Paradigmatic Relations in Word Formation 9004433406, 9789004433403

Citation preview

Paradigmatic Relations in Word Formation

Empirical Approaches to Linguistic Theory Series Editor Brian D. Joseph, The Ohio State University Editorial Board Artemis Alexiadou, University of Stuttgart Harald Baayen, University of Alberta Pier Marco Bertinetto, Scuola Normale Superiore Kirk Hazen, West Virginia University Maria Polinsky, Harvard University

volume 16

The titles published in this series are listed at brill.com/ealt

Paradigmatic Relations in Word Formation Edited by

Jesús Fernández-Domínguez Alexandra Bagasheva Cristina Lara-Clares

LEIDEN | BOSTON

This publication has been supported by the Spanish State Research Agency (SRA, Ministry of Economy and Enterprise) and European Regional Development Fund (ERDF) (Ref. FFI2017-89665-P). Library of Congress Cataloging-in-Publication Data Names: Fernández-Domínguez, Jesús, editor. | Bagasheva, Alexandra,  editor. | Lara-Clares, Cristina, editor. Title: Paradigmatic relations in word formation / edited by Jesús  Fernández-Domínguez, Alexandra Bagasheva, Cristina Lara-Clares. Description: Leiden ; Boston : Brill, 2020. | Series: Empirical approaches  to linguistic theory, 22106243 ; volume 16 | Includes bibliographical  references and index. Identifiers: LCCN 2020024561 (print) | LCCN 2020024562 (ebook) | ISBN  9789004433403 (hardback) | ISBN 9789004433410 (ebook) Subjects: LCSH: Grammar, Comparative and general—Word formation. |  Paradigm (Linguistics) Classification: LCC P245 .P37 2020 (print) | LCC P245 (ebook) | DDC  415/.92—dc23 LC record available at https://lccn.loc.gov/2020024561 LC ebook record available at https://lccn.loc.gov/2020024562

Typeface for the Latin, Greek, and Cyrillic scripts: “Brill”. See and download: brill.com/brill-typeface. issn 2210-6243 isbn 978-90-04-43340-3 (hardback) isbn 978-90-04-43341-0 (e-book) Copyright 2020 by Koninklijke Brill NV, The Netherlands. Koninklijke Brill NV incorporates the imprints Brill, Brill Hes & De Graaf, Brill Nijhoff, Brill Rodopi, Brill Sense, Hotei Publishing, mentis Verlag, Verlag Ferdinand Schöningh and Wilhelm Fink Verlag. All rights reserved. No part of this publication may be reproduced, translated, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission from the publisher. Authorization to photocopy items for internal or personal use is granted by Koninklijke Brill NV provided that the appropriate fees are paid directly to The Copyright Clearance Center, 222 Rosewood Drive, Suite 910, Danvers, MA 01923, USA. Fees are subject to change. This book is printed on acid-free paper and produced in a sustainable manner.

Contents Editorial Foreword vii Brian Joseph List of Tables and Figures IX Notes on Contributors XI 1 What Paradigms and What For? 1 Jesús Fernández-Domínguez, Alexandra Bagasheva, and Cristina Lara-Clares 2 Paradigmaticity in Compounding 21 Alexandra Bagasheva 3 Characterizing Derivational Paradigms 49 Bernard Fradin 4 The Level of Paradigmaticity within Derivational Networks 85 Petr Kos 5 Doublet Formation in Palestinian Arabic—Where Do Paradigms Interfere? 100 Lior Laks and Faten Yousef 6 What We Talk about When We Talk about Paradigms: Representing Latin Word Formation 128 Eleonora Litta and Marco Budassi 7 A Paradigmatic Approach to Compounding 164 Jan Radimský 8 Of Brownie Girls and Aussie Families: A New Look at Morphosemantic Paradigmaticity in Adj+ie/y Nominalisations 186 José A. Sánchez Fajardo and Elizaveta Tarasova 9 Neoclassical Word Formation in English: A Paradigm-Based Account of -scope Formations 213 Ana Díaz-Negrillo Index 263

Editorial Foreword The Ancient Greek word παράδειγμα, the source of the term that furnishes the first and key word in the title of this volume and provides the notion that unifies its nine substantial chapters, originally meant an architect’s or sculptor’s ‘model’. More germane to the volume at hand, it also meant ‘pattern’, and that is the sense that most directly informs the linguistic concepts of paradigm and concomitantly, paradigmatic relation, that are the focus of this latest contribution to the Empirical Approaches to Linguistic Theory (EALT) series, Paradigmatic Relations in Word Formation. This word παράδειγμα had other meanings in Ancient Greek as well, and interestingly, many of its extended meanings are relevant to the present volume: – For instance, it meant (and still does mean in Modern Greek) ‘example’ and to be sure, there are a good many examples in this book, drawn from a number of different languages, those well-represented in the literature (Arabic, Dutch, English, French, Italian, Latin, Russian), those less well represented (Dogon, Rikbaktsa, Vietnamese), and those in-between (Czech, Slovak). – It could also mean ‘lesson’, and there are certainly lessons to be learned from the studies in this volume about derivation, about inflection, about networks of words, about different kinds of connections among words— semantic, formal, and other—and so on. – Another meaning was ‘argument’, in the sense of an element of a proof in philosophical debate, a meaning which clearly resonates with the nature of argumentation in linguistics and thus in evidence in the chapters here. – Finally, it could also mean ‘precedent’, and while this volume is not totally unprecedented, in that there are many works on morphology and on derivation today as morphological investigation is becoming, or has become, a distinct subfield within linguistics in the past few decades, it is setting a precedent by addressing various open issues—what the editors call “unsettled matters” (pp. 12–13) in morphological analysis and theory. Each new volume of EALT offers something important to the scholarly world and with this work, editors Jesús Fernández-Domínguez, Alexandra Bagasheva, and Cristina Lara-Clares, along with the ten other authors whose work is showcased here, are offering an interesting and insightful group of studies that elucidate the role of the notion of “paradigm” in word-formation processes. Indeed, through these studies, it becomes evident how patterns—παραδεἰγματα—are

viii

Editorial Foreword

as essential for the understanding of word formation and derivational morphology as they are for the understanding of inflectional morphology. Brian D. Joseph

EALT Series Managing Editor Columbus, Ohio USA

25 July 2020

Tables and Figures 1.1 1.2 1.3 3.1 3.2 6.1 6.2 6.3 6.4 6.5 6.6

Tables

An inflectional paradigm of Latin 3 A gap in a derivational paradigm 5 Three configurations post-Full Reanalysis Hypothesis 5 Distinction between agglutinative vs. fusional languages 55 Partial classification of pets and farm animals in French 60 Derivational paradigm for base DIC-: verbs 134 Derivational paradigm for base DIC-: nouns 134 Derivational paradigm for base DIC-: adjectives 135 Sub-paradigm of CAL- 141 List of sub-paradigms with the same derivational behaviour of cal- 141 List of sub-paradigms with no -sc-form corresponding to a -facio/-fio formation 142 6.7 Sub-paradigm of terr- 143 6.8 Sub-paradigms based on base sue- 143 6.9 Sub-paradigm of ir- 144 6.10 Stem allomorphy in ago and duco 145 6.11 Most frequent WFRs in Classical Latin 146 6.12 Sketch of point iv. Towards the construction of a core derivational paradigm for Latin 149 6.13 Core derivational paradigm for Latin filled with lemmas with bases fac-, dic-, ag-, mitt- 150 7.1 Derivational paradigm 167 7.2 Examples of structures that may express an implicit “made of” relationship 176 8.1 Colour/origin-based Adj+ie nominalisations used in the study 203 8.2 Colour-based Adj+ie nominalisations, and their corresponding senses, extracted from normative and descriptive dictionaries 205 9.1 Distribution of formation types in the study 221 9.2 Distribution of [X -scope] instrument N formations 222 9.3 Distribution of [entity -scope] instrument N formations 224 9.4 Distribution of [mode of action -scope] instrument N formations 225 9.5 Distribution of [agent -scope] instrument N formations 226 9.6 Distribution of [-scope X] formations 227 9.7 [-scope X]event N: realizations 228

x

Tables and Figures

9.8 [-scope X]relation Adj: realizations 230 9.9 [-scope X]quality Adj: realizations 231

3.1 3.2 3.3 4.1 6.1 6.2 6.3 6.4 7.1 7.2 7.3

Figures Paradigm’s types 64 Inflectional paradigm 84 Derivational paradigm 84 Word-nest with kniha ‘book’ as the motivating word 87 Partial derivational paradigm of HORR- 133 Hierarchical schematic construction for suffixed verbs in -sc- 139 Partial hierarchical schema for V + V = V compounds 139 Partial hierarchical schema for [co(n)[x](i)t]V formations 145 A French paradigmatic system 168 Bidimensional paradigmatic system in compounding 171 Hierarchical organization of compounding paradigms 173

Notes on Contributors Alexandra Bagasheva Sofia University “St. Kliment Ohridski” Marco Budassi Università degli studi di Pavia Ana Díaz-Negrillo University of Granada Jesús Fernández-Domínguez University of Granada Bernard Fradin Université de Paris, LLF & CNRS Petr Kos University of South Bohemia, České Budějovice Lior Laks Bar-Ilan University Cristina Lara-Clares University of Granada Eleonora Litta Università Cattolica del Sacro Cuore, Milano Jan Radimský University of South Bohemia, České Budějovice José A. Sánchez Fajardo University of Alicante Elizaveta Tarasova IPU New Zealand Tertiary Institute Faten Yousef Bar-Ilan University

Chapter 1

What Paradigms and What For? Jesús Fernández-Domínguez, Alexandra Bagasheva, and Cristina Lara-Clares 1

Introduction

The paradigmatic character of the interword relationships has been present from the very beginnings of morphology as a branch of linguistics. The models of ancient Greek and Latin, to take a classic example, “[…] project morphological analysis primarily upwards from the word, and treat the association of words with paradigms or other sets of forms as the most fundamental morphological task” (Blevins, 2013, p. 375). As was the received wisdom in these ancient grammatical traditions, paradigmatic relations were mostly identified among word forms or what we would call today morphosyntactic words. The paradigm case of a paradigm was the set of forms in which a word could appear. It goes without saying that morphology was understood as the study of syntactically conditioned and appropriate forms. 2

How Paradigmatic Is Derivational Morphology?

More than one century after the appearance of de Saussure (1916), it may seem increasingly difficult to detect areas of modern linguistics that have not come under the influence of its postulates somehow or other. Even so, the crucial effect of this work remains outside the range of issues which periodically reemerge in the 20th and 21st centuries. The debate surrounding word-formation paradigms is no exception, as it inexorably takes us back to de Saussure’s (1916, p. 171) rapports associatifs:1

1  [Outside discourse, the words that have something in common are associated in the memory, and so groups are formed within which very diverse relations rule. Thus, the word enseignement will unconsciously arise in front of the mind a host of other words (enseigner, renseigner, etc., or armement, changement, etc., or éducation, apprentissage); for one reason or another, they all have something in common. Our translation]

© Koninklijke Brill NV, Leiden, 2020 | doi:10.1163/9789004433410_002

2

Fernández-Domínguez, Bagasheva, and Lara-Clares

[E]n dehors du discours, les mots offrant quelque chose de commun s’associent dans la mémoire, et il se forme ainsi des groupes au sein desquels règnent des rapports très divers. Ainsi le mot enseignement fera surgir inconsciemment devant l’esprit une foule d’autres mots (enseigner, renseigner, etc., ou bien armement, changement, etc., ou bien éducation, apprentissage); par un côté ou un autre, tous ont quelque chose de commun entre eux. Several of these features have turned out to be essential for the concept of paradigm, for example the fact that it involves relations outside discourse (i.e., in absentia), or that speakers tend to memorise and homogenise paradigm slots. Despite the vagueness in the understanding of paradigmatic structures and the sporadic reticence towards their application beyond inflectional morphology (Carstairs-McCarthy, 1987), the above definition is proof that de Saussure does not automatically exclude word formation from the span of paradigmaticity. He, as a token, draws on a series of derivatives for exemplification: enseignement-enseigner-renseigner. De Saussure’s description is in fact somehow loose and limits itself to stating that members of a paradigm will have something in common, although what this is exactly is left unsaid. Several properties have been discussed in the literature as central to the notion of paradigm and, consequently, to its potential suitability for the description of word formation. The following subsections consider a number of prominent attributes of paradigmaticity with the aim of offering a panorama of partially settled as well as live issues on derivational morphology. This catalogue does not intend to be exhaustive, but hopes to address a number of pertinent questions, presented in order of ascending complexity, for the description of word-formation paradigms. 2.1 Understanding Paradigms and Neighbouring Concepts Before considering their possible relevance as an object of study for inflection and word formation, it seems necessary to set the limits of the entities known as paradigms. Perhaps because of their wide use in linguistic theory, or because of language-specific limitations, decades of attempts have not sufficed for a unanimous definition. One basic feature of the paradigmatic axis of language is that, in contrast to the syntagmatic one, it is largely based on relatedness, i.e., the similarity or disparity of linguistic forms or meanings. In turn, it is essential to note that “[…] relatedness is a matter of degree” (van Marle, 1994, p. 2927), The use of the term paradigm has been at least three-fold. While an original understanding of it is inevitably linked to de Saussure’s (1916) associative

3

What Paradigms and What For ?

relations (see §‎2 below), paradigmaticity may also point to the nature of the organisation of lexemes in the lexicon: synonymy, homonymy, polysemy, hyponymy, partonymy, etc. A third sense refers to the lexical units with the potential to occupy the same lot in language structure which, as is well-known, typically happens in inflection. The following features have been claimed to characterise a paradigm (Bybee, 1985; Carstairs-McCarthy, 1987; Bauer, 1997; Antoniová & Štekauer, 2015): i) It is made up of morphologically related forms, ii) it is made up of forms that, though semantically derived from the same base, display some meaning divergence, iii) it presupposes a typical set of realisations that can be regularly realised from the base form, iv) it stems from a base word, from which a variable number of other words are systematically predictable, and v) it is restricted to affixation. This catalogue is purposefully broad with the goal of facilitating a joint treatment of inflection and derivation, even if these features materialise differently for each branch of morphology; let us not forget that inflectional paradigms and derivational paradigms are types of paradigms. Thus, if applied to inflection, criterion i) is taken to mean that the members of a paradigm should be formally related, as in the present indicative of Latin, where the roots (laud-, hab-, leg-) are common to all slots: Table 1.1

1st p. 2nd p. 3rd p. 1st p. 2nd p. 3rd p.

An inflectional paradigm of Latin

singular

plural

laudo laudas laudat laudamus laudatis laudant

habeo habes habet habemus habetis habent

lego legis legit legimus legitis legunt

This notwithstanding, formal relatedness is far from a hard-and-fast rule even in inflection, as evidenced by the numerous instances of suppletion found across languages, e.g., go-went in opposition to call-called, follow-followed, workworked, etc. in the expression of present-past in English (Veselinova, 2006). Series like this show how, in spite of formal disparity, suppletive units effectively fill in paradigm slots, which substantiates the idea that meaning, rather than form, is the decisive factor for the structuring of paradigms (Bauer, 2017). The

4

Fernández-Domínguez, Bagasheva, and Lara-Clares

application of criterion i) to word formation has far-reaching consequences which are discussed separately in §‎2.3. Criterion ii) has to do with the semantic connection between the lexical base and all other forms in the paradigm, which must be close enough so as to set up a network. Observance of this condition in inflection can be confirmed by contrasting forms like laudo, laudas, laudat, laudamus, laudatis, and laudant, which share the core of their meaning (here, present indicative of laudāre ‘to praise’) and are thus grammatically connected to each other. In the case of derivation, the meaning in common across paradigm members is that of the root, but the added meaning in each particular form will depend on the morphological process involved. For instance, perform is the starting point of a derivational paradigm where a range of semantic roles are possible: performingN (action), performance (process), underperform (diminution), performer (agent), performable (quality), etc. While the meaning expressed in the present of indicative is clearly inflectional and that in agent is clearly derivational, semantic distance is a thorny issue in the discussion of the inflectionderivation continuum, with cases like the inflectional vs. derivational status of -ly (see Giegerich, 2012). To what extent such semantic divergence concerns more inflection or derivation is once more a consequence of the debate on split morphology, beyond the scope of this article (see §‎3). 2.2 Gaps and Predictability—Actual vs. Potential Words Potential words have been of relevance to morphologists ever since Aronoff’s (1976) precept that “[…] the simplest task of a morphology […] is the enumeration of the class of possible words of a language” (pp. 17–18). Aronoff’s challenging statement led to the demarcation between actual, possible and potential words, the effects of which have been subsequently felt variously in word-formation research (Aronoff, 1983; Bybee, 1985, p. 85; Bauer, 2001, pp. 34– 43; Rainer, 2012). In the field of paradigmaticity, Horecký, Buzássyová, and Bosák maintained already in 1989 that “[…] new complex words do not come into existence as isolated units but rather as complete paradigms” (p. 42; see Štekauer, 2014, p. 360). The question therefore is how to cope with the empty spaces in those paradigms. Potential words have been pointed out as a fundamental difference between inflection and derivation: paradigmatic gaps or defective paradigms are relatively anomalous in inflection (note the expressive terminology) given the obligatory nature of inflection, as proved by the fact that these gaps will be filled by a non-related stem if a regular form is unavailable. The circumstances are different in derivation since, not being grammatically compulsory, its application cannot be anticipated and hence gaps are not the exception (Štekauer, 2014, p. 357). Gaps in derivation are not unknown to word formationists, who

5

What Paradigms and What For ?

have studied the phenomenon by considering phonological, morphological and semantic factors (Raffelsiefen, 1996; Kjellmer, 2000; Stump, 2018). Precisely due to the very nature of word formation, the suitability of the qualifier defective may be called into question as a descriptor for word-formation paradigms. Table 1.2 exemplifies two different types of gap filling: Table 1.2

A gap in a derivational paradigm

root

1st order

2nd order

capital character

capitalist ∅

capitalistic characteristic

In the first paradigm, capital is the lexical base upon which capitalist and capitalistic are generated and, because the three slots are filled, it is possible to sequentially follow the derivational chain. The opposite is true for the second paradigm where, due to the non-existence of *characterist, a gap emerges in the first order of derivation, with the consequence that the root (character) cannot be morphologically linked to the second order (characteristic). This goes against the principle of uniformity of paradigms. Several ways out have been proposed in this respect, one of which is a diachronic reanalysis whereby characteristic is not derived by -ist and then -ic, but by -istic right away. This proposal gets around the problem by turning a two-order paradigm into a oneorder one, and is labelled the Full Reanalysis Hypothesis (Stump, 2018, pp. 4–5). One consequence of such reanalysis is an adjustment of the paradigm’s internal configuration in order to accommodate two forms where there were three. At least three options exist for this new configuration: Table 1.3

Three configurations post-Full Reanalysis Hypothesis

root

1st order

#1

capital

#2

capital character

#3

capital

2nd order

capitalist capitalistic character characteristic capitalist  capitalistic characteristic

capitalist character

   capitalistic characteristic

6

Fernández-Domínguez, Bagasheva, and Lara-Clares

Reanalysis #1 implies a redistribution of the paradigm’s space, which is evenly split for each of the two forms. The major problem with this analysis is that the implicit paradigmatic structure is not parallel to that of paradigms like capital-capitalist-capitalistic, and this contradicts the principle of structural equivalence across paradigms. Reanalysis #1 must hence be disregarded. The alternative of reanalyses #2 and #3 is to align one of the forms character and characteristic with one of the three forms in the other paradigm, while the other form (character or characteristic) is aligned with two forms simultaneously. In reanalysis #2, for example, character and capital are equated as roots, and characteristic assumes all further derivation, in this case through -istic, which would correspond to -ist and -ic for capitalistic. In reanalysis #3, the root of one paradigm (character) corresponds to the root and the first order of derivation in the other (capital and capitalist), and the second orders of derivation are equivalent in both. However, reanalysis #3 means that an already suffixed unit (capitalist) is paralleled to a root (character), and that a unit of a second order (capitalistic) is paralleled with a unit of one of a first order (characteristic), with morphosemantic inconsistencies as a result. Albeit with different implications, the three possibilities in Table 1.3 share a reduction of the number of slots from three to two, thus generating a minimal derivational paradigm. Because the only previous gap in the paradigm is eliminated and the remaining slots are full, Table 1.3 represents a fully saturated paradigm, which means that there is no room for new coinages in it. Lexical gaps are in fact potential naming needs. These factors all make it advisable to accept paradigmatic gaps, as in Table 1.2, rather than to avoid them at the expense of reducing the number of slots (Table 1.3). The (non-)existence of lexical gaps is synchronically circumstantial and inherent to derivation, and clearing them off paradigms would exert the undesirable effect of precluding potential words, in turn restraining derivational morphology from its main purpose: lexical creation. This is precisely Bauer’s (1997) position when he maintains that “[…] the availability of the slots [is] more important for the paradigm than the forms which fill them” (p. 253; Dammel & Kempf, 2018, pp. 80–81). The above seems good reason to assert that lexical gaps are perhaps not a prerequisite, but a major boost for morphological productivity, since they favour the use of morphological processes to meet that need and instigate phenomena like analogy, morphological competition or blocking, among others. Gaps play a key role also in relation to the predictability of paradigm members, since predictability comes into play especially for empty slots. The study of predictability in word formation is not new and has been approached from various perspectives, e.g., onomasiological (Štekauer, 2005), psycholinguistic (Gagné, Marchak, & Spalding, 2010) and, more recently, in relation to derivational paradigms explicitly (Bonami & Strnadová, 2019). In a nutshell, the

What Paradigms and What For ?

7

debate concerns the degree to which the forms and meanings of the units that occupy the slots in a paradigm can be successfully anticipated. Contrary to widespread belief, it has been shown that cell predictability is as probable in word formation as it is in inflection, even if paradigmatic gaps are more frequent in the former (Bauer, 1997; Antoniová & Štekauer, 2015). 2.3 Usefulness of Paradigms in Word Formation One central matter that is at the same time a source of disagreement is which morphological processes should be a part of paradigmatic word formation. Viewpoints have ranged in their degree of restrictiveness. If one particular goal is to put derivation on the same level as inflection, then only affixation should probably be involved in the constitution of paradigms, as it is the vehicle of expression of inflection (e.g., Table 1.1; Matthews, 1974; Pounder, 2000, p. 47; ten Hacken, 2014). The array of morphological operations available for word formation, by contrast, is considerably broader and, with slight variation across languages, it includes affixation, back-formation, blending, clipping, compounding, conversion, or parasynthesis. As it may be manifest, limiting the span of derivational paradigms to affixation means dismissing the lexical output of not few processes, a central decision because these processes generate the bulk of complex words in some languages (Štekauer, Valera, & Körtvélyessy, 2012). Štekauer (2014) describes two conceptions of derivational paradigms in this respect. The first conception is a narrow one where only affixation participates and where the result is a paradigm with systematic, regular and predictable relationships, akin to an inflectional one, at the cost of limiting the number of potential words. This is the most adequate view, in Štekauer’s opinion. The second conception is a broad one which welcomes all the complex words that are produced by word-formation processes with a common stem, which thus increases the slots for potential words but is “[…] paradigmatically vacuous because it does not lead to a predictable and regularly organized system of complex words” (2014, p. 369). The involvement of different morphological processes in derivational paradigms has been treated differently in the specialised literature. Unsurprisingly, virtually every survey into word-formation paradigms has studied prefixation or suffixation in one way or another, and not few have strictly focused on these processes (e.g., Fehringer, 2003; Beecher, 2004; Janssen, 2005; Booij & Lieber, 2014; Antoniová & Štekauer, 2015; Boyé & Schalchli, 2016; Gaeta & Angster, 2018; Stump, 2018, among many others). It is less frequent to find surveys of paradigmaticity into other morphological process. After affixation, compounding has been the secondary centre of attention for specialists, some giving it full consideration and others dealing with it only in passing (Bybee, 1985,

8

Fernández-Domínguez, Bagasheva, and Lara-Clares

pp. 105–109; van Marle, 1985; Stump, 1991; Bauer, 1997, p. 253; Pounder, 2000, pp. 82–95; Booij, 2008; Dammel & Kempf, 2018). Among these, van Marle’s (1985, pp. 16–21, 121) contribution is noteworthy because, being a pioneer in the paradigmatic dimension of derivation, he already analyses the behaviour of compounds in this area. Bagasheva (2015) also discusses the advantages of word-formation paradigms for a uniform analysis of compound verbs in that they are created by a miscellany of processes, namely, compounding proper, conversion and back-formation but constitute a uniform lexical class of words with identical morphosyntactic properties. Conversion is dealt with generally in van Marle (1985), although this process is not the principal focus of the monograph. The key to the inclusion/exclusion of conversion in paradigmatic word formation is that converted units do not constitute for van Marle a morphological category, as the formal characteristics that they share come from their behaviour, namely, their positioning in the sentence as members of a word-class different from their original one. Units created by conversion would not therefore meet the formal criteria of word formation. Van Marle (1985) also claims that these units are not complex words from a semantic point of view either because their meaning similarities have more to do with their categorial value: “[t]he semantic characteristics that the nouns at issue have in common, to put it differently, only result from the fact that they are nouns” (pp. 144–145). The opposite is found in Becker (1993), where noun>verb and verb>noun conversion are discussed on a par with other concatenative morphological processes and are presented as a type of cross-formation, that is, a word-formation rule which is equally productive in both directions. Conversion is regarded as a part of derivational paradigms in this and other works (Antoniová, 2016; Bonami & Strnadová, 2019; Dammel & Kempf, 2018). Traditionally of lesser importance in the catalogue of word-formation processes, back-formation was first studied from a paradigmatic perspective in Becker (1993), where it is shown to be of paramount relevance due to its unconventional directionality of derivation. Becker (1993) discusses matters related to the diachronic and synchronic facets of back-formation, as well as the semantic and formal dependence between base and derivative. His main conclusion is that back-formation is not only relevant synchronically, but also fully pertinent for a paradigmatic theory because, just as there exists a morphological rule for pairs like work > worker, another valid rule will be used for burglar > burgle. In his words, “[b]esides differences in productivity and in the morphophonological operations (affixation vs. affix subtraction), they are of the same rule type” (Becker, 1993, p. 7; see Booij, 2012, pp. 40–41; Štekauer, 2015).

What Paradigms and What For ?

9

Remarkably enough, no in-depth study has been carried out to the best of our knowledge on the role of neoclassical compounding and combining forms in word-formation paradigms, and neither on the processes of lexical blending, acronymy, clipping or eponymy. Theoretical Constructs or Entities with Psychological Reality for Speakers Though not explicitly addressed as a null hypothesis, the existence of paradigms as psychologically real network-like structures in the mental lexicons of speakers has been amply tested in psycholinguistic experiments (see Libben & Jarema, 2004). In essence, psycholinguistic research on family size effects (see Bertram, Baayen, & Schreuder, 2000) and the nature of the relationship between constituents in compounds (e.g., Gagné & Shoben, 2002) tacitly presume the existence of some memory guiding principle which facilitates the speakers’ processing of complex words. It might be suggested that all the principles being tested are different paradigmatic dimensions as the ones Bauer (2017) identifies for the variety of paradigms. As can be gleaned from Gagné et al. (2010), meaning predictability is a central issue in psycholinguistic research; meanwhile predictability (see §2.2) is a golden standard desideratum for a paradigm. The debate over relational (thematic) vs. attributive relations in root nominal compounds (see for example Estes, Golonka, & Jones, 2011 and Estes, 2003) illustrates different possible meaning-based dimensions of paradigmatic structures. The very concept of second order schemas (as defined by Booij & Masini, 2015, p. 47) seems to fit in degrees of granularity and abstractness with the idea of local paradigms. “[V]iolation of Fregean compositionality can be accounted for by means of ‘second order schemas’, i.e., sets of two or more paradigmatically related constructional schemas” and these, being schemas, are psychological in nature. Even though the psychological reality of paradigms has not been directly tested, it is far from mere conjecture. We are a long way from gaining any direct insight as to the nature of paradigms in the mind, but research in this direction can more clearly focus on testing hypotheses about dimensions that generate and maintain paradigms in word formation. 2.4

2.5 Typology of Word-Formation Paradigms The question of the possibility of different entities being recognised as paradigms is still open. Bonami (2017) identifies two ways in which paradigmaticity can figure prominently in word formation. On the one hand it could be understood

10

Fernández-Domínguez, Bagasheva, and Lara-Clares

as “paradigmatic relations between words by opposition to syntagmatic relations between words and word parts” (or Štekauer’s (2014) broad understanding of a paradigm). On the other, “it may literally extend analytic strategies originally conceived for the study of inflectional paradigms to the domain of wordformation” (Bonami, 2017; or the narrow definition of a word-formation paradigm provided by Štekauer, 2014). The first conception not only postulates a difference in the nature of paradigms in inflection morphology and word formation, but also posits a different principle for word-formation paradigms. In a similar vein of reasoning, Spencer (2017) argues that paradigms in inflection and word formation are “logically distinct types of relation, only superficially similar”. Both broad views described above allow for a basic difference between inflectional and word-formation paradigms and, more importantly, both positions presume a diversity of paradigm types within word formation. Spencer (2017) distinguishes “an inflectional-type paradigm, Πi, from a derivationaltype paradigm, Πd” with none of the properties shared by the two types. The differentiation goes on further to the recognition of a “‘chain-Πd’” and “‘radial-Πd’”. From the brief expounding of different versions of understanding paradigmaticity, it transpires that the diversity of types of paradigms is the rule, rather than the exception. This diversity, at least in the field of word formation, is recognised by Bauer (2017) with the following different kinds of paradigms: a paradigm of functions; paradigms of morphological material; the paradigm of lexical material; a conjugation class (which fits the criteria for a paradigm); a word class (but again it is a paradigm); a paradigm of predictable relationships (e.g., ‘made of’), etc. (spelling as in the original). Bauer discusses the following dimensions of variability among paradigm types: predictability; closeness/openness of the set; extensivity; membership (one relevant item or more than one relevant item); and specificity. The common property of all these variable structures is relatedness on the basis of a conspicuous feature. Settling the controversy with the nature of paradigms (in word formation) requires that all possible features of relatedness and any inherent hierarchy thereof be exhaustively described, which might turn out to be unattainable as the concept of prime number is for a mouse (see Piattelli-Palmarini, 2000). It might be that (at least some types of) paradigms in word formation turn out to be an “irreducible component of cognition” (Piattelli-Palmarini, 2000, p. 358). The fact that at present scholars cannot agree on the nature and typology of (word-formation) paradigms does not in any way undermine their psychological reality as matrixes aiding speakers in operation with the architecture of language.

What Paradigms and What For ?

3

11

Paradigms in Inflection vs. Derivation

Approaching the span of morphological paradigms as an either-or choice between inflection and derivation has underlying theoretical implications. Bauer (1997, pp. 243–244) is conclusive: restricting paradigms to the region of inflectional morphology implicitly means endorsing the existence of a split morphology, which sees inflection and derivation as independent grammar modules (see Anderson, 1982; Plank, 1994; ten Hacken, 2014). In contrast, allowing for (some kind of) paradigms in derivation involves perceiving a continuum between inflection and derivation, with more and less prototypical paradigms occurring along the cline depending on their features. Reservations over the existence of paradigms in word formation have most often been voiced regarding the alleged irregularity of derivational morphology. As opposed to inflection, derivation has been conventionally described as full of exceptions that are caused by phenomena like morphological competition (van Marle, 1994, pp. 2929–2930; Plag, 1999; Bauer, 2009; Aronoff, 2016) or blocking (Rainer, 1988, 2016; Giegerich, 2001; cf. Kjellmer 2000, p. 224; Don & Lin, 2014, among others). This all results in paradigms which, at least from a purely formal point of view, are defective in nature because they very often lack slots for specific categories, which is rarer in inflection. An example is shown in ‎(1), which shows lexemes for the role agent and would thus be a part of the paradigm for that semantic category: (1) teach > teacher, novel > novelist, escape > escapee, milk > milkman, spyV > spyN, express > *expresser Two facts should be differentiated in this respect, however. The first is that one and the same derivational affix is rarely applicable beyond its derivative scope (thus, teacher but *noveler, *escaper, *milker); the second fact is that the primary purpose of word formation is to express a given meaning (here, agent), and hence the form of derivatives is of secondary importance only. This is why the specific suffix (-er, -ist, -ee, etc.) is irrelevant as long as the new lexeme materializes. The same reason justifies why word-formation paradigms are relevant not only to affixation, but to any other morphological process that may coin a semantically appropriate lexeme, e.g., compounding (milkman) or conversion (spyN). These facts aside, cases are not uncommon of unattested units, e.g., *expresser which, despite being potential words, currently represent gaps in the paradigm (see Bauer, 1997, p. 245, Štekauer, 2014, p. 357, Gaeta & Angster, 2018, pp. 5–6, Olsen, 2019).

12

Fernández-Domínguez, Bagasheva, and Lara-Clares

For van Marle (1994), three characteristics shape the homogeneity of inflectional paradigms, namely that they (a) share the same base, (b) share a large number of properties, and (c) their differences are relatively small. As seems evident, extending the notion of paradigmaticity to derivation will be problematic if the above are taken as defining conditions, insofar as the systems of inflection and derivation are markedly different and direct correlation is improbable. Following van Marle (1994), “[t]he only conclusion which is justified is that in derivation paradigmatic structure may manifest itself in a fundamentally different way from the way it does in inflection” (p. 2929). Again, it is suitable to go back to de Saussure, 1916 (p. 174): Un terme donné est comme le centre d’une constellation, le point où convergent d’autres termes coordonnés, dont la somme est indéfinie.2 De Saussure’s discussion certainly makes it clear that associative relations are more characteristically found in inflection, but nowhere in it do we find an explicit exclusion of derivation. In fact, just as it is explained that “[…] pour la conscience des sujets parlants le nominative n’est nullement le premier cas de la déclinaison, et les termes pourront surgir dans tel ou tel ordre selon l’occasion”3 (de Saussure, 1916, p. 715), the same principle may be extrapolated to the field of word formation. That is, that the lexeme expected to be the basis for derivation is not always attested first in actual language use. A case in point is that of back-formation, described by many as relevant diachronically only on the basis of the listing of the hypothetical lexical base (Becker, 1993; Štekauer, 2015). 4

Description of the Volume

As has been shown, a range of research questions have been recently addressed in relation to the paradigmatic nature of word formation, not all of which have received a satisfactory answer as of today. Unsettled matters include the 2  A given term is like the center of a constellation, the point where other coordinated terms converge, the sum of which is indefinite. [Our translation] 3  For the speaking subject, the nominative is not at all the first case of the declension, and the terms can arise in one or another order depending of the occasion. [Our translation]

What Paradigms and What For ?

13

nature of paradigmatic relations in word formation and the lexicon; the interconnectedness between derivational families and paradigms; the constitutive characteristics of a word-formation paradigm; the degree of predictability that word-formation paradigms can display; the specificity of paradigms depending on the variety of recognised word-formation processes and patterns (if any); etc. The growing interest for paradigmaticity in lexical morphology is evidenced not only by the quantity and quality of related publications, but also by the recent celebration of several thematic workshops specifically dedicated to the theory and applications of the concept of paradigm in word formation (more prominently derivation). The papers that are collected here were part of the workshop Revisiting Paradigms in Word-formation, hosted within the Word-Formation Theories III and Typology and Universals in Word-Formation IV Conference, held at Pavol Jozef Šafárik University (Košice, Slovakia) in June 2018. A wide audience gathered together there for the discussion of unsettled matters in the area of word-formation paradigms, some of which are a part of this volume. Chapter 2 is authored by Alexandra Bagasheva, who discusses i) the nature of paradigmaticity (and paradigms) in word formation, ii) how paradigmaticity, motivation and analogy are related, and iii) the status of compound verbs in relation to the way compounds inform our understanding of the role of paradigms in word formation (and possibly the reverse). Bagasheva maintains that paradigms are psychologically real and that the inherent paradigmaticity of word-formation products and the role of analogy account for the growing number of compound verbs in English. Besides, it is claimed that the fact that parasynthetic adjectives rarely, if ever, function as motivating bases for compound verb derivation only strengthens the assumption of the pervasiveness of paradigms in word formation as sets of analogical products where the principle of relational priming operates. Bernard Fradin (Chapter 3) explores whether derivational paradigms can be useful to make predictions about possible lexemes in a way similar to the predictive role played by inflectional paradigms in inflection. Using an extended view of the notion of paradigm proposed by Bonami and Strnadová (2019), it is assumed that a paradigm is a collection of morphological families structured by the same system of oppositions of content, whose nature can be either morphosyntactic (inflection) or morphosemantic (derivation) and which has a predictive potential provided the elements that belong to the morphological families are properly aligned. The chapter deals with the domain of relevance and the organization of derivational paradigms and discusses existing proposals, namely those made by Roché (2017a, 2017b), and concludes that

14

Fernández-Domínguez, Bagasheva, and Lara-Clares

it is necessary to account for the variety of derivational paradigms in order to make them more efficient as descriptive devices. In the fourth chapter, Petr Kos argues against the consideration of derivational networks as paradigmatic. The chapter applies Dokulil’s (1962) major onomasiological categories, i.e., the modificational, the transpositional and the mutational, to distinguish regular and predictable relations of paradigms from an open system of complex words related only by a common stem. The mutational category, Kos argues, is not canonically paradigmatic because i) there is a significant shift in meaning between the original stem and the resulting derivation, ii) the existence of naming units depends on the naming of the extralinguistic reality and, thus, it is difficult to think of pre-existing cells which would be filled by these new naming units, and iii) the predictability of a form lowers when the decision of which (probably synonymous) suffix will be used depends on the coiner. The chapter concludes, thus, that only the modificational and the transpositional categories comply with the features of paradigmaticity put forward by Bauer (1997). Lior Laks and Faten Yousef (Chapter 5) examine the correlation between derivational paradigms and doublet formation in Palestinian Arabic and investigate the role of paradigms in morphological change. They argue that morphological change takes place in order to achieve both structural and semantic transparency in paradigms. Their exploration of cases of variation reveals that only CiCeC verbs with transitive alternates in CaCCaC have tCaCCaC doublets. The CaCCaC-tCaCCaC paradigm is found to be the most productive transitiveto-intransitive paradigm, and they claim that CiCeC intransitive verbs that are related to CaCCaC transitive verbs change their form, adhering to the most common paradigm. The chapter thus shows that words are more likely to undergo morphological change when they are part of a derivational paradigm, as the morphological change establishes more uniform paradigms. Eleonora Litta and Marco Budassi (Chapter 6) aim at determining the main requirements for the physical representation of a derivational paradigm through data from the Word Formation Latin (WFL) lexicon (Litta & Passarotti, 2019). Considering the problems in the representation of derivational families in WFL, paradigms are here represented in a tabular form so as to adopt aspects of Construction Morphology (CxM, Booij, 2012), which highlights the importance of the ‘cell’ as a meaningful location for lexemes that can be filled to add morphological and semantic information. This chapter also assesses a number of concepts from CxM with regard to their validity for a description of the function of Latin paradigm cells throughout time and space. The seventh chapter is authored by Jan Radimský, who reviews a number of types of paradigmatic relations, namely word families, word series, paradigms and paradigmatic systems, as defined in Boyé and Schalchi (2016) and

What Paradigms and What For ?

15

Bauer (2017), and discusses their relevance for an analysis of compounding. Radimský argues that the study of compounds within a paradigmatic approach allows for an account of a number of phenomena that are challenging for rulebased approaches. The paradigmatic approach is, according to the author, so closely related to CxM that “paradigms” and “constructions” seem to actually be two different names for the same concept, even though some essential differences set the two frameworks apart. This chapter underlines the relevance of the concept of paradigm for the study of compounding and argues that a paradigmatic approach may be an appropriate answer for issues concerning both learned and native Romance Noun-Noun compounds. Chapter 8, by José Antonio Sánchez Fajardo and Elizaveta Tarasova, deals with Adj+ie/y nominalisations. Specifically, it focuses on the i) amount of polysemy of deadjectival -ie diminutives and ii) the semantic convergence of the output units whose formation and semantic content is driven by an ellipsed noun, an adjectival base (colour/origin) and the suffix -ie/y. The -ie/y units in this chapter are collected from dictionaries (Ayto, 1998; Thorne, 2014; MWD11; OED3), and examples of their use are extracted from the NOW Corpus. This chapter analyses the morphosemantic features by using the onomasiological approach to word formation (Štekauer, 2005), and several relevant principles of Construction Grammar. The results suggest a gradient regularity and predictability in the formation of Adj+ie/y nominalisations and they demonstrate a correlation between the semantic features of these ‘diminutive’ constructs and their impact on an endearment-pejoration scale. Díaz-Negrillo (Chapter 9) explores neoclassical compounds and their morphologically-related lexical formations through a word-based treatment. The suitability of a paradigm-based approach is tested, with a focus on [X -scope] formations. Over 460 formations extracted using OED data are classified into 14 semantic categories which may set the base for the configuration of a set of categories for paradigm-based accounts of neo-classical compounds. Issues like paradigmatic pressure, competition or analogy are discussed in this chapter, which concludes that a word-based approach to neoclassical word formation has a number of advantages compared to morpheme-based approaches. Acknowledgement This publication has been supported by the Spanish State Research Agency (SRA, Ministry of Economy and Enterprise) and European Regional Development Fund (ERDF) (Ref. FFI2017-89665-P).

16

Fernández-Domínguez, Bagasheva, and Lara-Clares

References Anderson, S. R. (1982). Where’s morphology? Linguistic Inquiry, 13, 571–612. Antoniová, V. (2016). Derivational paradigm – is there any? A contrastive research. SKASE Journal of Theoretical Linguistics, 13(2), 91–105. Available at: http://www .skase.sk/Volumes/JTL32/pdf_doc/05.pdf [Accessed 28 September, 2018]. Antoniová, V., & Štekauer, P. (2015). Derivational paradigms within selected conceptual fields – Contrastive research. Facta Universitatis, Series: Linguistics and Literature, 13(2), 61–75. Aronoff, M. (1976). Word-formation in generative grammar. Cambridge, MA: MIT Press. Aronoff, M. (1983). Potential words, actual words, productivity and frequency. Proceedings of the 13th International Congress of Linguists, Tokyo, 163–171. Aronoff, M. (2016). Competition and the lexicon. In A. Elia, C. Iacobino, & M. Voghera (Eds.), Livelli di Analisi e fenomeni di interfaccia. Atti del XLVII congresso internazionale della società di linguistica italiana (pp. 39–52). Rome: Bulzoni. Ayto, J. (1998). Oxford dictionary of slang. Oxford: Oxford University Press. Bagasheva, A. (2015). Paradigmatic word-formation, metonymy and compound verbs in English and Bulgarian. In J. Audring, N. Koutsoukos, F. Masini, & I. Raffaelli (Eds.), Morphology and semantics—MMM9 on-line Proceedings, 1–11. Available at: http://www.lilec.it/mmm/wp/wp-content/uploads/2015/02/2276-3980-1-SM.pdf [Accessed 10 January, 2018]. Bauer, L. (1997). Derivational paradigms. In G. Booij & J. van Marle (Eds.), Yearbook of morphology 1996 (pp. 243–256). Dordrecht: Kluwer. DOI: 10.1007/978-94-017-3718-0_13 [Accessed 10 January, 2018]. Bauer, L. (2001). Morphological productivity. Cambridge: Cambridge University Press. Bauer, L. (2009). Competition in English word-formation. In A. van Kemenade & B. Los (Eds.), The handbook of the history of English (pp. 177–198). Malden, MA: Blackwell. Bauer, L. (2017). Notions of paradigm and their value in word-formation. Paper delivered at ParadigMo 2017: First Workshop on Paradigmatic Word-formation Modeling, Toulouse, 19–20 June 2017. Becker, T. (1993). Back-formation, cross-formation, and ‘bracketing paradoxes’ in paradigmatic morphology. In G. Booij & J. van Marle (Eds.), Yearbook of Morphology 1993 (pp. 1–25). Dordrecht: Kluwer. Beecher, H. (2004). Derivational paradigm in word formation. Available at http:// citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.94.9071&rep=rep1&type=pdf [Accessed 10 September, 2018]. Bertram, R., Baayen, R. H., & Schreuder, R. (2000). Effects of family size for complex words. Journal of Memory and Language, 42, 390–405. Blevins, J. P. (2013). Word-based morphology from Aristotle to modern WP (Word and Paradigm models). In K. Allan (Ed.), The Oxford handbook of the history of linguistics (pp. 375–396). Oxford: Oxford University Press.

What Paradigms and What For ?

17

Bonami, O. (2017). Predictability in inflection and word-formation. Paper delivered at ParadigMo 2017: First Workshop on Paradigmatic Word-formation Modeling, Toulouse, 19–20 June 2017. Bonami, O., & Strnadová, J. (2019). Paradigm structure and predictability in derivational morphology. Morphology, 1–31. DOI: 10.1007/s11525-018-9322-6 [Accessed 24 September, 2018]. Booij, G. (2008). Paradigmatic morphology. In B. Fradin & G. Dal (Eds.), La raison morphologique. Hommage á la mémoire de Danielle Corbin (pp. 29–38). Amsterdam & Philadelphia, PI: John Benjamins. Booij, G. (2012). The grammar of words. An introduction to linguistic morphology. Oxford: Oxford University Press. Booij, G., & Lieber, R. (2014). On the paradigmatic nature of affixal semantics in English and Dutch. Linguistics, 42(2), 327–357. Booij, G., & Masini, F. (2015). The role of second order schemas in the construction of complex words. In L. Bauer, L. Körtvélyessy, & P. Štekauer (Eds.), Semantics of complex words (pp. 47–66). Berlin: Springer. Boyé, G., & Schalchli, G. (2016). The status of paradigm. In A. Hippisley & G. Stump (Eds.), The Cambridge handbook of morphology (pp. 206–234). Cambridge: Cambridge University Press. Bybee, J. (1985). Morphology. A study of the relation between meaning and form. Amsterdam & Philadelphia, PI: John Benjamins. Carstairs-McCarthy, A. (1987). Allomorphy in inflection. London: Croom Helm. Dammel, A., & Kempf, L. (2018). Paradigmatic relationships in German action noun formation. Zeitschrift für Wortbildung / Journal of Word-Formation, 2, 52–85. Available at: https://www.peterlang.com/fileasset/Journals/ZWJW_2-2018_book.pdf [Accessed 24 January, 2019]. de Saussure, F. (1916). Cours de linguistique générale. Lausanne-Paris: Payot. Dokulil, M. (1962). Tvoření slov v češtině 1, Teorie odvozování slov. Prague: Nakladatelství Československé akademie věd. Don, J., & Lin, J. (2014). A syntagmatic analysis of ‘paradigmatic’ morphology. In R. Kager, J. Grijzenhout, & K. Sebregts (Eds.), Where the principles fail: a festschrift for Wim Zonneveld on the occasion of his 64th birthday (pp. 29–40). Utrecht: Utrecht Institute of Linguistics OTS. Estes, Z. (2003). Attributive and relational processes in nominal combination. Journal of Memory and Language, 48, 304–319. DOI: 10.1016/S0749-596X(02)00507-7. Estes, Z., Golonka, S., & Jones, L. (2011). Thematic thinking: the apprehension and consequences of thematic relations. In B. Ross (Ed.), Psychology of Learning and Motivation (Vol. 54, pp. 249–294). Burlington: Academic Press. Fehringer, C. (2003). Morphological ‘gangs’: Constraints on paradigmatic relations in analogical change. In G. Booij & J. van Marle (Eds.), Yearbook of Morphology 2003 (pp. 249–272). Dordrecht: Kluwer.

18

Fernández-Domínguez, Bagasheva, and Lara-Clares

Gaeta, L., & Angster, M. (2018). Stripping paradigmatic relations of out the syntax. Morphology, 1–22. DOI: 10.1007/s11525-018-9326-2 [Accessed 24 September, 2018]. Gagné, C. L., & Shoben, E. (2002). Priming relations in ambiguous noun-noun combinations. Memory and Cognition, 30(4), 637–646. Gagné, C. L., Marchak K. A., & Spalding, T. L. (2010). Meaning predictability and compound interpretation: A psycholinguistic investigation. Word Structure, 3, 234–251. Giegerich, H. J. (2001). Blocking and the Elsewhere Condition: Lexical morphology and the speaker. Transactions of the Philological Society, 99(2), 65–98. Giegerich, H. J. (2012). The morphology of -ly and the categorial status of ‘adverbs’ in English. English Language and Linguistics, 16(3), 341–359. Horecký, J., Buzássyová, K., & Bosák, J. (1989). Dynamika slovnej zásoby súčasnej slovenčiny. Bratislava: Veda. Janssen, M. (2005). Between inflection and derivation: Paradigmatic lexical functions in morphological databases. In J. D. Apresjan & L. L. Iomdin (Eds.), East-West Encounter: Second International Conference on Meaning and Text Theory (pp. 187– 196). Moscow: Slavic Culture Languages. Kjellmer, G. (2000). Potential words. Word, 51(2), 205–228. DOI: 10.1080/00437956 .2000.11432501. Libben, G., & Jarema, G. (2004). Conceptions and questions concerning morphological processing. Brain and Language, 90, 2–8. DOI: 10.1016/j.bandl.2003.12.005. Litta, E., & Passarotti, M. (2019). (When) Inflection Needs Derivation: A Word Formation Lexicon for Latin. In N. Holmes, M. Ottink, J. Schrickx, & M. Selig (Eds.), Words and Sounds (pp. 224–239). Berlin, Boston: De Gruyter. Matthews, P. H. (1974). Morphology. An introduction to the theory of word-structure. Cambridge: Cambridge University Press. MWD11: Merriam-Webster dictionary online, 11th edition, Merriam-Webster, Inc. https://www.merriam-webster.com [Accessed 10 December 2017]. NOW: Davies, M. (2013). Corpus of News on the Web: 3+ billion words from 20 countries, updated every day. http://corpus.byu.edu/now/ [Accessed 30 August, 2018]. OED3: Oxford English Dictionary, 3rd edition, Oxford University Press. https://oed.com [Accessed 14 December, 2017]. Olsen, S. (2019). The relatedness of meaning in derivational patterns. In J. Audring, N. Koutsoukos, & C. Manouilidou (Eds.), MMM12 Online proceedings (pp. 64–76). https://mmm.library.upatras.gr/mmm/article/view/3250/3510 [Accessed 26 February, 2020]. Piattelli-Palmarini, M. (2000). An “irreducible” component of cognition. In R. Singh, C. Krimbas, D. Paul, & J. Beatty (Eds.), Thinking about evolution: Historical, philosophical, and political perspectives (pp. 353–376). Cambridge: Cambridge University Press.

What Paradigms and What For ?

19

Plag, I. (1999). Morphological productivity: Structural constraints in English derivation. Berlin: Mouton De Gruyter. Plank, F. (1994). Inflection and derivation. In R. E. Asher (Ed.), The encyclopedia of language and linguistics (Vol. 3, pp. 1671–1678). Oxford: Pergamon. Pounder, A. (2000). Processes and paradigms in word-formation morphology. Berlin & New York, NY: de Gruyter. Raffelsiefen, R. (1996). Gaps in word-formation. In U. Kleinhenz (Ed.), Interfaces in phonology (pp. 194–209). Berlin: Akademie Verlag. Rainer, F. (1988). Towards a theory of blocking: the case of Italian and German quality nouns. In G. Booij & J. van Marle (Eds.), Yearbook of Morphology 1998 (pp. 155–185). Dordrecht: Foris. Rainer, F. (2012). Morphological metaphysics: virtual, potential, and actual words. Word Structure, 5(2), 165–182. Rainer, F. (2016). Blocking. In Aronoff (Ed.), Oxford Research Encyclopedias: Linguistics. DOI: 10.1093/acrefore/9780199384655.013.33. Roché, M. (2017a). Les familles dérivationnelles: comment ça marche? Toulouse: Université Toulouse 2 Jean Jaurès. Roché, M. (2017b). Un exemple de réseau constructionnel: ethnique, toponymes, gentilés. Toulouse: Université Toulouse 2 Jean Jaurès. Spencer, A. (2017). Participles: inflectional paradigms, derivational paradigms or something else? Paper delivered at ParadigMo 2017: First Workshop on Paradigmatic Word Formation Modeling, Toulouse, 19–20 June 2017. Štekauer, P. (2005). Meaning predictability in word-formation: Novel, context-free naming units. Amsterdam, Philadelphia, PI: John Benjamins. Štekauer, P. (2014). Derivational paradigms. In R. Lieber & P. Štekauer (Eds.), The Oxford handbook of derivational morphology (pp. 354–369). Oxford: Oxford University Press. Štekauer, P. (2015). Backformation. In P. O. Müller, I. Ohnheiser, S. Olsen, & F. Rainer (Eds.), Word-formation. An international handbook of the languages of Europe (pp. 340–351). Berlin, New York, NY: de Gruyter Mouton. Štekauer, P., Valera, S., & Körtvélyessy, L. (2012). Word-formation in the world’s languages. A typological survey. Cambridge: Cambridge University Press. Stump, G. (1991). A paradigm-based theory of morphosemantic mismatches. Language, 67(4), 675–725. Stump, G. (2018). Some sources of apparent gaps in derivational paradigms. Morphology, 1–31. DOI: 10.1007/s11525-018-9329-z [Accessed 24 September, 2018]. ten Hacken, P. (2014). Delineating derivation and inflection. In R. Lieber & P. Štekauer (Eds.), The Oxford handbook of derivational morphology (pp. 10–25). Oxford: Oxford University Press.

20

Fernández-Domínguez, Bagasheva, and Lara-Clares

Thorne, T. (2014). Dictionary of contemporary slang. London: Bloomsbury Publishing. van Marle, J. (1985). On the paradigmatic dimension of morphological creativity. Dordrecht: Foris. van Marle, J. (1994). Paradigms. In R. E. Asher (Ed.), Encyclopedia of language and linguistics (Vol. 6, pp. 2927–2930). Oxford: Pergamon. Veselinova, L. N. (2006). Suppletion in verb paradigms. Bits and pieces of the puzzle. Amsterdam & Philadelphia, PI: John Benjamins.

Chapter 2

Paradigmaticity in Compounding Alexandra Bagasheva 1

Introduction

Paradigmaticity as a concept of a potential heuristic value in word formation has been in the recent upsurge of interest discussed mainly in relation to affixation, which admittedly is most readily analysed in such terms. An interesting question in the study of paradigmaticity in word formation is the unresolved argument about the way in which paradigmaticity implicates and is implicated by the concepts of motivation and analogy. Closely associated with these controversies is the open debate of the requirements for regularity, synchronic transparency of derivational pattern/rule and predictability as properties of paradigms in word formation. These and many other still open questions (for an overview of such issues see Bauer, 1997, 2017; Štekauer, 2014) in the study of paradigms in word formation render research in this area topical (see Lipka, 1981; van Marle, 1985; Bochner, 1993; Bauer, 1997, 2017; Pounder, 2000; Beecher, 2004; Hathout, 2011; Štekauer, 2014; Spencer, 2017; etc.). In relation to the above rough outline of a very few basic controversial issues in word formation, the current chapter aims to answer the following research questions: 1) how are paradigmaticity, motivation and analogy related (if) and 2) do compounds have an impact on the way we think of paradigms in word formation and how do the latter inform our understanding of the former? In order to meet its objectives, the chapter proceeds as follows: §2 reviews the rebirth of the concept of paradigm in word formation and touches upon the typology of paradigms and possible concomitant differences between paradigms in inflectional morphology and word formation; §3 is focused on the concepts of motivation and analogy and their relevance for understanding paradigmaticity, in §4 some controversies in analysing compound verbs and parasynthetic adjectives (in English) and plausible solutions to those associated with a paradigmatic approach are presented; §5 concludes.

© Koninklijke Brill NV, Leiden, 2020 | doi:10.1163/9789004433410_003

22 2

Bagasheva

The Rebirth of Paradigms in Word Formation

The discussion of paradigms was initially exclusively or predominantly the prerogative of inflectional morphology (see Anderson, 1992; Aronoff, 1994; Stump, 2001; Bonami & Stump, 2016; etc.). For the past several decades the interest in paradigms in word formation has been steadily growing (Bochner, 1993; Bauer, 1997, 2017; Booij, 1997; Pounder, 2000; Hathout, 2011; Roché & Plénat, 2014; Štekauer, 2014; etc.). While in inflectional morphology a paradigm is conceived of as a set of paradigm functions and realization rules (at least in one strand of inflection morphology, see Bonami & Stump, 2016), the nature of paradigmatic relations in word formation is not yet exactly clear, irrespective of differences in details stemming from different theoretical persuasions (Štekauer, 2014). Furthermore, in the former a paradigm is clearly understood as “a central principle of morphological organization” (Stump, 2001, p. 32), while the status of paradigmaticity as a universal principle in word formation or as an alternative resorted to when irregularity and lack of a rule require explanation is still under discussion. Nonetheless, it is clear today that paradigms are widely recognized as at least theoretical constructs and analytical heuristics to be fruitfully applied in word formation. Admittedly, for the most part paradigms research in word formation is mostly exploratory, even though the basic utility of paradigms is claimed to be explanatory, i.e., it is assumed that paradigmaticity is the mechanism accounting for the architecture of the grammar (Bonami & Stump, 2016) and it is basically assumed to apply to affixation. Meanwhile psycholinguistic research on the morphological family size effect (see e.g., Moscoso del Prado Martin, Kostic, & Baayen, 2004) and the processing of compounds (Libben & Jarema, 2006; Gagné & Spalding, 2009; Gagné, Marchak, & Spalding, 2010) have provided ample evidence for the psychological reality of the word-formation paradigm and the strongly paradigmatic organization of the mental lexicon, including compounds as a special type of complex words. There are two distinct lines of development of the idea of paradigmaticity in word formation: i) one comes from word-formation research proper, where paradigmaticity is evoked to explain irregularities, discrepancies, mismatches between meaning and form, bracketing paradoxes, generally in all instances where a standard syntagmatic approach relying on rules or directed relations between derivatives in a pair fails (Booij & Lieber, 2004; Stump, 2010); and ii) the second line is an extension of (or at least a closely approximate parallel to) paradigms in inflectional morphology. One of the numerous and varied

Paradigmaticity in Compounding

23

explanations of bracketing paradoxes postulates a mismatch between the morphosemantics and the morphotactics of a lexeme. From the standpoint of paradigm function morphology, Stump (2010, p. 207) contends that it is fallacious to expect a derived lexeme’s logical structure to be isomorphic to its morphological realization: the meaning of the lexeme ATOMIC SCIENTIST is related to the meaning of the lexeme ATOMIC SCIENCE, but this relation is not mediated by ATOMIC SCIENTIST’s morphological realization as [atomic [scient-ist]]. This contention arises from the second line of development of the idea of paradigms in word formation, namely the extension of the concept as a central productive organizing principle from inflection morphology to word formation. Bonami (2017) succinctly summarizes the two distinct ways in which an approach to word formation can be paradigmatic. On the one hand, it may focus on relations in absentia (paradigmatic) between words against mainstream focusing on combinability relations (syntagmatic) between word parts (for similar views see van Marle, 1985; Becker, 1993). On the other hand, it may literally extend analytic strategies originally conceived for the study of inflectional paradigms (see among many others Matthews, 1972; Aronoff, 1994; Stump, 2001; Bonami & Stump, 2016) to the domain of word formation. The problem arises with the second understanding of derivational paradigmaticity since Plank (1994) maintains that inflectional categories form “a more or less closed system, being alternatively realized by a limited number of terms”, while derivational categories are “not part of any well-organized categorial system”. (p. 1672) We can easily object that derivational paradigms are also based on expressing features of content categories or ontological types with relevance for language as Cruse defines these (Cruse, 2000), for example, the categories Agent, Patient, Instrument, Location, Abstraction, Iterativity, Cause, Result of action, Diminutiveness, Augmentativeness, etc. After all, Bolozky (1999) maintains that “lexical formation is first and foremost semantically based and concept driven” (p. 7), which are the categories which word formation operates with. The problem is that these are elusive, open-ended and very difficult to formalize. In practice, the first approach emphasizes the fact that often in word formation the application/definition of a rule is not possible (i.e., syntagmaticity fails and we need to resort to principles and relations in absentia), while the second highlights the conceptual-semantic nature of the essential categories we, both speakers and analysts, operate with in word formation.

24

Bagasheva

There is no compelling reason why we should be looking for a concept of paradigm that will be equally potent in both inflection and derivation. It sounds intuitively true that “the inflectional paradigm interfaces primarily with syntax, while the word-formation paradigm interfaces almost exclusively with lexical semantics/conceptual structure” (Spencer, 2017). Consequently, there might be distinct properties of what might unequivocally be recognized as paradigm in its instantiations in different contexts. Acknowledging this, Spencer (2017) recognizes two distinct types of paradigms (with Bauer’s (2017) contention that “only the systematicity of relations of content matters to the identification of paradigmatic systems” remaining a must), which share the prerequisite of sameness of relational content. The two types might be dubbed inflectional-type paradigm, Πi, and a derivational-type paradigm, Πd. The latter type can be further differentiated into two subtypes: a ‘chain-Πd’ where a syntagmatic process holds precedence for the establishment of derivational sequences where a so-called ‘recursive’ application operates: employ → re-employ → re-employable → re-employability and a ‘radial-Πd’ subtypes which subsumes sets of complex lexemes all related directly to a single base: employ → {employer, employee, employable, employment, re-employ… Spencer, 2017

Even this differentiation and the subsequent recognition of different types of word-formation paradigms, however, remains captive to the idea of a single base as the source of all related forms in a paradigm no matter what the relation exactly is. This belief seems to impoverish the concept of paradigm. More compelling is the idea that paradigmatic models involve word-formation relations that may be oriented both ways to avoid any causal relations or pure syntagmaticity, which necessarily starts from input through linear combinability into a well-formed output, or have an unspecified direction. More importantly these relations are not limited to base derivative pairs: they are multidirectional and may involve any number of members, since openness is one of the defining characteristics of word-formation paradigms. In practice, this requests the identification of the properties that define word-formation paradigms without translating any such from those established in inflectional morphology. This is a task still in its infancy for two main reasons. First, most research within word formation is restricted to derivational (i.e., affixation) paradigms and inconspicuously subscribes to ideas of successive derivative pairs or series which result in word nests or families and second, for the most part such accounts are process-based, which ultimately means they remain linear. The reason for that

Paradigmaticity in Compounding

25

is the tacit resistance to do away with the classical well-established processes in word-formation research: affixation, conversion, truncation, composition, reduplication, etc. The concept of word-formation paradigm allows for the disjunction between word-formation processes and their canonical products. It collapses further the opposition between rule- and process-based interpretations of word-formation phenomena. It devalues the opposition between schema- or analogy-based interpretations and process-based interpretations. As Booij (2010, p. 94) maintains, if instead of trying to differentiate the scope of application of abstract rules (a process- or operation-focused approach, e.g., Bach, 1989, p. 46) and schemas, we subscribe to an analogy-based approach, it would be possible to apply the adequate degree of granularity of analysis to be able to describe a wide variety of word-formation data. The concept of the wordformation paradigm operationalizes Booij’s idea of analogy and rule-based processes as constituting endpoints along a scale of schematicity (Booij, 2010, pp. 93–95). The word-formation paradigm, conceived of as a network of lexicosemantically (i.e., conceptually) motivated relations between words is indispensable in analysing the growing numbers of compound verbs in English against the lack of a “productive schema” (Lamberty & Schmid, 2013) and explaining the fact that parasynthetic adjective compounds are abundant but rarely, if ever, serve as input for compound verbs. 3

Rules and Paradigmaticity: the Role of Motivation and Analogy

3.1 Paradigmaticity and Motivation Syntagmaticity and regularity (including concepts of productivity) underlie all process-informed approaches to word formation. A syntagmatic approach excludes the possibility of associations outside the components involved in the determinant-determinatum structure resulting from the application of a syntagmatic process of morphotactic ordering and morphosemantic computation (disregarding issues of (non-) compositionality as this goes beyond the scope and immediate objectives of this chapter). As Arndt-Lappe (2015) claims, “[r]egular processes are the product of an abstract formalism that operates independently from individual lexemes, on symbolic features that are shared by pertinent lexemes” (p. 103). Meanwhile, according to Saussure (1968, p. 296), the meaning of complex words is motivated and potentially computable on that basis. Quoting Saussure, Becker (1993) defines morphology as “the system of paradigmatic relations between words, new words being formed in to existing ones” (p. 1). More importantly, the scholar contends that “[i]n a

26

Bagasheva

paradigmatic morphological theory, words need not have inherent morphological structure. They have their structure through their relations to other words” (Becker, 1993, p. 3). Analysing the structure of complex words necessitates the establishment of their relations to other words, which is one of the simplest and theory-neutral definitions of motivation. Motivation is taken to be the overriding principle of language structure in cognitive accounts of the architecture of language (Lakoff, 1990 and a cohort of authors whose ideas can be found in Evans, Bergen & Zinken, 2006), as a harnessing mechanism over wild arbitrariness. In their introduction to a volume dedicated to motivation in language, Radden and Panther (2004, p. 2) refer to Heine’s (1997, p. 3) contention that since “[h]uman behavior is not arbitrary but […] driven by motivations”, language structure, which is a product of behavior, “must also be motivated”. Such a position can be directly read off Becker’s claims, even though the scholar does not explicitly associate motivation and paradigms in word formation. The term motivation has multiple interpretations in linguistics from intrinsic and extrinsic, through ecological, genetic, experiential, perceptual, cognitive, communicative to other unidentified motivations (for overviews, see Radden & Panther, 2004 and Panther & Radden, 2011). A linguistic sign, identified as the target, is identified as motivated “to the extent that some of its properties are shaped by a linguistic or non-linguistic source and language-independent factors” (Panther & Radden, 2011, p. 9). This definition suggests that any product of word formation is motivated, at least linguistically. The interdependence between motivation and paradigmaticity lies in the fact that both undermine the importance of any immediate process of derivation and allow for multiple contributing factors for the creation of a single lexeme (sign). In a paradigmaticity informed analysis motivation figures prominently. Analyzing complex words from the point of view of paradigmatic structuring allows for disregarding the process of word formation (e.g., affixation, conversion, etc.) and directionality of derivation (input-output causal chain) and excludes the necessity of a one to one mapping between the morphotactics and morphosemantics of the output (Stump, 2010). Štekauer (2014) states that within the onomasiological approach to word formation “the derivational paradigm in the narrow sense is conceived as an ordered set of all complex words directly derived from a single basic (motivating) word” (Štekauer, 2014, p. 363, emphasis added). In this narrow sense in a derivational paradigm, the motivating word is the point of departure for an ordered series of synchronically transparent derivations whose products constitute a nest (a subset of a word-formation family). A paradigmatic approach necessarily requires the identification of conceptual and/or formal links among lexemes, i.e., either a recurrence of form

Paradigmaticity in Compounding

27

(different fractions thereof) or an elaboration or alternative construal of content. Thus, paradigmaticity can be interpreted as the specification of all linguistic and cognitive factors that motivate related lexemes. Construal as used in mainstream cognitive linguistics (for an overview see Croft & Cruse, 2004) is defined as “alternative expressions for what appears to be truth-functionally equivalent situations” (Croft & Cruse, 2004, p. 40). Indispensable in both conceptualization and language and central in identifying alternative but equivalent construals is the concept of the frame. Frames account for relations of motivation. A frame as the central type of knowledge structure with direct links to lexical items is a “system of concepts related in such a way that to understand any one of them you have to understand the whole structure in which it fits” (Fillmore, 2006, p. 373). The scholar (Fillmore, 2006, p. 378) defines the relatedness between frames, construal mechanisms and lexical items as a mutually implicating one in which frame is “the structured way in which the scene is presented or remembered, we can say that the frame structures the wordmeanings, and that the word evokes the frame”. Barsalou and Hale (1993) believe “[h]uman knowledge […] to be frames all the way down” (p. 131). Frames “represent all types of categories, including categories for animates, objects, locations, physical events, mental events and so forth” (Barsalou, 1992, p. 29). In essence, frames constitute the gestalts against which the lexico-semantic relations in one type of word-formation paradigm are established. These relations are based on different profilings of a background frame (on the role of frames in word formation see Koch, 1999). 3.2 Analogy and Paradigmaticity Two other types of relations, besides the conceptual ones operationalized as frames and construal mechanism at word level, underlie paradigmatic relations in word formation. The ubiquitous of analogy makes these other types of paradigmatic relations in word-formation paradigms possible. Analogy is either recognized as a productive mechanism in synchronic morphology (Krott, 2009) or it is, according to Arndt-Lappe, appealed to in all cases “in which linguistic behaviour is not rule-governed, but exceptional, unproductive, unpredictable, or irregular” (Arndt-Lappe, 2015, p. 92). Analogy is generally interpreted as a proportional correlation, a heuristic device utilized by speakers for creating new lexemes and by linguists as a profitable tool of analysis. An unavoidable caveat, even for word-based approaches, which accept analogy as a productive synchronic mechanism, remains the fact that “the basis for the computation of similarity is not part of the equation” (Arndt-Lappe, 2015, p. 99). It is very difficult to establish which of the multiple similarity relations that exist between words in the mental lexicon may form the basis for

28

Bagasheva

analogical formations. In the words of Bauer (1983) “[i]f instances of wordformation arise by analogy then there is in principle no regularity involved, and each new word is produced without reference to generalizations provided by sets of other words with similar bases or the same affixes: a single existing word can provide a pattern, but there is no generalization” (p. 294). Leech, Mareschal, and Cooper (2008, p. 358) define analogy as a domaingeneral cognitive skill, which arises naturally from normal functioning of a memory system. More importantly, the ability to use and generate analogies “is argued to arise from priming of relations that hold between terms in the analogy” (ibid.). It is claimed that one of the crucial constraints acting on this ability is the “maximum relational complexity that can be processed in parallel in working memory” (ibid., p. 359). The authors go on, quoting (Halford, Wilson, & Phillips 1998, p. 803) to define complexity as “the number of related dimensions or sources of variation” (ibid.). Translated into paradigm analytical terms, this implies that we may not be able to determine whether it is the morphotactic, morphosemantic or any emergent property of the “relation between constituents”, and probably they often work in tandem, but relational analogy is exactly the heuristic that active coiners and comprehenders use, which is extremely relevant for compounds as evidenced from the bulk of psychological work on processing compounds (see Gagné, 2002; Marelli, Gagné, & Spalding, 2017; etc.). After all, as Bochner (1993) claims, word-formation paradigms are generated on the basis of “cumulative patterns”, i.e., exposure to existing compounds triggers the elaboration of paradigms on the basis of naturally arising word-formation general analogical abilities. 4

Compound Verbs, Parasynthetic Adjectives and Word-Formation Paradigms

4.1 Compounds and Paradigms The concept of word-formation paradigm is narrowly understood to apply to affixation (derivation proper) (see Štekauer, 2014) where two dimensions intersect: i) a horizontal dimension which captures derivational series, i.e., derivatives related through cycles of derivation, and ii) a vertical dimension which captures diversity in the formal encoding of an identified conceptual category (rivalry among alternative forms), e.g., location bakery, storage, etc. When we add to this bedroom or barber shop, it becomes feasible to assume that the same principles of paradigmaticity may apply across the whole spectrum—inflectional morphology, affixation, and compounding. If differences are detectable in the paradigms in inflectional morphology and word

Paradigmaticity in Compounding

29

formation (including both affixation and compounding), they are a matter of degree, not of principle. A stumbling block appears to be the nature of meaning computation in compounding, where for the most part meaning is configured, not directly computed. We find convincing and further on adopt Bundgaard, Ostergaard, and Stjernfelt’s (2006) reformulation of compositionality, which can well account for the nature of the semantic fusion detectable in compound verbs (and other types of compounds). According to the authors themselves, they “propose a non-trivial and non-standard compositional theory likely to capture the general way in which semantic parts of a compound configure into a semantic whole” (Bundgaard et al., 2006, p. 364; emphasis in the original). The essence of the new compositional theory is summed up in the authors’ contention that “any attempt to define a combinatorial rule in terms stemming exclusively from the linguistic system as such (qua a self-contained formal system) is doomed to fail” (ibid., p. 369). For this reason, it is quite natural to resort to something outside the linguistic systems to account for meaning generating mechanisms in compounds (including compound verbs). The analytically profitable “something” is the frame as defined above. Combining the threads of the discussion so far, we are ready to acknowledge that the word-formation paradigm should be thought of as correlation between potential, possible and actual words. The idea of potentiality is of paramount importance for the emancipation of the word-formation paradigm. As Štekauer (2014, p. 369) contends, [p]otentiality is a crucial term for the concept of derivational paradigm. It guarantees a high level of predictability and regularity […] in the sense that the existing gaps in the system can be filled anytime with actual words. The question of how we detect the gaps invites an elaboration of the dimensions of variability within a paradigm so that expected combinations thereof can be used for determination of the gaps. The hypothesis is put forward here that a word-formation paradigm is to be understood as a network of lexical constructions (words) held together by conceptual-semantic, formal (exclusively constituent-based) and/or pattern relations (a term which in an attempt to avoid confusing tautology actually refers to analogy as relational priming as an underlying principle). Each of these types of underlying principles constitutes a different subtype of word-formation paradigm. The differences lie in the degree of openness of each subtype, degree of predictability and immediate principle involved. The three organizing principles are as follows:

30

Bagasheva

a) meaning-based subtype (where conceptual/ontological categories underlie the projection of the paradigm’s members), b) form-based subtype (where series of compounds establish a formal pattern with at least one identical constituent), and c) analogy-based subtype (where an exemplar is used as a model for pattern imitation). The first subtype is relatively closed, since there are a restricted number of conceivable construals associated with an ontological type. The semantic relations are based on conceptual plausibility determined by the ontological types: “thing quality quantity place time state process event action relation manner” (Cruse, 2000, p. 49). Not all possible relations are actualized in a paradigm, only those that are triggered by “pragmatic pressure” (Booij & Lieber, 2004, p. 350). This subtype comprises paradigm members which are based on different profilings of a background frame as the central type of knowledge structure with direct relations with lexical items. As a method of analysis, frame semantics necessarily involves the study of the unidirectional backgrounding/foregrounding relations between concepts and the lexical items evoking and evoked by them. Each actualized lexeme out of the set of potential words represents a uniquely profiled portion of a scene/ frame. Admittedly, this understanding of ‘frame’ is aimed at pinning down how a lexical item correlates with conceptual knowledge, i.e., it specifies interface conditions between the representation of knowledge in the human mind and the latter’s abstraction and encoding in the symbolic inventory of language, or a schematic lexical concept (Evans, 2009). A way to narrow down the scope of the analytical concept ‘frame’ and understand how a lexical concept is a schematization of a conceptual concept (tautology intended) is to supplement Fillmore’s definition with the more restrictive understanding offered by Goldberg. (1) a. A word sense’s semantic frame (what the word ‘means’ or ‘evokes’) = profile + background frame. b. A word sense’s profile: what the word designates, asserts. c. A word sense’s background frame: what the word takes for granted, presupposes (Goldberg, 2020, p. 40). Such a definition captures the most essential feature of frames, namely their ability to invite inferences in a relevance-constrained manner in language processing. Nouns profile or construe (in Langacker’s (2008, p. 43) sense) more or less static, conceptually autonomous wholes, while verbs profile “participants interacting energetically in a ‘force-dynamic’ event” (Langacker, 2008, p. 103).

Paradigmaticity in Compounding

31

The conceptual-semantic word-formation paradigm can be equated with the background frame, which can be construed or profiled alternatively and surface as a nominal lexical concept, verbal or adjectival one. As an illustration we can compare the profiling of the noun bear hug and the verb to bear hug: in the verb the whole background is activated, i.e., the projected conceptualization involves the activity, its force-dynamic dimension, potential participants (at least two), etc., while in the noun the projected conceptualization profiles in “summary scanning” manner (Langacker, 1987, p. 144) the type of activity. 4.2 Compound Verbs While compound nouns and adjectives constitute non-disputed compounding objects, the status of compound verbs in English is highly contested and most approaches deny the existence of genuine compound verbs (for overviews see Lamberty, 2014, among others; for the opposite view see Erdmann, 2000, 2009; Wald & Besserman, 2002; Bagasheva, 2012, 2014a, b). The issue of paradigmaticity in word formation is crucial in the study of compound verbs since it changes the very conception of the genuineness of verbal compounds. What counts as a compound verb receives two different answers when we apply a syntagmatic approach which focuses on the nature of the derivation (i.e., the immediate process) and a paradigmatic approach (which avoids directionality and a source-output causal relation). Ever since Marchand’s (1969, p. 101) identification of pseudo compounds such as to babysit, this line of reasoning has persisted in the study of compounds. Lamberty (2014, p. 226) dismisses as nongenuine verb compounds to babysit, to chain-drink, and to cold-shoulder on grounds of their being “related to a nonverbal base lexeme and treated as derivations”. Meanwhile, the aim of her extensive corpus analysis is to establish criteria that “qualify a lexeme as a potential verbal compound”. Interestingly, most verb pseudo-compounds are disqualified on the basis of bottom-up analysis, which focuses on process of derivation, while a top-down functional analysis identifies such lexemes as compound verbs since they function as the predicator in clauses or as temporally marked relational concepts, and depending on their transitivity ranking, they project different argument configurations. In essence, a paradigmatic approach is a compromise between a top-down and a bottom-up analysis. It appears that the recognition of verb compounds as compounds is an artifact of the preferred analysis. In an analysis which does not recognize a cause-effect relationship between a process of word formation and nature of the products as a necessary and sufficient condition for the identification of certain complex lexeme, in a paradigmatic approach, for example, lexemes will still be recognized as compounds, even if they do not obligatorily

32

Bagasheva

arise from compounding. Lieber and Štekauer (2009, p. 2) voice a similar possibility of disjunctively approaching compounds and compounding by wondering “whether compounds exist as a distinct species of word formation”. Lamberty and Schmid (2013, p. 591) claim that “speakers of English apparently do not have a productive schema for the creation of genuine verbal compounds at their disposal”, yet they are exposed to such compounds and process them with ease. The authors find an explanation for this seeming paradox (compound verbs are not processed as the result of compounding but are freely formed and easily understood) in the speakers’ deployment of “different processing strategies […], trying to take recourse to possible base nouns or adjectives and interpreting meanings on the basis of analogies to similar lexical items in the network” (ibid.; emphasis added). This network is actually the super-paradigm structure in which all three subtypes of word-formation paradigm intersect and it renders full compound status to compound verbs, irrespective of the way in which they arise. In such a view the three traditionally recognized types of compound verbs: i) incorporating verbs—gift-wrap, spoon-feed, rough-dry, husband-hunt, boyfriend-drop, name-ambush, massproduce, etc.; ii) compounding/composition proper verbs—kick-start, stir-fry, sleepwalk, sleep-talk, etc.; and iii) converted verbs—moonlight, piggyback, brownbag, redshirt, bear hug, etc. can be studied as the uniform class of lexemes that they actually constitute. Incorporating compound verbs are usually claimed to be associated with synthetic or parasynthetic compound nouns or adjectives as their source, while converted ones are based on a root [NN/AdjN] compound. This traditional trichotomy of compound verbs is patently syntagmatic and is based on the identification of a process, i.e., a rule-based procedure of unifying an input and output. Many scholars (among whom are Lieber, 1981, 2004; Mayo, Schepping, Schwarze, & Zaffanella, 1995; Lieber & Baayen, 1997; and Plag, 1999) maintain that regularity in the semantics of derived words depends on the derivational process/pattern involved. Lieber (2004), in particular, develops a theoretical model which captures and formally implements the semantics of (English) word formation in a system of lexical semantic representation based on decomposition. In Lieber’s model the semantic representation of a lexical item is believed to have two harmonious but separate components: a semantic-grammatical skeleton and a semantic-pragmatic body, the latter being only partially amenable to systematic analysis, if at all. In compound lexemes the skeletons of two stems are co-indexed. The model provides “a framework of lexical semantic representation that makes it possible to explain in a uniform way the interpretation of both root compounds and synthetic compounds and that does not appeal to the internal structure of those compounds” (Lieber, 2004, p. 48; emphasis added). Promising as this sounds,

Paradigmaticity in Compounding

33

it fails to provide for a uniform analysis of compound verbs. First, the typology of compounds based on morphosyntactic features is maintained (root vs. synthetic, with the latter requiring a (de)verbal base) and second it excludes all converted compound verbs. Conversion is understood as relisting, defined in its own turn as the process in which “an item already listed in the lexicon is re-entered as an item of a different category” (Lieber, 2004, p. 90). Again, the uniformity of compound verbs is refuted. The paradigmatic approach does not differentiate between the three subtypes of compound verbs, neither does it predict different semantic properties of the verbs on the basis of source-process-output. Instead, the ecology of compound verbs is uniformly recognized within the super-paradigm structure— the profiling of a relational concept with background of the frame always included in its onomasiological structure. Not taking into consideration the syntagmatic notions of input, process and directionality, it is easy to recognize that there are no differences in the series of lexemes: to apple-polish, apple-polishing (n), apple-polisher (n), applepolished (adj.) and apple-polishing (adj.), namedropping (n), to name-drop (v), name-dropper (n) and namedropping (adj.), even though the first is claimed to have the verb as a motivating word (in the sense of Stekauer’s definition of the term motivating word) and the second a noun. The uniformity might be explained by the “flexible type-token, degrammmaticalized part of speech system in English” (Vogel, 2000) and “the ambicategorial status of participles, which are known to combine verbal and adjectival features” (Hilpert, 2015, p. 117; references there), without resorting to the heuristics of paradigmaticity. If we accept the validity of the above explanations, then there should be the form *to book-read,1 derivable from *book reading. Assuming that the motivating word is an ambivalent form which could be an activity noun, an adjective and a verbal participle, it is expected that a proper compound verb will also be coinable. Indeed, to lip-read meets that expectations, but *to book read does not. There cannot possibly be morphological or syntactic constraints, since any conditions deriving from word formation (back-derivation) or syntax (argument relation: incorporation of the affected) are identical for the two compounds. The question remains why should one be an actualized compound verb, while the other just a possible, potential word. The triviality of *book 1  While editing the chapter Cristina Lara-Clares (to whom my gratitude goes for this) noted the following can be found in the iWEB: Think CCIE or RHCE—two exams that are very hard to book-read your way through because you actually have to physically DO something to pass (https://community.infosecinstitute.com/discussion/comment/175280/#Comment_175280). This finding only comes to support the argumentation provided below, despite Lamberty’s (2014) claims for the non-existence of *to book read.

34

Bagasheva

reading as opposed to the nameworthiness of lip-reading seems to be the only plausible explanation. Nameworthiness has been promoted to the status of a constraint formulated by Goldberg (2010, p. 50) as follows, Conventional Frame constraint: For a situation to be labelled by a verb, the situation or experience may be hypothetical or historical and need not be directly experienced, but it is necessary that the situation or experience evoke a cultural unit that is familiar and relevant to those who use the word. Even this convincing interpretation fails in the face of to studentteach, also claimed to be derived from a secondary, synthetic nominal compound (or to babysit for that matter), which is an attested or actualized compound verb. The triviality of studentteaching does not block the creation and use of the compound verb. 4.3 A Detour on Parasynthetic Adjectives Almost identical is the issue associated with structures recognized as parasynthetic adjectival compounds, such as able-bodied. Admittedly good-looking (adj.) and good-looker (n) are the only attested members of the respective series, while the supposedly easily derivable *good-looking (n) and *to good-look (v) are not attested words. Promoting as a reliable criterion for individuation the non-attested status of the second constituent and the non-existence as a compound base of the combination of the other two constituents, Melloni and Bisetto (2010) assign such structures to a special class of parasynthetic compounds. They insist that such compounds are formed on the basis of an ordered sequence between compounding and affixation with the affix being attached to a non-existent compound formed by “the merger of two lexical stems” (2010, p. 199). The “binary steps of these forms” (ibid., p. 200) and the unique “ternary” structure of such compounds render them significantly different from synthetic ones. Melloni and Bisetto (2010) provide a derivational model for such compounds, which is diametrically opposed to the one provided by Stump (2010): [S]ynthetic compounds such as sheep-stealer, theater-goer, and evildoer, in which the agentive suffix does not join with the verbal base in isolation: *stealer, *goer, *doer […] might […] be brought into conformity with the hypothesis that [Rule of composition for Deriv: Deriv([X Y])=[X Deriv(Y)]] is invariably valid by appealing to the distinction between absolute and conjunct forms. An absolute form is used in

Paradigmaticity in Compounding

35

isolation from any compounded formatives; a conjunct form is used in the presence of compounded formatives. Many languages exhibit absolute/conjunct distinctions in their morphology. Stump, 2010, pp. 209–210

The first explanation (Melloni & Bisseto, 2010) is provided by following a strictly syntagmatic approach, with specific focus on the process of word formation and the specific nature of the input. The second explanation (Stump, 2010) comes from a paradigmatic approach, where paradigmatic features are an extension from inflection morphology (the second line of word-formation paradigmaticity in Bonami’s (2017) terms. The problem remains that such lexemes are recognized as an analytical problem for syntagmatic, process-, and/or rule-based approaches, and, second, that compounds stand out as special, requiring the positing of either special rules or special compound formatives, i.e., conjuncts, in the case of extended paradigmatic features from inflection morphology to word formation. More importantly, none of these solutions to the controversies involved answers the question of paramount importance for us: why can we studentteach but not *cardrive. In our opinion, it is only a matter of socio-pragmatic salience for a compound verb to surface as an onomatologically realized symbolic unit. Only the naming needs of speakers determine what potential, possible words will be actualized. There is no other more ecologically effective explanation of the facts than the recognition of a word-formation paradigm whose specific nature cannot be explained with features extended from paradigms in inflection morphology. Paradigmaticity in word formation has its specific features that arise from the nature of cell/gap filling. The most significant differences in the cell/gap filling processes in inflection morphology and in word formation comprise at least the following: i) novelty (cell filling in inflection morphology draws from a finite pre-given set in which changes take a long time to occur, since these changes affect the structure of communicative message. Gap filling in word formation affects the content of messages, which is usually situationally easily recoverable and innovation is welcome); and ii) degree of abstractness of the meaning side in inflectional and word-formation constructions. The basic tenet of constructionalism has it that language is a constructicon, a set of taxonomic networks where each construction constitutes a node in the network that forms a continuum from the fully concrete to the highly schematic (for relevant overviews see Croft, 2007; Sag, Boas, & Kay, 2012). The relations between constructions are ones of inheritance and motivation. The constructicon is acquired via language use and innovated via neoanalysis (the term is

36

Bagasheva

chosen over the more traditional reanalysis. The reasons for that are detailed in Traugott & Trousdale, 2013) and analogization. Both processes are localized within constructions, or more precisely in actualized constructs. Form and meaning can in the incremental stages of constructionalization undergo individual shifts, but the co-evolution hypothesis (Bybee, Perkins, & Pagliuca, 1994) holds true for the whole construction, preserving the biuniqueness of the symbolic complex. A construction is instantiated in actual language use by specified constructs that are fully phonetically specified and have contextually sensitive meaning, based on their conventionalized meaning or any appropriate extension thereof. A shift in any dimension of the construct might be strengthened via propagated use across the speech community into a modified or novel construction depending on the degree of dissimilarity from the initial one(s). Constructionalization of what are traditionally recognized as lexical objects involves fewer levels of abstraction and schematicity. Though incremental or gradient, this difference in the type of concepts expressed by a symbolic complex accounts for the different nature of paradigmaticity in different areas of the constructicon (inflectional morphology vs. word formation). Exclusively word-formation constructions have the properties of basic level concepts (on nature and typology of concepts see Howard, 1992 and Murphy, 2002), while inflection constructions are superordinate concepts. 4.4 Tying Up the Threads For the purposes of synchronic paradigmatic analysis, compound verbs are best studied as instantiating a constructional idiom2 (Michaelis & Lambrecht, 1996; Goldberg, 2006; Booij, 2007, 2010; Masini, 2009) represented at several levels of schematicity and abstraction as members of a super-paradigm of other lexemes with which they are in relations of one of the following types: a) inheritance relations (those from the conceptual-semantic paradigm), b) formal relations (or the traditional family with same constituent members, with positional peculiarities), and c) relational analogues. The constructional idiom is in the form of a construction schema, “a cognitive representation comprising a generalization over perceived similarities among instances of usage” resulting from “repeated activation of a set of co-occurring properties” (Kemmer & Barlow, 2000, p. xxiii). The elements of the construction schema are variously lexically specified. The unifying property of all compound verbs is that they are relational simulators in the sense defined by Wu and Barsalou (2009, p. 186), 2  For a detailed analysis of compound verbs in English and Bulgarian as members of the Constructicon, see Bagasheva (2012).

Paradigmaticity in Compounding

37

[b]ecause simulations are typically situated, the thematic relations that exist between situation components are implicitly available in them, and can be represented explicitly through the application of relation simulators. We assume that much of the background knowledge important for conceptual combination arises through these same mechanisms. The constructional idiom, whose distinguishing feature is that it functions as a relation simulator, determines the profiling of the whole as a verbal lexical concept. In the process of lexically instantiating the constructional idiom, compound verbs constitute subschema niches with variable membership, dependent on the analogical potential of the local niche, where niches are understood, as defined by Hüning (2009), as subgeneralized unities which “show semantic fragmentation: in the course of time they develop ‘semantic niches’, i.e., groups of words” (p. 183). The notion of simulation in processing language constitutes part of the “embodiment premise”, as formulated by Gibbs (2006), “[p]eople’s subjective, felt experiences of their bodies in action provide part of the fundamental grounding for language and thought” (p. 9). The crucial postulate of this premise is that “[m]eaning […] isn’t just abstract mental symbols; it’s a creative process, in which people construct virtual experiences—embodied simulations—in their mind’s eye” (Bergen, 2012, p. 46). One significant division in the compound verb constructicon stems from the nature of these embodied simulations or meaning configuring principles. These lexical items can be divided into attribute-value foregrounding verbs in which a frame specification and the background are both included in the profiling, i.e., a frame component is onomatologically realized (e.g., to hunt: job-hunt, apartment-hunt, house-hunt, husband-hunt, etc.) and concept modification compounds, which piggyback on an already configured conceptual complex (e.g., to moonlight, to bellyache, to redshirt, etc.). Compound verbs in English are consistently categorially and morphologically right-headed (for inflectional purposes), but display wide variability in terms of semantic endo/exocentricity. As special construal mechanisms compound verbs combine relational and situational properties. They capture the transformation of situational salience into ontological salience and as special type of conceptualizations, they occupy an intermediate position between simple relational symbolic units and conceptual core descriptions (Radden & Dirven, 2007) which motivates their rising usage (Bagasheva, 2012). Compound verbs are used in diverse semantic domains from purely physical activities like island-hop, spin-dry, smoke-dry to abstract domains like emotional states and decision-making, as well as interpersonal relations such as:

38

Bagasheva

hag-ride, cherry-pick, short-change, spoon-feed, browbeat, etc., so their extralinguistic motivation is not restricted in any meaningful way. Despite the neatness of the offered analysis,3 one problem remains unsolved; why don’t we *good-look or *record-break but mindread? It is not a trivial matter that the majority of (so-called parasynthetic) adjectival compounds in English do not seem to project micro-paradigms involving compound verbs. A random list of such adjectives, e.g., record-breaking; mouth-watering; thoughtprovoking; slow-moving; far-reaching; computer-based, etc. seem to conform to the claim, i.e., they are not associated with a corresponding compound verb, though they have a corresponding compound noun, e.g., good-looker, recordbreaker, slow-mover, etc. with the common semantics of ‘bearer of quality X’. The explanation provided by Brömser (1985), namely that parasynthetic adjectives correlate with collocations and are formed via constituent inversion, stumbles over the following facts: a) there are cases when inversion does not seem to be involved (e.g., one’s mouth waters—mouth-watering) and b) to hand-carve is a compound verb claimed to be derived from the past participle hand-carved. We are faced with the same systemically unmotivated case as in to lipread and to mindread as actualized compound verbs but *to bookread as an only potential, possible compound verb. Another offered explanation is blocking, i.e., the existence of well entrenched verb complement constructions with identical meaning, e.g., to look good, to break the record, etc. Again, this fails to account why to hand-carve or to mindread are possible, even though the collocations to carve with one’s hands and to read one’s mind are also entrenched expressions. The only plausible explanation can be sought in the nature of the paradigmatic relations (conceptual-semantic, formal, relational) and the fact that these form a hierarchy. Parasynthetic compound adjectives are derived and processed via analogy with an established exemplar via the reversal of constituent order of an alternative construction. Within the construction, analogybased paradigms comprise constituents not related conceptually/semantically, but analogically replicating an exemplar, in the manner in which this explains the cases of violation of the “no-direct-object constraint” in noun participial compounding “through analogy with […] usage patterns” (Hilpert, 2015, pp. 118–119). The analogical subtype of word-formation paradigm overrides the other two subtypes. Construal-grounded word-formation paradigms comprise re-profiled lexical items and are restricted exclusively by the underlying 3  The claims made here are based on preliminary conceptual-semantic analysis of 462 compound verbs extracted from COCA, BNC, OED, urban dictionary, and word spy. Details of the analyses can be found in Bagasheva 2012, 2014a, and 2014b.

Paradigmaticity in Compounding

39

ontological types and possible construals. Form- and analogy-based ones are constituted by replication either of positionally and functionally marked constituents or of complex relations. The non-actualization of a compound verb is most frequent in potential parasynthetic adjectives as motivating words, since analogy-based paradigms take precedence over the conceptual-semantic subtype as the former impose less processing pressure. Such a hypothesis is consistent with the claim made by Arndt-Lappe, Bell, Schäfer, and Schlücker (2016, p. 107) concerning all kinds of compounds and “the looseness of the link between their formal characteristics and their semantic interpretation”. Any processing problem that such a loose link between meaning and form can pose is resolved by the “maximization of opportunity” view of language processing, according to which “the system makes maximum and opportunistic use of the information that is available” (ibid., emphasis added). It is this opportunism that differentiates paradigms in inflectional morphology and word formation. Support for all the claims made so far comes from the iWEB where compound verbs are used (derived from parasynthetic adjectives), even though in the literature and the source of data used here they are not recognized or recorded. 1. a) If you’ve ever had someone mind-boggle you with a line such as4 … (https://junkee.com/make-friends-semester-two/115110) b) You never cease to mind-boggle the masses. (http://awbw.amarriner .com/prevmaps.php?maps_id=12310) 2. It becomes difficult to find balance when technology is being used to time-save how often do you check email and voice messages after hours and away (https://www.netnanny.com/blog/6-ways-to-find-balance-in-a -world-of-tech-addiction/) 3. My husband and I are trying to forward-think our lives (https://the kitchensgarden.com/2017/06/10/a-momentous-decision/) No prediction can be made as to which parasynthetic adjectives will give rise to compound verbs. The feat of lexical creation is accomplished by speakers and their inferencing abilities using an “emergent”,5 complex adaptive perceptual-symbol communicative system. The emergent, multi-agent nature of embodied language and the complex role of the individual as the internalizing and producing agent simultaneously in incessant contextualized interactions shape the nature of word-formation paradigms, where potentiality and opportunistic gap-filling are key properties. 4  I owe all examples from the iWEB to Cristina Lara-Clares. 5  The term emergent is used here as a shorthand for the understanding that language is a complex adaptive system (Beckner et al., 2009) and meaning in language is a synergetic, emergent phenomenon (Frank, 2015).

40

Bagasheva

The human mind (according to Libben, 2010) tries to maximize efficiency neither by reducing how much is stored, nor by reducing how much is computed. It seeks to both store and compute as much as possible. Libben (2014, p. 8) explains how this works for compounds as follows, as a result of acts of lexical processing, the constituents of compound words develop into new lexical representations. These representations are bound to specific morphological roles and positions […] within a compound word. The development of these positionally bound compound constituents creates a rich network of lexical knowledge that facilitates compound processing and also creates some of the well documented patterns in the psycholinguistic and neurolinguistic study of compounding. It could only be added that what is special about compounds is their specific, pre-syntactic status defined by Jackendoff (2009), for whom compounds are a proto-grammatical phenomenon. He identifies them as the area of modern language which preserves the semantico-pragmatic grounding of the protolanguage stage. He goes on to suggest that compounding is “the same phenomenon writ small” (Jackendoff, 2009, p. 114), where same refers to the nature of the semantic relations that guide discourse structuring and utterance comprehension which are not syntactically marked. 5

Conclusions

Ecological plausibility is accepted as a golden standard in all attempts to analyse the human mind and as Jackendoff (1983, p. 3) puts it: “to study semantics of natural language is to study cognitive psychology”, i.e., the human mind. This view is shared by Libben and Weber (2014, p. 205), who specifically insists that we should never forget the psychological nature of morphological structures. Driven by such considerations, Libben (2014) promotes a psychocentric view of compounds. In this view (which from our standpoint happens to be essentially paradigmaticity-informed), word formation operates in a manner which economizes on the number of patterns in the minds of speakers by maximizing on transcendence through morphological proliferation (Libben & Weber, 2014). Libben (2014, p. 8) claims that as a result of acts of lexical processing, the constituents of compound words develop into new lexical representations. These representations are bound to specific morphological roles and positions […] within a

Paradigmaticity in Compounding

41

compound word. The development of these positionally bound compound constituents creates a rich network of lexical knowledge that facilitates compound processing and also creates some of the well documented patterns in the psycholinguistic and neurolinguistic study of compounding. Paradigms are a convenient tool to model and analyse networks of lexical representations of compounds. They provide a uniform analytical tool for compound verbs despite their disparate derivational nature. Just as “theories of concepts generally ignore background situations, focusing largely on bottom-up, stimulus-based processing” (Yeh & Barsalou, 2006, p. 349), in word formation for a very long time most research targeted bottomup, linear analyses with a focus on processes on amodal symbols. In many modern theories of concepts, these are recognized as situated, and concept modifications in relation to an immediate situation are the rule, rather than the exception. Paradigms in word formation appear to be the alternative topdown approach that can capture the opportunistic nature of everyday human reasoning and non-specialist use of human language as a complex, adaptive, emergent system. Paradigms in the subsystem of compounds constitute three different subtypes, with the one based on pattern analogy taking precedence over the other two subtypes. The former’s primacy stems from the ecology of the human processing system. This explains the randomness of actualized compound verbs associated with parasynthetic adjectives as motivating words. References Anderson, S. (1992). A-morphous morphology. Cambridge: Cambridge University Press. Arndt-Lappe, S. (2015). Word formation and analogy. In P. O. Müller, I. Ohnheiser, S. Olsen, & F. Rainer (Eds.), Word formation: An international handbook of the languages of Europe (Vol. 2, pp. 89–149). Berlin: Mouton de Gruyter. iBooks. Arndt-Lappe, S., Bell, M., Schäfer, M., & Schlücker, B. (2016). Introduction: Modelling compound properties. Morphology, 26, 105–108. Aronoff, M. (1994). Morphology by itself. Cambridge: MIT Press. Bach, E. (1989). Informal Lectures on Formal Semantics. Albany: SUNY Press. Bagasheva, A. (2012). Reflections on compound verbs and compounding. Sofia: Sofia University Publishing House “St. Kliment Ohridski”. Bagasheva, A. (2014a). Frame semantics, metaphtonymy and compound verbs in English. In G. Rundblad, A. Tytus, O. Knapton, & C. Tang (Eds.), Selected papers

42

Bagasheva

from the 4th UK Cognitive Linguistics conference (pp. 1–17). London: UK Cognitive Linguistics Association. (ISSN 2046–9144). Bagasheva, A. (2014b). Compound verbs in English and Bulgarian and the relativity debate. In L. Filipović & M. Pütz (Eds.), Multilingual cognition and language use. Processing and typological perspectives (pp. 97–121). Amsterdam, Philadelphia: John Benjamin Publishing Company. Barsalou, L., & Hale, C. (1993). Components of conceptual representation: From feature lists to recursive frames. In I. Van Mechelen, J. Hampton, R. Michalski, & P. Theuns (Eds.), Categories and concepts: Theoretical views and inductive data analysis (pp. 97–144). San Diego, CA: Academic Press. Bauer, L. (1983). English word-formation. Cambridge: Cambridge University Press. Bauer, L. (1997). Derivational paradigms. In G. Booij & J. van Marle (Eds.), Yearbook of Morphology 1996 (pp. 243–256). Dordrecht: Kluwer. Bauer, L. (2017). Notions of paradigm and their value in word-formation. Paper delivered at ParadigMo 2017: First workshop on paradigmatic word formation modeling, Toulouse, 19–20 June 2017. Becker, T. (1993). Back-formation, cross-formation and ‘bracketing paradoxes’ in paradigmatic morphology. In G. Booij & J. van Marle (Eds.), Yearbook of Morphology 1993 (Vol. 6, pp. 1–25). Dordrecht: Kluwer. Beckner, C., Blythe, R., Bybee, J., Christiansen, M., Croft, W., Ellis, N., Holland, J., Ke, J., Larsen-Freeman, D., & Schoenemann, T. (2009). Language is a complex adaptive system: Position paper. Language Learning 59 (Suppl. 1), 1–26. DOI: 10.1111/j.1467-9922.2009.00533.x. Beecher, H. (2004). Derivational paradigm in word formation. Research Paper I. Retrieved from http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.94.9071& rep=rep1&type=pdf [Accessed September, 2019]. Bergen, B. (2012). Louder than words: The new science of how the mind makes meaning. New York: Basic Books. BNC: British National Corpus at: http://www.natcorp.ox.ac.uk/ [Accessed September, 2018]. Bochner, H. (1993). Simplicity in generative morphology. New York: Mouton de Gruyter. Bolozky, S. (1999). Measuring productivity in word-formation: the case of Israeli Hebrew. Leiden: Brill. Bonami, O. (2017). Predictability in inflection and word formation. Paper delivered at ParadigMo 2017: First Workshop on Paradigmatic Word Formation Modeling, Toulouse, 19–20 June 2017. Bonami, O., & Stump, G. (2016). Paradigm Function Morphology. In A. Hippisley & G. Stump (Eds.), The Cambridge handbook of morphology (pp. 449–481). Cambridge: Cambridge University Press.

Paradigmaticity in Compounding

43

Booij, G. (1997). Autonomous morphology and paradigmatic relations. In G. Booij & J. van Marle (Eds.), Yearbook of Morphology 1996 (pp. 35–54). Dordrecht: Kluwer Academic Publishers. Booij, G. (2007). The grammar of words: An introduction to linguistic morphology. Oxford: Oxford University Press. Booij, G. (2010). Compound construction: Schemas or analogy? A construction morphology perspective. In S. Scalise & I. Vogel (Eds.), Cross-disciplinary issues in compounding (pp. 93–107). Amsterdam, Philadelphia: John Benjamins Publishing House. Booij, G., & Lieber, R. (2004). On the paradigmatic nature of affixal semantics in English and Dutch. Linguistics 42(2), 327–357. Brömser, B. (1985). On the derivation of English verbal compounds. In W. Kürsch-Ner & R. Rüdiger Vogt (Eds.), Grammatik, Semantik, Textlinguistik: Akten des 19. Linguistischen Kolloquiums Vechta 1984 (Band 1, pp. 99–113). Tübingen: Niemeyer. Bundgaard, P., Ostergaard, Sv., & Stjernfelt, F. (2006). Waterproof fire stations? Conceptual schemata and cognitive operations involved in compound constructions. In Semiotica 161(1–4), 363–393. Bybee, J., Perkins, R., & Pagliuca, W. (1994). The evolution of grammar: Tense, aspect and modality in the languages of the world. Chicago, IL: Chicago University Press. CoCA: Davies, M. (2008–) The Corpus of Contemporary American English (COCA): 425 million words, 1990–present. Available online at https://www.english-corpora.org/ coca/ [Accessed September, 2018]. Croft, W. (2007). Construction grammar. In D. Geeraerts & H. Cuyckens (Eds.), The Oxford handbook of cognitive linguistics (pp. 463–508). Oxford: Oxford University Press. Croft, W., & Cruse, A. (2004). Cognitive linguistics. Cambridge: Cambridge University Press. Cruse, A. (2000). Meaning in language: An introduction to semantics and pragmatics. Oxford: Oxford University Press. Erdmann, P. (2000). Compound verbs in English: Are they pseudo? In G. Tops, B. Devriendt, & S. Geukens (Eds.), Thinking English grammar to Honour Xavier Dekeyser, Professor Emeritus (pp. 239–252). Leuven: Peeters. Erdmann, P. (2009). Compound verbs. In G. Rohdenburg & J. Schluter (Eds.), One language, two grammars? Differences between British and American English (pp. 38–59). USA: Cambridge University Press. Evans, V. (2009). How words mean: Lexical concepts, cognitive models and meaning construction. Oxford: Oxford University Press. Evans, V., Bergen, B., & Zinken, J. (2006). The cognitive linguistics reader. Sheffield: Equinox Publishing.

44

Bagasheva

Fillmore, Ch. (2006). Frame semantics. In D. Geeraerts (Ed.), Cognitive linguistics. Basic readings (pp. 373–400). Berlin, New York: Mouton de Gruyter. Frank, R. (2015). A complex adaptive systems approach to language, cultural schemas and serial metonymy: Charting the cognitive innovations of ‘fingers’ and ‘claws’ in Basque. In J. E. Díaz-Vera (Ed.) Metaphor and metonymy across time and cultures. Perspectives on the sociohistorical linguistics of figurative language (pp. 65–95). Berlin, New York: Mouton de Gruyter. Gagné, Ch. (2002). Lexical and relational influences on the processing of novel compounds. Brain and Language, 81(1–3), 723–35. DOI: 10.1006/brln.2001.2559. Gagné, Ch., Marchak, K., & Spalding, Th. (2010). Meaning predictability and compound interpretation: A psycholinguistic investigation. Word structure, 3(2), 234–251. Gagné, Ch., & Spalding, Th. (2009). Constituent integration during the processing of compound words: Does it involve the use of relational structures? Journal of memory and language, 60, 20–35. Gibbs, R., Jr. (2006). Embodiment and cognitive science. New York: Cambridge University Press. Goldberg, A. (2006). Constructions at work: The nature of generalization in language. Oxford: Oxford University Press. Goldberg, A. (2010). Verbs, constructions and semantic frames. In M. Rappaport Hovav, E. Doron, & I. Sichel (Eds.), Syntax, lexical semantics, and event structure (pp. 39– 58). Oxford: Oxford University Press. Halford, G. S., Wilson, W. H., & Phillips, S. (1998). Processing capacity defined by relational complexity: Implications for comparative, developmental, and cognitive psychology. Behavioral and Brain Sciences, 21, 803–865. Hathout, N. (2011). Une approche topologique de la construction des mots: propositions théoriques et application à la préfixation en anti-. In M. Roché, G. Boyé, N. Hathout, St. Lignon, & M. Plénat (Eds.), Des unités morphologiques au lexique (pp. 251–318). Paris: Hermès/Lavoisier. Heine, B. (1997). Cognitive Foundations of Grammar. New York, Oxford: Oxford University Press. Hilpert, M. (2015). From hand-carved to computer-based: Noun-participle compounding and the upward strengthening hypothesis. Cognitive linguistics, 26(1), 113–147. Howard, R. (1992). Classifying types of concept and conceptual structure: Some taxonomies. European journal of cognitive psychology, 4(2), 81–111. Hüning, M. (2009). Semantic niches and analogy in word-formation: Evidence from contrastive linguistics. Languages in contrast, 9(2), 183–201. Jackendoff, R. (1983). Semantics and cognition. Cambridge, MA: MIT Press. Jackendoff, R. (2009). Compounding in the parallel architecture and conceptual semantics. In R. Lieber & P. Štekauer (Eds.), The Oxford handbook of compounding (pp. 105–128). Oxford: Oxford University Press.

Paradigmaticity in Compounding

45

Kemmer, S., & Barlow, M. (2000). Introduction: A usage-based conception of language. In M. Barlow & S. Kemmer (Eds.), Usage based models of language (pp. vii–xxvii). Stanford, CA: CSLI Publications. Koch, P. (1999). Frame and contiguity: On the cognitive bases of metonymy and certain types of word-formation. In K.-U. Panther & G. Radden (Eds.), Metonymy in language and thought (pp. 139–167). Amsterdam, Philadelphia: John Benjamins Publishing House. Krott, A. (2009). The role of analogy for compound words. In J. Blevins & J. Blevins (Eds.), Analogy in grammar: Form and acquisition (pp. 118–136). Oxford: Oxford University. Lakoff, G. (1990). Women, fire and dangerous things: what categories reveal about the mind. Chicago: University of Chicago Press. Lamberty, A. (2014). Why we don’t cardrive or bookread, but slavedrive and lipread: A cognitive-linguistic approach to verbal compounds and pseudo-compounds in English. Darmstadt: Büchner-Verlag. Lamberty, A., & Schmid, H-J. (2013). Verbal compounding in English: A challenge for usage-based models of word-formation? Anglia, 131(4), 591–626. Langacker, R. W. (1987). Foundations of Cognitive Grammar, Vol. I: Theoretical Pre­ requisites. Stanford, CA: Stanford University Press. Langacker, R. W. (2008). Cognitive grammar: A basic introduction. Oxford, New York: Oxford University Press. Leech, R., Mareschal, D., & Cooper, R. (2008). Analogy as relational priming: A developmental and computational perspective on the origins of a complex cognitive skill. Behavioral and brain science, 31(4), 357–378. Libben, G. (2010). Compounds, semantic transparency, and morphological transcendence. In S. Olson (Ed.), New impulses in word-formation [Linguistische Berichte Sonderheft 17] (pp. 212–232). Hamburg: Buske. Libben, G. (2014). The nature of compounds: A psychocentric perspective. Cognitive Neuropsychology, 31(1–2), 8–25. Libben, G., & Jarema, G. (Eds.). (2006). The representation and processing of compound words. Oxford: Oxford University Press. Libben, G., & Weber, S. (2014). Semantic transparency, compounding, and the nature of independent variables. In F. Rainer, W. U. Dressler, F. Gardani, & H. C. Luschützky (Eds.), Morphology and meaning (pp. 205–221). Amsterdam, Philadelphia: John Benjamins Publishing House. Lieber, R. (1981). Morphological conversion within a restrictive theory of the lexicon. In M. Mootgart, H. Hulst, & H. Hoekstra (Eds.), The scope of lexical rules (pp. 161–200). Dordrecht: Foris Publications. Lieber, R. (2004). Morphology and lexical semantics. Cambridge: Cambridge University Press.

46

Bagasheva

Lieber, R., & Baayen, H. (1997). A semantic principle of auxiliary selection in Dutch. In Natural Language and Linguistic Theory, 15, 789–845. Lieber, R., & Štekauer, P. (2009). Introduction: status and definition of compounding. In R. Lieber & P. Štekauer (Eds.), The Oxford handbook of compounding (pp. 1–25). Oxford: Oxford University Press. Lipka, L. (1981). On the interrelation of syntagmatic modification and paradigmatic lexical structuring in English. In H. Geckeker, B. Schliebben, J. Trabant, & H. Weydt (Eds.), Logos semantikos: Studia linguistica. In honorem Eugenio Coseriu 1921–1981. (Vol. 3, Semantics, pp. 373–384). Berlin: Mouton de Gruyter. Marchand, H. (1969). The categories and types of Present-Day English word-formation (2nd ed.). München: Beck. Marelli, M., Gagné, C. L., & Spalding, T. L. (2017). Compounding as abstract operation in semantic space: A data-driven, large-scale model for relational effects in processing of novel compounds. Cognition 166, 207–224. Masini, F. (2009). Phrasal lexemes, compounds and phrases: A constructionist perspective. Word structure, 2(2), 254–272. Matthews, P. H. (1972). Inflectional morphology: A theoretical study based on aspects of Latin verb conjugation. Cambridge: Cambridge University Press. Mayo, B., Schepping, M. T., Schwarze, C., & Zaffanella, A. (1995). Semantics in the derivational morphology of Italian: Implications for the structure of the lexicon. Linguistics, 33, 883–938. Melloni, Ch., & Bisetto, A. (2010). Parasynthetic compounds: data and theory. In S. Scalise & I. Vogel (Eds.), Cross-disciplinary issues in compounding (pp. 199–217). Amsterdam, Philadelphia: John Benjamins Publishing House. Michaelis, L. A., & Lambrecht, K. (1996). Toward a construction-based theory of language function: The case of nominal extraposition. Language, 72, 215–247. Moscoso del Prado Martin, F., Kostic, A., & Baayen, H. (2004). Putting the bits together: An information theoretical perspective on morphological processing. Cognition, 94(1), 1–18. Murphy, G. (2002). The big book of concepts. Cambridge, MA: The MIT Press. OED2: Oxford English Dictionary (2009), 2nd edition, Version 4.0, Simpson, J. (Ed.). Oxford: Oxford University Press. Panther, K.-U., & Radden, G. (2011). Introduction: Reflections on motivation revisited. In K.-U. Panther & G. Radden (Eds.) Motivation in grammar and the lexicon [Human cognitive processing 27] (pp. 1–28). Amsterdam, Philadelphia: John Benjamins Publishing House. Plag, I. (1999). Morphological productivity: Structural constraints in English derivation. Berlin: Mouton de Gruyter. Plank, F. (1994). Inflection and derivation. In R. E. Asher (Ed.), The encyclopedia of language and linguistics (Vol. 3, pp. 1671–1678). Oxford: Pergamon.

Paradigmaticity in Compounding

47

Pounder, A. (2000). Processes and paradigms in word-formation morphology. Berlin: Mouton de Gruyter. Radden, G., & Dirven, R. (2007). Cognitive English grammar [Cognitive linguistics in practice 2]. Amsterdam, Philadelphia: John Benjamins Publishing House. Radden, G., & Panther, K.-U. (2004). Introduction: Reflections on motivation. In G. Radden & K.-U. Panther (Eds.) Studies in Linguistic Motivation [Cognitive Linguistics Research 28] (pp. 1–46). Berlin: Mouton de Gruyter. Roché, M., & Plénat, M. (2014). Le jeu des contraintes dans la sélection du thème présuffixal. In F. Neveu, P. Blumenthal, L. Hriba, A. Gerstenberg, J. Meinschaefer, & S. Prévost (Eds.), Proceedings of the 4e Congrès Mondial de Linguistique Française (CMLF 2014) (Vol. 8, pp. 1863–1878). Paris: Institute de Linguistique Française. DOI: 10.1051/shsconf/20140801143. Sag, I., Boas, H., & Kay, P. (2012). Introducing sign-based construction grammar. In H. Boas & I. Sag (Eds.), Sign-based construction grammar (pp. 1–29). Stanford, CA: CSLI Publications. Saussure, F. (1968). Cours de linguistique générale (Critical ed. by R. Engler, Vol. 1). Wiesbaden: Harrassowitz. Spencer, A. (2017). Participles: inflectional paradigms, derivational paradigms or something else? Paper delivered at ParadigMo 2017: First Workshop on Paradigmatic Word Formation Modeling, Toulouse, 19–20 June 2017. Štekauer, P. (2014). Derivational paradigms. In R. Lieber & P. Štekauer (Eds.), The Oxford handbook of derivational morphology (pp. 354–369). Oxford: Oxford University Press. Stump, G. (2001). Inflectional Morphology: A Theory of Paradigm Structure. Cambridge: Cambridge University Press. Stump, G. (2010). The derivation of compound ordinal numerals: Implications for morphological theory. Word Structure 3(2), 205–233. Traugott, E., & Trousdale, G. (2013). Constructionalization and constructional changes. Oxford: Oxford University Press. Urban dictionary. Available at http://www.urbandictionary.com/ [Accessed on a regular basis August, 2016–August, 2018]. van Marle, J. (1985). On the paradigmatic dimension of morphological creativity. Dordrecht: Foris. Vogel, P. (2000). Grammaticalization and part of speech systems. In P. Vogel & B. Comrie (Eds.), Approaches to the typology of word classes (pp. 259–284). New York: Mouton de Gruyter. Wald, B., & Besserman, L. (2002). The emergence of the verb-verb compound in twentieth century English and twentieth century linguistics. In D. Minkova & R. Stockwell (Eds.), Studies in the history of the English language: a millennial perspective (pp. 417–447). Berlin: Mouton de Gruyter.

48

Bagasheva

Word spy. Available at http://www.wordspy.com/ [Accessed on a regular basis December, 2016–February, 2018]. Wu, L., & Barsalou, L. (2009). Perceptual simulation in conceptual combination: evidence from property generation. Acta Psychologica, 132(2), 173–189. Yeh, W., & Barsalou, L. (2006). The situated nature of concepts. American Journal of Psychology, 119, 349–384.

Chapter 3

Characterizing Derivational Paradigms Bernard Fradin 1

Introduction

At the beginning of The grammar of words, Booij (2005, p. 8) characterizes the term paradigm very broadly as “a set of linguistic elements with a common property”, and goes on saying that “when we speak about morphology as the study of the systematic form-meaning correspondences between the words of a language, we take a paradigmatic perspective, since we take properties of classes of words as the starting point of morphological analysis”. Even though I totally agree with Booij’s viewpoint, my characterization of paradigm will be slightly more restrictive insofar as I shall speak of network instead of set to the extent that the elements of a paradigm are related with one another, and networks can be defined in a way that allows us to make predictions. Actually, what for should we pay attention to paradigms if they would not help us to better predict the behavior of word classes? Such a predictive power is indeed expected to follow from an appropriate description of paradigms. The first goal of this article is to recall the properties paradigms should abide in order to be predictive tools (§2). Usually tools have a domain of relevance within which their use is optimal. It is crucial to know what empirical conditions have to be satisfied for the notion of derivational paradigm to be efficiently employed and all the more so as the flimsiness of this notion is frequently pointed out. Otherwise it might be criticized or discarded for inappropriate reasons. Delimitating the domain of relevance of the notion of derivational paradigm is the second issue addressed here (§3). Taking advantage of the properties that have been regularly associated with derivational paradigms in the literature (Bauer, 1997; Štekauer, 2014), the conceptual organization of derivational paradigms is the next issue that will be addressed (§4). Since it is important to root discussions in empirical data, I will discuss recent proposals that argue for distinguishing several types of derivational paradigms (Roché, 2017a, b). Although it focuses on French, the discussion’s outcome is relevant for other languages as well, where similar data can be observed (§5). A conclusion will sum up the main ideas of the article and raise a few questions for the future (§6).

© Koninklijke Brill NV, Leiden, 2020 | doi:10.1163/9789004433410_004

50 2

Fradin

Properties of Paradigms

Paradigms will be viewed here as networks of morphologically related words. Morphological relatedness is defined as follows by Bonami and Strnadová (2018): Two words w and w’ are morphologically related if and only if there exists a nontrivial content relation Rc relating the two words […] and there exists a non-trivial form relation Rf  relating the two words […] and there are multiple pairs of words related by that same pairing of content relation and form relation. Although differences between derivational and inflectional paradigms is a much emphasized topic, it seems fruitful to focus on what these notions have in common instead on what distinguishes them. Bonami and Strnadová (2018) show that derivational paradigms share all the properties exhibited by inflectional paradigms: suppletion, defectiveness, heteroclisis, overabundance, etc., and that a quantitative approach to predictiveness works for both types of paradigms. This leads them to adopt an extended view of this notion, according to which a paradigm is a collection of morphological families structured by the same system of oppositions of content. More formally, “[a] morphological family is a tuple F = (wi ,…, wn) of words such as any member wi of the family is morphologically related to any other member wj” (Bonami & Strnadová, 2018). The content may be morphosyntactic or morphosemantic. In the first case, we have an inflectional paradigm, in the second a derivational paradigm. If we adopt an implicative view of morphology (Blevins, 2006, 2016), paradigms are networks constituted of implications between the pairs of word forms which structure these networks. The case of inflection is illustrated by a fragment of the Russian nominal declension (Annex, Figure 3.2), where the word forms corresponding to lexemes izba ‘izba’, lico ‘face’, and les ‘forest’ are contrasted in the dimensions of CASE and NUMBER.1 The forms in question are correlated by content relations such as: ‘x is the acc.sg word form corresponding to the nom.pl word form y’, ‘x is the acc.pl word form corresponding to the acc.sg word form y’, etc., that structure the paradigm. The case of derivation is illustrated by families of words correlated by relations such as ‘x denotes the set of 1  The morphological families constituting the paradigm in question are partial (and so is the paradigm): “A morphological family F is complete if there exists no larger morphological family that contains all members of F. A morphological family is partial if it is not complete” (Bonami & Strnadová, 2018; emphasis in the original).

Characterizing Derivational Paradigms

51

individuals acting as agent in events like those denoted by verb y’, ‘x denotes a set of individuals that have the property to be potential patient in the event denoted by verb y’, etc. These relations, which are captured by the tags Agent_N, Action_N, ADJ in Figure 3.3 of the Annex, are typical of the content relations associated with morphological derivational patterns. In the implicative perspective, establishing what the form in cell Ck can be, knowing what the form in cell Ci is (conditional entropy of cell Ck given Ci is known) becomes one of the main goals of studies.2 Paradigms conceived of in this way have a predictive potential. They allow to express morphological regularities in a non-categorical way, through the application of the same statistical tools to both types of paradigm (quantitative approach. See Bonami and Strnadová (2018) for an illustration). Paradigms (networks) presented in Figure 3.2 and Figure 3.3 in the Annex show that alignment is a crucial property of paradigms:3 each word of a given morphological family must have a correlate in the other morphological families. Without alignment, no prediction can be made. Bonami and Strnadová (2018) define alignment as follows: Given two ordered pairs of morphologically related words (w1, w2) and (w3, w4), we say that the two pairs are aligned if the same content relation holds between them: there is some content relation Rc such that Rc(w1, w2) and Rc(w3, w4). We call Rc the aligning relation. Examples of alignment relations are given in (1). We see that pliable ‘folding’ has the same content relation with its base verb as prédictible ‘predictable’ does, even though the exponent of the relation is different. This exponent variation is observed in languages with distinct inflectional classes, such as Russian (1a). (1) a. RNOM.SG-ACC.SG(izba, izbu), RNOM.SG-ACC.SG(les, les) b. RV-AGT(plier, plieur), RV-AGT(prédire, prédicteur)
 c. RNZN-A(pliage, pliable), RNZN-A(prédiction, prédictible)

2  Actually, Bonami and Beniamine (2016) show that the implicative entropy, which pays attention to the shape of the form filling a given paradigm’s cell, is a more accurate predictor than conditional entropy. I refer to Bonami and Beniamine (2016) and Bonami and Strnadová (2018) for the definition and application of this concept. 3  The terms paradigm and network are conceptually equivalent here, even though their domain of reference is distinct (linguistics / graph theory). Nodes in networks correspond to cells in paradigms.

52

Fradin

If alignment is crucial, what kind of content relationships may guarantee a sound alignment in derivational paradigms? This is one of the main issues that have to be tackled in the rest of this article. This issue will be addressed in §4, which investigates the organization of paradigms, and continued in §5 where I will discuss the existence of various types of derivational paradigms that have recently been distinguished in the literature. But before undertaking this task, we have to delimit the domain of relevance of the notion of derivational paradigm. This will be done in §3. In the present section, it has been argued that it was justified and fruitful to deal with the derivational paradigms examined so far with the same tools as the ones employed with inflectional paradigms. This does not imply, however, that these two kinds of paradigms are equivalent. Some of their properties show that they have to be kept distinct. The first one is the fact that the elements are more tightly linked in inflectional paradigms than in derivational paradigms. This property has to do with ‘morphological proximity’ (Hathout, 2011, p. 255), which is a scalar assessment of the morphological relationship defined above: “Two forms jointly sharing semantic and formal properties are morphologically proximate” and this proximity increases in proportion to the number of properties and their specificity. Think of the difference between a feature like CASE, which is specific to nouns and whose values may be quite numerous4 and primarily motivated by distinctions internal to the language itself (Sapir, 1921), e.g., ACC, and content ‘agent’ which may be associated, without any formal correlate, to many units in addition to derived nominals in -eur, for instance, agentive verbs or subject oriented adverbs e.g., voluntarily (Geuder, 2003).5 The second property is the fact that, in inflectional paradigms, the content relations link elements that both belong to the word form type; in derivational paradigms instead, the elements in question are of the lexeme type (Matthews, 1974). While the first property is generally satisfied, it may happen that the second be not, which has some consequence for the applicability of the notion of paradigm, as we will see now.

4  These values (e.g., ACC, DAT, INS, LOC, ALL, etc.) should not be confused with the semantic function they may have in actual contexts. 5  This echoes Bybee’s remarks according to which “the results of derivational processes usually have a few lexical counterparts, that is, lexical items in which the same combination of meanings are expressed monomorphologically [= lexematically]. For example, sad for unhappy, pilot for flyer” (Bybee, 1985, p. 83). It must be noted that in Bybee’s text specificity denotes the range of applicability of morphological patterns: “The greater specificity in derivational meaning restricts the applicability of derivational processes” (Bybee, 1985, p. 86).

Characterizing Derivational Paradigms

3

53

The Domain of Relevance of Derivational Paradigms

The distinction between content words (units) and grammatical words (units) has to be drawn for every language. Content words introduce discourse referents under the form of individuals, properties or eventualities and are used to make claims about the world. Grammatical words provide us with information about the grammatical obligatory coding: the markers expressing syntactic functions, speech act properties, temporal deixis, quantification, etc., and their interaction (Mel’čuk, 1993, ch. 5).6 Grammatical words “are […] realizations of grammatical category features—pure combinations […] of syntactic and grammatical category features with no other semantics” (Zwicky, 1990, p. 230). In isolating languages, both types of words are realized as independent units. Zwicky (1990) hypothesizes that inflection is the morphologization of grammatical information, “the realization of grammatical category features on individual lexemes”. Morphologization means first, that some words may convey grammatical information and thereby change their shape in function of the nature of this information; and second, that this change occurs in a systematic way: the information is organized along dimensions associated with a range of values (e.g., nb: {sg, pl}), and these dimensions are relevant for all units of a given lexical category.7 Nominal declensions are a typical illustration of this case (cf. Figure 3.2). Once one assumes Zwicky’s conjecture, the domains of inflectional and derivational morphology become clearly distinct from a conceptual point of view, and this distinction is difficult to deny. The common view is that, conceptually, derivational morphology is used to build new items, which will enrich the lexicon in case they become lexicalized. Inflectional morphology, on the other hand, is syntax-dependent (at least for contextual inflection) and makes explicit the phonetic correlates of morphosyntactic content. However, the way paradigms are organized qua semiotic systems does not reflect this clear-cut conceptual distinction. We already mentioned this lack of parallelism for fusional languages (Bauer, 1997; Bonami & Strnadová, 2018). At a higher level, various situations can be observed. Languages without inflectional morphology (isolating languages) may nonetheless have derivational paradigms provided they have derivational patterns. A case in point seems to

6  The proposed distinction intersects with the traditional distinction between lexical vs. grammatical units. As Bickel and Zúñiga (2017, pp. 165–167) underline, the later distinction is far from clear-cut. 7  Carstairs-McCarthy (2010)’s account of the origin of inflection classes complements Zwicky’s view.

54

Fradin

be Vietnamese, if we rely on Panfilov, Baumann, and Haspelmath (2001), from which the following examples are borrowed. (2) Prefixation (N~N) chín trị → nhà chín trị ‘politics’ ‘politician’
 khoa học → nhà khoa học ‘science’ ‘scientist’ (3) Suffixation (N~V) công nghiệp → công nghiệp hóa ‘industry’ ‘industrialize’ tự động → tự động hóa ‘automat’ ‘automatize’ (4) Prefixation (A~V or V~V) bẩn → đánh bẩn ‘dirty’ ‘pollute’ thức → đánh thức ‘wake’ ‘awake’ khổ → làm khổ ‘painful’ ‘torment’ The existence of larger paradigms depends on the existence of families built around units such as chín trị ‘politics’, công nghiệp ‘industry’, tự động ‘automat’, thức ‘wake’, etc. Except the fact that these paradigms are not morphological by construction, they would perfectly meet the alignment condition imposed on derivational paradigms. Languages possessing inflectional morphology may have paradigms similar to those mentioned in Figure 3.2 and Figure 3.3 or different: everything depends on the way these paradigms are organized as systems of signs. I limit myself here to presenting a case found in agglutinative languages. The variety of sign systems depends on the way the sound/meaning relationship is settled in the language in question. Plungian (2001, p. 669) argues that the distinction between agglutinative and fusional languages involves three independent parameters: (a) the way in which inter-morpheme boundaries are handled within a word form; (b) the extent to which non-phonologically conditioned variation of stem and grammatical markers is attested, and (c) the extent to which a symmetry between semantic and formal organization

55

Characterizing Derivational Paradigms Table 3.1

Distinction between agglutinative vs. fusional languages

Type

1

2

3

4

5

6

7

8

(a) clear-cut boundaries (b) one form only (c) 1/1 correspondence

+ + +

− + +

+ − +

+ + −

− − +

− + −

+ − −

− − −

of grammatical markers is observed. The way these parameters combine is given in Table 3.1. Column 1 corresponds to maximally agglutinative languages, and 8 to maximally fusional languages.8 According to Plungian, these parameters result from “the greater autonomy of morphemes in agglutinative languages”, which make “agglutinative affixes […] much closer to roots than non-agglutinative affixes as far as their phonological, morphological, syntactic and grammatical properties are concerned” (Plungian, 2001, p. 674). Rules combining units in agglutinative languages are then more syntactic than morphological, i.e., more like rules that combine words. Several well-known properties of agglutinative languages follow from the syntactic autonomy of their markers. The one that interests us here is the fact that affixes with a semantic content may be adjoined immediately after grammatical markers, as shown in the following examples. (5) Dogon (Plungian, 2001, p. 675) a. dɔnɔ-dɛ sell-prs.hab(3sg) ‘he sells’ dɔnɔ-dɛ-nɛ sell-prs.hab-agt ‘one who sells, seller’ b. wal-a be (Plungian, 2000, p. 184) cultivate-ant be:pst(3sg) ‘(s)he had cultivated’ wal-a be-nɛ cultivate-ant be:pst(3sg)-agt ‘an ex-farmer’

8  It should be kept in mind that a language may be agglutinative/fusional in one domain and not (or less so) in another.

56

Fradin

(6) Rikbaktsa (Silva, 2011, pp. 310–311) kɑ-kɔrɔ-t͡ʃi=ɽɑ 1poss.sg-burn-pl=f.pl ‘those[female] who burned me’ kɑ-t͡ʃĩhĩ=t͡ʃɑ=bo ikt͡ʃɑ ∅-mi 3poss.sg-smoke_dry=nf=all ‘see’ 1sg-npass+aux ‘I go and see my buccaneer’ In Rikbaktsa, the clitics t͡ʃɑ ‘non-feminine’ and ɽɑ ‘feminine plural’ change the sentence into a nominal entity. In a parallel way, the Dogon agentive marker nɛ, dɛ transforms a sentence into an agent noun. Languages of this type (tend to) have no derivational morphology. Any sentence headed by an action verb can potentially be transformed into a definite description denoting the agent (patient, instrument, etc.) of the action: agent formation is not handled by a derivational process. In Rikbaktsa, the markers that trigger the relevant interpretation convey a meaning, which is not clearly morphosemantic. Could units the meaning of which is ‘seller’ in Dogon or ‘buccaneer’ in Rikbaktsa appear in a network similar to the derivational networks/paradigms mentioned in §2? The answer seems to be negative, if we stick to the definitions of morphological family and alignment given above. The problem is that one element of the relationship is not a word but a phrase or a sentence and, more generally, that this relationship is not between lexemes (the second property evoked in §2 is not respected). This case would not be in the realm of the notion of paradigm defined in §2. Nevertheless, a well-formed network based on alignment relations involving prs.hab(3sg) verbs and the forms denoting agents can be conceived of in the case of Dogon. Transposition offers another, very common, example of markers completely encapsulated within an inflectional paradigm that create lexemes endowed with a distinct syntactic category, e.g., participles, masdars (nominal non-finite form of a verb, Mel’čuk (1994, pp. 215–219)), etc. I refer to Haspelmath (1996), who drew the attention to this issue, and Spencer (1999, 2013, 2018), who discusses transposition at length in many works. The more general problem these forms raise “is that an inflected form unexpectedly participates in derivational processes” (Blevins, 2001, p. 215).9

9  Blevins shows that the way the common view distinguishes between inflection and derivation is erroneous, but discussing the changes he proposes, however important, would lead us too far afield.

57

Characterizing Derivational Paradigms

4

The Organization of Derivational Paradigms

If we agree that the semantic content between pairs of words constitutes the foundation of derivational paradigms (Štekauer, 2014), the question that comes to mind is: what is the nature of this semantic content? This section will try to give an answer to it. 4.1 Nature of the Semantic Content Among the derivational paradigms that have been proposed in the literature, very few are rooted in purely lexical meaning (‘meaning-rooted’ for short). Such meaning implies that the relationship involves no information that would be syntactically relevant in the language in question. But what counts as ‘syntactically relevant’ is not easy to define and may vary from language to language; it is wiser to consider this notion as a cline. At first glance, paradigm schemes (7)–(9) seem to be good illustrations of what meaning-rooted paradigms are, even though one could quibble about the fact that the concept Agent may happen to be relevant for syntax while concept ‘plant_N’ may not (more on this below). (7) fruit_N, pomme, ‘apple,

plant_N, pommier, apple-tree,

place_N pommeraie apple orchard’

(8) V, wandel, ‘walk,

male_AGT, female_AGT wandelaar, wandelaarster walker(male), walker(female)’

(Boyé & Schalchli, 2017)

(Booij, 2002, p. 102)

(9) proper_N, DA, supporter_N/DA, doctrine_N (Roché, 2007) Proudhon, proudhonien, proudhoniste, proudhonisme ‘Proudhon, Proudhonian, Proudhonist, Proudhonism’ The important distinction here seems to be between derivational paradigms that are event-related and those that are not. Event-related paradigms tend to be organized around verbs or events denoting lexemes. This is the case in (8), where the gender dimension is cross-classified with that of agentivity. In (7), on the contrary, the paradigm scheme involves concepts that directly stem from basic properties of fruits, namely the fact that they are the natural product of plants (here apple-trees) and that they later grow in some specific places (an apple-orchard). This content can be captured through inference patterns

58

Fradin

such as those in (10), where LOC is the locative function which relates a figure and a ground specified by loc(z).10 (10) a. fruit(x) ⇒ ∃y[plant(y) ∧ produce(y,x,e)] b. plant(x) ⇒ ∃z[grow(x,e) ∧ LOC(e, loc(z))]

On a par with (7), paradigm scheme (9) is not verb related, but unlike (7) the relations it involves are artifactual instead of being natural. Once we know that Proudhon was a philosopher, the fact that he had a doctrine and followers that support him and his doctrine is either expected or at least in the nature of things.11 In a way parallel to (10), one might formulate inferences such as (11), the difference between the two being that while the truth of (10) is categorical, that of (11) is generally less than 100%. (11) a. philosopher(x) ⇒ ∃y[doctrine(y) ∧ teach(x,y,e)] b. philosopher(x) ⇒ ∃y[follow(y,x,e) ∧ people(y)]

On the basis of the above examples, it seems fair to claim that the more the element denoted by the pivot item of the paradigm (usually the origin) exhibits natural kind properties, the less the paradigm is event-related. As shown in (8), (10), and (12) below, derivational paradigms often introduce concepts such as agent, instrument, action, etc. that cannot be characterized independently of the event they are part of. Their definition is holistic and involves parameters such as initial causation, control, performing actor, and so on. These parameters are ultimately settled through linguistic tests and criteria, which have been worked out for years in many studies (Cruse, 1973; Delancey, 1985; Dowty, 1991; Van Valin & LaPolla, 1997; Koenig, Mauner, Bienvenue, & Conklin, 2008, just to mention a few). The grammatical incidence of each criterion vary from language to language (Mithun, 1991) and this is why it is more appropriate to consider the concepts in question as partly grammatically rooted instead of simply lexically rooted.

10  The value of loc can be any of the locative relations denoting a stative ground, e.g., inessive, adessive, superessive, subessive, etc. (cf. Mel’čuk, 1994, pp. 280–282; Kracht, 2002). Spatial relationships are captured more accurately and efficiently by vector semantics (Gärdenfors, 2000, 2014; Zwarts & Winter, 2000). But introducing that framework is beyond the scope of this article. 11  Scheme (9) includes denominal adjectives (DA), which have the same form as the noun. Forms ending in -iste have an axiological value and the N is basic, whereas for those in -ien, the adjective is basic.

Characterizing Derivational Paradigms

(12) V, place_N, action_N, agt_N, ins_N semer, —, semailles, semeur, semoir ‘sow, —, sowing, sower, seed drill’

59 (Roché, 2017a)

The first partial conclusion we can draw is that no pure semantic content can be observed in event-related paradigms. The semantic content the latter involves is established on the basis of criteria, which are generally relevant to other domains of the grammar (e.g., pronoun selection, activity hierarchy, etc.). 4.2 Within vs. Out of Human Reach A closer inspection of event-related paradigms reveals that most of them are human-centered: their elements denote activities that are typically those that humans control, at least partly, or that they benefit from. This points towards the ‘force dynamic approach’ (Croft, 1991; Gärdenfors, 2000; Talmy, 2000), where an agent is the initiator and/or effector of the action that affects a patient, with the eventual help of an instrument. The force dynamic approach involves a ‘causal chaining’ and assumes that, cross-linguistically, typical verbs “reflect segment of causal structure, not any other kind of structure” (Croft, 1991, p. 191). This causal chaining seems to be supported by psycholinguistic experiments (Wolff, 2003, 2007; Gärdenfors, 2014).12 Indeed, most derivational paradigms proposed in the literature involve either an agentive event, e.g., ‘weld/welding’ or a functional artifact/substance e.g., ‘spade’/‘paint’ that requires an agent to be properly used. The discussion in §5 will shed more light on this point. It is important to realize that human-centeredness has to be dissociated from force dynamics. Such a dissociation happens when the interaction with humans is not expressed through a specific verbal predicate, as is (generally) the case with causal chaining. Let us consider the case of animals. They can be grouped into three categories: (a) domestic animals, with which humans have mostly been having strong interactions for a long time; (b) wild animals with which humans are in predatory interactions (hunting, fishing); (c) wild animals with which the human interaction is reduced or nonexistent. What we observe, is that the more man interacts with an animal species, the finergrained the distinction of various sub-types of individuals within the species is. These sub-types depend on distinctions useful to humans in their trade with the animals of the species in question. For instance, pets are usually sorted in 12  In their thorough survey of theories dealing with argument realization, Levin and Rappaport Hovav (2005) argue that the force dynamic approach better accounts for the data than others.

60

Fradin

Table 3.2 Partial classification of pets and farm animals in French

DOG CAT GOAT HORSE COW

Species

Male

Female

Young

chien chat chèvre cheval vache

chien chat bouc cheval taureau

chienne chatte chèvre jument vache

chiot chaton chevreau poulain veau

function of their sex and age, and in livestock farming more distinctions operate: sex, age, reproductive capacity, finality, etc. (cf. Boyé & Schalchli, 2017). Each of these sub-types could be classified in the way illustrated in Table 3.2. A classification like this one constitutes a paradigm since the same content relations link the elements of each family, e.g., ‘x is the Male item corresponding to the Female item y’, ‘x is the Young item corresponding to the Species item y’, etc. However, it does not constitute a morphological paradigm for the following reasons: (i) The elements in the cells either are independent simple lexemes, e.g., cheval ‘horse’/jument ‘mare’, or their formal link is not an exponent of a dedicated meaning, as the content relation requires and in contradistinction to what we observed in (7)–(9). It can be a diminutive exponent, e.g., chat-on, chevr-eau, or take the shape of the feminine marker used in some conversions, e.g., chienn-e, chatt-e (Roché, 2005).13 (ii) The concepts which structure the paradigm of Table 3.1 do not follow from basic properties of the animals listed in the first column: it is impossible to postulate inferences on the model of (10), since they would be either meaningless, e.g., Species(x) ⇒ Young(x) or uninformative, e.g., Species(x) ⇒ male(x) ∨ female(x). These concepts are not abstract nodes in a larger scheme, as is the case with the force dynamic approach. They are rather imposed by the organization of the domain of activity (livestock farming, hunting, etc.), such as man designed it. This is all the more true as classifications become more specific whenever the activity in question requires more purpose-oriented distinctions. The classification given in Table 3.2, and other more sophisticated ones, rather fall in the realm of semiotic analyses such as those that were proposed in the 13  In Old French, -eau was a diminutive suffix used, among others, to form names of young animals, e.g., lionc-eau lion-dim, fauconn-eau hawk-dim. This extension is common for diminutives (Jurafsky, 1996).

Characterizing Derivational Paradigms

61

heydays of structural semantics (Pottier, 1963; Lehrer, 1974; Rastier, 1987). These approaches are better suited to account for them.14 As a rule, paradigms like those of Table 3.2 have little to do with derivational morphology, since the elements they involve are mostly, and sometimes exclusively, simple lexemes. The argumentation that has just been developed about domestic animals extends to wild animals interacting with humans. Other animals, and natural kinds in general, are domains on which the man’s grip is less strong or even inexistent. This offers the opportunity to reverse the perspective and examine which morphological possibilities exist for nouns denoting these kinds without worrying about human-centeredness. One striking fact about nouns denoting a (wild) natural kind is the reduced number of derived lexemes in comparison with the amount we observe for morphological families based on verbs or artifacts. For instance, whereas about 80 lexemes can be derived from Russian zvenet’ ‘to ring’ / zvon ‘ringing, chime’,15 only 30 can be from zver’ ‘beast’ (of which several are diminutives) (Tixonov, 1985). The same contrast is repeatedly observed for other families e.g., suslik ‘gopher’ (3), caplja ‘heron’ (1), žuravl’ ‘crane’ (10 of which 4 are diminutives), volk ‘wolf’ (4), voron ‘raven’ (9 of which 3 evaluatives), vs. bežat’ ‘to run’ (138; 153 with compounds), maslo ‘butter, oil’ (100; 148 with compounds). Even in French, derivational families based on artifacts are larger e.g., balle2 ‘bale’ (16) vs. renard ‘fox’ (4). The contrasts put to light here are not numerous enough to be conclusive, but they suggest a working hypothesis that deserves further investigation. A cursory survey of the lexemes that can be derived from the nouns denoting a natural kind shows that the range of possible types of derived V and A seems to be fixed by the grammar of the language in question. For example, in Russian, nouns of this type are generally correlated with (a) the so-called possessive adjective (which modifies Ns denoting inalienable property), e.g., lisa ‘fox’ → lisij e.g., lisij xvost ‘fox tail’, suslik ‘gopher’ → susličij; (b) and a (set of) relational adjective(s), which have a wider range of meanings, e.g., suslik → suslikovij, zver’ ‘beast’ → zverskij ‘brutal’, medved’ ‘bear’ → medvežatyj ‘strong (as a bear)’, volk ‘wolf’ → volčinyj ‘intended to hunt wolf’. French has no possessive adjectives but has relational adjectives. The same holds for English, but with far fewer relational adjectives, since compounds fulfill their role. As for relational adjectives, in all examined languages their meaning involves a property, real or invented, associated with the referent of the base noun 14  See for instance Bonan Garrigues and Elie (1971), who analyze the various denominations of riding animals in Old French. All distinctions but one that the authors postulate are purpose-oriented and specific to the semantic field in question. 15  The total figure of complex lexemes formed on zvon/zvenet’ reaches 92, if compounds are included (Tixonov, 1985, pp. 364–365).

62

Fradin

(shape, color, behavior, etc. cf. Fradin (2017b)), cf. Russian zverskij, medvežatyj; French chenille ‘caterpillar’ → chenillé ‘whose aspect resembles a caterpillar’, héron ‘heron’ → héronné ‘that looks like a heron’. As for verbs, both Russian and French build verbs whose meaning expresses a property associated with the referent of the base noun, e.g., Russian zveret’ (ipf) / ozveret’ (pf) ‘behave in a bestial way’ (← zver’ ‘beast’), lisit’ ‘flatter’ (← lisa ‘fox’), zmeit’sja ‘meander’ (← zmeja ‘snake’); French se lézarder ‘to crack’ (← lézard ‘lizard’), mur lézardé ‘crannied wall’, singer ‘to fake, to feign’ (← singe ‘monkey’), moucheter ‘speckle (a fabric)’ (← mouche ‘fly’), serpenter ‘meander’ (← serpent ‘snake’). But unlike Russian, French can easily form verbs, generally based on the young animal’s name, to express the meaning ‘give birth (to a young of kind X)’, e.g., chatonner/ chaton ‘kitten’, agneler/agneau ‘lamb’, faonner/faon ‘fawn’, vêler/veau ‘veal’. The (un)availability of such derivational patterns is entrenched in the grammar of the languages in question, which means that the meaning they convey can be impossible to express derivationally in other languages. This is the first point. But as we saw above, the situation is different with verbs derived from nouns denoting animal species, since the nature of the content relation attached to each derivational pattern seems to depend on the properties attributed to the base noun’s referent (as in (7a)). This contrasts with derived lexemes whose meaning implies that the activity they describe is human-centered, as is the case with the French derivational pattern ↔ , e.g., loup ‘wolf’ ~ louvetier ‘wolf hunter’, loutre ‘otter’ ~ loutrier ‘otter hunter’, etc. This is the second point. The third and most important one is that the non-human-centered morphological families derived from natural kind nouns cannot easily be stacked to form paradigms, because no alignment relations hold between the elements that constitute them. For instance, each pair of the following collection exhibits a distinct content relation and the first two seem to be specific to the pairs in question: serpent/serpenter ‘to meander’, mouche/moucheter ‘to speckle’, agneau/agneler ‘to lamb’, héron/héronner ‘to hunt heron’, singe/singer ‘to fake’.16 More work is needed to see whether generalizations can be established. 4.3 Three Types of Content Relations Two sources that feed content relations have been identified up to now: eventrooted schemes and properties associated with the entity denoted by the base noun. Developing an idea sketched above, I would like to suggest that 16  The situation looks like what we have with the series of derived English verbs in -ize (Plag, 1998), e.g., hospitalize, fossilize, satirize, ionize, etc. (Štekauer, 2014, p. 358), where the semantic relationship between the base and the derived V varies. However, for natural kinds the meanings are less predictable and the series may be very small.

63

Characterizing Derivational Paradigms

derivational paradigms imposed by the grammar constitute a third source of content relations. Some well-entrenched morphological patterns that correlate various types of lexemes offer correlations available in the language in question. Examples are given in (13) and (14) for Russian and English, respectively. Four properties distinguish these patterns: their content is rather abstract and to a large extent instructional; they apply across the board; they are characteristic “of the word-formation system of a language as a whole” (Štekauer, 2014, p. 363); their availability, in the sense of Bauer (2001), is great. (13) a. A~Abstract_N b. A~ADV c. N~DA d. Vipf~Vpf e. Vpf~Vipf

zvonkij~zvonkost’ zvonkij~zvonko zver’~zverskij zvonit’~pozvonit’ vyzvonit’~vyzvanivat’

(14) a. A~ADV b. V~Ving

apt~aptly to match~matching

‘sonorous, sonority’ ‘sonorous, sonorously’ ‘beast, brutal’ ‘ring up’ ‘give a call’

Actually, it would be more appropriate to conceive of the above patterns as meta-patterns inasmuch as most of them subsume particular, properly defined, derivational patterns. These meta-patterns are so central in the morphology of the language in question that speakers cannot avoid using them, much in the way compounding is central in German. In fusional languages, the domain covered by inflectional paradigms is coextensive with lexical categories: conjugational paradigms concern verbs and differ from nominal declensions, etc. This is why inflectional paradigms are category-rooted. Derivational paradigms are semantically rooted instead. This means first, that they are semantically motivated as the preceding section tried to show; and second, that derivational paradigms, obviously, do not cover all items of a given category: derivational paradigms are like islands, even though overlaps between paradigms are possible (see §5). A picture of the situation is given in Figure 3.1. The first type of paradigm corresponds to event-related networks, such as action and activity networks illustrated in (15) that will be discussed in detail in §5. Three examples of nonevent-related networks are given in (16): the ethnonym network, the monophytic area network,17 the status network, and the gender network (in Czech). Finally, grammatically entrenched networks are illustrated by (13) and (14).

17  This network was dubbed “nom de lieu monophyte”, i.e., name of place where one type of plant grows by Corbin (1987).

64

Fradin A~adv

Vpf~Vipf Activity nwk Action nwk N~da Vipf~Vpf

Status nwk

Ethnonym nwk A~abstract_N Monophyte area nwk N(m)~N(f)

Figure 3.1 Paradigms’ types

(15) Action network: dorer, dorure, doreur ‘gild, gilding, gilder’ Activity network: chapeau, chapellerie, chapelier ‘hat, hat making, hatter’ (16) Ethnonym network: Malaisie, Malaisien, malais ‘Malaysia, Malaysian, Malay’ Monophytic area network: houx, houssaie ‘holly, hollywood’ Status network: vizir, vizirat ‘vizier, status of vizier’ Gender network: biolog, bioložka (ces) ‘biologist(m), biologist(f)’ 5

Discussion of Existing Proposals

I will mainly discuss the proposal made by Roché (2017a, 2017b), according to which two varieties of paradigms are worth distinguishing: the ‘action network’ (réseau action) and the ‘activity network’ (réseau activité). These varieties will be discussed in turn in §5.1 and §5.2. It must be clear that this discussion is undertaken less because I fully agree with Roché than because his proposal is rich enough to fuel an exchange that could contribute to clarify what derivational paradigms are. I will confine myself to event-related paradigms and leave aside a variety of other paradigms discussed in the mentioned articles. Roché remains highly allusive about the way to distinguish the various nodes and to design the networks’ structure. These issues constitute the core of the present section. 5.1 The Conceptual Foundations of the Action Network Roché (2017a) contends that the verb is the pivot in the action network, both from the formal and semantic point of view. The schema he proposes to capture this idea is reproduced in (17) and some examples are provided in (18). In

65

Characterizing Derivational Paradigms

(18) the sequence of words begins with the V and then follows the second line of schema (17); a dash indicates a gap in the morphological family. Action network, version 1 (17)

place_N

action_N

Verb

(18) a. laver lavoir lavage ‘to wash wash-house washing

agent_N

instrument_N

laveur laveuse washer washing machine’

b. tondre — ‘to shear —

tonte tondeur tondeuse shearing shearer shears’

c. biner ‘to hoe

binage hoeing

— —

d. souder — ‘weld —

bineur hoer

binette hoe’

soudage soudeur — welding welder —’

As illustrated in (18), many morphological families lack an item for the ‘place’ or ‘instrument’ nodes. On the other hand, others include lexemes that instantiate a ‘result’ or ‘means’ nodes. This seems to support introducing two additional nodes in the action network, as in (19). But examples in (20) show that the more the network includes nodes, the less there will be morphological families whose members will fit into all these nodes. Action network, version 2 (19)

place_N

action_N

agt_N

Verb

ins_N

(20) a. peindre — peinture2 peintre — ‘to paint — painting1 painter —

mns_N

rslt_N

peinture1 peinture3 paint painting2’

b. riveter — rivetage riveteur riveteuse ‘to rivet — riveting riveter riveting machine

rivet rivet

c. paver — pavage ‘to pave — paving

pavé pavement paver2 pavement’

paveur — paver1 —

— —’

66

Fradin

Is there a principled way to fix the number of nodes of the action network and their nature? This is the main question we have to answer now. Notice that what is at stake is the abstract structure of a sub-type of derivational paradigms, what Carstairs (1987) called Paradigm 1.18 Paradigm 1 conditions the predictive capacity of paradigms, since it determines which slots are available for a given paradigm (Štekauer, 2014, p. 361). Insofar as the paradigms in question are event-related, the idea is to take advantage of the argument structure of the pivot verb to establish the nodes that belong to the network. If we agree that the pivot V denotes an event that involves a causal chaining, it follows that the V has an agent, a patient and, potentially, an instrument argument. Leaving aside the place_N for the moment, we see that the nodes constituting network (17) correspond to the possible arguments of the verb.19 (21) a. λx.∃y∃e[V(e) ∧ AGT(e,x) ∧ PAT(e,y)] agent b. λy.∃x∃e[V(e) ∧ AGT(e,x) ∧ PAT(e,y)] patient c. λz.∃x∃e1e2[V(e1) ∧ AGT(e1,x) ∧ use(e2) ∧ AGT(e2,x) ∧ PAT(e2,z) ∧ cause(e2,e1)] instrument d. ∩λe.∃x[V(e) ∧ AGT(e,x) …] action

In an event e involving both an agent and a patient, (21a) and (21b) respectively state that agent is argument x and patient argument y; we can infer that e corresponds to an eventive event (not a state), since it includes an agentive argument. The criteria for agenthood and patienthood have been much debated in the past, but most people nowadays agree on the properties that characterize them: sentience, control, effectedness, initiator, autonomous existence, for agent; non-sentience, affectedness, incrementality, non-autonomous existence, for patient (Dowty, 1991; Mithun, 1991; Van Valin & LaPolla, 1997; Ackerman & Moore, 2001; Fradin, 2005). The semantic role assigned to verbal arguments depends on the verb that heads the construction they occur in (Davis, 2001; Beavers, 2008, 2011, among others). The concepts of instrument and place mentioned in network (15) are more intricate. According to Koenig et al. (2008, pp. 180–181), one can ascertain that argument z is an instrument only if inferences (22) prove to be valid, for a given

18   Carstairs-McCarthy’s complementary notion Paradigm 2 denotes the set of inflected forms (word forms) that occupy the cells (nodes) of Paradigm 1. 19  Argument structures given in (19), where e is the event argument, are couched in a neoDavidsonian formalism (Carlson, 1998; Landman, 2000). Nothing crucial hinges on this choice.

Characterizing Derivational Paradigms

67

verb V. This test allows us to infer (23a), and thereby (23b–c), since ‘pen’ entails ‘writing instrument’. (22) a. XVY ⇒ XVY with Z b. XVY with Z ↛ X and ZVY (comitative reading excluded) c. XV-ed Y ⇒ X used Z to Vinf (23) a. Jane wrote a letter

⇒ Jane wrote a letter with a . b. Jane wrote a letter with a pen ↛ Jane and a pen wrote a letter.20 c. Jane wrote a letter ⇒ Jane used a pen to write a letter.

Since typical instruments are objects, Fradin and Winterstein (2012) introduce a condition (24) to set instruments apart from means. The referent of Ns governed by with in (25) will therefore be identified as means and not as instrument. (24) Reusability constraint: An instrument is an object that must exist as a separate entity before and after it has been used (as an instrument). (25) They built the wall with bricks. The cook flavored the sauce with oregano. On the contrary, microscope in (26a) is arguably an instrument (examples (26) and (27) come from Davis (2001, p. 143)). As such, it cannot occur in subject position (cf. (26b)) because foregrounded NPs entail control, but instruments are by definition deprived of control (see Schlesinger 1989’s Deliberation Condition). However, sophisticated instruments, which can function by themselves, do things humans cannot do and can even be endowed with control capacities, may occur in subject position, as illustrated in (27). Booij (1986) considers them ‘quasi-agents’. (26) a. The researchers examined the specimen with a microscope. b. *A microscope examined the specimen. (27) a. The researchers detected the earthquake with a seismograph. b. The seismograph detected the earthquake. 20  The inference works for comitatives (a) Jane wrote the letter with Bob ⇒ Jane and Bob wrote the letter.

68

Fradin

In (27), the detector is the seismograph: without it, the researchers cannot detect anything (but they interpret the measurement). In other words, the seismograph is the performer. The situation is the reverse in (26): a microscope is unable to examine anything by itself (it cannot be the performer); the examiners are unquestionably the researchers. At this point, it is important to see how means arguments enter the picture. A verbal argument is a means when its referent, by its mere existence or functioning, allows the event denoted by the verb to reach its term (telic event) or to continue its activity (unbound event). This is the case in (25a), since without the bricks the construction stops and without oregano the sauce lacks its flavor. But this is also the case in (28a), where a countable N is embedded in the PP[with]: since it is a heating device, the mere functioning of the stove keeps the room warm. In contradistinction to instruments, means of this type may occur as subject, as (28b) shows (cf. Fradin, 2012). This is linked with the fact that, in (28a), Bob is in no way the source of heat (even though he can control it, he is not the performer). (28) a. Bob heats his room with a wood-burning stove. b. A wood-burning stove heats his room. Coming back to (21), we see that formula (21c) identifies verbal argument z as an instrument, provided that the agent of an event e1 uses z (as a patient) and that this using event e2 causes event e1 to be achieved or to go on.21 As for the node ‘place’ appearing in the activity network, two distinctions have to be made. The first one is between verbs which imply a spatial relationship and those which do not. The second has to do with the argument that instantiates the figure in the spatial relationship. A sample of English and French verbs that imply a spatial relationship and thereby have a built-in locative structure is given in (29). (29) a. to store/entreposer, to enter/entrer, to display/disposer, to shelter/ protéger (de), to hide/cacher b. to stop/s’arrêter, to entwine/s’entortiller (autour de), to lie/reposer Verbs of each type are illustrated in (30a) and (30b) respectively. In (30a), the tea (figure) is located in the cupboard (ground) once the storing event has 21  Alternative formulation: ‘… makes event e1 be achieved or go on’. Koenig et al. (2008) discusses in depth the notion of instrument but fails to distinguish it correctly from the ‘means’ notion.

69

Characterizing Derivational Paradigms

culminated. This is captured in (31a), which represents a first type of place, by the fact that the verb’s patient argument instantiates the figure in the spatial relationship. In (30b), the whole event itself, i.e., the bus stopping, is the figure and it takes place at some place under the bridge (ground). The corresponding representation is (31b), where the figure argument is e. It says that z is the set of entities that have the property to be a ground where some event e takes place (s = state). (30) a. Elsa stores the tea in the cupboard. Elsa entrepose le thé dans le placard. b. The bus stops under the bridge. Le bus s’arrête sous le pont. (31) a. λz.∃y∃x∃e∃s[V(x,y,e) ∧ LOC(y,loc(z), s)] store/entrepôt b. λz.∃x∃e∃s[V(x,…,e) ∧ LOC(e,loc(z), s)] (bus)stop/arrêt

Action networks with a place_N corresponding to (31a) and (31b) respectively are given in (32). (32) Verb place_N action_N agent_N instrument_N a. entreposer entrepôt entreposage entreposeur — b. s’arrêter arrêt2 arrêt1 — —

However, the place node in the action network rarely corresponds to a built-in ground argument of the V, because verbs rarely imply a locative relationship (Fradin, 2017a). On the contrary, it is quite common to characterize a place by the type of event or activity, which takes place therein. This is all the more frequent as the activity in question is useful or socially valuable. Examples of this third type of place_N are provided in (33). Most instantiation of the ‘place’ node are of this type. (33) a. laver lavoir lavage ‘to wash washhouse washing b. rôtir rôtisserie ‘to roast rotisserie

laveur washer

— —’

rôtissage rôtisseur rôtissoire roasting seller of roast meat roaster’

For the sake of completeness, a few remarks are in order about the verbs which appear in networks. The term verb denotes two concepts: the morphological

70

Fradin

verb, defined by its inflectional paradigm; and the verbal lexeme, defined by the construction(s) that it heads (Fradin & Kerleroux, 2003). In (34), we have two verbal lexemes souder 1 and se souder 2, but one morphological verb. What we are interested in for derivational paradigms is the verbal lexeme. (34) a. L’ouvrier[AGT] a soudé le tuyau[PAT]. ‘The workman welded the pipe.’ b. Les os[PAT] du crâne se soudent entre 9 et 18 mois. ‘Cranial bones knit together between 9 and 18 months.’ By construction, two verbal lexemes corresponding to the same morphological verb give rise to distinct morphological families (Fradin, 2019). Therefore, their alignment possibilities are not equivalent. This is why souder 2 cannot appear in the action network, contrary to souder 1: (35) a. souder1 — soudage soudeur — ‘to weld’ b. se souder2 — soudure — — ‘to fuse, be knit together’ 5.2 The Organization of the Action Network The action network conceived of as a mere skeleton is too general to be useful. In what follows, a few suggestions are made to improve the situation. Insofar as the V plays a central role, it would be wise to calibrate the number and nature of the cells appearing in the network in function of the constructions the V is associated with. We know that only verbs of creation are expected to have a result argument; that only verbs whose semantic representation includes a spatial relationship may have a place argument; that causative verbs of emotion, e.g., befriend have a result state but no instrument (Davis, 2001), etc. If verbal properties are hierarchically represented (Koenig, 1999), a set of expected derivational networks could be associated at each relevant level in the hierarchy. The social relevance of a denomination for a place, a type of action, an agent, etc. may also determine whether the corresponding nodes should be integrated in the action network. For instance, insofar as hiding is not a socially entrenched activity, there is no point to have an AGT_N node for cacher ‘to hide’. The same holds for peigner1, which denotes the action of ‘untangle or arrange the hair by drawing a comb through it’ (Online Oxford Dictionary of English), but not for peigner2, which denotes the preparation of a natural substance (wool, hemp, cotton …) for manufacture with a comb.

Characterizing Derivational Paradigms

71

(36) a. cacher cache2 cachage ??cacheur — ‘to hide’ b. peigner1 ??peignerie peignage ??peigneur peigne ‘to comb’ c. peigner2 peignerie peignage peigneur peigne ‘to card, to comb’ Instead of conceiving of derivational paradigms as rigid schemes, which have to be satisfied to become predicative tools, it would be more appropriate to see how controlled variation can be introduced into the picture. The action network could be devised as a cluster of networks, the core of which would be (37a), but which could aggregate or ignore various other nodes in function of the construction types allowed by the V (or the V type). (37) a. verb b. verb c. verb d. verb e. verb etc.

action_N action_N place_N action_N action_N

agt_N, ins_N agt_N, ∅ action_N agt_N ins_N agt_N ins_N rslt_N agt_N ins_N pat_N

In a complementary way, it seems crucial to have an idea of the number of derivational families that fit each network (for French, using Démonette (Hathout & Namer, 2014)) and of the frequency of their occurrences in corpora. Derivational paradigms can help to make inferences only if these figures are known. 5.3 The Conceptual Foundations of the Activity Network The ‘activity network’ (réseau activité) is based on an activity noun, the corresponding verb (or verbal expression) being a secondary formation. For many activities, professional or for leisure, the (creation) verb does not even exist, e.g., pottery2 ‘craft of making pottery1’. The activity noun somehow is more than the mere nominalization of an agentive verb, since it generally does not denote a process but “an accumulation of processes or the systematic repetition of the same process” (Roché, 2017a, §2.1), cf. nouns such as harvest, sailing, dance. Whereas the action network puts to the fore a process and its participants, the activity network is motivated by the fact that human occupations are classified (or classifiable) in distinct types of activity. Activity networks involve human beings, generally agents, and objects (or substances) specifically linked with the activity in question. The schema proposed by Roché (2017a) is reproduced in (38).

72

Fradin

(38) The activity network Object

activity_N human_N activity_V

The initial object or substance, when it exists, is what makes the activity possible, e.g., paint for painting (39b) or what motivates it, e.g., journal for journalism (39c). But for many activity networks, especially those that can be performed without any object correlate at the beginning (e.g., paint) or at the end (e.g., jewel), this initial object is absent. (39) a. — boxe boxeur faire de la boxe ‘boxing’ b. peinture1 peinture2 peintre faire de la peinture ‘painting’ c. journal journalisme journaliste faire du journalisme ‘journalism’ d. bijou bijouterie1 bijoutier — ‘jewelry’ e. voilier voile2 — faire de la voile2 ‘sailing’ f. — judo judoka faire du judo ‘judo’ g. jardin jardinage jardinier faire du jardinage ‘gardening’ Roché does not pay much attention to the grammatical status of the activity nouns that appear in activity networks. He overlooked Van De Velde (1997),22 which convincingly shows that two types of activity nouns have to be distinguished. Those of the first category (Type I ActNs) denote an event, as attested by their possibility to appear as subjects of event predicates such as se produire ‘to happen, avoir lieu ‘to take place’, e.g., la promenade a eu lieu ce matin ‘the walk took place in the morning’. The only property that makes them distinct from accomplishment nouns in Van De Velde’s classification is that they remain in the singular when they have a measure specifier, as illustrated in (40a). For nouns denoting spatially extended entities, this possibility is observed only for mass nouns (cf. (40b)) (Van De Velde, 1997, p. 372). (40) a. deux heures de (promenade|discussion) ‘two hours of (walk|discussion)’

22  Van de Velde’s argumentation is based on French. But the distinction she draws seems so deeply rooted in the category noun that it probably exists in other languages as well. See also Huyghe (2011).

Characterizing Derivational Paradigms

73

b. trois cents grammes de (pain|sel) ‘three hundred grams of (bread|salt)’ Activity nouns of the second type (Type II ActNs) can also appear with a measure specifier, e.g., deux heures de judo ‘two hours of judo’, trois ans de journalisme ‘three years of journalism’, but unlike Type I they cannot be individualized as shown in (41a), which indicates that they can only be conceived of as atelic processes. (41) a. une (promenade|discussion) de deux heures ‘a (walk|discussion) of two hours’ b. *un ( judo| journalisme| jardinage) de huit ans ‘a (judo|journalism|gardening) of eight years’ In accordance with other mass nouns (cf. (42a)), Type II activity nouns take the so-called ‘partitive article’ (masculine du, feminine de la, plural des, cf. (42b)), while this possibility is not available for Type I ActNs (cf. (42c)). (42) a. Il a acheté (du beurre|de la confiture). ‘He bought (butter|jam).’ b. Faire du ( jardinage| judo) lui ferait du bien. ‘Doing (gardening|judo) would be good for her.’ c. *Faire de la (promenade|discussion) lui ferait du bien. ‘Doing (walk|discussion) would be good for her.’ According to Van De Velde (1997, p. 374), Type II ActNs “have all properties of non-countable nouns”. Even though both types without any doubt denote “activities, that is actions that proceed along the time axis in a homogenous way and do not have a necessary end” (“d’activités, c’est-à-dire d’actions se déroulant dans le temps de manière homogène, et sans comporter de terme necessaire”), she claims that “types I and II represent two poles […] which behave in opposite ways for what regards the manner that the nominalized activity is completed […] Type I groups together activities directed toward an end. […] On the contrary, activity nouns of type II denote a homogenous and continuous type of action without any other finality than itself”. (“les types I et II représentent deux pôles […] Le type I regroupe les activités orientés vers une fin. […] A l’opposé, les noms d’activité de type II dénotent un type d’action

74

Fradin

homogène et continu qui n’a plus d’autre finalité que soi-même”) (Van De Velde, 1997, 375–376, my translation, BF). In (40a), promenade and discussion are viewed as uncountable nouns. In (41a), they are viewed as countable nouns. But (41a) is possible because the end happened, because a limit has been reached. Expressions (41b) are impossible because the activity remains unaccomplished, but unlike actions called ‘accomplishment’, it can stay unaccomplished without stopping being what it is (Van De Velde, 1997, 375–376). Van de Velde sorts out two important properties of Type II activity nouns. The first is the specific syntactic structures (43) that they occur in ( faire ‘to do’, jouer ‘to play’, DEF = definite article). (43) a. faire de DEF activity2_N b. jouer à DEF activity2_N

faire du ( jardinage|piano) ‘do gardening’ jouer au bridge ‘play bridge’

These structures are the only way to predicate the activity of the agent. The second is that a Type II activity noun cannot take a complement in the same way as the corresponding verb can, when it exists, because what we obtain is a new kind of activity predicate. Whereas in (44a) à la gare can be parsed as an allative complement of the manner of motion verb courir, it is not the case in (44b), even though course is the activity noun correlated with courir: course à la gare is interpreted as a type of race in the same way as, for instance, course à pied ‘running, footrace’, course hippique ‘horse race’, course automobile ‘car racing’, course de haies ‘hurdle race’, are. Since such a race does not exist and is difficult to fathom, (44b) is utterly weird. (44) a. Pierre a couru à la gare. ‘Peter ran to the station.’ b. ?*Pierre fait de la course à la gare. ‘Peter made running to the station.’ This leads us to van de Velde’s main conclusion: constructions involving structures (43) allow one to create taxonomies in the domain of action, and they are the only ones to do so in French (Van De Velde, 1997, pp. 382–385). They offer us a way to classify activities. As regards activity networks, these constructions imply that the person they are predicated of is an agent. For instance, while (45a) is sound, (45b) is impossible because chuter ‘to fall’ is an unaccusative verb. The patientive nature of its subject is inherited by chute. This clashes with the conditions imposed by the V faire ‘to do’, which requires an agentive subject. This is why activity nouns

Characterizing Derivational Paradigms

75

appearing in structures (43) imply that an agent is involved (whose degree of agentivity may vary, cf. §4). (45) a. Pierre fait du saut à ski. ‘Peter is a ski-jumper.’ b. *Pierre fait de la chute. ‘Peter (does falling | is a faller).’ From the preceding discussion, we conclude that the activity noun in activity networks corresponds to type II activity noun. 5.4 Distinguishing the Two Networks The distinction between the two networks clearly appears whenever the same morphological family has to be split between the two. Examples (46)–(47) illustrate this case: (46) a. (object) activity_N human_N activity_V b. parachute parachutisme parachutiste faire du parachute ‘skydiving’ (47) a. V action_N agent_N instrument_N b. parachuter parachutage parachuteur parachute ‘to parachute’ e.g., parachuter des vivres ‘parachute supplies’ Obviously, skydivers parachute nothing, except themselves, and a ‘parachuteur’ is a plane (avion parachuteur) or a man who makes a particular parachuting take place. This example shows that networks can overlap (here through parachute) and this certainly happens for other networks as well. The more an activity is a long-established one, the more likely its morphological family will tend to instantiate some nodes through idiosyncratic terms. Fencing and rowing are cases at hand.23 (48) (object); activity_N, human_N, activity_V a. épée; escrime, escrimeur, faire de l’escrime ‘fencing’ b. aviron1; aviron2, rameur, faire de l’aviron2 ‘rowing’

23  In (48a), several weapons can instantiate the node ‘object’ in addition to épée ‘sword’, e.g., fleuret ‘foil’, sabre ‘sabre’. Such a situation is not peculiar to fencing (cf. palet ‘puck’ for hockey in (51)).

76

Fradin

Note that rameur ‘rower’ appears in an action network in (49): one can row without doing that as a pastime or hobby. (49) a. V, action_N, agent_N, instrument_N b. ramer, —, rameur, rame ‘to row’ To that extent, the existence of a phrasal activity verb with structure (41) is the touchstone of the activity network. Since *faire du sablage is out, sable ‘sand’ can only occur as a means_N in an action network. (50) a. V, action_N, agent_N, instrument_N, means_N b. sabler, sablage, sableur, sableuse, sable ‘to sand’ Activity networks contain no verbal lexeme with a structuring function, since the role of pivot is supported by the activity noun. However, the existence of a correlated verbal phrase appears to be the best criterion to assess whether a derivational paradigm belongs to the activity network type. A major interest of activity networks lies in the fact that they integrate nouns morphologically marked as agent (cf. (51)), even though the source verb is missing: no inflected form of the verb hockeyer was found in Google (June 2018), which precludes us to admit that this verb exists. (51) (object); activity_N, human_N, activity_V a. crosse; hockey, hockeyeur, jouer au hockey ‘hockey’ Activity networks are one of the instances of the numerous noun-based derivational paradigms that exist (Roché, 2017b). Many of them are small, with two nodes only, as the ‘monophytic_place network’, e.g., hêtre~hêtraie ‘beech, beechwood’ mentioned in §4. 6

Conclusion

What motivated the present study was the issue of the predictive capacity of derivational paradigms in comparison with the corresponding capacity exhibited by inflectional paradigms. Respecting the alignment relation is a key condition for predicting what the form in a given cell can be, knowing the forms instantiating other cells. For this condition to be satisfied, content relations must be properly distinguished. This is why concepts ‘instrument’ vs. ‘means’,

Characterizing Derivational Paradigms

77

or ‘action’ vs. ‘activity’, or else ‘activity’ vs. ‘place’ must be accurately defined and used (§5.1). In the same vein, it has been shown that it is the verbal lexeme that is relevant when a verb is involved as a base in a paradigm, not the morphological verb (cf. souder 1 vs. souder 2). But the question of the predictive capacity makes sense only if the issue of the foundation of derivational paradigms themselves has got an answer. What arguments can we provide to ascertain that a given network of nodes can be said to form a derivational paradigm (Paradigm 1, according to Carstairs (1987))?24 And how can we show that this paradigm should be distinguished from other, sometimes very similar, paradigms? The gist of this article was an attempt to give preliminary answers to these questions. Needless to say, these answers remain very tentative. The main points of this article are the following: (i) Several types of derivational paradigms exist and are articulated along two main distinctions: first, those that are event-related and those that are not; second, those that are human-centered and those that are not. Paradigms with the higher number of cells/nodes are those involving morphological families centered on human activities. They are also those that exhibit the higher number of members. In comparison, morphological families derived from nouns denoting natural kinds have fewer cells/ nodes and the number of their members is (relatively) low. An inquiry into a larger set of data is needed before drawing any firm conclusion. Non-event-related paradigms are illustrated by morphological families that include names of location or of status. (ii) Content relations may originate in three sources: event-rooted schemes that generally involve an agent, properties associated with the entity denoted by a paradigm’s pivot noun, and finally highly available derivational meta-patterns that play a salient role in the grammar of the language in question, e.g., adjective/noun of quality. (iii) The action network and the activity network, introduced by Roché (2017a), indeed seem to correspond to distinct types of human-centered derivational paradigms. While the action network is rooted in the event denoted by a verb, the activity network rests on activity nouns that have properties of uncountable nouns and are used to classify (human)

24  Kos (this volume) suggests that “the pre-existence of cells seems to be clearly manifested” by derivational relations belonging to Dokulil’s modificational or transpositional onomasiological categories (Dokulil, 1994). However, the reasons why it is so are not given.

78

Fradin

activities. Frequently, lexemes derived from the same origin belong to distinct morphological families and thereby should be attached to either one of these paradigms. At a more general level, however, the predictive value that can be associated with these two networks is weak, because the aligning relations cannot be settled once and for all. The number and nature of the nodes slightly vary. Since this situation repeats itself with almost all derivational paradigms, it would be worth to see whether (more) stable networks cannot be established on the basis of activity domains or sub-domains, which amounts to take into account the social anchoring of these networks. (iv) Derivational paradigms should not be conceived of as mere abstract schemes. To flesh them out, we must take advantage of the information attached to lexemes and give a more precise picture of the concepts constituting the dimensions along which families are aligned. The question of the number of cells/nodes a given type of derivational paradigm should have and of their nature is a very important one. In-depth qualitative and quantitative studies have to be undertaken to see better what the possibilities are. (v) It is not completely clear whether the notion of paradigm defined as a network of implications can be applied to all types of languages without changing the nature of the entities linked by the Aligning relation. If the goal is to make derivational paradigms more efficient qua descriptive devices, we have to account for their variety in detail in order to see better how the general picture has to be devised. The understanding of the parts depends on the understanding of the whole, but the understanding of the whole cannot be clarified while ignoring the functioning of the parts. This is the dilemma we have to face. This explains why, often enough, we had to investigate deeply into the behavior of the words that constitute derivational paradigms. In a complementary way, an investigation of the types of derivational paradigms existing at the level of languages should be carried out: which types exist in a given language? How stable are they diachronically? How tight are the links between the cells/nodes of the paradigm? Answering these questions is crucial to improve the tools that will allow us to account for the data. Abbreviations A acc adv

adjective accusative adverb

Characterizing Derivational Paradigms

agt, agt all aux DA dat f.pl gen hab ins loc mns N nb nf nwk nom npass pl poss prs rslt sg V Vipf Vpf

79

agent allative auxiliary denominal adjective dative feminine plural genitive habitual instrument locative function means noun number non-feminine network nominative non-passive plural possessive present result singular verb imperfective verb perfective verb

Acknowledgment

I would like to thank Fiammetta Namer and Fabio Montermini for their relevant remarks and insightful comments about a first version of this paper. References Ackerman, F., & Moore, J. (2001). Proto-properties and grammatical encoding. A correspondence theory of argument selection. Stanford: CSLI. Bauer, L. (1997). Derivational paradigms. In G. Booij & J. van Marle (Eds.), Yearbook of Morphology 1996 (pp. 243–256). Dordrecht: Kluwer. Bauer, L. (2001). Morphological productivity. Cambridge: Cambridge University Press.

80

Fradin

Beavers, J. (2008). Scalar complexity and the structure of events. In J. Dölling, T. Heynde-Zybatow, & M. Schäfer (Eds.), Event structures in linguistic form and interpretation (pp. 245–265). Berlin, New York: Mouton de Gruyter. Beavers, J. (2011). On affectedness. Natural language & linguistic theory, 29(2), 335–370. Bickel, B., & Zúñiga, F. (2017). The ‘word’ in polysynthetic languages: phonological and syntactic challenges. In M. Fortescue, M. Mithun, & N. Evans (Eds.), The Oxford handbook of polysynthesis (pp. 158–186). Oxford: Oxford University Press. Blevins, J. P. (2001). Paradigmatic derivation. Transaction of the philological society, 99(2), 211–222. Blevins, J. P. (2006). Word-based morphology. Journal of linguistics, 42(3), 531–573. Blevins, J. P. (2016). Word and paradigm morphology. Oxford: Oxford University Press. Bonami, O., & Beniamine, S. (2016). Joint predictiveness in inflectional paradigm. Word structure, 9(2), 156–182. Bonami, O., & Strnadová, J. (2018). Paradigm structure and predictability in derivational morphology. Morphology, 10, pp. 1–31. Retrieved from: 10.1007/s11525-018-9322-6 [Accessed 9 March, 2019]. Bonan Garrigues, M., & Elie, J. (1971). Essai d’analyse sémique. Cahiers de lexicologie (XIX), 70–93. Booij, G. (1986). Form and meaning in morphology: the case of Dutch ‘agent nouns’. Linguistics, 24, 503–517. Booij, G. (2002). The morphology of Dutch. Oxford: Oxford University Press. Booij, G. (2005). The grammar of words. Oxford: Oxford University Press. Boyé, G., & Schalchli, G. (2017). Realistic paradigm for derivational morphology. ParadigMo 2017, first workshop on paradigmatic word formation modeling, June 19–20 2017, Toulouse. Bybee, J. (1985). Morphology. A study of the relation between meaning and form. Amsterdam: John Benjamins Publishing Company. Carlson, G. N. (1998). Thematic roles and the individuation of events. In S. Rothstein (Ed.), Events and grammar (pp. 35–51). Dordrecht: Kluwer Academic Publishers. Carstairs, A. (1987). Allomorphy in inflexion. London: Croom Helm. Carstairs-McCarthy, A. (2010). The evolution of morphology. Oxford: Oxford University Press. Corbin, D. (1987). Morphologie dérivationnelle et structuration du lexique (Vol. 1–2). Lille: Presses Universitaires du Septentrion. Original edition, Tübingen: Niemeyer. Croft, W. (1991). Syntactic categories and grammatical relations: The cognitive organization of information. Chicago: The University Press of Chicago. Cruse, A. (1973). Some thoughts on agentivity. Journal of linguistics, 9(1), pp. 11–23. Davis, A. R. (2001). Linking by types in the hierarchical lexicon. Stanford: CSLI. DeLancey, S. (1985). Agentivity and syntax. In W. H. Eilford, P. D. Kroeber, & K. L. Peterson (Eds.), Papers from the parasession on causatives and agentivity (Vol. 2, pp. 1–12). Chicago: CLS.

Characterizing Derivational Paradigms

81

Dokulil, M. (1994). The Prague School’s theoretical and methodological contribution to “word formation” (derivology). In J. Panevová & Z. Skoumalová (Eds.), Cizojazyčné studie Miloše Dokulila. Obsah Výraz Význam 2 (pp. 179–210). Prague: Filozofická fakulta UK a Nadace Viléma Mathesia. Dowty, D. R. (1991). Thematic proto-roles and argument selection. Language, 67(3), 547–619. Fradin, B. (2005). On a semantically grounded difference between derivation and compounding. In W. U. Dressler, D. Kastovsky, O. E. Pfeiffer, & F. Rainer (Eds.), Morphology and its demarcations (pp. 161–182). Amsterdam, Philadelphia: John Benjamins. Fradin, B. (2012). Les nominalisations et la lecture ‘moyen’. Lexique, 20, 129–156. Fradin, B. (2017a). Deverbal nominalizations denoting places. ISMO. First International Symposium of Morphology. December 13–15, 2017, at Villeneuve d’Ascq. Fradin, B. (2017b). The multifaceted nature of denominal adjectives. Word Structure, 10(1), 27–53. Fradin, B. (2019). Competition in derivation: what can we learn from French doublets in -age and -ment? In F. Gardani, F. Rainer, H. C. Luschützky, & W. U. Dressler (Eds.), Competition in morphology (pp. 67–93). Berlin: Springer. Fradin, B., & Kerleroux, F. (2003). Troubles with lexemes. In G. Booij, J. De Cesaris, S. Scalise, & A. Ralli (Eds.), Topics in morphology. Selected papers from the Third Mediterranean Morphology Meeting (Barcelona, September 20–22, 2001) (pp. 177– 196). Barcelona: IULA-Universitat Pompeu Fabra. Fradin, B., & Winterstein, G. (2012). Tuning agentivity and instrumentality: deverbal nouns in -oir revisited. Paper delivered at Décembrettes 8, 6–7 December 2012, Bordeaux. Gärdenfors, P. (2000). Conceptual spaces: The geometry of thought. Cambridge, Mass.: The MIT Press. Gärdenfors, P. (2014). The geometry of meaning. Semantics based on conceptual spaces. Cambridge, London: The MIT Press. Geuder, W. (2003). Agentive adjectives and adverbs. In C. Beyssade, O. Bonami, P. Cabredo Hofherr, & F. Corblin (Eds.), Empirical issues in formal syntax and semantics 4/Questions empiriques et formalisation en syntaxe et sémantique 4 (pp. 173–188). Paris: Presses de l’Université de Paris-Sorbonne. Haspelmath, M. (1996). Word-class-changing inflection and morphological theory. Yearbook of Morphology 1995 (pp. 43–66). Dordrecht: Kluwer. Hathout, N. (2011). Une approche topologique de la construction des mots: propositions théoriques et application à la préfixation en anti-. In M. Roché, G. Boyé, N. Hathout, S. Lignon, & M. Plénat (Eds.), Des unités morphologiques au lexique (pp. 251–318). Paris: Hermès, Lavoisier. Hathout, N., & Namer, F. (2014). Démonette, a French derivational morpho-semantic network. Linguistic issues in language technology, 11(5), 125–168.

82

Fradin

Huyghe, R. (2011). (A)telicity and the mass-count distinction: the case of French activity nominalizations. Recherches linguistiques de Vincennes, 40, 101–126. Jurafsky, D. (1996). Universal tendencies in the semantics of the diminutive. Language, 72, 533–578. Koenig, J-P. (1999). Lexical relations. Stanford: CSLI. Koenig, J-P., Mauner, G., Bienvenue, B., & Conklin, K. (2008). What with? The anatomy of a (proto)-role. Journal of semantics, 25(2), 175–220. Kos, P. (this volume). The level of paradigmaticity within derivational networks. In J. Fernández-Domínguez, A. Bagasheva, & C. Lara-Clares (Eds.), Paradigmatic relations in word formation (pp. 85–99). Leiden: Koninklijke Brill NV. Kracht, M. (2002). On the semantics of locatives. Linguistics and philosophy, 25, 157–232. Landman, F. (2000). Events and plurality: The Jerusalem lectures. Studies in Linguistics & Philosophy (Vol. 76). Dordrecht, Boston, London: Kluwer Academic Publishers. Lehrer, A. (1974). Semantic fields and lexical structure. Amsterdam: North Holland. Levin, B., & Rappaport Hovav, M. (2005). Argument realization. Cambridge: Cambridge University Press. Matthews, P. H. (1974). Morphology. 2nd Edition 1991. Cambridge: Cambridge University Press. Mel’čuk, I. A. (1993). Cours de morphologie générale. Première partie: le mot (Vol. 1). Montréal: Presses de l’Université de Montréal—CNRS Editions. Mel’čuk, I. A. (1994). Cours de morphologie générale. Deuxième partie: significations morphologiques (Vol. 2). Montréal: Presses de l’Université de Montréal—CNRS Editions. Mithun, M. (1991). Active–agentive case marking and its motivations. Language, 67(3), 510–546. Panfilov, V. S., Baumann, H., & Haspelmath, M. (2001). Vietnamesisch (Viet-Muong). In G. Booij, C. Lehmann, J. Mugdan, & S. Skopeteas (Eds.), Morphologie Morphology. Ein internationales Handbuch zur Flexion und Wortbildung. An international handbook on inflection and word-formation (Vol. 2, pp. 1545–1554). Berlin, New York: Walter de Gruyter. Plag, I. (1998). The polysemy of -ize derivatives. Yearbook of morphology 1997 (pp. 219– 242). Dordrecht: Kluwer. Plungian, V. (2000). Dogon. (Vol. 64, Languages of the World/Materials). München: LINCOM Europa. Plungian, V. (2001). Agglutination and flection. In M. Haspelmath, E. König, W. Oesterreicher, & W. Raible (Eds.), Language typology and language universals. Sprachtypologie und sprachliche Universalien. La typologie des langues et les universaux linguistiques (Vol. 1, pp. 669–678). Berlin, New York: Walter de Gruyter. Pottier, B. (1963). Systématique des éléments de relation. Paris: Klincksieck. Rastier, F. (1987). Sémantique interprétative. Paris: PUF.

Characterizing Derivational Paradigms

83

Roché, M. (2005). Sur une classe d’adjectifs par conversion. In I. Choi-Jonin, M. Bras, A. Dagnac, & M. Rouquier (Eds.), Questions de classification en linguistique: méthodes et descriptions. Mélanges offerts au Professeur Christian Molinier (pp. 319– 343). Bern, Berlin: Peter Lang. Roché, M. (2007). Logique lexicale et morphologique: La dérivation en -isme. In F. Montermini, G. Boyé, & N. Hathout (Eds.), Selected proceedings of the 5th Décembrettes. Morphology in Toulouse (pp. 45–58). Somerville: Cascadilla Press. Roché, M. (2017a). Les familles dérivationnelles: comment ça marche? Toulouse: Université Toulouse 2 Jean Jaurès. Roché, M. (2017b). Un exemple de réseau constructionnel: ethnique, toponymes, gentilés. Toulouse: Université Toulouse 2 Jean Jaurès. Sapir, E. (1921). Language. New York: Harcourt, Brace and Company. Schlesinger, I. M. (1989). Instruments as agents: On the nature of semantic relations. Journal of linguistics, 25(1), 189–210. Silva, L. de J. (2011). Morphosyntaxe du rikbaktsa (Amazonie brésilienne). Thèse de doctorat, UFRL, Université Paris Diderot-Paris 7, Paris. Spencer, A. (1999). Transposition and argument structure. Yearbook of Morphology 1998 (pp. 73–101). Dordrecht: Kluwer. Spencer, A. (2013). Lexical relatedness. A paradigm-based model. Oxford: Oxford University Press. Spencer, A. (2018). On lexical entries and lexical representations. In O. Bonami, G. Boyé, G. Dal, H. Giraudo, & F. Namer (Eds.), The lexeme in descriptive and theoretical morphology (pp. 277–301). Berlin: Language Science Press. Štekauer, P. (2014). Derivational paradigms. In R. Lieber & P. Štekauer (Eds.), The Oxford handbook of derivational morphology (pp. 354–369). Oxford: Oxford University Press. Talmy, L. (2000). Toward a cognitive semantics: Concept structuring systems (Vol. 1). Cambridge, London: The MIT Press. Tixonov, A. N. (1985). Slovoobrazovatel’nyj slovar’ russkogo jazyka. 2 vols. Moskva: Russkij Jazyk. Van De Velde, D. (1997). Un dispositif linguistique propre à faire entrer certaines activités dans des taxinomies: faire + du + nom d’activité. Revue de linguistique romane, 62, 369–395. Van Valin, R. D. Jr., & LaPolla, R. J. (1997). Syntax. Structure, meaning and function. Cambridge: Cambridge University Press. Wolff, P. (2003). Direct causation in the linguistic encoding and individuation of causal events. Cognition, 88, 1–48. Wolff, P. (2007). Representing causation. Journal of experimental psychology: General, 136, 82–111.

84

Fradin

Zwarts, J., & Winter, Y. (2000). Vector space semantics: A model-theoretic analysis of locative prepositions. Journal of logic, language and information, 9(2), 169–211. Zwicky, A. M. (1990). Inflectional morphology as a (sub)component of grammar. In W. U. Dressler, H. C. Luschützky, O. E. Pfeiffer, & J. R. Rennison (Eds.), Contemporary morphology (pp. 217–236). Berlin, New York: Mouton de Gruyter.



Annexes

Figure 3.2 Inflectional paradigm

Figure 3.3 Derivational paradigm

Chapter 4

The Level of Paradigmaticity within Derivational Networks Petr Kos 1

Introduction

The derivational network can broadly be understood as a system of all derivatives coined from a simple underived word. In the Czech and Slovak tradition of the onomasiological approach to word formation, derivational networks are equated with derivational paradigms. So, in this manner, Štekauer (2014) states that within the onomasiological approach to word formation “the derivational paradigm in the narrow sense is conceived as an ordered set of all complex words directly derived from a single basic (motivating) word” (p. 363), providing Furdík’s (2004, p. 74) example of such a paradigm in Slovak based on the word škola ‘school’. Similarly, Dokulil (1986) perceives a word nest, his term for a derivational network, “connecting all words synchronically derived from the identical stem” (p. 207, my translation), to be an example of a paradigm in word formation. However, the question arises to what extent the relations within derivational networks meet “the requirement of systematic, regular, and predictable relationships” (Štekauer, 2014, p. 369), i.e., to what extent we may apply the analogy to inflection. There are various authors who discuss the distinction between inflectional and derivational morphology, or more specifically inflectional and derivational paradigms (e.g., Bauer, 1997; Booij, 2000; Stump, 2005; Bonami & Strnadova, 2019). However, from the examples they use in their discussions it follows that they do not take into account the fact that different types of derivation exhibit a different scale of similarity between inflection and derivation—the types distinguished by Dokulil (1962, 1986) as the modificational, transpositional, and mutational onomasiological categories. In this chapter I aim to argue that derivational networks, although sharing the same stem, cannot automatically be considered paradigmatic, as not all relationships within such sets are fully regular and predictable. Dokulil’s (1962, 1986) major relational onomasiological categories, namely, the modificational, the transpositional, and the mutational, will be discussed on the background

© Koninklijke Brill NV, Leiden, 2020 | doi:10.1163/9789004433410_005

86

Kos

of Bauer’s (1997) features of paradigmaticity, and I will demonstrate that only the modificational and transpositional onomasiological categories resemble inflection in terms of their level of paradigmaticity. The text is organized as follows: §2 deals with the theoretical delimitation of a derivational network, §3 discusses features of a paradigm following Bauer’s (1997) conception, and §4 describes the three major onomasiological categories as conceived by Dokulil (1962, 1986). The core of the paper is §5, in which the features of a paradigm are discussed within each onomasiological category in order to illustrate my original claim. 2

Derivational Networks

In the Czech and Slovak, or more broadly Eastern European, word-formation tradition, derivational networks are understood as sets of derivatives in which the derivational process is synchronically transparent. This synchronic transparency is what distinguishes it from an older concept of a word-formation family, which also includes lexemes containing the same stem but otherwise synchronically obscure.1 Also, derivational networks are not meant to be mere lists of derivatives with a common stem but aim to reflect the internal structure of the whole derivational system through “chains of gradual […] derivational steps” (Horecký, Buzássyová, & Bosák, 1989, p. 39, my translation). Dokulil’s (1986) example of a derivational network, a word nest in his terminology, based on the stem kniha ‘book’ (see Figure 4.1) includes, among others, the following chains of gradual derivational steps: (1) kniha → knihovna → knihovník → knihovnice ‘book’N ‘library’N ‘librarian’N ‘female librarian’N (2) kniha → knihovna → knihovní ‘book’N ‘library’N ‘related to library’Adj (3) kniha → knihovat → knihovač ‘book’N ‘to keep record books’V ‘a person who keeps record books’N → knihovačka ‘a female person who keeps record books’N 1  See the example of WF family based on med ‘honey’ used in Dokulil (1962), which also included medvěd ‘bear’ whose original motivation ‘honey-eater’ is obscure for contemporary speakers.

The Level of Paradigmaticity within Derivational Networks

87

(4) kniha → knížka → knížečka ‘book’N ‘bookDIM’N ‘bookDOUBLE DIM’N

Figure 4.1 Word-nest with kniha ‘book’ as the motivating word taken from Dokulil, 1986, p. 208

I suggest that such derivational steps within derivational networks are based purely on form and only some of the steps are predictable in terms of both form and meaning, which seems to be the basic precondition of any paradigm. 3

What Is a Paradigm?

I borrowed the title of the chapter from Bauer (1997, p. 244), who in his search for paradigms within derivational morphology starts his discussion with a list of features that are relevant for inflectional paradigms. The following is a summary of these features: – A paradigm is a series of morphologically related forms which share a base or base-type. – The forms in a paradigm are semantically related by more than the meaning of the base. Semantically, the forms in a paradigm differ in terms of relatively marginal features. – A paradigm provides a typical set of realizations that are generalizable (to some extent) beyond the individual base word chosen as an illustration.

88

Kos

– Paradigms are frequently arranged around a basic form, or set of forms, from which the other forms in the paradigm can be predicted. The approach taken in this paper is that canonical derivational paradigms are those sets of derivatives which meet all the four above-mentioned basic conditions, i.e., sets which share an identical base, are closely semantically related, and are fully predictable both in form and meaning. Only such paradigms comply with Horecký et al.’s (1989, p. 42) conclusion that “new complex words do not come into existence as isolated units but rather as complete paradigms” (Štekauer, 2014, p. 360). 4

Dokulil’s Onomasiological Categories

Bauer’s (1997) features of paradigmaticity within derivational networks will be discussed on the background of Dokulil’s major onomasiological categories, namely the modificational, the transpositional, and the mutational. The following is their brief description. 4.1 Modificational Onomasiological Category The modificational onomasiological category is characterized by “adding a modifying element to the contents of the given concept” (Dokulil, 1986, p. 212, my translation). Among the typical modifications within this category are diminutiveness, as in the Czech2 strom ‘tree’ → strom-ek ‘treeDIM’, augmentation, as in strom ‘tree’ → strom-isko ‘treeAUGM’, feminine nouns derived from masculine nouns, as in učitel ‘teacher’ → učitel-ka ‘woman teacher’ or tygr ‘tiger’ → tygř-ice ‘tigress’, immaturity lev ‘lion’ → lv-íče ‘lion cub’, mass nouns list ‘leaf’ → list-í ‘foliage’, repetitive action in verbs, as in psát ‘to write’ → psávat ‘to write habitually’, or other semantic features according to the meaning of prefixes, as in babička ‘grandmother’ → pra-babička ‘great grandmother’, věda ‘science’ → pseudo-věda ‘pseudo-science’, and others. The lexical meaning of the derivative belongs to the same ontological category as that of the stem, and there is no change in the word-class of the word, as modification is a word-class preserving category. 4.2 Transpositional Onomasiological Category Within the transpositional onomasiological category, the meaning of the motivating word also remains unchanged; what changes is its phenomenal category, i.e., the word-class, only. Typical instances of this onomasiological 2  All examples in this chapter are taken from Czech.

The Level of Paradigmaticity within Derivational Networks

89

category are the transposition of quality into substance bledý ‘pale’ → bled-ost ‘paleness’, action into substance pít ‘to drink’ → pit-í ‘drinking’, quality into circumstance hezký ‘nice’ → hezk-y ‘nicely’, etc. Dokulil’s approach may be illustrated with the example of deverbal nouns: “Action viewed through the prism of substance still remains to be action, what has changed is its conception—it is not understood as a mark of something else but as a phenomenon of its own” (Dokulil, 1982, pp. 258–259, my translation). However, not all changes in the word-class category are transpositional in Dokulil’s conception. For example, denominal adjectives, such as hudební ‘musical’, do not have a transpositional function only, but also denote varied relations to the base word; even if we generalize these relations as ‘referring to’ or ‘associated with’, we cannot see it as a mere projection of substance on the plane of attributiveness. Dokulil, 1982, p. 259, my translation

An example which can better illustrate this is the denominal adjective medový ‘honey’Adj, a derivative of the noun med ‘honey’. The adjective in the given context always expresses one aspect of the substance only, be it colour, taste, sweetness, also metaphorically, and purpose, as in medová barva ‘honey colour’, medová chuť ‘honey taste’, medový úsměv ‘honey smile’, and medová plástev ‘honeycomb’, respectively. It is for this reason that the formation of denominal adjectives falls under the mutational category. 4.3 Mutational Onomasiological Category The mutational onomasiological category deals with naming in its narrowest sense, as it seems to be the only one that has its starting point in the perception or conceptualization of the extra-linguistic reality. Within the mutational category, new naming units are formed by first categorizing the named referent into an existing conceptual category (e.g., human), which becomes the onomasiological base, and it is then determined by a feature which distinguishes the referent from other members of the conceptual category, which is the onomasiological mark (e.g., book binding), thus gaining knih-ař ‘bookbinder’— the onomasiological base is expressed by the suffix -ař, the onomasiological mark being kniha ‘book’. The semantic relation between the base and mark, i.e., binding, remains in this particular case unexpressed. As the stem of the derivative in the mutational category is the onomasiological mark, i.e., a distinguishing feature within the conceptual category expressed by the suffix, the meaning of the stem always differs from the overall

90

Kos

meaning of the derivative. In the case of knih-ař ‘bookbinder’, the stem kniha ‘book’ serves as a distinguishing marker for the person, an onomasiological base, who binds them. By the same token, the question of whether the mutational category is wordclass preserving or word-class changing appears irrelevant, as onomasiological marks are commonly nouns, adjectives, and verbs, and it is often a matter of the naming strategy of the coiner whether the onomasiological mark is a noun, as in knih-ař, or, for example, an adjective, as in červen-ka ‘robin redbreast’ (lit. red + suffix). 5

Paradigmaticity from the Perspective of Onomasiological Categories

The (non-)paradigmatic nature of Dokulil’s three major onomasiological categories will be described on the background of Bauer’s (1997) features of paradigmaticity listed above. 5.1 Lexemes Sharing a Base In principle, all lexemes in a derivational network are morphologically related and share the same base, so this feature does not require any detailed discussion in terms of Dokulil’s onomasiological categories. 5.2 Semantic Relatedness among the Lexemes Another feature of any central type of paradigm is that the members should be closely semantically related. The modificational onomasiological category, as has been suggested above, does not change the semantic category of the lexeme, since it adds or modifies a semantic feature which may be considered marginal. In Beard and Volpe’s (2005) terms, this covers expressive derivation, such as diminutives, which “are pragmatic in that they express neither a grammatical function nor a lexical meaning but merely a subjective evaluation of the speaker, the interpretation of which is contextually determined” (p. 199). Thus, a reference to an identical tree may be expressed by a neutral term strom ‘tree’, an endearing term strom-ek ‘treeDIM’, or an augmentative term strom-isko ‘treeAUGM’—the lexical meaning remains the same; what changes is the expressive value only. Still within the modificational category, the same can be said of Beard and Volpe’s (2005) switches in feature value, such as adding a feature of collectiveness to countable nouns, e.g., strom-oví ‘collective of tree’ or adding or modifying a gender marker in nouns referring to humans, e.g., an addition of a female

The Level of Paradigmaticity within Derivational Networks

91

gender marker to otherwise grammatically masculine učitel ‘teacher’, resulting in učitel-ka ‘woman teacher’. The ontological category of strom and učitel in these examples remains unchanged, what is added or modified is a semantic feature only. The semantic relatedness is not disrupted within the transpositional category, either. Transposition changes the syntactic function of a word, “while its lexical content remains untouched and is thus preserved in its entirety” (Dokulil, 1982, p. 258, my translation). This view is shared by ten Hacken (2015), who defines transposition as “a process that changes the syntactic category of a word, does not change its semantic category, and does not modify, add or delete any semantic features” (p. 196). However, within the mutational onomasiological category, by definition, the meaning of the original underived lexeme differs from the meaning of the new derivative, as within the naming process the original lexeme becomes the onomasiological mark, which distinguishes the entity named from other members of its conceptual category. The onomasiological mark is thus loosely associated with the named entity, expressed by the suffix. In Furdík’s (2004, p. 74) examples, škol-ák ‘school boy’ or školn-ík ‘school janitor’, are both persons in one way or another associated with the school, i.e., the ontological category of school is used to name a human being. Moreover, the type of association is not expressed, so not only do the meanings of the original lexeme and the new derivative differ, but also their semantic relation remains unclear. In this respect, these derivatives are analogical to NN ground compounds (for classification of compounds, see Bisetto & Scalise, 2005). Such coinages are idiomatic, so their resulting meanings remain to a degree unpredictable. Even in names where the onomasiological mark is an adjective, such as červen-ka ‘robin redbreast’ (lit. red + suffix), the way the colour is associated with the onomasiological base, in our case the bird, is not fully interpretable, since the interpretation of červen-ka as “the bird is red” is misleading, as it is only the breast of the bird that is red, not the whole bird. The stems of neighbouring derivatives within a derivational network are not necessarily semantically related, either, as the onomasiological mark may express different facets of the meaning of the original lexeme, e.g., škola ‘school’ in školn-ík ‘school janitor’ is understood as a building, whereas škola ‘school’ in škol-ák ‘school boy’ is understood as an educational institution. Again, the apparent semantic relation is based on the form of the stem only. All in all, the meaning of the derivatives within the mutational category always differs from the one of the original lexeme, and it is to a large extent idiomatic as the form of the original lexeme can never fully predict the meaning

92

Kos

of the derivative.3 Neither can the identity of the stem’s form guarantee the semantic relatedness between neighbouring derivatives within a derivational network. So, when Stump (2005) claims that unlike inflection, “operations of word-formation often express meanings that are at least partially unpredictable” (p. 55), providing dollarize as one of the examples, and then he goes on to state that other instances of derivation “show high semantic regularity; for instance, adverbs arising from adjectives through the suffixation of -ly generally have the meaning ‘in an X manner’, where the adjectival base supplies the meaning X” (p. 55), he speaks of the mutational and transpositional categories, respectively. 5.3 Sets of Generalizable Realizations Sets of generalizable realizations, in other words the pre-existence of cells for relevant morphosyntactic categories, is one of the most important features of inflectional paradigms—there is a pre-existing cell, e.g., 3rd person singular in the present tense in verbs or a pre-existing cell for the plural in countable nouns. Any new lexeme of a particular grammatical category thus automatically enters its appropriate paradigm and fills its pre-existing cells with appropriate forms. In derivational morphology, within the modificational onomasiological category, the pre-existence of cells seems to be clearly manifested by the modifying elements added to a given lexical base. Which modifying elements are appropriate for a base is determined by the grammatical features of the lexeme. Thus, in Czech, any noun expressing a human referent in principle distinguishes male and female gender, e.g., onomaziolog ‘male onomasiologist’, onomaziolož-ka ‘female onomasiologist’, nouns in principle accept diminutive markers, e.g., květina ‘flower’ → květin-ka ‘flowerDIM’, etc. Similarly to inflectional morphology, where “languages vary with respect to both the inflectional categories to which their morphology is sensitive and the morphosyntactic properties which those categories comprise” (Stump, 2005, p. 50), in derivational morphology languages vary with respect to which modificational features they accept. Hence, compared to Czech, Slovak much more freely forms augmentatives, such as ryba ‘fish’ → ryb-isko ‘fishAUGM’ or telo ‘body’ → tel-isko ‘bodyAUGM’ (see Böhmerová, 2011, p. 21). In the transpositional category, the pre-existence of cells is also apparent. In principle, adjectives can be nominalized, as Beard and Volpe (2005, p. 198) state that “[a] transposition common to many languages is A → N, whereby 3  For a more comprehensive treatment of idiomaticity within the mutational category, see Kos (2018).

The Level of Paradigmaticity within Derivational Networks

93

gradable adjectives become abstract nouns”, an example in English being good → good-ness, verbs can be nominalized (e.g., drink → drink-ing), and dynamic adjectives can turn into adverbs (e.g., nice → nice-ly). This shared property of inflection and transposition, i.e., the fact that both are “obligatory in the sense that for each word there is a paradigm of which the cells have to be filled” (Booij, 2000, p. 363), even leads Booij to consider certain types of class-changing morphology, i.e., transposition, as part of inflection (see Booij, 2000, p. 363). One of the major distinctions between inflection and derivation is that “derivation, unlike inflection, creates words for new concepts. However, one should realize that derivation has a secondary function in that it is also used to make stylistic variation possible” (Booij, 2000, p. 360). As the latter, secondary, function belongs mainly to the sphere of the transpositional category, it is the mutational category that has the function to create words for new concepts, i.e., the function of naming in the strict sense. As such it reflects the needs of the extra-linguistic reality, unlike transposition, which is purely intra-linguistic. Even though within the mutational category we can abstract certain paradigmatic series which follow the same pattern, such as the names for butterflies běl-ásek ‘white butterfly’, modr-ásek ‘blue butterfly’, and žluť-ásek ‘yellow butterfly’, following the pattern “A(colour)+[ásek] suff. ↔ butterfly which has an A colour”, or names for fruit brandies sliv-ovice ‘plumb brandy’, třešň-ovice ‘cherry brandy’, ořech-ovice ‘nut brandy’, etc., following the pattern “N(fruit)+[ovice] suff. ↔ spirit made of N”, such paradigms are rather fragmental and their structure within the system is not clearly given. The same applies to deverbal nouns denoting the agent of the motivating verb, which is a classic example of a derivational paradigm used in discussions on paradigms in derivational morphology (see e.g., Bauer, 1997; Bonami & Strnadová, 2019). Whether a verb is associated with an agent depends on extra-linguistic factors, such as the nature of an activity as a deliberate function in humans and human-made products, the ability to do the activity, excessive manifestation of such activity in humans, or the activity as a salient feature in natural organisms. Thus, for the verb pít ‘to drink’ we can only find pij-an ‘drinker’, someone who drinks alcohol excessively, pij-ák ‘blotting paper’, paper whose purpose is to “drink” ink, or pij-avice ‘leech’, whose salient feature is “drinking” blood based on the perception of the extralinguistic reality. Also chod-ec ‘pedestrian’ (lit. walker) is not anyone who walks, but the term is used only in the frame of a “traffic context” in opposition to drivers and cyclists. A similar frame restricts the Booij’s (2000, p. 360) example read-er, as it is relevant within the context “someone who reads this book” (my emphasis); read-er may also be a client of a library or an electronic device whose purpose is to read data. In other words, not everyone who reads something automatically becomes a reader.

94

Kos

Although the series of deverbal agent nouns is much larger than other, rather fragmental, series within the mutational category, it does not still reach the level of generality (see Booij, 2000, p. 363) as do the transpositional and modificational categories, not to mention inflection; as a consequence, by far not all verbs form deverbal agent nouns. 5.4 Predictability of Forms In both the modificational and transpositional categories, there is generally no, or very limited, choice in terms of the use of a suffix; in other words, the form of the stem often determines the form of the derivative. So, in the modificational category, in nouns ending in -ák as a rule the suffix turns into -áček (e.g., škol-ák ‘school boy’ → škol-áček ‘school boyDIM’) to express the diminutive or the same suffix turns into -ačka (škol-ák ‘school boy’ → škol-ačka ‘school girl’) to express the female gender, to give but one example. In the transpositional category deadjectival nouns as a rule take the productive suffix -ost, e.g., žlut-ost ‘yellowness’. Deverbal nouns as a rule end in the suffix -í, such as chozen-í ‘walking’, even though they may coexist with suppletions, such as chůze ‘walking’. As mentioned above, the mutational category does exhibit fragmented series of lexemes in which we can find systematic co-variation of form and meaning. However, synthetic languages, such as Czech, have a much higher variety of suffixes to express the onomasiological base (i.e., the named entity) since derivation is their predominant word-formation process, and the high variety helps to avoid polysemy and homonymy. Even within the above-discussed deverbal agent nouns we can thus find synonymous suffixes, such as zpěv-ák ‘singer’, tlumoč-ník ‘interpreter’, vlád-ce ‘ruler (of a country)’, holi-č ‘barber’ (lit. shaver), učit-el ‘teacher’, lete-c ‘pilot’ (lit. flyer), etc. Also, a similar variety of suffixes attaches to nominal bases, giving rise to lexemes expressing human, animate, and inanimate referents, such as those based on the stem voda ‘water’: (5) vod-ař ‘river maintenance worker’ vod-ák ‘plumber’ or ‘canoeist’ vod-ouš ‘sandpiper’ vod-ík ‘hydrogen’ vodn-ice ‘water pipe’ or ‘turnip’ Vodn-ář ‘Aquarius’ vodn-ík ‘a mythical character living in ponds and rivers’ With the exception of vod-ík ‘hydrogen’, which belongs to a limited series of nouns expressing basic chemical elements, such as kysl-ík ‘oxygen’, dus-ík

The Level of Paradigmaticity within Derivational Networks

95

‘nitrogen’, vápn-ík ‘calcium’, etc., the choice of the suffix in the naming process of the remaining nouns seems to be rather random. This randomness is also proved by instances of names for an identical concept, where the named entity may be expressed by a number of suffixes. Compare the different Czech local names for the linnet, a small bird feeding on seeds of flax and hemp, based on the same stem, konopí ‘hemp’: konop-ka, konop-ek, konop-ník, konop-níček. The choice is also free of any gender bias as the same (non-human) concept can be assigned the masculine gender (konop-ek, konop-ník) or the feminine one (konop-ka). Moreover, the use of diminutive suffixes as a means of expression of the onomasiological base in the mutational category is rather common, for example in naming natural organisms, and the diminutive forms do not require any pre-existing non-diminutive ones. For example, modř-inka ‘blue tit’ should not be seen as a derivative of modř-ina ‘bruise’, as the form could suggest. The fact that the same suffixes can be used either in the modificational or mutational category necessarily leads to polysemy. Thus -ice in vodn-ice can be either a female equivalent of vodn-ík ‘a mythical character living in ponds and rivers’ or a term denoting a turnip or a waterpipe, the former falling under the modificational category and the latter two under the mutational category. Note that the latter two have no male equivalents. As has been suggested, the same formal equivalence can be found in the use of diminutive suffixes. Hence, modř-inka can either be a diminutive of modř-ina ‘bruise’ (modificational category) or a term for the blue tit (mutational category), the latter having no non-diminutive equivalent. The difference between the two onomasiological categories lies in the fact that the forms in the modificational category, the suffixes -ice and -inka are the only ones to express the female gender and diminutive marker for the given stems, respectively, whereas in the mutational category they are one of several possible options in naming the given concepts. This differentiation between the modificational and mutational categories seems to offer an explanation to the question of productivity of the diminutive, mentioned in ten Hacken (2015, p. 200): As I demonstrate, the diminutive in Dutch is used in many cases to name a new concept. The special meaning cannot be accounted for as the result of lexicalization with subsequent specialization, because the required regular meaning does not exist. Thus, a train ticket can be called kaartje, but there is no corresponding meaning of kaart (‘map, card’). Even this highly ‘productive’ process must therefore be treated as semiproductive in Jackendoff’s sense.

96

Kos

I suggest that ten Hacken’s question of whether the diminutive is a productive or semi-productive process is based solely on the form of the suffix. Taking into consideration the onomasiological categories, however, the “true” diminutives (at least in Czech and apparently in Dutch) within the modificational category are fully productive, thus highly paradigmatic, whereas the diminutive forms coined under the mutational category are one of several possible options used in naming a new concept. 6

Conclusion

If we accept Stump’s claim that “[m]any of the arguments that motivate the postulation of paradigms in the inflectional domain have straightforward analogues in the domain of derivation” (Stump, 2005, p. 65), we may conclude that such straightforward analogues can specifically be found within Dokulil’s modificational and transpositional categories in the following areas. As in inflection, the lexemes formed within the modificational and transpositional categories are closely related. Within the modificational one, only a marginal semantic feature is added to the meaning of the stem, as e.g., the feature of diminutiveness. In the transpositional category, what changes is the syntactic function of a word and its lexical meaning remains untouched. Also, modification and transposition are analogical to inflection in the fact that there are pre-existing cells in the relevant categories for the given language. So, e.g., in Slovak, we can in principle form augmentatives from nouns, unless this is blocked by some morphological or pragmatic aspects; at the same time, we can in principle change the syntactic category of verb into noun. The pre-existence of cells in such cases is apparent from the fact that the speakers of the language do not consider such forms as neologisms. Within these two categories, as in inflection, there is also no, or very little, choice in the form of the suffix. As in inflection, we can occasionally find instances of two competing forms, very often a productive and unproductive one, as in English burnt and burned in the formation of the past tense or cacti and cactuses in the formation of the plural, we can find such occasional oscillation of two forms in these two categories too, such as Czech nov-ota ‘newness’ and nov-ost ‘newness’, only the latter being a productive one. The existence of more than two productive forms that would express one modificational or transpositional change is hardly thinkable. On the other hand, the fundamental reason why the mutational onomasiological category does not demonstrate features of canonical paradigmaticity is

The Level of Paradigmaticity within Derivational Networks

97

the substantial shift in meaning between the original stem and the resulting derivative. This change of meaning is the basic principle of mutation, in which “the phenomenon of one conceptual category is named on the basis of a different phenomenon of the same or a different conceptual category” (Dokulil 1982, p. 259, my translation). Further, the naming units formed within the mutational category in principle reflect the naming needs of the language user, so their existence is dependent on naming the extra-linguistic reality, so, as a consequence, the paradigmatic series within the mutational category are much more fragmented and the feature of generality assigned to inflection does not apply. Finally, as the onomasiological base within the mutational category may be expressed by a suffix or a free morpheme, and as, for example, in Czech or Slovak, due to their morphological type, the onomasiological base is predominantly expressed by a suffix, there is a high level of synonymy within suffixes. The decision of which suffix will be used to express the onomasiological base to a high degree depends on the decision of the coiner of the new naming unit, which significantly lowers the predictability of form. Note the fact that, for example, English, in keeping with its morphological type, in the mutational category much more frequently expresses the onomasiological base (the named entity) by a free morpheme, this leading to compounding; these English compounds are thus analogical to Czech or Slovak suffixal derivatives. However, if we accept that “[i]t may be assumed that the concept of the derivational paradigm is restricted to affixation in terms of word-formation processes participating in its constitution” (Štekauer, 2014, p. 369), we would also have to accept the fact that in analogical formations, one having a suffix and the other a compounding base as its onomasiological base, as e.g., the Slovak školn-ík ‘school janitor’ (lit. school + suffix) and the English school janitor, the former is based on a paradigm whereas the latter is not. In my opinion, both naming units demonstrate the same reduced level of semantic transparency between the onomasiological base and the mark and the same reduced level of predictability of form, which expresses the onomasiological base. Štekauer (2014, p. 369) arrives at the following conclusion: Conceived broadly, the idea of a derivational paradigm, constituted by all word-formation processes, is paradigmatically vacuous because it does not lead to a predictable and regularly organized system of complex words. A system of complex words produced by all the word-formation processes and related by a common stem is an open system, different from the closed system of paradigm.

98

Kos

In my opinion this approach is based purely on the form of lexemes as is the concept of derivational network. To a certain extent, the formation of all naming units is based on the abstraction of existing series of lexemes. However, in the discussion of paradigms within derivational morphology and to what extent they correlate with those in inflectional morphology, we need to take into account the notion of Dokulil’s onomasiological categories. The modificational and transpositional categories demonstrate the same or a very similar level of predictability and regularity in contrast to the much more fragmented and less clearly structured system of paradigms in the mutational category. As all naming units in derivational networks are related by a common stem only, they lack a systematic co-variation of form and meaning, so in the strict sense they are not paradigms. References Bauer, L. (1997). Derivational paradigms. In G. Booij & J. van Marle (Eds.), Yearbook of Morphology 1996 (pp. 243–256). Dordrecht: Kluwer. DOI: 10.1007/978-94-017-3718-0_13 [Accessed 25 August, 2019]. Beard, R., & Volpe, M. (2005). Lexeme-morpheme base morphology. In R. Lieber & P. Štekauer (Eds.), Handbook of word-formation (pp. 189–205). Dordrecht: Springer. DOI: 10.1007/1-4020-3596-9_8 [Accessed 31 August, 2019]. Bisetto, A., & Scalise, S. (2005). The classification of compounds. Lingue e Linguaggio, 4(2), pp. 319–332. Böhmerová, A. (2011). Suffixal diminutives and augmentatives in Slovak. Lexis[Online], 6, pp. 1–29. DOI: 10.4000/lexis.429 [Accessed 31 August, 2019]. Bonami, O., & Strnadová, J. (2019). Paradigm structure and predictability in derivational morphology. Morphology, 1–31. DOI: 10.1007/s11525-018-9322-6 [Accessed 31 August, 2019]. Booij, G. (2000). Inflection and derivation. In G. Booij, C. Lehman, & J. Mugdan (Eds.), Morphology. An International Handbook on Inflection and Word-Formation, (Vol. 1, pp. 360–369). Berlin: Walter de Gruyter. Dokulil, M. (1962). Tvoření slov v češtině 1, Teorie odvozování slov. Prague: Nakladatelství Československé akademie věd. Dokulil, M. (1982). K otázce slovnědruhových převodů a přechodů, zvl. transpozice. Slovo a slovesnost, 43(4), pp. 257–271. Dokulil, M. (1986). III Tvoření slov. In J. Petr (Ed.), Mluvnice češtiny 1 (pp. 191–526). Prague: Academia. Furdík, J. (2004). Slovenská slovotvorba (Ed. Ološtiak, M.). Prešov: Náuka.

The Level of Paradigmaticity within Derivational Networks

99

Horecký, J., Buzássyová, K., & Bosák, J. (1989). Dynamika slovnej zásoby súčasnej slovenčiny. Bratislava: Veda. Kos, P. (2018). O idiomatickém charakteru pojmenování v mutační kategorii. Časopis pro Moderní Filologii 100(1), pp. 9–23. Retrieved from: https://dspace.cuni.cz/ bitstream/handle/20.500.11956/100488/Petr_Kos_9-23.pdf?sequence=1&isAllowed=y [Accessed 31 August, 2019]. Štekauer, P. (2014). Derivational paradigms. In R. Lieber & P. Šteakauer (Eds.), The Oxford handbook of derivational morphology (pp. 354–369). Oxford: Oxford University Press. DOI: 10.1093/oxfordhb/9780199641642.013.0020 [Accessed 25 August, 2019]. Stump, G. (2005). Word-formation and inflectional morphology. In P. Štekauer & R. Lieber (Eds.), Handbook of word-formation (pp. 49–71). Dordrecht: Springer. DOI: 10.1007/1-4020-3596-9_3 [Accessed 31 August, 2019]. ten Hacken, P. (2015). Transposition and the limits of word formation. In L. Bauer, L. Körtvélyessy, & P. Štekauer (Eds.), Semantics of complex words (pp. 187–216). Cham: Springer International Publishing. DOI: 10.1007/978-3-319-14102-2_10 [Accessed 31 August, 2019].

Chapter 5

Doublet Formation in Palestinian Arabic—Where Do Paradigms Interfere? Lior Laks and Faten Yousef 1

Introduction

This study examines the correlation between paradigms and doublet formation in the verbal system of Palestinian Arabic (hereafter PA). Semitic morphology relies highly on non-concatenative morphology, where words are formed in patterns. The pattern consists of a prosodic structure, a vocalic pattern and affixes. For example, the verbs šireb ‘drink’, nizel ‘go down’ and wisˁel ‘arrive’ belong to the CiCeC pattern as they all share the vocalic pattern i-e and the difference in their meaning is based on the different stem consonants. We examine cases where the same stem consonants surface in two patterns with the same meaning, as demonstrated in (1). (1) a. rijli wirmat wu-alam šadi:d tˁabʕan, ma ruħt la-l-mustašfa ‘My leg became swollen, and great pain of course, I didn’t go to the hospital’ (http://www.alhilalclub.com/vb/archive/index.php/t-223078.html) b. wu-l-yo:m twarramat rijli wu-ruħt la-l-mustašfa ‘and today my leg became swollen and I went to the hospital’ (http://www.66n.com/forums/showthread.php?p=2801899) These examples in (1) consist of the third person feminine past form of the verbs wirem (1a) and twarram (1b). Both share the w-r-m stem consonants and denote ‘become swollen’, but are formed in different patterns: CiCeC and tCaCCaC. The two verbs share the same meaning and can be used in the same semantic and syntactic contexts. Examining more such cases of variation reveals that the change is mostly from the CiCeC pattern into other patterns, mostly tCaCCaC. In contrast, there are CiCeC verbs that do not have doublets in tCaCCaC. For example, riʕeb ‘become scared’ does not have a doublet like *traʕʕab, which could potentially be formed. Why does this change occur and what blocks it? We will argue that morphological change is highly motivated

© Koninklijke Brill NV, Leiden, 2020 | doi:10.1163/9789004433410_006

Doublet Formation in Palestinian Arabic

101

by transparency in paradigmatic relations between verb forms. The change takes place in order to achieve both structural and semantic transparency in inflectional and derivational paradigms. This paper is structured as follows. In §2, we provide background on the verbal system of PA and the relations between verb forms in inflectional and derivational paradigms. §3 provides background on variation and doublet formation. In §4, we account for the factors that trigger doublet formation, as well as for the factors that block it. In §4.1, we show that CiCeC verbs are formed in additional patterns because of the lack of structural transparency in their inflectional paradigms. In §3.2, we examine the derivational relations between the patterns and their semantic-syntactic features. We show that some CiCeC verbs change their pattern in order to be morphologically marked as intransitive verbs, yielding greater semantic transparency within the verbal forms. In addition, we show that morphological change is more likely to occur in cases where a verb is part of a derivational paradigm. §4 consists of concluding remarks. 2

Variation and Doublet Formation

Linguistic variation is inherent in human languages and is crucial to the study of the language faculty. The same speaker can use different forms to express the same basic meaning, and different speakers can express the same meaning using different forms. Linguistic variation is a situation where multiple forms are or can be used to express a single meaning. Specifically to morphology, variation is also known as “overabundance” or “polymorphy”, where a cell within a paradigm can be filled by more than one form (Anttila, 1997, 2007). The forms that fill the same cell are labeled “doublets” (see for example, Aronoff, 1976; Malkiel, 1977; Kroch, 1989; Taylor, 1994; Acquaviva, 2008; Embick, 2008; Corbett, 2010; Dal & Namer, 2010; Mörth & Dressler, 2014; Fradin, 2016) or “cell-mates” (Thornton, 2011, 2012a, 2012b). For example, the past form of the English verb learn can be realized by the doublets learned and learnt. Linguistic variation often results from a change that languages undergo at some point in time, and once it occurs, it can be maintained as part of the grammatical mechanism. The issue of linguistic variation and change has been addressed by linguists since the 19th century, following Neogrammarian accounts of sound change (Bloomfield, 1933; Hinskens, van Hout, & Wetzels, 1997). Various studies have associated variation and change with the speaker’s competence and considered variation an integral part of natural languages (see Kiparsky, 1968, 1995; Wang, 1969; Vincent, 1980; Reynolds, 1994; Guy, 1997; Booij, 2002;

102

Laks and Yousef

Maiden, 2004; Albright, 2005; Wedel, 2009; among others). Linguistic variation and change pose a challenge for any linguistic theory that aims to provide a synchronic analysis of linguistic knowledge. A change is by definition a diachronic process, and since intermediate grammars are not final, the status of change and variation is obscure in a deterministic model, where there is no room for random changes (Adam, 2002). Consequently, accounting for the criteria that trigger processes of change has been regarded as one of the important goals of linguistic theories (see, for example, Halle, 1962). The study of linguistic changes from a synchronic point of view can contribute to linguistic theory by providing a unique perspective on the properties and factors involved in a particular grammatical phenomenon and of the interrelations between them (Macken, 1992). A synchronically motivated analysis of change makes it possible to understand the current process that the grammar undergoes and provides an insight to the factors that motivate it. Variation has also been addressed with regard to paradigmatic relations, where a canonical paradigm is expected to demonstrate uniqueness of realization, such that for every stem in the paradigm, each cell must be filled in a unique way (Carstairs, 1987; Stump, 2001, 2010; Corbett, 2007, 2010; Stump & Finkel 2013, among others). Deviations from such canonical paradigms are represented by variation, where a cell is occupied by two or more synonymous forms (see Thornton, 2011, 2012a). This study will highlight the central role of both inflectional and derivational paradigms in morphological change. Variation is also associated with competition for grammaticality and use under certain approaches. On these views, the grammar generates numerous structures or words that express the same meaning and includes a mechanism for selecting one winner, marking the rest as ungrammatical (Embick, 2008). This means that if one variant is employed, another is not (see also Rainer, 1988; Joseph, 1998; Plag, 2000; Bauer, 2006; Kiparsky, 2010). This in turn leads naturally to the idea that distinct variants are competing with one another in the grammar (Weinreich, Labov, & Herzog, 1968; Pintzuk, 1991; Yang, 2002). Nevertheless, in some cases more than one competitor is selected as grammatical, with these variants in competition for interactional use. Thornton’s (2011, 2012a) case studies of cell-mates refers to cases of competing inflectional forms that demonstrate complete synonymy. This is typical mainly for inflection, as inflectional categories are determined by the morpho-syntax (Stump, 2016; Aronoff, 2019) and rival inflectional patterns are unable to differentiate themselves in their denotation. The picture is different with respect to competition between derivational processes and patterns. Aronoff (2016, 2019) shows that competing derivational affixes are less likely to be in real competition. Since there are no necessary paradigm cells to fill in derivation, in contrast

Doublet Formation in Palestinian Arabic

103

to inflection, it is impossible to count how many inputs or outputs there are for a given morphological process. Furthermore, semantic and pragmatic factors allow competing derivational patterns to differentiate themselves in many ways, where each pattern can have its own niche. Aronoff shows that a great deal of similarity exists between competing derivational affixes and ecological niche differentiation, in which natural selection drives competing species into different distribution patterns of resource use. The paradigmatic approach has been gaining a growing stand in derivational morphology, in addition to its well-established role in inflection. Many studies demonstrate the importance of paradigms in word formation (see Bauer, 1997; Booij, 1997, 2008; Pounder, 2000; Beecher, 2004; Booij & Lieber, 2004; Hathout & Namer, 2014; Štekauer, 2014; Blevins, 2016). The present analysis considers morphological variation and doublet formation within PA verbs. It will demonstrate that such change, as well as lack thereof, is not random, and can be predicted based on systematic guidelines, so that it can be expressed within a model of the speaker’s knowledge. Specifically, we will demonstrate the role of both inflectional and derivational paradigms in the likelihood of such morphological change to occur. In the next section, we first provide some background on the PA verbal system. 3

Verb Formation in Palestinian Arabic

Semitic morphology relies heavily on non-concatenative morphology, where each verb has a consonantal root and a vocalic pattern. The pattern determines the phonological shape of the verb, i.e., its vowels, prosodic structure and affixes (if any). The phonological shape of verbs, unlike that of nouns and adjectives, is essential for determining the shape of the other forms in the inflectional paradigm. Every verb must conform to one of the patterns (Ornan, 1971, 2003; Rosén, 1977; Berman, 1978, 2003; Bolozky, 1978, 1999; McCarthy, 1981; Schwarzwald, 1981, 2000; Ravid, 1990, 2003; Benmamoun, 1991, 2003; Aronoff, 1994, 2019; Bat-El, 1994, 2011; Ussishkin, 1999, 2005; among others). PA has ten verbal patterns which are presented in (2).1 1  The examples in this study are in their past form, which is the citation form, conventionally assumed to be the base of formation throughout the inflectional paradigm, as it is free of inflectional suffixes (see Ussishkin, 1999 and Bat-El, 2002, among others). However, the direction of derivation is irrelevant for the purposes of our analysis). In addition, there is dialectical variation with respect to the vowels of CiCeC verbs and in some cases, with respect to the length of the second vowel. We will use the i-e vowels throughout the paper.

104

Laks and Yousef

(2) Pattern Example CiCeC/ CaCaC šireb ‘drink’ CaCCaC tˁallaʕ ‘take out’ Ca:CaC sa:ʕad ‘help’ aCCaC arsal ‘send’ tCaCCaC tfarraq ‘scatter’ tCa:CaC tka:tab ‘correspond’ inCaCaC inhadam ‘be destroyed’ iCtaCaC intabah ‘pay attention’ iCCaCC izraqq ‘become blue’ istaCCaC istaxdam ‘use’ When examining the PA verbal system, it is important to note that two paradigmatic relations are involved: inflectional and derivational. Each pattern has inflectional paradigms of tense/aspect, and each such paradigm consists of conjugated forms according to gender, number and person.2 Patterns are, therefore, regarded as inflectional classes (see Aronoff, 1994), as each pattern has its own inflectional paradigms. Such paradigms share some properties but they are also distinct from each other in some respects. For example, the formation of imperfective forms in all patterns is based on prefixation, but the vowels of the prefixes are different to some extent. The relations between verbs in different patterns are derivational and are manifested mainly in transitivity alternations and other types of semantic relations (see Bolozky & Saad, 1983; Fassi Fehri, 1994; Holes, 1995; Goldenberg, 1998; Younes, 2000; Hallman, 2002, 2006; Watson, 2002; Ryding, 2005; Abu Chacra, 2007; Henkin, 2009, 2010; Izre’el, 2010; Glanville, 2011; Shawarbah, 2012; Ouhalla, 2014; among others). The formation of the same stem consonants in different patterns results in two (or more) different verbs, where the semantic relations between them can be of different degrees of transparency. Verbs that are formed in certain patterns share some typical semantic and syntactic features. For example, verbs in CaCCaC are usually active transitive verbs (e.g., ɣassal ‘wash’), and their intransitive alternates are formed in tCaCCaC (e.g., the reflexive verb tɣassal ‘wash oneself’). As shown, the two verbs are formed in two patterns and they both consist of the consonants ɣ-s-l. CiCeC/ CaCaC is considered the most basic pattern that hosts many verbs, which are

2  The participle tense paradigm does not distinguish for person, but has only four forms that differ in gender and number.

Doublet Formation in Palestinian Arabic

105

basic vocabulary items in Arabic, PA, as well as in many other dialects.3 It does not have specific semantic-syntactic features as it hosts active transitive verbs like katab ‘write’, rikeb ‘ride’, and šireb ‘drink’, active intransitive verbs like miše ‘walk’ and rakadˁ ‘run’, and intransitive inchoative verbs that denote a change of state, e.g., jimed ‘freeze’ and kiber ‘grow’, as well as other types of verbs. Some patterns tend to have a ‘usual mate’, namely another pattern, or several patterns, with which they have typical derivational relations. Such types of relations are demonstrated in (3). CiCeC/ CaCaC is related to CaCCaC in two ways. In (3a) the CaCCaC verb is the transitive causative counterpart of the CiCeC intransitive inchoative verb, while in (3b) the CaCaC verb is transitive and the CaCCaC verb is also transitive and denotes an intensification of the action of the verb. In (3c) the same CaCaC verb as in (3b) is in relation with an inCaCaC, which is inchoative. In (3d) the transitive-intransitive alternation is between CaCCaC (transitive) and tCaCCaC (intransitive). (3) Patterns Verb1 Verb2 a. CiCeC—CaCCaC jimed ‘freeze (intransitive)’ jammad ‘freeze X’ b. CaCaC—CaCCaC kasar ‘break’ kassar ‘break into small pieces’ c. CaCaC—inCaCaC kasar ‘break’ inkasar ‘become broken’ d. CaCCaC—tCaCCaC qallasˁ ‘shrink X’ tqallasˁ ‘shrink (intransitive)’ The examples in (3) demonstrate that CaCCaC is typically active and transitive, inCaCaC and tCaCCaC are typically intransitive and CiCeC/CaCaC is neutral with respect to transitivity. In addition, CiCeC can be derivationally related to CaCCaC and inCaCaC, but not to tCaCCaC, while CaCCaC is derivationally related to CiCeC/CaCaC and tCaCCaC but not to inCaCaC. The picture that emerges is that the verbal patterns are organized in terms of families, where some patterns are more closely related to others (see Berman, 1978, 2003; Bolozky 1978, 1999; Schwarzwald, 1981, 2002; Ravid, 1990, 2008; Doron, 2003 for Hebrew). It is crucial to note that such distinctions reflect strong tendencies rather than a clear-cut distribution. The derivational relations between verbs in different 3  CiCeC and CaCaC are considered the same pattern as they differ only in their vowels and they share the same inflectional paradigms. CiCeC is more typical for intransitive inchoative verbs, while CaCaC hosts other types of verbs, but as shown in the paper, this is only a tendency. Since most of the examples discussed in the paper are CiCeC~tCaCCaC doublets, we will mostly name the pattern CiCeC and not CaCaC.

106

Laks and Yousef

patterns are subject to irregularities to a great extent. For example, the verb tfarraj ‘watch’ is formed in tCaCCaC, but is active and transitive and does not fall into one of the typical semantic classes of this pattern (e.g., inchoative and reflexive). In general, the relations between verbs in different patterns share features that are more typical of derivation rather than inflection and, therefore, the pattern system is considered a system of derivational relations, where each pattern has its own inflectional classes. This study examines another type of relations between PA verbal patterns, where the same stem consonants surface in two patterns with no difference in their meaning. We now turn to the description of this phenomenon and account for the factors that are responsible for it, as well as the factors that block it. 4

Doublet Formation in Palestinian Arabic Verbs

The study addresses cases of morphological doublets in the verbal system, where two verbs share the same stem consonants and are formed in two different patterns. The two verbs share the same meaning and can be used in similar and even identical contexts. Similarly to the examples in (1), this is demonstrated for the verbs xireb (CiCeC) and txarrab (tCaCCaC) in their third person feminine past form. They both denote ‘become ruined’ and as shown in (4), they occur in similar sentences that were found on the web. (4) a. il-balad xirbat ‘The country became ruined’ (https://www.google.co.il/#q=%D8%AE%D8%B1%D8%A8%D8%AA+ %D8%B9%D9%8A%D8%B4%D8%AA%D9%86%D8%A7&start=20) b. il-balad txarrabat ‘The country became ruined’ (https://www.google.co.il/#q=%D8%AA%D8%AE%D8%B1%D8%A8% D8%AA+%D8%A7%D9%84%D8%A8%D9%84%D8%AF&start=30) CiCeC~CaCCaC variation is even attested in rare cases where the tCaCCaC form is dedicated to another meaning. As shown in (5), the third person feminine past forms of the verbs kiber (5a) and tkabbar (5b) are used to denote ‘become big’. kiber is more frequent in its usage for such a meaning, but such cases where tkabbar is also used to denote this meaning are also attested. This is so despite the fact that tkabbar has another meaning of ‘become arrogant/patronize’.

Doublet Formation in Palestinian Arabic

107

This shows that even in such cases some speakers show a tendency to use it with the same meaning as the CiCeC verb. (5) a. jru:ħi kibrat ‘My injuries became big’ (https://www.facebook.com/search/results.php?q=%D8%AC%D8% B1%D9%88%D8%AD%D9%8A+%D9%83%D8%A8%D8%B1%D8% AA&init=public)

b. jru:ħi tkabbarat ‘My injuries became big’ (https://twitter.com/smoal2merah)

More examples of doublet formation are presented in (6). (6) CiCeC wirem nišef sixen qiwi: sˁiɣer

tCaCCaC twarram tnaššaf tsaxxan tqawwa tsˁaɣɣar

‘become swollen’ ‘become dry’ ‘warm up’ ‘become strong’ ‘become small’

This analysis addresses three main questions with respect to doublet formation. First, which verbs are likely to change their pattern, and for what reasons (for example, why are these mostly CiCeC verbs and not verbs that are formed in other patterns?), with the goal of identifying the type of verbs that are more prone to undergo morphological variation than others. Second, which pattern is selected as the additional form of verbs that undergo change (for example, why a verb like qiwi: ‘become strong’ changes into tqawwa in tCaCCaC and not into another pattern like inCaCaC to yield *inqawa)—in order to demonstrate that the selection of the alternative pattern is governed by morphophonological and semantic-syntactic factors and the interaction between them. Third, why are there verbs that look like potential candidates for doublet formation but that do not undergo this process (i.e., why are there CiCeC verbs with no doublets in tCaCCaC)?, aiming to show that blocking of doublet formation is highly related to paradigmatic relations. While addressing these three questions, we will show that doublet formation can be predicted to a large extent and demonstrate the important role of paradigms in word formation and morphological change. Before we discuss the two main types of

108

Laks and Yousef

factors that play a role in doublet formation, some explanations regarding data collection are in order. 4.1 Data Collection Data collection is based on two sources: (i) Two dictionaries of PA, Elihay (2005) and Lubani (2006), as well as books and papers on PA (e.g., Levine, 1995; Jastrow, 2004); (ii) Google-based word searches for forms that are expected to have (and not to have) doublets. With regard to the second source of data, it is important to mention the pitfalls of using online web-searches. It is not always possible to know whether the online examples were produced by native speakers. This is even more crucial in Arabic in light of the issue of diglossia and the various dialects. It is not always possible to know whether the examples were produced by PA speakers or were written in Modern Standard Arabic. In the light of these methodological issues, searches are based on websites like blogs, chats, Facebook and Twitter, where speakers tend to write in colloquial Arabic and not Modern Standard Arabic (Warschauer, El-Said, & Zohry, 2002; Palfreyman & Al-Khalil, 2007; Abu Elhija 2012). In addition, searches were conducted on websites that are typically used by PA speakers and the examples were checked by native PA speakers. Using web-searches also raises the question to what extent the examples represent actual usage. How many examples should be found in order to determine that doublets indeed exist? While there is obviously no absolute answer to this question, it is important to note that the analysis is not based on the number of instances that are found, but mainly on the contrast between what is found, to different extent, and what is not found at all. Let us demonstrate this by comparing the CiCeC verbs wirem ‘become swollen’ and nije ‘be saved’. A web-search reveals that wirem has a tCaCCaC doublet (twarram), as shown in (1) in the Introduction. In contrast, the search reveals that niji does not have a tCaCCaC doublet (*tnajja). The fact that even a small amount of instances of twarram were found, in comparison to no examples of tnajja, supports the claim that some verbs undergo variation and others are less likely to do so. To conclude, we are well aware of the problems that this methodology has, but this seems to be the optimal way of revealing cases of doublets. Relying on a spoken corpus can also provide relevant instances, but the chances of finding enough examples are rather low. Using the web for morphological research has its own limits and morphologists should be cautious when relying on it (Lüdeling, Evert, & Baroni, 2007), but despite these limitations, studies have revealed several devices in which web content becomes a major asset in morphological analysis (Dal & Namer, 2015). We turn to examining the factors that play a role in doublet formation.

Doublet Formation in Palestinian Arabic

109

4.2 Inflectional Paradigms—towards Structural Transparency CiCeC verbs demonstrate relatively low structural transparency within their inflectional paradigms. This is manifested in both prosodic and segmental alternations. Verbs in this pattern are formed in tCaCCaC where no such alternations occur. 4.2.1 Reducing Prosodic Alternation Examining cases of doublet formation reveals that the change applies mainly to CiCeC verbs. Why is it so? The CiCeC pattern is considered prosodically more marked as its prosodic structure alternates within its inflectional paradigms (see Schwarzwald, 1996, 2008 and Bat-El, 2002 for similar cases in the verbal patterns of Hebrew). As mentioned in §2, each pattern has inflectional paradigms of tense, and each such paradigm consists of conjugated forms according to gender, number and person. The relations between the tense paradigms in the CiCeC pattern are structurally less transparent in comparison to the relations between such paradigms in other patterns. In contrast to the rest of the patterns, it does not preserve its syllabic structure throughout its inflectional paradigms. Examine, first, the prosodic structure of some of the other patterns in (7).4 As shown, the same syllabic structure of the stem remains intact throughout the entire paradigm. For example, all the inflectional forms of CaCCaC verbs share a CVCCVC stem and all the inflectional forms of aCCaC share a CCVC stem. This uniformity of the syllabic structure makes the morphology of these patterns highly transparent, as the transition from one tense to the other involves no change in the internal prosodic structure. The only changes that occur are the addition of a prefix and in some cases a change in some of the vowels (e.g., rattab—yiratteb ‘arrange’ in the CaCCaC paradigm).5 (7) CaCCaC Ca:CaC aCCaC tCaCCaC inCaCaC istaCCaC Past rattab sa:far akram tʕallam inkasar istaʕmal Participle mratteb msa:fer mikrem mitʕallem minikser mistaʕmel Imperfective yeratteb yesa:fer yikrem yitʕallam yinikser yistaʕmel ‘arrange’ ‘travel’ ‘respect’ ‘study’ ‘break’ ‘use’ 4  We present here the paradigms of some of the patterns for space consideration. The uniformity of the prosodic structure exists is all patterns apart from CiCeC/CaCaC. 5  The tense/aspect system of the PA verbal system (as well as other dialects) is more complex in the sense that different verb forms can denote different tenses or aspects (see for example, Levin, 1995; Holes, 2004; and Jastrow, 2004; among others). This is irrelevant for the current study and therefore we will use the simplified terms ‘past’, ‘participle’ and ‘imperfective’ in order to relate to the three morphological forms in the paradigm.

110

Laks and Yousef

This uniformity and transparency of prosodic structure do not exist in the inflectional paradigms of CiCeC/CaCaC. As shown in (8), the syllabic structure of the past and participle forms is different from those of the imperfective forms. The past and participle forms share a CVCVC/CV:CVC structure with no consonant cluster (šireb, ša:reb), while the imperfective forms share a CCVC structure preceded by a prefix, where a consonant cluster emerges (yišrab).6 (8) CiCeC CaCaC Past šireb katab Participle šare:b ka:teb Imperfective yišrab yikteb ‘drink’ ‘write’ The prosodic alternation makes CiCeC/CaCaC structurally more complex and less transparent than other patterns. This results in a phonological load expressed by prosodic shifting, that is, a change in the prosodic structure throughout the inflectional paradigm (Bat-El, 2002), so that CiCeC/CaCaC can be defined as the paradigmatically most marked patterns in terms of prosodic structure. The lack of structural transparency due to prosodic alternation is also responsible for the low productivity of CiCeC/CaCaC in the formation of new verbs. A recent study of verb innovation by Laks (forthcoming) shows that there are mainly two patterns, CaCCaC and tCaCCaC, which are active in the formation of new verbs. Verbs that are formed based on existing words, either native or non-native, are formed in one of these patterns. Such verbs can be formed based on either existing PA word or based on loan words. For example, the verb massaj ‘send a message’ is formed in CaCCaC based on the English word message. CaCCaC and tCaCCaC are selected also due to the fact that they are the only patterns that can host verbs with more than three stem consonants (9). Other patterns do not host such verbs since their prosodic structure does not allow them to do so.7

6  Some forms in the past inflectional paradigm do consist of a cluster. This happens due to vowel deletion when a suffix is attached to the stem (e.g., šireb-et → šribet ‘I/You drank’). However, the base of the past form that is free of affixes does not consist of a consonant cluster. In this study, we relate only to the bare stems in each tense, prior to their inflection. 7  We do not distinguish between CaCCaC and tCaCCaC in this section since the division of labor between them is based on semantic-syntactic criteria, to be discussed in §4.3.

Doublet Formation in Palestinian Arabic

111

(9) Quadrilateral CaCCaC and tCaCCaC PA verbs Base Derived Verb format ‘format’ farmat ‘format’ kænsel ‘cancel’ kansal ‘cancel’ servis ‘service’ sarvas ‘provide service’ biznes ‘business’ tbaznas ‘do business with each other’ bijama ‘pajamas’ tbajjam ‘out on pajamas’ Based on the prosodic properties of the patterns discussed above, the CaCCaC and tCaCCaC paradigms are therefore regarded as the default derivational paradigms for the formation of new verbs. We will return to this point in the discussion of the pattern that is selected as a doublet in §4.3. Turning back to the case of doublet formation, the claim made here is that CiCeC verbs tend to receive an additional form in order to avoid prosodic alterations within the inflectional paradigm. This is demonstrated below for the doublets nišef and tnaššaf ‘become dry’ (10). The verb forms in the paradigm of nišef demonstrate prosodic alternation. The stem consonants n and š are either separated by a vowel or surface as a cluster. In contrast, the forms in the tnaššaf  keep the same syllabic structure intact, resulting in structural transparency in the inflectional paradigm. The following shows the verb ‘become dry’ in different patterns: (10) Tense CiCeC Past nišef Participle na:šef Imperfective yinšaf

tCaCCaC tnaššaf mittnaššef yetnaššef

4.2.2 Reducing Segmental Alternation Within CiCeC verbs, there is a group of verbs, defined as weak verbs, that demonstrates even a stronger tendency for doublet formation. Weak verbs, which usually have one of the consonants y or w as a stem consonant, exhibit segmental alternation in their paradigms, making them morphologically defective (see Chekayri & Scheer, 2003; Holes, 2004; Faust, 2016, 2017 for Arabic and Schwarzwald, 1977, 1984; Sumner, 2003; Bat-El, 2005 for Hebrew). This is demonstrated for the verb wiseʕ ‘become wide’ (11a) that alternates with twassaʕ (11b). Again, both verbs can be used in very similar contexts. (11) a. wesʕat meʕedti wu-sˁirt ajwaʕ ‘My stomach became wide and I became hungrier’ (http://www.vb.eqla3.com/archive/index.php/t-352623-p-5.html)

112

Laks and Yousef

b. maʕqu:l twassaʕat meʕedti? ‘Is it possible that my stomach became wide?’ (http://rashagh.com/forum/t18559.html) Why are such verbs considered weak or defective? This defectiveness stems from historical phonological processes (e.g., diphthong contraction) that are no longer productive in verb formation and thus do not constitute an active part of the morpho-phonological knowledge of speakers. Consider, for example, the pair of CiCeC verbs nišef ‘become dry’ and wiseʕ ‘become wide’ and their inflectional paradigms in (12). Both verbs are formed in CiCeC, but while nišef is a regular verb with the three stem consonants n-š-f that surface throughout the paradigm, wiseʕ is a weak verb, since its initial stem consonant w does not surface in the imperfective form.8 (12) Tense Past Participle Imperfective

Regular paradigm nišef na:šef yinšaf

Irregular paradigm wiseʕ wa:seʕ yu:saʕ (*yiwseʕ)

Weak verbs like wiseʕ ‘become wide’ exist in PA and their inflectional paradigms remain intact together with their defectiveness with regard to one or more of the stem consonants. However, few such new verbs enter the language, indicating that phonological alternations like these are no longer an active process in PA. Such defective paradigms are frozen in the sense that they are stored as irregularities in the lexicon. Had they been a part of the active grammar, it would be expected from the morphological component of the grammar to form more such verbs productively. Since the morpho-phonology of such weak verbs is not an active part of the grammar, many such verbs change into regular forms in tCaCCaC as well. tCaCCaC allows all stem consonants to surface throughout the entire paradigm, as no phonological alternations exist as in CiCeC.9 Compare again the doublets wiseʕ and twassaʕ, both denoting ‘become wide’ (13). The stem consonant w is missing in the imperfective form of CiCeC, while it surfaces in all the conjugated forms of tCaCCaC. Note that both nišef and wiseʕ (12) have doublets in 8  There are also different vowels in the imperfective forms of the verbs, but this is irrelevant for the purposes of this study. 9  In case the last stem consonant is a y or w, it does not surface in all tenses of the paradigm, regardless of the pattern. The cases that are discussed here are the ones in which the first or second stem consonant is y or w.

Doublet Formation in Palestinian Arabic

113

tCaCCaC, due to the prosodic alternation discussed in 4.2.1, but the study reveals that there are more cases of doublet formation for weak verbs like wiseʕ, because of the segmental alternation that does not occur for verbs like nišef. (13) shows the verb ‘become wide’ in different patterns: (13) Tense Past Participle Imperfective

CiCeC tCaCCaC wiseʕ twassaʕ wa:seʕ mitwasseʕ yusa:ʕ yetwasseʕ

Doublet formation results in paradigm uniformity (Steriade, 1988) in the verbal system. There is no longer alternation of the stem consonants within the inflectional paradigm of verbs that changes their pattern. In (14), all the CiCeC verbs in the left column demonstrate defectiveness in at least one form of their inflectional paradigms, whereas the tCaCCaC forms in the right column show no such consonant segmental alternation. (14) CiCeC wirem xa:f qiwe yibes tˁiwel

tCaCCaC twarram txawwaf tqawwa tyabbas ttˁawwal

‘become swollen’ ‘become scared’ ‘become strong’ ‘become dry’ ‘become long’

The fact that defective forms are morphologically neutralized and paradigmatically aligned with other forms provides further support for the role of paradigmatic relations in word formation (van Marle, 1985; Spencer, 1988, 1991; Corbin, 1989; Anderson, 1992; Booij, 1996, 2008; Burzio, 1998, 2002; Steriade, 2000; Dal & Namer 2010; Thornton, 2012a, 2012b, among others). That is, the morphological mechanism takes into account information not only about the actual verb that is formed but also about its inflection paradigm, avoiding paradigms that include alternation in the consonants. Specifically, such examples demonstrate the role of irregularity in morphological change (Maiden, 2004; Maiden & O’Neill, 2010; Stump, 2010). Avoidance of consonant alternation also shows that the grammar avoids complexity as much as possible. The morphological mechanism of PA blocks forms that are more complex in the sense that they involve irregularity and have to be memorized. This is related to a more general question, whether or not the grammar contains a general principle regulating how words and

114

Laks and Yousef

phrases interact with one another (Embick & Marantz, 2008). Selecting defective forms based on paradigmatic irregularity and inactive morphological rules gives rise to complexity in the morphological mechanism of the grammar. Consequently, such forms are avoided, both in the construction of new verbs and in changing existing defective forms into non-defective forms. Specifically, within CiCeC verbs, weak verbs are more prone to have doublets in tCaCCaC, because in addition to the prosodic alternation that exists in CiCeC paradigms in general, weak verb also demonstrate segmental alternation. Doublet formation in tCaCCaC avoids both types of alternations. So far we have answered the first question of the study, namely, why is it CiCeC verbs that have doublets? CiCeC verbs, in contrast to other patterns, tend to receive an additional form in another pattern in order to achieve structural transparency throughout their inflectional paradigms. It is clear, then, why the change occurs. Two more questions remain: (i) why is the change into tCaCCaC out of all patterns? and (ii) why are there other CiCeC verbs with no doublets? In order to answer these questions, we turn to the semantic-syntactic features of PA patterns and the derivational relations between them. Derivational Paradigms and the Semantic-Syntactic Status of PA Patterns 4.3.1 Pattern Selection in Doublet Formation As shown in §3, the relations between PA patterns are manifested mainly in terms of transitivity alternations, where some patterns are typically transitive and others are typically intransitive. Specifically, CaCCaC tends to host transitive verbs and tCaCCaC is used for intransitive verbs, mainly inchoatives. This division of labor between the two patterns is attested even more strongly with respect to new verb formations. Verbs that are active verbs, mainly transitive ones, are formed in CaCCaC (15). For example, the verb fannaš is based on the English verb finish. 4.3

(15) Base sayn ‘sign’ formæt ‘format’ iks ‘X’ milћ ‘salt’

Derived verb sayyan ‘sign’ farmat ‘format’ akkas ‘put an X on somebody’ mallaћ ‘add salt’

tCaCCaC, by contrast, is selected mostly for intransitive verbs like inchoatives, reflexives and reciprocals (16). The transitive verb makyaj ‘put makeup on X’, for example, is formed in CaCCaC, while its reflexive intransitive counterpart,

Doublet Formation in Palestinian Arabic

115

tmakyaj ‘put on makeup on oneself’ is formed in tCaCCaC. In some cases, intransitive verbs can be formed directly in tCaCCaC without a CaCCaC transitive counterpart. For example, the reflexive verb tkažwal ‘put on casual clothes’ is formed based on the English word casual and there is no related transitive verb in CaCCaC (*kažwal). (16) Base a. Inchoatives nərvəs ‘nervous’ markaz ‘center’ kahraba ‘electricity’ ʔazme ‘crisis’

Derived verb tnarvaz tmarkaz tkahrab tʔazzam

‘become nervous’ ‘become centralized’ ‘get electrocuted’ ‘reach a crisis’

b. Reflexives ћija:b ‘veil’ kundara ‘shoe’ kæžuəl ‘casual’

tћajjab ‘put on a veil’ tkandar ‘put on a shoe’ tkažwal ‘put on casual clothes’

c. Reciprocals biznəs ‘business’

tbaznas

‘do business together’

The crucial point is that with respect to the formation of new verbs in PA, there are only two main patterns that host such verbs, and there is a clear semantic distinction with respect to the verbs that each pattern hosts. As mentioned in §4.1, the CiCeC-tCaCCaC paradigm is the default paradigm in verb formation, as other patterns, including CiCeC, are not productive. There is a sharp contrast between the low productivity of CiCeC in verb innovation and its high productivity within existing forms that are used. A recent study by Laks, Saiegh-Haddad and Haj (in preparation) has examined the distribution of PA patterns in narrative text production of 30 adult native speakers of PA between the ages of 25–35. Participants were asked to watch a short video clip that depicts different situations of conflict and then produce a text on a conflict that they experienced. The texts were transcribed and the verbs were coded according to pattern, stem consonants, transitivity and semantic types. The most frequent pattern in the texts is CiCeC/CaCaC, which is indeed considered the most basic pattern of verbs that are considered basic Arabic vocabulary items. It was found to be 44% of the verb types and 60% of the tokens. As noted, this pattern does not carry specific semantic functions and is considered neutral with respect to transitivity, as it hosts both transitive and intransitive verbs

116

Laks and Yousef

(see Berman, 1982, 2003 for a similar pattern in Hebrew). However, examining the verbs that were actually used in texts reveals a different picture. Most of the verbs are basic active verbs, either transitive or intransitive. Inchoative intransitive verbs are less common in this pattern, constituting only 14% of CiCeC/CaCaC verbs types. It follows that this patterns is less typical of inchoative verbs. This explains why CiCeC intransitive verbs have doublets in tCaCCaC and not in other patterns. tCaCCaC doublets are formed not only because of structural transparency (§4.2), but also in order to mark CiCeC verbs as typical intransitive verbs with an inchoative meaning. The change is into a pattern with high structural transparency and that is appropriate with respect to its semantic-syntactic features. CiCeC active transitive verbs, in contrast, do not have doublets as they are in an “appropriate” pattern with respect to their semantic-syntactic features. The picture that emerges is that the morphological mechanism conspires to rearrange the verbal system in a way that there is a stronger correlation between form and meaning. Intransitive inchoative verbs that are formed in a less typical pattern change their form so that the majority of such verbs belong to one typical pattern. To sum, we have shown that within CiCeC verbs, only the inchoative ones have doublets and that these doublets are formed in tCaCCaC and not in other patterns, as it is the most typical pattern for this semantic function. One question that remains is why there are some other inchoative CiCeC verbs that do not have doublets in tCaCCaC. We address this in the next sub-section. 4.3.2 Blocking of Doublet Formation So far, we have shown that some CiCeC verbs receive an additional form in the tCaCCaC pattern and that this is triggered by structural transparency within the inflectional paradigms of the verbs, avoiding both prosodic and segmental alternation. In addition, such doublet formation marks the verbs as typical intransitive verbs. However, not all CiCeC intransitive verbs have tCaCCaC doublets. For example, the intransitive verb riʕeb ‘become frightened’ does not alternate with *traʕʕab. If we compare these verbs to other intransitive verbs with doublets, there seems to be no phonological or semantic factors that would block such formation. Why, then, is doublet formation blocked in such cases? We claim that morphological change is also related to the derivational relations between verbal patterns. As noted in §3, while each pattern has its own inflectional paradigms, the relations between verbs with the same stem consonants in different patterns are derivational and are manifested mainly in terms of transitivity alternations, e.g., wassax ‘make dirty’ (CaCCaC)—twassax

Doublet Formation in Palestinian Arabic

117

‘become dirty’ (tCaCCaC), and ka:tab ‘write to X’ (Ca:CaC) and tka:tab ‘correspond’ (tCa:CaC). Examining the paradigmatic relationships between verbal patterns reveals that some derivational paradigms are more frequent than others in transitivity alternations. For example, the CaCCaC-tCaCCaC paradigm is more common than the CiCeC-tCaCCaC paradigm. Many CaCCaC verbs have tCaCCaC verbs that are derivationally related to them, but there are rare cases of derivational relations between CiCeC and tCaCCaC, e.g., wiled ‘give birth’—twallad ‘be generated’. In general, when the same CiCeC and tCaCCaC verbs share the same stem consonants, they do not have derivational relations (e.g., radd ‘reply’—traddad ‘hesitate’), and if there are some derivational relations, they are rare and less semantically transparent, unless these are doublets with the same meaning. In contrast, the relations between CaCCaC and tCaCCaC are semantically transparent. Similarly, the derivational relations in the CaCaC-inCaCaC paradigm are semantically transparent, where CaCaC hosts transitive verbs and inCaCaC is used for the formation of CaCaC passive/inchoative counterparts, e.g., katab ‘write’—inkatab ‘be written’. However, CaCCaC verbs are rarely related to inCaCaC verbs, hence these two patterns do not constitute a derivational paradigm. It follows that each pattern has its ‘usual mate’ pattern (or patterns) in terms of derivational relations (see Ravid, 2008 for Hebrew). That is, it is possible to predict that if we find a verb in pattern A, it is more likely to find a semantically related verb with the same stem consonants in pattern B than in pattern C. It should be noted though, that this reflects strong tendencies and not a dichotomy. As noted in §3, the relations between PA patterns cannot be fully predicted, which is typical of derivational relations in general. Turning back to the cases of doublet formation and lack thereof, we argue that the derivational relations between patterns also have an impact on the possibility of doublet formation. Examining cases of CiCeC-to-tCaCCaC doublet formation reveals that it is only CiCeC verbs with transitive counterparts in CaCCaC that have tCaCCaC doublets. As shown in §4.2.1, CaCCaC-tCaCCaC is the most productive transitive-to-intransitive derivational paradigm. New verbs that enter the language are formed almost exclusively in one of these patterns based on their (in)transitivity. This doublet formation makes the intransitive verbs adhere to the most frequent paradigm. CiCeC intransitive verbs, in contrast, tend to retain their form if they are not morphologically related to the CaCCaC-tCaCCaC. More examples of CiCeC intransitive verbs with no tCaCCaC doublet are presented in (17). In all these cases, there are no CaCCaC transitive counterparts with the same stem consonants.

118 (17) CiCeC rijef ‘shivered’ ɣidˁeb ‘get angry’ ɣime ‘faint’ miredˁ ‘become sick’

Laks and Yousef

“Missing” tCaCCaC doublet *trajjaf *tɣadˁdˁab *tɣamma *tmarradˁ

The relations between the verbal patterns and their impact on doublet formation are represented in (18). In (18a) we have the most typical transitiveintransitive paradigm. The transitive verb ћassan ‘improve X’ is formed in CaCCaC and its intransitive counterpart is formed in tCaCCaC, yielding tћassan ‘become improved’. There is no verb with the stem consonants ћ-s-n in CiCeC (*ћisen) and therefore its cell is shaded, and in such cases, there is indeed no motivation for doublet formation. In (18b), the intransitive verb xireb ‘become ruined’ is formed in CiCeC, and its transitive counterpart is formed in CaCCaC, yielding xarrab ‘ruin’. The existence of xarrab in CaCCaC allows the formation of an intransitive verb in tCaCCaC, yielding txarrab ‘become ruined’, which is the doublet of xireb. txarrab takes over xireb, to some extent, and this results in a more uniform paradigmatic verbal system, in which transitive verbs are formed in CaCCaC and their intransitive counterparts are formed in tCaCCaC. It is important to emphasize that the doublets are xireb and txarrab, but the formation of txarrab is based on xarrab. Such a formation is executed only by adding the prefix t- to the transitive verb. In (18c), we have a CiCeC intransitive verb riʕeb ‘become scared’ with no transitive counterpart in CaCCaC (*raʕʕab ‘make scared’). As a result, it also does not have a doublet in tCaCCaC, as there is no CaCCaC base for such a formation. The claim made here is that CiCeC intransitive verbs are not the basis of direct formation of doublets in tCaCCaC due to two reasons. First, tCaCCaC is not the ‘usual mate’ of CiCeC, as in most cases there are not direct semantic relations between verbs in these two patterns, namely they do not constitute a common derivational paradigm. Second, the formation of tCaCCaC verbs based on CiCeC verbs is less structurally transparent, as it would involve adding a prefix (t-), gemination and changing the vowel of the base. In contrast, the formation of tCaCCaC verbs based on CaCCaC verbs involves only adding a prefix, hence the structural relations within the CaCCaC-tCaCCaC paradigm are more transparent. The transitive verb in CaCCaC is, therefore, needed as a mediator for the doublet formation of CiCeC intransitive verbs in tCaCCaC.

Doublet Formation in Palestinian Arabic

119

(18) CiCeC Intransitive a.

CaCCaC Transitive ћassan ‘improve X’

tCaCCaC Intransitive tћassan ‘become improved’

b. xireb

‘become ruined’

xarrab

‘ruin’

txarrab ‘become ruined’

c. riʕeb

‘become scared’

*raʕʕab

‘make scared’

*traʕʕab ‘become scared’

In sum, the CaCCaC-tCaCCaC paradigm is the most productive transitive-tointransitive paradigm. CiCeC intransitive verbs that are related to CaCCaC transitive verbs change their form, adhering to the most common paradigm. CiCeC intransitive verbs with no CaCCaC transitive alternates do not change their form, as there is no motivation for this in terms of derivational paradigms. 5

Conclusions

This paper has examined the formation of morphological doublets in the verbal system of PA, the factors that trigger it and the factors that block it. We have shown that the morphological change is from CiCeC transitive verbs into tCaCCaC, such that the CiCeC and tCaCCaC doublets share the same meaning and can be used in the same context. Doublet formation is motivated by a morpho-phonological factor, namely achieving uniformity within the verbs’ inflectional paradigms by reducing both prosodic and segmental alternations. In addition, within the intransitive CiCeC verbs that are prone to receive additional forms in tCaCCaC, it is mostly verbs with CaCCaC transitive counterparts. Cases where such change is blocked can be accounted for by the lack of typical paradigmatic relations between verbal patterns. The study shows that words demonstrate greater tendency to undergo variation when they are part of a derivational paradigm. The morphological change establishes more uniform and steady paradigms, in which there is a clear morphological association between their members. The study demonstrates the central role of paradigms in morphological change. The likelihood of such change to occur depends to a great extent on paradigmatic relations between words in both inflectional and derivational paradigms.

120

Laks and Yousef

Acknowledgement This research was supported by the ISRAEL SCIENCE FOUNDATION (grant no. 1303/18). References Abu Chacra, F. (2007). Arabic: An essential grammar. London & New York: Routledge. Abu Elhija, D. (2012). Qaf in Nazarene electronic writing. e-Journal of non-Sequitur, 1(2), 110–116. Acquaviva, P. (2008). Lexical plurals. Oxford: Oxford University Press. Adam, G. (2002). From optimal grammar to variable grammar (Doctoral dissertation). Albright, A. (2005). The morphological basis of paradigm leveling. In L. Downing, T. Alan Hall, & R. Raffelsiefen (Eds.), Paradigms in Phonological Theory (pp. 17–41). Oxford: Oxford University Press. Available at: http://web.mit.edu/albright/www/ papers/latin.html. Anttila, A. (1997). Deriving variation from grammar. In F. Hinskens, R. van Hout, & L. Wetzels (Eds.), Variation, change and Phonological Theory (pp. 35–68). Amsterdam: John Benjamins. Anttila, A. (2007). Variation and optionality. In P. de Lacy (Ed.), The Cambridge handbook of phonology (pp. 519–536). Cambridge: Cambridge University Press. Aronoff, M. (1976). Word formation in generative grammar. Cambridge, MA: MIT Press. Aronoff, M. (1994). Morphology by itself. Cambridge: MIT Press. Aronoff, M. (2016). Competition and the lexicon. In A. Elia, C. Iacobino, & M. Voghera (Eds.), Livelli di Analisi e fenomeni di interfaccia. Atti del XLVII congresso internazionale della società di linguistica italiana (pp. 39–52). Rome: Bulzoni. Aronoff, M. (2019). Competitors and alternants in linguistic morphology. In F. Gardani, F. Rainer, H. C. Luschützky, & W. U. Dressler (Eds.), Competition in inflection and word formation (pp. 39–66). Berlin: Springer. Bat-El, O. (1994). Stem modification and cluster transfer in Modern Hebrew. Natural Language and Linguistic Theory, 12, 572–596. (2002). Bat-El, O. (2003). Semitic verb structure within a universal perspective. In J. Shimron, Language processing and acquisition in languages of Semitic, root-based, morphology (pp. 29–59). Amsterdam: John Benjamins. Bat-El, O. (2005). Competing principles of in paradigm uniformity: evidence from Hebrew imperative paradigm. In L. J. Downing, T. A. Hall, & R. Raffelsiefen (pp. 44– 64). Oxford: Oxford University Press. Bat-El, O. (2011). Semitic templates. In M. van Oostendorp, C. Ewen, E. Hume, & K. Rice (Eds.), Blackwell companion to Phonology (pp. 2586–2668). Malden, MA: Wiley-Blackwell.

Doublet Formation in Palestinian Arabic

121

Bauer, L. (1997). Derivational paradigms. In G. Booij & J. van Marle (Eds.), Yearbook of Morphology 1996 (pp. 243–256). Dordrecht: Kluwer. DOI: 10.1007/978-94-017-3718-0_13 [Accessed 20 March, 2019]. Bauer, L. (2006). Competition in English word formation. In van A. Kemenade & B. Los (Eds.), The Handbook of the History of English (pp. 177–198). Oxford: Blackwell Publishing Ltd. Beecher, H. (2004). Derivational paradigm in word formation. Retrieved from: http:// citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.94.9071&rep=rep1&type=pdf [Accessed 20 March, 2019]. Benmamoun, A. (1991). Causatives in Moroccan Arabic. In B. Comrie & M. Eid (Eds.), Perspectives in Arabic linguistics III (pp. 173–195.). Amsterdam: John Benjamins. Benmamoun, E. (2003). The role of the imperfective template in Arabic morphology. In J. Shimron (Ed.), Language processing and acquisition in languages of Semitic, root-based, morphology (pp. 99–114). Amsterdam: John Benjamins. Berman, R. (1978.) Modern Hebrew structure. Tel-Aviv: University Publishing Projects. Berman, R. (1982). Verb-pattern alternation: the interface of morphology, syntax, and semantics in Hebrew child language. Journal of Child Language, 9, 169–191. Berman, R. (2003). Children’s lexical innovations: developmental perspectives on Hebrew verb-structure. In J. Shimron (Ed.), Language processing and language acquisition in a root-based morphology (pp. 243–291). Amsterdam: John Benjamins. Blevins, J. P. (2016). Word and Paradigm Morphology. Oxford: Oxford University Press. Bloomfield, L. (1933). Language. London: Unwin. Bolozky, S. (1978). Word formation strategies in Modern Hebrew verb system: denominative verbs. Afroasiatic Linguistics, 5, 1–26. Bolozky, S. (1999). Measuring productivity in word formation: the case of Israeli Hebrew. Leiden, Boston: Brill. Bolozky, S., & Saad, G. N. (1983). On active and non-active causativizable verbs in Arabic and Hebrew. Zeitschrift für Arabische Linguistik, 10, 71–79. Booij, G. (1996). Inherent versus contextual inflection and the split morphology hypothesis. In G. E. Booij & J. van Marle (Eds.), Yearbook of morphology 1995 (pp. 1–16). Dordrecht: Kluwer. Booij, G. (1997). Autonomous morphology and paradigmatic relations. In G. E. Booij & J. van Marle (Eds.), Yearbook of Morphology 1996 (pp. 35–55). Dordrecht: Kluwer. Booij, G. (2002). Language variation and phonological theory: inflected adjectives in Dutch and related languages. In J. Berns & J. van Marle (Eds.), Amsterdam: Present-day dialectology, problems and findings (pp. 35–56). Berlin: Mouton de Gruyter. Booij, G. (2008). Paradigmatic morphology. In B. Fradin (Ed.), La raison morphologique. Hommage á la mémoire de Danielle Corbin (pp. 29–38). Amsterdam, Benjamins.

122

Laks and Yousef

Booij, G., & Lieber, R. (2004). On the paradigmatic nature of affixal semantics in English and Dutch. Linguistics, 42(2), 327–357. Burzio, L. (1998). Multiple correspondence. Lingua, 104, 79–109. Burzio, L. (2002). Surface-to-surface morphology: When your representations turn into constraints. In P. Boucher (Ed.), Many Morphologies (pp. 142–77). Sommerville: Cascadilla Press. Carstairs, A. (1987). Allomorphy in inflexion. London: Croom Helm. Chekayri, A., & Scheer, T. (2003). The appearance of glides in Classical Arabic defective verbs. Folia Orientalia, 39, 7–33. Corbett, G. (2007). Canonical typology, suppletion, and possible words. Language, 83(1), 8–41. Corbett, G. (2010). Canonical derivational morphology. Word Structure, 3, 141–155. Corbin, D. (1989). Form, structure and meaning of constructed words in an associative and stratified lexicon. In G. Booij & J. van Marle (Eds.), Yearbook of Morphology 2 (pp. 31–54). Dordrecht: Foris. Dal, G., & Namer, F. (2010). French property nouns based on toponyms or ethnic adjectives: A case of base variation. In F. Rainer, W. U. Dressler, D. Kastovsky, & H. C. Luschützky (Eds.), Variation and change in Morphology (pp. 53–74). Amsterdam: John Benjamins. Dal, G., & Namer. F. (2015). Internet. In P. O. Müller, I. Ohnheiser, S. Olsen, & F. Rainer (Eds.), Word-formation: International handbook of the languages of Europe (pp. 2372– 2386). New York: De Gruyter Mouton. Doron, E. (2003). Agency and voice: The semantics of the Semitic templates. Natural Language Semantics, 11, 1–67. Elihay, J. (2005). The Olive Tree Dictionary. Jerusalem: Minerva. Embick, D. (2008). Variation and morphosyntactic theory: Competition fractionated. Language and Linguistics Compass, 2(1), 59–78. Hathout, N., & Namer, F. (2014). Démonette, a French derivational morpho-semantic network. Linguistic issues in language technology, 11(5), 125–168. Henkin, R. (2009). The cognate curse in Negev Arabic: From playful punning to coexistence conflicts. Israel Studies in Language and Society, 2(2), 169–206. Henkin, R. (2010). Negev Arabic: Dialectal, sociolinguistic, and stylistic variation (Semitica Viva 48). Wiesbaden: Harrassowitz Verlag. Hinskens, F., van Hout, R., & Wetzels, W. L. (1997). Balancing data and theory of phonological variation and change. In F. Hinskens, R. van Hout, & W. L. Wetzels (Eds.), Variation, change and phonological theory (pp. 1–33). Amsterdam: Benjamins. Holes, C. (2004). Modern Arabic: Structures, functions and varieties. Washington D.C.: Georgetown University Press. Izre’el, S. (2010). Constructive constructions: Semitic verbal morphology and beyond. In G. Goldenberg & A. Shisha-Halevy (Eds.), Egyptian, Semitic and general grammar

Doublet Formation in Palestinian Arabic

123

(Workshop in Memory of H. J. Polotsky) (pp. 106–130). Jerusalem: The Israel Academy of Sciences and Humanities. Jastrow, O. (2004). The Arabic dialects of the Muthallath (central Israel). Jerusalem Studies in Arabic and Islam, 29, 166–175. Joseph, B. D. (1998). Diachronic morphology. In A. Spencer & A. M. Zwicky (Eds.), The handbook of morphology (pp. 351–373). Oxford: Blackwell Publishers. Kiparsky, P. (1968). Linguistic universals and linguistic change. In E. Bach & R. Harms (Eds.), Universals in linguistic theory (pp. 171–202). New York: Holt, Rinehart & Winston. Kiparsky, P. (1995). The phonological basis of sound change. In J. Goldsmith (Ed.), The handbook of phonological theory (pp. 640–670). Oxford: Blackwell. Kiparsky, P. (2010). Dvandvas, blocking, and the associative: The bumpy ride from phrase to word. Language, 86, 302–331. Kroch, A. (1989). Reflexes of grammar in patterns of language change. Language change, 1, 199–244. Laks, L. (forthcoming). Verb innovation in Hebrew and Palestinian Arabic: The interaction of morpho-phonological and thematic-semantic criteria. Brill’s annual Afroasiatic languages and linguistics. Levin, A. (1995). A dialect of Jerusalem. Jerusalem: The Hebrew University Magnes Press. [In Hebrew]. Lubani, H. (2006). The broadened dictionary of the colloquial and the foreign in Palestine. Beirut: The Library of Lebanon Publishers. Lüdeling, A., Evert, S., & Baroni, M. (2007). Using web data for linguistic purposes. In M. Hundt, N. Nesselhauf, & C. Biewer (Eds.), Corpus linguistics and the web (pp. 7–24). New York: Rodopi. Macken, M. (1992). Where’s phonology? In C. A. Ferguson, L. Menn, & C. Stoel-Gammon (Eds.), Phonological development: Models, research, implications (p. 249–269). Timonium: York Press. Maiden, M. (2004). Irregularity as a determinant of morphological change. Journal of Linguistics, 28, 285–312. Maiden, M., & O’Neill, P. (2010). On morphomic defectiveness: evidence from the Romance languages of the Iberian Peninsula. In M. Baerman, G. Corbett, & D. Brown (Eds.), Defective paradigms: Missing forms and what they tell us (pp. 181–210). Oxford: Oxford University Press. Malkiel, Y. (1977). The analysis of lexical doublets. The Romanists’ earliest contribution to general linguistics. In D. Feldman (Ed.), Homenaje a Robert A. Hall Jr.: Ensayos lingüísticos y filológicos para su sexagésimo aniversario (pp. 191–196). Madrid: Playor. McCarthy, J. (1981). A prosodic theory of nonconcatenative morphology. Linguistic Inquiry, 12, 373–418.

124

Laks and Yousef

McCarthy, J. (2005). Optimal paradigms. In L. J. Downing, T. A. Hall, & R. Raffelsiefen (Eds.), Paradigms in phonological theory (pp. 295–371). Oxford: Oxford University Press. Meir, I. (2008). The role of prosody in morphological change: The case of Hebrew bound numerals. Language variation and change, 20, 41–65. Mörth, K., & Dressler, W. U. (2012). German plural doublets with and without meaning differentiation. In F. Rainer, F. Gardani, H. C. Luschützky, & W. U. Dressler (Eds.), Morphology and meaning: Selected papers from the 15th International morphology meeting, Vienna (pp. 249–259). Amsterdam: John Benjamins Publishing Company. Ornan, U. (1971). Verbal-patterns, stems: inflections and derivations. Ha-Universita, 16, 15–22. [In Hebrew]. Ornan, U. (2003). The final word. Haifa: Haifa University. [In Hebrew]. Ouhalla, J. (2014). Causatives, anticausatives and lexicalization. In S. Benjaballah, N. Faust, M. Lahrouchi, & N. Lampitelli (Eds.), The form of structure, the structure of form: Essays in honor of Jean Lowenstamm (pp. 333–348). Amsterdam: John Benjamins. Palfreyman, D. & Al-Khalil, M. (2007). A funky language for Teenzz to use: Representations of Gulf Arabic. In B. Danet & S. C. Herring (Eds.), The multilingual Internet: Language, culture, and communication online (pp. 43–63). New York: Oxford University Press. Pintzuk, S. (1991). Phrase structure in competition: variation and change in Old English word order (Doctoral dissertation, University of Pennsylvania). Retrieved from: https://www.researchgate.net/publication/247176104_Phrase_Structures_in_ Competition_Variation_and_Change_in_Old_English_Word_Order [Accessed 19 March, 2019]. Plag, I. (2000). On the mechanisms of morphological rivalry: A new look at competing verb-deriving affixes in English. In B. Reitz & S. Rieuwerts (Eds.), Anglistentag 1999 Mainz (pp. 63–76). Trier: Wissenschaftlicher Verlag Trier. Pounder, A. (2000). Processes and paradigms in word-formation morphology. Berlin: Mouton de Gruyter. Rainer, F. (1988). Towards a theory of blocking: Italian and German quality nouns. In G. Booij & J. van Marle (Eds.), Yearbook of Morphology 1988 (pp. 155–185). Dordrecht: Foris. Ravid, D. (1990). Internal structure constraints on new-word formation devices in Modern Hebrew. Folia Linguistica, 24(3–4), 289–348. Ravid, D. (2003). Root perception in Hebrew and Palestinian Arabic. In J. Shimron (Ed.), Language processing and language acquisition in a root-based morphology (pp. 293–320). Amsterdam: John Benjamins. Ravid, D. (2008). Parsimony and efficacy: The dual binyan system of Hebrew. Paper delivered at the 13th International Morphology Meeting 2008 (IMM18), 10–13 May, Vienna.

Doublet Formation in Palestinian Arabic

125

Reynolds, W. (1994). Variation and Phonological Theory (Doctoral dissertation, University of Pennsylvania.). Retrieved from: https://www.researchgate.net/publi cation/45598536_Variation_and_Phonological_Theory [Accessed 20 March, 2019]. Rosén, H, (1977). Contemporary Hebrew. The Hague: Mouton. Ryding, K. (2005). A Reference Grammar of Modern Standard Arabic. Cambridge, United Kingdom: Cambridge University Press. Schwarzwald, O. R. (1977). The weak verb representation in the Modern Hebrew lexicon. Hebrew Computational Linguistics, 12, 25–36. [In Hebrew]. Schwarzwald, O. R. (1981). Grammar and reality in the Hebrew verb. Ramat Gan: Bar Ilan University Press. [In Hebrew]. Schwarzwald, O. (1984). Analogy and regularization in morphophonemic change: The case of the weak verbs in Post-Biblical Hebrew and colloquial Modern Hebrew. AfroAsiatic Linguistics, 9(2), 87–100. Schwarzwald, O. (1996). Syllable structure, alternations and verb complexity: The Modern Hebrew verb patterns reexamined. Israel Oriental Studies, 16, 99–112. Schwarzwald, O. (2002). Hebrew morphology. Tel Aviv: The Open University. [In Hebrew]. Schwarzwald, O. (2008). The special status of Nif’al in Hebrew. In S. Armon-Lotem, S. Rothstein, & G. Danon (Eds.), Approaches to Modern Hebrew linguistics (pp. 61–75). Amsterdam, John Benjamins Publishing. Shawarbah, M. (2012). A Grammar of Negev Arabic: Comparative Studies, Texts and Glossary in the Bedouin Dialect of the ‘Azazmih Tribe. Wiesbaden: Harrassowitz. Spencer, A. (1988). Bracketing paradoxes and the English lexicon. Language, 64, 663–682. Spencer, A. (1991). Morphological theory: An introduction to word structure in generative grammar. Cambridge: Cambridge University Press. Štekauer, P. (2014). Derivational paradigms. In R. Lieber & P. Štekauer (Eds.), The Oxford handbook of derivational morphology (pp. 354–369). Oxford: Oxford University Press. Steriade, D. (1988). Reduplication and transfer in Sanskrit and elsewhere. Phonology, 5(1), 73–155. Steriade, D. (2000). Paradigm uniformity and the phonetic-phonology boundary. In M. Broe & J. Pierrehumbrett (Eds.), Papers in laboratory phonology V: Acquisition and the lexicon (pp. 313–334). Cambridge: Cambridge University Press. Stump, G. (2001). Inflectional morphology. A theory of paradigm structure. Cambridge: Cambridge University Press. Stump, G. (2010). Interactions between defectiveness and syncretism. In M. Baerman, G. Corbett, & D. Brown (Eds.), Defective paradigms: Missing forms and what they tell us (pp. 181–210). Oxford: Oxford University Press. Stump, G. (2016). Inflectional paradigms. Cambridge: CUP.

126

Laks and Yousef

Stump, G., & Finkel, R. A. (2013). Morphological typology: From word to paradigm. Cambridge: CUP. Sumner, M. M. (2003). Testing the abstractness of phonological representations in Modern Hebrew weak verbs. Doctoral Dissertation, SUNY Stony Brook. Thornton, A. (2011). Overabundancy (multiple cells realizing the same cell): A noncanonical phenomenon on Italian verb morphology. In M. Maiden, J. C. Smith, M. Goldbach, & M. Hinzelin (Eds.), Morphological autonomy: Perspectives from Romance inflectional morphology (pp. 358–381). Oxford: Oxford University Press. Thornton, A. (2012a). Overabundance in Italian verb morphology and its interactions with other non-canonical phenomena. In T. Stolz, H. Otsuka, A. Urdze, & J. der Auwera (Eds.), Irregularity in morphology (and beyond) (pp. 251–269). Berlin: Akademie Verlag. Thornton, A. (2012b). Reduction and maintenance of overabundance: A case study of Italian verb paradigms. Word Structure, 5, 183–207. Ussishkin, A. (1999). The inadequacy of the consonantal root: Modern Hebrew denominal verbs and output-output correspondence. Phonology, 16, 401–442. Ussishkin, A. (2005). A fixed prosodic theory of non-concatenative templatic morphology. Natural Language and Linguistic Theory, 23, 169–218. van Marle, J. (1985). On the paradigmatic dimension of morphological creativity. Dordrecht: Foris. Vincent, N. (1980). Words versus morphemes in morphological change: the case of Italian -iamo. In J. Fisiak (Ed.), Historical morphology (pp. 383–398). The Hague: Mouton. Wang, W. (1969). Competing changes as a cause of residue. Language, 45, 9–25. Warschauer, M., El-Said, G. R., & Zohry. A. (2002). Language choice online: Globalization and identity in Egypt. Journal of Computer Mediated Communication, 7(4). Retrieved from: http://jcmc.indiana.edu/vol7/issue/4warschauer.html [Accessed 20 March, 2019]. Watson, J. E. (2002). The phonology and morphology of Arabic. Oxford: Oxford University Press. Wedel, A. (2009). Variation, multi-level selection and conflicts between phonological and morphological regularities. In J. Blevins & J. Blevins (Eds.), Analogy in grammar: Form and acquisition (pp. 83–100). Oxford: Oxford University Press. Available at: http://dingo.sbs.arizona.edu/~wedel/publications/PDF/WedelPhonology MorphologyConflict.pdf. Weinreich, U., Labov, W., & Herzog, M. I. (1968). Empirical foundations for a theory of language change. In W. Lehmann & Y. Malkiel (Ed.), Directions for historical linguistics (pp. 95–195). Austin, TX: University of Texas Press.

Doublet Formation in Palestinian Arabic

127

Yang, C. (2002). Knowledge and learning in natural language. Oxford: Oxford University Press. Younes, M. (2000). Redundancy and productivity in Palestinian Arabic verb derivation. In M. Mifsud (Ed.), Proceedings of the Third International Conference of Aida (pp. 27–32). Hal Lija: Association international de dialectologie arabe (AÏDA).

Chapter 6

What We Talk about When We Talk about Paradigms: Representing Latin Word Formation Eleonora Litta and Marco Budassi 1

Introduction

Recent years have seen a growing interest in research aimed at building new language resources and Natural Language Processing (NLP) tools for derivational morphology. The Word Formation Latin lexicon (WFL, Litta & Passarotti, 2019) is a word formation-based lexicon for Classical and Late Latin, in which derivational and compounding rules are modelled as directed one-to-many input-output relations between lexemes. WFL was conceived following the Item-and-Arrangement (I&A) model of morphological description (Hockett, 1954), where both bases and affixes convey semantic information and are stored in the dictionary. WFL shares the same lexical basis as the morphological analyser and lemmatiser for Latin Lemlat and connects lemmas collected from three main Classical and Late Latin dictionaries: Georges (Georges & Georges, 1913–1918), Oxford Latin Dictionary (OLD, Glare, 1982); Gradenwitz (1904) (for a total of 4332 lexemes), plus most of the Onomasticon (2650 lexemes out of 2878) provided by Forcellini (1940; Budassi & Passarotti, 2016). As a rule, choices on dubious derivation processes are generally made following the contents of the OLD (and Georges secondarily in those cases where its lexical coverage is larger). This methodology is applied especially to the identification of affixes, so, for example, prefixal word-formation rules are defined as such when OLD includes certain invariables and numerals as prefixes. For instance, prae, is listed as an adverb and a preposition, but also as a prefix, so all lexemes introduced by prae- are prefixed words, while retro is listed only as an adverb, a self-contained word, hence all lexemes formed with retro are considered compounds. WFL, like other similar resources treating various modern languages,1 uses a step-by-step morphotactic approach: each word-formation process is treated 1  See for example the lexical network for Czech DeriNet (Ševčíková & Žabokrtský, 2014), the derivational lexicon for German DERIVBASE (Zeller, Snajder, & Padó, 2013) and that for Italian derIvaTario (Talamo, Celata, & Bertinetto, 2016). © Koninklijke Brill NV, Leiden, 2020 | doi:10.1163/9789004433410_007

What We Talk about When We Talk about Paradigms

129

individually, and the output of a word-formation rule (WFR) usually contains one morpheme more than the input (with the exception of rules involving conversion of part of speech). Each output lexeme can only have one base lexeme, unless the output lexeme qualifies as a compound of more than one lexeme (multinominis adj. ‘having many names’ < multus adj. ‘many’ + nomen noun ‘name’). The WFL lexicon is available as a web application (http://wfl.marginalia.it), where the relationships between lexemes of the same ‘word-formation family’ are represented as edges in a tree-graph. In the graph, each node is occupied by a lexeme, and each edge is labelled with a description of the WFR used to derive the output lexeme from the input one. During the construction of WFL, however, it has become clear that representing word-formation processes via directed tree-graphs can, at times, raise some significant theoretical issues, especially in those cases where the derivational relation is ambiguous or seems to be skipping a step, as shown in Budassi and Litta (2017). This for example happens when philological evidence goes against the morphotactic input-output directional approach, like in the case of more complex verbs being attested before simpler ones (e.g., ex-horre-sc-o ‘to shudder’ was attested earlier than the simpler ex-horreo ‘to shudder, be terrified’; Haverling, 2000); or, in the case of backformation, where it is difficult to stick to the step by step simpler > more-complex model (e.g., ira-sc-or ‘to be angry’ < ir-at-us ‘angry’); or, in the case of formations where more than one affix is added at the same time without an intermediate step (decaulesco ‘to form a stem’ < caulis ‘stem’). In such cases, WFL uses a series of stratagems to work around the problem, the most striking of which is the creation of “fictional” lexemes that serve as links between existing words in order to justify the extra step: caulis < *caulesco < decaulesco. Although it is not implied that these fictional lexemes ever existed, their presence in the lexicon, even if justified by the step-by-step methodology, remains however unsatisfactory. It has been found that they do only confuse the user, who is accustomed to treating words preceded by asterisks as reconstructions of unattested—yet at some point existing—words, but they also create a cumbersome presence in the digital resource: indeed, when performing numerical investigations on single-word formation processes, such fictional lexemes always need to be kept in mind and removed from counts and analyses. The recent emergence of a strong interest in the application of Word and Paradigm (WP) models to derivational morphology led us to explore their potential in explaining those processes, and to plan a second phase in the development of the WFL lexicon during which derivational relationships could be organised and visualised as paradigmatic structures. The principal advantage offered by paradigmatic models for derivational morphology in explaining and describing ambiguous relations between words

130

Litta and Budassi

is that these are not limited to base-derivative pairs. Derivation processes do not necessarily require a linear direction (Jackendoff, 1975), but are rather based on cell-filling potentiality. For Štekauer (2014), “potentiality is a crucial term for the concept of derivational paradigm”, which means that in those cases where there is a gap to fill, a paradigmatic approach to word formation “guarantees a high level of predictability and regularity […] in the sense that existing gaps in the system can be filled anytime with actual words”, or not (p. 369). Hence, the paradigmatic approach to derivational morphology can account for those cases where it is unclear how the derivational process worked, without recurring to directionality. However, WP applications to derivational morphology are still in a gestational stage and concrete models applicable to all languages, or indeed to all word-formation processes occurring in one single language, have yet to be proposed and applied on a large scale (i.e., with a full lexicon in mind).2 In this paper, we use the data contained in the WFL lexicon, which offers the double advantage of being a complete but closed set of lexemes, as a starting point for the definitive design of a derivational paradigm. However, when confronted with empirical data, that is, once we try to port a derivational family from the tree-graph display into a paradigm, we are presented with a number of questions on how to model such a paradigm. The main aim of this investigation is to determine what the main requirements for a physical representation of a derivational paradigm need to be. By adopting a table-like representation, we are going to stress the importance of the “cell” as a placeholder for lexemes that has the potentiality to be filled at any time, provided there is already a pattern upon which the new word can be modelled. This pattern is provided through a combination of the morphological and semantic functions found through empirical evidence (i.e., existing words behaving in the same way). We have found that many concepts borrowed from Construction Morphology (CxM, Booij, 2010) are very useful when adding morphological and semantic information to the cell. The paper describes which concepts borrowed from CxM suit the description of the function of a paradigm cell, and which ones we need to forgo in order to account for Latin’s varied development throughout time and space. Finally, by drawing data about productive WFRs from WFL, we are going to attempt the design of a “core” derivational paradigm for Latin.

2  For a detailed survey of the most recent proposals, see Fradin (this volume).

What We Talk about When We Talk about Paradigms

2

131

Paradigm Design

Designing a derivational paradigm is not a straightforward operation. This is due to the philological and theoretical issues sketched above (and, in more detail, in Budassi & Litta, 2017), that can also be found in many languages. Issues like the prioritisation of prefixation or suffixation in derivational processes, simultaneous double affixation and backformation clearly show how a paradigmatic approach to derivational morphology seems highly preferable to a linear one. Nevertheless, questions which need to be approached carefully remain, such as how is the paradigm shaped and which features it requires, in order to portray derivational processes. By and large, a paradigm can be described as a set of recurring behaviours, which apply to more than one series of morphologically-related forms sharing a base. When studying Latin inflection, for example, the “paradigm” of a verb is generally given as a series of forms, e.g., laudō, āre, āvī, ātum, ‘to praise’ (contractions of laudāre, laudāvī, laudātum, respectively) indicating the behaviour of the verb in its main significant tenses, that is, those that show stems different from the one for the present tense. The paradigm indicates that laudō behaves in the same way as all first conjugation verbs featuring the same pattern, such as, for example, amō, āre, āvī, ātum, ‘to love’, and differently from other first conjugation verbs displaying a different one, such as lavō, āre, lāvī, lautum, ‘to wash’. From the paradigm, a model, one can infer the full range of forms of a particular verb. Equally, the appeal of a derivational paradigm is the possibility of stacking recurring word formations into a model, carrying descriptive and predictive powers. Hence, in the case of derivational morphology, we can have ag-o, perag-o, and act-us, but also duc-o, per-duc-o, and duct-us. Our main goal is to enclose as much of a derivational family as possible within the derivational paradigm itself. This effort, however, cannot rely on the assumption that certain lexemes are linearly derived from others. More specifically, a number of properties are deemed necessary to postulate a manageable derivational paradigm: i. the model requires to waive assumptions on directionality to accommodate those lexemes for which the derivational process is not of the simplex (or simpler) > complex type (as in the example of exhorreo/ exhorresco above); ii. the cell, its function and characteristics, need to acquire a central role in the paradigm. Since the potential of the cell to be filled is an essential feature in derivational paradigms, followed by predictability and regularity, it follows that the availability of slots (filled with potential words) is more

132

Litta and Budassi

important for the paradigm than the forms that fill them (Štekauer, 2014, p. 369); iii. each cell must contain information on semantic features, due to the underlying role of semantics in accounting for derivational processes;3 iv. cells need to preserve morphological integrity, that is to say that while representing semantic features we cannot overlook morphological features: the same semantic feature of an affix (e.g., -or) does not make two lexemes the same (dictor, from the simple base verb dico, is not the same as dictator, from the iterative dic-to). To fulfil the aforementioned requirements, we conceived a number of experimental visualisations. Firstly, we considered a 3D box model in the shape of a table resembling an inflectional paradigm (see Budassi & Litta, 2017). In the 3D view, no directionality is at play, and missing links can be accounted for on the basis of analogy.4 Figure 6.1 illustrates part of the derivational paradigm of the morphological family characterised by the base HORR- represented according to a 3D boxes model. The cell labelled ‘Base’ (dotted line) displays the fundamental lexical morpheme, which does not change throughout the paradigm. Horizontally, we have the paradigm relying on the un-prefixed stem of HORR-. Vertically, the simple verb horreo holds a relation with horresco, which is built by adding the -sc- suffix to the simpler verb form. Prefixed forms of horreo are placed perpendicularly to horreo, prefixed forms of horresco are perpendicular to horresco. This three-dimensional orientation facilitates the display of prefixed forms with no assumption as to which form was created first: the relationship existing between horreo and e.g., inhorreo or exhorreo on the one hand, horresco and e.g., inhorresco or exhorresco on the other, is indeed paradigmatic in nature. If we imagine a stage of Latin in which the box in Figure 6.1 containing exhorreo were blank, the presence of other prefixed -sc- verbs, such as inhorresco or abhorresco, would justify the creation of exhorresco without the need for an input exhorreo. Exhorreo, then, started to be used, thus filling an empty box in the paradigm (consider e.g., the pairs inhorreo-inhorresco or abhorreo-abhorresco. Nonetheless, this approach raises a few significant issues. Primarily, a purely mechanical perspective on derivational morphology, that is, a model of 3  This is true even though Latin is not an easy-to-analyse language under this perspective. The evolution in meaning of Latin affixes does not always allow one to label things coherently (e.g., suffix -sc- loses its semantic value throughout time, but the negation prefix in- always keeps its privative meaning). 4  The role of analogy in derivational paradigms is extensively dealt with in Bagasheva (this volume).

What We Talk about When We Talk about Paradigms

133

Figure 6.1 Partial derivational paradigm of HORR-

derivational morphology, where complex lexemes are obtained only by adding morphemes to simpler lexemes, is not sufficient. Semantics of affixes and their combination plays a key role in word formation and must not be neglected. Some blank boxes in paradigms may indeed be the result of semantic constraints: for instance, in Latin we have cohorresco ‘to shiver from emotion’ but not *cohorreo. This is perhaps due to the perfective value of the prefix co-. Since horreo has stative meaning, it is not compatible with the perfectivising prefix co-. On the contrary, horresco, which is not stative, can be prefixed with co-. Moreover, regardless of how appealing a 3D visualisation of the paradigm may be, it can only really be implemented within a digital environment. This brought us back to a simpler but more direct model to visualise derivational paradigms: the table. Tables are easy to read, they are the basis for the traditional depiction of the inflectional paradigm, as well as memorable and consultable. Furthermore, sticking to a table-like representation highlights the importance of the ‘cell’ into the paradigm. Table 6.1, Table 6.2, and Table 6.3 below show one of the first tables that were drawn for the paradigm formed from base DIC-, for ‘to speak’. For space reasons, the full table has been here split into three parts. In these tables, suffixes were listed horizontally with a mention of the generic semantic value these add to the lexeme, while prefixes were listed vertically. Un-suffixed verbs (V), nouns (N), and adjectives (A) were also listed vertically, while un-prefixed lexemes are listed in the first row of each table, bearing the simple base for that paradigm. The cells in these tables are the result of the intersection of un/suffixed with un/prefixed lexemes. For this design, we theorised the existence of ‘parallel tables’ containing all lexemes characterised by similar features, such as a different stem, so that all lexemes featuring a certain symbol (+, #, *, etc.) belong in the same parallel table.

134

Litta and Budassi

Table 6.1 Derivational paradigm for base DIC-: verbs

Prefix/suffix

V

-sc- inchoative

Base: dic-

dico

abadconcontradeexin- (entering) in- (negation) interperpraeproresuper-

abdico addico condico contradico dedico edico indico

-ur- desiderative

-it- frequentative

dicturio dictaturio(+)

dicto(+) dictito

edicto

interdico perdico praedico prodico redico superdico

Table 6.2 Derivational paradigm for base DIC-: nouns Prefix/ suffix

Base: dic-

abadconcontradeex-

N

-t(us) -ioprovided action with

-oragent masc.

-rix -tas/t- -ur-ul-(un)cul- -bulagent abstract/ occupation diminutive diminutive instrument fem. state

dictus dicio dictum dictio dictata(+) dictatio(+) abdicatus addictio

dictor dictator(+)

dicta- dicacitas† dictatura(+) trix(+)

addictor

condi- concio(#) dictor condictio contra- contradictio dictor edictum edictio

dictiuncula dicabula dictatiuncula(+)

contraditiuncula

135

What We Talk about When We Talk about Paradigms Table 6.2 Derivational paradigm for base DIC-: nouns (cont.) Prefix/ suffix

N

-t(us) -ioprovided action with

inindex (entering) in(negation) interinterdictum perpraepraedictum proresuper-

-oragent masc.

-rix -tas/t- -ur-ul-(un)cul- -bulagent abstract/ occupation diminutive diminutive instrument fem. state

indictio† indicium

indiculum

interdic- intertio dictor praedictio prodictio prodicprodicia tator(+)

Table 6.3 Derivational paradigm for base DIC-: adjectives Prefix/ suffix

A

Base: dicabadconcontradeexin(entering) in(negation) interperpraeproresuper-

-ori-ens/t- -bil-al-os-ci/ce-ivdeverbal participle capacity relation abundance resemblance

uldiminutive

dictatorius(+)

dicaculus†

addictus addicta contradictorius

interdictorius

dicibilis

dictiosus

dicax†

condicio- condicionabilis# nalis# contra­ dicibilis*

indicens incontra­ dicibilis*

indictionalis†

indicticius

indictivus

praedictivus

136

Litta and Budassi

With this exemplification, we moved closer to the model we were pursuing, despite it being just a different take on the concept of the previously discussed 3D level. Some relevant points remained unanswered: i. the combination of suffixes and prefixes, especially in those cases where more than one suffix is involved (e.g., dict-io ‘the action of speaking’ does not have the same meaning and function as dict-at-io ‘a dictated draft’); ii. the concept of а parallel paradigm imposes, once again, assumptions on the direction of derivation. Does prodictator belong in the dictat- level or in the pro-dic- level? iii. a model where the colliding point between a plain base and affixes is not enough to account for the behaviour of the base itself in derivation; iv. the semantics of derivation is not well represented, because it is only portrayed at the axis but not at the intersection (i.e., cell) level. Ultimately, here the “cell” is not satisfactorily labelled and framed under a theoretical perspective. The main problem lies in the absence of both a description of the role and a representation of the function of the cell. What is more, even if the cell is adequately labelled and framed, its position within the table would be totally arbitrary, since the arrangement of prefixes/suffixes in the horizontal/vertical space of the table would be meaningless. The conclusion that was drawn from yet another failed experiment was that it was necessary to find the correct way of labelling the cell, so that it could contain all of the information needed to free it from position, direction and design. 3

Construction Morphology

We have found that the emerging morphology model CxM, conceived by Geert Booij in a series of publications (Booij, 2010, among others), could offer some help in the theorisation of how we want to describe Latin derivational paradigms, the interaction between cells, and the way we want to label them.5 For Booij (2009), “word-formation patterns can be seen as abstract schemas that generalise over sets of existing complex words with a systematic correlation between form and meaning. These schemas also specify how new complex words can be created” (p. 201). This, to us, represents the exemplification of how we would like the cell to be epitomised:

5  See also Radimský (this volume).

What We Talk about When We Talk about Paradigms

137

i. ii.

The role of the cell can be described by an abstract label. The label needs to be generic enough to describe the relationship between the combination of the lexeme’s parts but specialised enough to specify the nature of their meaning. iii. The label should be drawn and theorised from recurrent patterns already existing in a language. iv. A placeholder for new words to be created at any time during its development. CxM revolves around the central notion of constructions, conventionalised pairings of form and meaning (Booij, 2010, p. 6). For example, walker is analysed in its internal structure as in (1). (1) [[walk]V er]N ↔ [someone who walk(s)V]N Constructions may be hierarchically organised: words and multi-words are descendants of more schematic constructions, schemas, i.e., mental repositories of all constructions, that motivate the structure of existing words and facilitate the coinage of new ones. The schema below, for instance, represents a generalisation of the construction of all words displaying the same morphological structure as walker, e.g., buyer, player, surfer, reader, etc.: (2) [[x]Vier]Nj ↔ [someone who SEM Vi]Nj In (2) the double arrow stands for the form-meaning correspondence, the variable ‘x’ stands for the phonological form of the base, and ‘SEM’ stands for a generalisation of the meaning of ‘x’, contained in the first part of the equation. Subscript capital letters (V, N, etc.) indicate the part of speech exemplified by the schema, while ‘i’ and ‘j’ are markers for semantic differences. Schemas such as the one in (2) can additionally have placeholders for further morphological specifications (be that class for adjectives, conjugations or declensions, and so on), generally marked in Greek alphabet letters. The specification of such markers is stored elsewhere. Although construction schemas might look very similar to WFRs, WFRs are procedural, i.e., they explain the procedure of change from one part of speech to the other, (e.g., V-to-N -er). Schemas, on the other hand, are declarative, which means that they describe static generalisations, and are purely output-oriented. This represents an advantage for Latin, for which the lack of native speakers makes the explanation of non-concatenative phenomena, or ambiguous derivations, arduous. Following the CxM model, then, Latin words can be analysed as undivided items, but with specific internal morphosyntactic structures.

138

Litta and Budassi

The concept of constructional schemas as a systematic pairing of word and meaning puts emphasis on the lexical value of the base. The base, in our paradigm, is what is shared by both a complex lexeme and a simplex within the same morphological family. At the same time, the fact that the construction is in itself declarative means that the cell can exist on its own without the need for a direct ancestor, so long as there is a ‘base’ from which the word can be created and a ‘template’ (presumably equivalent to that existing in other families) for the creation of the word at any point in the history of the language. So excandesco ‘to catch fire’ can exist without any implication of direct derivation from either candesco ‘to grow light/white’ or excandeo ‘to shimmer brightly’ because both verbs contain the same base, cand-, and a cell for a verb showing the schema [ex[x]V sc]V is already available in the paradigm. Another concept that we borrow and adapt from CxM theory is that of the ‘constructicon’, the collection of all lexical and grammar knowledge of a language. The constructicon is built as a hierarchical lexicon, where each construction schema can define a series of generalisations about subsets of words. The schema can be further specified so as to contain intermediate levels of abstraction in between the most general template and the individual existing lexemes. Accordingly, schemas can be further described to account for restrictions in the word-formation process and for changes in semantic value. This concept becomes useful in the construction of a new resource when we want to detail the evolution of meaning or morphological restrictions with information that does not fit within the restricted cell space. See, for example, in Figure 6.2 the detail with which it is possible to describe the morpho-semantic characteristics of -sc- suffixed verbs with a hierarchical schema. The top box in Figure 6.2 provides the motivation for the morphological and semantic properties of Latin verbs containing the -sc- suffix. This hierarchical construction presents two sub-levels: the first sub-level describes a semantic shift in the specific x-sc- construction indicating that the meaning of verbs containing suffix -sc- does not always have the semantic value of ‘to begin/to become’, which is its core meaning (Haverling, 2000). Furthermore, schemas can be further specified to mark restrictions in the word-formation process, indicating for example if a certain base displays certain morphological characteristics. However, as far as Latin is concerned, it might not be always possible to achieve a straightforward distinction down to the morphological sub-level in the schema. Once again, if we reject assumptions on the directionality of certain derivations, we could possibly state that dormisco ‘to fall asleep’ can be specified by [[x]V sc]V, but this last specification might not be so easy to achieve in other cases. For example, to state that gelasco ‘to freeze’ comes from

What We Talk about When We Talk about Paradigms

[[x]Vj sc]Vj acesco cassesco fervesco ...

139

[[x]j sc]Vi

[[x]j sc]Vi

[to begin to/become SEMj ]Vi

[to begin to/become SEMj ]Vi

[[x]Nj sc]Vj arboresco carbonesco noctesco ...

[[x]Aj sc]Vj crudesco fortesco iuvenesco ...

Figure 6.2 Hierarchical schematic construction for suffixed verbs in -sc-

the verbal root for gelo ‘to turn to ice, to freeze’ and not that for gelu ‘frost’, noun, is to make assumptions on directionality. The CxM model readily offers a number of strategies that can help to solve the theoretical issues that we found difficult to explain with an I&A model. One of these is the concept of schema unification: schemas can be unified to account for formations without the need to postulate non-existing words. Hence the schema [co(n)[x]]V + [[x]sc]V = [co(n)[x]sc]V justifies not only the concurrent existence of collibesco, concupisco, consudesco, which at the moment can be found in WFL among V-to-V -sc- derivations, and of cohorresco, currently listed as a result of V-to-V co(n)- (because cohorreo does not exist). Schema unification also facilitates compounding.6 Consider for example the following schema: [[x]Vi [y]Vj ]Vk [to SEMj SEMi ]Vk

[[x]Vi [facio]Vj ]Vk

[[x]Vi [fio]Vj ]Vk

[to makej SEMi ]Vk

[to be madej SEMi ]Vk

calefacio

calefio

candefacio ...

candefio ...

Figure 6.3 Partial hierarchical schema for V + V = V compounds 6  On paradigmatic relations in compounding, see also the chapters by Bagasheva (this volume) and by Radimský (this volume).

140

Litta and Budassi

The schema in Figure 6.3 does not only account for the construction of V + V = V compounds in Latin, but allows one to make a semantic distinction between compounds formed with a certain second constituent over another. Compounding in Latin, however, is a complicated issue: although the number of compounds is rather low compared to other languages (see Micheli & Litta, 2017), there is often a fine line between word formation and etymology, and between compounding and the grammaticalisation of the second constituent, between transparent constituents and less-than-transparent ones (consider for instance, all the adjectives ending in -ficus, or all the nouns ending in -fex, -plex, etc.), so theorising the application of CxM-inspired labels to the WFL paradigms deserves more attention. In addition, a CxM approach to paradigmatic word formation can help with describing certain semantic alternations such as caleo ‘to be hot’ ~ calesco ‘to become hot’ calefacio ‘to make something/someone hot’ ~ calefio ‘to be made hot’ or liqueo ~ liquesco ~ liquefacio ~ liquefio. In this simple case, two oppositions are identifiable: (i) that between a ‘basic’ (i.e., non-suffixed) verb and a suffixed verb (namely, caleo ~ calesco and liqueo ~ liquesco), and (ii) that between those two verbs and a compound verb in -facio (active) and -fio (passive), namely, calefacio/calefio and liquefacio/liquefio. When such alternations occur in the paradigm, there is almost always also an adjective in -idus. This happens coherently with what is described by Pultrová (2007): -idus belongs to the so-called “set of suffixes” (Suffixverband), that is, a group of suffixes by which nouns/adjectives/verbs are derived from one root of specific type. The peculiarity of these derivations is that it is indistinguishable which are primary and which secondary derivations (to put it another way, which words are derived from which within the given suffix family).7 In Latin, from the synchronic standpoint, such a system of suffixes is made of adjectival -idus, substantival -or, verbal -ēre/-ēscere suffixes. This issue can be faced differently if considered from the perspective of W&P + CxM. Table 6.4 sketches the relevant portion of a paradigm built on the base CAL-.

7  Note how, in Indo-European linguistics, among the “system of suffixes” is the so-called “Caland system of suffixes”, which includes suffixes *-ro-/*-mo-, stative verbs in *-eh1-, neuters in -es-, comparatives in *-i̯os-, superlatives in *-is-to- (Pultrová 2007).

141

What We Talk about When We Talk about Paradigms Table 6.4 Sub-paradigm of CAL-

CAL-

V

caleo Morphology [x]Vαi Semantics [to SEM]Vi

-id-

-sc-

-facio

calidus [[x]iid]Aj [having the characteristics of SEM i]Aj

calesco calefacio [[x]i sc]V [[x]Vi[facio]Vj]Vk [to become [to makej SEM j]Vk SEM i]j/ [to be SEM i]j

-fio calefio [[x]Vi[fio]Vj]Vk [to be madej SEM j]Vk

The rows labelled ‘Morphology’ and ‘Semantics’ in Table 6.4 show how the semantic (as well as morphological) relationship existing among caleo ~ calidus ~ calesco ~ calefacio ~ calefio is neither linear nor directional in nature. What is ‘x’ in CxM schemas is actually not caleo, but cal-, and the same thing is valid for SEM. This pattern occurs in quite a number of cases, that we have extracted from the WFL lexical base. Below is a summary of sub-paradigms modelled on exactly the same behaviour: Table 6.5 List of sub-paradigms with the same derivational behaviour of cal-

aboleo areo caleo candeo expergo ferueo frigeo labo langueo liqueo madeo niteo obduro obsoleo oleo pateo paueo pudeo

calidus candidus feruidus frigidus labidus languidus liquidus madidus nitidus

olidus pauidus

abolesco aresco calesco candesco expergiscor feruesco frigesco labasco/labesco languesco liquesco madesco nitesco obduresco obsolesco olesco patesco pauesco pudesco

abolefacio arefacio calefacio candefacio expergefacio/expergifico feruefacio frigefacio labefacio languefacio liquefacio madefacio nitefacio obdurefacio obsolefacio olifacio/olfacio patefacio pauefacio pudefacio

abolefio calefio candefio expergefio feruefio frigefio lebefio languefio liquefio madefio nitefio obsolefio patefio pauefio pudefio

142

Litta and Budassi

Table 6.5 List of sub-paradigms with the same derivational behaviour of cal- (cont.)

puteo putreo rigeo rubeo stupeo tabeo tepeo timeo torpeo torreo tremo tumeo ualeo umeo

putidus putridus rigidus rubidus stupidus tabidus tepidus timidus torpidus torridus tremidus tumidus ualidus umidus

putesco putresco rigesco rubesco stupesco tabesco tepesco timesco torpesco torresco tremesco tumesco ualesco umesco

putefacio putrefacio rigefacio rubefacio stupefacio tabefacio tepefacio timefacio torpefacio torrefacio tremefacio tumefacio ualefacio umefacio

putefio putrefio rigefio rubefio stupefio tepefio

tremefio tumefio ualefio umefio

Much more could be said about this sub-paradigm (e.g., there are some paradigms which do not have the -fio counterpart of -facio forms, such as torpeo ~ torpidus ~ torpesco ~ torpefacio, or which show neither the -idus adjectives nor the -fio form, but that are featured by a -fico and not a -facio form, such as augeo ~ auctus ~ augesco ~ augifico), but we would run the risk to exceed the limits of this paper. Remarkably enough, just note how in a few cases there is no -sc- verb corresponding to a -facio/-fio formation: Table 6.6 List of sub-paradigms with no -sc-form corresponding to a -facio/-fio formation

allicio caueo condoceo depono incendo perterreo

allicefacio cauefacio condocefacio deponefacio incendefacio perterrefacio

allicefio cauefio condocefio deponefio incendefio perterrefio

The last row in Table 6.6 shows the case of perterreo, which shows no -sc- form belonging to its “set of suffixes”. Note, however, that within a W&P model, things could come out much differently (Table 6.7).

143

What We Talk about When We Talk about Paradigms Table 6.7 Sub-paradigm of terr-

TERR-

V

-sc-

-facio

-fio

terreo terresco Morphology [x]Vαi [[x]i sc]V Semantics [to SEM]Vi [to become SEM i]j/[to be SEM i]j perperterreo perterrefacio Morphology [per[x]]Vαi [per [x]Vi[facio]Vj]Vk Semantics [to SEM]Vi [to makej SEM j through/intensively]Vk

perterrefio [per [x]Vi[fio]Vj]Vk [to be madej SEM j through/intensively]Vk

What emerges from Table 6.7 is that, although a form *perterresco is not attested in Latin, terresco ‘to terrify’ occurs in texts. This shows well how forms perterrefacio ‘to make extremely frightened’ and perterrefio ‘to be made extremely frightened’ could be in any case considered part of the above-described schema of “set of suffixes”, since the morpho-semantic slot destined to *perterresco was already filled in the language by terresco, a form belonging to the same derivational paradigm of perterreo, perterrefacio and perterrefio. In other cases, there is no simple verb in the paradigm: assuesco ~ assuefacio ~ assuefio, consuesco ~ consuefacio ~ consuefio, desuesco ~ desuefacio ~ desuefio, insuesco ~ insuefacio ~ insuefio. Note that these compounds cannot be connected in WFL due to the impossibility to recognise the first constituent morphotactically. In our model, they would be represented as part of the paradigm headed by the base SUE- as follows: Table 6.8 Sub-paradigms based on base sue-

sue-

Morphology Semantics adMorphology

V -scsuesco [[x]i sc]V [to become SEM i]j/[to be SEM i]j assuesco [ad [x]i sc]V

-facio

-fio

assuefacio [ad [x]Vi[facio]Vj]Vk

assuefio [ad [x]Vi[fio]Vj]Vk

144

Litta and Budassi

Table 6.8 Sub-paradigms based on base sue- (cont.)

sueSemantics

conMorphology Semantics deMorphology Semantics in1-a Morphology Semantics

V -sc[to become intensely SEM i]j/ [to be intensely SEM i]j consuesco [con [x]i sc]V [to become SEM i with]j/ [to be SEM i with]j desuesco [de[x]i sc]V [to become SEM i from]j/ [to be SEM i from]j insuesco [in1 [x]i sc]V [to become SEM i in]j/ [to be SEM i in]j

-facio

-fio

[to makej intensely SEM j]Vk

[to be madej intensely SEM j]Vk

consuefacio [con[x]Vi[facio]Vj]Vk [to makej SEM j with]Vk desuefacio [de[x]Vi[facio]Vj]Vk [to makej SEM j from]Vk insuefacio [in[x]Vi[facio]Vj]Vk [to makej SEM j in]Vk

consuefio [con [x]Vi[fio]Vj]Vk [to be madej SEM j with]Vk desuefio [de[x]Vi[fio]Vj]Vk [to be madej SEM j from]Vk insuefio [in[x]Vi[fio]Vj]Vk [to be madej SEM j in]Vk

a There are two homograph prefixes in- in Latin. In this paper, we distinguish between in1meaning ‘entering’, and in2- meaning negation.

An all-round model on derivational morphology can be used also to solve semantic as well as morphology-related questions. This claim seems questionable on the basis of a complete paradigm. But consider Table 6.9: Table 6.9 Sub-paradigm of ir-

IR-

Morphology Semantics

V

-at

-sc

[x]Vαi [to SEM i]Vj

iratus [[x]at]Aαi [possessing SEM i]Aαj

irasco(r) [[x]i sc]V [to become SEM i]j / [to be SEM i]j

The empty box in column V signals how linearity and directionality must be neglected in approaching similar issues. Otherwise, fictional lemmas or unattested lemmas should be posited to account for “further” steps in derivational terms (*iro).

145

What We Talk about When We Talk about Paradigms

CxM can also help in the case of stem allomorphy. Consider, for example, the following recurrent patterns: Table 6.10 Stem allomorphy in ago and duco

Base

V

PREFIX (per) + V

ADJ

ag-/actduc-/duct-

ag-o duc-o

per-ag-o per-duc-o

act-us duct-us

This short example can tell us a lot concerning the derivational patterns of allomorphic stems as ag-/act- or duc-/duct-: i. present indicative forms are derived from the ag-/duc- base; ii. the same stem appears also in prefixed verbs; iii. the act-/duct- base, on the other hand, is used in adjectives. In another example, cogito ‘to think’ and coacto ‘to compel’, within the same morphological family sharing the base (AG-, from ago ‘to drive’), present the same construction [co(n)[x]]V, but the morphology of the base completely changes both the shape and the meaning from one verb to the other; this is because cogito contains the stem of the present (ag-), subsequently incurring in vocalic change and in the loss of a due to the presence of another vowel in the prefix (co(n)-), while coacto uses the stem of the perfect tense (act-). The concept of a hierarchical lexicon is once again fundamental because all variations can be accounted for, even in the case of phonologically-conditioned allomorphy, as often happens in Latin with prefixed bases. In a paradigmatic model based on CxM, the morphology of the stem can not only be specified in subschemas (as shown in Figure 6.3), but also in the cell itself, provided that the alternation in morphological features is paired with a definite semantic shift, whereby all verbs prefixed with co- displaying the stem of the perfect [co(n)[x]xi (i)t]Vj [to SEM with iteratively]j

[co(n)[x]𝛼𝛼i (i)t]Vj

[co(n)[x]𝛽𝛽i (i)t]Vj

[to SEM with iteratively]j

[to SEM with iteratively]j

cogito

coacto

Figure 6.4 Partial hierarchical schema for [co(n)[x](i)t]V formations

146

Litta and Budassi

tense show similar semantic features that are, however, significantly different from those showing the stem of the present. 4

WFL Paradigms

The first requirement to build a derivational paradigmatic model for Latin is data since derivational paradigms are fundamentally evidence-driven. To this end, we used the data contained in WFL, which has the double advantage of being a complete but closed set of lexemes, already enhanced with metadata containing information on morphology and word formation. However, given that this metadata was added with a different take on derivational morphology, we had to employ a very specific methodology. As a first step, we extracted a list of the most frequent word-formation rules from the WFL database (Table 6.11). Table 6.11 Most frequent WFRs in Classical Latin

WFL rule type

No. of lemmas per WFL rule

WFL rule type

No. of lemmas per WFL rule

V-to-N -tio V-to-N -tor Conversion N-to-V conversion V-to-N conversion N-to-A -osus A-to-N -tas V-to-V conN-to-A -alis A-to-A in- (neg) V-to-V exN-to-A -arius V-to-A in- (neg) V-to-V in- (ent) A-to-N conversion V-to-V deV-to-N -trix V-to-V re-

2515 1415 1114 739 730 560 556 553 544 501 479 471 457 452 445 418 417 409

N-to-A -atus N-to-N -arius N-to-N -ulus/a/um V-to-A -bilis V-to-V adN-to-A -icus N-to-A conversion V-to-V -(i)to V-to-V perN-to-A -inus V-to-N -ia/-ium V-to-A -ivus V-to-V praeV-to-V subN-to-A -eus A-to-N -ia V-to-V obV-to-I -im

405 390 387 376 374 343 339 334 308 304 289 288 267 263 245 214 209 206

What We Talk about When We Talk about Paradigms

147

Table 6.11 shows a list of the most frequent WFRs in Classical Latin followed by the number of lexemes per rule contained in WFL.8 Rules are named by their part of speech category of change (V-to-N, V-to-V, A-to-N, etc.) together with the affix involved, if present. Hence the “V-to-N (t)io” WFR outputs all deverbal nouns characterised by suffix -(t)io. Conversion rules involve only a change in part of speech without the addition of any affix. This table excludes compounds, as these will be separately considered once the main model for derivation has been implemented. These numbers are valuable because they give us an idea of the most used formations in Latin. However, in order to adapt this information to the derivational paradigm the following must be considered: i. Values emerging from Table 6.11 do not necessarily correspond to the actual number of lexemes displaying the same formative elements. For example, the most productive WFR in Latin was that producing action nouns ending in -(t)io from verbs (see Litta, Passarotti, & Ruffolo, 2017). In WFL, this rule accounts for 2515 nouns, but in the full lexical basis third declension nouns ending in -(t)io are actually 2658. This is because a number of these have been connected to their input lexemes through other WFRs for specific reasons. For instance, exauguro ‘to deconsecrate’ > exauguratio ‘deconsecration’, but articulatio ‘joint structure’ > exarticulatio ‘dislocation’ because the verb exarticulo does not exist. Within the paradigm, however, exauguratio and exarticulatio will be inserted in the same cell labelled by the schema [[ex [x]]i (t)io]jN3 ↔ [action of SEM i out]jN3; similarly, articulatio and exarticulatio are characterised by two different formations ([[x]i (t)io]jN3 and [[ex [x]]i (t)io]jN3) and will need to be placed in two different cells. ii. In making assumptions on directionality of derivations, WFRs involving conversion will need to be disregarded. This means that the lexemes involved in the V-to-N conversion rule, for example sono ‘to make a sound’ > sonus ‘sound’, will merely be included in different cells as [x]V1 and [x] Nm, respectively; iii. Because the lexical basis used to build WFL considers past participles as being part of the inflexional paradigm of a verb, some nouns that are generally seen as having derived from (participial) adjectives are given as deriving straight from verbs, so, for example, in “V-to-A in (neg)” the resulting adjective will be listed as simply [ink[x]i]Aj ↔ [not SEM i]Aj. 8  For this study, we arbitrarily chose to cut the count at 200 lemmas; if a certain formation appears to be missing from the core paradigm drawn in this paper, it is because this appears less than 200 times in Classical Latin.

148

Litta and Budassi

Beyond these criticalities, however, WFL represents a solid starting point to approach the reconstruction of a core derivational paradigm for Latin. As a matter of fact, despite a few fluctuations caused by what is described in point i. above, quantitative data as offered by WFL are consistent enough to be employed. Consider WFRs “V-to-N tio”, “V-to-N tor”, “N-to-A osus” and “A-to-N tas”. These rules belong to the top 10 most frequent WFRs. Quantitative differences are as follows: i. V-to-N tio, value in Table 6.11: 2515, actual -tio lemmas: 2658; ii. V-to-N tor, value in Table 6.11: 1415, actual -tor lemmas: 1556; iii. N-to-A osus, value in Table 6.11: 560, actual -osus lemmas: 629; iv. A-to-N tas, value in Table 6.11: 556, actual -tas lemmas: 582. The difference between values reported in Table 6.11 and actual lemmas belonging to a given class is on average ca. 7.5%, in favour of the number of actual lemmas. This difference seems neglectable, and even more so given the fact that it favours actual lemmas. The data offered by WFL, for instance, tell us that at least 544 families in Classical Latin contain adjectives featured by suffix -alis, a relevant fact for the construction of a core paradigm. Table 6.11 can thus be confidently used as a starting point from which a core derivational paradigm can be drawn. The core paradigm was built following these procedures: i. suffixes and prefixes were aligned on the X and Y axes respectively, in order of appearance in the frequency list; ii. lexemes resulting from conversions are added to the paradigm, be that in a cell dedicated to simple verbs, nouns or adjectives (currently arbitrarily placed in the first three slots of the X axis), or to a more complex construction; iii. crossroad cells of X/Y axis projections constitute the first bulk of the core paradigm. Where there is no evidence for a construction, the cell is empty. This is only possible because Latin is a dead language and the creation of new words will never occur; iv. cells are labelled using schemas inspired by CxM models, containing both morphological and semantic information. However, these schemas are only generic and need further specification outside the table. This is because it is extremely difficult to pinpoint an always-valid semantic value for Latin affixes, since their value often evolved across time. In this initial design of a core paradigm for Latin, we have used conventional mark-up in order to distinguish cells that display a certain degree of “regularity” from another. Empty cells indicate lack of evidence for a given construction, which means that there are no words, in our lexical base, that could slot

What We Talk about When We Talk about Paradigms

149

Table 6.12 Sketch of point iv. Towards the construction of a core derivational paradigm for Latin

[x]Vj ↔ [SEM]j

[x]N ↔ [SEM]

[x]A ↔ [SEM]

[[x](t)io] N↔ [action of SEM] [[con [x]] [con [x]] [con [x]] [con [x]]A ↔ [co-SEM Vj ↔ N↔ j (t)io]N ↔ (simultaneity, [action of [to SEM [SEM] joint action)] SEM j with] with]j [in2 [x]] [in2 [x]]A ↔ [[in2 [x]] [not SEM] ↔ Vj j (t)io]N ↔ [to not [action of SEM]j SEM j] [ex [x]] [ex [x]] [ex [x]]A ↔ [[ex [x]] [without SEM] j (t)io]N ↔ V↔ N↔ [to SEM [SEM] [action of out/away]j SEM j] … … … …

[[x](t)or]N ↔ [[x]]j os] [one who A↔ SEM] [full of SEM j] [[con [x]]j [[con [x]]j os]A ↔ (t)or]N ↔ [one who [full of SEM j with] SEM j] [[in2 [x]]j [[in2 [x]]j (t)or]N ↔ os]A ↔ [one who [not full SEM j] of SEM j] [ex [x](t)or] [[ex [x]]j os]A ↔ N↔ [one who [SEM j] SEM j] … …

…

…

…

…

…

in a cell labelled by the intersection of two or more given affixes. Cells in bold indicate a consistent attestation of a given construction. Cells in italics mark a less consistent attestation of a given construction. Cells considered with a “consistent attestation” (thus in bold) represent those constructions which appear in more than 200 lemmas in WFL. Cells considered with a “less consistent attestation” (thus in italics) represent those constructions which appear in less than 200 lemmas in WFL. The full core paradigm achieved through this method is available in the Appendix. In order to test the value of the method that has been used to build the core paradigm, we have attempted to fill most of its cells with lemmas from four among the richest morphological families in Latin, that is, FAC-, DIC-, AG-, MITT- (Table 6.13). The data has been extracted once again from WFL. This time however, WRFs are completely disregarded, in favour of allocating lemmas to the right cell in the paradigm on the basis of constructions modelled on CxM theory. In this case, the Latin core derivational paradigm contained in the Appendix has been portrayed not as a grid but in a single column. This operation was performed by listing one row at a time (so the second row, starting

150

Litta and Budassi

with [con [x]]Vj ↔ [to SEM with]j, follows immediately the first row ending with [[x]]j im]I ↔ [in a SEM j manner]). Table 6.13 Core derivational paradigm for Latin filled with lemmas with bases fac-, dic-, ag-, mitt-

BASE [X]

fac-

dic-

ag-

mitt-

[x]V ↔ [SEM]V [x]N ↔ [SEM]N

facio factus factum

dico dictus dictum

ago actus actum

mitto missa missus

factio factor

dicio dictor

actio actor actuosus actuarius actrix actuaria actuarius

missio missor

[x]A ↔ [SEM]A [[x](t)io]N ↔ [action of SEM]N [[x](t)or]N ↔ [one who SEM]N [[x]]j os]A ↔ [full of SEM j]A [[x]j ari]A ↔ [having qualities of SEM j] [[x]j trix]]N ↔ [a female who SEM j]N [[x]j ari]N ↔ [[dealer in SEM j]N [[x]j bil]A ↔ [able/worthy to be SEM j]A [[x]j ic]A ↔ [belonging to SEM j]A [[x]j (i)t]V ↔ [to SEM j repetitively]V

factrix factionarius dicibilis

facticius facto factito

dicto dictito

conficio

condico

[[con [x]]j (t)io]N ↔ [action of SEM j with]N

confectio

[[con [x]]j (t)or]N ↔ [one who SEM j with]N [[con [x]]j trix]]N ↔ [a female who SEM j]N [[con [x]]j ari]N ↔ [SEM j]N [[con [x]]j ic]A ↔ [SEM j]A [[con [x]]j (i)t]V ↔ [to SEM j repetitively]V

confector

condicio condictio condictor coactor

confectrix

coactrix

[[x]j iv]A ↔ [related to SEM j]A [con [x]]V ↔ [to SEM with]V [con [x]]N ↔ [SEM]N

[[con [x]]j im]I ↔ [SEM j]I [[in2 [x]]j os]A ↔ [not full of SEM j]A [ex [x]]V ↔ [to SEM out/away]V

efficio

edico

actito activus cogo coactus coacta coactio

coactarius coacticius cogito coacto coactim inactuosus

missibilis missicius missito

committo

commission commissor

emitto

151

What We Talk about When We Talk about Paradigms Table 6.13 Core derivational paradigm for Latin filled with lemmas (cont.)

BASE [X]

fac-

[ex [x]]N ↔ [SEM]N

effectus effectum effectio effector effectuosus effectrix

[[ex [x]]j (t)io]N ↔ [action of SEM j]N [ex [x](t)or]N ↔ [one who SEM j]N [[ex [x]]j os]A ↔ [SEM j]A [[ex [x]]j trix]]N ↔ [a female who SEM j]N [[ex [x]]j (i)t]V ↔ [to SEM j repetitively]V [[ex [x]]j iv]A ↔ [related to SEM j]A [[ex [x]]αj i]N ↔ [result of action of SEM j]N [[ex [x]]j im]I ↔ [SEM j]I [in1 [x]]V ↔ [to SEM in]V [in1 [x]]N ↔ [SEM]N [in1 [x]]A ↔ [SEM]A [[in1 [x]]j (t)io]N ↔ [action of SEM j] [[in1 [x]]j (t)or]N ↔ [one who SEM j] [[in1 [x]]j trix]]jN ↔ [a female who SEM j] [[in1 [x]]j (i)t]V ↔ [to SEM j repetitively]V [[in1 [x]]j iv]A ↔ [related to SEM j]A [de [x]]V ↔ [to SEM away from]j [[de [x]]j (t)io]N ↔ [action of SEM j] [[de [x]]j ic]A ↔ [SEM j]A [re [x]]V ↔ [to SEM again]V [re [x]]N ↔ [SEM]N [[re [x]]j (t)io]N ↔ [action of SEM j again]N [[re [x]]j (t)or]N ↔ [one who SEM j]N [[re [x]]j ari]A ↔ [SEM j]A [[re [x]]j bil]A ↔ [able/worthy to be SEM j]A [[re [x]]j iv]A ↔ [related to SEM j]A [ad [x]]V ↔ [to SEM towards]V [ad [x]]N ↔ [SEM (intensifier)]N

effectivus efficentia

inficio infectus inficiens infectus infectio infector infectrix infecto infectivus

reficio refectus refectio

dic-

ag-

emissus edictio

emissio emissor

edicto

indico

immitto immissus

indictio

immissio immissor

indictivus dedico

redico

reago

refector

afficio affectus

mitt-

demitto demissio demissicius remitto remissa remissio remissor remissarius remissibilis

addico addicta addictus

adigo adactus

remissivus admitto admissus admissum

152

Litta and Budassi

Table 6.13 Core derivational paradigm for Latin filled with lemmas (cont.)

BASE [X] [ad [x]]A ↔ [SEM (intensifier)]A [[ad [x]]j (t)io]N ↔ [action of SEM j]N [[ad [x]]j (t)or]N ↔ [one who SEM j]N

[[ad [x]]j os]A ↔ [SEM j]A [[ad [x]]j trix]]N ↔ [a female who SEM j]N [[ad [x]]j (i)t]V ↔ [to SEM j repetitively]V [[ad [x]]j iv]A ↔ [related to SEM j]A [per [x]]V ↔ [to SEM through]V [per [x]]N ↔ [SEM]N

[per [x]]A ↔ [very SEM (intensive)]A [[per [x]]j (t)io]N ↔ [action of SEM j]N [[per [x]]j (t)or]N ↔ [one who SEM j]N [[per [x]]j trix]]N ↔ [a female who SEM j]N [[per [x]]j iv]A ↔ [SEM j]A [prae [x]]V ↔ [to SEM ahead]V [prae [x]]N ↔ [SEM]N [[prae [x]]j (t)io]N ↔ [action of SEM j]N [[prae [x]]j (t)or]N ↔ [one who SEM j]N [sub [x]]V ↔ [to SEM down/under]V [sub [x]]N ↔ [sub-SEM (rank)]N [[sub [x]]j (t)io]N ↔ [action of SEM j]N [[sub [x]]j i]N ↔ [result of action of SEM j]N [[sub [x]]j im]I ↔ [SEM j]I [ob [x]]V ↔ [to SEM against]V [ob [x]]N ↔ [SEM]N [[ob [x]]j (t)io]N ↔ [action of SEM j]N [[ob [x]]j (t)or]N ↔ [one who SEM j]N

fac-

dic-

ag-

mitt-

affectio affectator [[[ad [x]]j at] (t)or]N affectuosus affectatrix [[[ad [x]]j at] trix]]jN affecto(r) affectivus perficio perfica perfectus perfectum perficus (perfectus) perfectio perfector perfectrix perfectivus praeficio praefica praefectus praefectio praefector sufficio

addictio addictor

adactio

admissio admissor

perdico

perago

suffectio sufficentia

officio offectus offector

peractio

admissivus permito permissus permissum

permissio permissor

praedico

praemitto

praedictio

praemissio submitto submissus submissio

submissim omitto omissio

What We Talk about When We Talk about Paradigms

153

The first thing clearly emerging from Table 6.13 is that it is not a given that lemmas out of four of the richest morphological families are enough to fulfill the core paradigm (please refer to Appendix to see the entire core paradigm). For example, the whole area containing lemmas formed with the negative prefix in- seems not to be inhabited. However, this might be due to semantic restrictions on the bases that we have chosen, and more investigation on this matter is needed. Furthermore, Table 6.13 highlights that relying on how the data is displayed in WFL right now can be problematic when creating this new model for Latin derivational morphology. Word-formation rules, as they are conceived in the I&A version of WFL, reflect only the last part of the word-formation process, mainly the last affix added to the word, but do not reflect the entire structure of the word. For example activitas is included among -tas nouns, but it is actually formed as [[act]iv](i)ta]N, dictator is formed as [[dic](i)t]at]or], so it cannot be allowed to slot into the [[x](t)or]N cell. When we collect data on the fact that A-to-N -tas nouns are among the most frequent, it is actually necessary to climb up the derivation tree-graph in order to consider all processes that led to a specific formation. Therefore, because the cell has been drawn to reflect a WFR that is based solely on the addition of suffix -tas, the [[x]]Aj tas]N ↔ [characteristic of SEM j], it cannot represent many -tas formations. It seems therefore necessary to take a step back, and reconsider that way we want the paradigm to be assembled: a different approach, based on the number of elements constituting a word rather than the WFRs it results from may seem preferable to better account for Latin derivational morphology processes. 5

Conclusions: a New Word Formation Latin

In this paper, we described a new model to portray paradigmatic behaviours underlying derivational morphology. After a short introduction on the existing Word Formation Latin lexicon, a language resource based on a morphotactic approach to I&A principles, we discussed a group of theoretical issues concerning derivational morphology which the resource fails to describe in a consistent way. During the process of designing a full paradigmatic model for derivational morphology, however, we have established that the cornerstones of our theory must be the circumvention of directionality, the central role of the cell, and the involvement of semantics in support of morphology. A paradigmatic approach to derivational morphology on the one hand resolves a few relevant theoretical issues which a morphotactic approach demonstrated to fail to account for. On the other hand, a paradigmatic model must lead to a

154

Litta and Budassi

more fine-grained representation of words’ components. This led us to consider and add elements borrowed from CxM theories to the cells in our model. CxM theories fit to our purpose because they are evidence-driven in nature, and because they are based on a solid depiction of morphology and semantics at the same time. Additionally, we have endeavoured to attempt the construction of a ‘core’ paradigm for Latin derivational morphology with data induced from the I&A version of WFL as a starting point. However, some critical issues remain to be analysed: a better version of the core paradigm still needs to be theorised. This operation requires to pass through a deep consideration of data extracted from the I&A version of WFL, which will be a matter of investigation for future work. Also, a new version of WFL could be built according to the above-described theoretical framework. The new version of WFL would need to include a new graphical interface displaying morphological families inside an interactive grid. Ideally this grid could be rearranged by the user according to his own research needs. The focal point anyhow is that each cell inside the grid will contain a label based on CxM models, which will specify both the morphology and the semantics of a given lexeme. Such a grid, or table, in order to be dynamic and interactive, must include the following functionalities: i. The user will be able to search a given lexeme and obtain its morphological family mapped on the grid. ii. By clicking on the label in each cell, the user will be redirected to the relevant section within the resource, detailing the construction(s) within hierarchical schemas. iii. Hierarchical schemas will also be interactive. Clicking on every level or variation of the schema will produce a list of lexemes that follow that specific construction. iv. The grid will be consultable also in a ‘just-labels’ form. This will enable the user to select a certain number of cells and extract all those morphological families displaying the chosen number of constructions. In order to complete the compilation of the paradigm, an enhancement of the methodology used above will be tested on a larger scale and with more data, using techniques that need to be obligatorily inherited by computational linguistics methods, paired with thorough human verification of results: the exploitation of the metadata contained in WFL and in its lexical basis will be implemented in a semi-automatic pipeline in order to favour data re-deployment. The biggest challenge ahead, however, seems to be the realisation of the semantic portion of the CxM based WFL paradigm, which might require the

What We Talk about When We Talk about Paradigms

155

input of a larger community of scholars. The implementation of a crowdsourced platform to enrich the semantic portion of the new WFL is therefore a fundamental desideratum.

Acknowledgements

Both authors share responsibility for the contents of this article. Nevertheless, Paradigm Design was largely contributed by Marco Budassi, while Construction Morphology by Eleonora Litta. Sections Introduction, WFL Paradigms, and Conclusions are the work of both authors. We are grateful to Greta Franzini, Silvia Micheli, and Marco Passarotti for their invaluable advice on form and meaning. WFL received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No658332-WFL, and is available to browse at http://wfl.marginalia.it, and to download as part of the morphological analyser and lemmatiser for Latin Lemlat (Passarotti, Budassi, Litta, & Ruffolo, 2017) at https://github.com/ CIRCSE/LEMLAT3. References Bagasheva, A. (this volume). Paradigmaticity in compounding. In J. FernándezDomínguez, A. Bagasheva, & C. Lara-Clares (Eds.), Paradigmatic relations in word formation (pp. 21–48). Leiden: Koninklijke Brill NV. Booij, G. (2009). Compounding and Construction Morphology. In R. Lieber & P. Štekauer (Eds.), The Oxford handbook of compounding (pp. 201–216). Oxford: Oxford University Press. DOI: 10.1093/oxfordhb/9780199695720.001.0001. Booij, G. (2010). Construction Morphology. New York: Oxford University Press. Budassi, M., & Litta, E. (2017). In trouble with the rules: Theoretical issues raised by the insertion of -sc- verbs into word formation Latin. In E. Litta & M. Passarotti (Eds.), Proceedings of the Workshop on Resources and Tools for Derivational Morphology (DeriMo) (pp. 15–26). Milano: Educatt. Budassi, M., & Passarotti, M. (2016). Nomen Omen. Enhancing the Latin morphological analyser Lemlat with an onomasticon (pp. 90–94). In B. Alex & N. Reiter (Eds.), Proceedings of the 10th SIGHUM Workshop on Language Technology for Cultural Heritage, Social Sciences, and Humanities (LaTeCH 2016). Berlin: The Association for Computational Linguistics.

156

Litta and Budassi

Forcellini, E. (1940). Egidio Forcellini. 1940. Lexicon Totius Latinitatis ad Aeg. Forcellini lucubratum, dein a jos. Furlanetto emendatum et auctum; nunc demum Fr. Corradini et jos. Perin curantibus emendatius et auctius meloremque in formam redactum adjecto altera quasi parte Onomastico totius latinitatis opera et studio ejusdem jos. Perin. Padova: Typis Seminarii. Fradin, B. (this volume). Characterizing derivational paradigms. In J. FernándezDomínguez, A. Bagasheva, & C. Lara-Clares (Eds.), Paradigmatic relations in word formation (pp. 49–84). Leiden: Koninklijke Brill NV. Georges, K. E., & Georges, H. (1913–1918). Ausführliches lateinisch-deutsches Hand­ wörterbuch. Hannover: Hahn. Glare, P. G. W. (1982). Oxford Latin dictionary. Oxford: At the Clarendon Press. Gradenwitz, O. (1904). Laterculi vocum latinarum. Leipzig: Hirzel. Haverling, G. (2000). On sco-verbs, prefixes and semantic functions. A study in the development of prefixed and unprefixed verbs from Early to Late Latin. Göteborg: Göteborg University Press. Hockett, Ch. F. (1954). Two models of grammatical description. Words, 10, 210–231. Jackendoff, R. (1975). Morphological and semantic regularities in the lexicon. Language, 51(3), 639–671. Litta, E., Passarotti, M., & Ruffolo, P. (2017). Node formation: Using networks to inspect productivity in affixal derivation in Classical Latin. In A. Antonacopoulos & M. Büchler (Eds.), Proceedings of the 2nd International Conference on Digital Access to Textual Cultural Heritage (DATeCH2017) (pp. 103–108). New York, NY: ACM. Litta, E., & Passarotti, M. (2019). (When) Inflection Needs Derivation: A Word Formation Lexicon for Latin. In N. Holmes, M. Ottink, J. Schrickx, & M. Selig (Eds.), Words and Sounds (pp. 224–239). Berlin, Boston: De Gruyter. Micheli, S., & Litta, E. (2017). E Pluribus Unum. Representing compounding in a derivational lexicon of Latin. In R. Basili, M. Nissim, & G. Satta (Eds.), Proceedings of the Fourth Italian Conference on Computational Linguistics (CLiC-it 2017) (pp. 212–216). Rome: aAccademia University Press. Passarotti, M., Budassi, M., Litta, E., & Ruffolo, P. (2017). The Lemlat 3.0 package for morphological analysis of Latin. In G. Bouma & Y. Adesam (Eds.), Proceedings of the NoDaLiDa 2017 Workshop on Processing Historical Language (pp. 24–31). Gothenburg: Linköping University Electronic Press. Pultrová, L. (2007). The Latin adjectives with the suffix -idus. Studia Minora Facultatis Philosophicae Universitatis Brunensis, 12, 87–95. Radimský, J. (this volume). A paradigmatic approach to compounding. In J. FernándezDomínguez, A. Bagasheva, & C. Lara-Clares (Eds.), Paradigmatic relations in word formation (pp. 164–185). Leiden: Koninklijke Brill NV.

What We Talk about When We Talk about Paradigms

157

Ševčíková, M., & Žabokrtskỳ, Z. (2014). Word-formation network for Czech. In N. Calzolari, K. Choukri, T. Declerck, H. Loftsson, B. Maegaard, J. Mariani, A. Moreno, J. Odijk, & S. Piperidis (Eds.), Proceedings of the 9th International Conference on Language Resources and Evaluation (LREC 2014) (pp. 1087–1093). Reykjavik: European Language Resources Association (ELRA). Štekauer, P. (2014). Derivational paradigms. In R. Lieber & P. Štekauer (Eds.), The Oxford handbook of derivational morphology (pp. 354–369). Oxford: Oxford University Press. Talamo, L., Celata, Ch., & Bertinetto, P. M. (2016). DerIvaTario: An Annotated Lexicon of Italian Derivatives. Word Structure 9(1), 72–102. Zeller, B. D., Snajder, J., & Padó, S. (2013). DErivBase: Inducing and Evaluating a Derivational Morphology Resource for German. In H. Schuetze (Ed.), Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics (Vol. 1, pp. 1201–1211). Sofia: Association for Computational Linguistics.

158

Litta and Budassi

Appendix

Core paradigm for Latin derivation processes

A

B

C

D

E

F

BASE [X]

[x]Vj ↔ [SEM]j

[x]N ↔ [SEM]N

[x]A ↔ [SEM]A

[[x](t)io]N ↔ [action of SEM]N

[[x] (t)or]N ↔ [one who SEM]N

[con [x]]Vj ↔ [to SEM with]Vj

[con [x]]N ↔ [SEM]N

[con [x]]A ↔ [co-SEM (simultaneity, joint action)]A

[[con [x]]j (t)io]N ↔ [action of SEM j with]N

[[con [x]]j (t)or]N ↔ [one who SEM j with]N

[in2 [x]]Vj ↔ [to not SEM]Vj

[in2 [x]]A ↔ [not SEM]A

[[in2 [x]]j (t)io]N ↔ [action of SEM j]N

[[in2 [x]]j (t)or]N ↔ [one who SEM j]N

[ex [x]]V ↔ [ex [x]]N ↔ [to SEM out/ [SEM]N away]Vj

[ex [x]]A ↔ [without SEM]A

[ex [x] (t)or]N ↔ [one who SEM j]N

[in1 [x]]V ↔ [to SEM inside]Vj

[in1 [x]]N ↔ [SEM]N

[in1 [x]]A ↔ [SEM]A

[de [x]]V ↔ [de [x]]N ↔ [to SEM away [SEM]N from]Vj

[de [x]]A ↔ [SEM]A

[re [re [x]]N ↔ [x]]V ↔ [to [SEM]N SEM again]Vj

[re [x]]A ↔ [repeated SEM]A

[[ex [x]]j (t)io]N ↔ [action of SEM j]N [[in1 [x]]j (t)io]N ↔ [action of SEM j]N [[de [x]]j (t)io]N ↔ [action of SEM j]N [[re [x]]j (t)io]N ↔ [action of SEM j again]N [[ad [x]]j (t)io]N ↔ [action of SEM j]N

[ad [x]]V ↔ [to SEM towards]Vj

[ad [x]]N ↔ [ad [x]]A ↔ [SEM [SEM (intensifier)]N (intensifier)]A

[[in1 [x]]j (t)or]N ↔ [one who SEM j]N [[de [x]]j (t)or]N ↔ [one who SEM j]N [[re [x]]j (t)or]N ↔ [one who SEM j]N [[ad [x]]j (t)or]N ↔ [one who SEM j]

159

What We Talk about When We Talk about Paradigms

G

H

[[x]]j os]A ↔ [full of SEM j]A

[[x]]j tas]N ↔ [[x]j al]A ↔ [characteristic [typical of of SEM j]N SEM j]A

[[con [x]]j os]A ↔ [full of SEM j]A [[in2 [x]]j os]A ↔ [not full of SEM j]A

[[ex [x]]j os]A ↔ [SEM j]A

[[in1 [x]]j os]A ↔ [SEM j]A

I

J

[[x]j ari]A ↔ [having qualities of SEM J]A [[con [x]]j [[con [x]]j [[con [x]]j tas]N ↔ al]A ↔ ari]A ↔ [characteristic [typical of [having of SEM j]N SEM j]A qualities of SEM J]A [[in2 [x]]j [[in2 [[in2 [x]]j tas]N ↔ [x]]j al]A ↔ ari]A ↔ [not [characteristic [not typical having of non SEM j]N of SEM j]A qualities of SEM J]A [[ex [x]]j tas]N [[ex [x]]j al]A [[ex [x]]j ↔ [character- ↔ [typical of ari]A ↔ istic of SEM j]N SEM j]A [SEM J ]A [[in1 [x]A] j tas]N ↔ [SEM j]N

[[in1 [x]] j al]A ↔ [SEM j]A

[[in1 [x]]j ari]A ↔ [SEM J]A

[[de [x]]j os]A [[de [x]]j tas]N [[de [x]]j al]A [[de [x]]j ↔ [SEM j]A ari]A ↔ ↔ [character- ↔ [SEM j]A istic of SEM j]N [SEM J ]A [[re [x]]j os]A [[re [x]A] ↔ [SEM j]A j tas]N ↔ [SEM J]N

[[re [x]]j al]A [[re [x]]j ari]A ↔ [SEM J]A ↔ [SEM j]A

[[ad [x]]j os]A [[ad [x]]j tas]N [[ad [x]]j al]A [[ad [x]]j ↔ [SEM j]A ↔ [character- ↔ [typical of ari]A ↔ istic of SEM j]N SEM j]A [SEM J ]A

K

L

[[x]j trix]]N ↔ [a female who SEM J]N

[[x]j at]A ↔ [having SEM J]A

[[con [x]]j trix]]N ↔ [a female who SEM J]N

[[con [x]]j at]A ↔ [having SEM J]A [[in2 [x]]j at]A ↔ [not having SEM J]A

[[ex [x]]j trix]]N ↔ [a female who SEM J]N [[in1 [x]]j trix]]jN3:f ↔ [a female who SEM J]N [[de [x]]j trix]]N ↔ [a female who SEM J]N [[re [x]]j trix]]N ↔ [a female who SEM J]N [[ad [x]]j trix]]N ↔ [a female who SEM J]N

[[ex [x]]j at]A ↔ [having SEM J]A [[in1 [x]]j at]A ↔ [having SEM J]A [[de [x]]j at]A ↔ [having SEM J]A [[re [x]]j at]A ↔ [having SEM J]A [[ad [x]]j at]A ↔ [having SEM J]A

160

Litta and Budassi

Core Paradigm for Latin derivation processes (cont.)

A

B

C

[per [x]]V ↔ [to SEM through]Vj

[per [x]]A ↔ [[per [very SEM [x]]j (t)io]N (intensive)]A ↔ [action of SEM j]N [prae [x]]N ↔ [prae [x]]A [[prae [SEM]N ↔ [very SEM [x]]j (t)io]N (quality pre- ↔ [action of heminence)] SEM j]N

[prae [x]]V ↔ [to SEM ahead]Vj [sub [x]]V ↔ [to SEM down/ under]Vj

D

E

F

[per [x]]N ↔ [SEM]N

[sub [x]]N ↔ [sub-SEM (rank)]N

[ob [x]]V ↔ [to SEM against]Vj

A

[[per [x]]j (t)or]N ↔ [one who SEM j]N [[prae [x]]j (t)or]N ↔ [one who SEM j]N

[sub [x]]A ↔ [not very SEM]A

[[sub [x]]j (t)io]N ↔ [action of SEM j]N

[[sub [x]]j (t)or]N ↔ [one who SEM j]N

[ob [x]]A ↔ [SEM]A

[[ob [x]]j (t)io]N ↔ [action of SEM j]N

[[ob [x]]j (t)or]N ↔ [one who SEM j]N

M

N

O

[[x]j ari]N ↔ [[dealer in SEM J]N

[[x]j ul]N ↔ [small SEM J]N

[[x]j bil]A ↔ [[x]j ic]A ↔ [able/worthy to [belonging to be SEM J]A SEM J]

[[con [x]]j ari]N ↔ [SEM J]N

[[con [x]]j ul]N ↔ [SEM J]N

[[con [x]]j bil]A [[con [x]]j ic]A ↔ [[con [x]]j (i)t]V ↔ [able/worthy [SEM J] ↔ [to SEM J to be SEM J]A repetitively]V [[in2 [x]]j bil]A ↔ [unable/unworthy to be SEM J]A

P

[[in2 [x]]j ic]A ↔ [not belonging to SEM J]A

Q [[x]j (i)t]V ↔ [to SEM J repetitively]

161

What We Talk about When We Talk about Paradigms

G

H

I

J

K

L

[[per [x]]j os]A ↔ [full of SEM j]A

[[per [x] A]j tas]N ↔ [SEM J]N

[[per [x]]j al]A ↔ [SEM j]A

[[per [x]]j ari]A ↔ [SEM J ]A

[[per [x]]j at]A ↔ [having SEM J]A

[[prae [x]]j os]A ↔ [SEM j]A

[[prae [x]A]j tas]N ↔ [SEM J]N

[[prae [x]]j al]A ↔ [SEM j]A

[[prae [x]]j ari]A ↔ [SEM J ]A

[[per [x]]j trix]]N ↔ [a female who SEM J]N [[prae [x]]j trix]]N ↔ [a female who SEM J]N

[[sub [x]]j os]A ↔ [SEM j]A

[[sub [x] A]j tas]N ↔ [SEM J]N

[[sub [x]]j al]A ↔ [typical of SEM j]A

[[sub [x]]j ari]A ↔ [having qualities of SEM J]A [[ob [x]]j ari]A ↔ [SEM J ]A

[[sub [x]]j trix]]N ↔ [a female who SEM J]N

[[sub [x]]j at]A ↔ [having SEM J]A

[[ob [x]]j trix]]N ↔ [a female who SEM J]N

[[ob [x]]j at]A ↔ [having SEM J]A

[[ob [x]]j os]A [[ob [x]A]j [[ob [x]]j ↔ [SEM j]A tas]N ↔ [SEM J] ari]A ↔ [SEM j]A

[[prae [x]]j at]A ↔ [having SEM J]A

R

S

T

U

V

W

[[x]j in]A ↔ [typical of SEM J]A

[[x]j i]N ↔ [realisation of SEM j]N

[[x]j iv]A ↔ [related to SEM j]

[[con [x]]j [[con [x]]j i]N ↔ [SEM j]N iv]A ↔ [related to SEM j]A [[in2 [x]]j i]N [[in2 [x]]j iv]A ↔ ↔ [SEM j]N [related to SEM j]A

[[x]j i]N ↔ [result of action of SEM j]N [[con [x]]j i]N ↔ [result of action of SEM j]N [[in2 [x]]j i]N ↔ [result of action of SEM j]N

[[x]]j im]I ↔ [in a SEM j manner]I

[[con [x]]j in]A ↔ [SEM J ]A

[[x]j e]A ↔ [coming from/made of SEM J] [[con [x]]j e]A ↔ [SEM J]

[[in2 [x]]j in]A ↔ [SEM J ]A

[[in2 [x]]j e]A ↔ [SEM J]

[[con [x]]j im]I ↔ [SEM j]I [[in2 [x]]j im]I ↔ [not in a SEM j manner]I

162

Litta and Budassi

Core Paradigm for Latin derivation processes (cont.)

M

N

O

P

Q

[[ex [x]]j ari]N ↔ [[ex [x]]j ul]N ↔ [SEM J]N [SEM J]N

[[ex [x]]j ic]A ↔ [SEM J]A

[[ex [x]]j (i)t]V ↔ [to SEM J repetitively]V

[[in1 [x]]j ari]N ↔ [SEM J]N

[[in1 [x]]j ic]A ↔ [[in1 [x]]j (i)t]V ↔ [SEM J]A [to SEM J repetitively]V

[[ex [x]]j bil]A ↔ [able/ worthy to be SEM J]A [[in1 [x]]j ul]N ↔ [[in1 [x]]j bil]A [SEM J]N ↔ [able/worthy to be SEM J]A

[[de [x]]j bil]A ↔ [able/ worthy to be SEM J]A [[re [x]]j ari]N ↔ [[re [x]]j ul]N ↔ [[re [x]]j bil]A ↔ [SEM J]N [SEM J]N [able/ worthy to be SEM J]A [[ad [x]]j ari]N [[ad [x]]j ul]N ↔ [[ad [x]]j bil]A ↔ [SEM J]N [SEM J]N ↔ [able/ worthy to be SEM J]A [[per [x]]j ari]N [[per [x]]j ul]N ↔ [[per [x]]j bil]A ↔ [SEM J]N [SEM J]N ↔ [able/ worthy to be SEM J]A [[prae [x]]j ari]N [[prae [x]]j ul]N [[prae [x]]j bil]A ↔ [SEM J]N ↔ [able/worthy ↔ [SEM J]N to be SEM J]A [[de [x]]j ari]N ↔ [[de [x]]j ul]N ↔ [SEM J]N [SEM J]N

[[sub [x]]j ari]N ↔ [SEM J]N

[[sub [x]]j ul]N ↔ [SEM J]N

[[ob [x]]j ari]N ↔ [[ob [x]]j ul]N ↔ [SEM J]N [SEM J]N

[[sub [x]]j bil]A ↔ [able/ worthy to be SEM J]A [[ob [x]]j bil]A ↔ [able/ worthy to be SEM J]A

[[de [x]]j ic]A ↔ [SEM J]A

[[de [x]]j (i)t]V ↔ [to SEM J repetitively]V

[[re [x]]j ic]A ↔ [SEM J]A

[[re [x]]j (i)t]V ↔ [to SEM J repetitively]V

[[ad [x]]j ic]A ↔ [SEM J]A

[[ad [x]]j (i)t]V ↔ [to SEM J repetitively]V

[[per [x]]j ic]A ↔ [[per [x]]j (i)t]V ↔ [SEM J]A [to SEM J repetitively]V [[prae [x]]j ic]A ↔ [SEM J]A

[[prae [x]]j (i)t]V ↔ [to SEM J repetitively]

[[sub [x]]j ic]A ↔ [[sub [x]]j (i)t]V [SEM J]A ↔ [to SEM J repetitively] [[ob [x]]j ic]A ↔ [SEM J]A

[[ob [x]]j (i)t]V ↔ [to SEM J repetitively]

163

What We Talk about When We Talk about Paradigms

R

S

[[ex [x]]j in]A [[ex [x]]j i]N ↔ [SEM J]A ↔ [SEM j]N

[[in1 [x]]j in]A ↔ [SEM J ]A [[de [x]]j in]A ↔ [SEM J]A

T

U

[[ex [x]]j iv]A [[ex [x]]j e]A ↔ [related to ↔ [SEM J] SEM j]A

[[in1 [x]]j i]N [[in1 [x]]j iv]A ↔ ↔ [SEM j]N [related to SEM j]A [[de [x]]j i]N [[de [x]]j iv]A ↔ [SEM j]N ↔ [related to SEM j]A

[[in1 [x]]j e]A ↔ [SEM J]

[[de [x]]j e]A ↔ [SEM J]

[[re [x]]j in]A [[re [x]]j i]N ↔ [SEM J]A ↔ [SEM j]N

[[re [x]]j iv]A [[re [x]]j e]A ↔ [related to ↔ [SEM J] SEM j]A

[[ad [x]]j in]A [[ad [x]]j i]N ↔ [SEM J]A ↔ [SEM j]N

[[ad [x]]j iv]A [[ad [x]]j e]A ↔ [related to ↔ [SEM J] SEM j]A

[[per [x]]j in]A ↔ [SEM J ]A

[[per [x]]j i]N [[per [x]]j iv]A ↔ [SEM ↔ [SEM j]N j]A

[[prae [x]]j [[prae [x]]j [[prae [x]]j in]A ↔ [SEM J ] i]N ↔ [SEM j]N iv]A ↔ [related to SEM j]A [[sub [x]]j [[sub [x]]j i]N [[sub [x]]j in]A ↔ [SEM J ] ↔ [SEM j]N iv]A ↔ [related to SEM j]A [[ob [x]]j in]A [[ob [x]]j i]N [[ob [x]]j iv]A ↔ [SEM J] ↔ [SEM j]A ↔ [SEM j]N

[[per [x]]j e]A ↔ [SEM J]

[[prae [x]]j e]A ↔ [SEM J ]A

[[sub [x]]j e]A ↔ [SEM J]A

[[ob [x]]j e]A ↔ [SEM J]A

V

W

[[ex [x]]j i]N ↔ [result of action of SEM j]N [[in1 [x]]j i]N ↔ [result of action of SEM j]N [[de [x]]j i]N ↔ [result of action of SEM j]N [[re [x]]j i]N ↔ [result of action of SEM j]N [[ad [x]]j i]N ↔ [result of action of SEM j]N [[per [x]]j i]N ↔ [result of action of SEM j]N [[prae [x]]j i]N ↔ [result of action of SEM j]N [[sub [x]]j i]N ↔ [result of action of SEM j]N [[ob [x]]j i]N ↔ [result of action of SEM j]N

[[ex [x]]j im]I ↔ [SEM j]I

[[in1 [x]]j im]I ↔ [SEM j]I

[[de [x]]j im]I ↔ [SEM j]I

[[re [x]]j im]I ↔ [SEM j]I

[[ad [x]]j im]I ↔ [SEM j]I

[[per [x]]j im]I ↔ [SEM j]I

[[ob [x]]j im]I ↔ [SEM j]I

Chapter 7

A Paradigmatic Approach to Compounding Jan Radimský 1

Introduction

Although the concept of paradigm in morphology has traditionally been applied especially to inflection, it has attracted even more attention in studies on word formation in the past few years. Its importance has so far been recognized and confirmed in the domain of derivation only, while its relevance for the study of compounding still remains considerably less explored. In this paper, the relevance of paradigms in compounding will be given a thorough discussion, and it will be argued that a paradigmatic approach may be an appropriate answer to different challenging issues concerning specifically Romance Noun-Noun compounds. As suggested by Boyé and Schalchli (2016, p. 230), the discussion will start out with a general definition of paradigmatic relations in §2 and then, formal definitions as well as examples of their different subtypes, such as families, series, paradigms and paradigmatic systems will be considered, especially with regard to derivation and inflection. In §3, it will be argued that the same types of paradigmatic relations may be found in compounding, and relevant examples will be discussed. §4 will focus on specific problematic issues in compounding that may be accounted for using a paradigmatic approach. From a theoretical standpoint, this paper is anchored in recent studies on paradigmatic approaches to derivational morphology (Boyé, Hathout, & Roché, 2011; Boyé & Schalchli, 2016; Bauer, 2017; Bonami & Strnadová, 2019; or Hathout & Namer, 2018, amongst others) and at the same time, it acknowledges an important theoretical affinity between these approaches and Construction Morphology (Booij, 2010, 2016). To the best of the author’s knowledge, this chapter is the first attempt to apply the model of paradigmatic relations to compounding, including a thorough discussion of related terminological implications. It emphasizes the fact that Construction Morphology, which has been developing in the past decade, may also be viewed as a type of paradigmatic approach with its own theoretical machinery that is to a large extend compatible with other paradigm-based approaches.

© Koninklijke Brill NV, Leiden, 2020 | doi:10.1163/9789004433410_008

A Paradigmatic Approach to Compounding

2

165

Paradigmatic Relations in Derivational Morphology

In a general fashion, paradigmatic relations are viewed as relations between items that display a systematic co-variation between form and meaning.1 As Boyé and Schalchli (2016) put it, paradigms are defined as sets of items with “parallel variation between form and content” (p. 230); similarly, Bauer (2017, p. 15) argues that “two items are in a paradigmatic relationship with each other when one can replace the other in the stream of speech with a resultant difference in the message (i.e., they contrast)”. For the sake of the following discussion, we will stress the fact that the difference of form and that of meaning must be systematic. Indeed, the terms paradigm and paradigmatic relation have been used in several theoretical frameworks with different—though seemingly similar—meanings, as pointed out in a review made by Štekauer (2014, pp. 361–369). For instance, the onomasiological tradition rooted in the works by Dokulil and Horecký defines the paradigmatic relation as a relation between a base and all the derivatives made up of that base, irrespective of the fact whether it is systematic or not (Štekauer, 2014, p. 363). The present author’s perspective is rather rooted in the tradition going back to van Marle (1985) and developed in recent studies by French and Francophone morphologists (such as Boyé, Hathout, & Roché, 2011; Boyé & Schalchli, 2016; Bauer, 2017; Bonami & Strnadová, 2019), where paradigmatic relations are conceived as systematic covariations of form and meaning between words (or, rather, between lexemes), irrespective of their internal complexity. In other words, it will be assumed that derivational paradigmatic relations may hold between various sets of lexical items, be they simple or complex (including compounds or even phrasal lexemes), provided that these relations are systematic. Within the theoretical framework introduced in the previous paragraph, it is assumed that the general concept of “paradigmatic relation” between lexemes is to be understood as a macro-label that encompasses different specific phenomena, such as morphological families, series, paradigms, and paradigmatic systems. Although the concepts of a morphological family and series have already been outlined and discussed by Roché (2011) or Hathout (2011), amongst others, Bonami and Strnadová (2019) have recently proposed precise formal definitions that will be the starting point for the following discussion. Thus, in this section, we will discuss formal definitions of morphological families, series, paradigms, and paradigmatic systems and provide examples from both 1  Haspelmath and Sims (2010, p. 2) go so far as to state that morphology itself consists in the study of systematic covariations between the form and the meaning of words.

166

Radimský

derivational morphology and inflection, in order to get a more solid theoretical basis for the discussion of paradigmatic relations in compounding. Perhaps the simplest type of paradigmatic relationship is the one between the members of a morphological family. A morphological family has been defined by Bonami and Strnadová (2019) as a tuple of words F = (w1, … wn) such that “any member wi of the family is morphologically related to any other member of wj” (§2.1), which means that between any wi and wj there is a relation of form, a relation of content, and the same content-form relation can be identified in other pairs of words. Examples of inflectional (1, 2) and derivational (3, 4) families in French are given below. (1) égal, égale, égaux, égales equal.m.sg. equal.f.sg. equal.m.pl. equal.f.pl. (2) vieux, old.m.sg.

vieille, old.f.sg.

vieux, old.m.pl.

vieux old.f.pl.

(3) laver, laveur, lavage wash.V washer.N washing.N (4) gonfler, gonfleur, gonflement inflate.V air.pump.N inflating.N As a matter of fact, families exemplified in (1–4) are more precisely referred to as “partial families” by the authors, since any word belongs to multiple (partial) families which together form a complete family encompassing potentially both derivational (5a) and inflectional (5b) morphology, as exemplified in (5). However, even if the notion of “complete family” is understood as restricted to derivational morphology only, there still remains an important difference between complete derivational families (5a or 6) and partial families (1–4) lying in the fact that in the case of the former ones the form-content relation is not systematic in the sense that the same form-content relationship cannot be identified in other pairs of words. Understood in this way, complete derivational families display a systematic variation of form but not a systematic variation of meaning (cf. the concept of “derivational family” in Roché, 2011 or Boyé & Schalchli, 2016). (5a) laver, laveur, lavage, lavement, … wash.V.inf washer.N washing.N enema.N

A Paradigmatic Approach to Compounding

167

(5b) lave, laverai, laverais, … wash.V.1.sg.ind.pres wash.V.1.sg.ind.fut wash.V.1.sg.cond.pres (6) colourful, colourless, colourant, colouring, colourism This means that complete (derivational) families and partial families are to be kept apart carefully: we will follow Bonami and Strnadová (2019) in using the term “family” for partial families only, since they are much more important in derivational morphology than the complete ones. When referring to complete derivational families, the term “nest” (Štekauer, 2014, p. 364) will be used. The sets of lexemes in (5a) and in (6) are examples of derivational nests. A superposition of morphological families that exhibit the same formcontent relationship will be referred to as morphological paradigm. An example of derivational paradigm according to Hathout and Namer (2018) is provided in Table 7.1. Notice that in the case of a morphological paradigm, families (in the rows of the table) are organized in such a way that the same abstract content (in the columns of the table) corresponds to the same form (i.e., the same suffix in each column). As follows from the previous discussion, the concept of family and that of paradigm are closely related to each other: on the one hand, a paradigm is necessarily made up of families, and on the other hand, a family has to take part in a paradigm (otherwise, the form-content relationship between the members of the family would not be systematic). On a more abstract ground, a superposition of several paradigms where the same content is not always realized by the same formal means corresponds to a paradigmatic system. As Bonami and Strnadová (2019) put it, a paradigmatic system is formed by at least two ordered pairs of morphologically related words (w1, w2) and (w3, w4) that are aligned by the same “aligning relation” Rc, such that Rc (w1, w2) and Rc (w3, w4). An example of a simple paradigmatic system in French is given in Figure 7.1. The “vertical” cells of families, paradigms and paradigmatic systems that share the same abstract content have been labelled by the term morphological “series” (Hathout, 2011; Hathout & Namer, 2018). Thus, Figure 7.1 displays three Table 7.1

employ recruit announce

Derivational paradigm (according to Hathout and Namer, 2018)

employer recruiter announcer

employment recruitment announcement

employable recruitable announceable

168

Radimský

Figure 7.1 A French paradigmatic system Bonami & Strnadová, 2019

different series corresponding to an agent noun (laveur, formateur, gonfleur), an action noun (lavage, formation, gonflement), and a verb (laver, former, gonfler). This abstract content shared by all the members of a series has been referred to as “categorial value” (van Marle, 1985, p. 89). It is worth noting that paradigmatic relations are core properties of semischematic and schematic constructions in Construction Morphology (Booij, 2010; Booij, 2016), since constructions are defined as (systematic) pairings of form and meaning. Thus, the construction, such as (7a), that yields derived adjectives in (7b) underlies the paradigmatic relationship between the nouns air, beer, bread, and colour (Booij, 2016, p. 424), as these nouns may replace the variable [x] in (7a) (which is the systematic variation of form) that correlates with a systematic and predictable variation of the content. In other words, pairs such as air-airless, beer-beerless, … represent (partial) derivational families, while the set of lexemes in (7b) with the respective bases is a derivational paradigm. (7a) < [[x]Ni less]Aj ↔ [Property of being without SEM i]j > (7b) airless, beerless, breadless, colourless On the other hand, variations of form, such as those in (8), could hardly be conceived as paradigmatic because there is no systematic variation of meaning associated to them. In terms of Construction Morphology, the items in (8)

A Paradigmatic Approach to Compounding

169

cannot be described by a semi-schematic construction, and from the viewpoint of the paradigmatic approach, the set in (8) constitutes a word nest. (8) colourful, colourless, colourant, colouring In the following section, we will discuss the question whether the different types of paradigmatic relationships may also be applied to the analysis of compound words. 3

Paradigmatic Relations in Compounding

The crucial difference between paradigmatic approaches to inflection and derivation, on the one hand, and paradigmatic approaches to compounding, on the other, lies in the fact that in the former case, abstract contents that underlie morphological series are, roughly speaking, somehow “linked” to single affixes,2 while in the latter case, the abstract content underlying series is either implicit or derivable solely from the knowledge of paradigms. This is why the paradigmatic approach to compounding seems both challenging and inspiring. Let us illustrate this point on concrete examples. A very simple example of a morphological family in compounding is given in (9a). It represents a pair of words w1 (tube), w2 (steel tube) between which there is a relation of form and a relation of content that is systematic, since the same content-form relation, captured by the schema (9b), can be identified in other pairs of words, such as those in (9c). Thus, (9a) is a family, while the examples in (9c) represent a compounding paradigm. Notice that compounds, such as steel tube, steel frame, … in (9c) constitute a paradigmatic series based on the abstract content that may be described roughly as a “product made of steel”, while the bases in (9c), such as tube, frame, … form another series that may be described approximately as a “product that may be made of something”. However, the abstract content on which these series are based is not expressed overtly. (9a) tube, steel tube (9b) < [steel [x]Ni]Aj ↔ [SEM i made of steel]j >

2  For instance, the series of agent nouns in Figure 7.1 is linked to the suffix -eur, while the series of action nouns is linked to the suffixes -age, -tion, and -ment.

170

Radimský

(9c) tube, steel tube frame, steel frame sink, steel sink wire, steel wire plate, steel plate blade, steel blade Obviously, there are also other families with “steel-N” compounds, such as those in (10), that do not fit the paradigm (9c). Nevertheless, the sum of all “steel-N” compounds with the respective bases seems to be nothing more than a compounding “nest” which, as we have observed above in the case of derivational morphology, does not have any particular relevance for the paradigmatic approach. (10) industry, steel industry production, steel production company, steel company Now the question arises as to whether compounding paradigms may form more complex and abstract entities, such as paradigmatic systems. Indeed, this is actually the case, at least for some compounding paradigms. For instance, the “made of steel-N” paradigm in (9c) reproduced in (11a) is an instance of a more abstract “made of” paradigmatic system that may be retrieved in other paradigms, such as those in (11b) and (11c). (11a) tube, steel tube frame, steel frame … (11b) wall, concrete wall floor, concrete floor bunker, concrete bunker … (11c) soup, vegetable soup dish, vegetable dish … Notice that, just as in the case of derivational paradigmatic systems, the two abstract (vertical) series in (11a–c) based on the respective general semantic

A Paradigmatic Approach to Compounding

171

Figure 7.2 Bidimensional paradigmatic system in compounding

values of the bases (“product that may be made of a material”) and of the compounds (“product made of a material”) remains identical throughout the whole paradigmatic system although there is no formal overlap between the paradigms (11a), (11b), and (11c). As a matter of fact, the paradigmatic system introduced in (11a–c) has one more dimension. If each horizontal line corresponds to a two-member family made up of the couple “product that may be made of a material” (e.g., tube) and “product made of material” (e.g., steel tube), there is always also a corresponding family formed by the couple “material” (e.g., steel) and “product made of a material” (e.g., steel tube). The whole paradigmatic system is illustrated in Figure 7.2. The relationship between members of a family in Figure 7.2 is schematized by arrows. Thus, each level of the schema comprises two families, such as (i) steel-steel chain and (ii) chain-steel chain. In addition, there is a third, indirect relationship between the members of the series “Material” and “Product”, schematized by a flat line (e.g., steel and chain). The examples analysed above show that the different paradigmatic relations (i.e., families, series, paradigms, and paradigmatic systems) defined with respect to inflection and derivation may also be retrieved in the domain of compounding without there being a need to modify their formal definitions in any way. There is, however, an important difference that opposes the paradigmatic approaches to inflection to the paradigmatic analysis of word formation (including both derivation and compounding), which lies in the nature of abstract series. In the domain of inflection, series are usually identified with a purely grammatical (or highly grammaticalized) content, such as “1st person singular

172

Radimský

of the present indicative” or “accusative plural”, and so forth. Therefore, the cells in the paradigms are rather clearly structured, and their filling with the single forms is usually required by syntax, though there are many exceptions to this principle (e.g., avalent verbs usually have only one person, non-transitive verbs do not have the passive, etc.). Conversely, in the domain of word formation, series are usually identified with poorly grammaticalized abstract semantic values, such as action noun, agent noun, instrument, property noun, property adjective, potentiality noun, causative verb, and so forth. These semantic values seem to be less structured, and they are required by the needs of communication rather than by syntax, which makes word-formation paradigms potentially less regular and systematic, though the difference seems to be rather one of degree (cf. Bonami & Strnadová, 2019). In the final part of §2, we have already emphasized the strong link between the paradigmatic approach to word formation and the Construction Morphology (CM) approach. Indeed, any paradigm corresponds to a specific construction (a systematic pairing of form and meaning) that may be captured by a CM schema. The CM formalism may, therefore, be used for the analysis of paradigms. Moreover, constructions (as well as paradigms) are organized hierarchically, in the sense that upper-level constructions correspond to paradigmatic systems. An example of the hierarchical organization in English compounding paradigms is given in Figure 7.3. The vertical hierarchy in the paradigmatic system is organized from the bottom-up perspective because paradigms and constructions do not exist as general abstract entities but rather as abstractions over the existing lexicon. Therefore, language users first learn both simple and complex words, and subsequently they make generalizations about paradigms. Similarly, when a need for coining a new complex word emerges, speakers may either use a specific paradigm or, if such a paradigm does not exist, they may use a more general upper-level paradigmatic system (i.e., an upper-level schema), which opens space for morphological creativity. 4

Are Paradigms Useful in Compounding?

4.1 Overview This section will present several properties of compounds that are directly linked to the paradigmatic structure of compounding. The presentation will be organized around three axes. §4.2 provides examples of different “holistic properties” of compounds that do not derive from the compound constituents

173

A Paradigmatic Approach to Compounding

Right-headed subordinate N-N compounds < [[X]Ni [X]Nj]Nk ↔ [SEM j with any relation to SEM i]k >

Grounding N-N compounds “made of” < [[X]Ni [X]Nj]Nk ↔ [SEM j made of SEM i]k >

Verbal-nexus N-N compounds < [[X]Ni [X]N(

< [steel [X]Nj]Nk ↔ [SEM j made of steel]k >

< [[X]Ni production]Nk ↔ [production of SEM i]k >

steel door, steel wire

steel production, wire production

< [concrete [X]Nj]Nk ↔ [SEM j made of concrete]k >

< [[X]Ni creation]Nk ↔ [creation of SEM i]k >

concrete wall, concrete floor

job creation, database creation

Figure 7.3 Hierarchical organization of compounding paradigms

nor from a general rule but are related to paradigms. In §4.3, we will introduce some compound properties that derive from the vertical organization of paradigms, namely the position of the head in Romance compounds and the principles of insertion of phrases in compounds. In §4.4, an example of phenomena that derive from paradigm merging will be dealt with. 4.2 Holistic Properties of Compounds In Construction Morphology, it has been shown that constructions have properties that do not derive from their constituents (Booij, 2016, p. 428) nor from general “rules” but are related to particular constructions only. Thus, in terms of the paradigmatic approach to compounding, it may be assumed that some properties of compounds derive (only) from the knowledge of paradigms. Let us illustrate this feature on concrete examples. In a rule-based approach, it would be assumed that a compound, such as [[key]N [industry]N]N, is made up of the words [key]N and [industry]N. Nevertheless, data show that this is true only from a purely formal point of view. From the semantic standpoint, the meaning of the noun [key]N in the compound key industry in (12a) does not refer to any lexical meaning of the noun [key]N in other contexts, but it rather refers to its specific meaning in the paradigm (12b), captured by the schema (12c).

174

Radimský

(12a) It is SNP policy that electricity, gas, water, rail and a number of key industries, such as forestry, should be in the Scottish public domain. (BNC) (12b) role, key role issue, key issue factor, key factor element, key element feature, key feature (12c) < [key [X]Ni]Nj ↔ [Important SEM i]j > Therefore, this specific meaning of [key]N is related to the paradigm (12b) only. Obviously, it does not prevent the noun [key]N from appearing in other compounding paradigms: for instance, there is another compound [[key]N [industry]N]N in (13a) whose interpretation derives from the paradigm (13b) captured by the schema (13c). (13a) We’ve assembled a world-class team of gifted innovators to develop unique, proprietary technologies that are making massive changes in the key industry. (https://solidkeysusa.com/) (13b) coal, coal industry oil, oil industry film, film industry computer, computer industry textile, textile industry (13c) < [[X]Ni industry]Nj ↔ [Industry that produces SEM i]j > The phenomenon illustrated in (12), where the sense of a nominal component of a Noun-Noun (N-N) compound is derivable from the paradigm rather than from the component itself, captures well the situation of metaphoric modifiers of Italian attributive N-N compounds exemplified in (14). The English glosses in (14) describe both the literal sense and the metaphorical sense the noun has in N-N compounds (for more data and examples of compounds see Radimský, 2015, pp. 169–173).

A Paradigmatic Approach to Compounding

(14) chiave (key-important) lampo (lightning-quick) cardine (hinge-fundamental) pilota (pilot-leading, trial) fantasma (ghost-imaginary, false)

175

madre (mother-primary, parent) pirata (pirate-false, fraudulent) principe (prince-main, leading) quadro (frame-general)

A very similar phenomenon may be observed in the domain of learned compounding, where the paradigms are often rich and regular. In this case, many learned stems (confixes, noted C) exist as units in the mental lexicon only as abstraction over paradigms. Thus, we may ask, for instance, in which sense the Italian compound [[dermato]C[scopio]C]N (“dermatoscope”) contains the two bases dermato and scopio, which are exclusively bound forms, though they have a full lexical meaning. The answer would be that these stems exist only as abstractions over paradigms (15a) and (16a) captured by the respective schemas (15b) and (16b). (15a) dermatologo, dermatologia, dermatoscopia, dermatochirurgia, dermatopatologo, … (15b) < [dermato [X]Ni]Nj ↔ [SEM i with any relation to skin]j > (16a) telescopio, microscopio, oscilloscopio, periscopio, spettroscopio, endoscopio, bronchoscopio, coloscopio, gastroscopio, … (16b) < [[X]Ni scopio]Nj ↔ [Instrument that enables to examine SEM i or to examine something in a way defined by SEM i]j > This is what makes Peytard (1975, p. 86) claim that confixes, though being bound forms, have an operational autonomy (“autonomie de fonctionnement”). The holistic properties of paradigms outlined above were discussed essentially from the semasiological perspective. However, they may be relevant even from the onomasiological perspective. For instance, if a speaker faces the necessity to name a concept using an implicit “made of” relation, he in theory has several formal possibilities to choose from in any language. In Table 7.2, some examples of structures are given for English, French, and Czech. In general, the three basic structures N-N, A-N (or N-A) and N-PREP-N are available in the three languages although N-N compounding is rather restricted

176

Radimský

Table 7.2 Examples of structures that may express an implicit “made of” relationship

Language

Structure

Example

Comment

English

N-N A-N

steel wire, wood floor wooden floor

French

N-de-N N-en-N A-N

fil d’acier (lit. “wire of steel”, “steel wire”) fil en acier (lit. “wire in steel”, “steel wire”) [ocelový]A [drát]N (“steel wire”) [dřevěná]A [podlaha]N (“wooden floor”) [drát]N [z]PREP [oceli]N-gen. (lit. “wire of steelN-gen”, “steel wire”) [podlaha]N [ze]PREP [dřeva]N-gen. (lit. “floor of woodN-gen”, “wooden floor”) [dřevo]N[stavba] N (“wooden house”) [dřevo]N[plyn]N (“wood gas”) [panel]N[ák]suffix (“block of flats made of concrete panels”, lit. “panel + SUFF”) [dřev]N[ák]suffix (“clog”, lit. “wood + SUFF”)

open choice principle only when a construction with an appropriate adjective is available open choice principlea open choice principle open choice principle

Czech

N-z(e)-Ngen.

N-N suffixation

open choice principle

extremely restricted restrictedb

a Only the N-de-N construction is possible when acier (“steel”) has a metaphoric interpretation (“strong”), as in moral d’acier (“a high morale”, “nerves of steel”). Besides this, other restrictions may be also present with different semantic types of nouns, as Bernard Fradin suggested in a discussion. b The suffixation paradigm is by definition restricted, since the derived word only expresses the material and not the product itself (i.e., the onomasiological base). Thus, it may be used as open choice in particular contexts, where the onomasiological base is contextually anchored, or in lexemes where the onomasiological base has been already lexicalized. As for the latter case, approximately a dozen of lexicalized derivatives with the suffix -ák encoding the made-of relation can be retrieved in the corpus Syn2015, including also for instance plyšák (“a toy made of plush”), bramborák (“potato pancake”), slamák (“straw hat”), or zlaťák (“golden coin”).

in French and Czech.3 Nevertheless, the speaker who wants to name a concept using the “made of” relation is not really free to choose any of them. In Czech, the syntactic structures A-N and N-PREP-N represent the default (and, in most cases, the only) choice, while in English, the N-PREP-N (“wire of steel”) 3  On a more general level cf. also the brief comparison of availability of Noun-Noun structures in French, German and English in Bauer, Lieber, and Plag (2013, p. 626).

A Paradigmatic Approach to Compounding

177

structure is found only in specific contexts (e.g., “wire of steel or rubber?”) and the A-N structure is possible (and in that case, it is also the default choice) only when an appropriate adjective is available (as in “wooden floor”). In French, phrasal lexemes with the structure N-PREP-N seem to be the default (and in most cases the only) choice. It seems reasonable to assume that these restrictions do not derive from general rules; they derive rather from the existing paradigms (or, more specifically, “series” in paradigms) that already encode the “made of” relationship in a given language (cf. Figure 7.2 above). Thus, a good knowledge of language requires a good knowledge of paradigms, and the choice of an appropriate naming process will be guided by the knowledge of paradigms that exist in the mental lexicon. 4.3 Hierarchical Organization of Paradigms In §3, it was pointed out that paradigms are organized hierarchically, i.e., that single paradigms form more complex paradigmatic systems with common abstract properties. This feature may account for different phenomena, such as the position of the head in compounds or the principles that underlie the insertion of phrases in compounds. The principles that determine the position of the head in compounds have been discussed at least since the early 1980s. Since then, it has been shown that the position of the head is neither a universal (as claimed in the “righthand head rule” by Williams, 1981), neither a parameter, in the sense that the same language would only have left-headed or right-headed compounds but not both types (cf. Scalise, 1994, p. 194). Indeed, data from various languages allow for claiming that the position of the head in compounds may vary within the same language with respect to different parameters, such as the type of the compound (e.g., in Chinese, verbal compounds are right-headed, while nominal compounds are left-headed) or the semantic features of the compound (as in Nizaa).4 In the following paragraphs, it will be suggested that this variability in the head position may be accounted for by the vertical organization of paradigms. Italian has three types of N-N compounds with respect to the position of the head: left headed (17a), right-headed (17b), and reversible (17c) compounds. Although the left-headed type is the most common, right-headed compounds

4  Cf. a detailed discussion in Scalise and Fábregas (2010, pp. 116–118), Masini and Scalise (2012, pp. 81–82), and Radimský (2015, pp. 130–135).

178

Radimský

are also an available pattern,5 and reversible compounds (17c) seem to be particularly challenging for any type of rule-based approach. (17a) linea guida, “guideline”

trattamento rifiuti “waste treatment”

(17b) cantoterapia, drammaterapia, “singing therapy” “drama therapy”

arteterapia “art therapy”

(17c) autonoleggio vs. “car rental” radiogiornale vs. “radio news”

noleggio auto “car rental” giornale radio “radio news”

Left-headed compounds (17a) are the easiest to account for, as they represent the default Romance N-N compounding paradigm that may be captured by the schema (18). (18) [Ni-head Nj-non-head]Nk ↔ [SEM i with any relation to SEM j]Nk Italian right-headed compounds are commonly formed within the learned compounding pattern that involves at least one learned bound stem (confix— “C”), be it on the leftmost position (19a), on the rightmost position (19b), or on both positions (19c). (19a) [[agro]C[industria]N]N “agro-industry” (19b) [zanzari]N[cida]C]N “mosquito-killer” (19c) [[aracno]C[fobia]C]N “arachnophobia” Indeed, a detailed analysis of 1390 Italian compounds from 1980–1999 (Radimský, 2006) has showed that whenever a confix appears on any position in the compound, the output is always a right-headed compound. Thus, learned compounding, which may be captured by the general schemas (20a–c),

5  Indeed, relics from Latin (such as terremoto—“earthquake”) or calques from Germanic languages (such as scuola bus—“school bus”) are not taken into account in this discussion, cf. Radimský (2015, pp. 130–135).

A Paradigmatic Approach to Compounding

179

is triggered by the presence of specific input units (confixes) where each one forms its own right-headed paradigm. (20a) [Ci-non-head Nj-head]Nk ↔ [SEM j with any relation to SEM i]Nk (20b) [Ni-non-head Cj-head]Nk ↔ [SEM j with any relation to SEM i]Nk (20c) [Ci-non-head Cj-head]Nk ↔ [SEM j with any relation to SEM i]Nk Data allow us to hypothesize that in modern right-headed Italian N-N compounds, there is also a specific element that “triggers” the right-headed structure. It could be a former confix (a “second generation element”),6 such as radio (21a) or auto (21b), a scientific loan word from modern languages, such as laser (21c), or even a noun, such as terapia (“therapy”), that simply has a rich paradigm within the C-N pattern (22), so that its presence even in N-N compounds regularly triggers the right-headed structure (21d) as if it were a confix. (21a) [radio]N[drammaturgia]N]N “dramaturgy of radio broadcast” (21b) [auto]N[noleggio]N]N

“car rental”

(21c) [laser]N[chirurgia]N]N

“laser surgery”

(21d) [danza]N[terapia]N]N

“dance therapy”

(22) C-[terapia]N psicoterapia “psychotherapy” fisioterapia “physiotherapy” fitoterapia “phytotherapy” ippoterapia “hippotherapy” 6  The term “second generation element” (Iacobini, 2004, p. 74) denotes confixes that have undergone a substantial change in meaning and function typically by the process illustrated below, where the confix auto1 (“(by it)self”) forms a learned compound (automobile) whose form is subsequently truncated without changing its meaning and category. Thus, the new “second generation” element (here, auto2 “car”) is usually a noun, but it keeps triggering the right-headed compounding paradigm, as if it were a confix. The process illustrated in the example (i) probably took place first in French (cf. TLFi) and then in other Romance languages (cf. Zingarelli, 2011).   i) [vettura]N[[auto1]C[mobile]A]A (“self-propelled vehicle”) > [auto1mobile]N (“car”) > auto2 (“car”)

180

Radimský

To sum up, the fact that some Italian nouns trigger right-headed compounding schemas in N-N structures is a direct consequence of the family size effect that results from the presence of these forms in learned compounding schemas (C-N, N-C, or C-C). Some of these nouns form essentially right-headed compounds (17b), while others may even form reversible compounds (17c)— which is an interesting example of competition between paradigms. The vertical organization of paradigms may also explain the principles that underlie the insertion of complex constituents in compounds. As shown in previous studies, most of syntactic operations are impossible within Romance N-N compounds, since compounds are syntactic atoms (see especially Bisetto & Scalise, 1999, pp. 37–42), but this principle seems to have one exception: the insertion of complex constituents (such as compounds or phrases) in compounds has proven to be possible under certain circumstances (see Lieber & Scalise, 2006; Radimský, 2015, pp. 30–36). However, the nature of these “circumstances” remains far from being clear. The explanation put forward in Radimský (2015, 189–190 and 218–233) consists in the assumption that the element that triggers the interpretation of the compound usually does not allow for insertion of phrases because it serves to identify the compounding paradigm, while free insertion of phrases on the other element is possible. Thus, Italian verbal-nexus compounds (23a) allow for insertion of complex constituents on the argument position (N2), because the triggering element for their interpretation is a deverbal head (N1). Conversely, Italian attributive-appositive (ATAP) compounds (23b) allow for insertion of complex constituents on the head position (N1), as the triggering element that identifies the paradigm is the modifier (N2). (23a) [gestione [risorse umane]NA]

“human resources management”

(23b) [[settore tecnologico]NA chiave] “key technology sector” As shown in the cited source, this hypothesis is supported by extensive data extracted from the ItWac corpus (Baroni, Bernardini, Ferraresi, & Zanchetta, 2009, 2013). 4.4 Schema Unification Paradigms, in a broad sense of the word, may interact in many different ways. In §4.3, we have analysed an example of competition between two hierarchically organized paradigmatic systems, yielding left-headed and right-headed compounds. In this section, we will deal with the situation in which two

A Paradigmatic Approach to Compounding

181

paradigmatic systems “merge” into a single one. In the Construction Grammar terms, the simultaneous use of two-word formation schemas has been described as “schema unification” (Booij, 2016, pp. 433–434). It will be argued that this property makes it possible to account for the creation of French pseudorelational N-N-N compounds, such as (24). (24) [[assurance]N1 [décès]N2a-[invalidité]N2b]N “death and disability insurance” French, as well as Italian, has a well-established paradigmatic system that yields relational N-N-N compounds, such as (25a) (Radimský, 2018; Radimský, 2015, pp. 137–152). Romance relational compounds are made up of three nouns [N1-N2a-N2b], with a subordinate relationship between N1 and N2a–b, on the one hand, and an exocentric coordinate relationship between N2a and N2b, on the other. They have, amongst others, two interesting properties that seem to be mutually conditioned. First, [N2a-N2b] is not a lexical unit, as it cannot appear without the head noun (25b). Second, the head noun behaves as “bivalent” or “polyvalent”, since it does not allow a single nominal modifier (25c). Data from the FrWac corpus (Ferraresi, Bernardini, Picci, & Baroni, 2013; Baroni et al., 2009) allow us to hypothesize that this paradigmatic system is a well-established one.7 (25a) interface homme-machine “human-machine interface” (25b) *[homme-machine]N “*[human-machine]N” (25c) *interface homme-∅ *“[human]N-∅ interface” Interestingly, the pseudo-relational compounds, such as (24), have the same internal structure as relational compounds, including the fact that [N2a-N2b] is not a lexical unit (26). However, as shown in (27a–b), the head of pseudorelational compounds is not bivalent or polyvalent. Indeed, (27a–b) are two independent subordinate N-N compounds with a head and a single nominal complement.

7  In Radimský (2018) a sample of 880 French relational compounds (types) with 96 different heads (types) was extracted from the FrWac corpus and analyzed.

182

Radimský

(26) *[décès-invalidité]N “*death-disability” (27a) [[assurance]N1 [décès]N2a]N

“death insurance”

(27b) [[assurance]N1 [invalidité]N2b]N “disability insurance” The question thus arises as to why the monovalent head of pseudo-relational compounds takes a binominal [N2a-N2b] complement and yields trinominal pseudo-relational compounds. The explanation put forward in Radimský (2018) presumes that it is a result of a schema unification where two paradigmatic systems are merged. In other words, two subordinate N-N compounds, such as (27a) and (27b), that have the same head as well as the same headcomplement relationship (and, therefore, pertain to the same paradigm, or more precisely, even to the same family) merge according to the paradigmatic system of relational compounds into one pseudo-relational compound (24). Data has shown that French pseudo-relational compounds are not exceptional: in Radimský (2018), nearly 500 types (15000 tokens) have been identified in the FrWac corpus. 5

Conclusion

The aim of this paper was to assess whether modern paradigmatic approaches known from derivational morphology also have any relevance for the analysis of compounding. The answer is positive: compounding is a paradigmatic morphological phenomenon, and the study of compounding paradigms makes it possible to account for different phenomena that are challenging for rulebased approaches. Similarly to what has been claimed for inflectional and derivational paradigms, even in compounding, there are different types of paradigmatic relations that should be distinguished, and only some of them are relevant for a morphological analysis. On the one hand, compounds and their bases form (partial) families, paradigms, and complex paradigmatic systems, whose morphological pertinence has been demonstrated on examples in §4. On the other hand, they also form word nests (or complete families) that seem to be much less important. The paradigmatic approach is very closely related to the theoretical model of Construction Morphology, even so that it may sometimes seem that “paradigms” and “constructions” are just two different names for the same concept. However, despite the similarities, the differences in both the terminology and

A Paradigmatic Approach to Compounding

183

the epistemological background make the two approaches rather complementary and mutually inspiring.

Acknowledgements

This research has been supported by the research grant GAČR (Grant Agency of the Czech Republic) No. GA17-17253S: “N-N Compounding in Contemporary French”. References Baroni, M., Bernardini, S., Ferraresi, A., & Zanchetta, E. (2009). The WaCky wide web: a collection of very large linguistically processed web-crawled corpora. Language Resources and Evaluation, 43(3), 209–226. DOI: 10.1007/s10579-009-9081-4. Baroni, M., Bernardini, S., Ferraresi, A., & Zanchetta, E. (2013). itWaC. Institute of the Czech National Corpus, Prague: Charles University. Retrieved from: http://www .korpus.cz [Accessed April, 2018]. Bauer, L. (2017). Notions of paradigm and their value in word-formation. Paper delivered at ParadigMo 2017, First Workshop on Paradigmatic Word Formation Modeling, Toulouse, France. Abstract retrieved from http://w3.erss.univtlse2.fr/ ParadigMo2017/paradigmo_abstracts.pdf [Accessed April, 2018]. Bauer, L., Lieber, R., & Plag, I. (2013). The Oxford reference guide to English morphology. Oxford: Oxford University Press. Bisetto, A., & Scalise, S. (1999). Compounding. Morphology and/or syntax? In L. Mereu (Ed.), Boundaries of Morphology and Syntax (pp. 31–48). Amsterdam: John Benjamins. Bonami, O., & Strnadová, J. (2019). Paradigm structure and predictability in derivational morphology. Morphology, 1–31. Springer. DOI: 10.1007/s11525-018-9322-6. Booij, G. (2010). Construction morphology. Oxford: Oxford University Press. Booij, G. (2016). Construction morphology. In A. Hippisley & G. Stump (Eds.), The Cambridge handbook of morphology (pp. 424–448). Cambridge: Cambridge University Press. Boyé, G., & Schalchli G. (2016). The status of paradigms. In A. Hippisley & G. Stump (Eds.), The Cambridge handbook of morphology (pp. 206–234). Cambridge: Cambridge University Press. Boyé, G., Hathout, N., & Roché, M. (2011). Des unités morphologiques au lexique. Paris: Hermes.

184

Radimský

Ferraresi, A., Bernardini, S., Picci, G., & Baroni, M. (2013). frWaC. Institute of the Czech National Corpus, Prague: Charles University. Retrieved from: http://www.korpus.cz [Accessed April, 2018]. Haspelmath, M., & Sims, A. (2010). Understanding morphology. London: Hodder Education. Hathout, N. (2011). Une approche topologique de la construction des mots: propositions théoriques et application à la préfixation en anti-. In M. Roché, G. Boyé, N. Hathout, S. Lignon, & M. Plénat (Eds.), Des unités morphologiques au lexique (pp. 251–318). Paris: Hermès, Lavoisier. Hathout, N., & Namer, F. (2018). ParaDis: a Families-and-Paradigms model for derivation. The case of prefixed privative verbs in Italian. In A. Bagasheva & J. Fernández-Domínguez (Chairs), Workshop Revisiting paradigms in wordformation, conducted at the conference Word-Formation Theories III & Typology and Universals in Word-Formation IV, Košice (Slovakia), 27–30 June 2018. Iacobini, C. (2004). Composizione con elementi neoclassici. In M. Grossmann & F. Rainer (Eds.), La formazione delle parole in italiano (pp. 69–95). Tübingen: Max Niemeyer Verlag. Lieber, R., & Scalise, S. (2006). The Lexical Integrity Hypothesis in a new theoretical universe. Lingue e linguaggio, 1, 7–32. Bologna: Università degli studi di Bologna. Masini, F., & Scalise, S. (2012). Italian compounds. Probus, 24(1), 61–91. Peytard, J. (1975). Recherches sur la préfixation en français contemporain. Paris: Honoré Champion. Radimský, J. (2006). Les composés italiens actuels. Paris: Cellule de recherche en linguistique. Radimský, J. (2015). Noun+Noun compounds in Italian. A corpus-based study. České Budějovice: University of South Bohemia. Radimský, J. (2018). Les composés relationnels français du type ‘relations parentsenfants’. Linguistica Pragensia, 2, 131–153. Prague: Charles University. Roché, M. (2011). Quelle morphologie? In G. Boyé, N. Hathout, & M. Roché (Eds.), Des unités morphologiques au lexique (pp. 15–40). Paris: Hermes. Scalise, S. (1994). Morfologia. Bologna: Il Mulino. Scalise, S., & Fábregas, A. (2010). The head in compounding. In S. Scalise & I. Vogel (Eds.), Cross-disciplinary issues in compounding (pp. 109–126). Amsterdam: John Benjamins. Štekauer, P. (2014). Derivational paradigms. In R. Lieber & P. Štekauer (Eds.), The Oxford Handbook of Derivational Morphology (pp. 354–369). Oxford: Oxford University Press. Syn2015. Křen, M. Cvrček, V., Čapka, T., Čermáková, A., Hnátková, M., Chlumská, L., Kováříková, D., Jelínek, T., Petkevič, V., Procházka, P., Skoumalová, H.,

A Paradigmatic Approach to Compounding

185

Škrabal, M., Truneček, P., Vondřička, P., & Zasina, A. (2015). SYN2015: reprezentativní korpus psané češtiny. Ústav Českého národního korpusu. Prague: Charles University. Available at http://www.korpus.cz [Accessed April, 2018]. TLFi : Trésor de la langue Française informatisé, ATILF—CNRS & Université de Lorraine. Retrieved from: http://www.atilf.fr/tlfi [Accessed April, 2018]. van Marle, J. (1985). On the paradigmatic dimension of morphological creativity. Dordrecht: Foris. Williams, E. (1981). On the notions ‘lexically related’ and ‘head of a word’. Linguistic Inquiry, 12(2), 245–274. Zingarelli, N. (2011). Lo Zingarelli 2011, vocabolario della lingua italiana, CD-ROM. Bologna: Zanichelli.

Chapter 8

Of Brownie Girls and Aussie Families: A New Look at Morphosemantic Paradigmaticity in Adj+ie/y Nominalisations José A. Sánchez Fajardo and Elizaveta Tarasova 1

Introduction

According to Oxford English Dictionary (OED3), -ie and -ey are spelling variations of the suffix -y, whose primary function is to form adjectives from nouns, verbs and other adjectives to convey the meaning of ‘full of, having the quality of’, as in messy, milky, or ‘inclined to, apt to’ as in sticky. It is also noted that some of the formations of this kind may have depreciatory reference, e.g., boozy, tinny. Much of the information on the etymology of the suffix seems to relate it to the pragmalinguistic functions of endearment and diminution, but there is still some fuzziness in regards to its origin (Shields, 2001). Marchand (1969, p. 298) claims that the earliest formations date back as far as the 15th century in Scotland, whereas other authors believe that it might have originated from Old English -ig (Charleston, 1960, p. 123). The information on how the uses, form and meaning of the suffix developed over time is also inconclusive (Bauer, Lieber, & Plag, 2015, p. 388). According to OED3, the use of the suffix for forming diminutives, e.g., auntie/aunty, nightie, Tommy, is not registered until Middle English. Bauer, Lieber, and Plag (2015) state that “[i]t is related to the Germanic suffix that gives us Dutch -tje and German -chen” (pp. 388–389). It is also unclear which spelling should be viewed as original. For example, OED3 lists -y as the original suffix, and -ie as its spelling variation, whereas Merriam-Webster Dictionary (MWD11) does the opposite. More­ over, the latter seems to segregate the suffixes and attributes the formation of diminutives (ducky) to -y, while the formation of nominalisations (townie) to -ie. An obvious question in this regard is whether -y and -ie are different spelling variations of the same suffix or whether they are different homophonous suffixes. Examples like browny in reference to the colour, e.g., browny blondish hair, and brownie in reference to a kind of cake demonstrate that things are not as clear as they may seem on the surface. It is also unclear whether formal phonological properties of the suffix(es) in cases like jockey, sonny, brownie, and horsie are notably different, and if so, whether this may be connected with the various functions of the suffix and the shades of meanings it realises. This © Koninklijke Brill NV, Leiden, 2020 | doi:10.1163/9789004433410_009

Of Brownie Girls and Aussie Families

187

may or may not be connected with the differences in spelling. The analysed literature does not focus on this question, but it would be interesting to see if the /i/ sound has a higher pitch in cases like dearie when it is used as a term of endearment, and deary in expressions like “Deary me!”. Since phonological issues are not in the centre of this investigation, we will not focus on such matters here; however, we believe that research in this area may be beneficial as it would allow for a better understanding of how the developments in the semantics of the suffix may be connected to the changes in its formal properties and cognitive categories. One point that seems clear is that, irrespective of the origin and history of the development of the original suffix, at the current moment of time “the adjective-forming suffix which gives cheery, crazy, dotty […] is a different suffix, despite its homophony with the diminutive suffix -ie and […] similar connotations” (Bauer, Lieber, & Plag, 2015, p. 391). Thus, in this investigation, we focus on the diminutive suffix -ie and, following Bauer, Lieber, and Plag (2015), we assume the position that {-ie} is the base form of the diminutive morpheme, while {-y} and {-ey} are its variations. The purpose of our research is to outline semantic networks that we believe exist within -ie nominalisations, in order to see how they fit with the general understanding of the place and function of the -ie suffix in the morphology of present-day English. We also aim to investigate the connection between word-building paradigms or patterns of formation of -ie nominalisations and semantic characteristics that units comprising these paradigms demonstrate. This is an often neglected type of derivation that is sometimes described on the boundary with evaluative word formation. In this case study in particular, the notion of paradigm is extended outside other patterns where no or little evaluative meaning occurs. 2

Units under Investigation

The suffix -ie, when used in the formation of complex nouns, has been traditionally associated with child talk (tummy) and hypocoristic use (Jimmie). This certainly is the case if we come across a sentence like: “Our goalie Robbie hurt his pinkie when playing footy because he was distracted by a mossie”. However, as our data show, -ie derivatives are not necessarily limited to these, as seen in the following excerpts (1–3) extracted from News on the Web Corpus (NOW): (1) Preparing to fight the mossie bite. Riverine Herald (AU-08.01.2017) (2) Turnovers and loose defensive zone coverage were a problem at times for the Warriors, but goalie Andrew Vigliatore was there to cover up most mistakes. The Beacon Herald (CA-09.02.2018)

188

Sánchez Fajardo and Tarasova

(3) The showdown will give footy fans a chance to get their first look at Capras key recruits Jerry Key, Dave Cowhan, Connor Broadhurst and Blake Goodman. Rockhampton Morning Bulletin (AU-06.02.2018) Moreover, the semantic composition of -ie in the nominalising function has proven to be more complex; MWD11 includes the following definition of -ie: 1. Little one; dear little one: birdie, sonny. 2. a. One belonging to; one having to do with: townie. b. One who is: preemie. 3. One of (such) a kind or quality: cutie, toughie. These meanings are usually complemented by the connotations the suffix may convey, which range from diminution (doggie) and endearment (dearie) to pejoration (Chinky, Jewbie). The use of the suffix in the formation of nouns from adjectives, verbs and other nouns (tough > toughie, stick > sticky, handkerchief > hanky) is not sufficiently described in the literature, and only a few authors look at the issues of -ie suffixation (Marchand, 1969; Cannon, 1987; Plag, 1999; Lipka, 2002; Bauer, 2008; Bauer, Lieber, & Plag, 2015). This is probably connected with the fact that -ie units are mainly used in informal speech and their status in the system of the English language is obscure. That is, these constructs are not easy to class according to conventional word-formation mechanisms, probably owing to the overlapping categories they may be ascribed to (i.e., diminutives, hypocoristics, derogatory terms, slang). Thus, assuming a semantics-oriented approach could provide some information on how these words come into existence. These units are interesting for investigation in terms of both their morphology and semantics. According to Scalise (1984), complex words like these belong to the domain of evaluative morphology, a special kind of morphology that can account for formal strategies applied to formation of evaluative constructions, which “[…] cannot be numbered among word-formation processes” (Grandi, 2017, Summary and Keywords). Scalise’s hypothesis about evaluatives representing a separate block of rules that borders on inflection and derivation has received a lot of attention and has by now developed into a separate field of linguistic studies (Grandi, 2017; Grandi & Körtvélyessy, 2015). However, the arguments of the opponents of this view are hard to disregard (see Bauer, 1996; Stump, 1993 for discussion). The presented research is focusing on -ie nominalisations derived from adjectives (henceforth, Adj+ie), e.g., brownie, whitey, toughie, cutie, and is concerned with the regularities in their formation as well as the meanings they realise. Our investigation takes the position in which, on the one hand, we follow Bauer’s (1996) opinion that the observed patterns of behaviour may be explained “in terms of general cognitive processes and particular uses” (p. 31).

Of Brownie Girls and Aussie Families

189

On the other hand, we accept Grandi and Körtvélyessy’s (2015) arguments for the importance of the evaluative morphology framework in the analysis of evaluative constructions, to which Adj+ie nominalisations belong. Grandi and Körtvélyessy (2015) point out that despite the observed inconsistencies, the multifunctionality of evaluative suffixes “can be organised in recurrent and regular networks of semantic and pragmatic relations” (p. 5), both crosslinguistically and within one language. Stated differently, although we generally aim at modelling regularities and common paradigms, we are also aware that not all the cases under study might fit into a unified approach. This premise reflects the general concept of derivational paradigm that we discuss here. In our study, we follow Fernández-Domínguez’s (2019) approach, who presents a derivational paradigm dynamically, as “the fact that, if a construction is coined via a synchronic syntactic procedure, it will be placed together with the previous constructions created by that rule” (p. 212). Nonetheless, a number of studies suggest that the notion of paradigmaticity as a universal principle or as a logical choice for irregular patterns is still under discussion (see Bagasheva, this volume). The inherited property of paradigmaticity, particularly in derivational morphology, has given rise to a strong correlation between the analytic strategies of word formation and the heuristic structure of meaning. Therefore, the number of constructions (or derivatives as it is our case study) constitute a solid referential body that allows for the construction of the governing rule. In this study, our approach is driven by the analysis of the semantic content of Adj+ie nominalisations, and we aim to investigate how the semantic content of the output lexemes is influenced by the semantics of the base, the semantics of the suffix, as well as other linguistic and extra linguistic factors. Our interest in the semantics of Adj+ie nominalisations is driven by the fact that most of them are polysemous, with some of the derivatives used in up to half a dozen meanings. Bardsley and Simpson (2009) provide the example of roughie/roughy, which they report to be used in the following senses in Australian and New Zealand Englishes: For Australia these include: a rough person, a rough-leaved pineapple, a Tommy Rough (fish), a shrewd trick, an unbroken horse, an unpleasant job. In New Zealand, a wild cattle beast, a poor or unqualified tradesman, a poorly performing racehorse, a poor sports game, and a disobedient or poorly-trained sheep dog are amongst further uses. In both countries, roughie is also used for a racehorse that starts at long odds. (p. 51) This wide-ranging extension of meaning is owed to various semantic and syntactic conditions underlying the suffixation of adjectival bases, as well as

190

Sánchez Fajardo and Tarasova

to the evaluative nature of the suffix. Such semantic variability is especially acute in cases where pejoration or marginalisation restricts the usage of some of these derivatives: blackie, commie, whitey, etc. Interestingly, pejoration is not always pointed out in the definitions of these derivatives compiled from English dictionaries (MWD11; The American Heritage Dictionary (TAHD5); OED3; Cambridge English Dictionary (CED4)), but is often observable when the context is considered, as shown in examples (4–6) from NOW: (4) They would call me ‘blackie’ and ‘burnt crisp,’ she said. The Hindu BusinessLine (IN 11-08-18) (5) The second lieutenant became known as the commie cadet, not the best nickname in the U.S. military. Toronto Star (CA 21-06-18) (6) … leading many of us to suspect that BLMUK thinks black lives matter only when whitey is doing the killing and that otherwise it’s nothing to write home about. Spectator.co.uk (UK 09-05-18) 3

Limiting the Scope of the Study

There is a considerable amount of research on the formal properties of diminutives, some of which we draw upon in the course of this study. In particular, we follow Schneider (2003) in that we differentiate between levels of analysis, i.e., diminutive formation (formal level) and diminutisation (semantic level). This allows us to focus on the concept of diminution, which we, following Schneider (2003), view as “the interface between concepts of quantification and qualification, in that it combines aspects of size and attitude, and more particularly of smallness and appreciation and depreciation” (p. 4). The important point in this regard concerns the concept of diminutisation on the semantic level (Schneider, 2003). Consideration of the semantic level allows for seeing the connection between using the morphological means of diminutive formations for knocking down the size of the referent (diminution proper) and for expressing the attitude of the speaker towards the referent with the intent of expressing affection (endearment) or marginalising its value (pejoration). Words like toughie, brownie, sweetie are sometimes analysed as hypocoristics of existing words rather than new coinages, because it is often hard to draw clear-cut borders between the two (Bardsley & Simpson, 2009, p. 49). This distinction is not a matter of concern for this investigation. However, the development of clear-cut criteria that would allow for a distinction between hypocorisms and new coinages may be helpful for cases like jockey and mickey/micky, and may help shed light on the changes in the form and meaning of the original -y suffix in the course of its existence in the language.

Of Brownie Girls and Aussie Families

191

One of the aims of this investigation is to demonstrate to what degree the semantic content of the analysed derivatives is influenced by the semantics of the base and that of the suffix. The analysis presented here focuses mainly on the notions of ‘origin’ and ‘colour’, e.g., Hunkie < Hungarian, blackie < black, to establish a correlation between morphosemantic paradigms. The Adj+ie nominalisations that fit these two input categories are used to determine the morphological and semantic paradigms underlying the formation of these deadjectival suffixed nouns. The difficulty of the analysis lies in the variety of the units involved, which are found to comply with the following word-formation formula: Deadjectival base + (ellipted noun) + -ie/y = diminutivised deadjectival nominalisation As proposed in §2, we use the term Adj+ie to refer to the resulting ‘diminutivised deadjectival nominalisation’ for the purposes of this analysis; however, we admit that the term does not necessarily add much accuracy to the description of units involved in this word-formation process. The use of the term Adj+ie is dictated by its explicit reference to the original nature of these forms, regardless of their ultimate senses. With the purpose of our investigation in mind, we focus on two research questions: (a) What semantic qualities are the ellipted noun (referent), the adjectival base, and the suffix -ie characterised by? (b) Are there any specific trends that are motivated by the semantic scale of diminution, pejoration, and endearment? Even though the main focus of this investigation is on -ie nominalisations derived from adjectives denoting colour and origin, the analysis of all Adj+ie constructions in our dataset reveals a general morphological dynamic regardless of the semantic category of the base. An adjectival phrase (a syntactic structure) is reduced to a nominalisation unit by means of two mechanisms: ellipsis and suffixation, as in (7), in which dutchie and blackie are reduced derivatives of Dutchman and black person respectively. Occasionally, adjectival bases might also undergo a process of clipping, as in (8), in which fund- is a clipped adjectival base, from a fundamentalist religious person. The collected data also suggest that the relevant -ie forms are never premodified. (7) A Dutchie is a dutchie, a blackie is a blackie, a charro is a charro. News24 (NZ 13-01-13) (8) … can you imagine the rhetoric of the Bible thumping fundie, and Catholic, preachers in the lead up. News24 (NZ 13-01-13) Based on the multiple correlations among some of these syntactic and morphological components, our study is centred on examining the semantic content

192

Sánchez Fajardo and Tarasova

of -ie nominalisations. In particular, we look at how meanings of these derivatives might be activated due to the contribution of the following components: (a) the meanings of the attached suffix (additive meaning); (b) the meaning of the adjectival base (basal meaning); or (c) the meaning of the ellipted or omitted noun that is part of the original collocation, which is implicitly conveyed (inherited meaning). Thus, our hypothesis is that Adj+ie nominalisations, particularly those related with the notions of ‘origin’ and colour’, show consistent variations in meaning depending on the referent of the ellipted noun. This omitted unit provides ‘visible’ semes or nuances to the nominalisations, whose categories are also complemented by ‘additive meanings’ such as pejoration, endearment, and diminution contributed by the suffix. 4

Theoretical Prerequisites

Even though there is no unified opinion on the origins of the suffix, it has been suggested that -ie in its functions of forming diminutives and nominalisations emerged as an “expressively motivated morpheme” (Marchand, 1969, p. 298), and being that, it is tightly linked to the concept of “symbolic value” in morphology (cf. Jespersen, 1933). Interestingly, some of the theories that relate morphological symbolism to historical origin are based on parent-child interaction (Pinker, 1984, p. 249) and “panchronic laws” (Shields, 2001), which may characterise “the overall direction of linguistic change” (Fox, 1995, p. 194). The work of panchronic laws in regards to the suffix -ie is demonstrated on the example of cross-linguistic acquisition and use of the word ending /i/ in various Indo-European languages, and in the formation of clipped personal names (Sánchez Fajardo & Rodríguez González, 2018). The linguistic and cognitive evidence about the evolution of diminutive affixes in general is provided by Jurafsky (1996), who demonstrates how diminutive formations in various languages have a strong relation to the sense ‘child’, which he, following Wierzbicka’s (1985) intuition, believes to be at the heart of many pragmatic uses of diminutives. Shields (2001) builds on Jurafsky’s (1996) suggestion and explains that the English suffix -ie/y adopted diminutive signification from the word baby, and this signification was generalised to other word forms (e.g., nominalisations) and to related meanings with time. 4.1 Functions and Meanings The analyses of the use of the suffix -ie in adult talk traditionally class it as a ‘familiarity marker’ (e.g., Quirk, Greenbaum, Leech, & Svartvik, 1985, p. 1584). This reveals the nuance of informality, or perhaps non-seriousness associated

Of Brownie Girls and Aussie Families

193

with the suffix (Dressler & Merlini, 2001, p. 144), which is strongly connected with its hypocoristic use and thus is mainly associated with noun-based nominalisations, e.g., barbie for ‘barbecue’, steppie for ‘stepmother’. Wierzbicka (1985) claims that the way speakers use hypocoristics in modern Australian English reflects good humor, love of informality and tendency to reduce the size of things in language by the speakers. In their discussion of hypocoristics in Australian and New Zealand Englishes, Bardsley and Simpson (2009) look at the uses of hypocoristic forms at different historical periods, and their data demonstrate the developments in the function and meaning under the influence of various factors. It is obvious that at least some of the meanings associated with the hypocoristic use are present in Adj+ie nominalisations. The wide range of meanings and functions realised by the -ie suffix is also a reflection of its complexity in the formation of derivatives, as the derivation process clearly involves both semantic and pragmatic factors. Schneider (2003) finds a functional equivalence between the formation and structure of deadjectival derivatives such as sweetie, cutie, or darkie, with a well-known and more frequent paradigm of denominal units, i.e., lovie, dearie, townie, etc. One of the functions of -ie formations in adult communication is to express adults’ relationships to other adults. Schneider’s (2003) analysis of the semantic pattern, particularly in the case of vocatives, is based on the interlocutors that take part in a communicative stretch: […] the adjective denotes a quality or characteristic trait, the diminutive denotes a person who has this quality or trait. When formations of this type are used to realise vocatives, the speaker focuses on a particular trait of the addressee or assigns this trait to the addressee. The speaker’s attitude towards the addressee, reflected in the selection of such a descriptor, can be positive or negative, depending on the quality or trait which is focussed on. Accordingly, such diminutives can be appreciative or depreciative. Schneider, 2003, p. 156

This observation is connected with what Dixon (2004) describes as the representation of the ‘smallness’ for a particular communicative purpose. The choice of representing a referent through a diminutive is induced by the association of diminution with attitudinal forms such as affective, expressive, or evaluative. This perception of diminutive units is in conjunction with the componential analysis of these derivatives, which reveals that “[+small] is part of the denotation, while the attitudinal meaning is part of the connotation or ‘associative meaning’” (Schneider, 2003, p. 11).

194

Sánchez Fajardo and Tarasova

The idea of smallness, whether it is present as part of the denotational meaning or as a reflection of the speaker’s attitude demonstrates the longstanding uncertainty in regards to what should be classed as merely diminutives (Bybee, 1985; Anderson, 1982). The historical connection between the notions of smallness and the speaker’s attitudes is discussed by Strang (1968), who suggests that diminutives may “have begun by meaning ‘a small one of its kind’, but have undergone a development whereby they come to express not merely an assessment of size, but also, or even exclusively, the speaker’s response to small things, a response ranging from affection through condescension to contempt” (Strang, 1968, p. 138). This perspective of the semantic extension underlying the formation of such units confirms the need to examine diminutives through a multilayered mechanism, in which morphosemantic features are interpreted through a ‘pragmatic module’, and contextual factors are considered (Schneider, 2003, p. 12). 4.2 Matters of Productivity Of all the diminutive suffixes in English, most of which have restricted productivity, the suffixes -ie and -ette seem to be exceptions (Bauer, Lieber, & Plag, 2015). Cannon (1987, p. 185) and Bauer, Lieber, and Plag (2015, p. 389) state that -ie is a highly productive suffix in English, and some quantitative analyses (cf. Cannon 1987, Sinclair, 1991) support this claim. At the same time, its productivity is far more visible in Scottish, Australian, and New Zealand Englishes than in British or American varieties. Leisi (1969, p. 89, as cited in Schneider 2003, p. 77) reports that the use of this suffix in British or American Englishes is restricted to child language, or it may be used as a familiar substitute for the second element in compounds, e.g., bookie (< bookmaker), undies (< underclothes). The limited use of -ie in British and American Englishes explains why the analyses conventionally associate it with child talk or hypocoristic names; however, as shown in Cannon (1987) and Bauer, Lieber, and Plag (2015), the existing empirical evidence suggests the necessity of expanding the borders of the analysis. When talking about productivity of a word-formation pattern, it is important to consider its availability and profitability. Availability of the pattern is understood as the scope of a word-formation rule and is associated with the qualitative aspects of productivity, while profitability of the pattern refers to the use of this rule, and is quantifiable (Bauer, 2001). The interrelation and interaction between availability and profitability is discussed by Fernández-Dominguez (2015), who demonstrates how they work together in the process of developing paradigmatic and syntagmatic relations in the language respectively. In the case of -ie nominalisations (both deadjectival and denominal), we can say that the pattern can be considered available; however, its profitability is

Of Brownie Girls and Aussie Families

195

questionable due to the fact that domains of its use are quite restricted. The corpus-based data collected for this research show that -ie nominalisations are majorly used in slang and informal communication. The majority of the analysed examples realise a wide range of meanings, most of which are emotionally coloured (see §6 for discussion), with only a marginal number of cases sneaking into more regular use, e.g., brownie. Applying a traditional understanding of productivity, which refers to “rulegoverned word-formation processes, which are carried out by the creation and comprehension of new words” (Fischer, 1998, p. 17) to our discussion, it is possible to say that these nominalisations might be characterised by a clear dichotomy between what is considered [+small] and what is not size-related. Even though the number of Adj+ie nominalisations is not very large, they tend to be polysemous with a number of meanings (not necessarily connected with size) associated with them. The extensions of meanings allow for the expansion of use of such items in different contexts, thus compensating for the limited formal productivity of Adj+ie nominalisations as compared to denominal -ie derivatives, which are productive. One of the most visible differences between deadjectival (stiffie/stiffy) and denominal (baccy/baccie) nominalisations is that the former might also be dependent on the associative meaning of the ellipsed nouns. In the examples (9), (10), and (11), collected from the first 100 hits on the NOW, the word stiffy/stiffie is conditioned by the meanings of ‘a stiff/erected penis’ (9), ‘a stiff disc’ (10), and ‘an invitation on “stiff” surface’ (11). Alternatively, baccy in (12) does not undergo any referential change, but it does convey an attitudinal modification towards the interlocutor, i.e., familiarity. (9) Les mauvaises langues have it that Emmanuel Macron gets a stiffie every time he sees Angela Merkel. Deccan Chronicle (IN 14-05-2017) (10) Duke3D sat on 35 3.1/2 inch floppy disk drives (Commonly referred to as a stiffy). Lazygamer (ZA 14-02-2014) (11) [A] printed invitation by post once had. Perhaps a formal, old-fashioned “stiffy” is the only way to elicit a response? Telegraph (UK 24-03-2014) (12) She’d bought father’s baccy at Mr. Crewe’s store, but forgot to buy his papers. Western Star (CA 16-11-2014) Thus, rather than lexical, Adj+ie nominalisations demonstrate what could be considered a special kind of productivity on the level of semantics, as these derivatives are prompted by a mix/transference of semantic data coming from the omitted noun, adjectival base, and the meaning(s) associated with the suffix -ie, which is not necessarily obvious in denominal -ie units. 4.3 Scalarity of Meanings Adj+ie nominalisations under study are units derived from basic ‘origin’ and ‘colour’ adjectives (e.g., darkie, whitey, Chinky, Aussie) and belong to a particular

196

Sánchez Fajardo and Tarasova

semantic group, as they are traditionally associated with race and origin slurs. Slurs, being “[…] expressions that can be used to demean and dehumanise targets based on their membership in social groups based on, for example, race, ethnicity, religion, gender, or sexual orientation” (Ritchie, 2017, pp. 155–156), are characterised by pejoration and marginalisation of the referent. This contributes to the perception of slurs as being derogatory, and is connected with societal and cultural norms of language use. Discrimination of people on the basis of their skin colour and/or nationality is not acceptable in contemporary society, hence, most of the units in this category are listed in dictionaries as derogatory/rude. At the same time, a number of studies on this matter (e.g., Potts, 2007; Ritchie, 2017; Whiting, 2013) demonstrate that pejorative/offensive words referring to members of a certain community, for example, LGBT community, when used by these community members in communication with each other (e.g., faggot), undergo the process of appropriation and are reclaimed by the members of this community as solidarity terms and/or familiarity markers. In our analysis of origin- and colour-based Adj+ie nominalisations, we take the position that pejoration noted in the semantic compositions of these constructs is a relativised category, since pejoration is not exactly contributed by the suffixes “in a uniform way” (Finkbeiner, Meibauer, & Wiese, 2016, p. 3). Based on that, derivatives formed by means of the same input categories of adjectival bases and the same derivational material may have the opposite connotations. For example, blackie in reference to a dark-skinned person is reported to be offensive (OED3), while brownie in reference to a female of Hispanic or Asian descent bears very positive connotations (Urban Dictionary).1 Thus, we propose that an endearment-pejoration scale be used for the analysis of diminutive nominalisations of this kind. The necessity of applying this scale is linked to two global concepts: (i) morphosemantic paradigms of Adj+ie nominalisations, and (ii) communicative intentionality of the speaker towards the referent. Morphosemantic paradigms of deadjectival nominalisations lie at the core of the present study, and their analysis is aimed at accounting for morphological and semantic changes and general trends noted in the paradigms that the analysed derivatives form. Consideration of communicative intentionality of the speaker towards the referent allows us to account for specific communicative situations and situational contexts that are reflected in the semantics of these forms (both in terms of their denotational and connotational/associative meanings). Thus, a 1  It is a common understanding that Urban Dictionary cannot be considered a reliable academic source. We share this opinion and all the information that we use from this source is treated with caution.

Of Brownie Girls and Aussie Families

197

black person may not always be offended by being called blackie because the pejorative value of the term depends on the context in which communication takes place. Stated differently, pragmatic factors that are involved in a kind of metaphorical relation between the semantic activation of these nominalisations and the referents need to be taken into account (cf. Dressler, 1994). Finally, one of the most relevant features of pejorative suffixation relates to the concepts of iconicity/sound symbolism and phonaesthesia, i.e., “a similarity of form […] is correlated with a similarity of meaning” (Perniss, Thompson, & Vigliocco, 2010, p. 2). The influence of phonaesthesia and phonotactics on the process of lexicalisation is not new: some prior studies have established a correlation between phonological structures and meaning in etymologically unrelated words (cf. Bolinger, 1950, 1965; Markel & Hamp, 1960). This has also been corroborated by Crystal (2003) in his description of the so-called ‘submorpheme’. Nevertheless, not much has been produced on the acquisition, and restriction of meaning in morphemes that are often associated with slang. Even though this issue is not sufficiently discussed, there are indications showing that morphological systematicity and iconicity have a deep impact on vocabularies from cognitive and lexical perspectives (cf. Gasser, Sethuraman, & Hockema, 2005; Dingemanse, Blasi, Lupyan, Christiansen, & Monaghan, 2015). Based on these premises, it is possible to predict that the suffix -ie in English is also expected to comply with systematic paradigms and, therefore, an iconic structuring of meaning. However, it is important to point out that for a long time close front vowels have been associated with the idea of smallness,2 and since the primary meaning and function of the suffix is that of diminution, we should expect that size will be accounted for in one way or another. At the same time, it is also noted (e.g., Jurafsky, 1996; Schneider, 2003) that one and the same diminutive formation may express either appreciation or depreciation depending on the context a given form is used in. We thus believe that iconicity of the phonological form of the suffix does not have to be limited only to the idea of smallness in terms of the size, but also in terms of the value of the concept in the eyes of the interlocutors. Depreciatory use of one and the same form brings with it marginalisation of its referent, which is strongly connected to the idea of pejoration.

2  It is important to note though that the idea of ‘smallness’ has been also put into question by Bauer (1996), Körtvélyessy & Štekauer (2011), and Körtvélyessy (2015).

198 5

Sánchez Fajardo and Tarasova

Methodology

As stated above, the purpose of this research is to analyse the semantic networks, as well as word-building paradigms or patterns of formation of -ie nominalisations derived from adjectives of origin and colour. In our research, we use a qualitative analysis of the data, since we are more interested in the availability of the Adj+ie pattern and the aspects of meaning that nominalisations of this kind realise. A total of 214 examples of nominalisations ending in the -ie suffix were collected from a range of printed prescriptive and descriptive dictionaries of English (MWD11; OED3; Ayto, 1998; Dalzell & Victor, 2006; Thorne, 2014), as well as the reference sources available online (OED3; MWD11; CED4). The collected set contained 54 examples of Adj+ie nominalisations (as opposed to denominal and deverbal formations), which were chosen for further analysis. The input categories of the adjectives and the mechanisms involved into the formation of such nominalisations were initially analysed in the descriptive framework. The groups of nominalisations derived from adjectives denoting origin and colour were chosen for further investigation at this stage of our research; however, we hope to expand the range of semantic categories of the input adjectives in the future. We are particularly interested in this group of Adj+ie nominalisations since they allow for viewing the extension of meaning through the endearment-pejoration scale, which can be further applied to other kinds of -ie nominalisations. The second stage was the analysis of meanings of the selected origin- and colour-based Adj+ie nominalisations, which involved the analysis of meanings registered in the printed and online dictionaries. As Adj+ie nominalisations are characterised by a wide range of meanings due to the variety of contexts they are used in, the analysis of contextual meanings was necessary for this investigation. The latter was done through extraction of the contexts that these units occur in from NOW. The analysis of the meanings of Adj+ie nominalisations through their contexts allowed for seeing a wider range of senses that these units can realise as compared to the number of the meanings registered in the consulted reference sources. The analysis of the contexts also allowed us to confirm how communicative intentionality of the speaker and the referent contribute to the extension of associative meanings, which in time may lead to the extension of the denotational meanings. The next stage of our analysis employed an onomasiological approach to word formation (Štekauer, 2005), the central idea of which is the unity between the form of the naming unit and its meaning. It is of interest to view the analysed units through their existence in paradigmatic relations in the vocabulary.

Of Brownie Girls and Aussie Families

199

As we are focused on morphosemantic paradigms, we used this approach to identify the paradigmaticity that underlies the semantic connection between the base (black- as in blackie) and the omitted noun (person). This part of the data-processing stage intended to examine the senses by applying a componential analysis of the lexical units followed by a grouping of the stylistic traits conveyed by the lemmas (e.g., pejorative, informal, diminutive), and a semantic categorisation of the omitted noun (person, thing, animal, etc.). The last stage involved interpretation of the data and establishing plausible motivations that underlie the formation of such nominalisations through the analysis of identified trends noted in their morphology and semantics. The necessity to identify the underlying motivations is dictated by the need to represent the observable trends in the word-formation paradigms of Adj+ie nominalisations, which is important for a better understanding of semantic motivations within the paradigm, as well as for seeing how the meanings associated with such formations are transferred onto the -ie suffix. The interpretation of the results is centred around the following key issues: the degree of polysemy (or the possibility of homonymy) on the Adj+ie nominalisation; the transfer of meanings of the input components (the adjectival base, the omitted noun, and the suffix) onto the nominalisation; and the iconicity of the suffix -ie in terms of pejoration. In our analysis, we apply the endearment-pejoration scale, which is believed to be helpful for understanding the circumstances that influence activation of different meanings of the analysed items. The analysis and the discussion of the results that follow are aimed to demonstrate how pragmatic factors influence the development of paradigmatic relations within origin- and colour-based Adj+ie nominalisations. 6

Findings and Discussion

6.1 Categories of Deadjectival Nominalisations and Bases A necessary preamble to a more in-depth analysis of the data is that the number of Adj+ie nominalisations in English, as opposed to Noun+ie/y units, is relatively low. Out of the 214 -ie nominalisations extracted from the OED3, only 54 entries are derived from adjectives (see Annex). This number does not include the units with unclear word-formation patterns like sticky, trendy, woody, wrinkly, etc., which we exclude from the analysis at this stage. Although these units end in /ɪ/, it is not clear whether that is the diminutive -ie or the adjectival -y, where the adjective then becomes a new noun through conversion: a trendy person > a trendy. The omission of the noun (‘person’) does not necessarily imply derivation of the adjectival base, but rather it might be induced by the

200

Sánchez Fajardo and Tarasova

Adj+ie paradigm whereby an authentic derivation mechanism is activated simultaneously to the ellipsis of the noun. In terms of their morphological structure, these derivatives are mostly disyllabic nouns; however, Bauer, Lieber, and Plag (2015, p. 392) point out that there may be occasional exceptions to the general patterns, whereby the base may retain more than one syllable when derivatives of this kind anchor to a non-initial syllable of the base, e.g., littlie, second handy. However, such cases are exceptional rather than regular; therefore, we take the position that adjectival bases are necessarily monosyllabic, which explains why some polysyllabic bases are truncated to comply with the ‘disyllabic trend’ for the output unit. Of the 54 entries of this type of nominalisations in OED3, 10 have derived from truncated or clipped bases: veggie (< vegetarian), preppy (< preparatory), indie (< independent), fundie (< fundamentalist), conchie (< conscientious), Aussie (< Australian).3 Bauer, Lieber, and Plag (2015, p. 392) also talk about possible phonological changes to the root, e.g., conchie (< conscientious), Ozzie (< Australian), Gyppie (< Egyptian). Even though we are not looking at such changes here, research into the factors promoting them may be helpful for a better understanding of cognitive processes involved into the formation of these units. In terms of their phonology, -ie nominalisations in general are characterised by the primary stress on the first syllable. This can result in alterations in the quality of the vowel, which changes from reduced/schwa to its full form, as can be seen in the example of suspicious > sussy (Bauer, Lieber, & Plag, 2015, p. 392). This, in fact, fits with a wider pattern of the initial stress in adjectives which do not have a full vowel when the stress falls on the second syllable, e.g., the simple vs. remote. Another interesting feature that distinguishes Adj+ie nominalisations from denominal and deverbal ones (baccy < tobacco, walkie-talkie) is their tendency for polysemy. As Adj+ie constructs are based on three semantic and lexical components, i.e., the base, the omitted or ellipted noun, and the resulting nominalisation (output), e.g., smartie = smart (adj.) + -ie (suffix) + (person/ device/object), it is possible to suggest that this tendency can be explained by the transfer (or inheritance) of the semantics of the input forms (even though they might not be present in the resulting structure) onto the derivatives. We attribute the tendency to polysemy to the semantic content of the ellipted noun rather than the adjectival base. As argued by Tarasova (2013), the main difference between adjectives and nouns lies in their conceptual structure. 3  The specifications of the etymologies and the word class of the base have been taken from OED3 and MWD11.

Of Brownie Girls and Aussie Families

201

This difference is connected to the fact that adjectives are relational entities (Langacker, 1987), and information about relations that they can potentially realise is included in their meaning. Therefore, when encountering an adjective, we relate it to something it can potentially modify (most commonly nouns). Notions behind nouns, however, can be conceived of as such and are more packed with information than adjectives (Paradis, 1999, 2000). The information contained in the conceptual structure of nouns does not only include the knowledge about the concept that a noun names, but also (a) information on how a given concept fits into the hierarchy of conceptual structure of the world as reflected in the language, and (b) how this concept can be related to other concepts. Extending this to Adj+ie nominalisations, we can speculate that the informational load in these structures is comprised of the meaning of the adjectival base as well as the meanings transferred from the ellipted nouns. It is also possible to suggest that ellipted nouns may be responsible for the tendency to polysemy in these constructs. The omission of the accompanying noun is also expected to trigger a nounforming process, in which two mechanisms are involved: (a) the adjectival base undergoes a conversion or zero derivation process whereby an ‘embellishing’ suffix (Bauer & Huddleston, 2002, p. 1636) is added; or (b) the suffix -ie is embedded with a functional noun-forming value. The former complies with the formation of hypocoristics, the process in which the omitted nouns are considered part of a noun phrase, similarly to the formation of zoo (< zoological garden). In examples like this, there is a reduction of original material that does not affect the denotation of the clipped words. The clipped word (zoo), according to (a), might result from a preliminary conversion (adjective → noun) of bases and a clipping (or not) of the remaining adjectival base. The latter mechanism, i.e., (b) concerns the paradigmatic value of the suffix in Adj+ie nominalisations, in which the noun-forming morpheme is added to adjectival bases in a similar way to other nominalising suffixes, e.g., forgetful-ness or cruel-ty. Even though denominal derivatives, e.g., baccy, cheekie, do not demonstrate a change of the word class, it is still clear that -ie has ‘noun’ as part of its meaning, and it does not matter whether the base is characterised by the presence of the same seme or not. Adj+ie nominalisations are also connected with other word-formation processes such as compounding or affixation in relation to the degree of denotational change. That is, the denotation conveyed by a base differs from that of its resulting units. Alternatively, non-suffixed clippings (i.e., perf < perfect, lab < laboratory) do not demonstrate changes in their denotation and the semantic core of the residue coincides with that of the source.

202

Sánchez Fajardo and Tarasova

It is also noted that the meanings realised by the suffix -ie are informed by the semantics of the paradigms of adjectival bases (occasionally clipped), i.e., origin and colour in our case, as well as the semantic content of ellipted nouns. Consistent re-occurrence of the suffix in nominalisations derived from semantically homogeneous groups of adjectives (and even more so—ellipted nouns) in reference to similar concepts may contribute to the iconicity of the suffix in terms of its strong association with meanings of appreciation and, especially, depreciation. The latter seems to be a growing trend and may be connected with the analogical correspondence between this word-formation process and the social function and meaning of embellished suffixes of this kind (Peters & Burridge, 2012, p. 253). Cognitively, iconicity has been related to the correspondence between a sign and its referent, which is reflected somehow in the social and linguistic acceptance of salient features that characterise this referent. In other words, here we mean that it is more iconic to have word segments to which meaning elements can be assigned rather than assigning the meaning to the word as a unit. Hence, morphological units and word-formation processes are also likely to become iconic if their signatum or meaning is also paradigmatic. This goes along with the extrapolation of paradigmatic constructions (Fernández-Domínguez, 2019, p. 212) and the onomasiological approach of derivational regularities. 6.2 Paradigms in Colour/Origin-Based Adj+ie Nominalisations This section aims to show the morphological and semantic paradigms that origin- and colour-based Adj+ie nominalisations are characterised by. The group of colour-based nominalisations is not as frequent as expected (see Table 8.1), but their complexity lies in the convergence of three units: the colour/origin base (e.g., black-), the ellipted noun (e.g., ‘person’), and the nounforming suffix (i.e., -ie). On the level of structural representation, the process of forming units like blackie appears to be purely derivational: black- + -ie = blackie. However, the process of lexical ellipsis is also reinforced by a complex of cognitive operations (which in this case could be rendered by the pejorative meaning of -ie) and the semantic composition of the base and the omitted noun. As mentioned earlier, it is possible that the formation of this type of nominalisations is informed by the iconicity of the suffix. Since most of the derived nouns extracted from OED3 are clearly colloquial (slang) or dialectal, we can suggest that there exists a correlation between the suffix -ie and low-prestige word stock. If Bauer and Huddleston’s (2002) idea on the ‘embellishment’ nature of this suffix were applicable, doublets like (a) and (b) would be full

Of Brownie Girls and Aussie Families

203

Table 8.1 Colour/origin-based Adj+ie nominalisations used in the study

Colour-based

Origin-based

blackie brownie bluey greenie pinkie reddie whitey orangie fairie darkie blondie

Aussie (also Ozzie) Chinky (also Chinkie) Chonky (also Chonkie) Dutchie Eyetie Frenchie Heebie Jewbie Hunky Russki Paki

synonyms, which they are not due to a set of associations connected with -ie units as opposed to unsuffixed forms. Stated differently, even though green and greenie might be viewed as the result of conversion, the attachment of -ie is expected to add evaluative content towards the same referent. (a) green ‘environmentalist’ (n.) vs. greenie ‘environmentalist’ (n.) (b) white ‘white person’ (n.) vs. whitey ‘white person’ (n.) Thus, unsuffixed deadjectival nominalisations are less iconic than the suffixed ones, i.e., black (n.) and blackie (n.) but the latter is more likely to convey evaluative/attitudinal meanings than the former. Clearly, the suffix might add to the base ambivalent meanings, which are best represented by applying endearment-pejoration scale (e.g., blackie ‘black bird’ is neutral, while blackie ‘black person’ is pejorative). What appears to be a mere extrapolation of meaning via a suffixed base could also be analysed as a case of iconicity gain, in which the morpho-phonological segment represented by the suffix -ie acquires semantic associations. In the discussed case of blackie when it is used in reference to a person, the iconicity of the suffix lies in the denotational sense of -ie as a noun-forming morpheme, as well as the connotational traits as a pejorative or derogatory source. The endearment-pejoration scale constitutes one of the most significant features of the suffix -ie. The endearment side of the scale is best seen in hypocoristics, in which the endings -ie/-i/-y/-ey are conventionally added to clippings of personal names to express familiarity (Booij, 2005; Sánchez Fajardo

204

Sánchez Fajardo and Tarasova

& Rodríguez González, 2018): Andy < Andrew, Susie < Susan.4 As indicated by Schneider (2003) and Dixon (2004), the appreciative and depreciative value of the suffix -ie is related to its natural representation of ‘smallness’, which explains why it is typically referred to as a diminutive. The missing link between ‘smallness’ and ‘pejoration’ is found in the sense of ‘unimportance’, and the suffix -ie is then embedded with slangy, or even derogatory, value. This cognitive process is even more complex when such nominalisations show high levels of polysemy, entailing a direct correspondence between the omitted noun or referent and the type of semantic value activated. For instance, our data demonstrate that there is a connection between the activation of pejorative meaning and the sense of [human]. This iconic value of the suffix -ie can be seen in the connotational correlations between nominalisations derived from basic colour adjectives (whitey, blackie, greenie/greeny, etc.) and non-basic colour terms ( fairie, darkie, blondie). This suggests that the semantic paradigm is observable in these two types of bases, regardless of the denotational meaning. As seen in Table 8.2, the only morphological trait that restricts this analogy is multisyllabism, as all the colour/origin bases are monosyllabic, with the exception of orang-.5 The uneven levels of homonymy and polysemy are also typical of these nominalisations. Table 8.2 shows the colour-based lemmas under study and their senses, in which some units such as brownie and blackie are clearly far more polysemous than others (e.g., fairie, whitey). It can be observed that there is at least one derogatory meaning under each lemma, and this pejorative sense refers to someone’s physical appearance (blackie1, brownie5, bluey2, greenie1, pinkie1, pinkie3, whitey1, darkie1, darkie2, darkie3, reddie2), political beliefs (pinkie11, greenie1), or origin (orangie1). The fact that ‘someone’ or [human] is connected to this derogatory meaning reveals the existence of paradigmaticity in the coinage of nominalisations of this kind. A preliminary analysis of the data also corroborates that there exists a unidirectional transposition of qualities from the ellipted noun to the base. This transfer is determined by a metonymic association between the colour and its corresponding referent. Cognitively speaking, language users are able to identify the new units through the trait of ‘colour’ on condition that supporting context is provided. In the sentence ‘I saw some blackies’, the word blackie 4  This obviously is not the case for personal names that end in -ie/y, e.g., Mary, Laurie, which are not hypocoristic. 5  As can be seen from Table 8.2, some gaps are obvious, for example there is no greyie and yellowie nor magentaie, turqoisey, indigory, etc. It is not clear whether this absence is determined by their phonological structure, their etymology, or the fact that they lack a referent.

205

Of Brownie Girls and Aussie Families

Table 8.2 Colour-based Adj+ie nominalisations, and their corresponding senses, extracted from normative and descriptive dictionaries (OED3; MWD11; TAHD5; Ayto, 1998; Dalzell & Victor, 2006; Thorne, 2014) and corpora (COCA; NOW; BNC)

Colour-based Senses Adj+ie

Colour-based Senses Adj+ie

blackie1 blackie2 blackie3 brownie1 brownie2 brownie3 brownie4 brownie5

‘a black person’ ‘a type of mango’ ‘a black-feathered bird’ ‘an elf’ ‘a girl scout’ ‘a chocolate cake’ ‘bread with currant’ ‘one that is of mixed black and white ancestry’ ‘one that is attractive’ ‘a drink of whiskey’

pinkie4 pinkie5 pinkie6 pinkie7 pinkie8 pinkie9 pinkie10 pinkie11

‘a bruised eye’ ‘the vagina’ (T&T) ‘a warning ticket’ (Ca.) ‘cheap wine’ (Ca.) ‘a model of Land Rover’ (UK) ‘a fifty-pound note’ ‘a type of goose’ ‘one who is a left-wing supporter’

pinkie12 orangie1 reddie2

reddie3

‘a red-haired person’

whitey1 whitey2

‘a white person’ ‘marihuana’

greenie2 greenie3 greenie4 pinkie1 pinkie2

‘a bundle of possessions carried by a bushman’ ‘a nickname for a red-headed person’ ‘a five-pound note’ ‘one that is environmentalist’ ‘paper money’ ‘a one-pound note’ ‘thick nasal mucus’ ‘one that is white’ ‘little finger’a

‘a type of marihuana’ ‘one that is from the Netherlands’ ‘a fifty-pound note’

fairie1 darkie2 darkie2 darkie3 darkie4

pinkie3

‘an albino’

darkie5

‘a white person’ ‘a black person’ ‘an aboriginal’ (Aus.) ‘a Polynesian person’ (NZ) ‘term of address for attractive, dark-skinned woman’ (T&T) ‘excrement’ (Aus.)

brownie6 brownie7 bluey1

bluey2 bluey3 greenie1

a Though listed here, it has not been used in the study as it might come “from Dutch pinkje small pink, diminutive of pink, from Middle Dutch pinke” (MWD11).

206

Sánchez Fajardo and Tarasova

could refer to people or birds, entailing not only the semantic ambiguity of these elements, but also the pragmatic use of the word in the utterance: ‘offensive’, ‘derogatory’, or just ‘distinct’. Thus, colour bases are particularly sensitive to semantic ambiguity and pejoration, especially when [race], [political belief], or [origin] are present. Without a doubt, these domains have been traditionally associated with language inappropriateness and/or insult, but these cases also comply with other widely-known derogatory words: fattie, oldie, lezzie, mutie, to name just a few. Establishing the endearment-pejoration scale is a difficult task as not all these words might have the same impact on language users, and a number of pragmatic and sociolinguistic factors may be involved. In a nutshell, regardless of the adjectival base, nominalisations of this kind are also embedded with ‘slang-sounding’ features, whose degrees of pejoration (or just vulgarism) depend on the sociolinguistic connotations of the base and the ellipted noun, with [race] and [human] being expected to convey pejoration and derogation. This can also be seen in cases where nationality-based nominalisations are used as collective terms for different ethnic minority groups (Hunkies for Eastern Europeans, Chinky for Asians, and Paki for South Asians). However, not all origin -ie words express pejoration, and examples like Frenchie and Dutchie seem quite neutral/positive, while Aussie and Russkie can be offensive in some contexts but not others. This demonstrates the necessity of identifying the factors that influence the perception of such nominalisations by the members of such groups as well as non-members. This also supports the necessity of representing the senses that the -ie suffix can realise on the scale. A significant feature of origin-based nominalisations, as compared to colour-based ones, is their level of explicitness. All these constructs, e.g., Heebie, Frenchie, possess the trait [human], which results in a less ambiguous understanding by language users. Yet, the colour-based nominalisations are less explicit owing to relatively varied semantic forms, which may include [complexion], [race], [clothes], etc. At the same time, the explicitness of origin-based nominalisations in reference to people may be responsible for the lower level of polysemy in this group. In general, the vast majority of the senses conveyed by origin- and colourbased Adj+ie nominalisations reflect the following traits or semes: [human], [negative], [concrete], and/or [race/skin/complexion]. The distinction between [male] and [female] is irrelevant: the cases of brownie ‘elf’ (masculine) and brownie ‘scout’ (feminine) are generally induced by the features of the referent, not necessarily by the suffix. The combination of these semantic elements reflects the iconic salience of the suffix -ie in the formation of

Of Brownie Girls and Aussie Families

207

deadjectival nominalisations, and reinforces the establishment of wordformation paradigms within this group. 7

Conclusions

Though more infrequent than denominal and deverbal -ie formations, Adj+ie nominalisations, in particular colour- and origin-based ones, demonstrate semantic complexity, which we attribute to the set of features associated with the units involved in their formation, i.e., the adjectival base, the ellipted (or omitted) noun, and the suffix -ie. Morphologically, these derivatives tend to be disyllabic, which explains why some polysyllabic bases, such as Australian (> Aussie), undergo a clipping process. A distinctive feature of colour-based Adj+ie nominalisations is their high level of polysemy, which is induced by the relational structure of adjectives and the semantic content of ellipted nouns. This also has an impact on the denotational value of the resulting nominalisations, which are expected to vary from their original bases. Because the ellipted noun is always the same in origin-based -ie units, the trait [human] is more explicit in this group, and is viewed as part of the denotational meaning. Owing to the strong association with pejorative meaning that Adj+ie nominalisations may demonstrate, the suffix -ie is believed to be reinforced with gradable levels of iconicity. The examination of the senses that Adj+ie nominalisations convey suggests that the iconic value of the suffix can be accounted for by means of the application of the pejoration-endearment scale within the category of ‘diminution’. Colour- and origin-based -ie nominalisations do not convey the literal sense of ‘diminution’ in terms of the size, but they comply with the other two ensuing categories. The expression of ‘pejoration’ and ‘endearment’ seem to be more predictable in the case of colour-based units, especially those in which their semantic structures contain the primary traits of [human], [race]/[clothes], [positive]/[negative]. Colour-based nominalisations are, in general, more sensitive to polysemy and ambiguity, but they are less context-dependent in terms of conveying connotational meanings when they are used in reference to people, as opposed to origin-based units. Iconicity of the suffix is understood as its strong link with the senses that are not present as part of the denotational meanings of the output lexemes but are instead associated with the concepts they denote. This may also be the reason for the limited (or restricted) formation of these units in conventional discourse, and can also be seen as the result of a complex of

208

Sánchez Fajardo and Tarasova

semantic variation: semes or semantic traits activated → pejorative iconicity → semantic regularity → restricted word-formation mechanism in the conventional language. This corroborates the role of semantic variation in systemic paradigmaticity, and how evaluative meaning is dependent on both suffixcentred cognitive operations and semantic compositionality of derivatives. References Anderson, S. R. (1982). Where’s morphology? Linguistic Inquiry, 13, 571–612. Ayto, J. (1998). Oxford dictionary of slang. Oxford: Oxford University Press. Bagasheva (this volume). Paradigmaticity in compounding. In J. FernándezDomínguez, A. Bagasheva, & C. Lara-Clares (Eds.), Paradigmatic relations in word formation (pp. 21–48). Leiden: Koninklijke Brill NV. Bardsley, D., & Simpson, J. (2009). Hypocoristics in New Zealand and Australian English. In P. Peters, P. Collins, & A. Smith (Eds.), Comparative studies in Australian and New Zealand English: Grammar and beyond (pp. 49–69). Amsterdam & Philadelphia: John Benjamins. Bauer, L. (1996). Evaluative morphology: In search of universals. Studies in language, 21, 533–575. Bauer, L. (2001). Morphological productivity. Cambridge: Cambridge University Press. Bauer, L. (2008). Derivational morphology. Language and linguistics compass, 2, 196–210. Bauer, L., & Huddleston, R. (2002). Lexical word-formation. In R. Huddleston & G. K. Pullum (Eds.), The Cambridge grammar of the English language (pp. 1621–1721). Cambridge: Cambridge University Press. Bauer, L., Lieber, R., & Plag, I. (2015). The Oxford reference guide to English morphology. Oxford: Oxford University Press. BNC: Davies, M. (2004). British National Corpus. http://corpus.byu.edu/bnc/ [Accessed 27 August, 2018]. Bolinger, D. L. (1950). Rime, assonance, and morpheme analysis. Word, 6, 117–136. Bolinger, D. L. (1965). Forms of English: Accent, morpheme, order. Cambridge, MA: Harvard University Press. Booij, G., (2005). The grammar of words: An introduction to linguistic morphology. Oxford: Oxford University Press. Bybee, J. (1985). Morphology. A study of the relation between meaning and form. Amsterdam: John Benjamins. Cannon, G. (1987). Historical change and English word-formation. New York: Peter Lang. CED4: Cambridge English Dictionary Online, 4th edition, Cambridge University Press. https://dictionary.cambridge.org/es/ [Accessed 26 January, 2018].

Of Brownie Girls and Aussie Families

209

Charleston, B. M. (1960). Studies on the emotional and affective means of expression in Modern English. Bern: Francke. COCA: Davies, M. (2008). The corpus of Contemporary American English: 400+ million words, 1990–present. https://corpus.byu.edu/coca/. [Accessed 24 August, 2018]. Crystal, D. (2003). A Dictionary of linguistics and phonetics. Oxford: Blackwell. Dalzell, T., & Victor, T. (2006). The concise new Partridge dictionary of slang and unconventional English. London, New York: Routledge. Dingemanse, M., Blasi, D. E., Lupyan, G., Christiansen, M. H., & Monaghan, P. (2015). Arbritariness, iconicity, and systematicity in language. Trends in cognitive sciences, 19, 603–615. Dixon, R. M. W. (2004). Adjective classes in typological perspective. In R. M. W. Dixon & A. Y. Aikhenvald (Eds.), Adjective classes: A cross-linguistic typology (pp. 1–49). Oxford: Oxford University Press. Dressler, W. (1994). Morphopragmatics. Diminutives and intensifiers in Italian, German, and other languages. Berlin, New York: Mouton de Gruyter. Dressler, W. U., & Merlini Barbaresi, L. (2001). Morphopragmatics of diminutives and augmentatives: On the priority of pragmatics over semantics. In I. Kenesei & R. M. Harnish (Eds.), Perspectives on semantics, pragmatics and discourse: A Festschrift for Ferenc Kiefer (pp. 43–58). Amsterdam, Philadelphia: John Benjamins. Fernández-Domínguez, J. (2015). The syntagmatic and paradigmatic axes of productivity: an application to the onomasiological model of word-formation. SKASE Journal of Theoretical Linguistics, 12, 218–237. Fernández-Domínguez, J. (2019). Compounds and multi-word expressions in Spanish. In B. Schlücker (Ed.), Complex lexical units: Compounds and multi-word expressions (pp. 189–220). Berlin, Boston: De Gruyter Mouton. Finkbeiner, R., Meibauer, J., & Wiese, H. (Eds.). (2016). Pejoration. Amsterdam, Philadelphia: John Benjamins Publishing. Fischer, R. (1998). Lexical change in Present-Day English. Tübingen: Gunter Narr Verlag. Fox, A. (1995). Linguistic reconstruction: An introduction to theory and method. Oxford: Oxford University Press. Gasser, M., Sethuraman, N., & Hockema, S. (2005). Iconicity in expressives: An empirical investigation. In S. Rice & J. Newman (Eds.), Experimental and empirical methods in the study of conceptual structure, discourse, and language (pp. 163–180). Stanford CA: CSLI Publications. Grandi, N. (2017). Evaluatives in morphology. In M. Aronoff (Ed.), Oxford research encyclopaedia of linguistics. Oxford: Oxford University Press. DOI: 10.1093/acrefore/ 9780199384655.013.250 [Accessed 20 September, 2018]. Grandi, N., & Körtvélyessy, L. (2015). Introduction: Why evaluative morphology? In N. Grandi & L. Körtvélyessy (Eds.), The Edinburgh handbook of evaluative morphology (pp. 4–17). Edinburgh: Edinburgh University Press.

210

Sánchez Fajardo and Tarasova

Jespersen, O. (1933). The selected writings of Otto Jespersen. London: Allen & Unwin LTD. Jurafsky, D. (1996). Universal tendencies in the semantics of the diminutive. Language, 72, 533–578. Körtvélyessy, L. (2015). Evaluative morphology from a cross-linguistic perspective. In N. Grandi & L. Körtvélyessy (Eds.), The Edinburgh handbook of evaluative morphology (pp. 187–194). Edinburgh: Edinburgh University Press. Körtvélyessy, L., & Štekauer, P. (2011). Introduction. Lexis [Online], 6. http://lexis .revues.org/399 [Accessed 20 April, 2018]. Langacker, R. W. (1987). Foundations of Cognitive Grammar, Vol. 2: Descriptive Application. Stanford, CA: Stanford University Press. Leisi, E. (1969). Das heutige Englisch. Wesenszüge und Probleme. 5th ed. Heidelberg: Winter. Lipka, L. (2002). English lexicology: Lexical structure, word semantics & word formation. Tübingen: Gunter Narr Verlag. Marchand, H. (1969). The categories and types of Present-day English word-formation. A synchronic-diachronic approach. München: C. H. Beck’sche Verlagsbuchhandlung. Markel, N. N., & Hamp, E. P. (1960). Connotative meanings of certain phoneme sequences. Studies in Linguistics, 15, 47–61. MWD11: Merriam-Webster dictionary online, 11th edition, Merriam-Webster, Inc. https:// www.merriam-webster.com [Accessed 10 December, 2017]. NOW: Davies, M. (2013). Corpus of News on the Web: 3+ billion words from 20 countries, updated every day. http://corpus.byu.edu/now/ [Accessed 30 August, 2018]. OED3: Oxford English Dictionary, 3rd edition, Oxford University Press. https://oed.com [Accessed 14 December, 2017]. Paradis, C. (1999). Adjectives and boundedness. Cognitive Linguistics, 12, 47–65. Paradis, C. (2000). Reinforcing adjectives: A cognitive semantic perspective on grammaticalization. In R. Bermudez-Otero, D. Denison, R. M. Hogg, & C. B. McCully (Eds.), Generative theory and corpus studies (pp. 197–220). New York: Mouton de Gruyter. Perniss, P., Thompson, R. L., & Vigliocco, G. (2010). Iconicity as a general property of language: Evidence from spoken and signed languages. Frontiers in psychology, 1, 1–15. DOI: 10.3389%2Ffpsyg.2010.00227 [Accessed 26 September, 2018]. Peters, P., & Burridge, K. (2012). English in Australia and New Zealand. In R. Hickey (Ed.), Areal features of the Anglophone world (pp. 233–258). Berlin: de Gruyter. Pinker, S. (1984). Language learnability and language learning. Cambridge, MA: Harvard University Press. Plag, I. (1999). Morphological productivity: Structural constraints in English derivation. Berlin: Mouton de Gruyter. Potts, C. (2007). The centrality of expressive indices. Theoretical Linguistics, 33, 255–268.

Of Brownie Girls and Aussie Families

211

Quirk, R., Greenbaum, S., Leech, G., & Svartvik, J. (1985). A comprehensive grammar of the English language. London: Longman. Ritchie, K. (2017). Social identity, indexicality, and the appropriation of slurs. Croatian Journal of Philosophy, 17, 155–180. Sánchez Fajardo, J. A., & Rodríguez González, F. (2018). Motivations and morphological variations in clipped personal names: A cross-linguistic approach. Lingua, 206, 35–48. Scalise, S. (1984). Generative morphology. Dordrecht: Foris. Schneider, K. P. (2003). Diminutives in English. Tübingen: Max Niemeyer Verlag. Shields, K. Jr. (2001). On the origin of English diminutive suffix -y, -ie. Studia Anglica Posnaniensia: International Review of English Studies, 37 (Annual), 141–144. Sinclair, J. Mc. H. (1991). Corpus, Concordance, Collocation. Oxford: Oxford University Press. Štekauer, P. (2005). Onomasiological approach to word-formation. In P. Štekauer & R. Lieber (Eds.), Handbook of word-formation (pp. 207–232). Dordrecht: Springer. Strang, B. (1968). Modern English structure. 2nd rev. ed. London: Arnold. Stump, G. (1993). How peculiar is evaluative morphology? Journal of Linguistics, 29, 1–36. TAHD5: The American Heritage Dictionary Online, 5th edition, Houghton Mifflin Company. https://ahdictionary.com [Accessed 27 November, 2017]. Tarasova, E. (2013). Some new insights into the semantics of English N+N compounds. (Unpublished doctoral dissertation). Victoria University of Wellington. http://hdl .handle.net/10063/2768 [Accessed 19 September, 2018]. Thorne, T. (2014). Dictionary of contemporary slang. Bloomsbury Publishing. Urban Dictionary. (n.d.g.). https://www.urbandictionary.com/ [Accessed 19 September, 2018]. Whiting, D. (2013). It’s not what you said, it’s the way you said it: Slurs and conventional implicatures. Analytic Philosophy, 54, 364–377. Wierzbicka, A. (1985). Different cultures, different languages, different speech acts. Journal of Pragmatics, 9, 145–178.



Annex

List of -ie/y/ey nominalisations extracted from OED3 argy-bargy, baccy, baggy, bevvy, bicky, birdie1, birdie2, bookie, botty, brassy, bridie, brickie, buttie, cabbie, choccy, ciggy, civvy, civvies, cocky, conchie, corby, cossie, crannie, creepy-crawly, crikey, crony, cuddy1, cuddy2, cutie, daddy, darkie, darbies, dicky1, dicky2,

212

Sánchez Fajardo and Tarasova

dicky3, divvy, dobby, ducky, earlies1, earlies2, facies1, facies2, filly1, filly2, flattie1, flattie2, flattie3, floozie, flunkey, fogey, folkie, foodie, footie1, footie2, footsie, freebie, Frenchie, fundie, Gypsy, girlie, golden oldie1, golden oldie2, goodie1, goodie2, goody-goody, greenie, groupie, heinie, hippy, homie, hoody, hottie1, hottie2, housie-housie, hughie, hussy, indie1, indie2, jacksie, jerky, junkie1, junkie2, laddie, leccy, leftie1, leftie2, lippy, lolly1, lolly2, lolly3, lovey, meanie, mommy, mumsy, nappy, navvy, neddy1, neddy2, newie, newbie, newsy1, newsy2, nightie, noddy1, noddy2, nooky, noogie, noughties, numpty, offie, Okie1, Okie2, oldie, outie, Ozzie1, Ozzie2, Paddy, pappy, patsy, Pilly, pikey, pinkie, Pommy, popsy, preppy, prezzie, pussy1, pussy2, queenie, quickie1, quickie2, roughie1, roughie2, roughie3, sarnie, schoolie1, schoolie2, Scottie1, Scottie2, scarpie, script kiddie, scrumpy, selkie, sharpie1, sharpie2, sharpie3, sheltie, shilly-shally, shortie1, shortie2, sickie1, sickie2, skinny, smiley, smokie, softie, sparky, sticky, stiffy, stogie, stovies, stubby, surfie, sweetie1, sweetie2, talkie, tattie, techie, telly, ticky-tacky, tinny, toastie, toady1, toady2, tootsie1, tootsie2, townie1, townie2, trackie, tranny1, tranny2, tranny3, tranny4, trekkie, trendy, trusty, tummy, tweeny, twinky1, twinky2, veggie, wacky backy, walkies, walkie-talkie, waspie, wedgie1, wedgie2, weepie, welly1, welly2, westie, wheelie, whitey, willy, woody, wrinkly, yachtie, yardie1, yardie2, yorkie, yuppie, yippie.

Chapter 9

Neoclassical Word Formation in English: A Paradigm-Based Account of -scope Formations Ana Díaz-Negrillo 1

Introduction

While a paradigmatic arrangement of morphologically-related forms is probably the default conceptualization in inflection, lexical morphology, in particular derivation, has not always been in the tradition of describing forms as paradigms, at least in English. The increasing interest in paradigm-based approaches to lexical morphology is shown, for example, in relatively recent publications (Bauer, 1997; Pounder, 2000; Štekauer, 2014; Bonami & Strnadová, 2019, inter alia), volumes like the present one, and large research projects like the one on derivational networks across European language families (see Štekauer, Bagasheva, & Körtvélyessy, 2020). Supporters of a paradigm-based approach to lexical morphology assume that the differences between inflection and derivation are a matter of degree, rather than a discrete distinction, which therefore does not prevent a paradigm-based conceptualization of derivation (Bauer, 1997; see Štekauer, 2014 for the relevance of derivational paradigms in morphology). Moreover, word-based approaches to lexical morphology solve a number of issues that remain in morpheme-based views of morphology, for example, directionality among derivatives, and derivatives which lack base forms. These, and many others, make paradigm-based approaches worth exploring even if there are opposing views among the supporters of paradigm-based word formation as regards, for example, which word-formation processes are eligible for consideration, or as regards unresolved questions, like what semantic categories are to be used for the conceptualization of paradigms. The latter is only evidence that the area is still underdeveloped and very much in need of research. This paper intends to contribute to paradigm-based approaches to lexical morphology by providing empirical evidence from neoclassical word formation. The paper explores a word-based treatment of the lexical formations which are morphologically-related to the base -scope, and discusses the

© Koninklijke Brill NV, Leiden, 2020 | doi:10.1163/9789004433410_010

214

Díaz-Negrillo

suitability of paradigms for the conceptualization of neoclassical compounds and their derivatives. The study is based on 444 entries from the Oxford English Dictionary (hereafter, OED) giving rise to 469 formations, which have been classified into 224 sets of morphologically-related formations and 14 semantic categories. This paper is organized as follows: after the present introduction, §2 discusses the foundations and major issues in paradigm-based word formation. §3 is about neoclassical compounds and some of the foreseeable issues at play in a paradigm-based account of these constructions. Section 4 describes the methodology used. §5 presents the paradigm designed for -scope neoclassical formations. Finally, §6 recapitulates on some of the issues disclosed in the study and gathers some conclusions. 2

Paradigms in Lexical Morphology

Most paradigm-based approaches to lexical morphology have limited their scope to derivation. According to the specialized literature, the concept of derivational paradigm may be defined in the following terms (see, for example, Bauer, 1997; Booij, 2007; Štekauer, 2014 for overviews and identification of the properties of paradigms): a) Derivational paradigms consist of derivationally related forms organized around the same base. b) The derivationally-related forms in a paradigm are displayed in cells, or slots, which stand for the variety of contrasting semantic categories exhibited by the units that occupy those cells. This means that paradigms are semantically-rooted and established as systems of meaning-form contrasts. c) Derivational paradigms are descriptive tools, given that they are used for displaying the morphological pattern of a given base or base-type, but also predictive tools, given that they are also used for establishing the potential morphologically-related lexemes. As to the first property above, it seems widely accepted that derivational paradigms consist of all the derivationally-related forms which emerge from one base. Indeed, the very notion of ‘derivational’ paradigm restricts its application in the lexicon, in that, even if derivation covers a large extension of the lexicon, there are areas in the latter which are covered by other mechanisms, like borrowing, etc. A first implication of this statement is that semanticallyrelated forms which may potentially occupy a cell in the paradigm of a given lexeme, but which are not derivationally-related, are in principle to be left outside of the paradigm. Cases in point for English are, for example, suppletive

Neoclassical Word Formation in English

215

bases of a non-native origin, e.g., dental and feline, which express relation and quality with respect to tooth and cat respectively. Absence of these forms in the respective paradigms of tooth and cat on the grounds or their suppletive character may result in defectiveness, given that there may be no alternative derivationally-related non-suppletive formations which may occupy these cells. It is still to be agreed, however, whether dental relates synchronically to tooth in the derivational system just as was relates to be in the inflectional system and should thus be accepted as part of the same paradigm (see Bonami & Strnadová, 2019, pp. 177–179 on suppletion in inflectional and derivational paradigms). Relatedly, what morphological mechanisms are covered in derivational paradigms also seems to be a matter of discussion. While, for example, Pounder (2000) does not seem to restrict the mechanisms involved in the composition of derivational paradigms, Štekauer (2014, p. 369) limits them to affixation. According to him, affixation ensures predictable relations to a greater extent than non-affixational processes. Again, restricting the scope of derivational paradigms may have consequences in terms of defectiveness in the paradigm, given that a number of gaps will be left unfilled, not just by the formations resulting from non-affixational processes, but also by affixed formations deriving from the non-affixational formations in point. It is still to be agreed whether, eventually, limiting derivational paradigms to affixation will actually affect the predictability potential expected from derivational paradigms. A second property of derivational paradigms is that they are semanticallyrooted, which in practice means that the cells or slots in a paradigm stand for the variety of contrasting semantic categories exhibited by the various derivationally-related units. The semantic configuration of paradigms is crucial for the last property mentioned above, given that the descriptive power and, more crucially, the degree of predictability afforded by derivational paradigms rest upon the regularity of the semantic relations. Bonami and Strnadová (2019) refer to common content relations as ‘alignment’, such that randomrandomize and class-classify will be aligned through the causative relation. In principle, the more stable the semantic relations, the greater the paradigm’s potential to predict the derivational forms of a given lexeme. A first issue in this respect is that what semantic categories should pattern derivational paradigms is still to be agreed. As Fradin (this volume) states, while inflectional categories are organized around word-classes (traditionally, declensions and conjugations) and widely recognized grammatical oppositions, derivational paradigms are organized around lexical meaning, which is open to a variety of classifications. Bauer (1997, p. 253) explains that “[…] there are constellations of positions in matrices which […] can nonetheless be considered slots

216

Díaz-Negrillo

which must be available for any word of the appropriate word-class.” Among others, he mentions deverbal nominalizations, adverb formations derived from adjectives, negative forms derived from certain adjectives, etc. However, semantically coarse-grained slots are likely to gather semantically contrastive formations in the same cell, which may be undesirable. Another issue in semantic categorization is that semantic categories in derivation often overlap, e.g., hospitalize represents causation and, probably in more finely-grained scheme, also location (after Bonami & Strnadová, 2019), and wandelaar (Dutch) represents both agent and masculine (Booij, 2002, p. 102 after Fradin, this volume). All in all, it seems crucial that a set of semantic categories is proposed and clearly defined for progress on derivational paradigms to be made. Bagasheva (2017) has suggested a series of universal semantic categories for affixational processes which have been used in a large project on paradigmbased derivation, and which will certainly shed light on the development of a paradigm-based account of derivation (Štekauer et al., 2020). As mentioned at the beginning of this section, in word formation the emphasis has been on derivational paradigms. However, recent attempts at a paradigm-based account of compounding have also been made (see, for example, Bagasheva, this volume; Radimský, this volume). Bagasheva (this volume, pp. 28–29) argues that it seems feasible to assume that “[…] the same principles of paradigmaticity may apply across the whole spectrum—inflectional morphology, affixation, and compounding” and, just as across inflectional and derivational paradigms, the differences will be a matter of degree rather than of principle. The present paper aims to demonstrate how a paradigm-based account of neoclassical compounds and their derivatives seems a most suitable approach, given the very nature of these compounds. First, the sometimes unclear morphological status of some of their constituent elements (i.e., combining forms which behave as affix-like roots) leads to an unresolved discussion of whether the formations should be accounted for as derivation or compounding. And second, neoclassical compounds very often undergo derivational processes, yielding as a result a range of morphologically-related sets of formations within the continuum of neoclassical word formation. Hence, a fullyfledged account of neoclassical compounds and their morphologically-related formations from a paradigm-based perspective will not only avoid unresolved theoretical issues, but will also set in the framework of a comprehensive account of the scope of neoclassical word formation. It was advanced in the Introduction that whether paradigms are the most suitable conceptualization for word formation is a matter of debate, and one very much dependent on theoretical views of morphology. Advantages cited by their proponents are that derivational paradigms offer a neat form-meaning

Neoclassical Word Formation in English

217

mapping of lexical formations, stand as a powerful predictive tool of potential morphologically-related lexical formations, as recently mentioned, and that a conceptualization of word formation in terms of paradigms avoids a number of theoretical issues many of which arise from the stress often given to structural aspects (see Booij, 2007 for an overview). Thus, as mentioned in Booij (2007), and stressed in Litta and Budassi (this volume), directionality becomes irrelevant in a paradigm-based approach to derivation. Derivationally-related lexemes are no longer presented as linear sequences of formations arranged in a directional fashion, as in models like Item-and-Arrangement. Relatedly, if we accept that derivational paradigms are not restricted to affixation, in the non-linear arrangement of paradigms issues concerning directionality in conversion, back-formation and truncation (e.g., -ist and -ism words) also become irrelevant. Finally, as will be proposed here, if we accept that lexical paradigms should not be restricted to derivation, then issues concerning the structural status of some morphological units, including affixoids and combining forms, will also be avoided. Ultimately, the most central components in a paradigmbased approach to word formation are the morphologically-related lexical formations and the semantic realizations aligning them. 3

Neoclassical Compounds

Neoclassical compounds are widely described as ‘compounds’ because they consist of lexical elements which are equivalent to the bases in compounds, except that in neoclassical compounds at least one of the bases is bound, which is explained in its classical origin. As a result of these properties (lexical status, bound structure and classical origin), the bound element(s) in compounds have been referred to as ‘combining forms’ (hereafter, CFs) (Bauer, 1983, 1998), ‘stems’ (Kastovsky, 2009), ‘bound roots’ (Scalise, 1986; ten Hacken, 1994; Adams, 2001; Baeskow, 2004), or ‘neoclassical formatives’ (ten Hacken, 2012). In addition to CFs, they also often consist of a linking vowel, typically -obut in some cases also -i-, going back to the Ancient Greek and Latin thematic vowels respectively. Semantically, most neoclassical compounds are largely associated with nominal semantic categories, like entity, e.g., X-lith or event, e.g., X-cide, etc., given that most of them form nouns. They tend to be endocentric and right-headed, e.g., anthropomorph, but coordinative neoclassical compounds also exist, e.g., alveolo-palatal. They tend to undergo recursion (proctocolectomy, echocardiography), showing a variety of semantic relations among their left-hand constituents and the left-constituents and the head. Vagueness in

218

Díaz-Negrillo

compound semantics has been reported in that there seems to be unpredictable semantic ellipsis (McCray, Browne, & Moore, 1988, p. 166; from Bauer, Lieber, & Plag, 2013, pp. 485–486). So, as cited by the authors, cardiophobia is not fear of hearts but fear of heart diseases. Neoclassical compounds also often tend to undergo derivation (e.g., electrocardiographically), patterning as a result a gamut of semantic possibilities. A final distinctive feature of neoclassical compounds is that they pertain to the domain of learned or specialized vocabulary. Thus, for example, the CF s -scope and -ectomy are largely used in Medicine and -lith in Biology and Pathology. As a result of their specialized use, it is often the case that outsiders of these disciplines have to look up neoclassical compounds, or their constituents, in terminological dictionaries. Still, not all neoclassical compounds are as specialized as the above and, therefore, as infrequent in everyday language. Semantic extension may take place in the CF s and, as a result, they may acquire a non-specialized meaning. Lüdeling and Evert (2005), in relation to German -itis, which in Medicine means ‘inflammation of a particular body part,’ reported that this form has become particularly productive with a more common use meaning ‘excessive or in excess,’ e.g., telefonitis. Similarly, Lasserre (2015) reports that French -phobie shows more than one meaning: one within the specialized domain of Psychology (‘fear’), and another non-specialized meaning (‘hostility’). Neoclassical Compounds at the Borderline between Derivation and Compounding Neoclassical compounds have been claimed to be largely heterogeneous in terms of their varied structural make-up (Bauer, 1998). This poses the question of whether some formations should be better described as derivatives rather than as compounds. Their heterogeneity has been claimed to result from their gradual integration into the English language, through initial stages of borrowing of foreign formations and the foreign word-formation rule (e.g., anthropomorph), till their current integration into the English language involving nativization of the pattern (e.g., Kremlinologist) (ten Hacken, 2012, p. 82). Hence, there exist formations nowadays like bio-engineering or Egyptology, where a putative CF combines with a free element. These formations introduce the challenge that, if contrasted with native formations, there is an argument for interpreting them as derivatives, at least on structural grounds. This stands as a first feature which neoclassical compounds share with derivatives. A second property of neoclassical compounds which they share with derivatives is that some CF s show high productivity, which is commoner among affixes than among compound bases. This is the case of -logy, usually identified as a 3.1

Neoclassical Word Formation in English

219

CF, but which has also been described as a suffix (Prćić, 2008, p. 13; Kastovsky, 2009, p. 5). Third, semantic specialization and development of suffix-like meanings have been claimed for individual CF s. In French, in the framework of lexeme-based morphology, suffix-like status has been claimed for -logue on the grounds of its isolated use and semantic specialization. The element combines mainly with nouns meaning ‘specialist’ which, arguably, makes it equivalent to -ist (Amiot & Dal, 2007; see also Lüdeling & Evert, 2005, for German -itis, and Lasserre, 2015 for French -phobie mentioned above, for extension among CF s). A fourth property shared by derivation and neoclassical compounding is phonological. Some final combining forms (hereafter, FCF s), namely -cracy, -graphy, -logy, -phile, and possibly some others too, impose antepenultimate stress, just as some non-native stress-influencing suffixes (Plag, 2018, p. 154). Depending on the phonological make-up of the CF, the stress may fall on the linking vowel (-o-), e.g., meritócracy, filmógraphy, embryólogy (acute accent indicates primary stress), or on the syllable before the linking vowel, e.g., nécrophile. Finally, also on phonological grounds, in addition to some CF s being stress-influencing, the presence of the linking vowel may make at least FCF s look like vowel-initial elements (e.g., -ology, -icide), which is one defining feature of non-native suffixes. These two phonological properties are particularly important because, if CF s behave phonologically as non-native suffixes, they may be perceived as such by speakers of English, who may ultimately feel at ease using them morphologically as non-native suffixes. All in all, it is controversial whether some of these formations should be described either as compounds or as derivatives. First, because not all CF s behave similarly and, second, because, synchronically, the same bound element may appear in formations which look like compounds, and is hence interpretable as a CF, and also in other formations which look like derivations, and is hence interpretable as an affix. Therefore, the interpretation of neoclassical compounds in terms of discreet categories is only problematic. Alternatively, a paradigm-based approach may stand as a valid framework for the conceptualization of neoclassical compounds, largely because, being word-based and semantically-oriented, it avoids the major issues discussed above, which are largely motivated by morpheme-based accounts of the individual elements and of the formations. In addition, a paradigm-based approach to neoclassical compounds seems to be a suitable descriptive framework given that the formations tend to undergo (recursive) derivation. Hence, for a sound and comprehensive account of neoclassical compounds, a model which makes possible the characterization and semantic potential of neoclassical compounds and their derivatives seems desirable.

220 4

Díaz-Negrillo

Method

The most basic form of the paradigm and which we take as a starting point is -scope. This unit is often cited as an FCF (Bauer & Huddleston, 2002, p. 1661; Bauer, Lieber, and Plag, 2013, p. 455; Plag, 2018, p. 153), and therefore we can accept that other morphologically-related formations involve either combination of -scope with initial combining forms (hereafter, ICF s), prefixes or suffixes. According to the OED, -scope represents modern Latin -scopium which in turn goes back to Ancient Greek σκοπεῖν, meaning “to look at, examine,” and is used in words denoting scientific instruments for enabling the eye to view or examine or make observations. In the dataset, -scope formations cover the domains related to viewing including Medicine, Astrology, Film and Photography, and Engineering. Semantic extensions may cover figurative viewing, as in the domain of divination. From the formal point of view, -scope tends to be preceded by the linking vowel -o- which is evidence of the Greek origin of the CF.1 It may combine with bases of Greek origin (e.g., laryngoscope), Latin origin (e.g., fluoroscope), or English origin (e.g., sniperscope). The FCF -scope may be preceded by bound and free bases as shown in the previous examples, which may occur in endocentric neoclassical compounds recursively or in combination with prefixes (e.g., phonoelectrocardioscope, atomic force microscope, hyperstereoscopy). These formations, in turn, tend to undergo suffixation and combine with suffixes of classical origin (e.g., ultramicroscopic), English origin (e.g., microscoping), or both (e.g., ultramicroscopically). They can undergo conversion too (e.g., stethoscopeV). The OED has been used to retrieve all the formations in the study. First, all the formations including the unit -scope were queried. Due to space limitations, the scope of the study was restricted to non-recurrent compound or prefixed formations where -scope is followed by one or more suffixes. This means that -scope formations with co-occurring ICF s/prefixed compounds (e.g., hygrobaroscope) have been left out. Obsolete meanings or formations were also excluded from the study.2 Still, some entries or meanings included in the study may be marked as ‘rare’ or ‘historical’ in the OED because they may denote instruments which are no longer in use. Finally, formations or meanings which pose difficulties in their semantic interpretation have been excluded too (e.g., 1  Exceptions include sniperscope and snooperscope. They are formed in combination with English words, which may explain the behaviour of the formations (see Díaz-Negrillo, 2014 for a diachronic account of variation in neoclassical compounds’ defining features). 2  Obsolete meanings also include those marked as ‘disused’ in the OED.

Neoclassical Word Formation in English

221

Table 9.1 Distribution of formation types in the study

Types of formations

Formations (n)

[X -scope] [-scope X] Total

173 289 462

theatroscope) or whose meanings move away from the central meaning of viewing (e.g., microscoped “carefully selected”; nanoscopic “having or involving dimensions in the range 1 to 100 nanometres”).3 All in all, the study covers 444 OED entries or words. The formations were then analysed semantically and morphologically into data-driven semantic categories in two stages, according to the two types of formations in the dataset: i) [X -scope] formations, covering prefixed/compound formations, and ii) [-scope X] formations, covering suffixation and converted formations. Given the nature of the formations, the first stage involved a subcategorization of the formations in terms of their left-hand element, and the second stage covered a categorization of the formations yielded by the last derivational stage. All in all, the dataset of the study is distributed as shown in Table 9.1. All the formations were then arranged into sets. A set consists of a compound/prefixed word and all its possible suffixed derivatives. All in all, the paradigm consists of 223 sets of formations. Of these, 94 sets (42.15%) consist of more than one formation, reaching a maximum number of 13 formations, as in telescope and all its possible suffixed derivatives. The semantically classified sets of formations finally generated the word-formation paradigm of -scope, which is displayed in Appendix 1, for [X -scope] formations, and in Appendix 2, for [-scope X]) formations.4 The paradigm is organized horizontally into semantic categories and morphological realizations, and vertically into sets of formations.

3  Nanoscopic and mesoscopic seem to be analogical with one of the extended meanings of microscopic, namely, “very small, minor, or insignificant”. Formations and meanings related to size and subsequent semantic extension have been excluded from the dataset. 4  Given its size, the paradigm had to be shown in two appendices. The first column in Appendix 2 is intended as a summary of Appendix 1.

222 5

Díaz-Negrillo

-scope Formations in a Paradigm-Based Perspective

This section gives an overview of the paradigm of -scope morphologicallyrelated lexical formations. First, there will be an account of [X -scope] formations and then of [-scope X] formations. For a better illustration of the semantics of the formations, examples from the dataset will be given at all times. To ease the semantic decomposition of the examples, whenever the lefthand element in the formation is not an English word, an English counterpart will be given. The OED definition of the examples will be given too. OED quotations may also be provided, if the formation in question is not defined in the OED, or if it clarifies the semantic decomposition of the examples.5 5.1 [X -scope] Formations [X -scope] formations yield nouns denoting instruments for viewing, which we therefore identify with the semantic category instrument. Morphologically, the formations may consist of combinations of -scope with an ICF, a prefix or a word. Regardless of its morphological status, the left-hand element represents a number of semantic categories, which yield a classification of [X -scope] formations in the study. Each of these semantic categories represents a cell in the paradigm (see Appendix 1). Table 9.2 shows the distribution of [X -scope] formations. A detailed account of the formations by semantic categories follows the table. Table 9.2 Distribution of [X -scope] instrument N formations

X

n

entity mode of action purpose agent place Total

144 19 4 3 3 173

5  On many occasions a quotation from the creator of the instrument is far more clarifying of the origin of the formation than the actual definition, which tends to define the instrument in scientific terms. Derivatives in the OED often lack a definition.

Neoclassical Word Formation in English

223

[entity -scope] instrument N The left-hand element in these formations refers to the object or phenomenon intended to be viewed with the instrument in question. The latter may be referred to a) directly, Bronchoscope n. (broncho- bronchus) “an instrument for inspecting the interior of the bronchi.” Helioscope n. (helio- sun) “an apparatus for observing the sun without injury to the eye […].” b) through one of its distinctive properties, Microscope n. (micro- small) “an optical instrument used to view very small objects […].” Galvanoscope n. (galvano- galvanic) “an instrument for ascertaining the presence of galvanic electricity.” c) through a property acquired as a result of the action of the instrument, that is, the instrument causes the entity to be X, Erythroscope n. (erytho- red) “an optical contrivance by which the green of foliage is caused to appear red, while all other green objects retain their natural hue.” Pseudoscope n. (pseudo- false) “a binocular optical instrument for producing a stereoscopic image in which the appearance of depth is reversed by transposing the images seen by the left and right eyes.” 1852 C. Wheatstone in Philos. Trans. (Royal Soc.) 142 11 As this instrument conveys to the mind false perceptions of all external objects, I have called it the Pseudoscope. d) or through its location, Aeroscope (aero- air) “any of various instruments for estimating the concentration of particles in the air […].” [entity -scope] formations cover the highest number of formations in the study. A quantitative breakdown of the formations in this cell is in Table 9.3.

224

Díaz-Negrillo

Table 9.3 Distribution of [entity -scope] instrument N formations

ENTITY

n

Named entity Quality Resultant quality Locative Total

114 17 9 4 144

[mode of action -scope]instrument N The left-hand element in these formations makes reference to the instrument’s functioning. In particular, it may refer to a) a component of the instrument characterising the instrument’s mode of action, Cycloscope n. (cyclo- circle) “an apparatus for measuring the velocity of revolution, by means of a revolving ruled cylinder […].” Fibrescope n. ( fibre n.) “a fibre-optic device used to view inaccessible internal structure, esp. in medicine to examine passages in the body.” b) a central action characterising the instrument, Reflectoscope (reflecto- reflect v.) “a device which projects an image of an object on to a screen by reflecting light from it.” Stroboscope (strobo- a twisting or a whirling around) “a scientific toy which produces the illusion of motion by a series of pictures viewed through the openings of a revolving disc.” c) the manner in which the characterising action is performed, Tachistoscope (tachisto- swiftest) “an instrument by means of which objects may be presented to the eye for a brief measured period, a fraction of a second […].” Tachyscope (tachy- swift) “a kind of kinetoscope, in which a series of representations of an object in successive phases of motion are rapidly revolved, so as to present the appearance of actual motion.”

225

Neoclassical Word Formation in English

d) a directional reference in relation to the functioning of the instrument, as in the examples below, where the left-hand element refers to the direction of the illumination of the object, Episcope n. (epi- upon) “a magic lantern for projecting images of opaque objects.” Diascope n. (dia- through) “a magic lantern or projector by which the image on a transparent slide or the like is projected on to a screen, wall, etc.” A quantitative breakdown of the formations in this cell is in Table 9.4. Table 9.4 Distribution of [mode of action -scope] instrument N formations

MODE OF ACTION

n

Entity Action Manner Direction Total

9 4 4 2 19

[purpose -scope]instrument N The left-hand element in these formations denotes the ultimate purpose of the instrument, additionally to viewing, Finderscope n. ( finder n.) “a small, low-magnification telescope with a wide field of view attached to a larger telescope for use as an aid for locating and aiming the telescope at an object of interest.” Resectoscope n. (resect v.) “an endoscopic instrument used to perform resections […].” [agent -scope]instrument N The left-hand element in these formations may refer to a person who is connected with the instrument in question. In particular, it may refer to

226

Díaz-Negrillo

a) the user of the instrument, Sniperscope n. (sniper n.) “a small device which converts infra-red radiation to a visible image and may be fixed to a gun so that it can be aimed in the dark.” Snooperscope n. (snooper n.) “a device which converts infra-red radiation to a visible image; esp. a pair of such devices fitted together and worn on the head to provide binocular vision in the dark.” b) the inventor of the instrument, Debusscope n. (M. Debus) “an optical contrivance consisting of two mirrors placed at an angle of 72°, so as to give four reflections of an object or figure placed between them and form composite figures for purposes of decorative design, etc.” 1862 Timbs Year-bk. of Facts 144 M. Debus has invented this new form of kaleidoscope […]. A quantitative breakdown of the category is in Table 9.5. Table 9.5 Distribution of [agent -scope] instrument N formations

AGENT

n

User Inventor Total

2 1 3

[place -scope]instrument N The left-hand element in these formations refers to the place in which the instrument is to be used. Polemoscope n. (polemo- war) “an optical instrument fitted with a mirror or pair of mirrors set at an angle to the line of sight, for use in viewing objects not directly before the eye.” Nature 18 Mar. 68/1 Some of these trench periscopes recall the polemoscope, described by Helvelius [sic] in the seventeenth century for military purposes.

227

Neoclassical Word Formation in English

Trenchscope n. (trench n.) 1918 E. S. Farrow Dict. Mil. Terms 629 Trenchscope, a simple periscope, used in the trenches (permitting a safe view to the front), consisting of parallel mirrors in a long wooden box, both set 45° to the long axis of the box. This semantic category also includes a formation where the left-hand element refers to the relative position with respect to another contrivance the instrument in question is to be used with. Alternatively, this formation may be classified as mode of action. Hyposcope n. (hypo- under) “a form of periscope for attachment to a rifle or for use as a hand instrument.” 5.2 [-scope X] Formations [X -scope] formations may undergo suffixation, which assigns a variety of semantic categories to the resulting formations, here referred to as [-scope X] formations. Table 9.6 below shows the distribution of [-scope X] formations according to their semantic category, in other words, their distribution into paradigmatic cells (see Appendix 2). One and the same [-scope X] formation may be in more than one category as a result of polysemy, as will be explained below. A detailed account of each of the semantic categories follows the table. Table 9.6 Distribution of [-scope X] formations

Semantic category

n

event agent experiencer relation quality possession means manner/viewpoint action Total

95 19 1 91 27 2 34 8 12 289

228

Díaz-Negrillo

[-scope X]event N6 These formations refer to the action or activity of visual observation, often performed with a -scope instrument. Otoscopy n. (oto- ear) “examination of the ear using an otoscope; the use of an otoscope.” Cranioscopy n. (cranio- cranium) “examination of the size and configuration of the skull […].” This category covers action and activity nouns. The most frequent realization is -y, and much less so -ing, as shown in Table 9.7. Table 9.7 [-scope X]event N: realizations

EVENT

n

-y -ing Total

91  4 95

There is a high proportion of event formations which parallel an instrument formation in the same set, e.g., otoscope—otoscopy (62.10%, n = 59/95), while there are others which do not, e.g., cranioscopy, for which *cranioscope is not recorded in the OED. Hence, claiming the realization -y for the latter cases may not be fully satisfactory because it implies derivation from a [X -scope] formation which is actually inexistent. Alternatively, two different realizations may be proposed, namely -y for those event formations which parallel an instrument formation, and the FCF -scopy for those formations which do not. The latter seems uneconomical, however, specially given the semantic proximity of the two kinds of formations, so that is why this description marks only one single realization. Still, the actual display of the word-formation paradigm into cells is evidence of the event formations which are strictly derivational and those which are not in view of the existence or not of a paradigmatic 6  Another name for this category is ‘action,’ as in, for example, Bagasheva’s (2017, pp. 53–56) semantic classification for affixation. We use ‘action’ as a term for converted verbs in the paradigm, as will be shown below (see also Štekauer, 1996, p. 46 for a semantic identification of converted verbs with ‘action’).

Neoclassical Word Formation in English

229

gap.7 Moreover, in a word-based account issues having to do with morphological paths stand as peripheral. [-scope X]agent N These formations refer to persons who in some way or another are associated to the instruments and techniques denoted in the instrument and event categories. Sigmoidoscopist n. (sigmoid having a double curve like the letter S, here referring to the last curving portion of the colon) “one who uses a sigmoidoscope.” Ophthalmoscopist n. (ophthalmo- of or relating to the eye) “an expert in ophthalmoscopy; (also) a person who uses an ophthalmoscope.” This category covers nouns. The semantic and morphological relation between these formations and instrument and event formations is shown in the paradigm in that 90% of the formations in the category (n = 18/20) parallel both an instrument and an event formation (n = 13), or at least an event formation (n = 5) in their respective sets. The other two formations, hydroscopist and oneiroscopist, may be interpreted as analogical with other s­ copist formations in the paradigm. In other words, these two formations seem to emerge as a result of pressure from the paradigm category (vertical pressure) instead of from pressure from their set (horizontal pressure), that is, from derivation in this case.8 [-scope X]experiencer N These formations refer to persons experiencing mental processes. Misoscopist n. (miso- to hate) “a person who hates sights.”

7  For three formations, a parallel entity formation is missing in the dataset, but still recorded in the OED. The reason for excluding them from the dataset is that one of them (mixoscope) is obsolete, and the other two (cerebroscope and uranoscope) do not denote instruments and were interpreted as semantically unrelated. 8  These two formations are unrelated to scientific observation or examination. They belong to the domain of divination, and therefore are an instance of semantic extension from the core meaning of the rest of the formations in the category. This may support their nature as analogical forms.

230

Díaz-Negrillo

The formation has no instrument or event counterparts, which suggests that, again, analogy may be at play. This may explain the scarce representation of this category in the paradigm. [-scope X]relation Adj This type of construction covers formations which show a relational meaning with respect to a -scope instrument and/or its use. Embryoscopic adj. (embryo- embryon) “of or relating to embryoscopy.” Megascopic adj. (mega- very large) “of or relating to the megascope or its use.” This category covers adjectives and is realized by -ic, -ical, and -ial derivatives, which are quantitatively distributed as shown in Table 9.8: Table 9.8 [-scope X]relation Adj: realizations

Realizations

n

-ic -ical -ial Total

75 15  1 91

All the -ical and -ial formations in the data have a semantically equivalent -ic formation, which suggests that -ic is the default realization in this category and -ical and -ial are subsidiary competing realizations. Despite the semantic connection of the present category with the categories event and instrument, not all the formations representing relation have a parallel instrument or event formation in their respective sets (n = 2/91). The reason may be that the formations in questions result from analogy with other formations in this category, as in the examples below: Geoscopic adj. (geo- earth) “relating to visual examination of geological or geographical features.” Nooscopic adj. (noo- mind) “relating to the study of the mind.”

Neoclassical Word Formation in English

231

[-scope X]quality Adj These formations express current or resultative qualitative meaning. Periscopic adj. (peri- around) “of a lens, eyeglass, etc.: giving a wide field of view.” Pseudoscopic adj. (pseudo- false) “relating to or characteristic of a pseudoscope; involving or exhibiting reversal of the apparent depth of an image.” The OED entries considered for this category may also have been considered for the category relation because of their polysemous nature, as shown for example in the definition of pseudoscopic above. 37.03% (n = 10/27) of the formations in this category are polysemous with relation formations. As in previous categories, some formations in this category do not have a parallel instrument or event formation in their respective sets, which may be explained in terms of analogy. This is the case of 25.92% (n = 7/27) of the formations in this category. Below are some examples: Basiscopic adj. (basis- base) “looking or turned towards the base.” Platyscopic adj. (platy- broad) “designating a type of magnifying glass containing a combination of lenses, giving a wide field of view with reduced aberration.” This category covers adjectives and is realized by -ic, -ical, -ed, and -ing derivatives, which are distributed as shown in Table 9.9: Table 9.9 [-scope X]quality Adj: realizations

Realizations

n

-ic -ical -ed -ing Total

21 3 2 1 27

232

Díaz-Negrillo

-ed formations, namely kinescoped and rotoscoped, show resultative qualitative meaning. The only -ing formation, telescoping, shows simulative meaning in addition to quality. [-scope X]possession Adj This construction covers formations which can be paraphrased as “fitted with a [X -scope],” as shown in the examples below: Stethoscoped adj. (stetho- chest) “equipped with a stethoscope.” Telescoped adj. (tele- far-seeing) 1862 G. W. Thornbury Life Turner II. 170. Looking through his telescoped hand. The formations are realized by -ed adjectives. [-scope X]means Adv9 These formations refer to the means by or instrument with which an entity is observed. Gastroscopically adv. (gastro- belly) “by means of a gastroscope.” Otoscopically adv. (oto- ear) “by means of otoscopy; by means of an otoscope.” All the formations in this category have a parallel instrument and/or event formation in their respective sets. The category is realized by -ly adverbs. As will be shown, means is not the only category realized by -ly adverbs, given the polysemous nature of the formations. Still, means is the most representative category for -ly adverbs in the paradigm both quantitatively and also semantically. This is because the next category represents extended meanings from means, which exist only for a few of the means formations (n = 8/34). [-scope X]manner/respect Adv These formations refer to the manner in which an entity is observed or the point of view from which an utterance is formulated. 9  An alternative name for this category is instrument (see Fradin, this volume).

Neoclassical Word Formation in English

233

Gyroscopically adv. (gyro- spiral) “by or with a gyroscope; in a gyroscopic manner.” Rhinoscopically adv. (rhino- nose) “by means of a rhinoscope or rhinoscopy; as regards rhinoscopy.” All the formations in this category are also classified in the category means, as can be gathered in the definitions just above. Also, all the formations in this category have morphologically-related formations in the categories instrument and/or event in their respective sets. [-scope X]action V These formations mean to visually examine an entity using a particular instrument. Cystoscope v. (cysto- bladder) “to examine (the bladder) with this instrument [a cystoscope].” Spectroscope v. (spectro- spectrum) “to examine by means of a spectroscope.” This category covers converted verbs and they all have parallel instrument formations in their respective sets. 6

Discussion and Conclusions

This paper set out to explore a paradigm-based approach to neoclassical compounds and their morphologically-related lexical formations. To achieve this aim, the paper has investigated the network of 462 morphologically-related formations yielded by the FCF s­cope, based on the lexicographic evidence of 444 OED entries. A word-based approach to neoclassical word formation shows a number of advantages compared to morpheme-based approaches. The major conclusions are discussed below. From the semantic point of view, the study draws on 14 semantic categories. The categories were driven from the dataset, which became especially necessary for [X -scope] formations. Ready-made classifications of compounds, like Levi’s widely used classification (1978), proved unsuitable for the formations in question. This is, first, because for [X -scope] formations the study required semantic categories which were compatible with the categories to be used for [X -scope] formations and, second, because there were formations which were

234

Díaz-Negrillo

not classifiable into Levi’s categories and which seemed to be specific of neoclassical compounds, of instrument formations, or of both (this point still needs further investigation) e.g., mode of action. As a result, the present study may be a first step towards the configuration of a set of semantic categories for a paradigm-based accounts of (neoclassical) compounds. The most representative semantic category among [X -scope] formations is entity, covering formations denoting objects or phenomena under examination, which is readily expected from words denoting instruments for viewing. A peculiarity of the category, however, is that the entity in question is not always denoted directly, but through reference to a current or resulting property of the object, or through its location. This seems to be consistent with McCray et al.’s (1988) claim discussed above that there exists certain semantic vagueness in neoclassical compounds. So microscope is not an instrument which is small, but an instrument which is used for examination of small entities. Admittedly, the semantic nature of [X + FCF] formations is largely unexplored and deserves more extensive and detailed examination both in terms of the semantic disambiguation of non-recurrent formations, like the ones explored in this study, and the semantic relations existing in recursive formations. Another interesting fact about [X -scope] formations is that the semantic categories used cover a variety of meanings, ranging from more concrete to more abstract or relational meanings. Thus, the semantic demarcations often set for compounding and derivation, but which do not always apply systematically, may be avoided. For [-scope X] formations, widely recognized categories were used, though admittedly some may seem more fine-grained than others. In some cases, as discussed in the Introduction, a formation seemed to spread over more than one semantic category, for example, rotoscoped and kinetoscoped denote resulting qualities, and telescoping has simulative meaning in addition to quality. Coarse-grained classifications in these cases have been preferred for economy reasons. In other cases, however, the actual semantic nature of the formations in the study seemed to require a semantic category as detailed as means, which gathers 79% of the adverb formations (-ly), while manner/view was set for the rest of the formations of the same kind. For progress in paradigmbased word formation, we advocate for hierarchical semantic classifications allowing for categorizations of various degrees of granularity depending on the semantic nature of the formations. Documentation of the semantic categories is crucial, given the variety of existing classifications and interpretations of one semantic label. Among [-scope X] formations the most representative semantic categories are event and relation. The categories gathering fewer formations also

Neoclassical Word Formation in English

235

seem to be those whose members show meanings which are extended from more frequent categories. In other words, less frequent categories gather extended meanings of polysemous formations, e.g., quality as opposed to relation, or manner/respect as opposed to means. Meaning extension may be accompanied by loss of specialization of the resultant meaning, which is shown in the formations related to the domain of divination. This is consistent with claims that semantic extension results in a non-specialized meaning (cf. Lüdeling & Evert, 2005; Lassarre, 2015). From the morphological point of view, the paradigm covers compounding, prefixation, suffixation, and conversion. Compound/prefixed formations, which are all of nominal nature, represent over one third of the formations in the study. The line between compounding and prefixation in the study is difficult to draw, especially when it comes to cases like auto- or pseudo- formations which are sometimes described as derivatives and sometimes as compounds. As discussed in the literature review, this is precisely one of the reasons why we believe paradigm-based approaches are better suited to account for neoclassical word formation than morpheme-based accounts. Suffixation abounds in the paradigm. It yields nouns (-y), adjectives (largely -ic/-ical) and adverbs (-ly). Conversion is less frequent but still applies to the chains which seem to be more dynamic in the paradigm in terms of their number of morphologicallyrelated formations, e.g., telescope (n = 12), microscope (n = 11), periscope (n = 8), spectroscope (n = 8), etc. This seems to show that, even if secondary in terms of number of formations compared to some suffixational processes, and contrary to views discussed earlier (cf. Šteakauer, 2014, p. 369), conversion does not seem to be accidental and, therefore, may stand as predictable as other suffixational processes in the paradigm. Moreover, conversion is the only mechanism in the paradigm which yields verbs, i.e., action formations.10 The morphological relevance of conversion in the paradigm as a predictable mechanism yielding action formations suggests that conversion should be considered among derivational/word-formation paradigms; otherwise, verbal derivation would be overlooked, which may have an effect on the predictability power of the paradigm. Analogy stands as another variable in word creation which is easily detected when a word-formation paradigm is laid out. This paper identifies as analogical those formations which look like complex words, that is, there seems to be a syntagmatic relation with at least another formation in the same morphologically-related set, while, in fact, they seem to come about as 10  One suffixed -scope related verb was found in the OED (i.e., microscopize), but is marked as obsolete.

236

Díaz-Negrillo

a result of paradigmatic pressure from other formations in the same semantic category. This is because the base form is non-existent and, therefore, missing in the paradigm, e.g., geoscopic, which lacks a related geoscope, or basiscopic, which lacks a related basiscope. This is a serious problem for morpheme-based views of morphology firmly grounded on directionality. In terms of realizations, a case in point is that of -y formations. In the previous section, we have argued that -y event formations are of two kinds. There are formations which parallel an instrument formation in their respective sets, and therefore seem to be derived from them, and others which do not seem to follow a derivational path. For economy reasons, because of the close semantics and because, ultimately, we believe that morphological realizations stand at the background in paradigm-based approaches, we have decided to propose a single realization for both types, namely -y. However, -scopy could have been proposed for the latter type of formations, identifying as a result event formations with three realizations: -y, -scopy, and -ing. Also in terms of realizations, sometimes there is overabundance of realizations in the same semantic category. Overabundance can be of the kind exhibited by e.g., -ic, -ical, -ial in relation and quality, where lexicalization does not seem to be at play and the formations seem to be largely interchangeable, as marked lexicographically in the OED. These can be interpreted as cases of morphological competition. In other cases, however, the category may consist of formations which are not interchangeable and therefore cannot be considered as rival formations, e.g., -ing and -ed vs. -ic/-ical, quality formations. The latter is not the result of lexicalization but rather a consequence of the granularity of the semantic categories or paradigmatic cells, as discussed just above. As claimed in the Introduction, empirical studies of this kind are very much in need for progress in paradigm-based approaches in word formation. All in all, this has been an attempt at a paradigm-based approach to neoclassical word formation, which may raise issues of discussion and may hopefully clarify others. Admittedly, exploration of left-ward word formation, including recursive compounding and its interaction with derivation, is to be explored for a complete picture of neo-classical word formation. References Adams, V. (2001). Complex words in English. Harlow: Pearson. Amiot, D., & Dal, G. (2007). Integrating neoclassical combining forms into a lexemebased morphology. In G. Booij, B. Fradin, A. Ralli, & S. Scalise (Eds.), Online proceedings of the Fifth Mediterranean Morphology Meeting (MMM5). University of Bologna.

Neoclassical Word Formation in English

237

Baeskow, H. (2004). Lexical properties of selected non-native morphemes of English. Tübingen: Gunter Narr. Bagasheva, A. (2017). Comparative semantic concepts in affixation. In J. SantanaLario & S. Valera (Eds.), Competing patterns in English affixation (pp. 33–65). Bern: Peter Lang. Bagasheva, A. (this volume). Paradigmaticity in compounding. In J. FernándezDomínguez, A. Bagasheva, & C. Lara-Clares (Eds.), Paradigmatic relations in word formation (pp. 21–48). Leiden: Koninklijke Brill NV. Bauer, L. (1983). English word-formation. Cambridge: Cambridge University Press. Bauer, L. (1997). Derivational paradigms. In G. Booij & J. van Marle (Eds.), Yearbook of Morphology 1996 (pp. 243–256). Dordrecht: Kluwer. DOI: 10.1007/978-94-017-3718-0_13 [Accessed 25 August, 2019]. Bauer, L. (1998). Is there a class of neoclassical compounds, and if so is it productive? Linguistics, 36, 403–422. DOI: 10.1515/ling.1998.36.3.403 [Accessed 25 August, 2019]. Bauer, L., & Huddleston, R. (2002). Lexical word-formation. In R. Huddleston & G. K. Pullum (Eds.), The Cambridge grammar of the English language (pp. 1621–1721). Cambridge: Cambridge University Press. DOI: 10.1017/9781316423530.020 [Accessed 25 August, 2019]. Bauer, L., Lieber, R., & Plag, I. (2013). The Oxford reference guide to English morphology. Oxford: Oxford University Press. Bonami, O., & Strnadová, J. (2019). Paradigm structure and predictability in derivational morphology. Morphology, 29, 167–197. DOI: 10.1007/s11525-018-9322-6 [Accessed 25 August, 2019]. Booij, G. (2002). The morphology of Dutch. Oxford: Oxford University Press. Booij, G. (2007). Paradigmatic morphology. In B. Fradin & G. Dal (Eds.), La raison morphologique. Hommage à la mémoire de Danielle Corbin (pp. 29–38). Amsterdam, Philadelphia: John Benjamins. Díaz-Negrillo, A. (2014). Neoclassical compounds and final combining forms in English. Linguistik Online, 68, 3–20. Available at: https://bop.unibe.ch/linguistik-online/ article/view/1631/2754 [Accessed 25 August, 2019]. Fradin, B. (this volume). Characterizing derivational paradigms. In J. FernándezDomínguez, A. Bagasheva, & C. Lara-Clares (Eds.), Paradigmatic relations in word formation (pp. 49–84). Leiden: Koninklijke Brill NV. Kastovsky, D. (2009). Astronaut, astrology, astrophysics: About combining forms, classical compounds and affixoids. In R. W. McConchie, A. Honkapohja, & J. Tyrkkö (Eds.), Selected proceedings of the 2008 Symposium on New Approaches in English Historical Lexis (HEL-LEX 2) (pp. 1–13). Somerville, MA: Cascadilla Proceedings Projects. Lasserre, M. (2015). What are you afraid of? The construction of meaning in X-(o)phobie lexemes. Acta Linguistica Hungarica, 62(4), 477–495. DOI: 10.1556/064.2015.62.4.5 [Accessed 25 August, 2019].

238

Díaz-Negrillo

Levi, J. N. (1978). The syntax and semantics of complex nominals. New York: Academic Press. Litta, E., & Budasi, M. (this volume). What we talk about when we talk about paradigms: representing Latin word formation. In J. Fernández-Domínguez, A. Bagasheva, & C. Lara-Clares (Eds.), Paradigmatic relations in word formation (pp. 128–163). Leiden: Koninklijke Brill NV. Lüdeling, A., & Evert, S. (2005). The emergence of productive non-medical -itis. Corpus evidence and qualitative analysis. In S. Kepser & M. Reis (Eds.), Linguistic evidence. Empirical, theoretical, and computational perspectives (pp. 351–370). Berlin: Mouton de Gruyter. McCray, A. T., Browne, A. C., & Moore, D. L. (1988). The semantic structure of neoclassical compounds. In R. A. Greenes (Ed.), SCAMC’88—Proceedings of the 12th Annual Symposium on Computer Applications in Medical Care (pp. 165–168). Washington & New York: EEE Computer Society Press. OED3: Oxford English Dictionary, 3rd edition. Oxford University Press. https://www .oed.com/ [Accessed 18 August, 2019]. Plag, I. (2018). Word-formation in English (2nd ed.). Cambridge: Cambridge University Press. Pounder, A. (2000). Processes and paradigms in word-formation morphology. Berlin: De Gruyter. Prćić, T. (2008). Suffixes vs. final combining forms in English: A lexicographic perspective. International Journal of Lexicography, 21(1), 1–22. DOI: 10.1093/ijl/ecm038 [Accessed 25 August, 2019]. Radimský, J. (this volume). A paradigmatic approach to compounding. In J. FernándezDomínguez, A. Bagasheva, & C. Lara-Clares (Eds.), Paradigmatic relations in word formation (pp. 164–185). Leiden: Koninklijke Brill NV. Scalise, S. (1986). Generative morphology. Dordrecht: Foris. Štekauer, P. (1996). The theory of conversion in English. Bern: Peter Lang. Štekauer, P. (2014). Derivational paradigms. In R. Lieber & P. Šteakauer (Eds.), The Oxford handbook of derivational morphology (pp. 354–369). Oxford: Oxford University Press. DOI: 10.1093/oxfordhb/9780199641642.013.0020 [Accessed 25 August, 2019]. Štekauer, P., Bagasheva, A., & Körtvélyessy, L. (Eds.). (2020). Derivational networks in European languages. Berlin: De Gruyter. ten Hacken, P. (1994). Defining morphology: A principled approach to determining the boundaries of compounding, derivation, and inflection. Hildesheim: Olms. ten Hacken, P. (2012). Neoclassical word formation in English and the organization of the lexicon. In Z. Gavriilidou, A. Efthymiou, E. Thomadaki, & P. Kambakis-Vougiouklis (Eds.), Selected papers of the 10th International Conference on Greek Linguistics (pp. 78–88). Komotini: Democritus University of Thrace.

239

Neoclassical Word Formation in English



Appendix 1. [X -scope] Formations

ENTITY– INSTRUMENT

aeroscope aethrioscope anamorphoscope anemoscope angioscope anomaloscope anorthoscope aphengescope arthroscope astroscope auriscope baroscope bioscope bronchoscope cardioscope chromascope chronoscope cinematoscope colonoscope colposcope conchoscope cryoscope cystoscope diaphanoscope dichroscope dissocioscope

MODE OF ACTION– INSTRUMENT

PURPOSE– AGENT– PLACE– INSTRUMENT INSTRUMENT INSTRUMENT

autoscope

ceraunoscope chromatroposcope

cycloscope debusscope diascope dipleidoscope

240

Díaz-Negrillo

(cont.)

ENTITY– INSTRUMENT

dromoscope dynamoscope ebullioscope electroscope embryoscope endoscope engyscope erythroscope fetoscope

floriscope fluoroscope galvanoscope gasoscope gastroscope glottiscope gonioscope graphoscope gyroscope haematoscope hagioscope helioscope heliochromoscope hodoscope hygroscope hypnoscope iconoscope inductoscope introscope iridoscope iriscope isoscope

MODE OF ACTION– INSTRUMENT

PURPOSE– AGENT– PLACE– INSTRUMENT INSTRUMENT INSTRUMENT

episcope

fibrescope

finderscope

hyposcope

Neoclassical Word Formation in English

241

(cont.)

ENTITY– INSTRUMENT

keratoscope kinesiscope kinetoscope kinescope koniscope lactoscope laparoscope laryngoscope leucoscope lithoscope lychnoscope magnascope meatoscope mediastinoscope megaloscopec megascope melanoscope microbioscope microscope monoscope mutoscope nepheloscope nephoscope odoroscope oesophagoscope ophthalmoscope oscilloscope otoscope pantoscope pedoscope periscope peritoneoscope phacoscope

MODE OF ACTION– INSTRUMENT

myrioscope

otheoscope

PURPOSE– AGENT– PLACE– INSTRUMENT INSTRUMENT INSTRUMENT

242

Díaz-Negrillo

(cont.)

ENTITY– INSTRUMENT

MODE OF ACTION– INSTRUMENT

phantascope pharyngoscope phenakistoscope phoneidoscope phonoscope

phoroscope

phosphoroscope photoscope pioscope pluvioscope pneoscope polariscope

poroscope praxinoscope proctoscope pseudoscope psychoscope pyroscope radarscope radioscope rectoscope

retinoscope rheoscope rhinoscope rotoscope schistoscope scintilloscope scleroscope scotoscope seismoscope

PURPOSE– AGENT– PLACE– INSTRUMENT INSTRUMENT INSTRUMENT

polemoscope

polyscope

reflectoscope

roentgenoscope

resectoscope

243

Neoclassical Word Formation in English (cont.)

ENTITY– INSTRUMENT

MODE OF ACTION– INSTRUMENT

selenoscope sideroscope sigmoidoscope skiascope

spectroscope sphygmoscope stalagmoscope statoscope stauroscope stereoscope stethoscope stomatoscope synchroscope

telescope thermoscope thoracoscope toposcope trichinoscope urethroscope vaginoscope vibroscope videoscope vitascope volumescope zymoscope

PURPOSE– AGENT– PLACE– INSTRUMENT INSTRUMENT INSTRUMENT

sniperscope snooperscope

Stanhoscope

stroboscope tachistoscope tachyscope teinoscope

trenchscope

vectorscope

vortoscope

244

Díaz-Negrillo



Appendix 2. [-scope X] Formations

INSTRUMENT EVENT -y

-ing

AGENT

EXPER.

RELATION

-ist

-ist

-ic

-ical

abdominoscopy aeroscope

aeroscopy

aeroscopic

aethrioscope anamorphoscope anemoscope angioscope anomaloscope anorthoscope

anorthoscopic anthroposcopy

aphengescope arthroscope

arthroscopy

arthroscopist

arthroscopic

astroscope

astroscopy

auriscope

auriscopy

autoscope

autoscopy

autoscopic

bacterioscopy

bacterioscopic

baroscope

baroscopic baroscopical

bioscope

bioscopic

bronchoscope bronchoscopy

bronchoscopic

cardioscope ceraunoscope cerebroscopy

bronchoscopical

-al

245

Neoclassical Word Formation in English

QUALITY -ic

-ical

-ed

-ing

POSS.

MEANS

MANNER/ RESPECT

ACTION

-ed

-ly

-ly

conv.

anorthoscopically

arthroscopically

bacterioscopically

basiscopic

246

Díaz-Negrillo

(cont.) INSTRUMENT EVENT -y

-ing

AGENT

EXPER.

RELATION

-ist

-ist

-ic

-ical

chromascope chromatoscopy chromatroposcope chronoscope

chronoscopy

chronoscopic

colonoscope

colonoscopy

colonoscopic

colposcope

colposcopy

cinematoscope

conchoscope cranioscopy cryoscope

cranioscopist

cranioscopic cranioscopical

cryoscopy

cryoscopic

cystoscopy

cystoscopic

cycloscope cystoscope

dactyloscopy debusscope

deuteroscopy diaphanoscope

deuteroscopic

diaphanoscopy

diascope dichroscope

dichroscopic

dipleidoscope dissocioscope dromoscope dynamoscopy

-al

247

Neoclassical Word Formation in English

QUALITY -ic

-ical

-ed

-ing

POSS.

MEANS

MANNER/ RESPECT

ACTION

-ed

-ly

-ly

conv.

colonoscopically

cryoscopically cystoscope

dermatoscopic

diascopic

248

Díaz-Negrillo

(cont.) INSTRUMENT EVENT -y

-ing

AGENT

EXPER.

RELATION

-ist

-ist

-ic

dynamoscope ebullioscope

ebullioscopy

ebullioscopic

electroscope

electroscopic

embryoscope embryoscopy

embryoscopic

endoscope

endoscopic

endoscopy

engyscope enteroscopy episcope erythroscope

fetoscope

fetoscopy

fibrescope finderscope floriscope fluoroscope

fluoroscopy

fluoroscopic

fundoscopy

fundoscopic

galvanoscope galvanoscopy

galvanoscopic

gasoscope gastroscope

gastroscopy

gastroscopist

gastroscopic geoscopic

glossoscopy glottiscope gonioscope

gonioscopy

gonioscopic

graphoscope gyroscope

gyroscopic

-ical

-al

249

Neoclassical Word Formation in English

QUALITY -ic

-ical

-ed

-ing

POSS.

MEANS

MANNER/ RESPECT

ACTION

-ed

-ly

-ly

conv.

electroscopically

endoscopically

esoscopic exoscopic

fluoroscopically

gastroscopically

gyroscopically

gyroscopically

250

Díaz-Negrillo

(cont.) INSTRUMENT EVENT -y

-ing

AGENT

EXPER.

RELATION

-ist

-ist

-ic

-ical

haematoscope haematoscopy hagioscope helioscope

helioscopy

hepatoscopist

helioscopic

heliochromoscope hepatoscopy hieroscopy hodoscope hydroscopist hygroscope

hygroscopy

hygroscopic hygroscopical

hypnoscope hyposcope iconoscope inductoscope inoscopy introscope iridoscope iriscope isoscope keratoscope

keratoscopy

kinesiscope kinetoscope

kinetoscopic

kinescope

kinescoping

koniscope koroscopy lactoscope laparoscope

laparoscopy

laryngoscope laryngoscopy leucoscope

laparoscopist

laparoscopic

laryngoscopist

laryngoscopic

laryngoscopical

-al

251

Neoclassical Word Formation in English

QUALITY -ic

-ical

-ed

-ing

POSS.

MEANS

MANNER/ RESPECT

ACTION

-ed

-ly

-ly

conv.

hagioscopic

hygroscopically

kinescoped

kinescope

laryngoscopically

laryngoscopically

252

Díaz-Negrillo

(cont.) INSTRUMENT EVENT -y

-ing

AGENT

EXPER.

RELATION

-ist

-ist

-ic

-ical

-al

lithoscope lychnoscope magnascope mastoidoscopy meatoscope mediastinoscope

mediastinoscopy

mediastinoscopic

megaloscopec megascope

megascopic

melanoscope metalloscopy

metalloscopic

metoposcopy microbioscope microscope

microscopy microscoping microscopist

microscopic microscopical misoscopist

mixoscopy

mixoscopic

morphoscopy

morphoscopic

monoscope

mutoscope

morphoscopical

mutoscopic

myrioscope necroscopy

necroscopic necroscopical

nepheloscope nephoscope nooscopic

microscopial

253

Neoclassical Word Formation in English

QUALITY -ic

-ical

-ed

-ing

POSS.

MEANS

MANNER/ RESPECT

ACTION

-ed

-ly

-ly

conv.

lychnoscopic macroscopic macroscopical

macroscopically

megascopic megascopical

megascopically

microscopic

microscopically microscopi- microscope cally

mutoscope

254

Díaz-Negrillo

(cont.) INSTRUMENT EVENT -y

-ing

AGENT

EXPER.

RELATION

-ist

-ist

-ic

-ical

odoroscope oesophagoscope

oesophagoscopy

oesophagoscopic

omoplatoscopy omphaloscopy oneiroscopy

oneiroscopist

ooscopy ophthalmoscope

ophthalmoscopy

ophthalmoscopist

ophthalmo- ophthalscopic moscopical

organoscopy orniscopy ornithoscopy oscilloscope

oscilloscopic

otheoscope otoscope

otoscopy

otoscopic

pantoscope pedoscope pelviscopy periscope peritoneoscope

periscopic peritoneoscopy

phacoscope phantascope pharyngoscope phenakistoscope phoneidoscope

pharyngoscopy

periscopical

-al

255

Neoclassical Word Formation in English

QUALITY -ic

-ical

-ed

-ing

POSS.

MEANS

MANNER/ RESPECT

ACTION

-ed

-ly

-ly

conv.

ophthalmoscopically

ophthalmoscope

orthoscopic oscilloscopically

otoscopically pantoscopic

periscopic

phenakistoscopic

periscopical

periscopically

periscopically

periscope

256

Díaz-Negrillo

(cont.) INSTRUMENT EVENT -y

-ing

AGENT

EXPER.

RELATION

-ist

-ist

-ic

phonoscope phoroscope phosphoroscope photoscope

photoscopic phytoscopy

pioscope pluvioscope pneoscope podoscopy polariscope

polariscopy

polariscopist

polariscopic

polemoscope polyscope poroscope

poroscopy

poroscopic

proctoscope

proctoscopy

proctoscopic

pseudoscope

pseudoscopy

pseudoscopic

radioscope

radioscopy

radioscopic

rectoscope

rectoscopy

praxinoscope

psychoscope pyroscope radarscope

reflectoscope resectoscope retinoscope rheoscope

retinoscopy

retinoscopist

retinoscopic rheoscopic

-ical

-al

257

Neoclassical Word Formation in English

QUALITY -ic

-ical

-ed

-ing

POSS.

MEANS

MANNER/ RESPECT

ACTION

-ed

-ly

-ly

conv.

platyscopic

polariscopically

pseudoscopic

pseudoscopically

retinoscopically rheoscopic

258

Díaz-Negrillo

(cont.) INSTRUMENT EVENT -y

-ing

AGENT

EXPER.

RELATION

-ist

-ist

-ic

rhinoscope

rhinoscopy

rhinoscopic

roentgenoscope

roentgenoscopy

roentgenoscopic

rotoscope

-ical

rotoscoping scatoscopy

schistoscope scintilloscope scleroscope

scleroscopic

scotoscope seismoscope

seismoscopic

selenoscope sideroscope sigmoidoscope sigmoidoscopy skiascope

sigmoidoscopist

skiascopy

sigmoidoscopic skiascopic

sniperscope snooperscope spectroscope spectroscopy

spectroscopist

spectroscopic

spectroscopical

sphygmoscope stalagmoscope Stanhoscope statoscope stauroscope

stauroscopic stethoscopy

stereoscope

stereoscopy

stethoscope

stereoscopist

stereoscopic

stethoscopist

stethoscopic stethoscopical

stomatoscope strioscopy

strioscopic

-al

259

Neoclassical Word Formation in English

QUALITY -ic

-ical

-ed

-ing

POSS.

MEANS

MANNER/ RESPECT

ACTION

-ed

-ly

-ly

conv.

rhinoscopically rhinoscopically roentgenoscopically

roentgenoscope

rotoscoped

rotoscope

sigmoidoscopically

sigmoidoscope

skiascopically

spectroscopically

spectroscop- spectroically scope

stauroscopically stereoscopic

stereoscopically stereoscopically stethoscoped stethoscopically

stethoscope

260

Díaz-Negrillo

(cont.) INSTRUMENT EVENT -y stroboscope

-ing

AGENT

EXPER.

RELATION

-ist

-ist

-ic

stroboscopy

-ical

stroboscopic stroboscopical

synchroscope tachistoscope

tachistoscopic

tachyscope teinoscope telescope

telescopy

telescoping

telescopist

telescopic

telescopical

thermoscopic

thermoscopical

teratoscopy thermoscope thoracoscope thoracoscopy toposcope

toposcopy tracheoscopy

toposcopic tracheoscopist

tracheoscopic

trenchscope trichinoscope uranoscopy

uranoscopic

urethroscope urethroscopy urinoscopy

urethroscopic urinoscopist

urinoscopic

vaginoscope vectorscope vibroscope

vibroscopic

videoscope vitascope volumescope vortoscope zooscopy zymoscope

zooscopic

-al

261

Neoclassical Word Formation in English

QUALITY -ic

-ical

-ed

-ing

POSS.

MEANS

MANNER/ RESPECT

ACTION

-ed

-ly

-ly

conv.

stroboscopically

tachistoscopically

telescopic

telescoping telescoped

telescopically thermoscopically

uranoscopic

zooscopic

telescopically telescope

Index action 4, 51, 56–59, 63–66, 69–77, 88–89, 105, 134–136, 147, 149–152, 158–163, 168, 169n, 172–173, 223–228, 233–235 activity 31, 33, 60, 62–64, 68–78, 93, 228 adjective 32–34, 38, 58, 61, 77, 89–93, 103, 133, 135, 137, 140, 142, 145–148, 168, 172, 176–177, 186–188, 191, 193, 195–196, 198–204, 207, 216, 230–232, 235 parasynthetic 21, 25, 28, 32, 34, 38–39, 41. See also compounding (adjectival); lexical base (adjectival) affix see prefix; suffix agent 4, 11, 23, 34, 39, 51–52, 56–59, 65–71, 74–77, 93–94, 134–135, 168, 169n, 172, 216, 222, 225–227, 229 analogy 13, 21, 25, 27–30, 32, 36–39, 41, 91, 96–97, 132n2, 132, 202, 204, 221n1, 229–231, 235 Arabic 1oo–119 Aronoff, Mark 4, 11, 22–23, 101–103 back-derivation 8, 12, 33, 129, 131, 217 base see lexical base Bauer, Laurie 3, 6–7, 10–11, 28, 86, 165, 188, 215, 218 Becker, Thomas 8, 25–26 blocking see restriction Bonami, Olivier 9–10, 23, 50–51, 165–168, 215 Booij, Geert 9, 25, 49, 67, 93, 136–137, 164, 168, 173, 217 British National Corpus 38n, 174, 205 Bybee, Joan 3, 36, 52 cell see meaning (predictability) clipping 191–192, 200–203, 207 competition 11, 28, 102, 180, 236 overabundance in 50, 101, 236 compounding 40, 61, 139–140, 143, 164, 169, 172–173, 182 adjectival 34, 38 learned 15, 175, 178–180, 218 left-headed 177–178, 180 nominal 38, 91, 170–171, 180–181

neoclassical 213–236 parasynthetic 34 processing of 22, 28, 40–41 pseudo- 31, 181–182 right-headed 177–179 root 32, 91, 170–175 synthetic 21, 32–34 verbal 25, 28–34, 36–38, 41, 140 conceptualization 27, 31, 37, 89, 213–214, 216, 217, 219 constraint see restriction construction 14–15, 29, 35–38, 66, 68, 70–71, 74, 130, 136–146, 148–149, 154, 164, 168–169, 172–173, 181–182, 188–189, 202, 230, 232 Construction Morphology see construction conversion 8, 11, 32–33, 146–147, 199, 201, 203, 220–221, 233, 235 Czech 63–64, 86–98, 128n declension see under inflection derivation order of 5–6, 26, 51, 85, 167 vs. inflection 3–4, 10–13, 22–24, 35–36, 50, 53, 87–88, 93, 102–106, 166, 169, 215 de Saussure, Ferdinand 1–2, 12, 25 diminutive 23, 60–61, 88, 90, 92, 94–96, 134–135, 187, 190, 192–194, 199, 204 Dokulil, Miloš 85–91, 97 Dutch 95, 186, 205, 216 ellipsis 191–192, 195, 200–202, 206–207, 218 English 3, 13, 31–32, 37–38, 61–63, 68, 93, 96–97, 101, 110, 114–115, 175–176, 186–190, 193–194, 198, 215, 218–220, 222 entity 56, 67, 77, 91, 217, 222–225, 229n1, 232–234 event 27, 30, 51, 57–59, 62–63, 66–69, 72, 77, 217, 227–234, 236 female 56–57, 60, 73, 86, 88, 90–92, 94–95, 100, 106, 150–152, 159, 161, 206 French 60–62, 68, 70–71, 72n, 74, 166–168, 175–177, 179, 181–182, 218–219

264 Gagné, Christina 9, 22, 28 gap 4–7, 11–12, 29, 35, 39–40, 65, 130, 204, 215, 228–229. See also word (existing, potential) German 63, 128n, 176, 186, 218–219 grammaticalization 140, 171–172. See also inflection Greek 217, 220 Hathout, Nabil 52, 167 head see under compounding (left-headed; right-headed) Hebrew 105, 109, 116 incorporation see compounding (verbal) inflection 7, 28, 36, 51, 53, 56, 101–103, 131–132, 166, 169, 171, 188, 215–216 conjugation 63, 104, 109, 112, 131, 137, 215 declension 50, 53, 63, 137, 147, 215 paradigm  3–4, 10–13, 22–24, 35, 50, 52, 63, 70, 76, 87–88, 92–93, 96, 103–106, 109–114, 119 word form 50, 52, 54, 66n1. See also derivation (vs. inflection) instrument 23, 56, 58, 65–70, 75–76, 134–135, 172, 175, 220, 222–234, 236, 239–260 Italian 128, 174–175, 177–181 Latin 3, 128–155, 178, 217, 220 lexeme see word lexical base 3–5, 8, 12, 23–24, 28, 34, 61, 86–89, 92, 110–111, 114–115, 118, 128–136, 138, 140–141, 143–145, 150–154, 165, 175, 182, 187, 189, 213–214, 231, 235–236 adjectival 32, 92, 189, 191–192, 195–196, 199–207 nominal 32, 61–62, 76, 94–95, 168–170, 193 non-native 110, 215, 217–218, 220 onomasiological 89–91, 94–95, 97, 176 verbal 33–34, 77 lexicon 3, 9, 13, 22, 27–28, 33, 53, 112, 128–130, 138, 145, 153, 172, 175, 177, 214 Lieber, Rochelle 30, 32–33 location 23, 27–28, 58, 68–69, 77, 216, 223–224, 234

Index male 57, 60, 73, 88, 91–92, 95, 206, 216 manner 30, 74, 92, 150, 161, 224–225, 227, 232–235 meaning  3–4, 23, 29–30, 57, 60–61, 88–92, 100–101, 136–138, 173–174, 179, 188–189, 192–199, 218, 220–221, 234–235 compositionality 9, 25, 29, 196, 202, 208, 222 connotational 187–188, 193, 196, 203–204, 206–207 denotational 102, 193–194, 196, 198, 201, 203–204, 207 extension 23, 36, 60, 189, 194–195, 198, 218–221, 229n2, 235 predictability 6–7, 9–10, 13–15, 21, 27, 29, 62n, 85, 87–88, 91–92, 94, 97–98, 130–131, 168, 207, 218 shift 36, 96–97, 138, 145–146, 166. See also semantics; word (potential) naming unit see lexeme noun 8, 30–33, 52, 61–62, 89–90, 94, 96, 133–135, 139–140, 147, 153, 168, 172–174, 180–181, 187–188, 191–192, 195, 199–202, 204, 207, 217, 222, 228–229, 235 countable 65, 73–74, 90, 92 mass 72–73, 77, 88, 99 nominalization 71, 92–93, 186–208, 216. See also compounding (nominal); lexical base (nominal) onomasiological approach 26, 85, 165, 175, 198, 202 onomasiological category 33, 77n, 85–86, 88–92, 94–98, 176 Oxford English Dictionary 38, 186, 190, 196, 198, 200, 202, 205, 214, 220–222, 228, 231, 233, 236 pejorative 15, 188, 190–192, 196–199, 202–204, 206–208 phonology 8, 54–55, 103, 107, 110, 112, 119, 137, 145, 186–187, 197, 200, 203, 219 polysemy 94–95, 189, 195, 199–201, 204, 206–207, 227, 231–232, 235 pragmatics 30, 32, 35, 40, 90, 96, 103, 189, 192–194, 197, 206

Index

265

prefix 54, 88, 104, 109–110, 118, 128, 131–136, 144–145, 148, 153, 220–222, 235 productivity 6, 8, 25, 27, 32, 94–96, 110, 112, 115, 117, 147, 194–195, 218 purpose 89, 222, 225

transparency 14, 21, 26, 86, 97, 101, 104, 109–111, 114, 116–118, 140, 225 transposition 56, 77n, 85–86, 88–89, 91–94, 96, 204. See also onomasiological category

quality 4, 38, 77, 89, 186, 193, 215, 224, 227, 231–232, 234–236

van Marle, Jaap 2, 8, 12, 165, 168 verb 8, 25, 37, 51–52, 56–61, 62–63, 65–71, 76–77, 89, 93–94, 100–119, 129, 131–135, 138–140, 142–143, 145, 147, 168, 173, 200, 207, 216, 228n, 233, 235 intransitive 101, 104–105, 114–119, 172 transitive 104–106, 114–119. See also compounding (verbal); inflection (conjugation); lexical base (verbal)

restriction 28, 33–34, 38, 52n2, 67, 96–97, 100–101, 106–107, 116–119, 133, 138, 153, 175–177, 190, 194–195, 204, 208 result 23, 65, 70, 231–232 rivalry see competition root 3–6, 32, 55, 103, 139–140, 200, 216–217 Russian 50–51, 61–63 schema see construction semantic shift see under meaning (shift) semantics 24, 30, 32, 40, 58n1, 61, 132–133, 136, 141, 143–144, 153–154, 188–189, 195, 202, 218 sense see meaning slot see gap Slovak 85–86, 92, 96–97 Štekauer, Pavol 7, 10, 22, 26, 29, 32–33, 63, 85, 88, 97, 130, 165, 198, 228n Stump, Greg 5, 22–23, 34–35, 92, 96 suffix 6, 11, 15, 34, 54, 60n, 89–92, 94–97, 110n1, 131, 134–136, 140, 142, 147–148, 153, 167, 169n, 176, 186–208, 219–221, 227, 235 syntagmatic relation 2, 10, 22–25, 31–32, 35, 194, 235 ten Hacken, Pius 91, 95–96, 217

word complex 4, 7, 9, 22, 24–26, 61n2, 85, 88, 97, 133, 136, 138, 165, 172, 187–188, 235 existing 25, 28, 110, 114–115, 129–130, 136–139, 190 new 4, 6, 11, 14, 25, 27–28, 88–89, 91–92, 97, 130, 136–137, 148, 172, 190, 195, 204 possible 4, 29, 33, 35, 38 potential  4–7, 11, 29–31, 33, 35, 38, 131 simple 60–61, 85, 129, 131–133, 138, 143, 165, 172. See also lexical base; lexicon word form see under inflection word formation directionality in 8, 24, 26, 31, 33, 129–132, 138–139, 144, 147, 153, 204, 213, 217, 236 family 24, 26, 36, 50, 54, 60–61, 71, 75, 77–78, 86, 105, 129–132, 164–167, 170–171 rule 8, 25, 94, 97, 128–129, 138, 146, 153, 188, 194–195, 201, 218