. PL), shows an article system more consistent than Mar´ıa’s, whose ‘protoarticles’ are frequently neutral with respect to gender and number and are invariably vowels like /e/ or /a/. 5. The elision of final –s makes the word end in an open syllable. Such words are generally very common in early child speech and in the acquisition of Spanish in particular (Bernhardt and Stemberg 1998; Lle´o 1996). 6. See Mariscal (1997) on ‘protoarticles’. 7. Two months earlier, at 1;7, the singular forms appear: MAR: caca e p´e  [: en el pie], caca e pie [: en el pie] ‘dirt on the:SG foot’. 8. Kilani-Schoch (1998: 31–45) reports the use of the numeral deux ‘two’ as a lexical plural marker indicating plurality in general during the early acquisition of French. Sedlank et al. (1998: 66) describe the use of the same numeral zwei ‘two’ and the quantifier alle ‘all’ to express plurality in the premorphological stage in Austrian German. 9. There are are only twelve different types; 76 tokens concern the adjective amarilli (adult form amarillo) ‘yellow’. 10. We have excluded 15 tokens of lexicalized bibis ‘dolls.’ 11. We follow here the most accepted theory among Spanish morphologists (RealAcademia Espa˜nola 1973, Fern´andez Ram´ırez, 1951, Alarcos Llorach, 1994; Alcina y Blecua, 1975, G´omez Torrego, 1993, etc.), and not Matthews’ proposal for Italian. 12. Mag´ın’s fossilized form vamos was removed from the table, because it is very much used, but always with an imperative value. 13. Repetitions and frozen forms, like no (es)tan, se van, and vamos have been excluded. 14. With the exception of the form vamos ‘go:PRES:1P’, the first plural form of verbs used by both children. This form has an imperative or exclamative value (i.e. ‘let’s go’) which is extended by Mag´ın to other verbs, as we have shown. It cannot be considered a plural form in the strict sense, as it is not opposed to any singular in the child’s system. 15. It can be considered that agreement, at this age, is not a device for structuring information, but a memorised storage of forms which go together with other forms. We will not elaborate on these considerations here. 16. Due to the scarcity of plural verbs (cfr. table 5) we didn’t include them in this account.

366

Carmen Aguirre and Victoria Marrero

17. We should consider that many utterances containing a single plural marker are grammatically correct. Most of them constitute answers to questions with a direct object not requiring a determiner. 18. Within the generativist framework, this delay in number acquisition has been used as an argument to explain the phenomenon of root infinitives in languages such as Dutch, German and French. Hoekstra & Hyams (1998) base their proposal concerning this phenomenon on the underspecification of the number category. Whenever number is not specified in DP (Determiner Phrase) a non-inflected verb form – an infinitive – occurs, and so number is considered to be crucial in these languages for expressing finiteness. In contrast to this, finiteness is expressed via person morphology in the null-subject languages Spanish, Catalan and Italian so that the absence of number is not a problem. Similarly, Bel (1998) proposed that number can be left unspecified in Catalan and Spanish child language, as there is no need to identify the subject (person agreement is present) and it is not necessary for aspect assignment either (achieved by tense). Drawing on the ideas of Hoekstra and Hyams (1998) she considers that number may be acquired but that the child is not accurate enough to produce it. It is optional and may be “underspecified” (using Hoekstra and Hyams’s terms). The result is that it occasionally appears, but there are times when it is absent even when necessary. Thus the frequency of the plural in child language remains much lower than in the adult target for a long period: children seem to “avoid” plurals. A very different explanation has been offered by Grinstead (2000), who assumes that all functional categories must be present in order to assign case to the subject. This author defends that during certain periods person and tense are present while number is absent, and consequently lexical subjects cannot appear until number is acquired. 19. “Los criterios sem´anticos que asignan el g´enero en los nombres animados tienen que ver esencialmente con el sexo de sus referentes” (The semantic criteria which select gender in animate nouns refer essentially to the sex of their referents) (Ambadiang 1999: 4848). 20. At 1;7, the girl uses closed final syllables, even the same final –s (esta, este ‘this: MASC’, ‘this:FEM’), so she doesn’t seem to have any phonological limitations.

References Aguirre, Carmen 1995 La adquisici´on de las categor´ıas gramaticales en espa˜nol. Ph. D. diss. Universidad Aut´onoma de Madrid. 2003 Early verb development in one Spanish child. In Development of Verb Inflection in First Language Acquisition:A Cross-Linguistic Perpspective, Dagmar Bittner, Wolfgang U. Dressler, and Marianne KilaniSchoch (eds.), 1–26. Berlin/New York: Mouton de Gruyter. Alarcos Llorach, Emilio 1994 Gram´atica de la lengua espa˜nola. Madrid: Espasa Calpe.

Number morphology in Spanish first language acquisition

367

Alcina Franch, Juan, and Jos´e Manuel Blecua 1975 Gram´atica espa˜nola. Barcelona: Ariel. Ambadiang, Th´eophile 1999 La flexi´on nominal: G´enero y n´umero. In Gram´atica descriptiva de la lengua espa˜nola, Ignacio Bosque and Violeta Demonte (eds.), 4843– 4914. Madrid: Espasa. Aparici, Melina, Montserrat Cort´es, and Gloria D´ıaz 1996 El orden de adquisici´on de morfemas en catal´an y en castellano. In Estudios sobre la adquisici´on del castellano, catal´an, eusquera y gallego, Miguel P´erez Pereira (ed.), 165–174. Santiago de Compostela: Publicaciones de la Universidad de Santiago de Compostela. Bel, Aurora 1988 Teoria ling¨uistica i adquisici´o del llenguatge. An`alisi comparada de trets morfol`ogics en Catal`a y Castell`a. Ph. D. diss. Universidad Aut´onoma de Barcelona. Bernhardt, Barbara H., and Joseph P. Stemberg 1998 Handbook of Phonological Development: From the Perspective of Constraint-Based Nonlinear Phonology. San Diego: Academic Press. Braine, Martin D. S. 1963 The ontogeny of English phrase structure: The first phase. Language 39: 1–13. Caselli, Maria Cristina, Laurence Leonard, Virginia Volterra, and Maria Grazia Campagnoli 1993 Toward mastery of Italian morphology: A cross-sectional study. Journal of Child Language 20: 377–393. Dressler, Wolfgang U., and Annemarie Karpf 1995 The theoretical relevance of pre- and protomorphology in language acquisition. Yearbook of Morphology 1994, 99–122. Dressler, Wolfgang U., Willi Mayerthaler, Oswald Panagl, and Wolfgang U. Wurzel 1987 Leitmotifs in Natural Morphology. Amsterdam: Benjamins. Ezeizabarrena, Mar´ıa Jos´e 1997 Morfemas de concordancia con el sujeto y con los objetos en el castellano infantil. In Contemporary Perspectives on the Acquisition of Spanish, Ana Teresa Perez-Leroux and William R. Glass (eds.), vol. I, 21–36. Somerville, MA: Cascadilla Press. Fern´andez, Almudena 1994 El aprendizaje de los morfemas verbales: Datos de un estudio longitudinal. In La adquisici´on de la lengua espa˜nola, Susana L´opez Ornat, Almudena Fern´andez, Pilar Gallo, and Sonia Mariscal, 29–46. Madrid: Siglo XXI. Fern´andez Ram´ırez, Salvador 1951 Gram´atica espa˜nola: el nombre. Edition of 1986. Madrid:Arco-libros.

368

Carmen Aguirre and Victoria Marrero

G´omez Torrego, Leonardo 1993 Manual de espa˜nol correcto. Madrid: Arco Libros. Grinstead, John 2000 Case, inflection and subject licensing in child Catalan and Spanish. Journal of Child Language 27: 119–155. Guasti, Maria Teresa 1993 Verb syntax in Italian child grammar: Finite and non-finite verbs. Language Acquisition 3: 1–40. Hern´andez Pina, Fuensanta 1984 Teor´ıas psico-socioling¨u´ısticas y su aplicaci´on al espa˜nol como lengua materna. Madrid: Siglo XXI. Hoekstra, Teun, and Nina Hyams 1998 Aspects of root infinitives. Lingua 106: 81–112. Kilani-Schoch, Marianne 1998 The acquisition of number in French. In Studies in the Acquisition of Number and Diminutive Marking, Steven Gillis (ed.), 25–49. (Antwerp Papers in Linguistics 95.)Antwerp: UniversiteitAntwerpen. Lle´o, Conxita, Michael Prinz, Christliebe Mogharbel, and Antonio Maldonado 1996 Early phonological acquisition of German and Spanish: A reinterpretation of the continuity issue within the principles and parameters model. In Children’s Language, Carolyn Johnson and John H. Gilbert (eds.), vol. 9, 11–31. Mahwah, NJ: Erlbaum. L´opez Ornat, Susana, Almudena Fern´andez, Pilar Gallo, and Sonia Mariscal 1994 La adquisici´on de la lengua espa˜nola. Madrid: Siglo XXI. Mariscal, Sonia 1997 El proceso de gramaticalizaci´on de las categor´ıas nominales en espa˜nol. Ph. D. diss. Microfilm. ISBN:84-7477-740-2. Univ. Aut´onoma de Madrid. Marrero, Victoria, and Carmen Aguirre 2001 Adquisici´on y desarrollo de los plurales en espa˜nol. Paper read at the III Encuentro Internacional sobre la Adquisici´on de las lenguas del Estado, September 24–26, 2001, Universidad de M´alaga. 2003 Plural acquisition and development in Spanish. In Linguistic Theory and Language Development in Hispanic Languages, Silvina Montrul and Francisco Ord´o˜nez (eds.), 275–296. Somerville: Cascadilla Press. McWhinney, Brian 2000 The Childes Project: Tools for Analyzing Talk. 3rd ed. Mahwah, NJ: Lawrence Erlbaum. Matthews, Peter 1974 Morphology. Cambridge: Cambridge University Press.

Number morphology in Spanish first language acquisition

369

Mueller Gathercole, Virginia C., Eugenia Sebasti´an, and Pilar Soto 1999 The early acquisition of Spanish verbal morphology:Across-the-board or piecemeal knowledge? International Journal of Bilingualism 3 (2-3): 133–182. Newport, Elissa L. 1990 Maturational constraints on language learning. Cognitive Science 14: 11–28. P´erez Pereira, Miguel 1991 The acquisition of gender: What Spanish children tell us. Journal of Child Language 18: 571–590. Pizzuto, Elena, and M. Cristina Caselli 1992 The acquisition of Italian morphology: Implications for models of language development. Journal of Child Language 19: 491–557. Poeppel, David, and Ken Wexler 1993 The full competence hypothesis for early German phrase structure. Language 69: 1–33. RAE (Real Academia Espa˜nola) 1973 Esbozo de una nueva gram´atica de la lengua espa˜nola. Madrid: Espasa Calpe. Sedlak, Maria, Sabine Klampfer, Brigitta M¨uller, and Wolfgang U. Dressler 1998 The acquisition of number in Austrian German: A case study on the early stages. In Studies in the Acquisition of Number and Diminutive marking, Steven Gillis (ed.), 51–76. (Antwerp Papers in Linguistics 95.) Antwerp: Universiteit Antwerpen. White, Lydia 1982 Grammatical Theory and Language Acquisition. Dordrecht: Foris.

Relations between the development of the category of nominal and verbal number in two French-speaking children∗ Marianne Kilani-Schoch 1. Introduction From a typological perspective, number in French is an interesting category in that it concerns both the systems of noun and verb, whose properties are highly different (Stephany 1998, 2002). The system of the noun is predominantly isolating and mainly expresses number by analytic/periphrastic techniques, i.e. determiners. Synthetic marking on the noun itself only occurs in a few unproductive inflectional relics. By contrast, the verb system is typologically more mixed. Several tenses and moods are expressed by suffixation whereas person marking mainly rests on proclitics. Moreover, while the first conjugation class ending in /e/ has a tendency towards stem invariability (see below) and has evolved towards the isolating type, the other conjugation class ending in /r/, which is characterized by stem alternation and suppletion is more of the inflecting-fusional type (KilaniSchoch and Dressler 2005). The picture is rendered even more complex by the well-known phenomenon of morphonological liaison which additionally occurs in both systems and is currently characterized as inflectional (Morin 2003, 2005). Another intricacy inherent in the periphrastic marking of nominal plural which has to be taken into consideration is the fusion of the prepositions a` /de with the definite article les (`a/de&les → aux/des). The purpose of this paper is to study the role of these typological differences in the acquisition of number in two French-speaking children and to investigate how number marking on the noun relates to number marking on the verb. Specifically, since inflectional morphology is richer in the French verb than in the noun – i.e. inflectional productivity is found in the verb only – and since verb morphology has a fair amount of properties of the inflecting-fusional type, it seems tempting to predict that children learning French as their native language will not only detect morphology in the verb earlier than in the noun (Bittner, Dressler, and Kilani-Schoch 2003: xxv; Dressler, Kilani-Schoch, and Klampfer 2003: 414; Dressler 2005) but will also express plural more easily in

372

Marianne Kilani-Schoch

the former than in the latter (Slobin 1973, 1985). This rough prediction must be refined by taking the complex relations of the analytic and synthetic operations in the two subsystems of inflectional morphology into account. Indeed, the dominant pattern of French appears to be the periphrastic or analytic technique of articles and subject pronouns rather than the synthetic one of plural and person-number marking on noun and verb. Compared to clitic subject pronouns, articles are morphotactically fairly transparent. It therefore comes as no surprise that morphologically marked plural first emerges on determiners. The further development of determiners – especially of their fusional forms, be they liaison or portmanteau forms – is less clear, however. What is also unclear is their relation to the development of verb plural. In this paper, I will therefore try to tackle the problem of the relative emergence of nominal and verbal synthetic techniques. In particular, the question whether synthetic plural marking by liaison and portmanteau forms on nouns emerges earlier or later than synthetic plural marking by liaison and by suppletive or amplified stems1 on verbs will be addressed.

2. The data This study is based on the corpora of two children2 from Lausanne (Switzerland): Sophie (SOP) (1;6–3;8, 60 recordings, 30 hours, 13,000 utterances) and Emma (EMM) (1;4–2;11, 40 recordings, 19 hours, 8,000 utterances).3 Sophie can be characterized as following a prosodic strategy4 rather than a segmental one: she gives massive evidence of phonological substitutions and varied use of fillers which only disappear between 2;6 and 3;0. Emma, an early talker (MLU of 2.4 already at 1;7 and of 3.3 at 1;10), behaves more like a segmental child (Peters and Menn 1993) but also favors an imitative strategy. In the present paper, I will focus on the larger corpus of Sophie’s speech and use Emma’s corpus mainly for comparison. The phases of pre- and protomorphology correspond to the ages presented in table 1. Table 1. Phases of morphological development Premorphology Protomorphology Morphology proper

Sophie 1;6–1;10 1;11–2;3 2;4

Emma 1;4–1;7 1;8–2;1 2;2

Development of nominal and verbal number in two French-speaking children

373

Premorphology is the phase where no grammatical morphology has yet emerged. In protomorphology the children start to construct morphological patterns, e.g. first mini-paradigms with three members in the verb (Kilani-Schoch 2003). Morphology proper can be said to have developed when the nucleus of grammatical morphology has been acquired and has become productive.

3. Number in the noun 3.1. The grammatical category of number in the French noun 3.1.1. Obligatory expression of number Number is grammaticized in French and (in the spoken language) its expression (plural) is obligatory with determiners, a few nouns and a few masculine adjectives ending in –al (see tables 2 and 3). While in nouns and adjectives, number is mainly expressed by stem modification (see table 2), in determiners it is expressed by suppletion, i.e. no rule can be formulated, in spite of the fact that these plurals have an identical output schema /Ce/. Indeed the input of the alternation is very heterogeneous and the alternations between the vowels have no phonological motivation (see table 3).

Table 2. Synthetically marked plural singular travail [travaj] cheval [S@val] oeil [œj] boeuf [bœf] social [sOsjal]

plural travaux [travo] chevaux [S@vo] yeux [jø] boeufs [bø] sociaux [sOsjo]

gloss work horse eye ox social

Table 3. Singular and plural of determiners definite article indefinite/partitive possessive demonstrative

singular masc. le [l@], fem. la [la] masc. un [œ˜], du [dy] masc. mon [m˜o] masc. ce [s@] / cet [sEt], fem. sg. cette [sEt]

plural les [le] des [de] mes [me] ces [se]

374

Marianne Kilani-Schoch

3.1.2. Agreement Agreement with the governing noun in number (and gender) is overtly marked only on the determiner and a few adjectives ending in –al (used either attributively or predicatively). Most adjectives show agreement only in case of liaison (see 3.1.4).

3.1.3. Fusion The prepositions a` and de fuse with the definite article (table 4): Table 4. Fused forms PREP & DEF.ART a` & le MASC:SG a` & les PL de & le MASC:SG de & les PL

fused forms au [o] aux [o] du [dy] des [de]

examples au lit ‘in (the) bed’ aux oreilles ‘in/on the ears’ du pont ‘from/of the bridge’ des montagnes ‘from/of the mountains’

3.1.4. Liaison The appearance of a consonant in connected speech before a word with an empty onset belonging to an arbitrary lexical class is called liaison (table 5). Table 5. Liaison with liaison les amis [lezami]

gloss the friends

petit enfant [p@tit˜af˜a]

little child

without liaison les haches [leaS] not *[lezaS] les onze [le˜oz] not *[lez˜oz] petit h´eron [p@tier˜o] not *[p@titer˜o]

gloss the axes the eleven little heron

Liaison results in morphonological variation in the expression of number, especially with the definite and indefinite/partitive article overtly marked for plural (see [le] ∼ [lez], [de] ∼ [dez] above). Its distribution is mainly syntactically conditioned (Morin and Kaye 1982, 2005; Bybee 2001). It is obligatory between a determiner or a prenominal adjective in the singular or plural and the noun (e.g. petit enfant [p@tit˜af˜a] ‘little child’, petits enfants [p@tiz˜af˜a]) as well as in some frozen phrases (e.g. de haut en bas [d@ot˜aba] ‘from the top downwards’).

Development of nominal and verbal number in two French-speaking children

375

In contemporary French, most plural marking consists in liaison, e.g., determiners and prenominal modifiers such as (table 6): Table 6. Plural liaison gloss

example

gloss

fused form

aux 5 [o]

to, on

on the ears

possessive

leurs [lœr]

their

quantitative

d’autres [dotr]

others

interrogative

quel(le)s

which

adjective pl.

grandes

great

aux oreilles [ozOrEj] leurs amis [lœrzami] d’autres ours [dotr@zurs] quels animaux [kElzanimo] grandes amies [gr˜adzami] vs. grande(s) femme(s) [gr˜adfam]

some bears which animals great:FEM:PL friends vs. tall:FEM:SG/PL woman/women

Current analyses of liaison (see Morin and Kaye 1982; Morin 2003, 2005) consider prenominal liaison as being prefixed to the right-hand word. They are based on innovating examples such as quoi comme z-arbres lit. ‘what for trees’ (‘what kind of trees’), en plein z-examens ‘amidst exams’ where the liaison consonant cannot have been triggered by the preceding word, and on examples such as un robuste mais petit [p@ti] . . . t-enfant ‘a robust but small child’ with an unexpected liaison after a pause. All these examples imply that ‘the liaison consonants have become completely dissociated from the preceding word, to which they were historically related’ (Morin 2003: 389).

3.1.5. Productivity There is no common, frequent or productive inflectional pattern for marking plurality synthetically on the noun. As mentioned above (3.1.1), very few nouns (less than 50) have a plural form different from the singular. Loan words are generally integrated and become invariable, e.g. pl. des lieds [lid] (not lieder), pl. des maximums (rather than maxima). Therefore neologisms are never marked for number. The most generally available place of number information in the noun phrase is the article (and the subject pronoun in the verb phrase).

376

Marianne Kilani-Schoch

3.2. The development of number in the noun phrase 3.2.1. Lexical expression of the plural Since the category of number is generally not overtly expressed on the noun but is rather marked by the determiner, the acquisition of plural in French is interrelated with the development of the noun phrase. In Sophie’s corpus, both the indefinite and the definite article are at first rare. The singular form un of the indefinite article is not regularly used before the end of 1;11 and the plural forms les and des of the definite and indefinite/partitive articles only catch up at 2;4 reaching 75% of use in prenominal contexts at 2;5 (see 3.2.3). By 2;0, the number of articles exceeds the number of bare nouns used without articles (26% of nouns used with articles vs. 16% articleless nouns, 58% of fillers, see Kilani-Schoch and Dressler 2002). Sophie’s earliest expression of number is lexical. She starts using the numeral deux ‘two’ at 1;6 and it seems to serve as her prototypical (lexical) plural marker throughout the period studied. Since it is never contrasted with any other numeral in the corpus it does not seem to have a distinct numeral meaning but only indicates plurality in general (example 1). (1)

Sophie 1;6 MOT: tu aimerais le ballon ? ‘would you like the ball ?’ SOP: e` do e dodo a deux. FILL NUM FILL yy FILL NUM ‘FILL two FILL beddy-byes [?] FILL two’ MOT: les deux ? ‘both of them ?’

During a long period of ten months (from 1;6 until 2;3), bare nouns unaccompanied by an article and unmarked for number (‘N:SG’) occur in most plural contexts (table 7).

Table 7. Marking of plural in Sophie’s corpus from 1;6 to 2;2 PL contexts Tokens 255

‘N:SG’ in PL contexts Tokens % 190 75

Numeral deux contexts Tokens % 27 11

ART:PL in PL Tokens 4

% 1.6

Fillers Tokens 34

% 13

Development of nominal and verbal number in two French-speaking children

377

The plural definite article occurs only in a few isolated, rote-learned occurrences (example 2). (2)

Sophie 1;8 MOT: qu’est-ce qu’ils vont faire apr`es ? ‘what are they going to do afterwards ?’ SOP: les dents. ART.DEF:PL teeth ‘the teeth’ MOT: ils vont se brosser les dents, oui. ‘they are going to brush their teeth, yes.’

The word tooth is a noun which usually occurs in the plural. This is probably why Sophie has memorized that plural form. In Emma’s development of the category of number, which is quicker than Sophie’s, there is no evidence for a phase of lexical plural marking (see 3.2.3 below).

3.2.2. Noun plurals Isolated plural noun forms such as chevaux ‘horses’ appear from 1;9 on in Sophie’s corpus. Since they are rote-learned forms which at first occur only in the plural and may even be used for singular reference, they must not be interpreted as plural forms in the child’s language. It is only by 2;3 that Sophie uses both the plural and the singular form of the lexeme cheval ‘horse’ in the same recording and seems to distinguish their meaning (examples 3). (3)

Sophie 2;3 (a) a vu un has seen ART:INDEF:MASC:SG for j’ai vu un ‘I’ve seen a horse.’ (b) MOT: qu’est-ce que tu vois l`a ? ‘what do you see there ?’ SOP: bovaux for ‘horses.’

azal horse:SG cheval.

chevaux.

Overall, the proportion of both types and tokens of nouns synthetically marked for plural is small in both children’s data and even lower in Emma’s corpus than in Sophie’s (table 8).

378

Marianne Kilani-Schoch

Table 8. Synthetic noun plurals in Sophie’s and Emma’s data Subject

N:PL in PL contexts

Types/ tokens Sophie 1;6–3;0 4/23 Emma 1;5–2;11 3/14 Sophie’s input 4/39 Emma’s input 4/27

% of noun tokens 0.5% 0.5% 0.7% 0.5%

% of plural contexts 5% 3%

Number of noun tokens

Number of PL contexts

4816 2545 5523 5415

436 549

3.2.3. From fillers to articles In comparison with the development of number in other languages (Stephany 2002), the more strongly grammaticized means of expressing number seem to emerge more slowly in the development of French. This is at least in part due to the structure of the language to be acquired, in which nominal number marking depends on the acquisition of articles. Therefore, in a study of the development of number by a formulaic child (see fn. 4) such as Sophie, who uses a lot of fillers, it is important to consider the development of the precursors of plural articles and study their process of grammaticization. The increase of grammaticized fillers, in this case of unambiguous phonetic approximations of the plural definite article (see below),6 can be interpreted as a first step towards the expression of plurality. This is reached shortly before the age of 2;4, where there are more plural contexts containing grammaticized plural fillers (and first plural articles) than bare nouns unmarked for number (table 9). Still, articles remain infrequent at 2;3.7 Very soon the grammaticization process of prenominal plural fillers reaches a point in which all fillers are accurate phonetic realizations of the target definite or indefinite plural article. A grammaticized filler usually consists of the vowel of the adult morpheme, e.g. /e/ for plural definite or indefinite les, des /le, de/ Table 9. Plural articles and fillers in Sophie’s corpus between 2;3 and 2;68 Sophie 2;3.9 2;3.22 2;4 2;5 2;6

Number of N:PL contexts 16 15 68 43 45

Filler:PL /e/ 4 8 23 13 14

%

ART:PL

%

‘N:SG’

%

14% 53% 34% 30% 31%

– 2 24 18 22

– 13% 35% 42% 49%

12 5 17 7 9

75% 33% 25% 16% 20%

Development of nominal and verbal number in two French-speaking children

379

Table 10. Plural articles in Emma’s corpus at 1;8 Emma 1;8 1;9

Number of N:PL contexts 20 31

Filler:PL

%

ART:PL

%

‘N:SG’

%

1 1

5% 3%

17 28

85% 90%

2 2

10% 6%

or /a/ for definite la while, in earlier phases, fillers are mainly schwa and /a/. Schwa is predominant until 1;10. In the period discussed here, the use of plural articles also increases. At 2;4, plural articles are for the first time as numerous as grammaticized fillers. They are also more diversified with the indefinite article des being used besides the definite article les. By 2;6, Sophie produces obligatory linguistic plural markers in almost all plural contexts. New types of determiners are added to the definite and indefinite articles les and des: 3rd sg. plural possessive (ses ‘their’), demonstrative (ces ‘these’), invariable quantifiers such as un petit peu de +N ‘a little bit of’, plein de +N ‘lots of’9 (all at 2;4). Remaining fillers are all grammaticized fillers (/e/ for les), which disappear after 2;9. At approximately 2;6, there are also first correct uses of portmanteau plural forms in Sophie’s corpus (e.g. aux yeux ‘in/on the eyes’). The relatively slow and gradual development of plural determiners is not peculiar to Sophie. It agrees with other findings on the acquisition of French in which their regular occurrence is also observed in the middle of the third year (Gr´egoire 1937; M¨uller 1994; Bassano 2000). Nevertheless, Emma, the other child studied here, displays a different course of acquisition. Indeed, she generalizes the use of plural articles already by 1;8, only two months after her first expression of number by rote-learned articles and grammaticized fillers (table 10). As will become apparent below this difference between the children does not apply to other aspects of the acquisition of number so that there is a correlation between plural development in the two corpora after all.

3.2.4. Liaison Obviously, liaison as a more general phenomenon depends on the syntactic development of the noun phrase and emerges in Sophie’s corpus by 2;0. However, isolated frozen examples with the singular determiner un can be noticed from 1;8 on and become more frequent by the end of 1;11. Plural liaison, on the other hand, does not occur before fillers grammaticize and articles have become frequent by 2;4. First examples of (redundant) plural liaison are also found with grammaticized fillers (example 4).

380

(4)

Marianne Kilani-Schoch

Sophie 2;4.12 cochons m´e (l)es aut(res) [me ezot kOSo˜ ] yy FILL:ART.DEF:PL:LIAIS|z other pigs ‘FILL [= the] other pigs’

Earlier in development, before articles are used, liaison in singular or plural noun phrases occurring in the input may lead to a resegmentation of the onset of words in children’s speech (Gr´egoire 1937; Morel 1994; Chevrot and Fayol 2001). Such an example occurs in Sophie’s data at 1;9 (example 5). (5)

Sophie 1;9 MOT: l`a y a le hibou, les oiseaux [lajal@ibu lezwazo] ‘there, there is the owl, the birds’ SOP: dadeau for oiseau(x) [dado] [wazo] ‘bird(s)’ MOT: oui les oiseaux [wi lezwazo] ‘yes, the birds’

The form [dado] results from a phonological substitution of the plural liaison consonant /z/ (z → d), attached to the following word: dadeau10 = z(w)azeau = les oiseaux [lezwazo] ‘the birds’. Sophie reproduces the syllabic structure occurring in the prosodic unit [le-zwa-zo] applying the process of “stopping” to the two fricatives so that there is consonant harmony in the child’s form as there is in the adult form. The initial consonant has neither a morphological function nor a meaning so that in later occurrences it can be replaced by another consonant (see below). This follows from the fact that it is not the morphological unit les which is being produced by the child but the beginning of the second syllable of the input sequence les oiseaux which the child takes to be the first syllable of the lexeme oiseau. The consonant may also occur with a singular referent as in example (6) (see the mother’s recast of the child’s utterance). In this example, there is clear evidence of the child’s analysis of plural liaison as the onset of the following word.

Development of nominal and verbal number in two French-speaking children

(6)

381

Sophie 1;11.29 a dedans z azeau [: oiseau]11 [ad@d˜azazo] [: wazo] FILL inside LIAIS|z:PL bird ‘FILL inside bird’ MOT: oui il va dedans l’oiseau. [wi il va d@d˜a lwazo] ‘yes, it goes inside the bird’. SOP:

Examples of resegmentation, such as those given in (7) to (9), can be found throughout Sophie’s corpus. (7)

Sophie 2;5 des petits-n-ours (un ours → n-ours) for des petits ours [deptinurs] [deptizurs] ‘small bears’

(8)

Sophie 2;5 une autre . . . z-oreille (les oreilles → z-oreille) for une autre oreille [ynotrOrEj] [ynotr. . . zOrEj] ‘another ear’

(9)

Sophie 2;5 au n-oeil (un oeil → n-oeil) for a` l’oeil [onœj] [alœj] ‘in/on the eye’

From 1;11 to 2;1 Sophie renders the noun oiseau with varying initial consonants (examples 10).12 The target plural form les petits oiseaux ‘the little birds’ eventually occurs at 2;11. (10)

1;11 un [kazo], 2;0 un [tajo], 2;1 un [tazo] for un oiseau ‘a bird’

Consonant substitutions do not seem random here: [k] preserves the velar feature of the glide [w], [t] may be interpreted as the stopping of the liaison consonant of the plural article le[z] or as preserving the consonant occurring in the phrase petit oiseau [p@titwazo] ‘little bird’. The fact that vowel-initial words of standard French may have several variants comprising different initial consonants in the speech of French-speaking children from the end of the second through the end of the third year has also

382

Marianne Kilani-Schoch

been observed in other studies (Morel 1994; Chevrot and Fayol 2001). The consonants added to the noun by the children studied in this paper are /k/, /n/, /d/, /t/, /z/, /r/ and /l/. Not all of these belong to the inventory of possible prenominal liaison consonants; the latter do not include /l/, and /k/ only occurs in very rare literary instances.13 The consonants used by the children are probably partly determined by the phonemic context of the constructions in which the respective nouns are used on different occasions rather than corresponding to different representations in the lexicon. Thus, in example (11), the initial velar stop of the noun oiseau (resulting in [kazo]) may have been influenced by the mother’s previous use of the word coq [kOk] ‘cock’. (11)

Sophie 1;11 MOT: oui, toi t’avais dit un coq . [wi twatavEdiœ˜kOk] ‘yes, you had said a cock.’ MOT: et qu’est-ce qu’il y a encore ? ‘and what is there besides ?’ MOT: et c¸ a qu’est-ce que c’est ? ‘and this what’s this ?’ SOP: un kazeau [: oiseau]. [œ˜kazo] ‘a bird.’

Interestingly, substitutions of the plural liaison consonant due to false segmentation occur after these nouns have at first been used correctly with liaison after the determiner or prenominal modifier (compare the earlier correct examples in 12 to the later non-standard ones in 13–14). As is to be expected, the late examples of nonstandard forms of nouns cooccur with correct examples of liaison resulting in high variability.14 (12)

Sophie (a) 2;4 des autres ‘others’ (b) 2;5 grandes oreilles ‘big ears’ (c) 2;8 deux heures ‘two hours’ des assiettes ‘plates’

(13)

Sophie (a) 2;8 les-l-images ‘the pictures’ DEF.ART:PL-DEF.ART:SG-picture at 2;10 besides correct les images

Development of nominal and verbal number in two French-speaking children

383

(b) 2;9

des-l-´el´ephants ‘elephants’ INDEF.ART:PL-DEF.ART:SG-elephant (c) 2;9 les-l-habits ‘the clothes’ DEF.ART:PL-DEF.ART:SG-clothes (d) 2;9 tous les-l-enfants ‘all the children’ all DEF.ART:PL-DEF.ART:SG-child (e) 2;10 les-l-autres ‘the others’ DEF.ART:PL-DEF.ART:SG-other (14)

Sophie 2;10 des d-erv`esi for des exercices ‘exercises’ [dezEgzErsis]

The picture found in Emma’s data is not very different despite the fact that plural liaison occurs very early from 1;8 on (example 15a). According to the early emergence of articles in her speech (see table 10 above) correct liaison varies with the erroneous addition of diverse consonants to vowel-initial nouns (examples 15b–d). (15)

Emma (a) 1;8 1;9

les autres ‘the others’ les yeux ‘the eyes’ deux ans ‘two years’ 1;10 aux oreilles ‘on the ears’ 2;1 les oreilles ‘the ears’ (b) 1;9 deux ans ‘two years’ besides deux-d-ans (with stopping of /z/) (c) 1;10 les yeux ‘the eyes’ followed in the same recording by the sequences: zyeux, lyeux, mes yeux (‘my eyes’), zeux (d) 2;1 les oreilles and les-r-oz-oreille ‘the ears’ (with /r/ and ?aux (prep. a` & def.art.pl. les) (16)

Emma 2;3 MOT: (ils ont) des plus grandes+oreilles [deplygr˜adzOrEj] ‘(they have) larger ears’ (for standard de plus grandes oreilles) EMM: noreilles [nOrEj] (une oreille → n-oreille) for (z)oreilles [zOrEj]

384

(17)

Marianne Kilani-Schoch

Emma (a) 2;5 des-l-olives ‘olives’ INDEF.ART:PL-DEF.ART:SG-olive (b) 2;9 les-l-hippopotames ‘the hippopotamuses’ DEF.ART:PL-DEF.ART:SG-hippopotamus

There is evidence in both children’s speech that plural liaison does not yet function as a grammatical mark of plural meaning. Thus, in example (16), Emma changes her mother’s liaison in the plural noun phrase plus grandes oreilles ‘larger ears’ into a form of the noun oreille which is derived from the singular indefinite noun phrase une oreille ‘an ear’. Furthermore, both children continue to use l-substitutions, which block prenominal plural liaison, throughout their third year (examples 18–20). These substitutions most certainly result from a reanalysis of the consonant [l] of the reduced singular definite article as the initial consonant of the stem: les-l-habits = les+l’habit (see examples 13 and 17 above). (18)

Sophie 2;9 un-l-autre livre for un autre livre ‘another book’

(19)

Sophie 2;10 une autre l-image for une autre image ‘another picture’

(20)

Emma 2;9 le-l-hippopotame for l’hippopotame ‘the hippopotamus’

The examples of erroneous attachment of plural liaison /z/ to a plural form of the noun unpreceded by an article (example 21) and also to singular forms in Sophie’s corpus are further evidence for a wrong segmentation of noun phrases (examples 22 and 23). Although the attachment of /z/ is similar to the earlier attachment of other consonants it also shows that Sophie now bases herself on plural noun phrases of the input, especially so with plural-dominant nouns. It also seems to show that Sophie does not apply stopping any more.

Development of nominal and verbal number in two French-speaking children

385

(21)

Sophie 2;6 SOP: l`a on n’ a pas d-z-oreilles for pas d’oreilles [la˜onapadzOrEj] [padOrEj] there one has not PREP-LIAIS:PL-ear ‘there we don’t have any ears.’ MOT: il a pas d’oreilles lui ? [ilapadOrEjlOi] ‘he doesn’t have any ears this one?’

(22)

Sophie 2;6 SOP: cass´e [=? cach´e] r-une [= une] autre . . . z-oreille for [kaserynotrzOrEj] une autre oreille [ynotrOrEj] ‘hidden another ear’ MOT: y a une autre oreille cach´ee l`a ? [jaynotrOrEjkaSela] ‘is there another ear hidden there?’

(23)

Sophie 2;10 comme un z-Indien for comme un Indien [kOmœ˜n˜Edj˜E] [kOmœ˜z˜Edj˜E] ‘like an Indian’

How are we to interpret these data in terms of the development of liaison? In Morin’s (2003: 395–397) analysis of prenominal liaison as nominal prefixation, in which the prefixed consonant “formally indicates the presence of a preceding adjective or determiner belonging to a specific class” (p. 396), /t/ indicates the class of petit ‘small’, grand ‘big’, etc., /z/ the class of les ‘the’, des ‘some’, gros ‘big’, /n/ the class of ancien ‘old’, aucun ‘none’, un ‘one’, mon ‘my’. The data analyzed in this paper seem to show that the children have not yet recognized that the various prefixed forms of one and the same noun are determined by the context (z-avion ‘plane’, after les ‘the’, des ‘some’, gros ‘big’, t-avion after petit ‘small’, n-avion after un ‘a’, aucun ‘no’, etc.). Rather, they seem to use the different prefixed forms of standard vowel-initial nouns as optional phonetic variants.

386

Marianne Kilani-Schoch

3.2.5. Fused/portmanteau forms The irregular plural marking found in the portmanteau form aux (`a & les ‘at, in, on + DEF.ART:PL) triggers a substitution [o(z)] → [a(z)] in both corpora (examples 24–27). (24)

Sophie 1;11 asyeux [asjø]15 in eyes

for aux yeux [ozjø] ‘in.the:PL eyes’

(25)

Sophie 2;1 atalettes [atalEt] for aux toilettes [otwalEt] to toilets ‘to.the:PL toilet:PL’

(26)

Emma 1;10 a` dents to teeth

for aux dents ‘in.the:PL teeth’

Emma 2;0 azyeux in eyes

for aux yeux ‘in.the:PL eyes’

(27)

These substitutions cannot be explained phonologically since the homophonous singular form au [o] ‘in.the:SG’ occurs in both corpora in its standard form (examples 28). (28)

(a)

Sophie 2;0 au dodo ‘to bed’ 2;1 a` [//] au foot (with self-correction) ‘to the football’ (b) Emma 1;9 (jouer) au chien ‘play at (being a) dog’ 1;10 au cou ‘on the neck’

Thus, in constructions such as those cited in examples (24) to (27), the form [a] may either result from an overextension of the preposition a` taken from contexts

Development of nominal and verbal number in two French-speaking children

387

in which it is not fused with the article (e.g. a` la maison ‘(at) home’, a` l’´ecole ‘at/to school’) or may be an instance of the frequently used filler /a/. The form azyeux (for aux yeux) can be explained by a reanalysis of the plural liaison consonant16 as the initial consonant of the noun16 (see 3.2.6).

3.2.6. Commission errors It is at the end of the third year that first morphological commission errors occur in the expression of noun plural in Sophie’s data. At 2;11, the adult plural form chevaux used by Sophie until then is substituted by a form following the general invariable pattern of the French noun, the singular form cheval combined with the plural form of the definite article (example 29). In example (30), Sophie contrasts this with a singular construction expressed by the singular form of the definite article combined with the same (standard singular) form of the noun. Another example of the use of the singular form of a noun (exceptionally) marked for plural (in standard French) with the plural indefinite article is cited in (31). (29)

Sophie 2;11 les . . . cheval for les chevaux [leS@val] [leS@vo] the:PL horse:SG the:PL horse:PL

(30)

Sophie 2;11 et le b´eb´e cheval [el@bebeS@val] and the:SG baby horse:SG

(31)

Sophie 2;10 c’est des [sEdeZurnal] this is ART:INDEF:PL c’est des [sEdeZurno] this is ART:INDEF:PL ‘these are newspapers’

journal

for

newspaper:SG journaux newspaper:PL

This type of overregularization of the singular form of nouns in plural constructions is well documented in French child language (see Mayerthaler 1981: 56; Clark 1985: 705, among others). The reverse, i.e. use of plural forms in singular

388

Marianne Kilani-Schoch

contexts, such as un chevaux, un journaux may also occur (Mayerthaler 1981: 56; Clark 1985: 705). Such an instance is found in the corpus of Emma at 2;2 (example 32). (32)

Emma 2;2 un ART:INDEF:SG un ART:INDEF:SG ‘a little horse’

petit small petit small

chevaux for horse:PL cheval horse:SG

As has also been found with other children learning French (Clark 1985: 727), there is ample evidence for errors with portmanteau forms toward the end of the third year in Sophie’s development and in the beginning of the third year with Emma. While the earlier substitutions of portmanteau forms rather constituted simplifications of the target by omission of the article, in this later period, they consist of the full occurrence of the preposition and the plural article. The children use the prepositions a` and de together with the definite article instead of the fused forms and thus mark the locative relation transparently (examples 33–36). Besides being transparent ways of marking the locative relation, these constructions, which cannot be imitations of the adult language, represent regularizations of irregular plural forms by generalizing the definite plural form les once this has become frequent. They therefore constitute first evidence for a morphological processing of plurality as well as a preference for transparent coding. (33)

(34)

Sophie 2;10 le plus [l@plymOvEd@leSa˜ piño˜ la] ART:DEF:SG more de les (for fused des [de]) of ART:DEF:PL ‘the worst of the mushrooms there’

mauvais bad champignons l`a mushrooms there

Sophie 2;11 MOT: la maman de qui ? ‘the mother of whom?’ SOP: de les (for des [de]) petits gar¸cons l`a of ART:DEF:PL little boys there ‘of the little boys there’

Development of nominal and verbal number in two French-speaking children

(35)

Sophie 2;10 MOT: c’est juste les miettes ? ‘these are just the crumbs?’ SOP: mais a` les deux for but for the:PL two [mEaledø] aux deux/ pour les deux [odø]/ [purledø] to.the:PL two/ for the:PL two ‘but for both’ MOT: pour les deux les miettes ? ‘for both, the crumbs?’

(36)

Emma 2;2 (le coffre) de les jouets for (the chest) of the:PL toys des jouets of.the:PL toys ‘(the chest) of the toys’

389

It is important to point out that correct contracted forms occur alongside nonstandard, unfused ones containing a preposition eventually accompanied by an article (examples 37–38). Contracted standard forms occur in Emma’s data even earlier than in Sophie’s since Emma favors an imitative strategy. Therefore, the children seem to already have the knowledge that the forms aux/des and a` /de are related. According to Clark (1985: 727) variability in the use of portmanteau forms may last until the age of 5 years or even longer. (37)

Sophie (a) 2;11 a` cˆot´e des pieds at side of.the:PL feet de Papa of Daddy ‘next to Daddy’s feet’ (b) 2;10 mal aux dents ache at.the:PL teeth ‘toothache’ (c) 3;0 aux yeux, aux oreilles ‘in the eyes, on the ears’

390

(38)

Marianne Kilani-Schoch

Emma (a) 2;2

tous les ami des abeilles ‘all the:PL friend/s of.the:PL bees’ (b) 1;10 aux oreilles ‘on the ears’ (c) 2;1 FAT: ils jouent a` quoi ? ‘they are playing (with) what ?’ EMM: aux jouets ‘with the toys’

3.2.7. Summary As has been shown in the analysis of the data of two French-speaking children, the acquisition of plural in the nominal domain mainly consists in the gradual development of determiners according to their relative transparency: more transparent forms emerge first, while the development of fused forms takes place more slowly and extends over a long period of time during which achievements change gradually. In the earliest period, determiners are either omitted or their positions are marked by fillers and in one child’s data (Sophie) the notion of plural is expressed by a numeral. Despite important differences in absolute chronology in the early periods of development and the fact that Emma does not pass through a long early period with plural expressed by a numeral, a parallel relative chronology is found in both children as far as the formal means chosen for expressing the notion of plural in the nominal domain are concerned; also, the difficulties the children encounter with marking nominal plural in their native language seem to be comparable. Fillers – more precisely grammaticized fillers – are among the first formal means to be found in both children’s data, although there are very few of these in Emma’s corpus. The occurrence of fillers precedes regular marking of plural by the definite article by one month. Both children give evidence of regularizing portmanteau forms only once they have reached the developmental stage of morphology proper, i.e. the stage when articles are reliably used in obligatory contexts. As far as formal marking of plural in liaison contexts is concerned, both children show a U-shaped development, i.e. errors occur after some earlier correct realizations. Both corpora give evidence of errors concerning noun plurals even in the stage of morphology proper (although for a longer time with Sophie and with a different directionality as compared to Emma: Sophie les cheval, Emma un petit chevaux). By the end of the recordings, at 3;0 for Sophie and 2;9 for Emma, differences in the developmental rhythm of the two children seem to have almost vanished.

Development of nominal and verbal number in two French-speaking children

391

As we will see in the next section, the parallels between the two children are even more pronounced in the development of the category of plural in the verb.

4. Number in the verb 4.1. The grammatical category of number in the French verb 4.1.1. Obligatory expression of number The expression of number is also obligatory with the verb. However, in the first conjugation class (chant-er [Sa˜ te] ‘sing’) and in a limited number of verbs of the second conjugation class, its marking by suffixes is limited to the first and second person plural (table 11) (Kilani-Schoch and Dressler 2005: 140). Number marking on the verb is even more restricted in everyday interactions where the first person plural is usually replaced by the clitic on constructed with the 3rd person singular.This is also the case in the data studied in the present paper (see 4.2.1 below). Indeed, in everyday speech, the synthetic verb forms of the present indicative, imperative, imperfect, conditional, and present subjunctive are exclusively distinguished by proclitic pronouns (je, tu, il/ils, elle/elles, on chante /Sa˜ t/ ‘I, you:SG, he/they, she/they, we sing’), the only exception being the 2nd person plural /Sa˜ te/, which is, however, infrequently used in the speech situations studied here. The 3rd person masculine and feminine singular forms of the personal pronouns are homophonous with the respective 3rd person plural forms unless there is liaison with the following verb (see 4.1.2). For the reasons stated, synthetic plural verb forms belonging to the first conjugation class are expected to be scarce in the children’s data (as well as in the input) and, indeed,

Table 11. Present Indicative first conjugation class, chanter ‘to sing’, and second conjugation class, courir ‘to run’ Person 1st

Singular chante [Sa˜ t]

2nd

chantes [Sa˜ t] chante [Sa˜ t]

3rd

Present Indicative Plural cours (chant-ons [kur] [Sa˜ t-˜o] → ((on) chante) cours chantez [kur] [Sa˜ t-e] court chantent [kur] [Sa˜ t]

cour-ons [kur˜o] → ((on) court) cour-ez [kure] courent [kur]

392

Marianne Kilani-Schoch

Table 12. Present Indicative 3rd person of the auxiliaries avoir ‘to have’ and eˆ tre ‘to be’ and the semi- auxiliary aller ‘to go’ avoir SG a [a]

PL ont [˜o]

eˆ tre SG est [E]

PL sont [s˜o]

aller SG va [va]

PL vont [v˜o]

only a single example of such a plural form was found in the children’s speech (Emma) and two of them in the input (one in Sophie’s and Emma’s input each). In the second conjugation class, number marking is more important than in the first conjugation class since the 3rd person singular of the present indicative is formally distinguished from the 3rd person plural with most verbal lemmas (e.g. elle finit [Elfini] ‘she finishes’ vs. elles fini-ss-ent [Elfini-s] ‘they finish’). There are two types of 3rd person plural forms in this second class: a) a set of highly frequent and strong suppletive, monosyllabic rhyming verbs, namely ont ‘have’, sont ‘are’, vont ‘go’, font ‘do’ and b) a larger set18 of verbs characterized by stem amplification (henceforth stemamplified set), some of them with additional stem modification (see the second part of table 13). The rhyming verbs also function as auxiliaries (or as a semi-auxiliary in the case of faire ‘do’) in the four periphrastic verb forms of the compound past, pluperfect, past future, and compound future. Both person and number are distinguished (table 12). For some of the stem-amplified verbs the consonant of the amplification is identical with the infinitive coda of the stem, e.g. dorment - dorm-ir, mettent - mett-re. For most of them, however, the consonant is not predictable and

Table 13. Examples of second conjugation class Pres. Ind. 3rd person SG dort /dOr/ met /mE/ finit /fini/ boit /bwa/

PL dorment /dOr-m/ mettent /mE-t/ finissent /fini-s/ boivent /bwa-v/

INF dor-m-ir /dOr-m-ir/ me-tt-re /mE-t-r/ fini-r /fini-r/ boi-re /bwa-r/

gloss ‘to sleep’ ‘to put’ ‘to end’ ‘to drink’

prend /pr˜a/ peint /p˜E/ assied /asje/

prennent /prEn/ peignent /p˜E/ asseyent /asE-j/

pren-d-re /pr˜a-d-r/ pein-d-re /p˜Ed-r/ asseoir /aswar/

‘to take’ ‘to paint’ ‘to sit’

Development of nominal and verbal number in two French-speaking children

393

is lexically constrained (table 13). Minor rules describing the allomorphy can nevertheless be formulated for many of these verbs (Kilani-Schoch and Dressler 2005: 175–176). Depending on conjugation class and person, verbs may be overtly marked for agreement in number (and person) with their subject. The many partial systematicities of the expression of plural by verb forms or allomorphs of clitic subject pronouns (see 4.1.2 on liaison) raise the question how these intricacies are acquired by children.

4.1.2. Liaison In the third person of the first conjugation class, the only plural marker is liaison between the clitic subject pronoun and a verb beginning in a vowel (ils aiment [ilzEm] ‘they love’ vs. ils parlent [ilparl] ‘they speak’). There are no examples of preverbal liaison with a first conjugation class verb form in Sophie’s data and only a single one occurs in Emma’s (example 39). (39)

Emma 2;3 elles arrivent ‘they are coming’

Redundant liaison in the 2nd person plural is also rare (2 tokens at 2;7 in Emma’s data) (see 4.2.5 below). A few examples of liaison with verbs belonging to the second class can be found in Sophie’s data (example 40). At 2;8, Sophie uses a single instance of the 2nd person where liaison redundantly marks plural (see example 52 in 4.2.5). (40)

Sophie 2;4 ils ont pas e´ cout´e ‘they have not listened’ (lit.).

4.2. The development of number in the verb 4.2.1. Emergence of number distinctions As has been amply demonstrated for many languages (see Stephany 1997; Lieven 1998; Vihman 1999), the development of number in the French verb – strongly intertwined with person – is a progressive one in which not all categories are acquired simultaneously.The development is not linear either, i.e. new forms do not usually recur with any strong consistency. Examples in which the subject clitic

394

Marianne Kilani-Schoch

pronoun is missing or without number agreement on the verb can still be found in the last recordings of the period studied in Sophie’s development (see 53). Until 1;8 with Emma and even until 2;4 with Sophie, exclusively singular verb forms are found in the data, or at least verb forms which are unmarked for (person and) number (sg. and pl. aim(ent) [Em] ‘like’). Such verb forms may have to be interpreted as expressing plural according to the situational context (example 41) (see also 4.2.6 below). (41)

Sophie 1;9 e tatane FILL elephant ?les e´ l´ephants the:PL elephants

est is sont are

l`a for there l`a there

Subject clitics are limited in number and develop slowly: in Sophie’s corpus, no more than 23% of subject clitics are used in preverbal contexts even at 2;5 (N = 335). It is only at the end of the period of observation (3;0 for Sophie and 2;11 for Emma) that subject clitics are used in 76% of preverbal contexts (N = 183). With Emma, whose development proceeds more rapidly, 80% of subject clitic positions are filled already at 2;2 (N = 185) and 95% at 2;8 (N = 176). This shows that according to Brown’s (1973) 90% criterion of marking in obligatory contexts, neither one of the children studied reaches productivity with subject clitics before the end of her third year. Notice that on is always used instead of nous so that there are no instances of unstressed nous in either child corpus. There is no 1st person pluralsubject clitic nous in the input either and no instance of eux (MASC:3PL) in Emma’s data.

4.2.2. Rhyming plurals The earliest plural verb forms all belong to the set of rhyming plurals although with Emma also 2nd plural imperatives emerge early. The earliest and most frequently used rhyming plurals are 3rd pers. sont ‘(they) are’ and ont ‘(they) have’. Probably due to liaison phenomena, both children seem to be uncertain about the proper form of the auxiliary and sometimes correct themselves (examples 42–43). Notice that the children have first selected the auxiliary eˆ tre ‘to be’ and correct themselves by choosing avoir ‘to have’. These self-corrections suggest that there is a strong connection between the two verbs in the mental lexicon. Examples where the children replace avoir by eˆ tre in their utterances do not occur. In contexts without liaison the two verbs are kept apart.

Development of nominal and verbal number in two French-speaking children

(42)

Sophie 2;4 e´ son [=? ils sont] pas e´ cout´e [//] [es˜opaEkute] FILL are not listened pas e´ cout´e i(l)s ont [iz˜opaEkute] they have not listened ‘they didn’t listen’

(43)

Emma 1;11 ils sont mang´es [//] they are eaten ils t’ont mang´e they you have eaten ‘they have eaten you’

395

It is interesting to note that in the beginning of her fourth year Sophie still seems uncertain about the sequence of 3rd pers. plural subject pronoun followed by the plural verb form ont ‘have’ and happens to omit the obligatory liaison (example 44): (44)

Sophie 3;1 i-ont pas de balles for [i˜o] i(l)s ont pas de balles19 [iz˜o] ‘they do not have balls/any ball’

After the sporadic occurrence of the form ont ‘(they) have’ at 2;4 (see example 42), the 3rd pers. plural form sont ‘(they) are’ of the copula eˆ tre ‘to be’ is the only plural verb form occurring between 2;5 and 2;8 in Sophie’s data (15 tokens), with the exception of a single token of the 3rd pers. plural of faire ‘to make’ in font une baignade ‘(they) are taking a bath’. Since according to Benveniste and Adam (1999: 100) eˆ tre ‘to be’ is the most frequent verb lemma occurring in spoken French after avoir ‘to have’ these findings are not surprising. Also in Emma’s corpus, the copula eˆ tre ‘to be’ occurs with the highest frequency (see table 14) and the form sont emerges at 1;9 as the first rhyming plural. As with Sophie, the copular function of sont appears earlier than the auxiliary one (1;11), but the time-lag between the two uses is shorter in Emma’s development, namely two months instead of three.

396

Marianne Kilani-Schoch

Table 14. Development of 3rd pers. rhyming plurals Sophie 2;4 ont AUX ‘have’ 2;5 sont COP ‘are’ 2;6 font ‘do’ 2;8 sont AUX ‘are’ ont full verb ‘have’ 2;10 vont full verb ‘go’ vont semi-AUX ‘are going to’

Emma 1;9 sont COP ‘are’ 1;10 ont AUX ‘have’ 1;11 sont AUX ‘are’ 2;1 font ‘do’ 2;3 ont full verb ‘have’ vont full verb ‘go’ 2;6 vont semi-AUX ‘are going to’

With both children, the 3rd pers. plural of aller ‘to go’ emerges after that of eˆ tre ‘to be’ and avoir ‘to have’, whether as a semi-auxiliary or a full verb.20 In Emma’s corpus there are only two tokens of the 3rd pers. plural of aller at 2;6 and 2;9, both of which function as a semi-auxiliary. (45)

Emma 2;6 ils vont dormir ‘they are going to sleep’

In both children’s corpora, the number of tokens of the rhyming plural of avoir is also very small (tables 15 and 16). As far as Sophie’s data are concerned, this scarcity mirrors the input where ont and vont are the least frequent of the rhyming verb plurals representing less than 10% of them. With Emma, however, the low frequency of the rhyming plural of avoir does not mirror the input, where ont represents almost 30% of the rhyming verb plurals. By far the most 3rd person plural forms belong to rhyming verbs in both the children’s speech and their input. It is therefore not surprising that they emerge before other 3rd person plurals. However, the relation to the input is not straightforward since the age of appearance of rhyming plurals does not exactly match input frequencies. Thus, in Sophie’s corpus, the auxiliary ont occurs earlier than sont and font, which have the highest token frequency in the input. Although vont, both as a full verb and a semi-auxiliary, occurs more frequently in the input than the auxiliary sont, it emerges last, not only in Sophie’s but also in Emma’s corpus. In Emma’s data, the auxiliaries ont and sont occur earlier than font. Thus, there seems to be only a moderate correlation between the frequency of rhyming plural forms occurring in the input and age of first appearance (cf. Wijnen, Kempen, and Gillis 2001; Gillis 2003). Notice that in both input corpora the frequency ranking of these forms is the same except for ont as a full verb (example 46).

Development of nominal and verbal number in two French-speaking children

397

Table 15. Tokens of rhyming plurals in the children’s data and the input Lemma 3P

Sophie 2;4–3;0 Emma 1;8–2;11 Sophie’s input 2;4–3;0 Emma’s input 1;8–2;11 Sophie’s input 1;6–3;0 Emma’s input 1;5–2;11

eˆ tre ‘be’ sont full verb 41

eˆ tre ‘be’ sont AUX

aller ‘go’ vont AUX

3

aller ‘go’ vont full verb 1

12

2

70

27

4

2

1

0

2

9

44

78

7

5

6

7

7

24

134

60

2

28

21

7

12

42

172

174

13

9

21

22

14

59

312

69

2

28

21

8

12

45

185

8

avoir ‘have’ ont full verb 3

avoir ‘have’ ont AUX

faire ‘do’ font

Total of rhyming pl.

Table 16. Proportion of rhyming plurals in the children’s data and the input (tokens) Subjects

Total of verb tokens

Sophie 2;4–3;0 Emma 1;8–2;11 Sophie’s input 2;4–3;0 Emma’s input 1;8–2;11

4693

77

Verb pl. tokens/ total verb tokens 1.6%

4606

68

6913

8276

(46)

(a)

Total verb pl. tokens

Total rhyming pl.

Rhyming pl./total verb tokens

70

1.5%

Rhyming pl./total verb pl. tokens 91%

1.5%

44

0.96%

65%

153

2.2%

142

2%

93%

209

2.5%

152

1.8%

73%

Sophie’s input sont > font > ont (AUX) > vont > vont (AUX) > sont (AUX) > ont (b) Emma’s input sont > font > ont (AUX) > vont (AUX) > vont > ont > sont (AUX)

398

Marianne Kilani-Schoch

4.2.3. Imperatives At first sight, it may be striking that not only the first plural verb forms occurring in Emma’s corpus at 1;8 are imperative plurals but that all imperative forms but one in her data are plural imperatives. The two examples of imperatives to be found in Sophie’s corpus are also plural. However, most, if not all of these forms can be considered as formulaic, i.e. rote-learned contextually bound forms whose meaning is not immediately linked to the lexical meaning of the verb. In the adult language, these forms either function as discourse markers (e.g. attendez ‘wait’, regardez ‘look’, tenez ‘hold’) or as formulae tied to specific situational contexts (partez ‘start’, allez ‘go on’21 ). The interactional function or illocutionary force of these forms is more important than their semantic content. The children use them as attention-getting or directing devices without taking their appropriateness in specific frameworks of participants into account. Thus, in example (47), Emma oddly addresses her father by using the polite 2nd plural verb form reformulating her utterance in which she first uses the appropriate 2nd singular. (47)

Emma 1;8 attends [/] attends [/] attends; attendez wait:IMP:SG wait:IMP:PL

It may well be that by replacing the appropriate singular form by the longer plural one, Emma is trying to increase the illocutionary force of the directive. The alternation may also have a playful value as in example (48) where she is playing around with rote-learned forms successively transforming them into other such forms. (48)

Emma 1;8 attention partez, attention start:IMP:PL parte parte start:SUBJ:SG/3PL [twice]

pars, start:IND:SG, pars parti. start:IND:SG gone.

Another illustration of the rote-learned nature of these plural forms serving as interjections is example (49), where the plural form of aller ‘to go’ used in addressing a baby is similar to adult usage and lacks a number meaning. The same form also occurs twice in Sophie’s corpus (at 2;5 and 2;7) and represents her only imperative plural form. The hortative 1st person plural allons! ‘let’s go!’, which occurs in Emma’s data at 1;10 also belongs to this set of rote-learned verb forms (example 49).

Development of nominal and verbal number in two French-speaking children

399

Table 17. Formulaic plural imperative and hortative forms in the children’s data and their input Sophie

form.verb pl. tokens Emma 1;8 1;10 1;11 2;1 2;2 2;3 2;4

2;5 allez ‘go’ 2;7 allez total Sophie’s input allez/allons

(49)

1 1 2 7

form.verb pl. attendez ‘wait’ partez ‘start’ allez ‘go on’ allons ‘let’s go’ allez allez allez partez allez regardez ‘look’ tenez ‘hold’ allez

total Emma’s input allez

tokens 1 1 2 1 1 4 2 1 1 1 1 1

17 26

Emma 2;2 allez, b´eb´e, bois! ‘come on, baby, drink!’

The interpretation of the plural imperative and the hortative forms discussed in this section as rote-learned, formulaic forms without any morphological processing is supported by the fact that they occur in Emma’s data only between 1;8 and 2;4 but not thereafter (table 17).

4.2.4. Stem-amplified 3rd person plural forms Besides rhyming plurals, stem-amplified ones are the only plural forms marked on the verb stem. The latter appear much later than the first, with a time lag of four or even five months (at 2;2 in Emma’s corpus and at 2;9 in Sophie’s). Examples are very few, in the children’s data as well as the input (table 18). In these verbs, the 3rd person sg. and pl. only differ by the absence vs. presence of a final consonant, eventually accompanied by a nasal vs. oral vowel. The singular forms emerge before the plural forms in both corpora (examples 50, 51).

400

Marianne Kilani-Schoch

Table 18. Stem-amplified 3rd person plurals in the children’s data and the input Sophie

stem-amplified pl. tokens

2;9 2;10 2;11 total Sophie’s input

dorment ‘sleep’ mordent ‘bite’ disent ‘say’ 12 lemmas

2 1 1 4 20

Emma 2;2 2;5

stem-amplified pl. tokens descendent 1 ‘descend’ prennent ‘take’ 1

2;10

disent ‘say’

1

total Emma’s input

10 lemmas

3 10

(50)

Sophie (a) 1;6 dort /dOr/ ‘sleeps:3S’ 1;11 dit /di/ ‘says:3S’ 2;0 mord /mOr/ ‘bites:3S’ (b) /dOrm/ ‘sleep:3P’, /diz/ ‘say:3P’, /mOrd/ ‘bite:3P’

(51)

Emma (a) 1;10 descend /dEs˜a/ ‘descend:3S’ (b) 1;11 prend /pr˜a/ ‘take:3S’ (c) 2;2 dit /di/ ‘say:3S’ (d) /dEs˜ad/ ‘descend:3P’, /prEn/ ‘take:3P’, /diz/ ‘say:3P’

The difference in emergence of rhyming plurals and those marked by an amplified stem in Sophie’s and Emma’s data (tables 14 and 18) seems to be related to the different frequencies with which these two types of plural occur in the input. It seems plausible to assume that the lower frequency of stem-amplified plurals in the input will delay their appearance in the children’s language.22

4.2.5. Other plural forms An even smaller number of verb plurals is constituted by 2nd person plurals, which only occur in the second half of the third year in both children’s data and are rarely used. They concern the 2nd pers. pl. of the auxiliaries avoir ‘to have’ and eˆ tre ‘to be’ in Emma’s data but only the former in Sophie’s (example 52). Emma also uses two full verbs in this form. Plural verb forms other than those described in sections 4.2.2 and 4.2.3 are limited to the 2nd pers. also in the input and occur very rarely (3 types/4 tokens and 2 types/5 tokens in Sophie’s and Emma’s input, respectively) e.g.,

Development of nominal and verbal number in two French-speaking children

(52)

401

Sophie 2;8 vous avez pas perdu you:2P have:2P not lost ‘you haven’t lost’

4.2.6. Commission errors Due to the high degree of syncretism characteristic of French verbal inflection, the amount of number errors in the domain of verbs is extremely low in both children’s data (7 and 6 tokens in Sophie’s and Emma’s corpora, respectively) and mainly consists of agreement errors concerning singular verb forms constructed with (right-dislocated) plural lexical subjects (example 53). (53)

Sophie 3;0 elle est o`u PRO:FEM:SG is:SG where elles sont o`u PRO:FEM:PL are:PL where ‘where are my slippers?’

mes my:PL mes my:PL

pantoufles? for slippers pantoufles? slippers

Use of singular forms for plural ones even occurs in immediate repetitions of adult plural forms. Thus, at 2;2, Emma repeats adult ils dorment ‘they are sleeping’ as il dort ‘he is sleeping.’ Commission errors other than scarse examples of plural stem-amplified forms instead of singular ones (example 54) and of bare infinitives instead of 3rd person plural present (example 55) are not found in the children’s data studied in the present paper (cf. Damourette and Pichon 1933: 22, Morin 1987). (54)

Sophie 2;4 prennent [prEn] for prend [pr˜a] ‘take:3P’ ‘take:3S’

(55)

Sophie 2;3 MOT: qu’est-ce qu’ils font chez Mamie ? ‘What do they do at Granny’s?’ SOP: lire livre for read:INF book ils lisent un livre they read:3P a book’ ‘They read a book.’

402

Marianne Kilani-Schoch

4.2.7. Summary As far as the development of marking the category of plural on the verb is concerned, three main types of structurally different plural verb forms, rhyming, formulaic and stem-amplified ones, were found to emerge at different times of development in the children’s speech. In addition, formulaic verb forms have been seen to play a role with the more imitative child Emma. In both children’s speech, rhyming plurals and formulaic plurals emerge earlier than stem-amplified ones, the former because of their relatively high frequency in the input (tables 15 and 16) and the latter in spite of their much lower frequency because they are unanalyzed verb forms serving as discourse markers with a pragmatic meaning and are not integrated into the syntactic structure of the utterance (table 17).

5. Conclusion The aim of this paper has been to study the relations between the development of nominal and verbal number in two French-speaking children from a typological perspective. In spite of the difference in the speed of development between the two children many parallels exist not only as far as the chronological development of the category of plural in the noun on the one hand and the verb on the other is concerned but also in the chronological relation between the development of the two plural categories. In both corpora, number marking on the verb emerges later than number marking on the noun. Morphologically marked number starts to develop in the nominal domain with grammaticized fillers and first plural articles (Sophie: 2;0; Emma: 1;6). First verb plurals, i.e. rhyming and formulaic plural forms (Sophie: 2;4 and 2;5; Emma: 1;9 and 1;8 respectively), begin to emerge in both corpora after plural articles have become frequent (Sophie: 2;3, Emma: 1;8) and first synthetic plural markings on the noun have appeared in Sophie’s data (2;3) (table 19), i.e. not before protomorphology. Stem-amplified plurals emerge even later (Sophie: 2;9, Emma: 2;2), i.e. several months after the first expression of number: ten months after the appearance of the numeral deux in Sophie’s corpus and four months after fillers and articles first have occurred in Emma’s. Hence, the contrastive study of the development of number in the nominal and verbal domains seems to show that the child needs to be exposed to plural verb forms for a much longer time than to plural noun phrases in order to begin marking number in her own speech. What happens in the mean-time

Development of nominal and verbal number in two French-speaking children

403

Table 19. Input frequency and age of emergence

Plural articles Rhyming verb plurals Formulaic verb plurals Synthetic noun plurals Stem-amplified verb plurals

Sophie’s input23 19%

Age of emergence

2%

Age of emergence Emma’s input 2;0 19% (frequent: 2;3) 2;4.20 1.8%

0.05%

2;5

0.3%

1;8

0.7%

2;3

0.5%

1;9

0.1%

2;9

0.2%

2;2

1;6 (frequent: 1;8) 1;9

is the emergence of mini-paradigms of verbs (expressing tense distinctions), a developmental step which has been shown to be crucial for the early development of morphology (Dressler, Kilani-Schoch, and Klampfer 2003; Kilani-Schoch 2003). When taking the findings on language acquisition generally and the typological properties of the French noun and verb into account (see sect. 1 above) the precedence of noun plural over verb plural is unexpected because synthetic marking of inflectional categories tends to develop earlier than periphrastic marking (Bassano et al. 2004: 56; Dressler, Stephany, Aksu-Ko¸c and Gillis 2007). It can, however, be explained by the more limited functional relevance of the plural category in the verb as compared to the noun (in contrast to the functionally more relevant verbal categories of person, tense, aspect, and mood). Moreover, plural is inherent in the noun but contextually determined in the verb and accordingly more dependent on syntax than with the noun (Booij 1996). Therefore, the demands on syntax are higher with the category of number in the verb than in the noun. In order to get number right with the noun more local constructions (NPs) are sufficient while with verbs the whole predicative construction has to be taken into consideration. Since the children’s syntactic constructions are, however, at first limited in scope, a better explanation for the priority of nominal vs. verbal number distinctions may be the different frequency distributions of plural articles and plural verbs in the input. There is a remarkable convergence in the speech of the caretakers of the two children in this respect: while the frequencies of plural determiners are quite high, those of 3rd person plural verb forms are very low (tables 20 and 21).

404

Marianne Kilani-Schoch

Table 20. Number of plural verb forms (1st , 2nd , and 3rd pers.) in the input

Sophie’s input Emma’s input

Number of overtly marked pl. verb forms 343 214

Number of verb tokens

%

13773 8465

2.5 2.5

Table 21. Number of plural articles /e/ (les, des, ces, mes, tes, ses) Sophie’s input Emma’s input

Plural articles 1027 786

Number of articles 4502 4161

% 19 19

Another finding of this study concerns the emergence of the two kinds of synthetic plural marking by liaison, namely prenominal and preverbal liaison, which depend on the development of determiners and subject clitics, respectively. In Sophie’s corpus, plural liaison between an article and a noun emerges at the same time as that between a subject pronoun and a verb, namely at 2;4. In Emma’s corpus, however, plural liaison in the noun phrase emerges as early as plural articles, namely at 1;8, while the only example of preverbal liaison is found much later (at 2;3). The reason for this difference in the development of the two children seems to be that both types of liaison develop late with Sophie, whereas Emma shows the typical chronological hierarchy because of her precocious development of the noun phrase so that liaison between subject clitics and the verb has to wait for the later development of subject clitics to have occurred. In conclusion, the higher frequency of noun plurals in the input as compared to 3rd person plural verb forms, both type- and tokenwise, can largely explain why the children come to grips with the nominal number system earlier than with the verbal one. However, within each of these domains, it seems to be the degree of homogeneity (i.e. less ambiguity), which accounts for which forms emerge earlier. Thus, in the domain of the noun, monosyllabic articles (i.e. the rhyming paradigm with the biunique vowel /e/ in the plural including the grammaticized fillers [e(z)]) emerge before fusional and non-diagrammatic plural modification al → aux and portmanteau forms aux/des. Among verb forms, the monosyllabic rhyming forms ont, sont, font, vont, which form a tightly-knit group of highly frequent (semi)-auxiliary verb forms, emerge before the inhomogeneous stemamplified plurals. Finally, prenominal plural liaison develops before preverbal plural liaison because there can be no liaison before the development of subject clitics.

Development of nominal and verbal number in two French-speaking children

405

Notes ∗

1. 2. 3.

4. 5.

6.

7. 8. 9. 10. 11.

I would like to thank W. U. Dressler, Y. Ch. Morin, U. Stephany and M. D. Voeikova for many helpful comments and suggestions, and U. Stephany for her help with the English language. Any errors and other shortcomings are, of course, my own responsibility. On preverbal liaison, see 4.1.2 below. I am grateful to Sophie’s and Emma’s parents for collecting the data and checking the transcription. Emma’s data are more limited than Sophie’s. Emma was generally recorded only twice a month and some of the recordings are very short (especially those at 1;6, 1;7, 2;0; at 1;7 diary notes complement the recordings). Probably, the two types of recorded and diary data are partly responsible for the greater heterogeneity of some of the findings concerning her language development. This is called a ‘formulaic strategy’ by Peters and Menn (1993: 745). See also Peters (1997: 159) and Bates, Dale, and Thal (1995). See 3.2.5. The form aux is an example of gender syncretism occurring in the plural (Biermann 1982: 241). The masculine gender aux indicates plural only with vowelinitial nouns, compare masc.sg. a` l’homme [alOm] pl. aux hommes [ozOm] PREP ART N fused PREP&ART liaison N ‘to the man’ ‘to the men’ with masc. sg. au gar¸con [ogars˜o] pl. aux gar¸cons [ogars˜o] fused PREP&ART N fused PREP&ART N ‘to the boy(s)’. In the feminine gender the plural is indicated throughout by allomorphy or both allomorphy and liaison: fem. sg. a` la, a` l’ - fem. pl. aux, e.g. fem. sg. a` la vache [alavaS ] pl. aux vaches [ovaS ] PREP ART N fused PREP&ART N ‘to the cow’ ‘to the cows’ fem. sg. a` l’amie [alami] pl. aux amies [ozami] PREP ART N fused PREP&ART N ‘to the friend’ ‘to the friends’. Grammaticization is the process by which the child is in the process of restricting a target to grammatically defined categories. Early fillers correspond to purely prosodic or phonological attributes of the target while grammaticized fillers refer to grammatically defined classes of target morphemes such as subject clitics (cf. Peters 2001; Dressler and Kilani-Schoch 2001; Kilani-Schoch and Dressler 2002). On the relation between fillers and determiners in general see also Bassano (2000), Kilani-Schoch and Dressler (2002). The use of the numeral deux ‘two’ is not counted here. These two markers can also be used with (singular) collective nouns. In addition, the labio-velar glide /w/ is deleted. See Gr´egoire (1937: 47) for a similar example.

406

Marianne Kilani-Schoch

12. As far as nouns beginning in a vowel in standard French are concerned, plural and singular dominance seem to account only partially for the distribution of /z/ (or with stopping /d/) deriving from the plural definite article les [lez] vs. /n/ deriving from the masculine singular indefinite article un [œ˜]. In both children’s data, /z/ mainly occurs with plural-dominant nouns (in the input) (e.g. ans ‘years’, yeux ‘eyes’, oreilles ‘ears’, Indiens ‘Indians’) and only with one singular-dominant noun (oiseau ‘bird’), while /n/ is used with both plural-dominant nouns (oreille(s), oeil/yeux) and singular-dominant ones (avion ‘plane’). 13. The lateral liquid /l/ does occur in preverbal liaison, however (e.g. il vient [i-vj˜E] ‘he comes’ vs. il aime [i-lEm]. It is most probable that, for the child, the definite article /l/ before vowel initial nouns (e.g. l’ours ‘the bear’) has the same status as the other initial consonants occurring in the same context (Morin, p.c.). 14. On variability in child language see Taelman and Gillis (2001). 15. In addition the liaison consonant has been devoiced. 16. This morphological analysis implies that devoicing of [z] has applied in Sophie’s form asyeux. It must be mentioned, however, that devoicing is not a common process in this child’s speech. 17. This reanalysis of the lexical item yeux ‘eyes’ is very common in French as shown by the neologism zyeuter ‘to look at’. 18. The set of stem-amplified plurals represents about 550 lemmas, i.e. roughly 79% of subclasses and paradigms of the second conjugation class. 19. In some regional varieties of French, e.g. in Quebec, there is no liaison between 3rd pl. masc. ils and the following verb. 20. This delay is not related to the acquisition of the periphrastic future which is frequent in Sophie’s corpus already at the beginning of morphology proper (2;4). 21. The other interactional meaning ‘come on’ does not occur in the corpus. 22. Questions raised by the low number of stem-amplified plurals are the following: Is this due to language-specific factors, the limited size of data or rather the interactional context of the recordings. A preliminary rough comparison with adult-adult interaction shows that stem-amplified 3rd pers. plural forms are infrequent in face-to-face interactions as well and therefore seem to be context-dependent. In the samples of the GARS corpus of spoken French (Blanche-Benveniste, Rouget, and Sabio 2002), 3rd pers. plural forms represent 0.4% of the total number of words (59,000) only. In the longest sample of 8,000 words considered separately, rhyming plurals represent 0.83% and stem-amplified plurals 0.25% of the total number of words (7% and 2.2% of the total number of verb forms, respectively). 23. Raw numbers are given in tables 20 and 21.

Development of nominal and verbal number in two French-speaking children

407

References Bassano, Dominique 2000 Early development of nouns and verbs in French: Exploring the interface between lexicon and grammar. Journal of Child Language 27: 521–559. Bassano, Dominique, Sabine Laaha, Isabelle Maillochon, and Wolfgang U. Dressler 2004 Early acquisition of verb grammar and lexical development: Evidence from periphrastic constructions in French and Austrian German. First Language 24, 1: 34–70. Bates, Elizabeth, Philip S. Dale, and Donna Thal 1995 Individual differences and their implications for theories of language development. In The Handbook of Child Language, Paul Fletcher and Brian MacWhinney (eds.), 96–151. Oxford: Blackwell. Biermann, Anna 1982 Die grammatische Kategorie Numerus. In Apprehension: Das sprachliche Erfassen von Gegenst¨anden, Hansjakob Seiler and Christian Lehmann (eds.), part I, 229–243. T¨ubingen: Narr. Bittner, Dagmar, Wolfgang U. Dressler, and Marianne Kilani-Schoch (eds.) 2003 Development of Verb Inflection in First Language Acquisition. Berlin: Mouton de Gruyter. Blanche-Benveniste, Claire, and Jean-Pierre Adam 1999 La conjugaison des verbes: virtuelle, attest´ee, d´efective. Recherches sur le fran¸cais parl´e 15: 87–112. Blanche-Benveniste, Claire, Christine Rouget, and Fr´ed´eric Sabio (eds.) 2002 Choix de textes de fran¸cais parl´e. Paris: Champion. Booij, Geert 1996 Inherent versus contextual inflection and the split morphology hypothesis. Yearbook of Morphology 1995, 1–16. Brown, Roger 1973 A First Language. London: Allen & Unwin Ltd. Bybee, Joan 2001 Frequency effects on French liaison. In Frequency and the Emergence of Linguistic Structure, Joan Bybee and Paul Hopper (eds.), 337–360. Amsterdam: Benjamins. Chevrot, Jean-Pierre, and Michel Fayol 2001 Acquisition of French liaison and related child errors. In Research on Child Language Acquisition: Proceedings of the 8th Conference of the International Association for the Study of Child Language, Margareta Almgren, Andoni Barre˜na, Mar´ıa-Jos´e Ezeizabarrena, Itziar Idiazabal, and Brian MacWhinney (eds.), vol. 2, 760–774. Somerville, MA: Cascadilla Press.

408

Marianne Kilani-Schoch

Clark, Eve V. 1985

The acquisition of Romance: With special reference to French. In The Crosslinguistic Study of Language Acquisition, Dan I. Slobin (ed.), vol.1, 687–782. Hillsdale, NJ: Erlbaum. Damourette, Jacques, and Edouard Pichon 1933 Des mots a` la pens´ee: Essai de grammaire de la langue fran¸caise. Vol. III. Paris: D’Artrey. Dressler, Wolfgang U. 2005 Morphological typology and first language acquisition: Some mutual challenges. In Morphology and Linguistic Typology. On-line Proceedings of the Fourth Mediterranean Morphology Meeting, Geert Booij, Emiliano Guevara, Angela Ralli, Salvatore Sgroi, and Sergio Scalise (eds.), 7–20. http://morbo.lingue.unibo.it/mmm/ Dressler, Wolfgang U., and Marianne Kilani-Schoch 2001 On the possible rise and inevitable fall of fillers. Journal of Child Language 28: 250–253. Dressler, Wolfgang U., Marianne Kilani-Schoch, and Sabine Klampfer 2003 How does a child detect morphology? Evidence from production. In Morphological Structure in Language Processing, Harald Baayen and Robert Schreuder (eds.), 391–425. Berlin: Mouton de Gruyter. Dressler, Wolfgang U., Ursula Stephany, Ayhan Aksu-Ko¸c, and Steven Gillis 2007 Discussion and conclusion. In Typological Perspectives on the Acquisition of Noun and Verb Morphology, Sabine Laaha and Steven Gillis (eds.), 67–72. (Antwerp Papers in Linguistics 112.) Antwerp: Universiteit Antwerpen. Gr´egoire, Antoine 1937 L’apprentissage du langage. Paris: Droz. Gillis, Steven 2003 A case study of the early acquisition of verbs in Dutch. In Development ofVerb Inflection in First Language Acquisition, Dagmar Bittner, Wolfgang U. Dressler, and Marianne Kilani-Schoch (eds.), 171–203. Berlin: Mouton de Gruyter. Kilani-Schoch, Marianne 2003 Early verb inflection in French: An investigation of two corpora. In Development ofVerb Inflection in First Language Acquisition, Dagmar Bittner, Wolfgang U. Dressler, and Marianne Kilani-Schoch (eds.), 269–295. Berlin: Mouton de Gruyter. Kilani-Schoch, Marianne, and Wolfgang U. Dressler 2002 Filler + infinitive and pre- and protomorphology demarcation in a French acquisition corpus. Journal of Psycholinguistic Research 2001, 30, 6: 653–685. 2005 Morphologie naturelle et flexion du verbe fran¸cais. T¨ubingen: Narr.

Development of nominal and verbal number in two French-speaking children

409

Koehn, Caroline 1994 The acquisition of gender and number morphology within NP. In Bilingual First Language Acquisition, J¨urgen Meisel (ed.), 29–51. Amsterdam: Benjamins. Lieven, Elena V. M. 1998 Introduction. In Developing a Verb Category: Cross-Linguistic Perspectives, Elena V. M. Lieven (ed.), 627–632. (Linguistics 36, 4). Mayerthaler, Willi 1981 Morphologische Nat¨urlichkeit. Wiesbaden: Athen¨aum. Morel, Eliane ´ 1994 Le traitement des liaisons chez l’enfant: Etudes exp´erimentales. TRANEL 21: 85–95. Morin, Yves Charles 1987 Remarques sur l’organisation de la flexion des verbes fran¸cais. I.T.L. Review of Applied Linguistics 77/78: 13–91. 2003 Remarks on prenominal liaison consonants in French. In Living on the Edge. 28 Papers in Honour of Jonathan Kaye, Stefan Ploch (ed.), 385–400. Berlin: Mouton de Gruyter. 2005 La liaison rel`eve-t-elle d’une tendance a` e´ viter les hiatus? R´eflexions sur son e´ volution historique. Langages 158: 8–23. Morin, Yves Charles, and Jonathan Kaye 1982 The syntactic bases for French liaison. Journal of Linguistics 18: 291– 330. M¨uller, Natascha 1994 Parameters cannot be reset: Evidence from the development of COMP. In Bilingual First Language Acquisition, J¨urgen Meisel (ed.), 235– 269. Amsterdam: Benjamins. Peters, Ann M. 1997 Language typology, prosody and the acquisition of grammatical morphemes. In The Crosslinguistic Study of Language Acquisition, Dan I. Slobin (ed.), vol. 5, 135–197. Mahwah, NJ: Erlbaum. 2001 Filler syllables: What is their status in emerging grammar? Journal of Child Language 28: 229–242. Peters, Ann M., and Lise Menn 1993 False starts and filler syllables:Ways to learn grammatical morphemes. Language 69, 4: 742–777. Slobin, Dan I. 1973 Cognitive prerequisites for the acquisition of grammar. In Studies of Child Language Development, Charles A. Ferguson and Dan I. Slobin (eds.), 175–208. New York: Holt, Rinehart & Winston. 1985 Crosslinguistic evidence for the language-making capacity. In The Crosslinguistic Study of Language Acquisition, Dan I. Slobin (ed.), vol. 2, 1157–1256. Hillsdale, NJ: Erlbaum.

410

Marianne Kilani-Schoch

Stephany, Ursula 1992 Grammaticalization in first language acquisition. Zeitschrift f¨ur Phonetik, Sprachwissenschaft und Kommunikationsforschung 3: 289– 303. 1997 The acquisition of Greek. In The Crosslinguistic Study of Language Acquisition, Dan I. Slobin (ed.), vol. 4, 183–333. Hillsdale, NJ: Erlbaum. 1998 A crosslinguistic perspective on the category of nominal number and its acquisition. In Studies in the Acquisition of Number and Diminutive Marking, Steven Gillis (ed.), 1–23. (Antwerp Papers in Linguistics 95.) Antwerp: Universiteit Antwerpen. 2002 Early development of grammatical number – a typological perspective. In Pre- and Protomorphology: Early Phases of Morphological Development in Nouns and Verbs, Maria D. Voeikova and Wolfgang U. Dressler (eds.), 7–23. Munich: Lincom. Taelman, Helena, and Steven Gillis 2001 Variation in children’s early production of multisyllabic words: The case of truncations. In Early Lexicon Acquisition: Normal and Pathological Development, Sophie Kern (ed.). Lyon: Universit´e Lumi`ere Lyon 2. CD ROM. Vihman, Marilyn V. (ed.) 1999 Cross-linguistic studies of early grammar. In First Steps in Morphological and Syntactic Development: Cross-Linguistic Evidence, Marilyn M. Vihman (ed.), 105–110. (International Journal of Bilingualism 3, 2 and 3). Wijnen, Frank, Masja Kempen, and Steven Gillis 2001 Root infinitives in Dutch early child language: An effect of input. Journal of Child Language 28, 3: 629–660.

Noun plurals in early Palestinian Arabic: A longitudinal case study Dorit Ravid and Rola Farah 1. Introduction Few studies have looked at the acquisition of spoken Arabic among young monolingual speakers in situ, that is, in a situation where both the subjects of the research and those conducting the fieldwork not only know the language natively but also live and work in the country where the language is acquired and used as a first and dominant language. The current study is a longitudinal examination of the development of noun plurals in one young monolingual native speaker of Palestinian Arabic (PA), Tawfik (a boy), between the ages of 1;8 and 2;8. Arabic is a fascinating topic for the investigation of child language development. One reason is because it constitutes the classical case of diglossia, where language users speak one local dialect yet command a written variety learned at school and shared by all Arabic speakers. Recent psycholinguistic and educational studies of Arabic indicate that the impact of diglossia is early and far-reaching (Ibrahim and Aharon-Peretz 2005; Saiegh-Haddad 2004). Another reason is the complexity of inflectional systems in Arabic, which makes their acquisition a challenge for the Arabic-speaking toddler. Finally, few studies have been done on the acquisition of early child Arabic. The current study investigates the acquisition of noun plurals – a topic which has attracted much research in the last decade.

1.1.

Models of processing and acquiring inflectional morphology

Two contrasting models of representation of regular and irregular morphology and of its acquisition provide different accounts for this U-shaped learning. A dual-route model assumes the existence of two distinct processing systems to handle the two facets of morphology (Clahsen 1999; Marcus 2000; Sahin, Pinker, and Halgren 2006): a rote-learning mechanism stores both regular and irregular forms at the onset of acquisition so that the child appears to use them correctly. This is supplanted by a mechanism that applies a concatenation rule to all forms (N stem + plural suffix, V stem + past tense suffix) resulting in overregularization (e.g., mans, knowed). Finally, a mature morphology contains two

412

Dorit Ravid and Rola Farah

systems: One handles regular inflection by abstract, symbolic rule that can be extended to nonce forms, and another stores irregular forms by lexical memory whose retrieval blocks the application of the rule (Halle and Mohanan 1985; Jaeger et al. 1996). Pinker and his associates further suggest that irregular forms are handled by an associative network of subregularities that clusters together minor patterns such as drink / drank, sing / sang and may cause occasional mistakes such as bring / brang under experimental conditions even in adults (Pinker 1999; Pinker and Prince 1994; Pinker and Ullman 2002; Prasada and Pinker 1993). A second account of the acquisition of regular and irregular forms and of its processing in an adult system denies a split in processing mechanisms, and proposes single-route models to handle both types of inflectional morphology. One type of these are listing models, in which all forms are listed in the lexicon and related by phonological and semantic features of various strengths (Bybee 1988, 1995; Stemberger and MacWhinney 1988; Stemberger 1994). Another type of single mechanism is an associative memory system (connectionist model) which regards each entry as a set of features that are shared by many other entries that overlap in their representation. Learning consists of strengthening or weakening connections between the features and is strongly affected by factors of similarity, consistency and frequency (McClelland and Patterson 2002). The network learns to compute the most probable inflected form for any input string by using learning algorithms that capture the statistical regularities between input and output strings. The system can generalize automatically to both regular and irregular new forms that resemble forms it already knows since the same mechanism computes predictable, minor-class, and exceptional forms (Daugherty and Seidenberg 1994; Plunkett and Marchman 1991, 1993; Rumelhart and McClelland 1986). The two models may not have the same explanatory power for inflectional systems based on different principles since the notions “frequent” and “regular” are not necessarily synonymous. English past-tense verbs and noun plurals constitute a case where speakers are exposed to both a regular, highly frequent suffix (-ed, -s) that may be perceived as the rule, as well as to a much smaller number of true exceptions (e.g., mouse/mice, do/did), suppletive forms (go/went), and minor patterns with clusters of forms (blow/blew, grow/grew, know/knew, etc.). Such systems can be accounted for by both models: dual-route models assume a default rule that applies everywhere to symbolic categories unless blocked by idiosyncratic rote-learned items or by items from minor morpho-phonological classes; while connectionist models attribute learning to the changing weights of frequent, transparent, productive, consistent, and regular forms in the changing input to children during the process of language acquisition. Note however that there are other cases, such as the German plural, consisting of phonologically

Noun plurals in early Palestinian Arabic: A longitudinal case study

413

unrelated plural allomorphs, with no single clearly dominant form.Across the literature on German-learning children, and for all Austrian ones described so far, -s plurals in German, claimed to be the default case (Bartke, Marcus, and Clahsen 1995; Clahsen 1999) are neither the first ones to emerge, nor are they the only ones to be over-generalized. Learning German plurals might be better accounted for by single-route or schema-based models, or else by assuming a gradual continuum between fully productive and unproductive plurals (Laaha et al. 2006). Arabic too is a language that does not conform to an English-like regular/irregular split and where the complexity of the system requires a different modeling of plural acquisition (Idrissi, 1997). This paper describes the route taken by a young speaker of Palestinian Arabic in Israel in learning to pluralize nouns, with the view to highlight some facets of Arabic plural acquisition in a spontaneous, naturalistic, longitudinal context.

1.2.

Number in Arabic

Modern Standard Arabic (MSA) is the written variety of the language, common to all literate Arabic speakers in the world, used in the media, in literature, at school, and for all literate activities, and almost exclusively used in its spoken form in the electronic media. It is a unified, codified pan-Arab variety of Arabic, the modern descendant of Classical Arabic (Holes 1995). MSA is not, however, the spoken language acquired by children: Arabic constitutes the classical case of diglossia, which involves a highly divergent and often grammatically more complex literary variety, usually older than the spoken variety, learnt by formal education and used for formal, mostly written purposes (Ferguson 1959).Arabicspeaking children acquire the local dialect at home, and learn MSA in school (Omar 1973; Saiegh-Haddad 2003, 2004), though MSA constitutes a constant alternative to the indigenous vernacular and a source of linguistic information due to its status and omnipresence in all kinds of media and literacy activities (Badry 1983; Holes 1995). The focus of our discussion here is the acquisition of nominal number categories in a Palestinian Arabic dialect spoken in the north of Israel (Levin 1994). Number in Arabic can be characterized as a four-way grammatical system: Singular, dual, collective and plural.

1.2.1. Noun plurals Plural (non-dual) formation in MSA refers to countable entities which are more than two in number and takes two distinct paths: Linear suffixation of a stem

414

Dorit Ravid and Rola Farah

with masculine -u:n (or -i:n) and feminine -a:t, though regular, is restricted to a short list of atypical nouns (and adjectives) such as proper names, transparently derived forms and names of the letters of the alphabet. The second type of pluralization is the widespread classical Semitic device of non-linear root-andpattern affixation, the phenomenon referred to as broken plurals, which involves an internal modification of the singular stem, e.g., kaleb / kila:b ‘dog / s’. Broken plurals are formally represented as patterns (vocalic tiers) interdigitated by root consonants, indicated by C’s. For example, both jma:l ‘camels’ and bwa:b ‘houses’ share the plural pattern CCa:C. There are four shape-defined categories in MSA, including 31 subtypes of broken plurals, formed on literally every canonical noun type in MSA (McCarthy and Prince 1990). This basic split plural system found in MSA applies also, with some modification, in the current spoken dialects, where the sound masculine suffix is –i:n and the sound feminine suffix is –a:t (Holes 1995; Versteegh 1997). Sound plurals in Palestinian Arabic (PA) are formed by attaching to the singular stem one of two suffixes, determined by the gender of the singular nouns, and found in all spokenArabic varieties (Holes 1995: 139).The masculine suffix is -i:n as in falla:H / fallaHi:n ‘farmer / s’. The feminine suffix is -a:t, as in kubba:y / kobaya:t ‘glass / es’. The feminine plural suffix is more productive than the masculine plural suffix, it may also attach to masculine stems (e.g., balo:n / balona:t ‘balloon / s’), and to foreign borrowings that cannot be accommodated by a broken plural pattern. It may thus be considered as the default plural suffix as in other spoken Arabic varieties (Holes 1995). Broken plurals follow the four shape-determined categories assembled in McCarthy and Prince (1990), encompassing the 31 categories described in Wright (1971); however they are governed by the same general dialectal phonetic changes that apply to PA as a whole. The three main phonetic sound changes that are relevant to plural formation are the following (Levin 1994): (1) Short vowels delete in open unstressed syllables, so that the Standard iambic plural pattern CiCa:C is realized as CCa:C in PA (e.g., MSA jamal / jima:l, PA jamal / jma:l ‘camel / s’). Note that i and e are not phonemically distinguished in PA. (2) Long vowels shorten in unstressed syllables, so that Standard plural pattern CaCa:Ci:C is realized as CaCaCi:C in PA. For example, compare MSA maka:ti:b with PA makati:b ‘letters’. (3) The glottal stop tends to delete, so that MSA PaCCa:C is realized as CCa:C (e.g., MSA ba:b / Pabwa:b, PA ba:b / bwa:b ‘house / s’).

Noun plurals in early Palestinian Arabic: A longitudinal case study

415

1.2.2. Duals The dual constitutes a separate inflectional category in Arabic, specifying two of what the noun base designates, e.g., ktab:en ‘book,DUA’. In Classical Arabic, the dual was a grammatical concord category like singular and plural, and entailed obligatory marking across the board on all lexical classes – nouns (and pronouns), adjectives, and verbs. Thus the following translated sentence describing two boys would have dual suffixation (marked by the numeral 2) on each of its components: These2 boys2 were2 close2 friends2 (Blanc 1970). The dialectal, or vernacular, dual, in contrast, is not a concord category, is marked only on nouns, and entails plural agreement. In the dialects, the dual takes the form of the suffix –e:n (or -ay:n) on masculine nouns, and -te:n (or -tay:n) on feminine nouns (Abu-Haidar 1979). Most scholars studying Arabic and its dialects share the view that the dual category is no longer functioning as an obligatory inflection. Despite its morphological form, the vernacular dual is, according to Blanc, “functionally equivalent to nouns preceded by a numeral or another quantifier . . . [It] has . . . become little more than an equivalent of the numeral 2; overt reference to two-ness is no more obligatory . . . than is overt reference to three-ness, four-ness, n-ness” (1970: 43). This view is shared by Cowell (1964), who claims for Syrian Arabic (the dialect of Damascus, close to northern Palestinian Arabic) that the dual inflection is more comparable to numerals than to the plural inflection. The dual is not obligatory in the dialects: a noun does not have to be inflected for the dual every time two of anything is used, and the plural may be used instead, unless it is the very two-ness of the noun that is important. It is possible to designate two-ness by the free numeral two, and even by using the dual noun with the numeral following in apposition, e.g., ktabe:n tne:n ‘book,DUAL two’. Duals, like the free numeral 2, also refer to a small, unspecified number, a small quantity (Blanc 1970). Moreover, most dialects have both a “real” dual (described above) and a pseudo-dual, which is used for paired parts of the body and also for the countable plurals of these words (Versteegh 1997; and compare Hebrew on such pseudo-duals, e.g. Schwarzwald 1996). According to Blanc, the largest, most open sets using the dual morphemes are the Syro-Mesopotamian dialects, of which PA is one.

1.2.3. Collectives Another number category in addition to duals and plurals is that of collectives. In MSA, the collective is a separate morphological category used to refer to uncountable entities or to living things – fruit, vegetables, grains, flowers, grasses, trees, fish, seafood, animals and humans – as an undifferentiated group, e.g.,

416

Dorit Ravid and Rola Farah

samak ‘fish,COLL’ (Cowell 1964; Holes 1995). The three-way singular / plural / collective distinctions are preserved in some of the more conservative Arabic dialects, but in many cases the collective seems to be less and less used, with the plural replacing it in collective contexts. Moreover, there is a trend in the dialects towards the development of an analytic singular / plural distinction by using a free lexeme meaning “one”, “a piece”, “a single example of”, to yield a partitive form from the collective. Another way in which the collective is used towards achieving this distinction is through adding the suffix –a to collective forms to form the singular. Collective noun stems can also occur in the plural, in which case the meaning is individualized and particularized (Abu-Haidar 1979).

1.3. The acquisition of noun plurals in Arabic Studies of plural formation in another Semitic language spoken in the same geographical area, namely Hebrew, have shown that number and gender are marked early on (Berman 1981, 1985; Kaplan 1983; Levy 1980; Ravid 1995), although mastery of all stem changes takes until age 12 (Lavi 2006; Ravid and Schiff, 2009). Observation in the Israeli context has indicated this to be the case in PA as well. However, relatively little is known about children’s acquisition of contemporary Arabic by native-speaking monolingual children. Studies on the acquisition of Arabic include Omar’s (1973) book on Egyptian Arabic, Badry’s (1983, 2005) studies on children’s learning of word-formation in Standard Moroccan Arabic. There are dissertations by Al-Akeel (1998) on syntactic comprehension in Saudi Arabic, and by Idrissi-Bouyahyaoui (1987) on metalinguistic awareness in Arabic children and adults. There is also some work comparing Bedouin children’s narratives to those of adults (Henkin 1996, 1998). In recent works, Saiegh-Haddad (2003, 2004) analyzes the impact of the linguistic distance between PA and written Arabic on children’s reading and writing development. The domain which to date has given rise to substantial psycholinguistic research is that of noun pluralization (Berman and Ravid 2000). Non-developmentally oriented studies on this issue include representation and organization of plurals in Algerian Arabic in non-brain-damaged and aphasic adults (Mimouni, Kehayia & Jarema, 1998) and computer simulations of morphological acquisition in neural network models from a connectionist perspective (Plunkett and Nakissa 1997). The two published studies on the acquisition of plural nouns in spoken Arabic come from Israel. One experimental study is reported in Ravid and Farah (1999), testing sound and broken plurals in 48 children aged 2–5, all native speakers of

Noun plurals in early Palestinian Arabic: A longitudinal case study

417

PA as their only mother tongue from urban, educated families. The participants were asked to give the plural form of 42 noun stimuli in three categories: masculine -i:n suffixed sound plurals (e.g., tabba:x / tabbaxi:n ‘cook / s’); feminine -a:t suffixed sound plurals (e.g., ta:wle / tawla:t ‘table / s’); and broken plurals in 7 formal classes (e.g., sˇobba:k / sˇabbabi:k ‘window / s’; jamal / jma:l ‘camel / s’). Findings revealed that the unmarked form of pluralization was sound feminine plurals (henceforth: SFP) suffixed by -a:t. The category of SFP was found to be learned early on and reach a plateau by age 3, and was moreover the preferred plural option in erroneous responses. The other type of sound plurals suffixed by masculine -i:n was found to be the least preferred option of pluralization in both correct and erroneous responses. So-called “irregular” broken plurals are located between these two extremes, with a learning curve similar to that of sound masculines but also with an increasing presence in the erroneous responses of the oldest age groups. A second experimental study (Ravid and Hayek 2003) focused on the acquisition of three number categories – sound feminine noun plurals, duals, and collectives – in the same dialect of Palestinian Arabic spoken in the north of Israel. The study population consisted of 58 children aged 3;6–4;6, 5–6, 6–7, and 7–8, all monolingual speakers of Palestinian Arabic as their mother tongue. 30 noun stimuli denoting fruits and vegetables familiar to young children were presented in pictures to elicit plural, dual and collective forms. Analysis of the correct results showed two different acquisition patterns – clear development in production of sound feminine nouns and dual nouns, and no increase above the initial 50% in producing collective nouns. Dual contexts elicited constructions preceded by the numeral ‘two’. Collectives and sound feminine plurals tended to replace each other. Against this background, and given the variety of number categories inArabic and the patterning of sound and broken noun plurals in the development of PA reported so far, the aim of the current study was to investigate number categories in child PA in non-experimental settings. This time we focused on the longitudinal development of a single Arabic-speaking child, Tawfik, between the ages of 1;8 and 2;8, when a great deal of inflectional morphology is learned.

2. Methodology At the time of his recordings, Tawfik was a single child, born to well-educated parents living in an urban environment in the north of Israel. He is a monolingual speaker of a northern Israeli dialect of PA. Until the age of 2 he spent most of his time with his grandparents when his parents were at work, and from

418

Dorit Ravid and Rola Farah

the age of 2 he started attending an Arabic-speaking kindergarten. Palestinian Arabic (PA) is the only language Tawfik has been exposed to in his immediate environment. He was audio-recorded approximately once a week in his home by the second author, a native speaker of the same PA dialect as Tawfik. Each recording session lasted between 45–60 minutes, in various everyday situations. Most of the time Tawfik’s mother was present at the recording session, in other cases his father as well. Every recording was transcribed on the same day. For the purposes of our analysis we focused on 37 recording sessions from the time Tawfik was 1;8.6 until he was 2;8. Recording started at the age of 1;8.6, when Tawfik started producing 2-word utterances. In the first two weeks of recording Tawfik produced no plural noun forms; his first plural forms occurred in the third recording session, well within the first recording period (see below). The following measures were counted in each recording: – – – – – – – – – – –

Number of utterances produced by Tawfik. Singular noun types. Singular noun tokens. Accumulating singular noun types. Plural noun types, correct and incorrect, including Sound masculines Sound feminines Broken plurals, by category Duals Collectives Accumulating plural noun types (as above).

3. Findings This paper traces the developmental distribution of plural nouns in Tawfik’s speech in the year of recording. While type frequency is certainly important for the consolidation of different classes and categories, the occurrence of tokens at this early stage of learning an inflectional system is critical (Nicoladis, Palmer and Marentette 2007). Therefore, this chapter analyzes both plural tokens and types. Table 1 first presents general information on each of the 37 sessions: number of utterances produced by Tawfik, singular tokens, and plural tokens by number category. It shows the following: The number of utterances by Tawfik is about 120 per session, with fewer utterances on the average at the beginning of the year of recording, and more utterances towards age 2;8. The number of singular

Noun plurals in early Palestinian Arabic: A longitudinal case study

419

noun tokens per session is about 40, closer to 30 at the beginning and climbing higher towards the end of the study – indicating the growth of Tawfik’s nominal lexicon. Noun plurals, however, are very few, about 10%–20% of the overall number of nouns produced each time (i.e., singulars and plurals together). They start appearing around age 1;9, and consist almost entirely of sound feminines and broken plurals. No sound masculines occur in Tawfik’s speech (except a single one close to the end of the study period). Collectives and duals are few (note that only spontaneous productions were included in these counts). Since we are dealing with a large body of data, and in order to permit a more coherent examination of the development of nouns in general and noun plurals in particular, the 37 recording sessions were further divided for the purposes of analysis into six periods, each approximately equivalent to two months, presented in table 2 below. We will analyze the developmental changes occurring during each of these six periods, as a window on Tawfik’s morphological and lexical development.

3.1. Tawfik’s singular lexicon in development Before embarking on the main topic of this article, it is interesting to examine the development of the noun lexicon in PA in terms of Tawfik’s singular noun types and tokens. Since the acquisition of inflectional classes and lexical development are inherently related, this examination will also highlight developmental aspects related to the occurrence of plural nouns. The full list of singular nouns produced by Tawfik between 1;8 and 2;8 appears in Farah (2001). Table 3 shows the new singular noun types Tawfik produced in each of the 6 periods. For the purposes of this chapter, we will focus on some relevant aspects of this list. First, note how Tawfik assembles his noun lexicon: In the first period (1;8;6–1;10.17), he has 98 singular noun types.These include, inter alia, concrete inanimate count nouns such as seki:ne ‘knife’, taxet ‘bed’, tu:te ‘strawberry’, and mafak ‘screwdriver’, as well as uncountable nouns such as doxa:n ‘smoke’, `aku:b ‘(a kind of) vegetable’ and `atme ‘darkness’. He also has, in this period, some animal names such as ´arnab ‘rabbit’, ´asad ‘lion’, dob ‘bear’, and fi:l ‘elephant’, and some people with designated roles (Ravid 2006), such as mama ‘mommy’, m`alme ‘teacher’, s¸ayad ‘hunter’, and xu:re ‘priest’. In the second period of analysis (1;10;24–2;0.12) 72 new singular types join Tawfik’s noun lexicon, among them more concrete count nouns such as tofa:©a ‘apple’, ra:dyo ‘radio’, xa:tem ‘ring’ and bara:d ‘refrigerator’, uncountable nouns such as labane ‘labane (a kind of yoghurt)’ or tra:b ‘sand’, with more animal names (bat¸a ‘duck’, fara:ˇse ‘butterfly’), and people’s roles (dokto:r ‘doctor’, boli:s ‘police-

420

Dorit Ravid and Rola Farah

Table 1. Number of utterances, singular tokens, sound masculine and sound feminine plural tokens, broken plural tokens, collective tokens and dual tokens per session, by age Age Utterances Singulars Masc. Plur. Fem. Plur. Brken Plur. Collect. Duals 1;8.6 109 43 0 0 0 0 0 1;8.20 139 57 0 0 0 0 0 1;9 99 34 0 2 3 0 0 1;9.17 103 31 0 2 3 0 0 1;10.10 131 55 0 3 0 1 0 1;10.17 112 33 0 4 5 0 0 1;10.24 128 42 0 1 3 1 0 1;11.5 126 33 0 4 2 1 1 1;11.14 131 53 0 1 2 0 0 1;11.21 170 32 0 3 2 0 0 1;11.28 150 37 0 1 3 5 0 2;0.5 120 29 0 3 3 0 0 2;0.12 129 31 0 3 3 2 0 2;0.19 135 33 0 2 4 0 0 2;0.26 163 105 0 1 4 0 0 2;1.2 100 57 0 3 3 0 0 2;1.9 133 81 0 3 6 0 0 2;1.16 112 67 0 1 6 0 0 2;2 123 60 0 3 2 2 0 2;2.7 95 36 0 2 2 1 0 2;2.17 112 53 0 2 5 1 0 2;2.27 120 57 0 1 5 2 0 2;3.7 102 54 0 2 3 1 0 2;3.16 92 42 0 2 1 1 0 2;3.28 113 49 0 2 2 1 0 2;4.10 152 51 0 0 5 1 0 2;4.17 139 49 0 1 3 0 0 2;4.27 148 44 0 2 1 2 0 2;5.7 130 38 0 2 5 2 0 2;5.17 195 51 0 3 3 4 0 2;5.28 173 61 0 4 4 0 0 2;6.8 149 56 0 4 4 0 0 2;6.16 133 66 0 3 3 3 0 2;6.27 145 63 1 1 8 1 0 2;7.7 169 92 0 1 1 0 1 2;7.17 133 37 0 3 12 1 0 2;8 138 63 0 4 6 0 0

Noun plurals in early Palestinian Arabic: A longitudinal case study

421

Table 2. The recording year divided into six periods Recording sessions 1–6 7–13 14–19 20–25 26–31 32–37

Period 1 2 3 4 5 6

Child’s age 1;8;6–1;10.17 1;10;24–2;0.12 2;0;19–2;2 2;2;7–2;3.28 2;4;10–2;5.28 2;6;8–2;8

Table 3. A quantitative examination of Tawfik’s singular nouns in development Recording sessions 1–6 7–13 14–19 20–25 26–31 32–37

Number of new singular types 98 72 77 41 39 46

Child’s age 1;8.6–1;10.17 1;10.24–2;0.12 2;0.19–2;2 2;2.7–2;3.28 2;4.10–2;5.28 2;6.8–2;8

man’, and ra:`i: ‘shepherd’. There are a few temporal and abstract-imageable nouns, such as kaf ‘blow’ and `i:d ‘holiday, festival’ (Ravid 2006). By the third period (2;0;19–2;2)Tawfik has 77 new singular nouns, expanding his lexicon to specific vegetables (bando:ra ‘tomato’, xya:ra ‘cucumber’) and household objects (ˇsarˇsaf ‘sheet’, maxade ‘pillow’, kamara ‘camera’, la:mba ‘lamp’) as well as place names (©adi:re ‘cowshed’, mostaˇsfa ‘hospital’), and many animals’ names (namle ‘ant’, temsa:© ‘crocodile’, s¸ar¸sur ‘cockroach’). The fourth period brings 41 singular noun types, with more and more specific nouns such as tre:n ‘train’, ©ama:m ‘shower’, and sya:g ‘fence’; people with specific roles such as `aru:s ‘bride’, `ari:s ‘bridegroom’, ´ami:re ‘princess’, and qazam ‘dwarf’; and some more abstract-imageable nouns such as dowe:ra ‘circle’, moraba’ ‘rectangle’, and motalat ‘triangle’. The 5th period of recording (2;4;10–2;5.28) brings 39 singular types with more people’s roles such as malek ‘king’ and wazi:r ‘minister’, heavenly bodies such as ´amar ‘the moon’, negme ‘star’, event nouns (´oros ‘wedding’, ©afle ‘party’, shoVol ‘work’), and specific animals’names (Vo:re:la ‘gorilla’, dayna¸su:r ‘dinosaur’, xanzi:r ‘pig’). Finally, the sixth period (2;6;8–2;8), has 46 singular noun types as diverse as dafa:y ‘stove, oven’, peˇsre ‘peel’, ki:s ‘bag’, xa:l ‘uncle’, Vora:b ‘raven’, ¸tawu:s ‘peacock’, bu:me ‘owl’, sˇames ‘sun’ and maktabe ‘library’. Table 3 presents a quantitative summary of Tawfik’s singular noun development.

422

Dorit Ravid and Rola Farah

Table 4. The distribution of noun plural categories in Tawfik’s plural tokens (both correct and incorrect) over the 6 periods Period 1 2 3 4 5 6

Child’s age 1;8–1;10 1;10–2;0 2;0–2;2 2;2–2;3 2;4–2;5 2;6–2;8

Plural tokens 23 44 40 34 42 57

SM 0 0 0 0 0 2%

SF 48% 36% 33% 32% 29% 28%

Broken 48% 42% 62% 53% 50% 59%

Dual 0 2% 0 0 0 2%

Coll. 4% 20% 5% 15% 21% 9%

Against this background indicating rich and variegated lexical development, and using the classification into the six periods, we now focus on the distribution of the noun plurals in Tawfik’s production across the year of recording.

3.2. A developmental analysis of Tawfik’s plural tokens Table 4 below presents the percentage distribution of morphological categories in Tawfik’s noun plural tokens in the 6 recording periods over the recording year: Sound masculines, sound feminines, broken plurals, dual nouns and collective nouns. Note that table 4 presents both correct and incorrect forms. Table 4 shows that Tawfik had 240 plural tokens, the most frequent of which was the broken plural category, e.g., correct sna:n ‘teeth’for singular sen ‘tooth’, at the age of 2;2. The second most frequent category in this period was the sound feminine, e.g., correct faraˇsa:t ‘butterflies’ for singular fara:ˇse ‘butterfly’, at the age of 1;9. Collectives (e.g., sˇo:k ‘thorns’, at the age of 1;10.10) were few, duals (e.g., ¸ta:bte:n ‘balls,DUAL’, age 2;7;7) negligible, and there were no sound masculine plurals at all until the last period, when at the age of 2;6;27 Tawfik produced his single example of a sound masculine plural – s¸ayya:di:n ‘hunters’. Many of Tawfik’s forms are overregularizations (e.g., sound feminine ¸tabla:t for correct broken plural ¸tbu:le ‘drums’, from singular ¸tabel ‘drum’, at the age of 1;9.17). Table 5 presents the percentages of correct noun plurals out of the total number of plurals in each morphological category, during the 6 recording periods. Later on, Tawfik produces irregularizations, i.e., erroneous broken plurals where regular sound feminines are required (see table 5 below). Table 5 makes it clear that many sound feminine nouns in each of the 6 periods are incorrect. They are, in fact, over-regularizations, mostly of broken forms, e.g., besa:t for correct broken plural besas ‘cats’from singular bese ‘cat’, age 2;6.8. Most of the broken plural forms and all of the other plural forms were correct, as will be discussed below.

Noun plurals in early Palestinian Arabic: A longitudinal case study

423

Table 5. Raw numbers of all plural noun forms in the five morphological categories and percentage of correct plural forms in each category, by recording period Child’s age 1;8–1;10

Plural tokens 23

1;10–2;0

44

2;0–2;2

40

2;2–2;3

34

2;4–2;5

42

2;6–2;8 Total

57

SM 0 0 0 0 0 0 0 0 0 0 1 100%

SF 11 27% 16 63% 13 62% 11 64% 12 83% 16 69%

Broken 11 100% 18 100% 25 84% 18 78% 21 91% 34 100%

Coll. 1 100% 9 100% 2 100% 5 100% 9 100% 5 100%

Duals 0 0 1 100% 0 0 0 0 0 0 1 100%

Table 6. Raw numbers of all plural types in the five morphological categories (both correct and incorrect) over the 6 periods Child’s age 1;8–1;10 1;10–2;0 2;0–2;2 2;2–2;3 2;4–2;5 2;6–2;8 Total

Plural types 18 27 21 19 16 20 121

SM 0 0 0 0 0 1 1

SF 10 12 6 7 6 9 50

Broken 7 8 14 8 5 6 48

Coll. 1 6 1 4 5 3 20

Duals 0 1 0 0 0 1 2

Since broken plural forms constitute the majority of the noun plurals produced by Tawfik, we analyzed their internal distribution into broken plural categories. The most frequent broken noun plural category was CCa:C (33%), e.g., kla:b ‘dogs’. This was followed by a number of other broken plural categories such as CCu:C (8.5%), e.g., kfu:f ‘gloves’, and CCu:Ce (1.7%), e.g., ¸tbu:le ‘drums’, and CuCuC (0.9%), e.g., kutub ‘books’. All in all, the 127 broken plural tokens contained 14 broken plural categories (see Farah 2001 for details).

3.2.1. Tawfik’s lexical development as reflected in plural types To complement the plural token analysis, table 6 presents another aspect of the development of Tawfik’s nominal lexicon – his new plural types in each recording period.

424

Dorit Ravid and Rola Farah

The type analysis first shows that the distribution of plural types closely follows that of plural tokens: most of the new plurals in all 6 periods are either sound feminines or broken plurals. Table 6 indicates how early on inflectional morphology emerges in Arabic: one-third of Tawfik’s plural types in the first period have corresponding singulars (e.g., the words for stones, snakes and men). Another interesting observation is the prominence of plural-dominant items in Tawfik’s early nominal lexicon – e.g., thorns, bananas, flowers, fish, socks, shoes, fingernails, teeth, eggs, candies, bells, ants, gloves, apples. Nouns which naturally occur in groups of similar items might be the platform for the understanding of plurality. Future studies are necessary to determine the exact relationship between CDS and children’s output in learning Arabic plurals.

3.3.

Error patterns

Over the year of recording, Tawfik produced 35 erroneous noun plural forms out of a total of 240 tokens, that is, 14.58% of his noun plural forms. Theoretically speaking, for each plural morphological category there were a number of possible error types. For example, erroneous dual forms can theoretically take sound masculine, sound feminine, broken or collective forms. In reality, table 5 shows that the only errors Tawfik made were in the categories of sound feminines and broken plurals, which seem to be his only productive plural categories. For the sound feminine category the possible error types were: – – – –

Sound masculine Broken plural Dual Collective

For the broken plural category, the possible error types were the following: – – – – –

Sound masculine Sound feminine Another broken plural form Dual Collective

Table 7 presents the actual error types in the 6 recording periods. For each of the two plural categories with errors we present the actual number of erroneous forms in each period and its error type.

Noun plurals in early Palestinian Arabic: A longitudinal case study

425

Table 7. The two productive noun plural categories in which plural errors occurred Child’s age

1;8–1;10 1;10–2;0 2;0–2;2 2;2–2;3 2;4–2;5 2;6–2;8

Erroneous sound feminines Erroneous Broken forms Plurals 0 0 0 3 100% 0 0

Erroneous broken plurals Erroneous Sound Fem forms 7 100% 6 100% 10 60% 4 100% 4 4 100%

Other Broken Pl

40% 50%

Table 7 shows that the overwhelming majority of errors were in the broken plural category, and in most of the cases there was overregularization towards the sound feminine category, e.g., sound feminine seki:na:t for correct broken plural sakaki:n ‘knives’, from singular sek:ine ‘knife’, at the age of 2;2;27. However, in the 3rd , 4th and 5th periods Tawfik also irregularized1 from the sound feminine towards the broken plural category (e.g., broken Vna:m for correct sound feminine Vanama:t ‘sheep,Pl’ from singular Vaname ‘sheep’, at age 2;2;17); and within the broken category from one broken plural type to another (e.g., broken plural dku:ne for correct broken plural dakaki:n ‘shops’, singular form doka:ne ‘shop’, at age 2;2;7). The double 50% figure in the broken plural category in the 5th period indicates that Tawfik irregularized within the broken plural category to two different broken types. Taken together, these and less expected new errors in the irregular direction indicate that in his 3rd year, Tawfik is acquiring knowledge not only about sound feminines, but also about the numerous yet less systematic patterns of broken plurals (see a similar trend in the development of Hebrew abstract nominals in Ravid and Avidor 1998).

4. Discussion This study highlights children’s perception of noun reference through their learning of how to express number distinctions in a sparsely investigated language that challenges its speakers by marking number in multiple and often opaque ways. To the best of our knowledge, this is the first longitudinal case study tracing the route an Arabic-speaking child during one year (1;8;6–2;8) takes in learning about the noun plural category. Our analysis indicates that the first incorrect plurals – indicating the emergence of the singular/plural distinction – occur by the time he is 1;9. The results of this initial analysis indicate that two major

426

Dorit Ravid and Rola Farah

plural categories are actually active, or productive, in Tawfik’s early lexicon, as reflected in his erroneous forms: the broken plural category, which occurs most frequently in his productive speech, and the sound feminine category, which lags somewhat behind but makes up in being the target of most erroneous forms. All other plural forms – sound masculines, duals and collectives – range between 4%–22% of all plural forms and are always correct. It thus seems that at this developmental point, most developmental “work” is taking place in the sound feminine and the broken plural categories alone. A number of interesting observations emerge when comparing these results to the cross-sectional study described in Ravid and Farah (1999). In that study, we examined the acquisition of two types of noun plurals – sound (feminine and masculine) and broken – by PA-speaking children aged 2;3–6;2, in a structured experiment where children had to give the plural form of a given singular noun. We found two different acquisition patterns for the three plural categories: In the sound feminine noun plurals, learning was already complete in the second youngest group, the 3-year-olds. In contrast, in the other two types – sound masculines and broken plurals – learning was more gradual and was still taking place in the oldest group, the 5-year-olds. We concluded in our 1999 study that “it is not the sound pluralization device in general that is easier than the broken plural, but rather the feminine sound plural that is easier to learn than both the sound masculines and the broken plurals”. The current study supports the status of the sound feminine as a central noun plural category in PA, given its widespread distribution inTawfik’s speech and the fact that the overwhelming majority of his erroneous plural forms are in sound feminine form. Two additional important observations emerge in the current paper. First of all, this study presents a completely different picture of the sound masculine category. In our 1999 paper we noted that the reason which makes sound feminine plurals so easy to learn might be either semantic or structural in nature. In the cross-sectional study, the masculine plural type was the most coherent of all plural types both semantically and structurally, denoting human agent nouns and mostly following the singular agentive pattern CaCCa:C to which the plural suffix -i:n is attached with no modification of the stem (except for loss of vowel length). Moreover, both masculine and feminine sound plurals share linear suffixation, which should have made them equivalent. Yet sound masculines lagged even behind broken plurals. This was most probably due to the semantic restriction (agentive humans only) on sound masculine plurals. The central role of sound feminines in early child Arabic finds support in Ravid and Hayek’s (2003) experimental study. Of the three constructions investigated in that study, sound feminines were the most accessible in both structural and semantic terms. They were very rarely left unmarked for the plural category

Noun plurals in early Palestinian Arabic: A longitudinal case study

427

or erroneously took a periphrastic expression of number. Despite the potential of exchanging a masculine suffix –i:n for the required feminine suffix –a:t on sound plurals, this rarely happened. The current paper emphasizes the enormous importance of examining language development in a natural setting: Tawfik used only a single sound masculine plural form out of 240 throughout this year of intensive and extensive acquisition. It seems that linguistic input in the sound masculine category is restricted (perhaps for the reasons outlined above) and that sound masculines might be theoretically easy to learn but are actually non-existent in early Arabic child language. The dual-route model, which splits morphological forms into ‘regular’ and ‘irregular’ would not be supported by this finding: both sound feminines and sound masculines are ‘regular’ in the same sense – they are both constructed of a stem and a suffix, and the semantics of the application is clear. However, according to the dual-route model, regular morphology is not supposed to be subject to frequency effects; but according to our findings, ‘regular’ sound masculine nouns might be subject to such effects. Finally, the facts presented in this paper highlight the centrality of the broken plural category in PA. Despite its apparent ‘difficulty’ as a root-and-pattern category which necessitates root extraction and pattern application in each plural formation, this is the most pervasive category in Tawfik’s early lexicon. Again, this finding is supported by Ravid and Hayek (2003), where erroneous broken plurals occurred only in the older age groups, replacing collectives and sound feminines (e.g. *mwa:z ‘bananas’ for collective mo:z and sound feminine plural moza:t). The challenge faced by Arabic-speaking children in their acquisition of broken plurals is lexical and morphological in nature: in order to make useful predictions about the plural forms of singular items, children should be able to organize their nominal lexicon into broken plural patterns; this is only possible given many encounters with numerous and varied nouns. Our analysis of Tawfik’s acquisition of plural nouns in Palestinian Arabic goes counter the simplistic split between ‘regular’ and ‘irregular’ morphological forms, as broken plurals are highly ‘irregular’ in their formation and diversity, yet constitute a critical part of early acquisition in Arabic. Moreover, these findings underscore the powerful impact of target-language typology on the process of acquisition: What may seem ‘irregular’ and ‘difficult’ to speakers of one language might be in fact easily accessible to children who are very early on sensitive to the ‘typological imperatives’ of their language.

428

Dorit Ravid and Rola Farah

Notes 1. We use the term ‘irregularization’ for this type of plural operation, since the regular category is the feminine sound, while broken plurals, as explained above, consist of a variety of ‘irregular’ patterns.

References Abu-Haidar, Farida 1979 A Study of the Spoken Arabic of Baskinta. London: Brill. Al-Akeel, Abdulrahman I. 1998 The acquisition of Arabic language comprehension by Saudi children. Ph. D. diss. University of Newcastle upon Tyne, UK. Badry, Fatima 1983 Acquisition of lexical derivational rules in Moroccan Arabic: Implications for the development of standard Arabic as a second language through literacy. Ph. D. diss. UC Berkeley. 2005 Acquisition of Arabic word formation: A multi-path approach. In Perspectives on Arabic Linguistics XVII–XVIII: Papers from the Seventeenth and Eighteenth Annual Symposia on Arabic Linguistics, Mohammad T. Alhawary and Elabbas Benmamoun (eds.), 243–272. Amsterdam: John Benjamins. Bartke, Susanne, Gary F. Marcus, and Harald Clahsen 1995 Acquiring German noun plurals. In Proceedings of the 19th Annual Boston University Conference on Language Development, Dawn MacLaughlin and Susan McEwen (eds.), vol. 1, 60–69. Boston: Cascadilla Press. Berman, Ruth A. 1981 Children’s regularizations of plural forms. Stanford Papers and Reports on Child Language Development 20: 34–44. 1985 Acquisition of Hebrew. In The Crosslinguistic Study of Language Acquisition, Dan I. Slobin (ed.), vol. 1, 255–371. Hillsdale, NJ: Erlbaum. Berman, Ruth A., and Dorit Ravid 2000 Acquisition of Israeli Hebrew and Palestinian Arabic: A review of current research. Hebrew Studies 41: 7–22. Blanc, Haim 1970 Dual and pseudo-dual in Arabic dialects. Language 46: 43–55. Bybee, Joan 1988 Morphology as lexical organization. In Theoretical Morphology: Approaches in Modern Linguistics, Michael Hammond and Michael Noonan (eds.), 119–141. San Diego: Academic Press.

Noun plurals in early Palestinian Arabic: A longitudinal case study 1995

429

Regular morphology and the lexicon. Language and Cognitive Processes 10: 425–455.

Clahsen, Harald 1999 Lexical entries and rules of language: A multidisciplinary study of German inflection. Behavioral and Brain Sciences 22: 991–1060. Cowell, Mark W. 1964 A Reference Grammar of Syrian Arabic (Based on the Dialect of Damascus). Washington, D.C.: Georgetown University Press. Daugherty, Kim G., and Mark S. Seidenberg 1994 Beyond rules and exceptions: A connectionist approach to inflectional morphology. In The Reality of Linguistic Rules, Susan D. Lima, Roberta L. Corrigan, and Gregory K. Iverson (eds.), 353–388. Amsterdam: John Benjamins. Farah, Rola 2001 The acquisition of noun plurals in Palestinian Arabic:A case study. MA thesis. Department of Communications Disorders, TelAviv University, Israel. [in Hebrew] Ferguson, Charles A. 1959 Diglossia. Word 15: 325–340. Halle, Morris, and K. P. Mohanan 1985 Segmental phonology of Modern English. Linguistic Inquiry 16: 57– 116. Henkin, Roni 1996 Negev Bedouin and sedentary Palestinian narrative styles. In Israel Oriental Studies XVI: Studies in Modern Semitic Languages, Shlomo Izre’el and Shlomo Raz (eds.), 169–191. New York: Brill. 1998 Narrative styles of Palestinian Bedouin adults and children. Pragmatics 8: 47–78. Holes, Clive 1995 Modern Arabic: Structures, Functions and Varieties. London: Longman. Ibrahim, Rafiq, and Judith Aharon-Peretz 2005 Is literary Arabic a second language for native Arab speakers?: Evidence from a semantic priming study. Journal of Psycholinguistic Research 34: 51–70. Idrissi, Ali 1997 Noun formation in Arabic. Current Issues in Linguistic Theory 153: 123–145. Idrissi-Bouyahyaoui, Bouchaib 1987 Metalinguistic awareness inArabic speaking literate and illiterate children and adults: A psycholinguistic study. Ph. D. diss. University of Edinburgh, UK.

430

Dorit Ravid and Rola Farah

Jaeger, Jeri J., Alan H. Lockwood, David L. Kemmerer, Robert D. Van Valin, Jr., Brian W. Murphy, and Hanif G. Khalak 1996 A positron emission tomographic study of regular and irregular verb morphology in English. Language 72: 451–497. Kaplan, Dafna 1983 Order of acquisition of morpho-syntactic elements in Hebrewspeaking nursery schoolers. MA thesis. Dept. of Communications Disorders, Tel Aviv University. [in Hebrew] Laaha, Sabine, Dorit Ravid, Katharina Korecky-Kr¨oll, Gregor Laaha, and Wolfgang U. Dressler 2006 Early noun plurals in German: Regularity, productivity or default? Journal of Child Language 33: 271–302. Lavi, Naama 2006 The route to acquiring noun plurals in Hebrew-speaking children. MA thesis. Dept. of Communications Disorders, Tel Aviv University [in Hebrew]. Levin, Arie 1994 The grammar of the Arabic dialect of Jerusalem. Jerusalem: The Hebrew University. [in Hebrew] Levy, Yonata 1980 The acquisition of gender. Ph. D. diss. The Hebrew University, Jerusalem. [in Hebrew] Marcus, Gary F. 2000 Children’s overregularization and its implications for cognition. In Models of Language Acquisition: Inductive and Deductive Approaches, Peter Broeder and Jaap Murre (eds.), 154–176. Oxford: Oxford University Press. McCarthy, John J., and Alan S. Prince 1990 Foot and word in prosodic morphology: The Arabic broken plural. Natural Language and Linguistic Theory 8: 209–283. McClelland, James L., and Karalyn Patterson 2002 Rules or connections in past-tense inflections: What does the evidence rule out? Trends in Cognitive Sciences 6: 465–72. Mimouni, Zohra, Eva Kehayia, and Gonia Jarema 1998 The mental representation of singular and plural nouns in Algerian Arabic as revealed through auditory priming in agrammatic aphasia patients. Brain and Language 61: 63–87. Nicoladis, Elena, Andrea Palmer, and Paula Marentette 2007 The role of type and token frequency in using past tense morphemes correctly. Developmental Science 10: 237–254. Omar, Margaret K. 1973 The Acquisition of Egyptian Arabic as a Native Language. The Hague: Mouton.

Noun plurals in early Palestinian Arabic: A longitudinal case study Pinker, Steven 1999

431

Words and Rules: The Ingredients of Language. London: Weidenfeld & Nicolson. Pinker, Steven, and Alan Prince 1994 Regular and irregular morphology and the psychological status of rules of grammar. In The Reality of Linguistic Rules, Susan D. Lima, Roberta L. Corrigan, and Gregory K. Iverson (eds.), 321–351. Amsterdam: John Benjamins. Pinker, Steven, and Michael T. Ullman 2002 The past and future of the past tense. Trends in Cognitive Sciences 6: 456–63. Plunkett, Kim, and Ramin Charles Nakissa 1997 A connectionist model of the Arabic plural system. Language and Cognitive Processes 12: 807–836. Plunkett, Kim, and Virginia A. Marchman 1991 U-shaped learning and frequency effects in a multi-layered perceptron: Implications for child language acquisition. Cognition 38: 43–102. 1993 From rote learning to system building: acquiring verb morphology in children and connectionist nets. Cognition 48: 21–69. Prasada, Sandeep, and Steven Pinker 1993 Generalization of regular and irregular morphological patterns. Language and Cognitive Processes 8: 1–56. Ravid, Dorit 1995 Language Change in Child and Adult Hebrew: A Psycholinguistic Perspective. New York: Oxford University Press. 2006 Semantic development in textual contexts during the school years: Noun scale analyses. Journal of Child Language 33: 791–821. Ravid, Dorit, and Avraham Avidor 1998 Acquisition of derived nominals in Hebrew: Developmental and linguistic principles. Journal of Child Language 25: 229–266. Ravid, Dorit, and Rola Farah 1999 Learning about noun plurals in early Palestinian Arabic. First Language 19: 187–206. Ravid, Dorit, and Lubna Hayek 2003 Learning about different ways of expressing number in the development of Palestinian Arabic. First Language 23: 41–63. Ravid, Dorit, and Rachel Schiff 2009 Morpho-phonological categories of noun plurals in Hebrew: A developmental study. Linguistics 47 (1): 45–63. Rumelhart, David, and James L. McClelland 1986 On learning the past tenses of English verbs: Implicit rules or parallel distributed processing? In Parallel Distributed Processing: Explorations in the Microstructure of Cognition, James L. McClelland,

432

Dorit Ravid and Rola Farah

David Rumelhart, and the PDP Research Group (eds.), 216–271. Cambridge, Mass: The MIT Press. Sahin, Ned, Steven Pinker, and Eric Halgren 2006 Abstract grammatical processing of nouns and verbs in Broca’s Area: Evidence from fMRI. Cortex 42: 540–562. Saiegh-Haddad, Elinor 2003 Linguistic distance and initial reading acquisition: The case of Arabic diglossia. Applied Psycholinguistics 24: 431–451. 2004 The impact of phonemic and lexical distance on the phonological analysis of words and pseudo words in a diglossic context. Applied Psycholinguistics 25: 495–512. Schwarzwald, Ora R. 1996 Words with +ayim endings in Hebrew. In Studies in Hebrew and Jewish Languages Presented to Shelomo Morag, Moshe Bar-Asher (ed.), 341–358. Jerusalem: Hebrew University. [in Hebrew] Stemberger, Joseph P. 1994 Rule-less morphology at the phonology-lexicon interface. In The Reality of Linguistic Rules, Susan D. Lima, Roberta L. Corrigan, and Gregory K. Iverson (eds.), 147–169. Amsterdam: John Benjamins. Stemberger, Joseph P., and Brian MacWhinney 1988 Are inflected forms stored in the lexicon? In Theoretical Morphology: Approaches in Modern Linguistics, Michael Hammond and Michael Noonan (eds.), 101–116. San Diego: Academic Press. Versteegh, Kees 1996 The Arabic Language. Edinburgh: Edinburgh University Press. Wright, William 1971 A Grammar of the Arabic Language. Cambridge: Cambridge University Press.

Subject Index Abbreviations for language codes Cr Es Fi Fr Ge Gk

Croatian Estonian Finnish French German Greek

abessive, Es 111–112, 114, 118, 134, 145, Fi 54, 65–66, 69, 71 ablative, Es 112, 114, 117, 134, 139, Fi 52–53, 62–65, 78, 80, 83, 86–87, Tu 15, 17, 19, 32–33, 35, 37, 43 absolutive, Yu 93, 98, 102, 104 accusative (see also object, direct), Cr 154–155, 163–167, 169, 171–172, 174–175, Fi 51, 55–57, 60–62, 68, 75, 79, 81, 86, Ge 271–273, 283, 285–287, 289–291, 293, Gk 5, 220, 222–223, 228–234, 236–247, 249, 251–256, Ru 3, 9, 180–190, 195–196, 203, 205, Tu 15–17, 19, 24, 29–37, 41–42, 44–45 addition, 2, Es 112, Fr 383, Gk 222, It 308, Sp 341 addressee, Ru 186–187 adequacy, system/typological, Ge 268 adequate/correct forms/use (see also errors), Cr 174, Es 124, 132, 142–143, Fr 379, 382–383, 389–390, Ge 275, 277–279, 282, 284–291, 293–294, Gk 240–243, 248–249, 252, 255–256, It 317–318, 321–324, 326, 331–332, 334–335, PA 411, 417, 422–423, 426, Ru 181–184,

It PA Ru Sp Tu Yu

Italian Palestinian Arabic Russian Spanish Turkish Yucatec Maya

195, 198, 203, 205, 207, Sp 357–358, 362, 366, Tu 30–31, 33, 36, 41–42, Yu 96–97 adessive, Es 112, 114, 117, 122, 134, 138–139, 146, Fi 50, 52–53, 55, 63–65, 68–70, 78–80, 83–87 adjective, Cr 153, 176, Es 111, 118, Fi 49, Fr 373–375, 385, Ge 265, 275, 280, 284–289, 291, 293–294, Gk 222–223, 240, 253, It 304, 306, 309–310, 312–322, 326–328, 335–336, PA 414–415, Sp 10, 341–345, 348–353, 361–362, 364–365, Tu 15–18, 36, 43, Yu 93–94, 103, 108 demonstrative, Fr 373, 379, Ge 286, It 319 adult-directed speech, 1, Es 123, Fi 70, Gk 9, 220, 225, 227, 233–237, Ru 203 adverb (see also case, adverbial; case, local; deictic forms; locative), Fi 7, 54, It 313, 318–319, 322, Tu 17, 37 deictic, Es 111, 129–130, Gk 238 local, Es 119–120, 129–131, 134, Fi 52–53, 56–57, 62–65, 69, 75–76, 86, Gk 238, It 312

434

Subject Index

affix/affixation (see also ending/ inflectional ending), 6, Gk 218, PA 414, Yu 92 agent/actor, Es 116, 132, 137–138, Gk 245, PA 426, Ru 181–182, 186–187, Tu 16–17 agglutinating (see also morphological types), 2–3, 6–7, Es 112–114, 118, 121, 125, 141–145, Tu 15 and inflecting-fusional, Gk 217, 234 agreement (see also errors, agreement), 2, Es 111, 118, Fr 374, 393–394, 401, Ge 279–280, 285–288, 293–294, Gk 249, 253, 258, It 304, 306, 309–310, 312, 316–317, 322, PA 415, Ru 198–199, Sp 11, 341–345, 347, 349–350, 355–366, Tu 15–17, 19–20, 33, 40–41, 43–44, Yu 93–94, 103, 107–108 allative, Es 112–114, 117, 121–122, 130–136, 139, 144, 146, Fi 50–55, 64–80, 83–87 alternation, morphophonological (see also gradation/grade alternation; stem alternation; suppletion), 2, Es 8, 112–113, 115, 140, 142, 145, Fi 50, 57, 61–63, 72, 75, Fr 371, 373–374, Ru 189, Sp 342 ambiguity (see also disambiguation), 3–4, Es 133, Fi 62, Fr 11, 378, 404, Ge 276–279, 283, 294, Gk 221, 227, 243, 246, It 311, 317, 328, Yu 104 analogy, 4, Fi 49, 60, 67, 72–73, 75, 77, 86, Ge 267, 282, 292, Gk 246, 250, It 315–317, 322–323, 326, 335, Ru 180, 183, 188, 191–192, Sp 351 analytic marking (see periphrastic marking) animacy (see also animacy hierarchy), Cr 153, 155, 158, Es 136, Fi 53, Ge 269, 282, Gk 221–223, 237, 239, 241–242, 245–246, It 306, 314, 336, PA 419, Ru 181, 183, 186, 188, 190,

203, Sp 349, 361–362, 366, Tu 40, 43, Yu 7–8, 91–102, 105–109 animacy hierarchy, Yu 7, 91, 95, 97, 103, 105–106, 108 apposition, Gk 240, 258, PA 415 Arabic (see also Palestinian Arabic), 411, 413–416 article (see also determiner; errors, omission; fillers), 3, Fr 371–381, 383–384, 386–390, 402–406, Ge 265, 268, 270–273, 275, 280–281, 284–290, 293–294, Gk 219, 223, 233, 237–246, 248–249, 251–255, It 303–304, 306, 309–310, 313, 319–322, 328, 332, 336, Ru 184, Sp 11, 342, 344, 346–348, 351–353, 359, 361, 363–365 definite, Fr 371, 373–374, 376–379, 384, 386–388, 390, 406, Ge 271–273, 280, 284–290, Gk 219–220, 223, 233, 237–246, 248–249, 251–255, It 319, Sp 11, 344 indefinite, Fr 373–374, 376, 378–379, 384, 387, 406, Ge 271–273, 284–288, Gk 242, It 309, 319 assibilation, Fi 50 bare form (see also root; stem), 4–5, Es 113, 125, 128–129, 132–133, 142–143, 147, Fi 62–63, Fr 376, 378, 401, Gk 248, It 335, Tu 16, 18–19, 22, 27, 30–31, 44, Yu 96 base form, 4–5, Es 115, 121, Fi 70–71, Ge 267, Gk 220, 222, 233–237, It 327–328, Ru 9, 185–186, 188, 190–192, 196, 199, 202–206, Tu 7, 27–28 benefactive (see also recipient), Es 133, 138, Fi 50, 53, 55, 69–70, 74, 78, Gk 238, 242, 246, Ru 183, 187, Tu 16, 19

Subject Index beneficiary (see benefactive) biuniqueness, Fr 404, Ge 268, 270, Gk 246, 253, It 312, 327, Sp 364 Bulgarian, 153 case adverbial, Es 111, 113–114, 127, 129, 133 grammatical, Es 111, 113, 116–117, 122, 126, 132, 134–136, 138, 141, 145, Fi 51–52, 54, 68–69, 71, 79, 86, Gk 241, Ru 180, 184, 187, 195–196, 202 internal, Tu 18–19, 32–33, 42 internal local, Fi 52–53, 55, 74, 79 local (see also locative), Es 111, 117, 120, 126, 129–131, 134–136, 138–139, 141, 144, 146–147, Fi 50, 52–56, 62–65, 67–71, 74–76, 78–80, 83–84, 86–87, Tu 42 marginal, Es 111, 117–118, 121, 129, 132, 134, 138–139, 144–145, Fi 50, 54–55, 65–67, 69, 71, 80, Tu 15, 18, 29, 34–35, 37, 42, 44 relational, Tu 18–19, 30, 32–33, 42 semantic, 2, Es 111, 122, Ru 184, 187, 195, 197 case functions, Cr 154, 175, Es 116–118, 129, Fi 49, 63, Gk 221–223, 237–247, 256, Ru 180, 203, Tu 18–19 case marking (see also inflectional marking; periphrastic marking; suffix/suffixation, case; synthetic marking), Cr 174, Es 140, Fi 7, 63, 72, Ge 265, 270, 272–273, 279, 284, 289, 291, 293–294, Gk 240–241, 244–246, Ru 182–183, 186, 188–190, 196, 199, 205, 210, Tu 15, 17–19, 23, 30–32, 37–40 child-directed speech (see also input), 1, 5, Cr 160, 162, Es 115, 122–123, 125–126, 128, 136, 140, Fi 80, Gk

435

220, 222, 225–227, 231, 233–237, 243, 254, 256, PA 424, Ru 191, 195, 210, Tu 19, 27, 44, Yu 97, 100, 106–107 class shift, Es 140, 147, Ge 268, It 304, 307, 320–322, 334–335 classifiers, numeral, Yu 7, 91–100, 106–109 clitics, Fr 11, 371–372, 391, 393–394, 404–405, It 303, 319, 332, Sp 351–352 cognition, Gk 217, 255, 258, It 303, 305, 334, Ru 181, 184, 198, 205 collective, Cr 157, Fr 405, It 306, PA 413, 415–420, 422, 424, 426–427 comitative (see also instrumental; case, marginal), Es 111–112, 114, 118, 121–122, 126–127, 129, 133–136, 138–139, 144–145, Fi 54, 65–66, Gk 238, Ru 182, Tu 15–16, 18–19, 26, 34–37, 42, 44 communicative relevance, Cr 171, Fi 74, Gk 232, 237, 239–241, 245, 255, Ru 179, Sp 352, 359, 361, 364 companion, Es 118, 137, 139, 146 completive, Yu 102, 104 connectionism (see theoretical models) consonant harmony (see also vowel harmony), Fr 380, Sp 344, Tu 15 constructivist model (see theoretical models) contrast (see also opposition), 1, Cr 156, 163, Es 126, 128, 133, 135, 137, 141–142, Fi 49, 57, 60, 72, 77, 80–83, 85–86, Fr 376, 387, Ge 277–279, 283, Gk 9, 220, 223–224, 226, 228–231, 236–241, 243–245, 247–256, 258, It 10, 306, 311–312, 321, 325, 328, 334–335, Ru 179, 181–182, 192–193, 195, 197–202, 207, Sp 341, 344–346, 349–350, 352–355, 361–362, 364, Tu 16, 21, 32, Yu 94, 103

436

Subject Index

correct (see adequate) Croatian, 2–3, 8, 153–177 cultural environment, Yu 8, 107 dative (see also benefactive), Cr 154–156, 163, 169, 171–172, Fi 53, Ge 270–273, 279, 281–287, 289–291, 293, Ru 182–190, 195, 210, Tu 15–17, 19, 24–26, 29–34, 36–37, 42–44 declensional class/pattern/system (see also inflectional class/pattern), 2, 5, 9, Cr 153, 155–158, 162, Gk 220–221, 223–224, 234, 240, 242, 246, 257–258, It 307, 309, Ru 180–181, 185, 188–190, 194–195, 197, 202–203, 207, 209–210 default (see also rule), 5, Cr 156, 167, Ge 10, 266–269, 282, 291–292, It 328, 330, PA 412–414, Ru 202 definiteness, (see also article; determiner; object), Fi 76, Gk 258, Sp 342, 352, Tu 40, Yu 106 deictic forms, Es 111, 129–130, Fi 64, Yu 92, 96, 107 demonstrative (see also adjective; determiner; pronoun), Es 113, 135–136, 147, Fr 373, 379, Ge 286, Gk 240, 253, It 312, 318–319, Sp 348, 351–352, Tu 19, 26–27, 30, 32, 36, 38, 43, Yu 92–93, 106–107 derivation, 4, Cr 153, Es 115, 128, 141, 148, Fi 54, 59, Ge 268, Gk 237, 253, PA 414, Ru 188, 191, 196, Tu 22, 36 determiner (see also article; demonstrative), 2–3, Fr 11, 371–376, 379, 382, 385, 390, 403–405, Ge 272, 278–279, 284–289, 291, 293–294, Gk 233, 249, 253, It 306, 312, Sp 10, 341–348, 351, 355, 357, 361, 363–364, 366, Yu 93, 102–104, 108 determiner phrase (see also noun phrase), Sp 366

developmental strategies (see language learning strategies) dialects, Cr 8, 153, 158–159, 161, Ge 280, 290, It 308, PA 411, 413–418, Yu 108 diglossia, PA 411, 413 diminutive, Cr 161, Es 128, Fi 84, Ge 275, Gk 222, 237, 246, 250, It 321, 327–329, 332–333, 336–337, Ru 188, 195, 210, Sp 364 direct object (see object, direct) direction, Cr 167, Es 146, Ru 184, 186–187, Tu 16–17 directional (see also case, local; goal–oriented; locative), Es 117, 129–130, Ge 290–291, Ru 184 disambiguation (see also ambiguity), 3, Fi 72, Gk 239 discourse markers, Fr 398, 402 distal (see deictic forms) dual, 2, PA 413, 415, 417–420, 422–424, 426, Sp 365 dual-process/dual-route model (see theoretical models)

elative, Es 112, 114, 117, 121–122, 130, 134–135, 138–139, 146, Fi 52–53, 62–65, 68–70, 78–80, 83–84, 87 emergence, 4, Cr 166–167, 169, 172, Es 8, 123–125, 127, 129, 133, 141, 144–145, 147, Fi 49, 54–57, 64, 67, 69, 86, Fr 11, 373, 378–379, 383, 393–395, Ge 265, 267, 272, 275, 277, 280, 285, 292–293, Gk 9, 217–221, 224, 231, 237–239, 241–245, 247, 252–258, It 10, 303–305, 310–311, 313, 318–319, 332, 334, 336, PA 424–425, Ru 9, 179, 182, 185, 191, 193, 195–197, 199, 201, 206–207, Sp 341–342, 359, 362–364, Tu 15–16, 22, 24, 29, 33, 36, 38–39, 41–42, 44, Yu 93–97

Subject Index order of, Cr 8, 160, 169–170, 172–173, 175, Es 121–122, 124, 130–132, 137–138, 142–147, Fi 55–56, 65, 67, 69, 71, 76, 78, 86, Fr 372, 390, 396, 399–400, 402–404, Ge 270, 277–279, 284, 286, 290, 292, Gk 231, 234, 245, 247, 253, It 321, 328, 335–336, PA 413, Ru 180, 182, 197, 205, Sp 347, 349, 354–355, 363, Tu 19–20, 32, 37, 39–40, 42, Yu 106 ending/inflectional ending (see also suffix/suffixation), 3–4, Cr 155, Fi 80, Fr 371, 373–374, Ge 265, 267, 269–270, 273, 278–279, 281–285, 289, 291–293, Gk 220–221, 223, 227, 229, 231, 234, 238–242, 244, 246–247, 249–252, 256–257, It 307–309, 314–317, 320–322, 326, 328, 330, 336, Ru 9, 179–184, 188–190, 196, 198, 202–204, 207, 210, Tu 29 English, 11, 266, 308, 412–413 entrenchment, Gk 218, 236, 254 ergative, Yu 1, 93, 102, 104 errors (see also adequate/correct forms/use; overgeneralization; self–correction; substitution), Cr 173–174, Es 132, 143, 146, Fr 383, 390, Ge 267, 280–282, 288–294, Gk 249, 258, It 304–305, 315–316, 320–324, 332, 334–335, PA 417, 422–425, Ru 180, 184, 199, 204, Sp 345, 349–350, 357–358, 362, Tu 34–37, 42, Yu 97 agreement, Fr 401, Sp 358–361, 363–364 commission, Fr 387–390, 401, Ge 273, 279–281, 286, 288, 293–294, Tu 18, 21, 35–36 of morpheme order, Tu 34 omission (see also suffix/suffixation), Es 112, Fi 57, 61, Fr 388, 390,

437

394–395, Ge 273, 279–280, 286–288, 293–294, Gk 243, 245, Ru 182–184, Tu 17, 21, 29–31, 33–34, 41, Yu 97–99 essive, Es 112, 114, 117–118, 134, 145, Fi 52–53, 65, 69, 71, 85 Estonian, 2–3, 8, 111–151, 181 experiencer, Ru 186–187, Tu 17, 19 experimental studies, Ge 266–268, 292, Gk 242, PA 12, 412, 416–417, 426 extragrammatical operation, Ge 274, Gk 219, It 10, 311–312, 317–318, Ru 179 feminine, Cr 153, 155–158, 160–161, 163–164, 175, Fr 373, 391, 405, Ge 265, 269, 271–272, 278–279, 281–283, 285–287, 293, Gk 220–223, 228–239, 242–249, 251–255, It 305–310, 312–315, 317, 320–321, 328–329, 336–337, PA 414–415, 417–420, 422, 424–428, Ru 180, 185, 188–190, 194, 202–203, 210, Sp 346, 349–350, 361 fillers (see also protoarticle), Fr 372, 376, 378–379, 387, 390, 402, 404–405, Ge 288, It 318–320, 322, Sp 344–345 Finnish, 2–3, 7–8, 49–89, 146, 181, 183 formulaic forms (see also frozen forms), Fr 378, 398–399, 402, 405, Gk 224, It 311, Ru 183 French, 2–3, 6, 11, 142, 272, 304, 347, 365–366, 371–410 frequency (see also productivity), 12, Cr 161, 166, 172, 175, Es 126, 144, Fi 70, 81–82, Fr 395–396, Gk 217–218, 226, 236–237, 244–245, It 332, PA 12, 412, 427, Sp 352, 355, 362–363, 366, Tu 17, 19, 23, 28, Yu 95, 98–99, 105 input (see also child–directed speech), Cr 161, 167, Es 125–127, 131, 135, 142–144, 146,

438

Subject Index

Fi 54, 65, 69, 73, 86, Fr 11, 396, 400, 402–404, Ge 10, 265, 278, 281, 291–292, Gk 226, 237, 253, It 303, 315, 317, 322, Ru 184, 188, 194, 198, Tu 19, 37–38, Yu 95, 100, 104–105 token (see also PBF), Es 125, 135, Fi 71, Fr 396, Ge 10, 265, 269–270, 275–276, 278, 281, 283, 291, Gk 218, It 312, 317, PA 418, Ru 180 type, 6, Ge 266–270, 275–276, 278, Gk 218, 250, It 312, 317, 328, PA 418, Ru 180, Tu 22 frozen forms (see also formulaic forms) Fi 53–54, 57, 62, 64–66, 70, Fr 374, 379, Gk 224, It 311, Ru 184, Sp 351, 365, Tu 32, 37 functional categories, It 10, 305, 335, Sp 366 functional load (see communicative relevance) fusion/fusional (see also inflecting–fusional), 2–3, Es 112, 118, 142–145, Fr 11, 371–372, 374, 404, Ge 265, 268, Gk 217–218, 234, It 304, 306, Ru 186, 188, Tu 40, 42

gender (see also agreement), 2–3, Cr 8, 153, 160–162, Fr 374, 405, Ge, 268, 270, 273, 280, 282, 285–286, 288, 291, 293–294, Gk, 217, 220–222, 225, It 304, 306–307, 309–310, 312, 314–315, 322, 327–328, 330, 336, PA 11, 416, Ru 186, Sp 11, 343–344, 350, 361–362, 364–366, Tu 15 and inflection, 3, Cr 42, 155, 157, Gk 9, 217, 220–221, 224, 227–237, 240, 245–246, 248, 252–254, 256–258, It 307, 317–318, 320–322, 328–329, PA 414, Ru 185, 188, 210, Sp 349–350

generalization (see also overgeneralization; rule; schema), 4, Es 133, Fr 379, 388, Ge 266–267, 282, Gk 9, 218–219, 256–257, It 328, 336, PA 412, Ru 183, Sp 355, 363, Tu 30, 36 genitive (see also possession), Cr 154–156, 158, 163–164, 166–167, 169, 171, Es 111–116, 118, 121–122, 126–138, 140, 142, 144, 147, Fi 50–51, 54–57, 60–62, 68, 79–86, Ge 270–271, 273, 283–285, 291, 293, Gk 220, 222–223, 229–233, 238–239, 241–246, 249, 252–254, 258, Ru 181–183, 185–190, 195–196, 199, 210, Tu 15–19, 25–26, 30, 33–37, 43–44 German/Austrian German, 2–3, 9–10, 44, 154, 181, 184, 265–302, 304, 347, 361, 365–366, 412–413 goal, Cr 154, 171, Es 117, 139, 146, Ru 184, Tu 16–17, 19, 32 goal-oriented (see also directional), Es 117, 129–130, 138–139, 146, Ge 290 gradation/grade alternation, Es 8, 112–116, 118–119, 125–126, 128, 135, 140, 145, 147–148, Fi 7, 50, 57, 59, 61–63, 72 grammaticization, 1, Es 141, Fi 52, Fr 373, 378–379, 390, 402, 404–405, Gk 245, 253–254, It 306, 335, Tu 38, 40, Yu 92, 108 Greek, 2–3, 5, 9, 42, 142, 181, 183, 185, 191, 217–264, 304

Hebrew, 415–416, 425 homonymy, Cr 155, 165–166, 169, Es 128, 133, 140, 142, 147, Fi 51, 63, 75, 79, Fr 386, 391, Ge 270, 280, 285, 288, Ru 185–186, 189, 203, Tu 16 Hungarian, 74, 146

Subject Index iconicity (see also onomatopoeia), 4, Ge 265, 267–270, 278, 281–282, 291–293, Gk 250, It 306, 312, Ru 185 illative, Es 112–114, 117, 121–122, 124–127, 130, 132–136, 139–140, 146, Fi 50, 52–57, 59–60, 64–65, 68–69, 71, 73–76, 78–83, 86, Ru 183 imitation, Cr 160, Es 119–120, 133, Fr 372, 388–389, 402, Ge 278–279, 294, Gk 247, 252, 258, It 311, Ru 190, 193 indirect object (see object, indirect) individual variation (see variation, individual) Indo-European, 2, 91, 142, 221, 306 inessive, Es 112, 114, 117, 121–122, 124, 127, 129–130, 132–134, 138–139, 146, Fi 52–53, 55, 62, 64–65, 68–70, 78–84, 87 infix, Cr 158 inflecting-fusional (see also root–inflecting), 2–3, Es 112, 142, Fr 371, Ge 268, Gk 217–218, 234, Ru 188 highly inflected/strongly inflecting, 3, Cr 153–154, Gk 217–218 weakly inflecting, 3, Ge 265, It 306 inflection irregular, 11, Es 121, 126, 140, 145, 147, Fr 368, 388, Ge 266, Gk 218, It 304–305, 307, 315, 329, 333, 336, PA 411–413, 417, 425, 427, Ru 202 nominal, 3, Cr 8, Es 145, Fi 86, Gk 9, 217–218, 220–221, 224–225, 227, 230, 234–235, 237, 246, 254, 256–257, It 305–306, 308, 330, Ru 179, 233, Sp 362, Tu 15–16, 19, 24, 39, 42 nominal and verbal, Gk 246, 255–256, Tu 41 pronominal Tu 7, 41

439

regular, 7, 11, Cr 154, Es 113, 121, 128, 135, 140, 145, 147, Fi 49, Ge 266, 282, 292, Gk 218, 234, It 303, 305–306, 321, PA 12, 411–414, 422, 427, Ru 198, Sp 342, 363, Tu 15, 19, 42 root- vs. stem-based, 2, Ru 186 verbal, Cr 153, 175, Es 124, 132, 137, Fi 65, 86, Fr 11, 371–373, 391–404, Ge 266, 274–275, 279–280, 294, Gk 217, 224, 227, 246, 253, 255–256, 258, It 303, 306, 312–313, 319, 322, 327, 336, PA 412, 415, Ru 179, 184–185, 191, 199, Sp 10–11, 341–343, 345, 347, 350–355, 357–361, 363–366, Tu 15–16, 19, 22, 36, 40–41, 43, Yu 93–95, 103–108 inflectional class (see also declensional class; inflectional pattern), Cr 160, Gk 227, It 304, 309, 327–329, 333, 335–336 macroclass, It 307–308, 336, Ru 188 microclass, It 307, 309, 315–318, 320–321, 330, 332, 334–337 inflectional development (see also morphology proper; premorphology; protomorphology; spurt) across-the-board, Gk 9, 218, 231, 237, 256, PA 11, Tu 7 item-based, Cr 8, 175, Es 145, Fi 65, 67, Gk 219 local, Gk 9, 220, 237, 255–256 order, 1, 6, 8, Cr 169–170, 172–173, 175, Es 121, 131, 145–147, Fi 55, 64, Ge 272–273, 284, Ru 182–184, 205, Sp 348, 364, Tu 20 inflectional ending (see ending/inflectional ending) inflectional forms marked, 2, 5, Fi 81, Fr 372–373, 402, Ge 285, 293, Gk 220, 225,

440

Subject Index

229–234, 237, 244, 247, 249, 258, Ru 183, 202, Yu 91 unmarked (see also base form), 5, 21, Es 118, Fi 49–51, 57, 62, 70–71, 81, 86, Fr 376, 378, 394, Ge 270, 273, 285, 293, Gk 221–222, 225, 228, 231, 233–236, 238–244, 246, 253, 255–256, 258, It 309, 330, PA 417, 426, Ru 183, 188, 202, 205, Sp 352, Tu 16, 31, 36, 40, 44, Yu 91, 103 inflectional imperialism, Ru 9, 181 inflectional marking (see also periphrastic marking/techniques; root–inflecting; suffix/suffixation; synthetic marking/techniques) addition of markers, 2, Gk 222, It 308, Sp 341 substitution of markers, 2, Gk 222 inflectional paradigm (see paradigm) inflectional pattern (see also declensional class; inflectional class), 1, 3, 6, Es 128, 140, 148, Fr 375, Gk 9, 217–218, 241, 246–247, 250, 253, 257, It 316–317, Ru 196 inflectional principle, Gk 256 inflectional spurt (see spurt) inflectional type (see morphological types) input (see also child–directed speech; frequency, input), 1, 6, Cr 158–162, 166–168, 171, 175–176, Es 111, 121, 123–129, 131–132, 136–138, 142–148, Fi 49, 54, 59, 65, 70, 73, 86, Fr 11, 373, 378, 380, 384, 391–392, 394, 396–397, 399–400, 402–404, 406, Ge 10, 265, 267, 269–270, 275–278, 280–281, 284–285, 290–294, Gk 9, 220, 224–226, 228–229, 231–232, 234, 236–237, 239, 245, 253–254, 256, It 303, 310, 312, 314–317, 320–323, 328–330, 332, 335–337, PA 412,

427, Ru 179–180, 184, 188–189, 194, 197, 202–203, 205, 207, Sp 342–343, 347, 352, 358–359, 362, Tu 15, 19–20, 28–29, 32–34, 37, 39–40, Yu 7, 95–97, 99–107, 109 instructive, Fi 54, 65–67, 69, 71, 79–80 instrument, Es 118, 127, 137–139, 145–146, Fi 55, 70, Ru 184, 187 instrumental (see also comitative; case, marginal), Cr 154, 156, 158, 163, 169, 172, Fi 50, 53, 78, Gk 238, Ru 182–184, 186–187, 195–196, 204–205, Tu 15–16, 18–19, 34–37, 42, 44 interfix, Cr 156 irregular (see inflection, irregular) irregularization (see also overgeneralization), PA 422, 425, 428 isolating, 2–3, 6, Fr 379, Ge 265, 268 Italian, 2–3, 10, 303–340, 361, 365–366 item-based development (see inflectional development) language learning strategies (see also imitation; rote–learning mechanism; variation, individual), Es 140, Fi 59, 61–62, 72, Fr 372, 389, 405, Gk 230, Ru 190, 199, 205–207, Sp 362 language typology (see also morphological types; syntactic types), 1–3, 219, 341, 427 language variety (see also dialects), It 308, 310, PA 411, 413, Yu 107 lexical spurt (see spurt) liaison, Es 125, 135, Fr 11, 371–375, 379–385, 387, 390–391, 393–395, 404–406 loanwords, Gk 254 location (see also case, local; locative; state), Es 114, 117, 127, 129, 131, 133–134, 136–139, 146, 154, 164, 167, Fi 50, 52, 55, 63, 65, Ru 184, 187, Tu 17–19, 32, 42

Subject Index locative (see also case, local), Cr 154–156, 164, 167, 169, 171, 174, Es 131, 137, 146, Fi 63, Fr 388, Ge 290–291, Gk 238, It 312, Ru 182–186, 195, Tu 15, 17–20, 29–30, 32–33, 36–37, 42–43 marked inflectional forms (see inflectional forms, marked) marker (see also case marking; discourse markers; plural markers; zero marking) inflectional/morphological, Cr 163, 174–175, Ge 268, Ru 181, 205 secondary, Ru 180 superstable, It 327, Ru 188–189 marking (see case marking; inflectional marking; number marking; periphrastic marking; synthetic marking; zero marking) masculine, 3, 5, 9, 11, Cr 153, 155–158, 160–161, 163–164, 174, Fr 373, 391, 405–406, Ge 269–273, 279, 281–282, 284–287, 289–291, Gk 220–223, 228–242, 244–248, 251–256, 258, It 306–310, 312–317, 320, 328–329, 336, PA 414–415, 417–420, 422, 424, 426–427, Ru 180–181, 186, 188–190, 194, 202–203, 210, Sp 342, 344, 348–350, 361 methodology (see also experimental studies; quantitative approaches), 8, It 310–312, PA 417–418 miniparadigm, 4–5, 7–9, Cr 162–164, 175, Es 121, 133–137, 141, 143, 147–148, Fi 49, 69, 80–87, Fr 373, 403, Ge 275, 279, 282, Gk 219, 227–228, 250, Ru 185, 192–197, 202, Tu 22, 24–27, 44 MLU, 9, Cr 159, 163, Es 119–123, 127, 137, 144, Fi 71, Fr 372, Ge 274, 294,

441

It 304, Ru 182, 186, 191–192, 203, 205–206, 208–209, Sp 343, Tu 21–22 models (see theoretical models) modular morphology (see also morphology proper), It 10, 310, 317, 336 morphological richness, 1–3, 8, Cr 153, 160, 166, 175, Fr 371, Ru 199, Sp 341 morphological types (see also agglutinating; inflecting–fusional; isolating), 1, 3 morphology proper (see also modular morphology), 4, Es 148, Fr 372–373, 390, 406, Ge 274–275, 293, Gk 219, Sp 341, 345–361, Tu 22, 24–25, 27–29, 34–36, 38, 41 morphophonology (see also alternation, morphophonological) Es 140, Fi 75

naming, Cr 165, 167, Fi 49, 70, It 323, 327, 334, Ru 190 nativism, Ge 273, 290 Natural Morphology (see theoretical models) neuter, 9, Cr 153, 155–158, 160–161, Ge 269–272, 281, 285–287, 290–291, Gk 220, 222–223, 227–234, 236–238, 246–256, 258, Ru 181, 188–190, 194, 209 nominative, 5, 9, Cr 154–155, 156, 158, 163–167, 169, 171–172, 175, Es 111–116, 118, 121–124, 126–128, 132–138, 140, 142–145, 147, Fi 49–51, 54, 56–58, 60–61, 68, 70–71, 75–87, Ge 270–273, 283, 285–287, 289–291, 293, Gk 220, 222–223, 227–234, 236–247, 249, 251–255, 258, Ru 180–182, 185–192, 195, 197–199, 202–203, 207, 210, Tu 15–16, 19, 30, 36, 44

442

Subject Index

noun phrase (see also determiner phrase), 2, Es 137, Fi 51, Fr 375–376, 379, 384, 402, 404, Ge 280, 294, Gk 220, It 306, 310, 312, 317, 320–321, 328, 332, Ru 199, Sp 355, 357, 359, Tu 19, 32, Yu 91–92, 98–100, 107 number (see dual, plural) number marking 2, Gk 251, It 303–306, 310, 321, 323–335, Tu 39–40 and case marking (see also case marking), Tu 19, 40 obligatory and optional, Yu 100, 105 on noun and verb, 11, Fr 371–372, 391–392, 402, Tu 40 periphrastic vs. synthetic, 11, Fr 378, Gk 258, It 304 transparent Es 141–142 numeral (see also classifiers, numeral; quantifier), 10, Es 111, 118, Fi 49, Fr 376, 390, 402, Ge 285, 291, Gk 252–253, 258, It 305, 309, 326–327, 330, 335, PA 415, 417, Ru 196, 199, Sp 342, 347, 351, 365, 373, Tu 20, 40, Yu 91–92, 106, 108–109 numeral languages, 2, 7–8, It 306 and transnumeral languages 2, It 306, Yu 102, 107 object, Cr 154, 165, 167, 171, 175, Es 127, 132, 142, 146, Fi 50, 51, 55, Gk 221, 237, It 303, 315–316, 323, PA 421, Ru 181, 183, 187, Tu 17, 19, 30, 32–33, 40, Yu 100–101, 103–104, 108 animate and inanimate, Ru 186 definite and indefinite, Tu 16, 32, 44 direct (see also accusative), Cr 154, Fi 51, 61, 86, Ge 272, 290, Gk 222, 238, 240–241, Ru 183, 186, 196, Sp 366, Tu 16, 30, 36, 44 indirect (see also dative), Cr 154, Ge 291, It 319, Tu 17, 31

negated, Ru 186–187 partial/total (see also partitive), Es 116–117, 128, 132, 137–138, 145–146, Ru 187 specific/nonspecific direct, Tu 16, 31, 44 obligatory context, Fr 390, 394, Ge 287, Gk 250–251, Tu 21–22, 32–34, 38, 41 oblique, 9, Cr 164, Es 118, 124, 126, 142, 144, Fi 49–50, 55, 62–63, 71–73, 75–78, 83, Ge 273, Gk 223, 231, 238–244, 246, 256, Ru 182, 185, 188, 196, 198, 203, 205, 207, Tu 16 omission errors (see errors, omission) onomatopoeia (see also iconicity) It 318–319, 322, Ru 198 opposition (see also contrast) case, 9, Cr 163–164, 174, Es 126, 128, 133, 135, 137, 141, Fi 57, 60, 69, 81–86, Ge 283, Gk 229–231, 236–241, 243–245, 252–256, 258, Ru 179–182, 188, 191–197, 200, 202, 206–207, Tu 16, 32 gender, Sp 349–350, 361–362 grammatical/inflectional/ morphological, 1, 4–5, 7, 9, Cr 162–163, 166, 175, Es 133, 137, 141, Fi 49, Gk, 9, 220, 223–224, 226, 228, 236, 254, 256, It 311–312, 317, 328, Ru 179, 181, 194–195, 199–202, 207–208, Sp 364, Tu 21 number, Cr 171–172, Es 142, Fi 77, 82–83, 85, Fr 387, Ge 267, 275, 277–279, 282, Gk 229–231, 237, 247–252, 254, 256, It 10, 306–307, 321, 325, 327–329, 331, 334–335, Ru 179–180, 185, 191, 194, 197–200, 202, 206–207, Sp 341, 344–346, 352–355, 362, 364, Yu 103

Subject Index order (see also word order) morpheme, Tu 16, 34, 39 phoneme, Fi 112 overgeneralization/overextension/ (over)regularization (see also analogy; generalization; irregularization), 6, 10, Cr 173, 175, Es 131, 140–141, 143–144, 147–148, Fr 386–388, 390, Ge 184, 266–267, 272–273, 277–278, 280–283, 285, 289, 291–293, 413, Gk 239, 242, 246, 250–251, It 304–305, 316–318, 321, 323, 335, PA 411, 422, 425, Ru 180–181, 198, 207, Tu 39, Yu 97, 100, 102, 104–105, 107

palatalization, Cr 153, It 307–308, 336–337, Ru 180, 188, 205 Palestinian Arabic (see also Arabic), 2–3, 6, 11, 411–432 paradigm (see also miniparadigm), Cr 155–156, 160, 163–164, 172, 175, Es 113–114, 133, 136, Fi 50–51, 76, Fr 404, Ge 270, 275, 285–287, Gk 9, 246, It 307–308, 311–312, 314, 316–317, 321, 327–337, Ru 186, 188–190, 192, 194–195, 197, 207, Sp 11, 341–342, 346–348, 365, Tu 15–16, 18, 24–28, 40, 42–43 mean size of, Gk 9, 220, 225, 255, It 316, Ru 202, 207, Tu 23, 44 paradigm formation (see also miniparadigm; pattern construction), Fi 72, 80–86, Gk 220, 227–233 participant, Fr 398 neutral (see also predicative), Es 127, 132, 137–138, 146 particles, Es 128–129, 138–139, Fi 64, Gk 249, Tu 43, Yu 104 partitive, Cr 164, Es 112–114, 117–119, 121–122, 124–128, 132–136, 138, 140–145, 147, Fi 49–52, 54–60, 63,

443

68, 72–73, 76–87, Fr 373–374, 376, Gk 222, PA 416, Ru 181, Tu 17 parts of speech, Gk 217–218, It 312, Sp 11, 363 patient, Ru 181, 186, Tu 16, 19 pattern (see inflectional pattern; paradigm; schema) pattern construction (see also rule formation; schema), Gk 218 PBF (percentage of base forms), 5, Es 121–123, Fi 7, 70–71, Gk 9, 220, 225, 233–237, 254, Ru 9, 185–186, 192, 202–206, 208, Tu 7, 22, 27–28, 44 periphrastic marking/techniques (see also number marking), 3, Es 114, 129, 142, 147, Fi 62–63, Fr 11, 371–372, 392, 403, 406, Ge 10, 265, 270, 275, 279, 284–288, 290–291, 293–294, Gk 233, 238–240, 243–245, 248–249, 251–253, 255, PA 417, 427 person (see also agreement), Cr 153, Es 132, 137, Fi 66, Fr 371–372, 391–394, 396, 398–401, 403–404, Ge 279–280, Gk 255, It 309–310, Ru 186, Sp 342, 345, 350–351, 353–355, 361, 363–364, 366, Tu 16–17, 40, Yu 103, 105, 108 personal pronoun (see pronoun, personal) phases, developmental (see also morphology proper; premorphology; protomorphology), Gk 219, Ru 195, 207 plural nominal vs. verbal, Fr 11, 372, 402, Sp 363–364 verbal, (see also plural forms, rhyming), Fr 400, 402–403 plural forms broken, PA 414, 416–420, 422–427 rhyming, Fr 392, 394–397, 400, 402–404, 406, Ge 266, 292 sound, PA 414, 417, 426–427

444

Subject Index

stem, Es 118, 143 stem-amplified, Fr 400, 402, 406 plural markers (see also number marking; suffix/suffixation, plural), Fr 376, 379, 393, Ge 265–267, 269, 272, 275, 277–279, 282, 291, 293, It 304, 316, 327–328, 330, Sp 341–342, 344–347, 359, 363–365 lexical (see also numeral; quantifier), Fr 365, 376 nominal, Yu 106, 108 single and multiple, Sp 347, 355–356, 358, 366 verbal, Yu 108 plural-dominant nouns, Fr 384, 406, Ge 278, 291, 294, It 320, 325, 330, PA 424 Polish, 189 polyfunctional device, Es 142 portmanteau, 11, Fr 372, 379, 386–390, 404 possessed, Cr 166, Es 127, It 309, Tu 17, 33, 44, Yu 91, 104–107 possession/possessive (see also genitive), Cr 154, 166–167, Es 116, 127, 131–132, 137–139, 146–147, Fi 51, 53, 55, 61, 66, 70, 78, 86, Ge 270, 273, 283, 291, Gk 222, 228, 238, 241–246, Ru 182, 185–187, Tu 7, 15–17, 19–20, 24, 28, 33–34, 36–39, 41–42, 44, Yu 7–8, 91, 93, 100–109 possessive adjective/pronoun, Fi 78, Fr 373, 375, 379, Ge 286, It 306, 309–310, 315, Sp 348, 351–352 possessor, Cr 154, 166, Es 116, 127, 131–132, 137–139, 146–147, Fi 55, 61, 70, Gk 245, Ru 182, 185–187, Tu 17, 33, 35, 44, Yu 93, 104 postposition, Es 114, 129–131, 137, 147, Fi 51, 53–55, 61, 69, Tu 18, 43 predicative (see also participant, neutral), Es 116, 138

prefix/prefixation, Fr 375, 385, Ge 275, Yu 92 premorphology (see also transitional period), 4, Cr 162, 166, Es 121, 123–133, 137–138, 142, 144–147, Fi 49, 86, Fr 372–373, Ge 268, 274, Gk 219, 224–226, 254–255, 257, It 10, 305, 310–320, 322–327, 329–333, 335, Ru 179–180, 191, 193, 195, 202, Sp 341, 344–345, 362, 364–365, Tu 22–25, 27–29, 32, 34–35, 37–38, 41, 44 preposition, Cr 154, 164, 167, 171, 174, Fi 50, 54, Fr 371, 374, 386, 388–389, 405, Ge 283, 290–291, Gk 222, It 304, 313, 319, 332, Ru 182, 184, 187, 204 prepositional case, Ru 186–187, 189–190, 210 prepositional phrase, Cr 154, Ge 270, 272–273, 283, Gk 222, 238, It 313, Ru 182 productivity, Fr 394, Tu 27 and frequency, 6, 12, Ge 268–269, Gk 250, Tu 22 inflectional, 4, 6, Cr 8, 163, 172, 175, 179, Es 8, 114–116, 124, 126, 130, 133, 135, 140–148, Fi 54, 62–65, 77–78, Fr 371, 373, 375, Ge 265–266, 268–270, 273, 276–279, 281–282, 291–293, Gk 218, 220, 239–240, 250–252, 255, It 10, 305, 307–309, 311, 317–318, 320–322, 328–330, 332–337, PA 412–414, 424–426, Ru 180–182, 185–186, 188, 190–192, 194–198, 202, 205, 207–208, Sp 346, 350, 352, 362, 364, Tu 7, 15–16, 21–25, 29–31, 37, 41–42, Yu 93 pronominal inflection (see inflection, pronominal)

Subject Index pronoun (see also clitics; demonstrative; possessive adjective/pronoun), Cr 153, 176, Es 111, 113, 118–120, 141, 148, Fi 49, 51, 67, 69, Ge 273, 275, 280, 288, Gk 240, 245, 253, It 306, 336, PA 415, Sp 10, 341–342, 348, 355, 361, Tu 6, 15, 17–20, 23–24, 26–27, 30, 32, 36–37, 42, 44, Yu 108 object, It 313, 319, Yu 103 personal, Es 125, 135–136, 147, Fi 51, Fr 391, 394, Gk 228, 245, It 319, Sp 351–352, Tu 26, 33, 36, 42–43 relative, It 319 subject, Fr 11, 372, 375, 393–395, 404 protoarticle (see also fillers), Sp 365 protomorphology (see also transitional period), 4, 6, Es 130, 133, 136–141, 144, 146–147, Fi 49, 86, Fr 372–373, 402, Ge 268, 274–275, 277, Gk 219, 255, It 10, 305, 310–318, 320–324, 326–336, Ru 179, 182, 191, 194–195, 197, 199, 202, 207, Sp 11, 341, 345–351, 362–364, Tu 22, 24–25, 27–39, 42, 44 proximal (see deictic forms) quantifier, Fr 379, Ge 10, 291, 294, Gk 253, 258, It 305–306, 327, 330, 335, PA 415, Sp 342, 351, 365, Tu 20, Yu 93–94, 104 quantitative approaches (see also miniparadigm; MLU; PBF), Fi 67, 71, 78, Ge 268, 275, Gk 9, 220, 225–237, PA 421, Tu 20 quantity (see also gradation), It 314, 326–327, 335, PA 415, Ru 199 partial (see also partitive), Fi 51, 55, Tu 32

445

recipient (see also benefactive), Cr 154, Es 131, 133, 137–139, 146–147, Tu 19 reduction, Fi 59, Fr 384, It 330, Ru 189, 203 redundant marking (see also agreement), Es 126, Fi 63, Fr 379, 393, It 321, Ru 183, 199, Sp 11, 342, 345 referential child, Gk 224, Ru 190 referential function, 1, Tu 16–17, 30–31, 43, Yu 91 referential style, Cr 167, Ru 205 regular (see inflection, regular) rhyming plural verbs (see plural forms, rhyming) root, 4, Cr 153, Es 112, 124, Gk 250, It 335–336, Yu 93 root infinitives, Sp 366 root-inflecting, PA 3, 414, 427 rote-learned forms, 4, 6, Es 8, 121, 123–127, 133, 137, 141–144, 146–147, Fi 7, 49, 56–58, 77, 86, Fr 377, 379, 398–399, Ge 274, 278, 285, 291, 294, Gk 218, 243, 246, 250, 258, It 304, 320, 322–323, 330, 335, PA 412, Ru 179, 192, 195, 205, Sp 341, Tu 21, 34 rote-learning mechanism, PA 411 rule (see also extragrammatical operation; schema), 10, Ge 274, Gk 218–219 concatenation, Fr 411 default, PA 412 inflectional, 4, Cr 155–156, 158, Es 115, 125, 133, Fi 78, Ge 266, 268–269, 274, 292, It 307, 334, Ru 181, 191, Tu 20 (morpho)phonological, Fr 373, 393, It 307, Ru 189 symbolic, 5, Ge 266, Gk 9, 218–219, 257, PA 412

446

Subject Index

rule formation (see also pattern construction), Ge 292, Gk 218, Sp 364 Russian, 2–3, 5–6, 8–9, 27, 121, 142, 179–215, 221, 233–234, 304 salience, 4, Es 127, 146, Fi 50, 58–59, 72, 74, Ge 267, 269, 285, Gk 237, It 303, Ru 9, 180–181, 185, 202, Sp 346, 365, Tu 15–16, 20, 32, 42 schema (see also rule), 5, Cr 8, 164, 172, 175, Fi 49, 65, 67, 80, 84, Fr 373, Ge 267, Gk 9, 218–219, 250, PA 413, Tu 22 segmental child, Fr 372 segmentation, 3, Fi 57, 80, Fr 372, 380–382, 384, It 328 self-correction, Fr 394, Ge 290, It 305, 334–335, Ru 184, Sp 349, 362, 386, Tu 35 semantic roles (see also case functions), Es 116–118, 127, 131–133, 137–139, 145–148, Ru 184, 186–187, Tu 16, 19 single-route model (see theoretical models) sound plurals, PA 414, 416–420, 422, 424–428 source, Es 117, 129–130, 139, 146, Ru 187, Tu 17, 19, 32, 35 Spanish, 2–3, 8, 10–11, 93, 97, 102–103, 108, 341–369 speed of development, 1, Fr 402, Ge 10, 293 spurt (see also inflectional development) lexical vs. morphological, Ru 186, 207 morphological/inflectional, Fi 65, 83, Ru 186, 193–194, 197–198, 207 syntactic, Ge 274 stages of development (see morphology proper; phases, developmental; premorphology; protomorphology)

state (see also location; locative; directional), Es 112, 116, 130, 137–139 stem, Cr 158, Es 112–116, 118–119, 125, 128–130, 132–133, 140, 142–146, Fi 7, 50–51, 53, 57–61, 63, 65, 67, 69, 72–73, 75–78, 80, Fr 11, 372, 384, Ge 279, Gk 218, 221, 253, PA 411, 413–414, 416, 426–427, Ru 185, 188, 198, Tu 16, 19–20, 22, 27–28, 34, 43–44, Yu 93 bare (see bare form) verb, Fr 372, 392, 399–404, 406, Tu 43 stem alternation/change (see also alternation, morphophonological; gradation), 2, Es 8, 113–115, 118–119, 121, 125–126, 128, 132, 140, 143, 145–147, Fi 50, 55, 57, 59–63, 72, 74–78, Fr 371–373, 392, 399–404, 406, Ge 265, 269, 294, PA 416, 426, Ru 188–189, 198, 205, Sp 342 stem-inflecting (see also root–inflecting), 2–3, Ru 186, Tu 15 strategies (see language learning strategies) strongly inflecting (see inflecting–fusional) subject, grammatical (see also clitics; pronoun), Cr 154, 167, Es 111, 116, 132, 137, 145, Fi 49–51, Fr 393, 401, Ge 279–280, Gk 221–222, 238, 240–241, 245, It 306, Ru 183, Sp 343, 345, 357–359, 366, Tu 15–17, 19, 33, 40, 43–44 substitution (see also errors, commission) of inflectional markers, 2, Fi 66, Fr 388, Ge 281, 290, Gk 246, Tu 35 phonological, Fr 372, 380–382, 384, 386 suffix/suffixation (see also bare form; ending/inflectional ending), 3–4, Cr

Subject Index 155, 157, Es 113, 123–126, 128, 132, Fi 7, 50, 66, 72, 80, 86, It 306–307, 310, 335, Tu 15–17, 33–34 case, Cr 155–156, 158, 188, Es 8, 111–114, 118, 124–126, 128–131, 133, 140, 144–147, Fi 7, 49, 51–65, 67, 72–77, 86, Ge 273, Tu 43 derivational (see also diminutive), Cr 188, Es 128, Fi 54, 59, Ge 281 inflectional, Cr 188–189, Es 132, Ru 9, Tu 39, 44 plural (see also dual), Cr 156, Es 118, 142–144, 147, Fi 49–50, 54, 75, 77–78, Fr 11, Ge 265, 281, 292, It 305, 309, 331, 335–336, PA 411, 413–414, 417, 426–427, Ru 197–198, Sp 11, 342, 345, 348, 363–364, Tu 19, 40, Yu 7, 91, 93, 95, 100, 103–108 singular, PA 416 verbal, Fr 371, 391, PA 411, Sp 342, 363–364 superstable marker, Fi 86, Ge 291, 293, It 327, Ru 188 suppletion, Es 112, 135, 147, Fr 11, 371–373, 392, 412, Ru 180, 188, 197 syncretism, 2–3, Cr 153, 163, 172, Es 113, 127, 140, 147, Fr 401, 405, Ge 270, 280, 291, 293, Gk 223, 233, 238, 245, 258, Ru 186, 189–190 syntactic case (see case, grammatical) syntactic types ergative (see also Yucatec Maya), 1 nominative-accusative, 1–2, Fi 50, Ru 180, 186 synthetic marking/techniques (see also number marking), 2–3, Es 114, 129–130, 147, Fi 54, 63, 65, Fr 11, 371–373, 375, 377–378, 391, 402–404, Ge 10, 268, 293, Gk 233, 238–240, 243–245, 248–249,

447

251–253, It 304, 323, Ru 186, 188, Tu 15, 39–40 terminative, Es 111–112, 114, 117–118, 122, 134 thematic vowel, Gk 222, 234, 238, 240–242, 244–245, 250, 256, It 336, Sp 363 theme, Ru 187, Tu 19 theoretical approaches (see theoretical models) theoretical models cognitive-functional models, 5, Gk 217–219, Ru 179 connectionist model, Ge 267, PA 412, 416 constructivist model, 4–5, Ge 273, It 304, Ru 179 dual-route model, Ge 10, 265–268, 292, Gk 218, PA 11, 411–412, 427 generative/modular/nativist models, Ge 272–273, 290, Gk 218, Ru 182 lexically based models, Ru 182–183, 185 Natural Morphology, 7, 10, 12, Ge 267–268, 290, 292, 294, Gk 219, It 304–305, Sp 341, 364, Tu 22 phase model, 12, Gk 255–256 pragmatic model, Ru 183 Pre- and Protomorphology, 4–5, 7, 12, Es 111, 141, It 10, 304–305, Ge 10, Gk 217–219, Ru 179, 185, 197, Sp 10, 341 race model, Ge 269 schema model, Ge 267, PA 413 single-route model, 11, Ge 265–267, 292, PA 412–413 stage vs. phase models, 12, Gk 255–256 Usage-Based Theory, 5, 8, 10, Cr 168, Es 145, Fi 49, 84, 86,

448

Subject Index

Ge 292, Gk 217–219, 250, 257, It 303, Ru 181, 183 transitional period (see also phases, developmental), Ru 179, 206, Tu 22, 43 preinflectional to inflectional, Tu 21 pre- to protomorphology, 4, Es 141, Ge 267–268, It 310–311, 318, 322, Ru 191, 194–195, 197, 207 protomorphology to morphology proper, It 317, Ru 182 translative, Es 111–112, 114, 121–122, 129, 134, 138–139, Fi 52–53, 69, 71, 80, 83, 85 transnumeral languages (see also numeral languages), 7, Yu 91 transparency, 4, Cr 153, 160, 163, 175, Es 140–142, 147, Fi 69, 73, Fr 372, 388, 390, Ge 265–267, 268–270, 278, 280–281, 285, 291–293, Gk 246, It 10, 306–307, 312, 322–323, 330, 332, 335–336, PA 412, 414, Ru 180, 184–185, 198, Tu 17, 19, 42–43 trochaic stage, Es 8, 125, 128, 135, Fi 7, 57–60, 62, 74, 77, It 10 Turkish, 2–3, 5–7, 15–48, 141, 154, 217, 234

variation/alternation (see also alternation, morphophonological) grammatical/morphological, Fi 54, Fr 373, 398, Gk 243–244, 247–248, It 331, Ru 199, Sp 346, 350, 361–362, Tu 22–24, 28, 36–37 individual/intersubjective, Es 8, 123, 147, Fi 7–8, 57, 61, Gk 226, 234, 257, It 10, Ru 9, 182, 190, 198–199, 205–208, Sp 352 lexical, Fi 64, 69 phonetic, Fr 381, Sp 344 verbal inflection (see inflection, verbal) verb (see also inflection, verbal), 12, Cr 154, 166–167, Es 111, 114, 132, Fi 50, 65, Fr 375, 394–398, 400, 403–404, 406, Ge 273, 288, 290, Gk 222, 256, It 313, 318–319, PA 415, Ru 196, Tu 15, 17–19, 27, 30–33, 36–37, 42–43, Yu 103, 108 vocative, Cr 154–156, 158, 165–167, 169, 172, 175, Gk 220, 223, 231, 239, 241, It 312–313, Ru 192 vowel harmony (see also consonant harmony), It 315, 320, 322, 326, 335, Tu 15

underdifferentiation, 5, Fi 72, Gk 252, 256 undergoer (see patient) uniformity, 4, Ge 267, 280, Gk 237 unifunctionality, Ge 270 universal preferences, Ge 268, 270, It 312 unmarked inflectional forms (see inflectional forms, unmarked) Usage-Based Theory (see theoretical models) U-shaped development, Fr 390, Gk 246, PA 411

word order, 2, Cr 153, Fi 50, Fr 394, Gk 240, Ru 186, Tu 17, 19–20, 27, 30, 42

Yucatec Maya, 1–3, 7–8, 91–110

zero marking (see also errors, omission), 2, Es 114, 125, 140, Ge 265, 269–270, 272, 276, 278, 280–282, 291–292, Gk 244, Ru 189, 202, Sp 342, 365, Tu 32, 43, Yu 105