From Language to Discourse : Selected Papers of VII Linguistic Sharing Forum [1 ed.] 9781443867153, 9781443856911

Citation preview

From Language to Discourse

From Language to Discourse: Selected Papers of Linguistic Sharing Forum

Edited by

Clara Nunes Correia (Coordinator), Camile Tanto, Larysa Shotropa, Lúcia Cunha and Noémia Jorge

From Language to Discourse: Selected Papers of Linguistic Sharing Forum, Edited by Clara Nunes Correia (Coordinator), Camile Tanto, Larysa Shotropa, Lúcia Cunha and Noémia Jorge This book first published 2014 Cambridge Scholars Publishing 12 Back Chapman Street, Newcastle upon Tyne, NE6 2XX, UK British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Copyright © 2014 by Clara Nunes Correia (Coordinator), Camile Tanto, Larysa Shotropa, Lúcia Cunha, Noémia Jorge and contributors All rights for this book reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior permission of the copyright owner. ISBN (10): 1-4438-5691-6, ISBN (13): 978-1-4438-5691-1

TABLE OF CONTENTS

Editorial Note ........................................................................................... vii Opening Text What do Cross-population Studies Tell Us about the Role of Grammar in Language Development? ......................................................................... 2 João Costa Articles Contrast Encoding in European Portuguese: The Case of Contrastive Parallelism ................................................................................................. 22 Aida Cardoso Phonological Variation in La Selva Catalan: Dialect Loss or Maintenance? ........................................................................................ 40 Eva Bosch i Roura Child Acquisition of Exhaustivity: European Portuguese ......................... 57 Stéphanie Dias Vaz The Acquisition of the Pronominal System in Portuguese L2 by Basic Education Students in Cape Verde: A Bilingual Experience..................... 83 Ana Josefa Cardoso The Acquisition of the Morphological and Semantic Properties of the Portuguese Pretérito Perfeito and Pretérito Imperfeito by Native Speakers of Cape Verdean Creole........................................... 106 Ana Sofia Fonseca Null Object Features in Brazilian Portuguese.......................................... 128 Mirna Sodré Valverde-Hübner

vi

Table of Contents

Subject-Verb Inversion in Declarative-Exclamative Sentences in the Portuguese Language ..................................................................... 150 Rita Valadas Pereira Terminology Issues in Spanish and Portuguese Phonetics and Phonology: vibrantes or rhotics? ...................................................... 176 Iiris Rennicke Relations between Grammar and Reading: A Study of Anaphor Pronouns in Lower Secondary Education ................................................ 199 Joana Batalha Closing Text Language in Action: Epistemological and Methodological Issues .......... 224 Maria Antónia Coutinho

EDITORIAL NOTE

The Young Researchers’ Group of CLUNL was created in 2006 with the purpose of disseminating and sharing research and recent work or work in progress in the field of Linguistics. Since then, it has annually organized linguistics sharing forums which bring together Master Degree students and holders, as well PhD students from all over the world. The volume that is now being presented, published by Cambridge Scholars Publishing, contains selected texts from the VI and VII Linguistics Sharing Forums, which took place at FCSH/UNL, on the 24th and 25th November 2011 and 23rd November 2012. Out of eleven proposals, nine articles were selected and submitted for consideration by the Academic Committee, whom we would like to thank: Bernadete Abaurrre (Universidade Estadual de Campinas) Nélia Alexandre (Universidade de Lisboa) Cristian Bota (Université de Genève) Hugo Cardoso (Universidade de Macau) Armanda Costa (Universidade de Lisboa) Isabel Falé (Universidade de Lisboa) Alexandra Fiéis (Universidade Nova de Lisboa) Cristina Flores (Universidade do Minho) Aldina Marques (Universidade do Minho) Maria Helena Mira Mateus (Instituto de Linguística Teórica e Computacional) Márcia Nogueira (Universidade Federal de Ceará) Isabel Pereira (Universidade de Coimbra) Luísa Álvares Pereira (Universidade de Aveiro) Raquel Santana Santos (Universidade Estadual de Campinas) The above mentioned articles should be seen as Work in Progress, as they were written by young researchers who are not PhD degree holders. In addition to these texts that are being published here, this book also includes the publication of the articles by João Costa, as the Opening Text, and Maria Antónia Coutinho as the Closing Text.

viii

Editorial Note

We would like to express our sincere thanks to the Linguistics Centre of the Universidade Nova de Lisboa and the Faculdade de Ciências Sociais e Humanas for its on-going support for the initiatives carried out by the Young Researchers Group. The Editors

OPENING TEXT

WHAT DO CROSSPOPULATION STUDIES TELL US ABOUT THE ROLE OF GRAMMAR IN LANGUAGE DEVELOPMENT?1 JOÃO COSTA CENTRO DE LINGUÍSTICA DA UNIVERSIDADE NOVA DE LISBOA FACULDADE DE CIÊNCIAS SOCIAIS E HUMANAS/UNIVERSIDADE NOVA DE LISBOA

Abstract The need for grammatical knowledge in the explanation of language growth is argued for on the basis of crosspopulation results on the production and comprehension of subject and object relative clauses. It is argued that approaches dispensing with grammatical primitives fail to account for subtle differences, reported in several experiments, in performance between typically developing children, adults, children diagnosed with Specific Language Impairment and agrammatic adults. The studies reported in this paper replicate, for European Portuguese, findings that N. Friedmann and collaborators obtained for Hebrew, reinforcing the crosslinguistic robustness of certain aspects of syntactic development.

1. Introduction Traditional and recent debates on language acquisition discuss the interplay between memory, processing constraints and language as an autonomous model in the development of language. 1

This paper summarizes results of several papers and dissertations developed under the projects “Syntactic Dependencies from 3 to 10” (PTDC/CLELIN/099802/2008) and “Crosspopulation and Crosslinguistic Approaches to the Acquisition of Dependencies” (PTDC/MHC-LIN/4812/2012) funded by Fundação para a Ciência e Tecnologia, I.P. I would like to thank the members of the two Project teams for their collaboration.

João Costa

3

Whereas innatist approaches contend that the child develops language because she is endowed with a language faculty, and with language particular knowledge of language, other approaches – such as the emergentist approach advocated by Bates and MacWhiney (1988) and O’Grady (2001), among others – argue that language development is best explained without any reference to linguistic or grammatical tools. The goal of this paper is to come back to this debate by showing what recent results on comparing performance on similar tests and constructions by different populations can reveal about this debate. The main line of argumentation is the following: crosspopulation studies enable us to pin down the role played by grammar and by general processing constraints. It will be shown that it is very difficult to dispense with grammatical tools in accounting for the differential behavior of the populations at stake. The paper is structured as follows: In section 2, I outline the basis of the debate. In section 3, I present the summary of results of several studies on the production and comprehension of subject and object relative clauses by different populations: typically developing children, children diagnosed with Specific Language Impairment, hearing impaired children, agrammatic adults and adults without any pathology. Section 4 discusses the results and presents the main conclusions of the paper.

2. Emergentism and Nativism: the role of language As mentioned in the introduction, the literature on language development from the last decades debates the role of processing and grammar in explaining children’s linguistic behavior. A fair evaluation of the role played by processing constraints is crucial to make sure that grammatical knowledge is needed for the explanation of language growth. The alternative is to show that it follows from other mechanisms that are needed anyway. If one were to study only L1 acquisition, the illusion might be created that a certain condition in which children deviate from adults is necessarily explained in terms of language growth. Obviously, differences may also be explained on different grounds, and processing resources are a relevant path for exploring existing differences. In short, one needs to know what the infant’s head is like: does it contain a module with grammatical primitives or are these illusions (to use O’Grady’s (2013) terms) reflecting general processing constraints or the effect of parsing routines? If the latter is true, grammar can be dispensed

4

What do Cross-population Studies Tell Us

with, but it must be shown that language development can operate without language primitives. In Costa (2013), I put forward some arguments against the view that language and language growth are the happy combination of independently attested mechanisms. Just to mention one of the arguments advanced in that paper, we can consider cases of development of syntax without any obvious impact on the meaning of the utterances created. It is well known that languages differ with respect to the relative order between inflected verbs and adverbs (Emonds 1978, Pollock 1989). Languages like English exhibit the order Subj- Adv –V –O (as in “John completely finished his work”, contrasting with “*John finished completely his work”), whereas languages like Portuguese display the reverse order between the verb and the adverb (as in “O João terminou completamente o seu trabalho”, contrasting with “*O João completamente terminou o seu trabalho”). The standard analysis of these word order contrasts suggests that the relevant difference between the two language groups is that, in languages like Portuguese, French or Spanish, the verb moves to the inflectional domain, whereas this movement is absent in English, as depicted in (1): (1) Portuguese: [ Vi I [VP Adv [VP ti Obj English: [ I [VP Adv [VP V Obj This is a case of a syntactic operation without any impact on meaning. This type of movement has no consequences whatsoever on how the sentences are interpreted, and when language diachrony is studied, there are changes regarding verb movement that may correlate with verbal inflection, but have nothing to do with sound-meaning pairing. This is interesting, since it is a case of pure syntactic movement. It is very difficult to find any kind of semantic motivation for this operation. It is even more difficult to find any type of language-independent constraint modulating or regulating this type of robust and systematic crosslinguistic difference. Interestingly, children master this type of syntactic difference from very early on, as it is amply demonstrated in the literature. For instances, Guasti (1993/1994) shows that children acquiring Italian, in their earliest utterances in which it is possible to diagnose verb movement, do move the verb to I. This indicates that this parametric difference is set very early, and in the absence of any semantic evidence to do so.

João Costa

5

The literature on language development is abundant on other relevant examples of this type: for instances, children acquiring V2 languages place the verb in the C-domain whenever it is inflected, from very early on, showing that they know not only the correlation between finiteness and Ito-C movement, but also the necessary operations to place the verb in the second position – again a purely syntactic operation. Just to give another example, in a series of experiments we have been conducting we show that children acquiring Portuguese produce sentences with null complements in a target-like manner at the age of 3, but they overuse them, producing null objects in contexts in which they are not legitimate, such as reflexive contexts (Costa and Lobo 2006, Costa and Lobo 2007). That is, not only did they produce target-like sentences with null objects as in (2), but they also overused them in reflexive contexts, as in (3): (2) A:E o João? And the João “What about João?” B:Não vi Ø hoje. Not saw today “I didn’t see him today.” (3) A:E o João? And the João “What about João?” B:ªNão *(se) lavou hoje. Not se washed today “He didn’t wash himself today.” Interestingly, in other languages, children omit complements, but there is robust evidence to say that the type of omission is not uniform (Grüter 2006, among others). This reveals that children have the means to find out, from very early on, that null categories exist, that they are an option in some languages only, and that their behavior is not uniform – there is an array of empty categories with different properties, and children are able to distinguish them very early on.2 2

In Costa and Lobo (2013) and Costa, Grolla and Lobo (2013) we compare the mastery of empty categories in European Portuguese, Brazilian Portuguese and Japanese, and argue that uniform systems (in which the null categories are all of the same type) are acquired earlier than those in which children must find out which type of category is to be used in each context.

6

What do Cross-population Studies Tell Us

All these examples instantiate cases in which children’s performance appears to be guided by purely linguistic notions – movement, the inherent properties of functional categories, or the properties associated to different types of null categories. These categories are needed for a proper description of these subtle crosslinguistic differences, as extensively argued in the literature, and the fact that children’s behavior is sensitive to such distinctions is a fine piece of evidence in favor of the idea that they are dealing with grammatical categories. It is difficult to conceive that these categories can be derived from statistical learning, probabilities, parsing routines or sound-meaning pairings. If, on the one hand, the idea that grammatical primitives can be dispensed with is not convincing, on the other hand, it is worth discussing which behaviors can be explained by processing constraints. In the next section, I will show that some aspects of children’s performance in the production and comprehension of relative clauses may be explained by processing limitations, although this does not appear to be the case prima facie, but, on the other hand, I will show that this type of explanation is not sufficient.

3. Relative clauses: performance across populations It is well known that there is an asymmetry between extraction from the subject position and extraction from the object position in several types of A-bar dependencies including relative clauses. A vast amount of studies since the 1970s reports that extraction from the object position poses difficulties that subject dependencies do not raise (Adams 1990, Berman 1997, Brown 1972, Correa 1982, Correa 1995, de Villiers et al. 1994, De Vincenzi 1991, Friedmann, Belletti and Rizzi 2009; Friedmann and Novogrodsky 2004; Håkansson and Hansson 2000; McKee et al. 1998; Roth 1984; Sheldon 1974; Tavakolian 1981, Vasconcelos 1991). This subject-object asymmetry in A-bar movement constructions is interesting, since it allows the investigation of what exactly is difficult in these constructions, in particular in their comprehension. At the same time, good performance in subject relatives is another piece of evidence for the early mastery of several aspects of syntax. The first aspect to be noted is that producing and comprehending a relative clause – even a subject relative clause, which is unproblematic – involves having access to the highest nodes of the clausal structure, the CP domain. The fact that children produce and comprehend subject relatives indicates that they are able to use this part of the functional domain. Secondly, one can conclude that movement is not necessarily problematic,

João Costa

7

since movement is involved both in subject and in object dependencies. For these reasons, several authors have hypothesized that the source of the difficulty in the interpretation of object dependencies is the intervention of a noun phrase in between the displaced constituent and its original position, as depicted in (4) (cf. Friedmann, Belletti and Rizzi 2009, Friedmann and Costa 2010, among others): (4) [ Obj [ Subj [ V t

There are several pieces of evidence in favor of the idea that the intervention of a lexical noun phrase is the explanation for the problems with dependencies. Just to mention three of them: object relatives with an antecedent are understood with greater difficulty than free object relatives ((5a) vs. (5b)); D-linked object wh-questions are understood with greater difficulty than bare object wh-questions ((6a) vs. (6b)); dependencies with an intervening lexical subject are more problematic than those with a null subject ((7a) vs. (7b)): (5) a.Show me the man that the boy hugs. b.Show me who the boy hugs. (6) a.Which man does the boy hug? b.Who does the boy hug? (7) a.Este é o rapaz que um homem abraça. This is the boy that a man hugs b.Este é o rapaz que proarb abraçam. This is the boy that pro hug The subject-object asymmetry was tested in the acquisition of relative clauses in European Portuguese in Costa et al. (2009) and Costa, Lobo and Silva (2011), and, similarly to what has been found for other languages, children performed very well both in the production and comprehension of subject relatives, but they were worse with object dependencies. Similar results were obtained by Costa et al. (2013) for object relatives containing a preposition. Altogether, these results appear to confirm the idea that intervention effects induce difficulties in the comprehension of object dependencies, but they do not necessarily tell us whether this difficulty with object dependencies is due to grammatical limitations, processing constraints or both.

8

What do Cross-population Studies Tell Us

In Costa, Lobo and Silva (2011) we tested children aged between 3 and 6 for the production and comprehension of subject and object relatives. Comprehension was assessed with an offline task, a picture selection test.3 The results of this task are summarized in Table 1:

Subject relatives Object relatives

Children 96% 68%

Adults 100% 100%

Table 1: Results of the comprehension task for subject and object relatives. (apud Costa, Lobo and Silva 2011). These results appear to indicate that there is a clear developmental effect in the mastery of object dependencies. In the same study, we show that the same type of asymmetry emerges in production, as the results in Table 2 indicate:

Subject relatives Object relatives

Children 78%

Adults 99%

31%

45.5%

Table 2: Results of the production task for subject and object relatives. (apud Costa, Lobo and Silva 2011) Note that the fact that adults do not produce 100% object relatives is due to the alternative production of passive relatives. Instead of producing sentences like “I’d rather be the boy that grandpa visits”, they sometimes opted for the legitimate alternative “I’d rather be the boy that is visited by grandpa.” Children, on the contrary, did not have legitimate grammatical alternatives when they failed to produce object relatives. Their alternatives were ungrammatical, confirming a development in the acquisition of object dependencies. The mastery of the object dependency then appears to show that there is a developmental path to be followed: children have to learn to deal with this specific type of syntactic configuration. An alternative is to say that the object dependency is longer and imposes higher memory constraints, which would favor a view based on processing limitations. If processing is 3

The reader is referred to the papers cited for details concerning participant information and the structure of the tests.

João Costa

9

the key to explaining the problem with object dependencies, we may expect that the higher processing cost of object relatives emerges with adults as well. In the task we used, this was not the case, however. In order to control for the emergence of difficulties with object relatives in adults, Costa, Grillo and Lobo (2012) tested the comprehension of free and headed object relatives with an online task. Materials similar to those used in the picture selection task were used in a self-paced reading test followed by comprehension questions. In Table 3, the results of this test are presented:

Subject relatives Object relatives

Free 87.2% 90.4%

Headed 77.6% 64%

Table 3: Results of the comprehension task for free and headed subject and object relatives. (apud Costa, Grillo and Lobo 2012) What is interesting about these results is that, just like children do in offline tasks, adults exhibit an asymmetry between free and headed relatives and, more importantly, between subject and object dependencies. Results of this type may signal that the subject-object asymmetry is not a matter subject to development or linguistic growth, but rather a general effect of higher processing costs in object dependencies. If this were true, emergentist approaches could be right, and grammatical knowledge might be dispensed with for explaining these patterns of acquisition. In what follows, I will argue that, in spite of this prima facie evidence, grammatical knowledge is still required, which is confirmed by the results of crosspopulation comparisons. The results to be presented reproduce for European Portuguese results by Naama Friedmann for Hebrew (Friedmann 1994, Friedmann and Novogrodsky 2004, Friedmann and Szterman 2006, Friedmann and Costa 2011). I will present results from the following populations: children with SLI, hearing impaired children and agrammatic adults.

Children with Specific Language Impairment The same production and comprehension tests run on typically developing children and reported above were applied to three groups of children diagnosed with Specific Language Impairment in the studies by Ferreira (2008) and Fonseca (2011). Ferreira (2008) studied a group of 7 children aged between 6;0 and 11;0, and Fonseca (2011) ran the tests on a

10

What do Cross-population Studies Tell Us

group of 6 children aged between 5;4 and 9;5. These children were matched for MLU with the group of typically developing children described above, aged between 3 and 6. Crucially, they are older, and at an age in which typically developing children no longer exhibit problems in the production and comprehension of relative clauses. Let us look at the results of these two studies. Ferreira (2008) found that SLI children perform worse than typically developing children on production and comprehension of both subject and object relatives, but there is still an obvious asymmetry between the two types of relatives, as shown in the following Table s: Subject relatives Object relatives

69.6% 61%

Table 4: Results of the comprehension task for subject and object relatives by SLI children. (apud Ferreira 2008) Subject relatives Object relatives

34.3% 14.3%

Table 5: Results of the production task for subject and object relatives by SLI children. (apud Ferreira 2008) Fonseca (2011) obtained comparable results, although with a better performance on subject relatives, as shown in the following two Table s, summarizing the data from both production and comprehension. These results come from the application of exactly the same tests: Subject relatives Object relatives

95% 72%

Table 6: Results of the comprehension task for subject and object relatives by SLI children. (apud Fonseca 2011) Subject relatives Object relatives

38% 12%

Table 7: Results of the production task for subject and object relatives by SLI children. (apud Fonseca 2011)

João Costa

11

This performance of children diagnosed with SLI is interesting for several reasons. First, we can observe that their performance is comparable to the behavior of typically developing children, but delayed. In other words, children with SLI have the same type of subject-object asymmetry, replicating the performance of typically developing children, but the age at which this asymmetry can be observed is much higher. In this sense, there is no clear qualitative difference between the two populations. The children who participated in the test were screened for IQ and processing capacities with WISC II and did not exhibit any evidence of a delay in their IQ. For this reason, it is possible to say that their performance cannot be linked to general limitations in memory or processing capacities. Since their limitation is purely linguistic, we must conclude that their delay in the production and comprehension of relative clauses is due to difficulties in dealing with grammatical constraints. When the type of intervention configuration, depicted again in (8), is considered, and when we think of Grillo’s (2005) or Friedmann, Belletti and Rizzi’s (2009) hypothesis on why intervention is problematic, the data we are describing makes sense: (8) [ Obj [ Subj [ V t

These authors argue that intervention is problematic because of Relativized Minimality: it is the presence of an intervener sharing grammatical features on the path relevant for the chain created that yields the problems in creating the relevant dependency. The fact that it is the sharing of grammatical features that play a role is a strong indicator that the problem with this type of dependency is grammatical to a great extent. Accordingly, an approach dispensing with grammatical primitives cannot easily accommodate these facts, and will have serious troubles in accounting for the comparison between typically developing children and children diagnosed with Specific Language Impairment.

12

What do Cross-population Studies Tell Us

Hearing impaired children The same tests were run on a population of moderate and severe hearing impaired children in Mangas (2011), whose impairment was diagnosed after they were 1 year old. This population is interesting, since these are children who received late or deficient exposure to input. Similar to the group of children diagnosed with Specific Language Impairment, these children do not have any cognitive problems, and it is not expected that we find language problems that can be attributed to some kind of deficit in processing resources, unless the processor is dependent on linguistic knowledge or is grammatically informed. Mangas (2011) ran the same production and comprehension tests used in the studies referred to above on a group of 6 moderate and severe hearing impaired children aged between 7;4 and 10;10, and compared their performance with age matched controls and with the typically developing children of the study by Costa, Lobo and Silva (2011). Again, the group in the latter study is much younger than the hearing impaired children. In the following Table s, I report Mangas’ (2011) results for production and comprehension by the hearing impaired children and by the age matched controls:

Subject relatives Object relatives

Hearing impaired children 97% 70%

Age matched controls 99% 92%

Table 8: Results of the comprehension task for subject and object relatives for hearing impaired children and their age matched controls. (apud Mangas 2011) Again, we observe a subject-object asymmetry, but only in the hearing impaired informants. The age matched controls perform almost at ceiling levels in both conditions. Since there is no evidence to claim that there is a difference in processing capacities between the two populations, it is appealing to suppose that the two populations differ in their grammatical capacities. This has been proposed in Friedmann and Szterman (2006), who argue that hearing impaired children lack the tools to perform certain movement dependencies. The results from the production task, at first sight, do not fully confirm this hypothesis:

João Costa

Subject relatives Object relatives

13

Hearing impaired children 63% 30%

Age matched controls 97% 73%

Table 9: Results of the production task for subject and object relatives for hearing impaired childrenand their age matched controls. (apud Mangas 2011) As we can see in Table 9, hearing impaired children have serious problems in the formation of relative clauses in both conditions, but the subject-object asymmetry is still present. This is similar to what was found for SLI children. Qualitatively, we can compare the strategies used by the two groups when they failed to produce object relatives. The following Table summarizes the types of productions obtained for hearing impaired children in Mangas (2011) and for SLI children in Ferreira (2008):

Target Passive subject relative Subject relative with verb change Fragment DP resumption C-omission Argument omission Role-reversal

Hearing impaired children 30% 10% 5.2% 8.5% 25% 13% __ __

SLI children 14.3% ___ ___ ___ ___ 26.4% 35% 10.3%

Table 9: Types of responses in the object relative condition – production task – Hearing impaired and SLI. (apud Mangas 2011 and Ferreira 2008) Interestingly, we can see that the apparent similarity between the two populations fades a bit when the qualitative data are taken into consideration. Although a much higher sample would be needed in order to confirm such data, we may observe a tendency emerging. In SLI children, most problems have to do with the C-head and the role of the verbal arguments, whereas in hearing impaired children, almost half of the problems come from the realization of the C-head and from resumptive

14

What do Cross-population Studies Tell Us

strategies. Interestingly, resumption is not a strategy for the SLI group in Ferreira (2008). The relevance of resumption in hearing impairment has been independently documented in Friedmann and Szterman (2006) and Friedmann and Costa (2011). Both these studies show that hearing impaired children resort to resumption as a strategy to avoid the creation of a movement dependency. This is consistent with the idea that there may be a generalized problem with movement in hearing impairment, but not in SLI. Arguably, SLI children have difficulties with the identification of the dependency because of Relativized Minimality effects, but not due to difficulties with movement. These differences are interesting and relevant for the current discussion. If this is right, language can be affected in different submodules: movement and theta-role identification. Again, these are clearly grammatical notions – it is very hard to subsume them within the effect of some general type of parsing procedure. Since we are dealing with populations without cognitive impairments, the observation that they experience difficulties in language is per se a very good argument for the autonomy of the language faculty.

Agrammatic adults Let us now consider the performance of agrammatic adults. In what follows I am reporting on the data by Ferreira (2008), who tested Friedmann’s (1994) hypothesis that agrammatic adults have a deficit in syntactic structure through difficulties in accessing the higher nodes of the clausal structure. Again, the same tests for the production and comprehension of subject and object relative clauses were used. Ferreira (2008) assessed 7 informants, all diagnosed with Broca’s aphasia, aged between 29 and 60, with a post-onset time of at least 6 months. The following Table s summarize the results obtained for this group: Subject relatives Object relatives

100% 75%

Table 10: Results of the comprehension task for subject and object relatives by agrammatic adults. (apud Ferreira 2008) As in other populations, there is a subject-object asymmetry, but the overall results are good. The most interesting fact comes from the production experiment:

João Costa

Subject relatives Object relatives

15

46.7% 35%

Table 11: Results of the production task for subject and object relatives by agramatic adults. (apud Ferreira 2008) As shown in Table 11, the subject-object asymmetry vanishes in production. Agrammatic adults experience difficulties in the production of both types of relative clauses. Friedmann (1994) formulates the Tree Pruning Hypothesis, according to which agrammatism yields a failure in accessing the higher nodes of the functional domain. The cut may operate at C level or at the T level, but in either case, there are problems accessing the C-domain, which induces difficulties in the production of relative clauses of both types. As for comprehension, this hypothesis makes no clear predictions, but in general it is not expected that there should be any comprehension problems, since building up the tree is only crucial for sentence formation. Again, this type of evidence is very interesting for the discussion regarding the role of grammar. There are no grounds to claim that agrammatic patients undergo any kind of cognitive impairment. On the contrary, their cognitive skills are spared. The fact that their difficulties are qualitatively different from those found in other populations again indicates that different areas of very specific language aspects may be affected in particular linguistic behaviors, and that these are better understood when grammatical tools are used to describe what is happening in each context.

4. Discussion and conclusions The Table 12.summarizes the results described in this paper, which replicate, for European Portuguese, those obtained in several studies by Naama Friedmann for Hebrew. As can be seen, the crosspopulation results leads to two major conclusions. Firstly, there can be problems in the production of the syntactic dependencies discussed in the paper even in populations for which there are no clear grounds for suggesting that they have any kind of cognitive impairment or processing delay. By itself, this is a good argument for positing the existence of a grammar module responsible for part of the explanation of the different performances.

What do Cross-population Studies Tell Us

16

Secondly, it is clear that the types of difficulties stem from different properties of language. We observe difficulties with movement, problems with generating the functional domain of the clause and troubles with the identification of theta-roles due to intervention configurations, in which the intervener shares features with a displaced constituent. In order to account for these behaviors, linguistic primitives are needed: we need to refer to Relativized Minimality, to the CP domain of the clause, to syntactic movement, or to theta-role assignment. These are language specific notions, and it is hard to conceive how to explain them as side-effects of any independently needed processing or memory constraint. Naturally, the only escape-hatch would be to build a processor that is so linguistically informed that it becomes difficult to argue that grammar is dispensable with. Population Typically developing children Adults SLI children Hearing impaired children Agrammatic adults

Characteristic Subject-object asymmetry. Relativized Minimality induced intervention effects. Subject-object asymmetry in online tasks. Relativized Minimality induced intervention effects. Subject-object asymmetry. Delayed Relativized Minimality induced intervention effects. Subject-object asymmetry. Difficulties in movement dependencies attested by resort to resumption. No subject-object asymmetry in production. Difficulties in access to CP.

Table 12: Summary of results. Concluding, by putting together the results of these studies, I hope to have contributed to strengthening the view that language development is best accounted for if one assumes that our mind is linguistically informed.

References Adams, Catherine. 1990. “Syntactic comprehension in children with expressive language impairment”. British Journal of Disorders of Communication 25, 149-171.

João Costa

17

Bates, Elizabeth and Brian MacWhinney. 1988. “What is Functionalism?” Papers and Reports on Child Language Development 27: 137-52. Berman, Ruth. 1997. “Early acquisition of syntax and discourse in Hebrew”. In Psycholinguistic Studies in Israel: Language Acquisition, Reading and Writing, edited by Y. Shimron, 57-100. Magnes Press (in Hebrew): Jerusalem. Brown, H. Douglas. 1972. “Children’s comprehension of relativized English sentences”. Child Development 42, 1923-1936. Correa, Letícia M. 1982. “Strategies in the acquisition of relative clauses”. In Working Papers of the London Psycholinguistic Research Group, vol. 4, edited by J. Aitchison, and N. Harvey, 37-49. Correa, Letícia M. Sicuro. 1995. “An alternative assessment of children’s comprehension of relative clauses”. Journal of Psycholinguistic Research 24, 183-203. Costa, João. 2013. “The delusion of processing”. In Linguistic Approaches to Bilingualism 3(3), 296-300. Costa, João, Friedmann, Naama, Silva, Carolina, and Maya Yachini. 2013. “The boy that the chef cooked. The acquisition of PP relatives in European Portuguese and Hebrew”, ms. FCSH/Universidade Nova de Lisboa and Tel-Aviv University. Costa, João, and Maria Lobo. 2006. “A aquisição de clíticos em PE: Omissão de Clíticos ou Objectos Nulos?”. In Textos Seleccionados do XXI Encontro Nacional da APL, edited by F. Oliveira, and J. Barbosa, 285-293. Lisboa: APL. —. 2007. “Clitic Omission, null objects or both in the acquisition of European Portuguese?”. In Romance Languages and Linguistic Theory 2005, edited by S. Baauw, F. Drijkonongen, and M. Pinto, 59-72. Amsterdam: John Benjamins. —. 2013. “Crosslinguistic variation in the acquisition of null objects: Portuguese vs. Japanese”. To appear in Developments in the Acquisition of Clitics, edited by K. Grohmann et al. Cambridge: Cambridge Scholars Press. Costa, João, Lobo, Maria, and Carolina Silva. 2009. “Null objects and early pragmatics in the acquisition of European Portuguese”. Probus 21(2), 143-162 —. 2011. “Subject–object asymmetries in the acquisition of Portuguese relative clauses: Adults vs. children”. Lingua 121(6), 1083-1100. Costa, João, Grillo, Nino, and Maria Lobo. 2012. “Minimality beyond lexical restrictions: Processing and Acquisition of free relatives in European Portuguese”. Revue Roumaine de Linguistique/Roumanian Review of Linguistics LVII(2),143-160.

18

What do Cross-population Studies Tell Us

Costa, João, Grolla, Elaine, and Maria Lobo. 2013. “The acquisition of syntactic microvariation in silent categories”. Paper presented at GALA, Oldenburg University. Costa, João, Lobo, Maria, Silva, Carolina, and Elisa Ferreira. 2009. “Produção e compreensão de orações relativas em português europeu: dados de desenvolvimento típico, de PEDL e do agramatismo”. In Textos Seleccionados do XXIV Encontro Nacional da APL, edited by A. Fiéis, and M. A. Coutinho, 211-224. Lisboa: APL. de Villiers, Jill, de Villiers, Peter, and Esme Hoban. 1994. “The central problem of functional categories in the English syntax of oral deaf children”. In Constraints on Language Acquisition: Studies of Atypical children, edited by H. Tager-Flusberg, 9-47. Erlbaum: Hillsdale, NJ. de Vincenzi, Marica. 1991. Syntactic Parsing Strategies in Italian: The Minimal Chain Principle. Kluwer: Dordrecht, The Netherlands. Emonds, Joseph. 1978. “The verbal complex V'-V in French”. Linguistic Inquiry 9, 151-175. Ferreira, Elisa. 2008. “Compreensão e produção de frases relativas por crianças com perturbação específica do desenvolvimento da linguagem e por adultos com agramatismo”. MA diss., Universidade Nova de Lisboa. Fonseca, Ana Rita. 2011. Produção e compreensão de frases relativas por crianças com Perturbação Específica do Desenvolvimento da Linguagem. MA diss., Universidade Nova de Lisboa/Instituto Politécnico de Setúbal. Friedmann, Naama. 1994. “Morphology in Agrammatism: A Dissociation between Tense and Agreement”. MA diss., Tel Aviv University. Friedmann, Naama, and João Costa. 2010. “The child heard a coordinated sentence and wondered. On children's difficulty in understanding coordination and relative clauses with crossing dependencies”. Lingua 120, 1502- 1515. —. 2011. “Last resort and no resort: resumptive pronouns in Hebrew and Palestinian Arabic hearing impairment”. In Resumptive Pronouns at the Interfaces, edited by A. Rouveret, 223-240. Amsterdam/Philadelphia: John Benjamins. Friedmann, Naama, and Rama Novogrodsky. 2004. “The acquisition of relative clause comprehension in Hebrew: a study of SLI and normal development”. Journal of Child Language 31, 661-681. Friedmann, Naama, and Ronit Szterman. 2006. “Syntactic movement in orally-trained children with hearing impairment”. Journal of Deaf Studies and Deaf Education 11, 56-75.

João Costa

19

Friedmann, Naama, Belletti, Adriana, and Luigi Rizzi. 2009. “Relativized relatives: Types of intervention in the acquisition of A-bar dependencies”. Lingua 119, 67-88. Grillo, Nino. 2005. “Minimality effects in agrammatic comprehension”. In Proceedings of ConSOLE XIII, edited by Sylvia Blaho, Eric Schoorlemmer, and Luis Vicente, 106-120. http://www.sole.leidenuniv.nl/. Grüter, Therese. 2006. “Object clitics and null objects in the acquisition of French”. PhD diss., McGill Univ. Montreal. Guasti, Maria Teresa. 1993/1994. “Verb syntax in Italian child grammar: finite and nonfinite verbs”. Language Acquisition 3(1), 1-40. Håkansson, Gisela, and Kristina Hansson. 2000. “Comprehension and production of relative clauses: a comparison between Swedish impaired and unimpaired children”. Journal of Child Language 27, 313-333. MacWhinney, Brian. 1998. “Models of the emergence of language”. Annual Review of Psychology 49, 199-227. Mangas, Vera. 2011. “Produção e compreensão de frases relativas por crianças surdas”. MA Dissertation, Universidade Nova de Lisboa/Instituto Politécnico de Setúbal. McKee, Cecile, McDaniel, Dana, and Jesse Snedeker. 1998. “Relative children say”. Journal of Psycholinguistic Research 27, 573-596. Novogrodsky, Rama, and Naama Friedmann. 2006. “The production of relative clauses in SLI: a window to the nature of the impairment”. Advances in Speech-Language Pathology 8, 364–375. O’Grady, William. 2001. “An Emergentist Approach to Syntax.” http://www.ling.hawaii.edu/faculty/ogrady/. A summary of the detailed arguments for an emergentist theory of syntax found in O’Grady (2005). —. 2013. “The illusion of Language Acquisition”. In Linguistic Approaches to Bilingualism, 3(3), 253-285. Pollock, Jean-Yves. 1989. “Verb Movement, Universal Grammar, and the Structure of IP”. Linguistic Inquiry 20: 365-424. Roth, Froma. 1984. “Accelerating language learning in young children”. Journal of Child Language 11, 89–107 Tavakolian, Susan, 1981. “The conjoined-clause analysis of relative clauses”. In Language Acquisition and Linguistic Theory, edited by S. Tavakolian, 167-187. Cambridge: Cambridge MIT Press. Sheldon, Amy. 1974. “The role of parallel function in the acquisition of relative clauses in English”. Journal of Verbal Learning and Verbal Behavior 13, 272-281.

20

What do Cross-population Studies Tell Us

Vasconcelos, Manuela, 1991. “Compreensão e produção de frases com orações relativas: um estudo experimental com crianças dos três anos e meio aos oito anos e meio”. MA diss., Universidade de Lisboa.

ARTICLES

CONTRAST ENCODING IN EUROPEAN PORTUGUESE: THE CASE OF CONTRASTIVE PARALLELISM AIDA CARDOSO CENTRO DE LINGUÍSTICA DA UNIVERSIDADE DE LISBOA

Abstract This study focuses on the encoding of contrast in European Portuguese (EP), specifically by analyzing a corpus built from a political debate. Taking into consideration the nature of the corpus, the focus of the analysis in a type of structures – contrastive parallelism structures – that are relevant from the discursive point of view, since they are crucial in the construction of an argumentative discourse. Hence the main goal of this paper is to analyze contrastive parallelism structures in EP, discussing how contrast is prosodically encoded in this context and relating it with previous results for other languages. The data shows that contrastive parallelism has specific acoustic properties and that there is no one-to-one relation between pitch accents and these structures. Therefore, such results seem to indicate that the prosodic encoding of such structures is a gradient type. Keywords: Contrast, Parallelism, Prosody, European Portuguese.

1. Introduction Recently, the nature of the prosodic encoding of contrast has received special attention in the literature and the discussion of this topic has been often centered on the gradient or categorical realization of contrast. The study of contrast in different languages also raises the question whether there are, or not, universals in the prosodic encoding of contrast. Moreover, the fact that contrast studies are often centered on the analysis of focus and topic structures points out the relevance of interface studies

Aida Cardoso

23

involving prosody, syntax, semantics, and discourse in achieving a better understanding of contrast and its linguistic marking. In this paper, a specific type of structure is discussed, namely parallelism structures that convey contrast, or contrastive parallelism, as they will be called from now on. As regards the analysis of these structures, the main goal is twofold: on the one hand, to discuss their role as a cohesion mechanism in the construction of an argumentative discourse and, on the other hand, to discuss the prosodic encoding of the semantic-discursive value of contrast. In order to do so, the data analysis will be focused on two main research questions: (i) Which prosodic features are associated with structures of contrastive parallelism? (ii) Is the prosodic encoding of contrastive parallelism a gradient or categorical one? An introduction to the subject of analysis can be found in section 2 of this article, where the relevant theoretical aspects and the empirical findings concerning contrast and parallelism are discussed. In section 3, the methodological steps taken in this study are explained and some light is shed over the relevance of contrastive parallelism in argumentative discourse. In section 4, the results of the prosodic analysis are presented and discussed, followed by a brief conclusion in section 5.

2. Theoretical background 2.1. Contrast Over the past few years, there has been a growing debate about the prosodic realization of contrast. In this matter, it is important to note, first of all, that contrast is often studied with regard to structures of topic and focus and that, in this context, it can have different meanings: contrast can be defined in the literature as a category (e.g. Steedman 2000; Büring 2003), a subtype of focus or topic (e.g. Baumann, Grice and Steindamm 2006), or as a semantic-discursive value associated with topic and focus structures (e.g. Torregrossa 2012), which is the view adopted in this paper. Regarding the prosodic encoding of contrast, studies for different languages have presented data in favor of a gradient or categorical prosodic encoding of contrast. Thus, for authors such as Steedman (2000) and Büring (2003), the realization of contrastive focus and topic obligatorily corresponds to a

24

Contrast Encoding in European Portuguese

specific intonational contour. In fact, Büring (2003) argues that the definition of contrastive focus and of contrastive topic should be based on their prosodic realization, hence the author defines the former as a linguistic category realized by a falling pitch accent – the “A-accent” (Büring 2003, 512) – and the latter by a falling-rising pitch accent – the “B-accent” (Büring 2003, 512). On the other hand, Féry (2007) and Féry and Krifka (2008) propose that there is no one-to-one relation between intonation and contrast, although some specific intonational contours1 can be preferentially associated with structures that convey contrast, such as contrastive focus. Opposite to what is argued by Büring (2003), for example, Féry (2007) claims that topic, focus, and contrast are not phonological concepts but rather that their phonetic and phonologic properties can be cues to their interpretation. As regards the hypothesis of a gradient prosodic encoding of contrast, several recent studies for different languages have presented data supporting this hypothesis. For German, Braun and Ladd (2003) and Braun (2006) compared the prosodic features of contrastive and noncontrastive topics to conclude that contrastive topics present (i) a longer duration of the stressed vowel; (ii) a higher and longer f0 rise; (iii) higher values of f0 range; (iv) higher values of f0 peak height; and (v) later f0 peak alignment. Also for German, but regarding focus structures, Baumann, Grice and Steindamm (2006) discuss the role of categorical and gradient features in contrast marking to conclude that speakers use both. According to the findings described in this study, broad focus and narrow focus (including in the latter contrastive focus) differ through the presence of the pitch accent !H* in over 50% of the cases of broad focus and, crucially, by the complete absence of the same pitch accent in contrastive focus. Moreover, similar phonetic properties such as those highlighted by Braun and Ladd (2003) were also found as the focus domain narrows, namely a longer duration of the focalized elements, a higher f0 peak associated with the nuclear accent, a greater pitch extension to the peak of the nuclear accent, and, finally, a delay in the nuclear accent peak (Baumann, Grice and Steindamm 2006, 303). For Italian, parallel results are described with regard to contrastive foci by Torregrossa (2012), who highlights the role of longer duration and higher values of f0 range measured in focalized 1

Concerning this aspect, the authors refer to the intonational contours that have been associated, in the literature, with the structures that convey contrast in different languages, such as Büring (2003), for German, Steedman (2000), for English, and Frota (2000), for EP, that are also referred to in the present work.

Aida Cardoso

25

elements, as opposed to the fact that contrast does not seem to have a direct effect either on phrasing or on intonational contours. Finally, Borràs-Comes, Vanrell and Prieto (2010), in a study about the prosodic features of statements, contrastive foci, and echo questions in Catalan, also conclude that f0 range and f0 peak height are determining features and, therefore, that there is a gradient distinction between statements and contrastive foci. Lastly, in many of the studies mentioned above (Braun and Ladd 2003; Borràs-Comes, Vanrell and Prieto 2010; among others), it has been pointed out that the variation within and across speakers is an additional argument in favor of the gradient nature of contrast. Additionally, it is also important to mention, as Ladd (2008) has noted, that emphasis can play a relevant role in contrast marking as well. As the author puts it, emphasis can be seen as a “paralinguist possibility of gradiently modifying the realization so as to single out individual words” (Ladd 2008, 256). As regards the phonetic features associated with emphasis, Ladd (2008) and Ladd and Morton (1997) state that higher values of energy and of f0 range are associated with emphasis. For EP, although there are no specific studies about contrastive parallelism, some literature has discussed the prosodic features related to contrast. Viana (1987), for instance, analyzed foci structures and argued that focalized elements are realized by a “height accent” (Viana 1987, 87) that affects the height of the f0 peak (aligned with the stressed vowel of the focalized word) which, in turn, affects the range of f0 that precedes and follows the f0 peak. Frota (2000), on the other hand, argues in favor of a categorical realization of focus, which is reflected in prominence and intonational patterns. Hence, the focalized element is the more prominent, regardless of its position in the sentence, and it is always associated with the pitch accents H*+L or ^H*+L. Finally, Viana and Frota (2007), besides reaffirming that H*+L and ^H*+L are associated with focus, present data that allow the authors to say that the pitch accents H* and L+H* convey new information and are associated with emphasis as well. Finally, the pitch accent ^H* was found in contexts of emphasis and specification or correction of given information.

2.2. Parallelism Often taken as a rhetorical device, parallelism is also, and more importantly, referred to as a cohesion mechanism in early works about textual cohesion, such as Halliday and Hasan (1976). From a discourse / syntax perspective, parallelism can be defined as a cohesion mechanism involving common grammatical features, similar word order, and similar

26

Contrast Encoding in European Portuguese

syntactic structure, as proposed by Duarte (2003). Furthermore, the author emphasizes other important characteristics of parallelism that evidence the interface of parallelism with other cohesion mechanisms in the construction of a textual unit. In this context, Duarte (2003) mentions, for instance, that lexical cohesion plays an important role in structural parallelism, since the latter can often be associated with strategies like word repetition or the presence of semantic relations. From a prosodic point of view, parallelism is essentially described as a mechanism of tonal copy or, in other words, as intonational parallelism. Moreover, intonational parallelism is traditionally defined as tonal copy between consecutive intonational units. It is also worth noting that, more frequently than not, the study of intonational parallelism does not focus on constructions of structural parallelism. In fact, as works such as Palmer (1922), Crystal (1969) or Fox (1984) show, these authors analyzed the presence of tonal copy involving coordination, subordination, and parenthetical structures, for example, aiming to identify the specific intonational contours that could be involved in intonational parallelism. Subsequent studies (Bolinger 1989; Wichmann 2000) show a new perspective by ascribing a cohesive function to intonational parallelism. Crucially, Bolinger (1989) points out that the repetition of intonational contours is what gives intonational parallelism its cohesive function and, for this reason, it is more relevant than the specific type of intonational contour that is copied. The type of intonational contour, on the other hand, can be motivated by its discursive function or meaning: the repetition of an intonational contour characterized by “an abrupt fall in or from the syllable that is made to stand out by the fall” (Bolinger 1989, 3) can be associated with emphasis by giving a “dramatic or authoritative effect” (Bolinger 1989, 208) to a sequence, for instance. Furthermore, Bolinger (1989) argues that an emphatic realization of conjunctions (e.g. “and”, “or”) creates more tension (since conjunctions are elements not frequently accentuated in other contexts), thus drawing the attention of the hearer. More recently, Wichmann (2000) presented important findings on this matter. The author analyzed data from the Spoken English Corpus and found intonational parallelisms involving consecutive nuclear pitch accents, as has been traditionally described, but also found two new contexts of tonal copy: (i) intonational parallelism involving different tones and (ii) intonational parallelism between non-consecutive tonal units. Regarding the first context, Wichmann (2000) argues that intonational parallelism can be perceived between different tones if we consider not their phonological categories, but their phonetic properties instead. As the author explains, the tones L* H and H* H, for instance, can

Aida Cardoso

27

be perceived as parallel because, phonetically, they are both realized with a final rising movement. On the other hand, the possibility of intonational parallelism between non-consecutive tonal units can be motivated by phrasing. As the author explains, taking a list as an example, if each list item does not correspond to a single tonal unit, we can still find tonal copy, although not between consecutive nuclear pitch accents.

3. Methodology The corpus used here was constructed from a political debate aired by the public radio and television broadcaster in Portugal (RTP) on November 6th, 19752. This political debate featured the two candidates for Prime Minister of Portugal at the time: Álvaro Cunhal (AC) and Mário Soares (MS). The main reason why this debate was chosen is related to its unique characteristics. First of all, it represents an historical moment since it was decisive in the result of the first democratic elections after the end of a long period of dictatorship in Portugal. Secondly, the quality of the debate has been acknowledged in the literature, through the fact that both opponents’ speech is characterized by rational argumentation, aiming at enlightening the viewers, and by the absence of verbal attacks (Sena 2002). The debate lasted 3:31’07” and, besides the presence of the two political leaders, it was moderated by the journalists José Carlos Megre and Joaquim Letria, although only the speech of AC and MS was considered for analysis (3:18’10’’). It is also important to note that the speech time of AC and MS was balanced: AC had a total of 1:38’01” and MS a total of 1:40’09”. Regarding the transcription and alignment of the corpus, the debate was previously converted from video format (Video OBject) to audio format (WAVEform audio format) using Cool Edit Pro and the transcription (based on the transcription published in the newspaper Diário de Lisboa on November 8th, 1975 edition) was aligned with the acoustic signal using Transcriber. Considering the target structures of this study, it was first necessary to produce a survey of the relevance of parallelism structures in the corpus. 2

It should be mentioned that, for this study, I had access to a DVD copy of the original TV recording provided by the Centro de Documentação 25 de Abril (http://www1.ci.uc.pt/cd25a/wikka.php?wakka=HomePage), of the University of Coimbra. Due to the professional quality of the recording and to the fact that the conversion from VHS to DVD was made by a technician from the Centro de Documentação 25 de Abril, it was possible to work on data with a high acoustic quality and, therefore, to analyze the entire speech time of the debate.

28

Contrast Encoding in European Portuguese

In order to do so, all parallelisms were identified and annotated according to a typology involving six categories: construction parallelism (or syntactic parallelism); lexical parallelism (involving lexical repetition and lexical scales); temporal parallelism (concerning verbal features such as tense, aspect, person, and number); semantic parallelism (involving semantic relations such as whole/part, hierarchy, and similarity/opposition relations); rhyming parallelism (concerning rhymes and alliterations); and prosodic parallelism (defined, a priori, as a cohesion mechanism mainly associated with tonal copy, according to Bolinger (1989) and Wichmann (2000)). Through this annotation, a total of 391 parallelism structures were identified in the corpus (244 produced by AC and 147 produced by MS). A closer observation of the data allowed the identification of a subtype of construction parallelism that seems relevant in argumentative discourse, since it is found in crucial moments of the argumentation and counterargumentation. This specific type of construction parallelism – from now on, called contrastive parallelism (cf. example in (1)) – is defined as a structure in which parallel grammatical structures express a proposition that denies or restricts the truth-value of another proposition in the same “contextual set” (Stalnaker 1978; Reinhart 1982). (1) Ora, o Partido Socialista já escolheu o seu campo desde sempre. O Partido Socialista é um partido de esquerda, quer instaurar em Portugal uma sociedade socialista, portanto, uma sociedade sem classes, mas em liberdade, mas respeitando os direitos do homem, mas através da democracia e do consenso popular majoritário, não fará uma revolução, nem irá para um socialismo que transforme este País numa ditadura. (MS) Now, the Socialist Party has chosen which side is it on from the beginning. The Socialist Party is a left-wing party, it wants to establish in Portugal a socialist society, therefore, a society without classes, but in freedom, but respecting human rights, but through democracy and the majority popular consensus, it will not do a revolution, nor will it choose a socialism that turns this country into a dictatorship. (MS) A total of 47 cases of contrastive parallelism were found in the corpus (37 produced by AC and 10 produced by MS). It is important to note that each case of contrastive parallelism is an utterance composed of the contrastive parallelism structure itself and the context necessary to ensure its correct interpretation. For this reason, the length of the analyzed

Aida Cardoso

29

utterances varies between 3.2 seconds, for the shorter utterance, and 50.5 seconds, for the longest utterance. After identifying the cases of contrastive parallelism, it was considered relevant to annotate the major and minor intonational phrases in all the 47 utterances containing contrastive parallelism structures. The result was a total of 1097 major and minor intonational phrases (789 for AC and 308 for MS). From this total, a sample was then selected for prosodic analysis. The selection of the data for analysis had the purpose of guarantying a balanced quantity of data from AC and MS and, therefore, avoiding any bias effects in the results, since the previous annotation tasks showed that AC had made more parallelism structures than MS. Secondly, the sample includes examples that are representative of the types and diversity of syntactic structures present in the speech of the two debaters. Moreover, in this selection process, the intonational phrases were classified according to their function in the utterance. Hence, two types of prosodic constituents were considered: the target constituents (T), which are intonational phrases that make up the contrastive parallelism structures, and the context constituents (C), which are intonational phrases that can also be found in the utterance in which there is a contrastive parallelism and are a part of their “contextual set”. Furthermore, a third type of constituent was retrieved from the corpus with the purpose of comparing intonational phrases extracted from neutral declarative sentences (simple or complex), without marked word order nor associated with an emphatic prosodic realization, with the intonational phrases retrieved from context and, specially, with those retrieved from the target structures. These prosodic constituents were classified as control constituents (Ctrl). Type of Constituent Target Constituents (T) Context Constituents (C) Control Constituents (Ctrl) Total

AC 115 (28.4%) 46 (11.4%) 38 (9.4%) 199 (49.1%)

Speaker MS 116 (28.6%) 53 (13.1%) 37 (9.1%) 206 (50.9%)

Total 231 (57%) 99 (24.4%) 75 (18.5%) 405 (100%)

Table 1: Prosodic Constituents Selected for Analysis Finally, all the 405 major and minor intonational phrases (cf. Table 1) were prosodically annotated in Praat (Boersma and Weenink 2009), following the conventions of Towards a P_ToBI (Viana and Frota 2007). Each Praat file has: (i) a word tier, with the orthographic transcription; (ii) a break index tier, with the annotation of break indices of levels 3 and 4

30

Contrast Encoding in European Portuguese

(minor and major intonational phrases); and (iii) a tone tier, where the prenuclear and nuclear pitch accents, as well as the boundary tones, were annotated. The annotation of each intonational phrase also included global and local phonetic measures. Concerning the local measures, the f0 values, in semitones (ST), of high and low targets of pre-nuclear and nuclear pitch accents and also of boundary tones were extracted. As for the global measures, duration (in seconds), number of syllables (phonological), maximum and minimum energy levels (in decibels), and f0 (ST) maximum, minimum and range were extracted from each intonational phrase. It should be mentioned that the data collected from the prosodic annotation was then statistically analyzed using SPSS software, version 18.0.0.

4. Results3 The statistical analysis conducted on the data showed the existence of correlations between the type of structures and the prosodic parameters considered in this study and, also, it showed that there are differences between both speakers regarding several of the parameters. Hence, and with regard to this latter aspect, the statistical analysis of phonetic measures shows that there are significant differences between the two speakers in many of the parameters that were considered. As far as global measures are concerned, this is true for duration (U = 17865, p = .025), number of syllables (U = 13108, p < .001), energy maximum (U = 14509, p < .001), f0 maximum (U = 17078, p = .004), and f0 minimum (U = 15709, p < .001). In the case of local measures, the f0 maximum of the pre-nuclear pitch accents (U = 2314, p < .001), the f0 minimum of the prenuclear pitch accents (U = 2484, p = .001), and the f0 maximum of the nuclear pitch accents (U = 17422, p = .009) also present similar results. In the case of pitch accents and boundary tones, significant differences were found in the nuclear pitch accents (Ȥ2 (1) = 9.332, p = .009) and boundary tones (Ȥ2 (2) = 10.258, p = .001). These results are relevant, since they are in line with what has been described in previous studies about the variation across speakers found in the prosodic marking of contrast (e.g. Braun and Ladd 2003; Borràs-Comes, Vanrell and Prieto 2010). Now, considering the correlation between phonetic features and the three types of constituents analyzed – target (T), context (C), and control

3

For a more detailed description of the data presented in this section, as well as for consulting tables and graphics concerning all the phonetic and phonological parameters discussed, see Cardoso (2012).

Aida Cardoso

31

(Ctrl) –, it was also found that, in many cases, there is in fact a positive correlation and, what is more, two patterns emerge from the statistical analysis. Firstly, the results obtained for the duration and number of syllables point to the fact that these parameters showed closer values between T and Ctrl, as can be seen in Table 2. What is more, the statistical analysis reveals significant differences between constituents T and C (duration: (U = 8957.5, p = .002), number of syllables: (U = 9388.5, p = .010)) and also between C and Ctrl (duration: (U = 2341, p < .001), number of syllables: (U = 2600, p = .001)). Duration (seconds) Type of Speaker Constituent AC MS T 0.894 0.751 C 0.985 0.963 Ctrl 0.707 0.737

Number of Syllables Type of Speaker Constituent AC MS T 7.28 4.70 C 8.46 6.13 Ctrl 5.53 4.81

Table 2: Mean Values of Duration and Number of Syllables Although the proximity of values between T and Ctrl can be seen as somewhat unexpected, the lower values found in T may be related to a phrasing and prominence strategy aiming at emphasizing function words in contrastive parallelism structures (Bolinger 1989). In fact, throughout the cases of contrastive parallelism, there are many examples of conjunctions and connectors conveying negation and contrast (e.g. mas ‘but’, pelo contrário ‘on the contrary’, etc.) that form an independent intonational phrase and that are realized with an emphatic intonation. The results obtained for the energy and f0 parameters, on the other hand, show a different pattern from the one described for duration and number of syllables. Focusing firstly on the statistical results for the global measures, significant differences exist between T and Ctrl and also between C and Ctrl (but not between T and C) in terms of energy maximum (T and Ctrl (U = 6386, p = .001); C and Ctrl (U = 2660, p = .001)); in f0 maximum (T and Ctrl (U = 3149, p < .001), C and Ctrl (U = 1114, p < .001)); and the f0 minimum (T and Ctrl (U = 4910, p < .001), C and the Ctrl (U = 2478, p < .001). Also relevant is the fact that the f0 range is the only parameter showing differences between all three types of constituents (T and Ctrl (U = 6289, p < .001), C and Ctrl (U = 2012, p < .001), and T and C (U = 9420, p = .011)).

Contrast Encoding in European Portuguese

Semitones

32

18 16 14 12 10 8 6 4 2 0

16.608

16.020

14.897

14.408

10.829 9.942

AC MS

T

C

Ctrl

Type of Constituent

Figure 1: f0 Maximum – Mean Values

Semitones

As can be seen by the results described so far, the energy and f0 parameters that show a correlation between types of constituent and prosodic features point to a pattern that opposes Ctrl to T and C (for an example see figure 1). The data from the local measures seems to corroborate this (for an example see figure 2).

18 16 14 12 10 8 6 4 2 0

16,736

15,638

13,835

T

11,218 8,786

13,563

C

AC

MS

Ctrl

T ype of Constituent

Figure 2: f0 Maximum of Pre-nuclear Pitch Accent – Mean Values

Indeed, for local measures significant differences were found, once again, between the T and Ctrl constituents in all the analyzed parameters: the f0 maximum (U = 208, p < .001) and the f0 minimum (U = 285, p = .004) of the pre-nuclear pitch accents; the f0 maximum (U = 4281, p ]ݠ‬j]), the voiceless realization of intervocalic affricates ([d.dࢎ ‫[>]ݤ‬t.tࢎ ‫)]ݕ‬, and the distribution of stressed mid front vowels ([e]-[‫)]ܭ‬ among those speakers indicate that not all linguistic features are equally affected by dialect loss across generations and also that internal geographical variation plays a decisive role in this regard. Keywords: Catalan, Phonology, Phonetics, Language Variation.

1. Introduction1 During exploratory research aimed at the identification of noteworthy phonological features in the Catalan spoken in the region of La Selva, in North-Eastern Catalonia (Figure 1), it became clear that at least three kinds of language variation processes have taken place in the area (see Bosch 2012). In this paper, three features are considered that represent those three kinds of phonological variation: historical delateralization of the palatal lateral ([‫[>]ݠ‬j], yodization from now on), the voiceless realization

1

Research supported by MINECO and FEDER (FPI2011 pre-doctoral scholarship, project number FFI2010-22181-C03-02).

Eva Bosch i Roura

41

of intervocaalic affricates ([d.d ( ࢎ ‫[>]ݤ‬t.tࢎ ‫)]ݕ‬,, and the distri ribution of streessed mid front vowelss ([‫]ܭ‬-[e]).

Figure 1: Left ft: La Selva in thhe Catalan-speaaking area. Righht: La Selva in Catalonia

The rest of this sectionn states the go oals and the innterest of this research. Section 2 deeals with metthodology, Seection 3 with data and anallysis and, finally, a sum mmary is pressented in Section 4.

1.1. Go oals There arre three goalss to this paper. The first iss to describe the three aforementiooned phonologgical featuress and to pressent the relev vant data gathered byy means of 344 interviews held h with speaakers belongin ng to two generations from La Selvva, one of the 41 comarquues2 in Catalo onia. The second goall is to analyse the variatiion observed in the produ uction of utterances oof those featuures between the t two age ggroups of parrticipants: younger speeakers, aged 20 to 30 yeaars (Generatiion 1, G1), and a older speakers, agged 60 years or o more (Gen neration 2, G22). The final goal g is to show that thhe linguistic variables v takeen into considderation are differently d affected byy variation accross generattions and thaat tendencies towards dialect loss or maintenance are due to a variety of ffactors, some of which seem directlly related to geeographical variation.

1.2.. Interest of the research h In terms of intergenerrational variattion, this reseaarch is justified by the fact that Cattalan is underrgoing rapid processes p of chhange and thaat there is a strong tenddency towardds dialect levellling (Viaplanna 1999, 115; Viaplana 2002, 222; Perea 2007, 87-88). Obv viously, thesee propensitiess are not exclusive too Catalan, butt some specifiic relatively reecent processses in this 2

A comarca is an administraative division containing severral municipalitiies.

42

Phonological Variation in La Selva Catalan

language – namely linguistic normalization, standardization, and schooling in Catalan – have reinforced them. In this sense, an apparent-time study of language is especially relevant to the current project due to the differences between the linguistic histories of the two age groups under study: G1 speakers have been schooled in Catalan and have grown up during the normalization process, while G2 speakers were schooled solely in Spanish and, for the most part, have never learned the basics of Catalan writing and grammar. As for the reasons that make La Selva worthy of study, it must be noted, first of all, that the region belongs to Central Catalan (Enciclopèdia Catalana, 2008b). This dialect is at the basis of Standard Catalan, which makes it especially prone to dialect levelling and which has led to the neglect of the actual varieties that conform Central Catalan by scholars. Apart from that, the comarca of La Selva is in the area of Girona, where several authors have argued for the existence of a distinctive, though vaguely described, sub-dialect (Luna 1982, 158; Sala 1983, 21; Julià 1986, 130; Monturiol and Domínguez 2001, 25-26; Adam 2006, 421-422). Moreover, this sub-dialectal area appears to be coincident with the diocese of Girona rather than with the administrative province (Coromines 1953, 209; Adam 2006, 421-422). The borders of this diocese cross La Selva, and so some of its Western municipalities belong to the diocese of Vic or are on the border between the dioceses of Girona and Barcelona (Enciclopèdia Catalana, 2008a).

2. Methodology In this section, we provide details on 4 methodological aspects: the area studied, the participants involved, the characteristics of the linguistic interviews, and the data processing.

2.1. Area studied The present research analyzes data gathered through 34 interviews carried out in 17 of the 26 municipalities in the comarca of La Selva (Figure 2). This is a region of numerous and deep contrasts, which arise from the highly diverse geographical configuration of the area and which are reflected in its economy and its socio-demographic characteristics. Such diversity has an impact on the structure and the unity, or the lack thereof, of La Selva (see Bosch 2012, §4.1).

Eva Bosch i Roura

- - - - - GIRONA DIOCESE BORDERS EXCLUDED MUNICIPALITIES a. Sant Julià del Llor i Bonmatí b. Susqueda c. Brunyola d. Massanes e. Riells i Viabrea f. Fogars de la Selva g. Blanes h. Lloret de Mar i. Tossa de Mar

43

STUDIED MUNICIPALITIES 1. Amer 10. Caldes de Malavella 2. La Cellera de Ter 11. Vidreres 3. Anglès 12. Maçanet de la Selva 4. Osor 13. Sils 5. Sant Hilari Sacalm 14. Riudarenes 6. Arbúcies 15. Sant Feliu de Buixalleu 7. Santa Coloma de Farners 16. Hostalric 8. Vilobí d’Onyar 17. Breda 9. Riudellots de la Selva

Figure 2: Municipalities in La Selva

2.2. Participants Interviews were held with 34 participants3, distributed among the 17 municipalities mentioned in §2.1, with 1 young participant (G1) and 1 3

Admittedly, the amount of participants is small enough that it could be argued that any variation found might be attributed to the personal characteristics of the speakers. As previously mentioned, though, this research is a work in progress that arises from a preliminary, essentially descriptive, study on the phonological characteristics of Catalan spoken in La Selva. And, in any case, as stated by Labov (1994, 34), all personal characteristics should be considered within the social context.

44

Phonological Variation in La Selva Catalan

aged participant (G2) in each municipality. The profile of the participants is as follows: they are Catalan speakers from Catalan-speaking families; they were born either between 1982 and 1992 (G1) or in 1952 or before (G2); they were born and have lived almost exclusively in the studied municipality; they were born to at least one parent from the same municipality, while the other must have come either from that same municipality, the region of La Selva or a municipality nearby; they do not have university-level education. Of the 34 participants, 16 were women and 18 were men.

2.3. Interviews The linguistic interviews were conducted between April and July 2012. Each interview lasted from 40 to 90 minutes and was recorded both on video and audio. The interviews were divided into two parts. The first was a semistructured interview used to obtain fragments of casual speech. The second part was a formal survey, composed of images and precise questions that had to be identified and answered, respectively, with single words or simple sentences. This section was aimed at recording potentially interesting phonetic segments. The current research uses data from the second part of the interviews.

2.4. Data processing The recordings were analysed to identify the required segments, which were subsequently extracted, transcribed using IPA, and stored, using Phon (http://childes.psy.cmu.edu/phon/ version 1.5.2). The data obtained was then classified and organized for appropriate analysis.

3. Results and analysis In this chapter, the data regarding the phonological features that are the focus of this paper are presented and analyzed. In each case, the phenomenon is briefly described, the data obtained in the interviews is presented, conclusions are drawn, and plausible explanations are offered.

3.1. Historical yodization ([‫[>]ݠ‬j]) In some Catalan-speaking areas, including La Selva, the digraph ll has traditionally had two distinct realizations. The first, [‫]ݠ‬, appears when ll

Eva Bosch i Roura

45

derives from the Latin segments L- or -LL-, as in lluna [‫ݠޖ‬u.nԥ] (“moon”), from Latin LNjNA. The second realization is [j] – hence yodization –, and it appears when ll derives from the Latin segments -LY-, -C’L- or -G’L-, as in fulla [‫ޖ‬fu.jԥ] (“leaf”), from Latin FǁLưA (Pons 1992, 12; Recasens 1996, 322). However, in Standard Catalan, as reflected by the orthography, [‫ ]ݠ‬is used in all contexts. Table 1 contains the realizations produced by our participants4 for 15 words susceptible to undergoing yodization. Word

Realization [‫]ݠ‬

(a) fillol/a “godson”

(b) assolellat “sunny”

(c) cella “eyebrow”

(d) agulla “needle”

[j] [‫]ݠ‬ [j] [‫]ݠ‬ [j] [‫]ݠ‬ [j]

(e) raspall “brush”

[‫]ݠ‬ [j]

Age group G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2

Municipalities5,6 2, 5,10, 11, 12, 13, 14, 15 1, 7, 8, 1, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17 1, 7, 9, 13, 15, 17 5, 7, 9 2, 3, 4, 5, 6, 8, 10, 12, 14 1, 3, 4, 6, 8, 11, 13, 15, 16 1, 3, 5, 7, 8, 9, 11, 12, 13, 15 1, 5, 9, 11, 17 2, 4, 6, 10, 14, 17 2, 3, 4, 6, 7, 8, 10, 12, 13, 14, 15, 16 1, 2, 3, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 17 2, 8, 9, 10, 11, 13, 17 4, 5 1, 3, 4, 5, 6, 7, 12, 14, 15, 16 1, 2, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 17 6, 8, 9, 10, 11, 13, 16, 17 3, 4 1, 2, 3, 4, 5, 7, 12, 14, 15,

4 The words lluna (“moon”) and ocell (“bird”), not susceptible to historical yodization, were included in the survey to make sure that our participants produced [‫ ]ݠ‬properly. This lead to the exclusion of data from speaker G1 from Breda, who used [j] in both words. 5 The number-municipality correspondence can be seen in Figure 2. 6 In Tables 1-5, missing municipalities indicate that there is no record for that specific word, municipality, and age group.

Phonological Variation in La Selva Catalan

46 Word (f) palla “straw”

(g) barallar-se “to fight”

(h) parell “pair”

(i) orella “ear”

Realization [‫]ݠ‬ [j] [‫]ݠ‬ [j] [‫]ݠ‬ [j] [‫]ݠ‬ [j]

(j) abelles “bees”

[‫]ݠ‬ [j]

(k) mirall “mirror”

[‫]ݠ‬ [j]

(l) ull “eye”

[‫]ݠ‬ [j]

(m) estovalles “tablecloth”

[‫]ݠ‬ [j]

Age group G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2

Municipalities 1, 2, 3, 4, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 4, 7, 8, 9, 10, 11, 12, 13 5, 17 1, 2, 3, 5, 6, 14, 15, 16, 17 1, 2, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 17 2, 5, 6, 7, 8, 9, 10, 11, 13, 16, 17 3, 4, 5, 16 1, 3, 4, 12, 14, 15 1, 2, 3, 5, 6, 7, 8, 9, 10, 11, 12, 13, 15, 17 6, 7, 8, 9, 10, 12, 13, 17 4, 14 1, 3, 4, 11, 14, 15, 16 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 17 2, 4, 8, 9, 10, 11, 13, 16, 17 1, 3, 5, 6, 7, 12, 14, 15 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 17 2, 6, 8, 9, 10, 11, 13, 17 1, 3, 5, 7, 14, 15, 16 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 17 2, 6, 8, 9, 10, 11, 13, 15, 16, 17 1, 3, 4, 5, 6, 7, 12, 14 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 17 2, 7, 8, 9, 10, 11, 13, 15, 16, 17 1, 3, 4, 5, 6, 12, 14 1, 2, 4, 5, 7, 8, 9, 10, 11, 12, 13, 14, 15, 17 2, 6, 7, 8, 9, 10, 11, 13, 15, 17 1, 3, 4, 5, 12, 16

Eva Bosch i Roura

Word

(n) genoll “knee”

Realization [‫]ݠ‬ [j]

(o) vermell “red”

[‫]ݠ‬ [j]

Age group G1 G2 G1 G2 G1 G2 G1 G2

47

Municipalities 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 17 4, 5, 6, 7, 8, 9, 10, 11, 13, 15, 17 1, 2, 3, 12, 14, 16 1, 3, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 17 2, 4, 5, 6, 7, 9, 10, 11, 13, 15, 16, 17 4 1, 3, 12, 14

Table 1: Data on historical yodization in La Selva The data presented in Table 1 can be divided into three blocks. First, the 1(a)-(c) words show a prominent use of the variant with [j], but with a frequency ranging from 54.55% to a maximum of 70.43% of the utterances. Secondly, the 1(d)-(h) words have a frequency of use of the variant with [j] that ranges from 27.27% to 36.36% of the recordings. Finally, between 15.15% and 24.24% of the 1(i)-(o) words are produced with [j]. These figures vary enormously from one age group to the other. G1 speakers, on the one hand, use forms with [j] with a frequency from 0 to 60% of their realizations. The use of these variants by G2 speakers, on the other hand, ranges from 17.65% to 100% of their utterances. In both cases, the frequency of forms with [j] varies according to the 3 aforementioned blocks. In addition to the words presented in Table 1, it is worth commenting on the forms recorded for two pairs of words in which yodization is used to distinguish between meanings: poll (“louse”, from Latin PEDǍCǍLU) vs. poll (“chick”, from Latin PǍLLU) and conill (“rabbit”, from Latin 7 CUNƮCǍLǍ) vs. conill (“naked”). In the first case, the situation is as follows: while poll meaning “louse” undergoes yodization in most of the recordings,8 every single example of poll meaning “chick” is produced with a [‫]ݠ‬. In the case of conill, all of the 21 participants who knew that 7

This use of conill meaning “naked” is found in several areas of the region, but it is not recorded in dictionaries. Our hypothesis is that this is a figurative use of the word, derived from the same Latin etymon CUNƮCǍLǍ. 8 All the speakers say [‫ޖ‬poj], except for G1 in Amer, Santa Coloma de Farners, Vilobí d’Onyar, Riudellots de la Selva, Maçanet de la Selva, Sils and Riudarenes, and G2 in Arbúcies).

48

Phonological Variation in La Selva Catalan

conill was used with the meaning of “naked” produced the word with a [‫]ݠ‬, while out of those, 5 had previously used the yodizised form of conill “rabbit”. The data presented enables some general conclusions: G1 speakers have very nearly abandoned yodization altogether, and, while the use of variants with [j] among G2 speakers maintains some vitality, it is also in clear regression. There are some specific words where yodization is still regularly used, even among younger speakers, but those items seem to be undergoing a lexicalization process (see Recasens 1996, 323). There are a number of reasons for this regression. First of all, we find that yodization lacks prestige9 and that it is highly stigmatized as rural, old, and uneducated (Bau and Font 1992; Pons 1992, 21-22; Recasens 1996, 322-323), at least in Central Catalan. In addition to that, as we have seen, orthography favours the use of the same phonetic segment for both forms traditionally pronounced with [‫ ]ݠ‬and forms traditionally pronounced with [j], for the digraph ll is used in both cases. Finally, it must be noted that a new form of non-historical yodization affecting all instances of ll, regardless of etymology, is spreading across the Catalanspeaking areas.10 This is occurring especially in urban areas and in areas strongly influenced by Spanish, which would add to the lack of prestige of historical yodization.

3.2. Voiceless realization of intervocalic affricates ([d.dࢎ ‫[>]ݤ‬t.tࢎ ‫)]ݕ‬ In some areas of Central Catalan and, apparently, more significantly in the Northern regions (Recasens 1996, 215), intervocalic affricates may have two distinct realizations. The first is a voiceless realization, [t.tࢎ ‫]ݕ‬,11 used in segments derived from Latin –ATưCU (see Pons 1992, §2.1.3, §2.2; Recasens 1996, 215). The second is a voiced realization, [d.dࢎ ‫]ݤ‬, used in segments derived from the Latin segments -GJA or -DJA. Furthermore, some speakers may also lose affrication and produce a fricative [‫ ]ݤ‬in some specific words, such as platja (“beach”) or lletja (“ugly”), the intervocalic affricates of which derive from Latin - GJA or –DJA as well. Standard Catalan considers that only the voiced forms are correct, and the 9

The lack of prestige is obvious in the case of our G1 speaker from Maçanet de la Selva, who produces the hypercorrection [dԥz.mԥ‫ݠޖ‬a] for the word desmaiar (“to faint”, [dԥz.mԥ‫ޖ‬ja]). 10 The G1 speaker from Breda, as stated in Note 4, exemplifies this phenomenon. 11 Long affricates are represented by [t.tࢎ ‫ ]ݕ‬and [d.dࢎ ‫]ݤ‬, according to authors such as Lloret (1992) or Wheeler (2005, 11-13), who consider that only the occlusive part of the segment is lengthened.

Eva Bosch i Roura

49

segment is orthographically represented by the digraphs tg and tj.12 Tables 2 and 3 provide data on this phenomenon. Words in Table 2 show the forms produced by our participants for three words with the suffix -atge, derived from Latin -ATưCU. This data shows that the voiceless realization is preferred in all three cases (with a frequency of 51.72 to 76.47% of the occurrences), but that there is an enormous difference between the forms used by these two age groups. Thus, while practically all G2 speakers use [t.tࢎ ‫ ]ݕ‬in these three words, the frequency of this variable among G1 speakers is significantly lower, ranging from 17.65 to 52.94% of their occurrences. Words in Table 3 present another interesting situation, in which variation between voiced, voiceless and fricative realizations is used as a meaning distinguisher. More precisely, we can observe that the intervocalic affricates in mitges (“tights”) and mitjons (“socks”) occur in 70.59% and 79.41% of voiceless realizations, respectively, with [t.tࢎ ‫ ]ݕ‬being used in 100% of the utterances produced by G2 speakers and from 41.18% to 58.82% of the utterances produced by G1 speakers. The adjective mitja (“half”), on the other hand, is produced with [d.dࢎ ‫ ]ݤ‬in 52.94% of the cases, which represent 76.47% of the utterances by G1 speakers but only 29.41% of the utterences by G2 speakers. The other occurrences are produced with [t.tࢎ ‫ ]ݕ‬and [‫ ]ݤ‬with a frequency of 23.53% for each variable, the majority of them being obviously produced by G2 speakers. Therefore, we can observe a clearly different behaviour regarding the intervocalic affricates in the pair of nouns mitges/mitjons and in the adjective mitja, although all three words derive from the same Latin adjective, MƞDưU/MƞDưA. Similarly to mitja, 67.65% of the utterances of platja (“beach”, derived from Latin PLAGưA) are produced with a voiced affricate, which represent 76.37% of the occurrences by G1 speakers and 58.82% of the occurrences by G2 speakers. The voiceless variant is used in 26.47% of the utterances, and the fricative [‫ ]ݤ‬represents a mere 5.88% of the occurrences, which are produced by two G2 speakers. All of these figures lead to the conclusion that G2 speakers show highly stable behaviour and maintain the traditional division between voiced and voiceless forms, but that such a distinction may also be used for semantic purposes. Regarding G1 speakers, they show a strong tendency towards the use of the Standard voiced affricate in all contexts, although their behaviour is in fact rather erratic. The fricative realisations of the intervocalic affricate appear to be merely residual for both groups.

12

The use of tg or tj is merely an orthographical issue.

Phonological Variation in La Selva Catalan

50

Word

Realization [d.dࢎ ‫]ݤ‬

(a) formatge “cheese”

(b) viatge “journey”

(c) equipatge “luggage”

[t.tࢎ ‫]ݕ‬ [d.dࢎ ‫]ݤ‬ [t.tࢎ ‫]ݕ‬ [d.dࢎ ‫]ݤ‬ [t.tࢎ ‫]ݕ‬

Age group G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2

Municipalities 1, 2, 4, 6, 8, 10, 12, 13, 3, 5, 7, 9, 11, 14, 15, 16, 17 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 12, 13, 14, 15 11, 16, 17 1, 3, 4, 5, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16 1, 2, 3, 4, 5, 7, 8, 9, 10, 12, 13, 14, 15 12 6, 11, 16, 17 4, 6, 8, 9, 10, 11, 13, 14, 15, 16, 17

Table 2: Data on intervocalic affricates from Latin -ATưCU in La Selva Word

Realization [d.dࢎ ‫]ݤ‬

(a) mitges “tights”

[t.tࢎ ‫]ݕ‬ [‫]ݤ‬ [d.dࢎ ‫]ݤ‬

(b) mitjons “socks”

[t.tࢎ ‫]ݕ‬ [‫]ݤ‬ [d.dࢎ ‫]ݤ‬

(c) mitja “half”, fem. adj.

[t.tࢎ ‫]ݕ‬ [‫]ݤ‬

Age group G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2

Municipalities 1, 4, 6, 8, 10, 12, 13, 15, 16, 17 2, 3, 5, 7, 9, 11, 14 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17 1, 5, 7, 10, 12, 13, 16 2, 3, 4, 6, 8, 9, 11, 14, 15, 17 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17 1, 2, 3, 4, 7, 8, 9, 10, 12, 13, 14, 15, 16 5, 8, 9, 12, 14 11, 17 7, 11, 13, 15, 16, 17 5, 6 1, 2, 3, 4, 6, 10

Table 3: Data on intervocalic affricates from Latin -DJA in La Selva

Eva Bosch i Roura

51

As for the possible explanations behind the tendency of G1 speakers towards abandoning the voiceless affricate in favour of the Standard voiced one, two main factors should be considered. The first is the fact that voiceless affricates are increasingly stigmatized, due to their use by speakers highly influenced by Spanish (Pradilla, 2008, §8.3.6). The second and most crucial factor is schooling, which favours a levelling of spoken language to written language (see Carrera-Sabaté 2002). The decisiveness of schooling over stigmatization in the choosing of voiced intervocalic affricates is reinforced by the fact that G1 speakers tend to realize initial palatal sibilant /‫ݕ‬/ as a voiceless affricate in words such as xai (“lamb”) or xeringa (“syringe”), even though [‫ ]ݕ‬is both the traditional and the Standard form in that case (see Bosch 2012).

3.3. Distribution of stressed mid front vowels ([e]-[‫)]ܭ‬ Catalan has a contrast between close-mid front vowels ([e]) and openmid front vowels ([‫ )]ܭ‬in stressed positions. According to several authors (see, for example, Coromines 1971, 250; Recasens 1996, 82), however, the distribution of those vowels in the area of Girona (considering, as mentioned in §1.2, that this area coincides with that of the diocese) shows a behaviour that differs from that of Standard Catalan, and that does not seem to be accountable for in terms of phonological context. In agreement with this theory, the participants in the diocese of Girona should be affected by such a distribution, while those in the diocese of Vic should not. Tables 4 and 5 provide data in this regard. The stressed vowel in the words in Table 4 is an open-mid front vowel in Standard Catalan, but this data shows that the majority of our participants use [e]. And, on the contrary, the stressed vowel in the words in Table 5 is a closed-mid front vowel in Standard Catalan, but the data shows that most of our participants produce the words with [‫]ܭ‬. The most remarkable factor about the distribution of these vowels, though, is that there is in fact almost no intergenerational variation – providing percentages would therefore be totally superfluous – but that geographical variation is significant and easily identifiable. The situation is as follows.

Phonological Variation in La Selva Catalan

52

Word (a) convergència “convergence”

(b) independència “independence”

(c) València

Realization [e] [‫]ܭ‬ [e] [‫]ܭ‬ [e] [‫]ܭ‬

(d) Josep

[e] [‫]ܭ‬

Age group G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2

Municipalities 1, 2, 3, 7, 8, 9, 10, 11, 13, 14, 17 2, 3, 7, 8, 9, 10, 11, 12, 13, 14, 15, 17 4, 5, 6, 12, 15, 16 4, 5, 6, 16 1, 2, 3, 7, 8, 9, 10, 11, 13, 14, 17 2, 3, 7, 9, 11, 13, 15, 17 5, 6, 12, 15, 16 4, 6, 10, 12, 16 1, 2, 3, 7, 8, 9, 10, 11, 13, 14, 17 1, 2, 8, 9, 10, 12, 13, 14, 17 4, 5, 6, 12, 15, 16 4, 6, 7, 11, 15, 16 1, 2, 3, 7, 8, 9, 10, 13, 14 1, 2, 3, 7, 9, 10, 13, 14 4, 5, 6, 11, 12, 15, 16, 17 4, 5, 6, 8, 11, 12, 15, 16, 17

Table 4: Data on Standard stressed mid front vowel [‫ ]ܭ‬in La Selva Word

Realization [e]

(a) mesos “months”

(b) església “church”

(c) reis “kings”

[‫]ܭ‬ [e] [‫]ܭ‬ [e] [‫]ܭ‬

Age group G1 G2 G1 G2 G1 G2 G1 G2 G1 G2 G1 G2

Municipalities 5, 6, 15, 16, 17 5, 6, 15, 16 1, 2, 3, 4, 7, 8, 9, 10, 11, 12, 13, 14 1, 2, 3, 4, 7, 8, 9, 10, 11, 12, 13, 14, 17 4, 6, 7, 11, 12, 15, 16, 17 4, 5, 6, 7, 9, 15, 16, 17 1, 2, 3, 5, 8, 9, 10, 13, 14, 17 1, 2, 3, 8, 10, 11, 12, 13 4, 6, 7, 10, 11, 12, 13, 14, 15, 16 4, 5, 6, 7, 9, 13, 15, 16, 17 1, 2, 3, 5, 8, 9, 17 1, 2, 3, 8, 10, 11, 12, 14

Table 5: Data on Standard stressed mid front vowel [e] in La Selva

Eva Bosch i Roura

53

The utterances by speakers from Amer, La Cellera de Ter and Anglès agree with those expected in the Girona area, while those located in Osor, Sant Hilari Sacalm, Arbúcies and Breda are unequivocally Standard, and the speakers in both groups of municipalities show highly stable behaviour in this regard. In the rest of the municipalities, the behaviour of the participants is slightly more unsteady, but they favour either Girona realizations (Vilobí d’Onyar, Caldes de Malavella, Riudellots de la Selva, Sils, and Riudarenes) or Standard realizations (Sant Feliu de Buixalleu, and Hostalric). Only in Santa Coloma de Farners – the capital of the region –, Maçanet de la Selva, and Vidreres do participants show truly erratic behaviour. This geographical variation is consistent with the aforementioned internal diversity of the region (§2.1). Thus, the speakers using Standard forms are those in the municipalities of Osor, Sant Hilari Sacalm and Arbúcies, which belong to the diocese of Vic.13 As for Breda, even though it belongs to Girona, it is located, along with Sant Feliu de Buixalleu and Hostalric, on the border of the diocese and the province of Barcelona. This administrative fact, which by itself could easily favour the use of Standard forms in these municipalities, is reinforced by the socio-demographic characteristics of the area, where mobility is by far greater than in the Northern areas of the comarca – that is, the municipalities of Amer, La Cellera de Ter and Anglès. Finally, these socio-demographic factors favouring mobility also affect the rest of the municipalities, where speakers show a somewhat erratic behaviour and have a tendency towards Standard forms, even if to a lesser degree. As to why most G1 participants in the diocese of Girona maintain the traditional distribution of these stressed mid front vowels instead of turning to Standard forms, it should be noted, firstly, that Catalan from Girona is popularly considered prestigious and well-spoken (Soler 1986). Secondly, it is proposed that this phenomenon is one of the centrifugal forces that affect any process of dialect levelling, and that the maintenance of these forms would contribute to the building of the social and linguistic identity of the speakers (Viaplana 2002, 70 and 222).

13 In fact, only half of the municipality of Arbúcies belongs to the diocese of Vic. The data obtained and analyzed in Bosch (2012), however, show that our speakers in the whole of Arbúcies have a strong preference for Standard forms. This behaviour is favoured by factors related to spatial diffusion of linguistic features by accommodation (Trudgill 2003, accommodation; Britain 2008, §2.6).

54

Phonological Variation in La Selva Catalan

4. Summary The data presented shows that the participants in the comarca of La Selva have a strong tendency towards the use of the Standard variety of Catalan to the detriment of their own geographical variety. Nonetheless, it has also allowed us to observe that not all features are equally affected by this variation. Therefore, firstly, the data shows that historical yodization ([‫[>]ݠ‬j]) has almost completely disappeared from the speech of G1 speakers and is also rare among G2 speakers, mainly due to the stigmatization of the phenomenon. Secondly, data concerning intervocalic affricates shows that while G2 speakers maintain the traditional division that favours the use of voiceless [t.tࢎ ‫ ]ݕ‬for segments derived from Latin atƱcu but of voiced [d.dࢎ ‫( ]ݤ‬or, residually, [‫ )]ݤ‬for segments derived from Latin -gja or -dja, G1 speakers tend to use only the Standard voiced affricates in this intervocalic position. The paper has stated that this shift must be attributed to the influence of written language on younger speakers. Finally, the distribution of stressed mid front vowels ([e]-[‫ )]ܭ‬has proved to be a very resistant trait, for both G1 and G2 speakers are inclined to maintain the traditional forms. In this case, however, variation is found between municipalities, with both diocese borders and the socioeconomic diversity of the region acting as major triggering factors. The maintenance of the particular distribution of these vowels by speakers from the Girona area has been attributed to popular prestige as well as to it acting as an identity-building element.

References Adam i Aulinas, Montserrat. 2006. El català septentrional de transició: nova visió des de la morfologia. Barcelona: Institut d’Estudis Catalans. Bau, Imma, and Ramona Font. 1992. “La iodització a la plana de Vic.” In Segon Congrés Internacional de la Llengua Catalana (1986), Àrea 3, vol. IV, 79–85. Palma: Universitat de les Illes Balears. Bosch i Roura, Eva. 2012. “Aspectes fonològics del català de la comarca de la Selva.” Master Thesis, Universitat de Barcelona. Accessed 14 June, 2013. http://hdl.handle.net/2072/209950 Britain, David. 2008. “Space and Spatial Diffusion.” In The Handbook of Language Variation and Change, edited by J. K. Chambers, Peter Trudgill and Natalie Schilling-Estes, 603-637. Oxford: Blackwell. Carrera-Sabaté, Josefina. 2002. Escola catalana i variació fonètica: una evolució del vocalisme àton a Alguaire i a Lleida. Lleida: Pagès.

Eva Bosch i Roura

55

Coromines, Joan. 1953. “Algunes lleis fonètiques catalanes no observades fins ara.” Estudis Romànics 3:201–230. Accessed 15 January, 2013. http://revistes.iec.cat/index.php/ER/article /viewFile/ 37651/48276. —. 1971. Lleures i converses d’un filòleg. Barcelona: Club Editor. Enciclopèdia Catalana. 2008a. “Bisbat de Girona”, in L’Enciclopèdia. Barcelona: Enciclopèdia Catalana, SAU. Accessed 20 January, 2013 . http://www.enciclopedia.cat. —. 2008b. “Català. Dialectes”, in L’Enciclopèdia. Barcelona: Enciclopèdia Catalana, SAU. Accessed 20 January, 2013. http://www.enciclopedia.cat. Julià i Muné, Joan. 1986. “Algunes notes sobre el parlar de Cornellà de Terri”. In Cornellà de Terri: pregons i estudis, edited by A Miquel Callís, 129–139. Cornellà de Terri: Ajuntament de Cornellà de Terri. Labov, William. 1994. Principles of Linguistic Change. Social Factors. Oxford [etc.]: Blackwell. Lloret, Maria-Rosa. 1992. “Les africades i la representació fonològica nolineal: estructures de contorn.” Els Marges: revista de llengua i literatura 46:47-63. Accessed 18 January, 2013. http://www.raco.cat/index.php/Marges/article/view/111358. Luna, Xavier. 1982. El vocalisme del català a la zona de transició centralseptentrional (Alt Empordà). BA Dissertation, Universitat Autònoma de Barcelona. Monturiol, Joaquim, and Eloi Domínguez. 2001. El Parlar De La Garrotxa. Olot: Ràdio Olot. Perea, Maria-Pilar. 2007. “Dedialectalization or the Death of a Dialect: The Case of the Catalan Subdialect Spoken in the Costa Brava.” Dialectologia et Geolinguistica, 15:77-89. Accessed 18 January, 2013. doi:10.1515/DIALECT.2007.006. Pons i Griera, Lídia. 1992. Iodització i apitxament al Vallès: interpretació sociolingüística i psicolingüística dels canvis fonètics. Barcelona: Institut d’Estudis Catalans. Pradilla, Miquel-Àngel. 2008. “Ensordiment, espirantització i fenòmens que afecten les sibilants.” In Gramàtica del català contemporani (4th ed.), directed by Joan Solà, Maria-Rosa Lloret, Joan Mascaró and Manuel Pérez Saldanya, Ch. 8, Vol. 1, 287-318. Barcelona: Empúries. Recasens i Vives, Daniel. 1996. Fonètica descriptiva del català: assaig de caracterització de la pronúncia del vocalisme i consonantisme del català al segle XX (2nd ed.). Barcelona: Institut d’Estudis Catalans. Sala, Ernesta. 1983. El Parlar de Cadaqués. Girona: Diputació.

56

Phonological Variation in La Selva Catalan

Soler, Josep. 1986. “Model i límits de la llengua estàndard a Barcelona.” Miscel·lània Antoni M. Badia i Margarit 5. Barcleona: Publicacions de l’Abadia de Montserrat. The PhoneBank Project. Accessed August 8, 2013. http://childes.psy.cmu.edu/phon/. Trudgill, Peter. 2003. A Glossary of Sociolinguistics. Oxford: Oxford University Press. Viaplana, Joaquim. 1999. Entre la dialectologia i la lingüística: la distància lingüística entre les varietats del català nord-occidental. Barcelona: Publicacions de l’Abadia de Montserrat. —. 2002. Dialectologia (2nd ed. corr.). Valencia: Universitat de València. Wheeler, Max W. 2005. The Phonology of Catalan. Oxford [etc.]: Oxford University Press.

CHILD ACQUISITION OF EXHAUSTIVITY: EUROPEAN PORTUGUESE1 STÉPHANIE DIAS VAZ

CENTRO DE LINGUÍSTICA DA UNIVERSIDADE NOVA DE LISBOA2

Abstract The main objective of this study is to see how children who are native speakers of European Portuguese acquire exhaustivity. This study brings together results previously obtained in Vaz (2010a; 2010b) with a new study of universal quantification structures. For each test, 60 children were tested divided into three age groups (3, 4 and 5 years old) and 20 adults as the control group. The results show that there is a significant difference between 4 and 5 year olds and that exhaustivity is associated with syntactic structure, even though this interacts with pragmatic factors. It is concluded that exhaustivity is acquired at a different rate across structures (simple wh-questions > universal quantification > clefts > multiple wh-questions), which is attributed to the varying complexity of each display. These results are in accordance with results found in other languages. Keywords: Acquisition, Exhaustivity, Questions, Clefts, Universal quantification.

1

This work was developed under the supervision of Professor Maria Lobo, for the master’s thesis by Stéphanie Dias Vaz (Vaz 2012). 2 Master in Linguistics – Universidade Nova de Lisboa, Researcher at the Centro de Linguística da Universidade Nova de Lisboa, in the project ‘Dependências Sintáticas dos 3 aos 10’ (PTDC/CLE-LIN/099802/2008). This work was supported by the National Funding Organisation – FCT for the Strategic Project PEstOE/LIN/UI3213/2011.

58

Child Acquisition of Exhaustivity: European Portuguese

1. Introduction It has been shown by several authors that there is linguistic development regarding the exhaustivity value in the different syntactic structures associated with it: 3 and 4 year olds have non-target interpretation for structures associated with exhaustivity, such as multiple wh-questions, clefts and universal quantification structures. The transition from a non-exhaustivity stage to an exhaustivity stage can be explained in terms of either grammatical or pragmatic development. (cf. Schulz and Roeper 2011). This work aims to describe the linguistic development of 3 to 5 year old children acquiring Portuguese, regarding exhaustive interpretation in wh-questions, clefts and universal quantification (with the quantifiers todos and cada3) and compare them to the results founds by other authors (cf. Schulz and Penner 2002; Grebenyova 2006; Schulz 2010; Heizmann 2007; Schulz and Roeper 2011; among others).

2. Structures involving Exhaustive reading Human languages code some structures with an Exhaustivity feature, meaning that a given variable should be applied to all the individuals in a set. Multiple wh-questions, some clefts and universal quantifiers such as todos (every/all), all encompass an exhaustivity feature. A sentence like Todos os meninos estão de pé (all the boys are standing) is only true if “estar de pé” (standing) is exhaustively applied to every individual under the scope of the quantifier todos, and is false if a single boy was not standing. Multiple wh-questions can also elicit exhaustive answers in several languages4. In Portuguese, an interrogative such as Quem comeu o quê? (Who ate what?), with two interrogative pronouns demands as an answer every pair of individuals that fulfills the value of the variables in a given pragmatic situation (cf. Costa 1998): a mãe (comeu) a banana / the mother (ate) the banana; o pai (comeu) a maçã / the father (ate) the apple; a avó (comeu) a laranja / the grandmother (ate) the orange; o avô (comeu) o pêssego / the grandfather (ate) the peach. These answers are commonly 3

Every/all and each. Grebenyova (2006) and Schulz and Roeper (2011) report the occurrence of interlinguistic variation related to several characteristics of multiple wh-questions: multiple wh-questions exist in some but not every language. Some, but not all, languages front every wh-pronoun; there is also variation regarding the requirement of exhaustive reading.

4

Stéphanie Dias Vaz

59

known as pair-list readings, (PL), opposing single-pair reading (SP)5. Simple wh-questions are dependent on the pragmatic situation to determine whether an exhaustive or non-exhaustive answer applies. Therefore, a question made in a court, Quem estava em casa à hora do crime? / Who was home at the time of the crime?, or in a classroom environment, Quem fez o trabalho de casa?/ Who did the homework?, elicits an exhaustive answer, while for Quem tem uma caneta? / Who has a pen? or Onde é que há uma casa de banho? / Where is the bathroom, a single answer is pragmatically suitable (or an answer where only a few items are mentioned – a plural but not exhaustive answer), because it is assumed that an exhaustive answer is not required. According to Schulz and Roeper (2011), however, exhaustive answers and single or a “few” answers are the result of ambiguity in the semantics of wh-questions, as non-exhaustive wh-questions are associated with existential quantification and exhaustive wh-questions, the non-marked ones, associated with universal quantification. It is possible to find a correlation between one reading and the operators with which each can cooccur, and so, in English, triggers a non-exhaustive reading and all or alles in German trigger exhaustive reading. The semantic analysis (nonpragmatic) of exhaustivity in wh-questions is supported by the fact that no plural answers (non-exhaustive) were found for multiple wh-questions (Schulz and Roeper 2011). Some clefts (or pseudoclefts) have the semantic property of determining exhaustive reading. Looking at the following sentences: (1) a. Quem comeu o bolo foi o avô. Who ate the cake was the grandfather b. O avô comeu o bolo. The grandfather ate the cake The first sentence,1(a), a pseudocleft, is only appropriate if no one else besides the grandfather ate the cake, and (1b), a declarative sentence, is true whether only the grandfather or more people ate the cake. Exhaustivity is usually associated with a set of structures involving an operator-variable chain. One can argue about the origin of exhaustivity. 5

As argued by Schulz and Roeper (2011), there are contexts in which multiple whquestions require singular answers: wh- echo questions and wh-questions concerning a single event, as in Quem bateu em quem primeiro? / Who hit who first?. In every other context an exhaustive answer is required and a plural nonexhaustive answer is ruled out.

60

Child Acquisition of Exhaustivity: European Portuguese

Some authors argue that it is pragmatically determined (e.g. van Rooij 2004), others consider it a grammatical property associated with a given structure, coded under a formal feature within an operator (cf. Roeper et al. 2007; Heizmnann 2007). If this view is accurate, exhaustive readings would envolve an operator, predicting that different structures should be associated with exhaustive reading around the same stage of development.

3. Acquisition of structures 3.1. Acquisition of interrogative structures Schulz and Roeper (2007) and Schulz and Roeper (2011) reflect on the predictions made by the semantic and pragmatic analysis associated with wh-questions for language acquisition, and Schulz and Roeper (2007) argue that children, when in the process of language acquisition, start with a default value [specific]6 for wh-questions, focusing on three main prerequisites to acquiring exhaustivity: discourse context, a hidden [+variable] and lexical markers for singular or exhaustive readings. The authors formulate two acquisition hypotheses: a pragmatic hypothesis and a formal feature hypothesis. According to the pragmatic hypothesis, the singular reading is natural, linked to an SO answer. Predicting that the child may give plural, nonexhaustive answers (Singularity > several (plural) > exhaustive). In the Formal Feature hypothesis stance, the child starts by acquiring a [specific] feature, which allows her a singular reading and only later will she acquire non-specificity, requiring [+variable], triggering the exhaustive reading associated with wh-pronouns and stopping the plural answer. We now have two approaches, making a different prediction as to the acquisition of exhaustivity. Schulz and Roeper (2011) argue that children start by analyzing the wh-pronoun as a constant, allowing singular readings, only later acquiring the value of the variable associated with the pronoun. This transition does not follow the cognitive hypothesis that would imply an intermediate plural stage, since children go from singular to the exhaustive stage. The path to wh-pronoun acquisition becomes unavoidably complex. Every stage is guided by Universal Grammar. The initial stage is

6

Schulz & Roeper (2011) updated their analysis, treating the wh-pronoun interpretation in the early acquisition stages as a constant and not as a variable.

Stéphanie Dias Vaz

61

dependent on prerequisites which are pragmatically and cognitively natural – the singularity stage. As argued by Schulz and Roeper (2011), children acquiring a language with an exhaustivity marker (e.g. alles in German) should acquire exhaustivity sooner. The linguistic differences demonstrate the role that explicit markers play in speeding up language acquisition Schulz and Roeper (2011) predict that exhaustivity acquisition should take place later in multiple wh-questions than in simple wh-questions due to the increased level of complexity of multiple wh-questions in mapping out the link between different sets.

3.2. Acquisition of clefts Heizmann (2007) has developed an experiment investigating the acquisition of exhaustivity in clefts and comparing the results with those found for the acquisition of exhaustivity in wh-questions for English speaking children. The author assumes that the results are similar due to both structures featuring a similar operator contained in the universal quantifiers. The author makes a connection between clefts, wh-questions and universal quantification structures, arguing that all of them have the same quantificational variable, involving a set of individuals and a given property. Furthermore, according to this author, exhaustivity in clefts is not triggered only by its focus meaning, but also due to a quantificational variable. This analysis is preferable because focus alone does not imply exhaustivity since other focus constructions are not associated with an exhaustive reading. This means that while focus is necessary, it is not the single factor in exhaustivity acquisition. From a truth value judgment task, applied to children aged between 3 and 5 acquiring English, the author concludes that the non-exhaustive answers disappear around 5 years of age and no significant change was found between clefts and wh-questions. Children recognize specificity for clefts, associating this more frequently with exhaustive readings in a pragmatic setting similar to its equivalent for simple wh-questions, where that reading is not necessarily triggered.

62

Child Acquisition of Exhaustivity: European Portuguese

3.3. Acquisition of universal quantification structures Quantifiers such as todos/todas (all/ every) and cada (each) are universal quantifiers, and, as such, are associated with exhaustive readings. However, they display unique properties. Todos (all/ every) allows for collective or distributive reading, while cada (each) can only have a distributive reading, thus being more complex to children and closer to pairing structures. In addition, todos is morphologically marked for number and gender while cada is not. The acquisition of the properties of quantifiers has been the object of many studies, using different methods, starting with Inhelder and Piaget (1964). As to the acquisition of universal quantification structures, several difficulties have been observed: i) difficulty in the exhaustive reading, ii) difficulty in assigning scope to the quantifier, iii) difficulty in deciding between collective or distributive reading (cf. Brooks and Braine 1996; Roeper, Strauss and Pearson 2006, among others). However, this matter will not be discussed in this article.

3.4. Summary As observed in several studies, exhaustivity is subject to linguistic development, being fully acquired around the age of 5. Pragmatics cannot fully account for this development given the grammatical properties necessary to mastering exhaustivity The three types of structures (wh-questions, clefts and universal quantification structures) are associated with exhaustivity. These structures do not have the same syntactic and semantic properties, though. Whquestions and cleft involve A- movement, which may imply a higher degree of complexity compared to other structures. Moreover, as Heizmann (2008) and Schulz and Roeper (2011) have shown, multiple wh-questions are more complex than simple questions, since they involve the mapping of two sets. Furthermore, according to Heizmann (2008), clefts involve comparing two groups. In universal quantifier structure the child must determine the scope of the quantifier and the type of semantic reading (collective or distributive) associated with each quantifier. In an overview of the literature it can be concluded that by the age of 5 children display target behavior.

Stéphanie Dias Vaz

63

4. Experimental study 4.1. Wh-questions 4.1.1. Hypotheses If exhaustivity really is a grammatical feature and not a pragmatic implication, subject to development, one would expect that, in answering wh-questions, multiple or simple, children should go from singular to exhaustive answers and very few plural non-exhaustive answers should occur. If processing two quantificational variables is more complex than processing just one, leading to the mapping between two sets, then children should display a higher degree of difficulty in answering multiple wh-questions than simple wh-questions. 4.1.2. Method In studying the acquisition of exhaustivity in wh-questions, a test developed by Petra Schulz for COST A33 was applied. This is a picture matching task showing several members of a family. The images were presented on a computer screen. For each image the individual had to answer a wh-question, either verbally or by pointing. The test had the following conditions: Simple subject wh-question with simple interrogative pronouns (quem/who) inducing exhaustive plural answers – 8 items; Paired wh-questions inducing exhaustive answers (quem – o quê/ who what) – 8 items; Triple wh-questions inducing paired exhaustive answers (quem – o quê – a quem /who – what – to whom; quem – o quê – onde /who – what - where) – 4 items; Control items – simple wh-questions (4 items), paired (4 items) and triple wh-questions (2 items) inducing singular answers. Before the test a vocabulary pre-test was carried out, to make sure that the children could recognize the objects mentioned during the test and an image, the objective of which being to introduce the subject to the characters.7 7

The test was also composed of a second part containing simple wh-question conditions marked with plural wh-pronouns quais (which). These items were meant to test whether the plural morpheme would elicit more exhaustive answers,

64

Child Acquisition of Exhaustiivity: Europeann Portuguese

The test w was applied inndividually and d all answers w were logged on n a record sheet. Next are some examplees of the items for each conddition: Simple wh-questions:

Quem é quee está a andar de d bicicleta? /Who / is ridingg a bike? Target answ wers: Menina / Girl Menino / Boy questions: Paired wh-q

Quem é quee está a pintar o quê? / Who is painting w what? Target answ wers: Pai – parrede / Father - wall Menino – cadeira / Moother - Chair

similar to what happens in German G with allles. This data w was not considered for the present work..

Stéphanie Dias D Vaz

65

Triple wh-q questions:

g what to Quem é que está a mostrar o quê a quem? / Whho is showing whom? Target annswers: Mãe – sool – menina / Mother M – su un – girl Menino ––borboleta – pai p / Boy – bu utterfly – fathher 4.1.3. Particcipants The test was applied to t 60 children n acquiring Poortuguese, witthout any diagnosed sppeech or cognnitive impairm ment. The chilldren were div vided into 3 groups, acccording to theeir age – 3, 4 or 5 years oldd. 20 adults were tested as a controll group. The following tab ble displays tthe distributio on of the participants according to their t age grou up:

3y n= 220 4y n= 220 5y n= 220 Adults n= = 20

Miniimum age

Maximum aage

Avera age age

2;11 4;2 5;1 18;8

3;11 4;11 5;11 43;8

3;5 4;5 4 5;4 5 26 6;4

Table 1: Paarticipants in the compreh hension test foor wh-questio ons

66

Child Acquisition of Exhaustiivity: Europeann Portuguese

4.1.4. Resullts

Chart 1: Ovverall resultss for wh-question by age ggroup The resuults in the Chhart above allo ow us to obseerve the existeence of a developmennt effect betw ween groups, especially e bettween the 4 year y olds and 5 year oolds and see thhat simple wh h-questions arre acquired eaarlier than multiple whh-questions.

Chart 2: Tyypes of answeers (including g the target an nswer) found for each group

Stéphanie Dias Vaz

67

The results as well as the analysis of the answers to control conditions (where some children answer with only the subject or the object), show that exhaustivity is acquired independently from being able to give paired answers and that the child can combine the two. It also shows that the ability to give paired answers (correctly recognizing all the wh-pronouns) is acquired later than the ability to give exhaustive answers (whether all subjects or all objects).

4.2. Clefts 4.2.1. Hypotheses If exhaustivity in clefts is conditioned by grammatical structure and not pragmatics, children should exclude non-exhaustive readings for clefts more often than for simple wh-questions. 4.2.2. Method To test whether children would associate exhaustive readings to clefts, a test was built using T. Heizmann (2007) as a guideline, which used a truth value judgment. The participants had to say whether a sentence said by a puppet matched the picture or not (picture verification task). The test consisted of the following conditions: Non-exhaustive subject clefts – 4 items Non-exhaustive object clefts – 4 items Exhaustive subject clefts – 4 items Exhaustive object clefts – 4 items Target (yes) subject control item – 3 items Target (yes) object control item – 3 items Target (no) subject control item – 3 items Target (no) object control item – 3 items In a pilot stage, the several clefts were tested and the pseudoclefts (Wh- be x) were chosen, because they best suited the context The item presentation order was random. The images below show an item used in the experiment. For a pair of pictures such as the one below the experimenter could pose several questions where the relevant element was the object.

68

Child Acquisition of Exhaustiivity: Europeann Portuguese

Experim menter: Olha! É um avô com m um pincel. E eu quero saaber o que ele vai fazeer! Oh! Olha uma cadeira, uma mesa e uma jarra. O avô vai pintar algum ma coisa. / Loook! Grandfath her has a brushh! And I wantt to know what he’s going to do!! Oh! Look, a chair, a ttable and a jar! The grandfather is about to paaint something g! [Non-exhhaustive objecct cleft condition] Puppet: Ah…o que o avô pintou u foi a mesa. RE: Não/ Oh… O The grandfather painted the taable. Ans: no [Exhausttive object cleeft condition] Puppet: Ah…o que o avô pintou foi a mesa e a cadeira. RE: R Sim / Oh… The grrandfather paiinted the chairr and the tablee. Ans: yes [Controll condition – taarget (yes)] Puppet: Ah...o avô pintou a cadeira. RE: Sim //Oh… the grrandfather painted the cchair. Ans: yees [Controll condition – taarget (no)] Puppet: Ah…o avô pintou p a jarra. RE: Não / O Oh… The grandfather painted the cchair. Ans: noo.

Stéphanie Dias D Vaz

69

For a seqquence of imaages like the fo ollowing, seveeral questionss could be posed, wherre the relevantt item would be b the subject..

Experimenter: Olha! É um avô, um ma menina e um m príncipe. E eu quero saber o que eles vão fazeer! Oh! Está aqui a uma caddeira, uma meenina, um príncipe e um m avô. Alguém m vai ficar dee pé. / Look! Itt’s a grandfath her, a girl and a princee. And I want to know whaat they are aboout to do. Oh. A chair, the girl, a prrince and a graandfather. Som meone will haave to keep staanding. [Non-exhhaustive subjeect cleft condiition] Puppet: Ah…quem ficou f de pé fo oi o príncipe.. RE: Não / Oh… O the prince is stilll standing. Anns: no. [Exhausttive subject clleft condition]] Puppet: Ah…quem fiicou de pé foi a menina e o príncipe. RE: R SIM / Oh… the prince and girl are a still standiing. Ans: yes [Controll condition – taarget (yes)] Puppet: Ah…a meninna ficou de péé. RE: SIM /O Oh… The girll was left standing. Anns: yes. [Controll condition – taarget (no)] Puppet: Ah…o avô fiicou de pé. RE E: Não / Oh… … The grandfa father was left standingg. Ans: no.

70

Child Acquisition of Exhaustivity: European Portuguese

These conditions were built on the following assumptions. With the non-exhaustive clefts (conditions (i) and ii)), the aim was to understand whether children recognize exhaustivity in clefts and whether they distinguish them from simple declaratives, like the ones presented in control conditions target yes (conditions (v) and vi)). If children were to answer affirmatively to both non-exhaustive clefts and non-cleft sentences (non-exhaustive), they would not be recognizing the exhaustive feature associated with clefts. If, on the contrary, they were to reject non-exhaustive clefts and accept the declarative non-exhaustive sentence, they would be recognizing the exhaustive feature. Rejecting both would lead to the fact that exhaustivity is a pragmatic occurrence and not something associated with a set of structures. In conditions (iii) and iv), corresponding to exhaustive clefts, no problems were anticipated. Control conditions target no (cf. (vii) and (viii)) were a means of ruling out a yes bias. Conditions (i) and (ii) (non-exhaustive clefts), compared to conditions (v) and (vi) (non-exhaustive declarative sentences), enabled the evaluation of whether children would reject non-exhaustive cleft answers, while accepting them for declarative sentences. If exhaustivity is triggered by the pragmatic context, no differences were expected. If exhaustivity is a grammatical feature, then children should reject non-exhaustive answers for clefts, but not for simple declarative sentences. If the pragmatic hypothesis holds between different types (clefts and simple declarative sentences), one would expect an asymmetry between those conditions, since adults would reject non-exhaustive answers for clefts, but would accept them for simple declarative sentences, something that does not happen with children. 4.2.3. Participants The experiment was applied to 20 adults, the control group, and 60 children divided into three groups of 20 children each. One group for 3 year olds, another for 4 year olds and a final group for 5 year olds, as demonstrated in the table below:

Stéphanie Dias Vaz

3 y n = 20 4 y n = 20 5 y n = 20 Adults n = 20

Minimum age 2;11 4;0 5;1 19;8

Maximum age 3;10 4;11 5;10 59;5

71

Average age 3;4 4;7 5;5 39;5

Table 2: Participants in the exhaustivity comprehension test for clefts All children were monolingual learners of European Portuguese with no diagnosed speech or cognitive impairment. 4.2.4. Results

Chart 3: % of right answers for non-exhaustive and exhaustive cleft conditions The Chart above allows us to verify that there is a development effect between 4 and 5 years of age: the 3 and 4 year olds accept non-exhaustive answers for cleft conditions, with around 40% of target answers, while 5 year olds reject them with a 78% rate of target answers. Adults have a 90% target rate. The results show a significant difference between both conditions. While children do not reject non-exhaustive readings for simple declarative sentences, with an acceptance rate above 90%, the acceptance of non-exhaustive readings for clefts is much lower in each age group,

72

Child Acquisition of Exhaustivity: European Portuguese

with a clear decline for 5 year olds for simple declarative sentences; with an acceptance rate above 90%, the acceptance of non-exhaustive readings for clefts is much lower in each age group, with a clear decline for 5 year olds for simple declarative sentences. These results also show that children do associate this specific trait with clefts, since keeping the same pragmatic context yields different results between the clefts and the simple declarative answers. The fact that adults reject non-exhaustive readings for simple declarative sentences also shows that they are more sensitive to the pragmatic context than children.

Chart 4: % of yes answers (failure) for non-exhaustive cleft conditions and yes answers (success) in control conditions target yes

4.3. Universal quantification structures 4.3.1. Hypotheses If children possess knowledge of universal quantification, recognizing that the universal quantifiers involve all the elements of a set, one would expect that they would reject non-exhaustive readings for the quantifiers todos (all/ every) and cada (each) (extra subject conditions). If children have trouble with limiting the scope of the quantifier, one would expect that they would wrongfully reject the items with an extra object. If children do have more trouble in interpreting distributive readings than in interpreting collective ones, one would anticipate worse results

Stéphanie Dias Vaz

73

with cada than with todos, and also worse results for distributive readings than for collective readings with todos. 4.3.2. Method To check the acquisition of exhaustivity in universal quantification structures, a truth value judgment comprehension task was used. Like the test for clefts, this test entailed judging the suitability of what the puppet said and what was in the picture (picture verification task) The following conditions were used: Quantifier – Todos i) Exhaustive structures with collective reading – 6 items ii) Exhaustive structures with distributive reading – 6 items iii) Exhaustive structures with collective reading and extra object – 3 items iv) Exhaustive structures with distributive reading and extra object – 3 items v) Non-exhaustive structure with collective reading – 4 items vi) Non-exhaustive structure with distributive reading – 4 items Quantifier – Cada i) Exhaustive structures with collective reading – 6 items ii) Exhaustive structures with distributive reading – 6 items iii) Exhaustive structures with collective reading and extra object – 3 items iv) Exhaustive structures with distributive reading and extra object – 3 items v) Non-exhaustive structure with collective reading – 4 items vi) Non-exhaustive structure with distributive reading – 4 items vii) Non-exhaustive structure with singular reading – 4 items viii) Control condition target yes – 8 items ix) Control condition target no – 4 items These quantifiers were chosen given their unique properties. Todos (all/every) allows for distributive or collective reading, while cada (each) allows only for distributive reading, making todos harder to process since it is morphologically more complex and it implies pairing structures. The item presentation order was randomized. An example of the items is given below.

74

Child Acquisition of Exhaustiivity: Europeann Portuguese

Todos (all/evvery) Puppet: Toddos os meninos estão a pin ntar um quadrro. / all / eveery boy is painting a piicture. i) [Exhaustivve and collecttive reading co ondition] Anss: Sim/Yes

g condition] A Ans: Sim/Yes ii) [Exhaustiive and distribbutive reading

Cada (each)) Puppet: Cadda menina esttá a segurar num n balão. / Each girl is holding h a balloon.

Stéphanie Dias D Vaz

75

iii) [Non-exhhaustive and collective c read ding conditionn] Ans: Não/N No

iv) [Non-exhhaustive and distributive d reading conditioon] Ans: Não//No

v) [Non-exhhaustive and siingular readin ng condition] A Ans: Não/No

These coonditions weree built with the following asssumptions: With connditions (i) annd (ii) an attem mpt was made to understand d whether children reccognize exhauustivity in uniiversal quanti fication structures and distinguish between colleective and disstributive readdings. To wh hat extent can the disstributive meeaning, requiring pairing, be harder than the collective m meaning?

76

Child Acquisition of Exhaustivity: European Portuguese

Conditions (iii) and (iv), involving an extra object, enable the determination of whether children have trouble with restricting the quantifier scope. If children wrongfully reject contexts where there is an extra object, this can be interpreted as a sign of difficulty in restricting the quantifier scope. If, on the other hand, they do not reject the extra object condition, condition (iii) with the quantifier todos is true, but with quantifier cada is false, since with the quantifier cada the collective reading is ruled out. Condition (iv) – extra object in distributive context – is true in the presence of both quantifiers. In non-exhaustive conditions, (v) and (vi), with both quantifiers, and (vii), with the quantifier cada, the intention was to understand whether children distinguish between both readings and which is easier. 4.3.3. Participants The test was applied to 60 children acquiring Portuguese, with no diagnosed speech or cognitive impairment. The participants were distributed among three groups – 3, 4 and 5 year olds. 20 adults made up the control group.

3 y n= 20 4 y n= 20 5 y n= 20 Adults n= 20

Minimum age 3;1 4;0 5;0 18;1

Maximum age 3;11 4;11 5;11 42;8

Average age 3;6 4;5 5;5 27;3

Table 3: Participants in the Exhaustivity associated with universal quantification structures test

Stéphanie Dias Vaz

77

4.3.4. Results

Chart 5: % of right answers in the non-exhaustive conditions and the exhaustive collective reading with todos The results show that: In non-exhaustive conditions, there is a development effect between 4 and 5 years of age: the younger children accept non-exhaustive readings with the quantifier todos (63.75% right answers for 3 year olds and 58.75% for 4 year olds). In Chart 6 the following can be observed: In non-exhaustive conditions, we once again observe a development effect between 4 and 5 years of age, and a difference in results for todos and cada. Some adults and 5 years olds accept non-exhaustive answers with cada. Cada forces the consideration of every element individually: the participants were not particularly sensitive to the extra subjects. Todos, on the contrary, forces the consideration of all the elements. Some adults and 5 year olds accepted non-exhaustive conditions with cada.

78

Child Acquisition of Exhaustivity: European Portuguese

Chart 6: % of right answers for non-exhaustive and exhaustive conditions in the distributive reading context with todos and cada

Chart 7: % of right answers for the non-exhaustive conditions (in addition to distributive, collective or singular reading) with the quantifier cada

Stéphanie Dias Vaz

79

Overall, these results verify that: i) There is a development effect between 4 and 5 years of age. ii) The younger children, 3 and 4 years old, do not treat cada as a universal quantifier, since they allow singular readings of cada. iii) Adults do not always interpret cada as an exhaustivity marker, which has to do with the fact that it forces individual consideration of the elements. iv) Comparing all the non-exhaustive conditions, every age group performed better in collective Reading conditions.

4.4. Comparing the acquisition of exhaustivity across structures Looking at the following Chart, the results for European Portuguese across structures can be seen (considering just the conditions which elicited target no answers – non-exhaustive readings associated with distributive readings with the quantifier todos), for simple wh-questions, multiple wh-questions (paired and triple) and clefts (target no – nonexhaustive contexts). It should be taken into consideration that the participants were not the same for the every test, except in the case of the wh-questions.

Chart 8: Results across structures

80

Child Acquisition of Exhaustivity: European Portuguese

It can be seen that simple wh-questions was the structure which yielded the best results, thus indicating that it is the structure where exhaustivity is acquired first. Comparing structures, it is possible to see how the development effect seems to play a role in the acquisition of exhaustivity.

5. Conclusion These results confirm the results already found in other studies for different languages: children seem to master exhaustivity around the age of 5. According to Schulz and Roeper 2011, at an early stage, children interpret the interrogative operator as a constant and not as a variable, associating it with singular answers. Only later, after recognizing the value of the operator, are they able to give exhaustive answers. Since they go from a singularity stage straight to an exhaustive stage in pragmatic contexts which require exhaustive answers, this shows how children are also able to grasp the pragmatic conditions of the exchange. Going back to the hypotheses in 4.1.1, it can be argued that, before these results, the first hypothesis was confirmed, seeing as in wh-questions children give very few plural non-exhaustive answers. The second hypothesis also holds, because it was seen how processing two quantificational variables was more demanding than processing just one, implying that in the mapping between two sets, children have more difficulty with multiple wh-questions (paired and triple) than with simple wh-questions (cf. Chart 1). Through the cleft experiment the conclusion can be reached that exhaustivity starts to arise around 3 and 4 years of age, but it is around 5 years of age that it reaches a nearly target state. The hypothesis in 4.2.1 is also satisfied, since it was verified that exhaustivity is conditioned by the grammatical structure and not by pragmatics, as children exclude non-exhaustive reading more often for clefts than for simple declarative sentences (cf. Chart 4). Exhaustivity does not depend solely on grammatical features, but on pragmatic properties as well (or maybe due to the experimental context), since adults rejected some simple sentences (9%) because they were not exhaustive, even though an exhaustive reading was not required. From the results from the universal quantification experiment, it was concluded that 5 year olds understand universal quantification, or they would not be able to recognize that universal quantifiers involve every element of a given set, nor could they reject non-exhaustive readings with quantifiers such as todos and cada in extra subject conditions.

Stéphanie Dias Vaz

81

Hypothesis 5, stating that children could have trouble in restricting the quantifier scope, was not confirmed, given that, overall, they performed well in extra object conditions. Finally, children had more difficulty with distributive readings (involving the consideration of each element) than with collective readings: they had worse results with the quantifier cada than with todos and in the distributive reading of todos than with the same quantifier, todos, but in a collective reading.

6. Final remarks It can now be concluded that: i) There is a development effect regarding the exhaustive feature associated with given linguistic structures, namely wh-questions, clefts and universal quantification structures with the quantifiers todos and cada; ii) The development effect is at its strongest point between 4 and 5 years of age across the structures; iii) Even though exhaustivity is grammatically determined, it interacts with pragmatics, which is visible for the control group, which is usually more sensitive to the experimental conditions than children; iv) Exhaustivity is acquired sooner in some structures than in others (simple wh-questions > universal quantificational structures > clefts > multiple wh-questions). This may be so due to the greater complexity of some structures when compared with others: multiple wh-questions require pairing in addition to exhaustivity.

References Brooks, Patricia J., and Martin D. S. Braine. 1996. “What do children know about the universal quantifiers all and each?”. Cognition 60: 235-268. Grebenyova, Lydia 2006. “Multiple Interrogatives: Syntax, Semantics, and Learnability”. PhD diss., University of Maryland. Heizmann, Tanja. 2007. “Child Acquisition of Exhaustivity in Clefts”. In Proc. of the 31st Annual BUCLD, edited by Heather Caunt-Nulton, Samantha Kulatilake, and I-hao Woo, 298-309. Cascadilla: Somerville, MA. —. 2008. “Children's Acquisition of Exhaustivity in Clefts”. Talk at IASCL 11, International Congress for the Study of Child Language

82

Child Acquisition of Exhaustivity: European Portuguese

Symposium, Acquiring the Semantics and Syntax of Presuppositions. University of Edinburgh, UK. Inhelder, Barbel, and Jean Piaget. 1964. The Early Growth of Logic in the Child. London: Routledge, Kegan and Paul. Roeper, Tom, Petra Schulz, Barbara Z. Pearson, and Ina Reckling. 2007. “From Singleton to Exhaustive: the Acquisition of Wh-”. In Proc. of the 3rd Conference on the Semantics of Underrepresented Languages in the Americas, edited by Becker, and McKenzie, 87-102. University of Massachusetts Occ. Papers in Linguistics 33, Amherst, MA. Roeper, Tom, and Petra Schulz. 2011. “Acquisition of exhaustivity in whquestions: A semantic dimension of SLI?”. Lingua 121: 383-407. Van Rooij, Robert. 2004. “Questions and relevance”. In Questions and answers. Proceedings of 2nd CoLogNET-ElsNET Symposium, edited by R. Bernardi and M. Moortgat, 96–107. Vaz, Stéphanie. 2010a. Relatório de Bolsa de Integração na Investigação (BII) em Linguística Comparada. Universidade Nova de Lisboa. —. 2010b. Relatório de Estágio curricular em Ciências da Linguagem. Universidade Nova de Lisboa. —. 2012. “Aquisição de Exaustividade em Crianças Falantes de Português Europeu”. MA diss., Universidade Nova de Lisboa.

THE ACQUISITION OF THE PRONOMINAL SYSTEM IN PORTUGUESE L2 BY BASIC EDUCATION STUDENTS IN CAPE VERDE: A BILINGUAL EXPERIENCE ANA JOSEFA CARDOSO CENTRO DE LINGUÍSTICA DA UNIVERSIDADE NOVA DE LISBOA FUNDAÇÃO PARA A CIÊNCIA E TECNOLOGIA1

Abstract The project described in this paper aims to investigate the acquisition of the pronominal system in Portuguese as a second language (L2) by Cape Verdean children in the early stages of basic education. The data collection will be carried out in Cape Verde in two classes that will be part of a Bilingual Education pilot project and two other classes that will have traditional teaching exclusively in Portuguese. The data collected will be analyzed within a comparative perspective, to verify whether the results of children from bilingual classes differ from those in the control classes. The analysis of the data collected will be used to test hypotheses about the process of L2 acquisition, access to UG, role of representations of the L1 in the interlanguage and whether explicit teaching of the L1 in the context of bilingual education facilitates the acquisition of the L2, in particular as regards the pronominal system, which includes clearly distinct syntactic structures and parametric values. Keywords: Pronominal system, Cape Verdean, Portuguese, Bilingual education, L2 acquisition, Interlanguage.

1

Grant reference – Fundação para a Ciência e Tecnologia: SFRH/BD/72785/2010. This work was supported by the National Funding Body – FCT through Strategic Project PEst-OE/LIN/UI3213/2011.

84

The Acquisition of the Pronominal System in Portuguese L2

1. Introduction This paper is about a study which aims to investigate the acquisition of the pronominal system in Portuguese as a second language (L2) by Cape Verdean children in the first stage of basic education (first and second years of schooling). The data will be collected from two classes that will be part of a pilot project of Bilingual Education and two other classes that will have traditional teaching exclusively in Portuguese. Students are about 6 years old. Through analysis of the data, we intend to investigate (i) what is the role of syntactic representations of the mother tongue (L1) – Cape Verdean – in the acquisition of the L2 and (ii) whether, as expected, bilingual teaching allows for greater success in the acquisition of the L2. This study should also lead to a better understanding of the pronominal system in Cape Verdean and Portuguese. The paper is organized as follows: it starts with a description of theories of L2 acquisition which are relevant to the study, and then introduces the concept of bilingualism discussing the acquisition of bilingual status and the notion of diglossia. Next, it presents a brief description of the Cape Verdean pronominal system and of the Portuguese pronominal system, and describes the linguistic situation in Cape Verdean schools. Finally, it presents the pilot experiment in bilingual education and the proposed study.

2. Theories of L2 Acquisition According to the theoretical framework of Principles and Parameters (Chomsky 1981; Chomsky 1986), the acquisition of the mother tongue assumes the existence of an innate language faculty; language is acquired by processing the input provided by the environment surrounding the child. This innate faculty includes a Universal Grammar (UG), composed of principles and parameters, which determines and restricts the possible formats of grammars of natural languages. Hence, the principles that govern all languages are universal and correspond to what remains unchanged in the different systems, and the parameters correspond to the options for variation that distinguish different languages. According to this theory, the faculty of language is inherent to all human beings, who are genetically predisposed to acquiring a particular language, when exposed to it. It is assumed that languages are not learnt explicitly, because, even without explicit instruction, children recognize

Ana Josefa Cardoso

85

very early that some constructions are not possible and the “mistakes” that are produced in the early stages of language acquisition do not violate the principles of UG. UG allows learners of a language (either L1 or L2) to acquire grammatical properties that go far beyond the input to which they are exposed. For those who adopt this view (Schwartz and Sprouse 1994; Schwartz and Sprouse 1996), the acquisition of L2 is also a natural process, allowing learners to acquire grammatical properties not represented in L1. This position, however, is not consensual. There is a debate regarding the degree of access to UG during the acquisition of L2. During the process of acquiring a second language, the learner constructs new grammatical systems based on hypotheses which will produce changes according to the successive approximations to the target language, creating a transitory grammar, different both from the L1 and from the L2, which s/he will evaluate and restructure when s/he acquires new data, giving rise to various stages of interlanguage (Selinker 1972), which Xavier and Mateus (1990, 215) define as being “each of the grammars constructed by an individual in the process of acquiring a language.” With respect to this topic, there are three different positions. The first argues that the second language does not behave as a natural language, and therefore does not suffer any interference from UG. The individual relies on other sources, such as the existing knowledge of L1 and general learning strategies that allow the processing of formal abstract systems (Bley-Vroman 1997). According to this view, L2 acquisition and L1 acquisition are different processes. The loss of UG access after the critical period, which some authors consider to happen between 5 and 9 years (Gass and Selinker 2001, 101) and others assume not to occur until puberty (Mitchell and Miles 2004, 88), makes it difficult or even impossible to learn a second language with a degree of success that resembles that of a native speaker. The second position considers that there are no differences between the acquisition of L1 and L2 with regard to access to UG. The learners have full access to UG. It disagrees with the possibility of a critical period and argues that learners continue to use the principles and parameters available in UG. In the initial stage of the acquisition of L2, there is a transfer of parameter values represented in L1 and, in subsequent stages these parameters are reset in response to the properties of the new language, the L2 input. Proponents of this position (Schwartz and Sprouse 1994; Schwartz and Sprouse 1996) may admit that the acquisition of L1 and L2 are different processes, but argue that L2 learners have access to UG, even

86

The Acquisition of the Pronominal System in Portuguese L2

if they are unable to reset some parameters. For these authors, the evidence that there is access to UG would be that learners are able to acquire functional categories not existent in their L1, do not develop a grammar that does not respect the format imposed by UG and display knowledge that goes beyond the input received. The third encompasses those who argue that there is only partial access to UG in the acquisition of L2. For example, Tsimpli and Roussou (1991) argue that the parameters associated with the functional categories, where L1 and L2 are different, are not fully acquired. They propose, moreover, that access to UG occurs through the L1 and that it is the parameters already established in the L1 that serve as a basis for the development of the L2. However, Hawkins and Chan (1997), Hawkins and Hattori (2006) and Tsimpli and Dimitrakopoulou (2007) argue that, although the L2 grammars are partially deficient, some functional traits can be acquired even when they differ in L1 and L2. In this case, if there is a partial deficit, some parametric properties may be affected; learners will then use other learning mechanisms to accommodate the properties which differ from those in their mother tongue to their mental representations. L2 acquisition is always related to bilingualism, the type of bilingualism depending on the circumstances in which the languages are learned.

3. Bilingualism Currently, the majority of people speak more than one language. Thus, bilingualism should not be considered as a state of exception, as Wei (2010, 3) states: “For many people, bilingualism and multilingualism are a fact of life and not a problem”.

3.1. The acquisition of bilingual status Bilingualism can be individual or social. Bilingualism exists when the presence of two languages within a society has linguistic and social consequences. Social bilingualism is always related to individual bilingualism, since social conditions have a strong influence on the bilingual acquisition process and vary from individual to individual. Generally, a bilingual is someone who, in addition to his/her mother tongue, has an equivalent language and communicative competence in another language and is able to use either in all circumstances with similar efficacy.

Ana Josefa Cardoso

87

However, many researchers also consider as bilinguals those who have varying degrees of proficiency in two languages and incorporate the development and acquisition of second language in studies of bilingualism (Butler and Hakuta 2006, 115). Wei (2010, 4), for example, considers that: A multilingual individual is anyone who can communicate in more than one language, be it active (through speaking and writing) or passive (through listening and reading). Multilingual individuals may have become what they are through very different experiences: some may have acquired and maintained one language during childhood, the so-called first language (L1), and learned other languages later in life, while others have acquired two or more first languages since birth.

According to Marques (2003), bilingualism implies a mental organization that distinguishes three different types of relationships between two linguistic systems. According to Weinreich (1953, cit. Butler Hakuta 2006, 115), this relationship can be coordinated, subordinate or compound. In a coordinated relationship, the two languages are acquired at the same stage of the child’s language development, in different contexts, interacting with specific groups of native speakers, thus creating ideal conditions for developing two separate language systems and language skills which are equivalent to those of monolingual native speakers of each language. In a subordinate relationship, a language dominates the other; it is usually the first language that plays the dominant role. The second language is learned in the context of the first language, usually through translation, and will be subject to strong influence from the mother tongue. When there is a compound relationship, the two languages have a system of fusion: each of them is used indiscriminately in all contexts, and interference may occur in both directions. The type of bilingualism acquired is determined by the context, age and order of acquisition. Thus, the pace and the characteristics of bilingual child language development depends on the acquisition of the two languages. In simultaneous acquisition, the child acquires both languages identically to what happens with the monolingual speaker, which does not imply that the mastery of two languages is always identical both in quantity and in quality. The increasing awareness of the differences between the two languages leads to a true separation of the phonological and syntactic systems at about seven years of age (Sim-Sim 1998). In these cases, the child is able to use the alternating codes (code-switching)

88

The Acquisition of the Pronominal System in Portuguese L2

without any difficulty, distinguishing the systems of the two languages in contact with no intersection of traits. In sequential acquisition, the child already has the basic structures of the first language when he or she starts the process of second language learning. This is the case with children who are exposed to different languages at home and at school. Individual bilingualism is developed according to the need to use each of the languages. A bilingual does not have equal fluency in the two languages if the everyday circumstances do not require it. The bilingual competence of an individual can be only passive in relation to one of the languages, without sufficient competence for expression. Passive competence is “the ability to understand everything that a person says, without being necessarily able to speak the language used ... is limited to decoding of utterances received.” (Hagège 1997, 224). This situation often happens with some children of immigrants who use the language of the host country and who, even when parents speak to them in their mother tongue, respond in their favorite language. Active competence, on the other hand, is that which develops both for production and for comprehension; the individual is able to understand and to express himself/herself in two languages. In Cape Verde, we have a situation of diglossia which makes the learning of the L2 difficult, as many Cape Verdeans have only passive competence in Portuguese.

3.2. Diglossia Two languages or two different linguistic codes can coexist in the same society, and each of them has a particular communicative function. This form of social bilingualism was named diglossia by Ferguson (1959). Fishman (1967) states that the language which is more familiar, has less social prestige, and is used in everyday relationships, in popular culture, within the family and in the community is considered as the Low language, as it is a common or colloquial language. The language used in religion, in teaching, in literature and in official situations is considered as the High language, given that it is more formal and scholarly. The Low and High languages complement each other to satisfy all the communication needs in situations of diglossia (Garmandi 1983: 133). The existing bilingualism in Cape Verde is characterized by a situation of diglossia. Cape Verdean Creole, which is the mother tongue and national language, is used above all in informal communication situations and Portuguese is the official language, used in all formal situations.

Ana Josefa Cardoso

89

However, this situation of diglossia is gradually becoming less pronounced, because the use of Cape Verdean Creole has significantly increased and nowadays it already has a place in politics (used in parliament and election campaigns), the media (used in some TV and radio programs and newspapers) and culture; however, it still lacks a place in education, as a teaching language (Pires 2010, 143). Its use in the educational context is only as a resource to support the learning of the Portuguese language (Monteiro 2009). According to Lopes (2011), Cape Verdean Creole is the favorite language to communicate with on the internet, and through mobile phone messages, emails and Facebook. Cape Verdean bilingualism is only individual and is characterized by sequential acquisition: those who are competent in Portuguese, as a rule, have learned it at school and in a formal context. There are some families that use Portuguese at home and who also learned this language in an informal context, but this is not very common. Duarte (1998, 132) says that “a high percentage of school-age children come from non-Portuguese-speaking environments (...) In addition, these children, once they leave primary school, forget Portuguese, not only through lack of knowledge of the language, but through lack of practice.” The education system is monolingual in Portuguese (L2), excluding the mother tongue. There is no promotion of biliteracy, although, socially, Cape Verde is considered a bilingual country. This language policy goes against the recommendations of UNESCO. UNESCO suggested in 1953 that there were advantages in using the child’s mother tongue to teach initial literacy. This was the position of the early proponents of bilingual education, especially for children of linguistic minorities (Modiano 1968). The idea was not to develop biliteracy per se, but rather to advance literacy in the dominant societal language by teaching children to read in a language they understood.” (Garcia et al. 2007, 210)

According to Gomes (1998, 44), 75% of Cape Verdeans are speakers of two languages and only 25% speak Creole exclusively. However, linguistic proficiency in Cape Verdean is far superior to that in Portuguese. Most speakers are not very fluent in Portuguese; they usually do not use it on a daily basis, communicating predominantly through the medium of their mother tongue. Even those Cape Verdeans that are fairly proficient in the two languages have a better performance in their mother tongue than in Portuguese. Veiga (2004) considers that Cape Verdean bilingualism is under construction. It is necessary to generalize the use of the two languages.

The Acquisition of the Pronominal System in Portuguese L2

90

The Portuguese needs to conquer non-formal communication spaces, and Cape Verdean needs to exploit and conquer formal communication spaces, a process which is still underway. For Portuguese to be learned to a higher level of proficiency, it becomes necessary to develop better grammatical awareness of both languages, due to the differences between them - in particular, of the pronominal system that has already been identified as being difficult to acquire.

4. The Cape Verdean pronominal system The pronominal system of Cape Verdean (Santiago variety) displays the following properties (Veiga 1995; Veiga 2000; Baptista 2002; Pratas 2004; Pratas 2007). A. Cape Verdean lexicalizes the three kinds of pronominal forms proposed by Cardinaletti and Starke (1994): emphatic forms (NP), free forms (with a reduced structure for maximum projections – which have ambiguous behavior, behaving sometimes as NP, and others as head – No.) and clitic forms. See Table 1 and the examples (1) – (3). Pronominal forms Persons st

Plural

Singular

1

nd

Emphatic forms

Free forms

Ami

Mi

Clitic forms Subject N

Object -m bu / -u

2 (informal)

Abo

Bo

Bu

2nd (formal, masc.) 2nd (formal, fem.) 3rd (fem., masc.) 1st pl 2nd pl 3rd pl

Anho Anha Ael Anos Anhos Aes

Nho Nha El Nos Nhos Es

Nhu

Table 1: Cape Verdean Pronominal forms (1) a. Ami é bon. (emphatic form) IEmph am good “I am good.”

E Nu

-l -nu -s

Ana Josefa Cardoso

91

(2) a. Mi é balenti. (Free form) IFree am brave “I’m brave.” (3) a. N kume pon.(Clitic forms) ICL ate bread “I ate bread.” b. E da-m un beju. he/sheCL gave me a kiss S/he gave me a kiss.” As can be seen in table 1, there are no clitic forms for 2nd sg formal, 2nd pl and 3rd pl. If we apply the tests for clitichood, we can verify that these forms do not pass all of them. The only condition that they obey is the following: no element can intervene between the clitic and the verb. B. Cape Verdean is a partial null subject language, and null referential subjects are prohibited in matrix sentences (Pratas 2004; Pratas 2007; Costa and Pratas 2012; Alexandre, Santos and Duarte 2012; Cabredo Hofherr 2013). The subject has to be phonetically realized. (4) a. *Sta duenti. am sick ‘* Am sick.’ b. N sta duenti. ICL am sick “I am sick.” The subject pronoun is also required in embedded sentences even when it is coreferential with the subject of the matrix sentence: (5) Djon fla-m ma *(e) ta bai kasa. John told -1SG that 3SG TMA goes home “John told me he goes home.” The status of Cape Verdean regarding the null subject parameter has generated some controversy among researchers. Baptista (1997) Batista (2002) and Baptista and Bayer (2013) considers that this is a null subject language, while others, like Pratas (2002), Pratas and Costa (2012)

92

The Acquisition of the Pronominal System in Portuguese L2

Alexandre, Santos and Duarte (2012) and Cabredo-Hofherr (2013), argue that the Cape Verdean language is a partial null subject language. According to Baptista and Bayer (2013, 6), in Cape Verdean, null referential subjects may occur in root, matrix clauses. Following Frascarelli (2007), the authors examined occurrences in the 1st, 2nd and 3rd persons and they argue that “Null subjects are always interpreted in relation to the closest Aboutness-Shift Topic or Familiar (overt or silent) Topic”, as illustrated in example (6). (6) E bo, un sumana di trabadju, nu ta pagada dozi merés, man tanbê, kel dozi merés, [bu]i pode baba Somada faze konpra. Ki proi ta baba propi. proi Ba faze konpra bu bem ku bu balai di konpra, purkê, tudu kuza e baratu. ‘And you, a week of work, we were paid twelve cents, but as well with those twelve cents [you]i could go to Somada to do grocery shopping. That proi would go really. proi would go and do the grocery shopping you come back with your basket of groceries, because everything is cheap.’ Baptista and Bayer (2013, 5), example (26) On the other hand, Costa and Pratas (2012) argue that null referential subjects cannot occur in root contexts (7); however, null subjects can occur in the following situations: (i) “null subjects can be expletive in root contexts, in expletive associate environments and with weather-predicates” (8); (ii) “null subjects can be indefinite in root contexts (9) …, Cape Verdean allows embedded null subjects, obligatorily identified by a matrix quantified antecedent or a wh-antecedent.”(10). (7) *(N) papia txeu. (Costa and Pratas 2012 (11a)) 1SG speak much (8) Sata txobe na Lisboa. (Costa and Pratas 2012 (18)) PROG rain in Lisbon ‘It’s raining in Lisbon.’ (9) Li pode fumadu. (Costa and Pratas 2012 (19)) LOC may smoke.PASS ‘People may smoke here’ (10) Kenhai ki ta atxa ma Øi/*j kore faxi labanta mo. (Costa and Pratas 2012 (26a))

Ana Josefa Cardoso

93

who COMP TMA think COMP run fast raise hand ‘Whoever thinks he ran fast raise (their) hand.’ C. Regarding the object position there are three different situations: C1. Object clitics are incompatible with verbs that display the tense post verbal affix {-ba} (past), according to Baptista (2002), Pratas and Salanova (2005) and Pratas (2007). (11) a. N ta odjaba. ICL TMA saw “I saw.” b. N odja-l. ICL saw - him “I saw him.” c. N odjaba el. / * N odjaba-l. ICL saw himFree /ICL saw himCL “I saw him.” C2. Double Object constructions (DOCs) always display the order Verb-Indirect Object-Direct Object (V-IO-DO) (Fiéis and Pratas 2006). This occurs in all structures, even when the complements are realized as pronouns: V-NP-NP (12), V-NP-Pron (13), V-Pron-NP (14), and V-PronPron (15) (see also Alexandre 2009): (12) a. N da Manel dinheru V - NP - NP ICL gave Manel money “I gave (the) money to Manel.” b. *N da dinheru Manel (13) a. N da Manel el. V - NP - Pron ICL gave Manel 3SG “I gave it to Manel b. *N da-l Manel

94

The Acquisition of the Pronominal System in Portuguese L2

(14) a. N da-l dinheru V-Pron - NP ICL gave-3SG money “I gave him (the) money.” b. *N da dinheru el. (15) a. N da-l el V- Pron-Pron ICL gave-3SG 3SG “I gave it to him.” C3. The reflexive expression is constructed with a possessive + kabésa (‘ head ‘); it is used only when it is necessary to emphasize the reflexive meaning. In Cape Verdean there is no form corresponding to Portuguese se (‘himself’). However, with certain verbs, the expression kabésa is obligatory for a reflexive reading. See example (16). (16) Djon mata si kabésa. Djon kill POSS.3SG self “Djon kill himself.” In other cases, such as in example (17), the use of the expression kabésa is optional; it is used only when there is a need to emphasize the reflexive meaning. (17) Pedru laba (si kabésa) Peter washed “Peter washed himself.” The properties described do not occur in Portuguese, the lexifier of this Creole language.

5. The Portuguese pronominal system In Portuguese: (i) there are no pronominal emphatic forms; (ii) null referential subjects are allowed and, in certain contexts, are even preferred; (iii) clitic objects are not incompatible with tense affixes; (iv) there are no DOCs (ditransitive contexts are always V-DO-preposition-IO, except with pronominal objects are (with restrictions of person); (v) reflexive contexts

Ana Josefa Cardoso

95

are marked by a pronominal se (for descriptions of the grammar of Portuguese see, among others, Mateus et al. 2003). The pronominal system of Portuguese displays the following properties, which differ from Cape Verdean: A. The personal pronouns in Portuguese can be strong or clitic. Pronominal clitic forms occur in the complement position of verbs (see Table 2).

1 2nd 3rd

eu tu, você ele/ela

Syntactic Functions Direct Indirect Object Object me me,(a) mim te te, (a) ti o/a; se lhe; (a) ele/ela

1st 2nd 3rd

nós vós, vocês eles/elas

nos vos os/as; se

Person

Plural

Singular

st

Subject

nos, (a) nós vos (a) vós lhes, (a) eles/elas

Oblique Object mim; migo ti; tigo si, sigo, ele /ela nós, nosco vós, vosco si, sigo, eles/elas

Table 2: Portuguese Pronominal forms B. Portuguese is a pro-drop language. Subject personal pronouns can be omitted. See example (18). (18) (EU) gosto de chocolates. (I) like1SG Prep chocolates Both sentences are correct, because Portuguese is a null subject language and admits the two situations. In Cape Verdean, on the other hand, a sentence with a similar construction to (18) is ungrammatical, as shown in (19). This example is similar to that shown in (4a): (19) *(N) Gosta di xukulati. 1SG like Prep chocolates “I like chocolates.”

96

The Acquisition of the Pronominal System in Portuguese L2

C. Portuguese only has object clitics. C1. Clitics are compatible with all verb forms; what may vary is their position in the sentence (enclitic (20a-b), proclitic (20c), mesoclitic (20d)). (20) a. Eu vejo-o. 1SG see-3SG “I see him.” b. Eu via-o. 1SG saw-3SG “I saw him.” c. Eu não o vi. 1SG Neg 3SG saw “I didn’t see him.” d. Eu vê-lo-ei. 1SG see-3SG-FUT.1SG ‘I will see him.’ C2. Ditransitive constructions exhibit the unmarked order verb-direct object-indirect object (21), except when complements are pronominal with restriction of person (22a). (21) Eu dei uma flor à Maria. V- DO - DI 1SG gave a flower Prep Mary ‘I gave a flower to Mary.’ (22) a. Eu dei-lhe uma flor. V-Pron-DO ‘I gave her a flower.’ 1SG gave-3SGCLa flower b. *Eu dei uma flor lhe. 1SG gave a flower 3SGCL I gave a flower her

Ana Josefa Cardoso

97

C3. The pronouns that mark reflective contexts inflect for person and, in the case of 1st and 2nd person, also inflects for number (me, te, se, nos, vos, se). (23) a. O Pedro lavou-se. Peter washed 3SG Pron 3SG “Pedro washed himself.” b. Eu lavei-me. I washed 1SG Pron 1SG “I washed myself.” c.

Eles lavaram-se. They washed 3PL Pron 3PL “They washed themselves.”

The comparison between Cape Verdean and Portuguese regarding the grammatical properties that are illustrated in the examples above can be more easily attained by children through the metalinguistic awareness that bilingual teaching can provide. Otherwise, they have few opportunities to reflect on the language in order to develop their linguistic awareness.

6. The Linguistic situation in Cape Verdean schools In Cape Verde the language of instruction has been exclusively Portuguese; Cape Verdean has been virtually excluded. We say ‘virtually’ because it is known that teachers often use Cape Verdean to explain Portuguese to the children in the classroom. The high rates of school failure and the low performance in Portuguese are evidence of students’ fragilities in this curricular area, which is essential for the learning of all subjects, since these are taught in Portuguese. In spite of being the only language used in teaching, the use of Portuguese in situations of communication is far from natural, because language skills are developed almost exclusively between the four walls (in the classroom) and through situations created for that purpose, which are unrealistic and limit the acquisition of communication and language skills. In this way, the focus is on learning to read and write, giving little importance to students’ creativity.

98

The Acquisition of the Pronominal System in Portuguese L2

The acquisition of the Portuguese language has been deficient and ineffective. According to a recent study on education in Cape Verde (Gomes 2008, 408): Children reveal difficulties in learning Portuguese, at the level of orality, particularly in the pronunciation of words, and employ vocabulary lacking in diversity, given that their active vocabulary is restricted. In writing they have difficulties in orthography, agreement and punctuation. The texts reveal mainly problems with textual cohesion; they are ideologically confused and speech has hardly any logical sequence. Portuguese arouses little interest in children... (My translation)

it is believed that bilingual education can be positive for Cape Verdean children. Cognitive advantages such as metalinguistic awareness (Bialystok 2006; Bialystok 2007; Garcia 2010) divergent thinking, communicative sensitivity and ability to learn multiple languages have already been proved and could increase their competence in Portuguese. There are social advantages too, such as increased cultural competence, which make bilingualism an advantage for all children. Many experiments around the world, such as the Jamaican Bilingual Education Project (Carpenter and Denovish 2010, 173), have shown good results and have proved that this is the best practice for societies where there are two languages in people’s daily life. The only bilingual education experience in living memory in Cape Verde was the Experimental Bilingual Literacy Project, which started in 1987 and lasted 5 years. It was originally an adult literacy project in Creole, supported by Swiss Cooperation which followed UNESCO guidelines (IV International Conference on adult education, 1985) which specify the right of all individuals to being taught in their mother tongue. In 2010, a review of the Bases of the Cape Verdean Educational System took place, which gave greater relevance to language learning. Legislative Decree No. 2/2010 (May 7), provides for the introduction of the mother tongue in school education. Considering this background, I propose the implementation of a Bilingual Education experiment.

7. The study: a Bilingual Education experiment in Cape Verde The main goal of this project is to investigate the acquisition of Portuguese as a second language by two groups of children in two elementary schools, one in Praia, the capital of Cape Verde, and the other in Flamengos, in the interior of the island of Santiago. In each of the

Ana Josefa Cardoso

99

schools, one of the classes will be subject to monolingual teaching in Portuguese (the control class) and the other (the bilingual class) will be subject to a pioneering experience in bilingual teaching in Portuguese and in Cape Verdean, planned specifically for this purpose. The Ministry of Education and Sports of Cape Verde authorized the implementation of this project, has committed itself to lending its support and has appointed a monitoring team to follow the process. Data will be collected and analyzed with the aim of investigating the nature of the L2 acquisition process, namely regarding access to Universal Grammar and the role L1 representations in the construction of an interlanguage grammar. The focus will be on the pronominal system, which, as described above, displays different morphological and syntactic properties, and depends on different parametric values in Cape Verdean and in Portuguese. This will follow the Bilingual Education model proposed as the dual language education or Two-Way Bilingual Education (Lindholm 2001; Torres-Guzman et al. 2002), where Cape Verdean will be used alongside Portuguese, as a language of instruction to teach other subjects of the curriculum such as mathematics, science, etc. Teachers will follow the national curriculum, but the contents will be worked on in both languages. The Cape Verdean teacher will have a specific language program, but s/he will work on other curricular areas using the mother tongue. This model promotes biliteracy, which implies not only knowledge of more than one language and two systems of graphic representation, but also knowledge of cultural aspects which influence the structuring and use of written speech”, in its multiple genres, the learning of which assumes explicit instruction (Garcia 2010, 349). Biliteracy reinforces the effects of bilingualism, in terms of capacities of abstraction and explicit language awareness (Bialystok 2001, Kenner 2011) and control over the choice and the use of each of the languages.

7.1. The goal of the study Through the implementation of a bilingual education project in a Cape Verdean context, in accordance with international guidelines, and through our investigation into the acquisition of the pronominal system in L2 Portuguese, the intention is to achieve the following goals: Provide explicit instruction on Cape Verdean and Portuguese grammar in the classroom, thus promoting language awareness and biliteracy.

100

The Acquisition of the Pronominal System in Portuguese L2

Investigate the behavior of children from bilingual classes regarding the properties of the pronominal system which are different in the two languages and determine whether these options are fully acquired (as predicted by full access hypotheses). Understand the role of the syntactic representations of Cape Verdean (L1) in the acquisition of Portuguese (L2). Demonstrate, through data relating to the acquisition of the pronominal system in L2 Portuguese that formal and explicit teaching in the L1 increases proficiency in the L2 and improves academic success.

7.2. Research issues The elicitation and collection of data, both in the bilingual and in the control groups, will focus on the pronominal system, chosen due to the fact that it is determined by parametric values which clearly distinguish the two languages. Following the Principles and Parameters model, we will seek to answer the following questions: How do children behave regarding the null subject parameter (negative value for referential subjects in Cape Verdean, positive in Portuguese)? What effect does tense marking on the verb have on the realization of the object pronoun? Do learners use the strong pronominal form free; avoiding the clitic form (this is obligatory in Cape Verdean, but impossible in Portuguese)? How do children behave in the case of ditransitive verbs (Double Object constructions are obligatory in Cape Verdean, but impossible in Portuguese)? What strategies do children use in reflexive constructions (in Portuguese, they are usually built with a clitic which inflects for person and number; such a pronominal form does not exist in Cape Verdean)?

8. Conclusion In line with the Principles and Parameters theory, this study tests different hypotheses regarding UG access and L1 influence, by investigating L2 acquisition of the Portuguese pronominal system by children who are native speakers of Cape Verdean. Showing (as we expect) how important it is for L2 learners to also learn to read and write

Ana Josefa Cardoso

101

in their mother tongue, the proposed bilingual education experiment may contribute to giving impetus to a new model of education in Cape Verde, thus creating a new generation of real bilingual citizens. The study is innovative, given that there is no record of any previously implemented bilingual education program in Cape Verde in basic education. In addition to the more general objectives concerning the advantages already recognized in bilingual education, academic success and performance improvement in Portuguese, the proposed bilingual education project fosters the explicit teaching of the grammar of the L1 as a facilitator of the learning of the L2, contributing to an improvement in Portuguese language proficiency, in particular the pronominal system, which is considered difficult and is known to be acquired late, even for native speakers (Guasti, 2002). This study will help to pave the way for the widespread implementation of the Cape Verdean language in schools.

References Alexandre, Nélia. 2009. “Wh- constructions in Cape Verdean Creole: Extensions of the copy theory of movement”. PhD diss., Universidade de Lisboa. Alexandre, Nélia, Ana Lúcia Santos, and Inês Duarte. 2012. “Personal infinitives and the status of Capeverdean as a partial null subject language”. In Handout Formal approaches to Creole Studies, Lisbon. Auer, Peter and Li Wei. 2007. Handbook of Multilingualism and Multilingual Communication. Berlin/New York: Walter de Gruyter GmbH and Co. KG. Baptista, Marlyse. 1997. The Morpho-syntax of Nominal and Verbal Categories in Capeverdean Creole. PhD diss., Cambridge: Harvard University. —. 2002. The Syntax of Cape Verdean Creole, the Sotavento varieties. Amsterdam/Philadelphia: John Benjamins. Baptista, Marlyse, and Rachel Bayer. 2013. “Licencing Argument Drop in Cape Verdean Creole”. Handout SPCL Meeting. Lisboa. Bialystok, Ellen. 2001. Bilingualism in Development: Language, Literacy, and Cognition. New York: Cambridge University Press. —. 2006. “The impact of bilingualism on language and literacy development”. In Tej K. Bhatia, and William C. Ritchie. The Handbook of Bilingualism, 577-601. Oxford: Blackwell Publishing.

102

The Acquisition of the Pronominal System in Portuguese L2

—. 2007. “Cognitive Effects of Bilingualism: How Linguistic Experience Leads to Cognitive Change”. The International Journal of Bilingual Education and Bilingualism 10(3): 210-223. Bley-Vroman, Robert. 1997. “Features and patterns in foreign language learning”. Paper presented at Second Language Research Forum. Michigan State University. October 1997. Butler, Yuko G., and Kenj Hakuta. 2006. “Bilingualism and Second Language Acquisition”. In Tej K. Bhatia, and William C. Ritchie, The Handbook of Bilingualism, 114-136. Oxford: Blackwell Publishing. Cabredo-Hofherr, Patricia. 2013. Varieties of partial pro-drop. Handout SPCL Meeting. Lisboa. Cardinaletti, Anna, and Michal Starke. 1994. “The typology of structural deficiency: on the three grammatical classes”. Unpublished manuscript, University of Venice and University of Geneva. Carpenter, Karen, and Hubert Devonish. 2010. “Swimming Against the Tide: Jamaican Creole in Education”. In Bettina Migge, Isabelle Léglise, and Angela Bartens, Creoles in Education: an appraisal of current programs and projects, 167-182. Philadelphia/ Amsterdam: John Benjamins Publishing Company. —. 1981. Lectures on Government and Binding. Dordrecht: Foris. —.1986. Knowledge of language: Its Nature, Origin, and Use. New York: Praeger. Costa, João, and Maria Lobo. 2007. “Omissão de clíticos na aquisição do português europeu: dados da compreensão.” In XXIII Encontro Nacional da Associação Portuguesa de Linguística. Textos selecionados. Lisboa: APL. Costa, João, and Fernanda Pratas. 2012. “Embedded null subjects in Capeverdean”. In Journal of Linguistics 49: 33-53. Duarte, Dulce A. 1998. Bilinguismo ou Diglossia?. Mindelo: Spleen Edições. Duarte, Joana Duarte. 2010. Bilingual language proficiency. A comparative study. New York/Münster: Waxmann. Ferguson, Charles A. 1959. “Diglossia”. Word 15:325-340. Fiéis, Alexandra, and Fernanda Pratas. 2006. “Construções de duplo objecto não derivam de ditransitivas com preposição: evidência do Caboverdiano”. In Textos Seleccionados do XXI Encontro Nacional da APL, 387- 396. Lisboa: APL. Fishman, Joshua. 1967. “Bilingualism with and without diglossia; diglossia with and without bilingualism.” Journal of Social Issues 23(2): 29-38.

Ana Josefa Cardoso

103

Frascarelli, Mara. 2007. “Subjects, topics and the interpretation of referential pro: An interface approach to the linking of (null) pronouns”. Natural Language and Linguistic Theory 25:691-734. Garcia, Ofelia. 2010. Bilingual Education in the 21st Century. Malden/Oxford: Wiley-Blackwell. Garcia, Ofelia, Lesley Bartlett, and Jo Anne Kleifgen. 2007. “From biliteracy to pluriliteracy”. In Peter Auer, P. and Li Wei, Handbook of Multilingualism and Multilingual Communication, 207-221, Berlin: Walter de Gruyter GmbH and Co. KG. Garmandi, Julieti. 1983. Socio-Lingüística. Lisboa: Publicações Dom Quixote. Gass, Susan M. and Larry Selinker. 2001. Second Language Acquisition – an introductory course. New Jersey/London: Lawrence Erlbaum Associates Publishers. Gausti, Maria Teresa. 2002. Language Acquisition: The Growth of Grammar. Cambridge, Massachusetts: MIT Press. Gomes, Maria A. A. 1998. “Alfabetizar em Cabo Verde – Um Processo Isomórfico Sociocultural”. Master diss., Universidade Aberta. —. 2008. “Papia, Lé y Skrebe na Skóla Kauberdianu – A emergência de práticas identitárias”. PhD diss., Universidade Aberta. Hagège, Claude. 1997. A Criança de Duas Línguas. Lisboa: Instituto Piaget. Hawkins, Roger, and Hajine Hattori. 2006. “Interpretation of English multiple wh-questions by Japanese speakers: a missing uninterpretable feature account”. Second Language Research 22(3): 269–301. Hawkins, Roger. and Cecilia Y. Chan. 1997. “The partial availability of Universal Grammar in second language acquisition: the ‘failed functional features hypothesis’”. Second Language Research 13(3): 187–226. Hawkins, Roger. 2001. Second Language Syntax – A Generative Introduction. Oxford/ Massachusetts: Blackwell Publishers. Kabuto, Bobbie. 2011. Becoming Biliterate ņ Identity, ideology, and learning to read and write in two languages. New York: Routledge. Kenner, Charmian. 2011. Becoming Biliterate: Young Children Learning Different Writing Systems. Trentham Books Ltd. Lindholm-Leary, Katherine J. 2001. Dual Language Education. Clevedon: Multilingual Matters. Lopes, Amália M. 2011. “As Línguas de Cabo Verde – Uma radiografia linguística”. PhD diss., Universidade de Lisboa. Marques, Maria E. R. 2003. Português, Língua Segunda. Lisboa: Universidade Aberta.

104

The Acquisition of the Pronominal System in Portuguese L2

Martinet, André. 1991. Elementos de Linguística General. 3ª Edição. Madrid: Editorial Gredos. Mateus, Maria H. et al. 2003. Gramática da Língua Portuguesa. 6ª edição. Lisboa: Editorial Caminho. Mitchell, Rosamund & Florence Myles. 2004. Second Language Learning Theories (2nd edition). London: Hodder Arnold. [First edition, 1998]. Monteiro, Adelaide. 2009. “O Contexto sociolinguístico e situacional da aprendizagem de língua portuguesa em Cabo Verde na 1.ª fase do Ensino Básico”. Textos do Seminário Metodologias e Materiais para o ensino do Português como Língua Não Materna, 99-102. Lisboa: ILTEC. Pires, D.O.G. 2010. “A situação Linguística em Cabo Verde”. In IILP/AULP- Instituto Internacional da Língua Portuguesa/ Associação das Universidades de Língua Portuguesa. Interpenetração da Língua e Culturas de/em Língua Portuguesa na CPLP, 139-152. Lisboa: CPLP. Pratas, Fernanda. 2004. O Sistema Pronominal do Caboverdiano. Lisboa: Edições Colibri. [MA diss., Universidade Nova de Lisboa, 2002]. —. 2007. “Tense Features and Argument Structure in Capeverdean Predicates”. PhD diss., Universidade Nova de Lisboa. Pratas, Fernanda, and Andrés P. Salanova. 2005. “The allomorphy of Capeverdean object pronominals: a case for late insertion”. Paper presented at the 15th Colloqui de Gramàtica Generativa, Universitat de Barcelona, April 2005. Raposo, Eduardo. 1992. Teoria da gramática, a faculdade da linguagem. Lisboa: Caminho. Schwarz, Bonnie D, and Rex Sprouse. 1996. “L2 cognitive states and the full transfer/full access model”. Second Language Research 12: 40-72. —. 1994. “Word order and nominative case in nonnative language acquisition: a longitudinal study of (L1 Turkish) German interlanguage”. In Language acquisition studies in generative grammar: Papers in honor of Kennet Wexler from the GLOW 1991 Workshops, edited by T. Hoekstra, and B. Schwartz, 317-78. Amsterdam: John Benjamins. Selinker, Larry. 1972. “Interlanguage”. International Review of Applied Linguistics, X(3):209-231. Reprinted in 1974. Error Analysis: Perspectives on Second Language Learning, edited by J. Richards 3154. London: Longman. Silva, Carolina. 2008. “Assimetrias na Aquisição de Clíticos Diferenciados em Português Europeu”. MA diss., Universidade Nova de Lisboa.

Ana Josefa Cardoso

105

Sim-Sim, Inês. 1998. Desenvolvimento da Linguagem. Lisboa: Universidade Aberta. Torres-Guzmán, Maria, Jorge Abbate, Maria E. Brisk, and Liliana Minaya-Rowe. 2002. “Dual language programs: Key features and results”. Directions in Language Education, Nacional Clearing House for Bilingual Education 14:1-16. Tsimpli, Ianthi-Maria, and Maria Dimitrakopoulou. 2007. “The interpretability hypothesis: evidence from wh-interrogatives in second language acquisition”, Second Language Research 23(2): 215-242. Tsimpli, Ianthi-Maria, and Anna Roussou. 1991. “Parameter-resetting in L2?” Working papers in linguistics, 3: 150-169. Veiga, Manuel. 1995. O Crioulo de Cabo Verde, Introdução à Gramática. Praia: Instituto Caboverdiano do Livro. Veiga, Manuel. 2000. Le Créole du Cap-Vert – étude grammaticale descriptive et contrastive. Paris-Praia. Kartala-Instituto de Promoção Cultural. Veiga, Manuel. 2004. A Construção do Bilinguismo. Praia: Instituto da Biblioteca Nacional e do Livro. Wei, Li, and Melissa G. Moyer. 2010. The Blackwell Research Methods in Bilingualism and Multilingualism. Malden/Oxford/Victoria: Blackwell Publishing. White, Lydia. 2003. “On the nature of interlanguage representation: Universal Grammar in the Second Language”. In The Handbook of Second Language Acquisition, edited by C. J. Doughty, and M. H. Long, 19-42. Oxford: Blackwell Publishing. Xavier, Maria Francisca and Maria Helena Mira Mateus. 1990. Dicionário de Termos Linguísticos, vol. 1.. Lisboa: Edições Cosmos.

THE ACQUISITION OF THE MORPHOLOGICAL AND SEMANTIC PROPERTIES OF THE PORTUGUESE PRETÉRITO PERFEITO AND PRETÉRITO IMPERFEITO BY NATIVE SPEAKERS OF CAPEVERDEAN CREOLE ANA SOFIA FONSECA FACULDADE DE CIÊNCIAS SOCIAIS E HUMANAS DA UNIVERSIDADE NOVA DE LISBOA

Abstract This study aims to investigate the acquisition of the morphophonological forms and semantic values of the Portuguese Pretérito Perfeito (PP) and Pretérito Imperfeito (PI), by native speakers of Capeverdean Creole. It is assumed that aspect is a functional category and that these forms and values are acquirable in the second language (L2), thus implying full access to the Universal Grammar (UG) after first language (L1) acquisition. An instrument has been devised in order to assess whether the acquisition of the aforementioned aspectual distinction is influenced by the learners’ L1 or, rather, constrained by a universal phenomenon, according to the Primacy of Aspect Hypothesis (Andersen and Shirai, 1996). The results indicate that learners can associate the known semantic values with the new morphophonological forms and that they rely on the lexical aspect for the selection of the suitable morphology, especially at advanced levels. Nevertheless, we have not been able to prove or categorically refute the influence of L1 in this process. Keywords: Aspect, Functional Categories, Universal Grammar, Primacy of Aspect Hypothesis, Second Language.

Ana Sofia Fonseca

107

1. Introduction The acquisition of the morphophonological forms of functional categories in second language (L2), their mapping onto suitable semantic values, the role of Universal Grammar (UG) and of learners’ first language (L1) in this process are core issues for research in Second Language Acquisition. By investigating the acquisition of the morphological and semantic properties of the Portuguese Pretérito Perfeito (PP) and Pretérito Imperfeito (PI) (an aspectual distinction), by speakers of Capeverdean Creole, the intention of this study is to make a further contribution to this discussion and address two general questions: a) Does the acquisition of PP and PI morphology necessarily imply that the speakers know the semantic values associated with them? b) Is this acquisition influenced by the learners’ L1, or is it constrained by universal factors? In order to accomplish the purpose of this research, firstly, some information on the terminology and theoretical accounts adopted will be presented. This involves defining the concept of aspect; stating the paper’s assumptions on its acquisition in L2, within a UG approach; and describing a theoretical hypothesis that accounts for the stages of the acquisition of aspect in L2 (Primacy of Aspect Hypothesis). The Portuguese and the Cape Verdean aspectual systems will also briefly be compared and contrasted. Finally, the study itself will be discussed, presenting the hypotheses formulated, describing its methodology and discussing its results.

2. Terminology and Background 2.1. Grammatical and Lexical Aspect The term aspect refers to the internal structure of a situation within a particular tense (Campos 2002); in other words, it provides information on how the internal structure of a given situation is perceived: “Aspects are different ways of viewing the internal temporal constituency of a situation” (Comrie 1976, 3). There are two types of aspectual marking – viewpoint or grammatical aspect and situation aspect: “viewpoint aspect is signalled by a grammatical morpheme, while situation aspect is expressed by a constellation of lexical morphemes” (Smith 1991, 8). Therefore, the

108

Acquisition of the Morphological and Semantic Properties of Portuguese

aspectual meaning of a sentence results from the interaction between these two elements (Smith 1991, 4). Viewpoint or grammatical aspect is signalled by morphemes, such as perfective and imperfective tense morphemes, and “gives the receiver a full or partial view of a situation” (Smith 1991, 5). If a situation is viewed from the outside as a single whole, with initial and final endpoints, it has a perfective value. If, on contrary, it is viewed from within, without any specification of its beginning or ending, it has an imperfective value. To sum up: the perfective looks at the situation from outside, without necessarily distinguishing any of the internal structure of the situation, whereas the imperfective looks at the situation from inside, and as such is crucially concerned with the internal structure of the situation, since it can both look backwards towards the start of the situation, and look forwards to the end of the situation, and indeed is equally appropriate if the situation is one that lasts through all time, without any beginning and without any end. (Comrie 1976, 4)

As for situation aspect, it is determined, first of all, by the lexical aspect of the verb. This results from the combination of three binary semantic features: [±dynamic], [± telic] and [± durative]. Taking into account these features, Vendler (1967) classified verbs as achievements – realise, find, win the race… –, accomplishments – paint a picture, write a letter, make a table –, activities – run, walk, swim… – and states – like, want, have. Each class constitutes, therefore, a set of properties that contrast with one another: “The situation type names – State, Activity, Accomplishment, (…), Achievement – function as shorthand for the cluster of properties that distinguish the situation types” (Smith 1991, 28). The following table summarises what has just been described: Verb types States Activities Accomplishments Achievements

Dynamism + +

Telicity +

Duration + + +

+

+

-

Table 1: Aspectual Verb Classes

Ana Sofia Fonseca

109

It is important not to ignore the fact that the situation aspect of a sentence is not exclusively determined by the verb, but also by its arguments and adverbials, that is, the “constellation of lexical morphemes” referred to by Smith (1991, 5).

2.2. The Acquisition of Aspectual Distinctions within UG In this study, it is assumed that grammatical aspect is a UG-constrained phenomenon encoded in a functional category (Asp). According to Chomsky (1995), functional categories are responsible for the instantiation of inflectional morphology and encode the grammatical properties associated with a particular morpheme. They, therefore, constitute an interface of form and meaning and are the locus of parametric variation between languages. Whether these categories and their features are acquirable after L1 acquisition and to what extent L1 influences this process are central issues in Second Language Acquisition studies. Considering that “functional categories, features and feature values absent from the L1 grammar are instantiated in the interlanguage representation” (White 2003, 127), thus assuming full access to UG after L1 acquisition, we sought to investigate the role of the L1 in this process, by exploring two hypotheses. The first one, Full Transfer/Full Access, developed by Schwartz and Sprouse (1994, 1996), postulates that L1 grammar is the initial state of the learners’ interlanguage grammar and that L2 parameters will be initially set according to L1 specifications: “According to the FT/FA model, the entirety of the L1 grammar (excluding the phonetic matrices of lexical/morphological items) is the L2 initial state (hence the term ெFull Transferெ)” (Schwartz and Sprouse 1996, 41). However, at the same time, all UG parameters are available to the learners and, as they accommodate L2 input, their values will be reset. The second hypothesis, Full Access Without Transfer, by Epstein et al. (1996) and Flynn (1996) (cit. White 2003, 128), claims that the L1 is not implicated in the representation of interlanguage grammar. During L2 acquisition, parameter options are immediately set according to L2 values, as a result of the interaction between UG and L2 input.

2.3. The Primacy of Aspect Hypothesis As was previously mentioned, the aim is to investigate whether the learners’ L1 has any influence on the acquisition of the aspectual distinction we are focusing on, or whether this process is constrained by a

110

Acquisition of the Morphological and Semantic Properties of Portuguese

universal phenomenon as predicted by the Primacy of Aspect Hypothesis (Andersen and Shirai, 1996). This approach posits that the learners’ selection of verbal morphology will depend on the semantic properties of verbs (Andersen and Shirai 1994). Andersen and Shirai (1996) and (Shirai 2002) relate this phenomenon to the existence of prototypical categories: this approach assumes an internal structure within a category, with some members of the category being more basic, or more prototypical than others. The idea was originally proposed regarding natural categories as “bird”, for which prototypical members are “robin”, “sparrow”, etc., and peripheral members include “penguin”. (Shirai 2002, 457)

In other words, this approach postulates that, once L2 learners become unable to assimilate and use all the semantic values of a particular morpheme, they will cling to its most prototypical values: “For example, the prototypical past describes a situation which is [+ punctual], [+ telic] and [+ result], i.e. it tends to describe an instantaneous event that results in some observable result” (Shirai 2002, 457). This prototypical value of the past will, then, lead learners to associate it first with verbs evincing the same semantic properties. In effect, according to Shirai (1991), Andersen and Shirai (1994) and Shirai and Andersen (1995) (cit. Shirai 2002, 457): “the association observed between inherent aspect and verb morphology in L1 and L2 acquisition can be attributed to the acquisitional sequence from the prototype to peripheral members of the linguistic categories”. This acquisitional sequence is explained as follows: Prototypes of particular linguistic forms have very strong connection with the forms (in this case past and imperfective markers) and can easily receive enough activation and be produced. The past tense form, for example, has a very strong connection with achievement verbs, completion, punctuality, and so forth. If these features are involved, the past tense form easily gets enough activation and is more frequently produced in learners’ speech. Non-prototypical members do not have as strong connections, and therefore are less likely to be produced and/or tend to be acquired later than prototypical members. (Shirai 2002, 473)

This gradual acquisition of verbal morphology according to verb type occurs in eight stages (Andersen 1991): (i) L2 learners do not mark past tense or aspect grammatically; (ii) Perfect marking is associated with verbs expressing punctuality, i.e., achievements;

Ana Sofia Fonseca

111

(iii) Imperfect forms are associated with prototypical stative verbs; (iv) Perfect markers are used with accomplishment verbs and imperfect markers are used with activity verbs (all verbs are marked in the past tense). From this stage on, the acquisitional sequence of aspectual morphology evolves to non-prototypical uses of the forms: (v) Accomplishments can now be associated either to imperfect or perfect markers; (vi) Activities can be marked either with imperfect or perfect forms; (vii) Achievements can be associated with perfect or imperfect morphology; (viii) Stative verbs can be associated with perfect forms. Having described this aspectual acquisition model, we will focus on the differences between the two languages under study: Portuguese and Capeverdean Creole.

2.4. Aspectual Marking in Portuguese and Capeverdean Creole In order to clearly understand the hypotheses that have been formulated, it is important to analyse the differences between Portuguese and Capeverdean Creole, as far as their aspectual system is concerned. First of all, Portuguese has a very rich morphology and the value of its forms does not depend on the inherent lexical semantics of verbs. On the contrary, in Capeverdean Creole the contrast between perfective and imperfective is mainly conveyed by unbound morphemes and the meaning of its forms is strongly determined by the lexical aspect of verbs. In Portuguese, the contrast between perfective and imperfective is clearly conveyed by tense-mode-aspect morphemes. Although there are other verb tenses that express these values, we will only focus on the ones under study: the Pretérito Perfeito (PP) and the Pretérito Imperfeito (PI). It is important to understand that, despite the fact that they both encode a temporal meaning of past, their aspectual value is not the same. Accordingly, perfective meaning is encoded in PP markers: (1) A Maria escreveu a carta. (Oliveira 2003, 156) The Maria write PP -3SG the letter “Mary wrote the letter”.

112

Acquisition of the Morphological and Semantic Properties of Portuguese

As for the imperfective value, it can be encoded in PI markers: (2) Dantes, eu sabia andar de bicicleta. Before I know- PI-1SG ride of bicycle. “In the past, I used to know how to ride a bicycle.” As previously mentioned, in Capeverdean Creole the value of the morphemes is determined by the inherent semantics of the verb and the contrast between stative vs. non-stative verbs is crucial. Furthermore, temporal reference is strongly dependent on aspect (Pratas 2007). Some examples will now be provided in order to better understand this system. Stative verbs will only encode the perfective value in the presence of a Ø morpheme combined with other adverbials or objects. (3) Onti N sabe ma Djon gosta di Maria. (idem) Yesterday I know Ø that John like of Mary “Yesterday I came to know that John likes Mary.” However, in the absence of adverbials or objects, stative verbs combined with a Ø morpheme will express present tense and, consequently, denote imperfectivity. (4) Pursor sabe mas di ki alunus. (Pratas 2007, 99) Teacher knowØ more of that students. “The teacher knows more than the students”. As for non-stative verbs, the presence of a Ø morpheme will convey the perfective meaning: (5) E odja (idem: 67) He see Ø “He saw.” Imperfective meaning will be conveyed by the bound morpheme -ba, when associated with stative verbs: (6) E gostaba. (idem, 67) He likeTMA “He liked.”

Ana Sofia Fonseca

113

It is very important to stress that the absence of pre-verbal morphemes with non-stative verbs ending in –ba will convey a perfective meaning (anteriority in relation to the past): (7) E odjaba. (idem, 67) He seeTMA “He had seen.” Non-stative verbs will require other markers to express imperfective values. For instance, the combination ta v + -ba conveys frequency: (8) E ta odjaba.(idem, 67) He TMA seeTMA ‘He used to see’ The combination sata v+ -ba is the equivalent of a progressive past: (9) N sata skrebeba. I TMA writeTMA “I was writing.” Finally, the combination ta staba + ta + v+ -ba refers to a repeated or frequent action in the past: (10) E ta staba tudora ta odjaba. He TMA TMA sempre TMA seeTMA “He was always seeing.” We are aware that this plain description does not convey the complexity that characterises the expression of temporal reference in Capeverdean Creole. It is extremely important to bear in mind that this is a compositional category (Pratas 2007): Tense and Aspect are, in Capeverdean, a complex matter of cooperation between a relatively small set of morphemes, auxiliary forms, and a huge amount of possible combinations with verbal semantic properties, adverbial expressions, temporal clauses and discourse information. In other words, this construction of temporal reference is typically compositional. (Pratas 2007, 61)

The aim of this section was to provide a brief description of how grammatical aspect is marked in Portuguese and Capeverdean Creole, in

114

Acquisition of the Morphological and Semantic Properties of Portuguese

order to facilitate the understanding of the hypotheses that have been formulated. These will be presented in the following section.

3. The study 3.1. The Hypotheses Comparing both languages leads to the formulation of some hypotheses concerning the difficulties learners may evince when dealing with the aspectual distinction we are focusing on. The first hypothesis is the following: Learners will exhibit more difficulties in the acquisition of the PP when associated with stative verbs than when associated with nonstative verbs. These difficulties may be due to: a) the existence of prototypical categories, as posited by the Primacy of Aspect Hypothesis; b) L1 influence: in Capeverdean Creole stative verbs require the presence of adverbials or objects in order to convey perfective meaning (which does not happen in Portuguese). Our second hypothesis postulates that: Learners will exhibit more difficulties in the acquisition of the PI with non-stative verbs than with stative verbs. Once again, these difficulties may be accounted for by: a) the Primacy of Aspect Hypothesis; b) L1 influence: in Capeverdean Creole, the pre-verbal morpheme ta with non-stative verbs implies perfective value (which does not happen in Portuguese). Furthermore, in this language there are several morphemes which convey the different imperfective values, unlike Portuguese. Hopefully, it will be possible to understand whether the learners’ behaviour results from L1 influence or from the existence of prototypical categories, regardless of L1 properties, by analysing their performance as far as activity verbs are concerned. If learners do not evince significant

Ana Sofia Fonseca

115

difficulties in the interpretation of perfective markers when associated with activity verbs, these predictions may be explained by L1 transfer. Learners may be envisaging activity verbs as non-stative verbs and thus the interpretation of this verb type when associated with imperfective markers may be more problematic. If, on contrary, learners are able to interpret imperfective morphophonological forms associated with activity verbs, their performance may be accounted for by the Primacy of Aspect Hypothesis. Learners may be envisaging activity verbs as stative verbs and therefore they may evince more difficulties in the interpretation of this verb type when associated with perfective forms.

3.2. Methodology 3.2.1. Participants Data was collected from twenty two native speakers of Capeverdean Creole, aged between 12 and 16, who are learning Portuguese as a second language in Portuguese State Schools. They were divided into two groups – advanced and beginners – according to the proficiency tests applied in the schools they attend. The beginners group was composed of eleven learners, whose average age was 13.7 years of age, and their exposure time to Portuguese varied between 0 and 4 years (average exposure time of 2.2 years). The advanced group was composed of eleven learners, whose average age was 14.2 with most of them having been born in Portugal. Some of them had early contact with the Portuguese language and can determine different linguistic behaviour. Nevertheless, both the fact that Capeverdean Creole was the language they spoke at home since childhood and their performance in the proficiency tests (featuring errors resulting from L1 interference) led us to include them in the advanced group. A control group, composed of 11 native speakers of Portuguese, aged between 12 and 16, was also tested. 3.2.2. Methodology and Procedures In order to test the acquisition of the morphological and semantic properties of the PP and the PI, a production test was devised. Learners were asked to fill in the gaps in a text with an inflected verb form (the verb was provided in the infinitive). We selected 24 verbs that should be conjugated in the PP, equally distributed by the four aspectual verb types:six achievements:

116

Acquisition of the Morphological and Semantic Properties of Portuguese

prometer (to promise); recusar (to refuse); proibir (to forbid); levantar (to lift); sorrir (to smile); atirar (to throw); six accomplishments: “espalhar-se… por toda a cidade” (to spread all over the town); “levar o barco até à parede do palácio” (to take the boat to the wall of the palace); “puxar… até à varanda) (to pull …up to the balcony); “descer … até ao barco” (to go down to the boat); “correr até ao cais” (to run towards the pier); “contar” (to tell); six activities: passar (to pass past); navegar (to navigate); afastarse (to move away); viajar (to travel); correr (to run); seguir viagem (travel on); six states: ficar (to stay) morar (to live); querer (to want); admirar (to admire); ficar (to become); gostar (to like). In each aspectual class, three verbs were accompanied by adverbials/adverbial phrases requiring the use of perfective markers: na noite em que (in the night that…), aos 18 anos (when she was 18 years old), nessa noite (in that same night), assim que (as soon as), no dia seguinte (the following day), quando (when]), imediatamente (immediately). We also selected 24 verbs that should be conjugated in the PI, equally distributed by the four aspectual verb types: six achievements: sacudir os cabelos (to shake her hair); embarcar (to embark); suspirar (to sigh); abrir (to open); levantar-se (to get up); espreitar (to peer); six accomplishments: cantarolar a sua canção favorita (to croon his favourite song); relatar a sua viagem (to describe his journey); fazer uma trança (to make a braid); cantar a sua canção favorita (to sing her favourite song); seguir Guidobaldo até ao palácio (to follow Guidobaldo to the palace); recitar um poema (to recite a poem); six activities: sonhar (to dream); vigiar (to keep tabs on); escrever (to write); pentear (to comb); dormir (to sleep); navegar (to navigate); six states: ansiar (to yearn); sentir-se (to feel); temer (to dread); ter medo (to be afraid); recear (to fear); desejar (to crave).

Ana Sofia Fonseca

117

Here is a brief excerpt from the devised test: Na noite em que o cavaleiro da Dinamarca e o comerciante de Veneza se conheceram, ____________ (ficar) a conversar na varanda. Já há algumas horas que o cavaleiro da Dinamarca ____________ (relatar) a sua viagem, quando, de repente, reparou num palácio ao longe e perguntou de quem era. O comerciante respondeu: “Este palácio pertence a Orso e, há alguns anos, passou-se aqui uma história muito interessante.” E continuou: “Vanina, sobrinha de Orso, perdeu os pais ainda bebé. Por isso, durante muitos anos, ____________ (morar) no seu palácio. Um dia, Orso ____________ (prometer) a jovem em casamento a Arrigo, um velho, seu parente. Há anos que Orso ____________ (sonhar) repetidamente com esse acontecimento. Mas aos 18 anos, Vanina ____________ (recusar) casar com Arrigo (6). Então, Orso ____________ (proibir) a jovem de sair do palácio sozinha.

While analysing the data, all the verb forms unequivocally evincing the morphological tense marks required by the context were considered correct answers, regardless of spelling mistakes, grammatical concordance errors or pronoun misplacements.

3.3. Results 3.3.1. Inter-group comparison: global performance with the PP and the PI First of all, the performance of both groups (advanced and beginners) was compared with regard to PP and PI. There are statistically significant differences between the beginners’ group performance and the advanced group with the PP forms (p = 0.002 < 0.05). The number of correct answers within the first group is lower than in the latter. At the same time, there are no statistically significant differences between the control group performance and the advanced group (p = 0.619 > 0.05). With the PI markers, the results are once again very similar. In effect, there are statistically significant differences (p = 0.001 < 0.05) between the beginners’ group performance and the advanced group (the latter being more accurate in the use of the PI markers). When compared with the

118

Acquisition of the Morphological and Semantic Properties of Portuguese

control group, the advanced group performance is not significantly different (p = 0.054 > 0.05). 3.3.2. Within-group comparison: PP vs. PI Within group comparison revealed that, in both samples (advanced and beginners), the average number of correct answers with the PP (beginners: 14. 82 / advanced: 19. 91) is higher than the average number of correct answers with the PI (beginners: 12. 36 / advanced: 18.45). However, that difference is not statistically significant (p 0.161 and 0.239 > 0.05). The control group exhibited more success in the use of the PI (average 22.18) than in the use of the PP (average 21.18), but once again that difference is not statistically significant (p = 0.184 > 0.05). 3.3.3. Within-group comparison according to the stative/non-stative contrast Given that, according to the formulated hypotheses, it was important to relate the use of the aspectual forms to the lexical aspect of verbs, the learners’ performance with the PP and the PI was compared, within groups, taking into account the stative verbs/non-stative verbs contrast. Within the beginners’ group the results were the following: a) PP: the average number of correct answers with the PP morphology associated with non-stative verbs (ȝ=4) is higher than the average number of correct answers with the same morphology associated with stative verbs (ȝ=2.81). According to the t test for paired samples, this difference is statistically significant (p = 0.004< 0.05). b) PI: the average number of correct answers with the PI morphology associated with non-stative verbs (ȝ=3.15) is higher than the number of correct answers with the same forms associated with stative verbs (ȝ=2.90). However, according to the t test for paired samples, this difference is not statistically significant (p = 0.301 > 0.05). Within the advanced group the results were the following: a) PP: the average number of correct answers with the PP forms associated with non-stative verbs (ȝ=5.30) is higher than the average number of correct answers with the same morphology associated with stative verbs (ȝ=4). According to the t test for

Ana Sofia Fonseca

119

paired samples, this difference is statistically significant (p = 0.0015< 0.05). b) PI: the average number of correct answers with the PI forms associated with non-stative verbs (ȝ=4.54) is lower than the average number of correct answers with the same marker associated with stative verbs (ȝ=4.81). According to the t test for paired samples, this difference is statistically not significant (p = 0.079 > 0.05). Under the same condition, the control group evinced the same behaviour as the advanced group. 3.3.4. Testing the Primacy of Aspect Hypothesis In order to test the Primacy of Aspect Hypothesis, a within group comparison between the learners’ performance with the PP and the PI when associated with each aspectual type (achievements, accomplishments, activities and states) was made. With the beginners’ groups the results were the following: a) achievements: the average number of correct answers with the PP forms (ȝ=4.9091) is higher than the average number of correct answers with the PI (ȝ=2.8182). According to the t test for paired samples, this difference is statistically significant (p = 0.0025 < 0.05). b) accomplishments: the average number of correct answers with the PP forms (ȝ=3.36) is higher than the average number of correct answers with the PI (ȝ=3.09). However, according to the t test for paired samples, this difference is not statistically significant (p = 0.3175 > 0.05). c) activities: the average number of correct answers with the PP forms (ȝ=3.72) is higher than the average number of correct answers with the PI (ȝ=3.54). However, according to the t test for paired samples, this difference is not statistically significant (p = 0.4065 > 0.05). d) states: the average number of correct answers with the PP forms (ȝ= ȝ=2.81) is lower than the average number of correct answers with the PI (ȝ=2.90). However, according to the t test for paired samples, this difference is not statistically significant (p = 0.4195 > 0.05).

120

Acquisition of the Morphological and Semantic Properties of Portuguese

Within the advanced group the results were the following: a) achievements: the average number of correct answers with the PP forms (ȝ=5.81) is higher than the average number of correct answers with the PI (ȝ=4.27). According to the t test for paired samples, this difference is statistically significant (p = 0.0005 < 0.05). b) accomplishments: the average number of correct answers with the PP forms (ȝ=5.09) is higher than the average number of correct answers with the PI (ȝ= 4.18). However, according to the t test for paired samples this difference is not statistically significant (p = 0.0835 > 0.05). c) activities: the average number of correct answers with the PP forms (ȝ=5) is lower than the average number of correct answers with the PI (ȝ=5.18). However, according to the t test for paired samples, this difference is not statistically significant (p= 0.345 > 0.05). d) states: the average number of correct answers with the PP forms (ȝ=4) is lower than the average number of correct answers with the PI ( ȝ=4.8). According to the t test for paired samples, this difference is statistically significant (p= 0.0475 < 0.05). Within the control group, the results were the following: a) achievements: the average number of correct answers with the PP morphology (ȝ=5. 27) is lower than the average number of correct answers with the PI. However, according to the Wilcoxon test for paired samples, this difference is not statistically significant (p= 0.24 > 0.05). b) accomplishments: the average number of correct answers with the PP forms (ȝ=5.81) is higher than the average number of correct answers with the PI (ȝ=5.36). However, according to the Wilcoxon test for paired samples, this difference is not statistically significant (p = 0.129 > 0.05). c) activities: the average number of correct answers with the PP forms (ȝ=5.27) is lower than the average number of correct answers with the PI (ȝ=5.36). However, according to the Wilcoxon test for paired samples, this difference is not statistically significant (p = 0.1335 > 0.05). d) states: the average number of correct answers with the PP morphology (ȝ=4.81) is lower than the average number of

Ana Sofia Fonseca

121

correct answers with the PI forms (ȝ=5.72). According to the Wilcoxon test for paired samples, this difference is statistically significant (p= 0.002 < 0.05).

3.3. Testing the L1 influence In order to assess the L1 influence on the acquisition of the PP when associated with stative verbs, the learners’ performance, in the presence or absence of adverbials was compared, within groups. Within the beginners’ group, the average number of correct answers with stative verbs in the PP with or without adverbials is very similar: ȝ=1.35 and ȝ=1.45, respectively. Consequently, according to the t test for paired samples, this difference is not statistically significant (p = 0.398 > 0.05) . Within the advanced group, the average number of correct answers with stative verbs in the PP with or without adverbials differs significantly (p= 0.051= 0.05). What is more, the average number of correct answers in the absence of adverbials (ȝ=2.18 ) is higher than in the opposite situation: ȝ=1.81. Within the control group, the average number of correct answers with stative verbs in the PP with or without adverbials is again very similar (ȝ=2.45 and ȝ=2.36, respectively). Therefore, this difference is not statistically significant (p = 0.3695 > 0.05), according to the Wilcoxon test applied.

3.4. Discussion The results of this indicate that, first of all, beginners’ group performance differs significantly from the other groups’ performance (advanced and control). It is possible to state that the former exhibits some difficulties in the use of the PP and PI forms. On the other hand, the advanced group performance does not differ significantly from the control group, thus demonstrating that learners tend towards a gradual acquisition of these forms as their proficiency level increases (cf. hypothesis three). Nevertheless, within both groups under investigation, the fact that the minimal number of correct answers is wider for the PI than for the PP may indicate that the former’s acquisition is more problematic. Although there were no statistically significant differences between the learners’ performance with the PP and the PI within both groups (beginners and advanced), the average number of correct answers with the PP is higher than with the PI. This fact can point to a more solid

122

Acquisition of the Morphological and Semantic Properties of Portuguese

acquisition of the perfective form, as predicted by the Primacy of Aspect Hypothesis (Andersen and Shirai 1996). Furthermore, according to Slabakova and Montrul (2002), most studies on the acquisition of tense and aspect in Spanish L2 (Hasbún 1995; Lafford 1996; Liskin-Gasparro 1997; Ramsay 1990; Salaberry 1997, Salaberry 1999, cit. Slabakova and Montrul 2002) show that perfective morphology is acquired earlier than imperfective morphology. Salaberry (1999) accounts for this fact by claiming that the PP might be used as a default marker for the past tense. Within the beginners’ group this may well be a possible justification for the learners’ behaviour. Turning to the conclusions that can be drawn from the comparison between the learners’ performance, in each group, with the PP and the PI, as far as the stative verbs / non-stative verbs contrast is concerned, learners within both groups evinced more difficulties in the use of perfective morphology when associated with stative verbs, which is consistent with the first formulated hypothesis. As for the imperfective forms, although the average number of correct answers with stative verbs is higher than the average number of correct answers with non-stative verbs, this difference is not statistically significant. Consequently, the second hypothesis cannot be confirmed. When the hypotheses on learners’ performance were formulated, two factors were also presented that could account for each one of them: L1 influence and the existence of prototypical categories (as predicted by the Primacy of Aspect Hypothesis). As far as the first factor is concerned, the analysis of the participants’ behaviour with the PP associated with stative verbs leads to the conclusion that it cannot be related to the presence of adverbials (it should not be forgotten that in Capeverdean Creole stative verbs require the presence of adverbials in order to acquire perfective value, which does not happen in Portuguese). Within the beginners’ group there is no statistically significant evidence to support this view and, therefore, L1 influence is difficult to prove. On the other hand, within the advanced group, the learners’ performance with stative verbs in the PP in the absence of adverbials was even more successful than in the presence of adverbials. This may indicate that, for these learners, the presence of adverbials makes the use of perfective markers redundant and perhaps facultative, which may point to a possible influence of the L1 at this stage. As regards whether the learners’ behaviour was being constrained by the existence of prototypical categories (the second factor), the paper will now consider the results referring to their performance with the PP and the

Ana Sofia Fonseca

123

PI when associated with each verb type (achievements, accomplishments, activities and states). Within both groups (beginners and advanced), there is statistically significant evidence to claim that the learners’ behaviour is consistent with the Primacy of Aspect Hypothesis, as far as achievements are concerned: perfective markers will be associated, at first, with [-DUR] events. As for accomplishment verbs, although there was a higher success rate when this verb type was associated with perfective forms, as predicted by the Primacy of Aspect Hypothesis, the results are not statistically significant. With regard to activity verbs, the learners’ behaviour in each group differed. Within the beginners’ group there was a higher success rate when this verb type was associated with perfective forms, which contradicts the Primacy of Aspect Hypothesis predictions. On the other hand, the advanced group exhibited a better performance when this verb type was associated with imperfective morphology, in accordance with the Primacy of Aspect Hypothesis. Nevertheless, there is no statistically significant evidence to support any of these situations. Finally, the learners’ within both groups exhibited a better performance when states were associated with imperfective morphology, which is in line with the Primacy of Aspect Hypothesis. Nevertheless, in the beginners’ group, there is no statistically significant evidence to support this position, contrary to what happened within the advanced group. To sum up, the beginners’ group behaviour is in line with the Primacy of Aspect Hypothesis, except for activity verbs. Nevertheless, according to the statistical analysis carried out, a relationship can only be established between the tenets of this theoretical account and the learners’ performance with regard to achievement verbs. Within the advanced group, the learners’ behaviour was in accordance with the Primacy of Aspect Hypothesis with respect to all the verb types. However, there was no statistically significant evidence to support this position (this only happened with achievement and stative verbs). The fact that there were no statistically significant differences between the use of the perfective and imperfective forms with accomplishment and activity verbs may indicate that learners within this group may have acquired non-prototypical uses of those markers. That is to say, they may be at stage (vi) of the acquisition of aspect markers: activities can be associated either with perfective or imperfective morphology (Andersen 1991).

124

Acquisition of the Morphological and Semantic Properties of Portuguese

4. Conclusion The aim of this study was to investigate whether native speakers of Capeverdean Creole would be able to acquire the morphological and semantic properties of the Portuguese PP and PI simultaneously. In other words, it aimed at verifying whether they could associate new morphophonological forms with semantic values they already knew (perfective and imperfective). The results confirm hypothesis 1: learners will exhibit more difficulties in the acquisition of the PP with stative verbs than with non-stative verbs. As for hypothesis 2 – learners will exhibit more difficulties in the acquisition of the PI with non-stative verbs than with stative verbs –, although the results reflect this prediction, there is no statistically significant evidence to support it L1 influence in this process is hard to prove. On the one hand, results indicate that the absence of adverbials in sentences with stative verbs in the PP did not negatively constrain the learners’ behaviour. On the other hand, within both groups, the learners’ behaviour as far as activity verbs is concerned does not constitute evidence for or against L1 influence in this process. In order to draw unequivocal conclusions concerning this, the learners’ behaviour would have to be compared with other learners with different mother tongues. In trying to figure out whether the learners’ performance could be accounted for by the Primacy of Aspect Hypothesis, it was verified that the relation between lexical aspect and the selection of suitable morphology is stronger within the advanced group. Nevertheless, there is no statistically significant evidence to support this relation with all the verb types. This group’s behaviour (which differs significantly from the beginners’ group and is similar to the control group’s) indicates, furthermore, that learners can associate known semantic values with new morphological forms. Given that in the test applied the selection of the suitable morphology was determined by an unequivocal context, the success evinced by the advanced learners indicates that not only do they know the perfective and imperfective markers under study, but they also understand and distinguish their semantic meanings. This inseparability of form and meaning leads therefore to the conclusion that we an affirmative answer can be given to our first research question: Does the acquisition of the morphology of the PP and the PI necessarily imply that the speakers know the semantic values related to them and vice-versa? In fact, this success can constitute evidence for the acquisition of the relevant aspectual features of Portuguese L2. It should not be forgotten that

Ana Sofia Fonseca

125

Capeverdean Creole and Portuguese differ in the way these features are associated and in the way they interact with the semantic properties of the verb. The fact that learners can use the PP and the PI markers, regardless of the stative/non-stative contrast (fundamental in Capeverdean Creole), may indicate that they have acquired options made available by the UG that are not distributed in their L1 in the same way. Hence, these results constitute evidence for the theoretical account adopted here which posits that parameters can be set in response to L2 input, namely full access to UG. Although there is some evidence that the learners have acquired the relevant aspectual features of Portuguese L2, it was not possible to unequivocally assess the L1’s role in this process. Therefore, it was not possible to provide an irrefutable answer to the second research question: Is the L2 acquisition of aspectual distinctions influenced by the learners’ L1, or is it constrained by universal factors? In effect, it cannot clearly be stated that L1 grammar is the initial state of the learners’ interlanguage grammar (although the progression demonstrated could corroborate that idea) – full access/full transfer; nor can it be claimed that the learners’ L1 is not implicated in that representation – full access without transfer. It is considered that a larger sample could provide us with more reliable results. Hence it is highly recommended that this study is replicated with a larger sample composed of learners with different mother tongues.

References Andersen, Roger, and Yasuhiro Shirai. 1994. “Discourse motivations for some cognitive acquisition principles”. Studies in Second Language Acquisition, 16, (2): 133-156. —. 1996. “The primacy of aspect in first and second language acquisition: the pidgin-creole connection”. In William C. Ritchie, and Tej K. Bhatia, orgs., Handbook of Second Language Acquisition, 527-570. London: Academic Press. Andersen, Roger 1991.” Developmental Sequences: The emergence of aspect marking in second language acquisition”. In T. Huebner, and C. Ferguson, orgs., Crosscurrents in Second Language Acquisition and Linguistic Theories. 305-324. Amsterdam: John Benjamins. Campos, Maria Henriqueta. 2002. “Questões aspectuais: algumas especificidades do português”. In S. Grosse, and A. Schomberger, orgs., Ex oriente lux: Festschrift für Eberhard Gärtner zu seinem 60. Geburtstag, 73-88. Frankfurt am Main: Valentia.

126

Acquisition of the Morphological and Semantic Properties of Portuguese

Chomsky, Noam. 1995. The Minimalist Program. Cambridge: The MIT Press. Comrie, Bernard. 1976. Aspect. Cambridge: Cambridge University Press. Epstein, Samuel, Flynn, Susan, and Gita Martohardjono. 1996. “Second language acquisition: theoretical and experimental issues in contemporary research”. Behavioural and Brain Sciences, 19 (4): 677758. Giorgi, Alessandra, and Fabio Pianesi. 1997. Tense and aspect: From semantics to morphosyntax. Oxford: Oxford University Press. Montrul, Silvina, and Roumyana Slabakova. 2002. “The L2 acquisition of morphosyntactic and semantic properties of the aspectual tenses preterite and imperfect”. In A. T. Pérez-Leroux, and J. M. Liceras, orgs., The acquisition of Spanish morphosyntax,115-151 . Boston: Kluwer Academic Publishers. Oliveira, Fátima. 2003. “Tempo e aspecto.” In Maria H. M. Mateus et. al., orgs., Gramática da Língua Portuguesa, 127-178. Lisboa: Caminho. Pratas, Fernanda. 2007. “Tense features and argument structure in Capeverdean predicates”. PhD diss., Universidade Nova de Lisboa. Salaberry, Rafael. 1999. “The development of the past verbal tense morphology in classroom L2 Spanish. Applied Language, 20 (2):151178. Schwartz, Bonnie, and Rex Sprouse. 1994. “Word order and nominative case in non-native language acquisition: a longitudinal study of (L1 Turkish) German interlanguage”. In Teun Hoekstra and Bonnie D. Schwartz, orgs., Language acquisition studies in generative grammar, 317-368. Amsterdam: John Benjamins. —. 1996. “L2 cognitive states and the full transfer/full access model”. Second Language Research, 12 (1):40-72. Shirai, Yasuhiro. 2002. “The prototype hypothesis of tense-aspect acquisition in second language”. In Rex Salaberry, and Yasuhiro Shirai, orgs., The L2 acquisition of tense-aspect morphology, 456-478. Amsterdam: John Benjamins Publishing Company. Slabakova, Roumyana, and Silvina Montrul. 2002. “On viewpoint aspect interpretation and its L2 acquisition – A UG perspective”. In Rex Salaberry, and Yasuhiro Shirai, orgs., The L2 acquisition of tenseaspect morphology, 313-395. Amsterdam: John Benjamins Publishing Company. Smith, Carlota. S. 1991. The parameter of aspect. Dordrecht: Kluwer Academic Publishers. Vainikka, Anna, and Martha Young-Scholten. 1994. “Direct access to Xbar theory: evidence from Korean and Turkish adults learning

Ana Sofia Fonseca

127

German”. In Teun Hoekstra and Bonnie D. Schwartz, orgs., Language acquisition studies in generative grammar, 265-315. Amsterdam: John Benjamins. Veiga, Manuel. 1995. O crioulo de Cabo Verde: introdução à gramática. S. Vicente: Instituto Cabo-Verdiano do Livro e do Disco. Vendler, Zeno. 1967. “Verbs and Times”. Linguistics in Philosophy. Ithaca: Cornell University Press. White, Lydia. 2003. Second Language Acquisition and Universal Grammar. Cambridge: Cambridge University Press.

NULL OBJECT FEATURES IN BRAZILIAN PORTUGUESE MIRNA SODRÉ VALVERDE-HÜBNER UNIVERSIDADE DE BRASÍLIA

Abstract This article seeks to analyze the Brazilian Portuguese (BP) Null Object (NO) as regards the features that this construction presents, and discuss two issues repeatedly mentioned in the literature: i) the dominant role of animacy in the manifestation of the phenomenon and ii) the relation of BP´s NO with the VP ellipsis construction, given the challenges of defining the boundaries between the two phenomena based on the available literature. The article concludes that, in addition to animacy, NO presents other features in this language, organized in a hierarchy, including imperfectivity. Besides thus formulating a better characterization of this phenomenon, these typical NO elements will also serve to establish the distinction between NO and VP ellipsis, as suggested in Goldberg (2005). Keywords: Null object, VP ellipsis, Animacy.

1. Introduction Within generative theory, empty categories fill a major role in that, while not having a phonological manifestation, they do have a mental representation, contributing to the hypothesis that children do not learn language by imitating the hearing of sounds. In this context, there is the null object, a construction present in several world languages, such as Brazilian Portuguese, Hebrew, Greek, Swahili, Japanese, etc., in which the object position is phonetically null but it presents syntactic and semantic content. Syntactically, the NO has been classified as a null pronoun (see Farrell 1990 and Galves 1989), variable (see Raposo 1986), as a null clitic (see Kato 1993) and null epithet (see Huang 1991), among others. The semantic content can be retrieved by an

Mirna Sodré Valverde-Hübner

129

antecedent expressed in speech (see 1a), with this being the most restricted type of NO among languages, by the pragmatic context (deictic) as in (1b) and (1c) or also by a generic reading such as “people” (2): (1) a. – Comprou guaraná zeroi também? – Comprei __i, uai. ‘– Did you buy diet guarana too? – Yeah, I bought [-].’ (Valverde-Hübner 2012, 11) b. – Vou pegar __ pra você. ‘– I´ll get [-] for you’. (Cyrino 1997, 39) c. – Push __! (Cyrino 1997, 39) (2) a. La música relaja __. ‘Music relaxes [-].’ (Landa 1991, cit. Cyrino 1997, 45) b. Wild Guns est un jeu qui défoule __. ‘Wild Guns is a game that destresses [-].’ (Cummins and Roberge 2004, 3) c. Questo conduce __ alla seguente conclusione. ‘This leads [-] to the following conclusion.’ (Rizzi 1986, 501) d. This leads __ to the following conclusion. (Rizzi 1986, 501) e. Esse remédio deixa __ tonto. ‘This medicine makes [-] dizzy.’ (Maia, 1990, cit. Cyrino 1997, 45) This article focuses on the set of NOs with defined antecedent and specific interpretation (anaphoric NO), which are less recurrent among languages and quite productive in BP. Among the Romance languages, NO is a feature only of Portuguese and it has distinct characteristics in BP and in European Portuguese (EP), being far more productive in BP. A classic distinction pointed out in the literature is that while BP allows NO in syntactic islands, EP does not allow it, so that the sentences in (3) are grammatical in BP and ungrammatical in EP: (3) a. Eu informei à polícia da possibilidade de o Manuel ter guardado __ no cofre da sala de jantar.

130

Null Object features in Brazilian Portuguese

‘I informed the police of the possibility that Manuel could have stored [-] in the dining room’s safe.’ b. O rapaz que trouxe __ agora mesmo da pastelaria era o teu afilhado. ‘The boy who brought [-] right now from the pastry was your godson.’ c. Que a IBM venda __ a particulares supreende-me. ‘That IBM sells [-] to private customers surprises me.’ d. O pirata partiu para as Caraíbas depois de ter guardado __ cuidadosamente no cofre. ‘The pirate sailed to the Caribbean after having carefully stored [-] in a safe.’ e. Quando é que o Manuel vai oferecer __ ao Antônio? ‘When will Manuel offer [-] to Antonio?’ (Raposo 1986, cit. Cyrino 1997, 1) According to the literature, NO can only work in BP if its antecedent is inanimate. Where it is animate, the sentence will be ungrammatical, as shown in the examples in (4), where the exchange of an inanimate antecedent by an animate antecedent takes place. This change could only be grammatical if it is made by a lexical pronoun, as in (4c): (4) a. O Emílio perdeu a carteira e não consegue achar __ em lugar nenhum. ‘Emilio lost his wallet and can´t find [-] anywhere.’ b. *A Clara não quer que o filho veja TV, então ela sempre leva __ no parquinho. ‘Clara doesn´t want her son to watch TV, so she always takes [-] to the playground.’ c. A Clara não quer que o filho veja TV, então ela sempre leva ele no parquinho. ‘Clara doesn´t want her son to watch TV, so she always takes him to the playground.’ (Lopes 2006, 164)

Mirna Sodré Valverde-Hübner

131

However, I argue in Valverde-Hübner (2012) that inanimacy is not the only feature of BP’s NO, since (5), typically a case of NO1, is grammatical: (5) A Maria escova o cachorroi, antes de dormir e a Joana escova __i assim que acorda. ‘Mary brushes the dog before she goes to bed and Joana brushes [-] as soon as she awakes.’ (Valverde-Hübner 2012, 14) Thus, it seems that there are other elements involved in the occurrence of NO in addition to animacy, which will be presented later. Beyond the question of animacy, the literature on BP´s NO sometimes touches on the VP ellipsis construction, defined when all the internal elements of the VP are elided, not only the direct object: (6) A mãe do Marquinhos sempre leva elei no parquinhoj, mas seu pai não costuma levar__ij . ‘Marquinho´s mother always takes him to the playground, but his dad does not usually take [-].’(Valverde-Hübner 2012, 14) In (6), both the direct object ele and the adjunct no parquinho are elided in the second clause, featuring a VP ellipsis construction. Such a construction, as seen in this example, has no animacy restrictions (see also Cyrino 1997 and Goldberg 2005). The literature sometimes shows similar definitions for the two phenomena (see Cyrino 1997), sometimes different definitions (see Cyrino and Matos 2002), but the difficulty of identifying which one it is remains when only the direct object is elided in the deleted clause, so that the construction can be both NO and VP ellipsis, since there is only one direct object being deleted and, once it is deleted, as it is the only one internal to the VP, all the internal elements within the VP are deleted, as in (7): (7) João viu sua mãe e Pedro também viu __. ‘John saw his mother and so did [-] Peter.’ (Cyrino 1997, 61)

1

I use typically to qualify this statement because, as I will show later, there are ambiguous constructions between the null object and VP ellipsis, basically when there is only a direct object in the VP and it’s elided. In the example in question, there is an element internal to the VP overt in the deleted clause, as soon as she wakes up, so it is certainly a case of NO.

132

Null Object features in Brazilian Portuguese

I will try now to move towards the identification of the features of BP´s NO, taking the opportunity to test the relevance of animacy, as pointed out repeatedly in the literature, trying to list other features, which will contribute, in turn, to a more well-established distinction between NO and VP ellipsis in this language.

2. Background Throughout the literature on BP´s NO, there are two recurring issues that I would like to emphasize in this article. First and foremost is the concern regarding the NO´s inanimacy and secondly, the difficulty in distinguishing this construction from VP ellipsis. These issues will be addressed in the following sections.

2.1. NO and animacy Even the earliest works on BP´s NO, starting with the variationists, had already stated that BP´s NO is inanimate (see Omena 1978 and Duarte 1989). Duarte (1989), for example, made a quantitative analysis of the NO and the other variant forms of the direct object position (accusative clitic, lexical pronoun and anaphoric NP), using data from recordings of spontaneous speech and television language (interviews and soap operas) in search of what would select each variant. Making use of morphological, syntactic and semantic factors, in addition to stylistic and social elements, as determinants of NO, the author chose the semantic component feature inanimacy of the antecedent as the most relevant characteristic for its occurrence, because this would favor NO regardless of the syntactic structure. However, she admitted the occurrence of animate NOs. In fact, all the NO examples given in her text appeared with this feature (see (8) and (9)), which she explained as being a result of social and stylistic factors: (8) O Sinhozinho Malta está tentando convencer o Zé das Medalhas a matar o Roquei... mas ele é muito medroso. Quem já tentou matar __i foi o empregado da Porcinaj. Ontem ele quis matar __i, a empregada é que salvou __i. Ele estava prontinho pra dar o tiro, quando a Mina chegou lá, passou um pito nele e convenceu __j que ele não devia matar __i. ‘Mr. Malta is trying to convince Zé das Medalhas to kill Roque... but he is very frightened. Porcina´s employee already tried to kill []. Yesterday, he wanted to kill [-] and the maid saved [-]. He was

Mirna Sodré Valverde-Hübner

133

ready to shoot when Mina appeared, complained to him and convinced [-] he shouldn´t kill [-].’ (Duarte 1989, 20) (9) No princípio ele não concordava comigo, mas depois eu convenci __ de que ele não devia agir assim. ‘At first, he didn´t agree with me, but later I convinced [-] that he shouldn´t act like this.’ (Duarte 1989, 32) For the author, the inanimate NO would almost entirely pass unnoticed and be naturally accepted in all types of syntactic structures, but its animate variant would also have reasonable acceptance by informants, “except by those with a higher level of education, for whom it causes awkward constructions that themselves make use of in natural speech”2 (32). Thus, the animate NO is considered as common by Duarte. Attention should also be given to the imperfective aspect of the verb matar3, appearing repeatedly in the NO clauses, the relevance of which will be mentioned later. Similarly, Corrêa (1992), even claiming that the use of the null object in BP is limited to the inanimate feature of its antecedent, presents an example of an animate NO (see (10)) to show that adults with a higher level of education also produce the phenomenon, although she did not call attention to animacy in this case: (10) (...) e eles saíram à procura do rapaz. Quando encontraram __, levaram __ para dentro da lanchonete. ‘(...) and they went looking for the boy. When they found [-], they took [-] into the snack bar.’ (Corrêa 1992, 4) There are still other publications differing in this regard that simply stated that BP´s NO is inanimate, without considering the possibility of it being animate (see Omena 1978). From a generative perspective, the hypothesis of the NO´s inanimacy remains. Cyrino (1997), one of the earliest and most important works on BP´s NO to adopt this theory, characterizes the construction in terms of its inanimate antecedent, explaining that there had been a reanalysis of the fall of the neutral clitic, which also had this feature: 2

In the original: “exceto aqueles que se situam num nível de escolaridade mais alto, a quem causam estranheza construções de que eles próprios fazem uso na fala natural”. 3 I´m considering infinitive verbs as imperfectives, without entering into the existing discussion as to whether they must be treated like this or in another way.

134

Null Object features in Brazilian Portuguese

(11) P: - Que é isto sobrinho? R: - Eu o não sei, em minha consciência. Q: - What is this, nephew? A: - I do not know it on my conscience.’ (Cyrino 1997, 169) As the absence of the clitic would not compromise the interpretation, over time, according to the author, all clitics with this feature have been deleted in this language, which would have generated the NO. If the clitic had the animate feature, in this process of disappearing, instead of using the NO, the speaker would have represented the direct object with the lexical pronoun, so that the reference would be made: (12) a. João trouxe a Mariai, mas Pedro não beijou elai. ‘John brought Mary, but Peter didn´t kiss her. b. ??João trouxe a Mariai, mas Pedro não beijou __i. ‘John brought Mary, but Peter didn´t kiss [-].’ (Cyrino 1997, 146) Cyrino states that the sentence in (12) is ungrammatical with NO (12b) because, in her opinion, sentences with animate and specific antecedents generate unacceptable constructions in BP with the NO, a fact indeed largely confirmed by her diachronic data. That is, to this author, there is no NO with an animate and specific/referential antecedent, because BP´s NO would be predominantly inanimate. Thus, she includes the specificity feature as relevant as well, along with inanimacy in BP, relativizing the latter as she admits BP´s NO could be animate and nonspecific (see (13)), though not animate and specific: (13) A FEBEM4 é um dos elos dessa corrente que cria o menor infrator; não é ela o único responsável, o único elo que cria__, e como tal ela não consegue recuperar__. ‘FEBEM is one of the links in the chain that creates the juvenile offender, it is not solely responsible, the only link that creates [-], and as such, it cannot recover [-].’ (Cyrino 1997, 148) However, in Cyrino’s diachronic data, there is an occurrence of animate and specific NO:

4

FEBEM is a Brazilian institution dedicated to the recovery of underage lawbreakers.

Mirna Sodré Valverde-Hübner

135

(14) P: - Descobriram o assassino do crime? R: - Para mim a polícia não descobre __. ‘Q: - Did they find the murderer? R: - For me, the police can´t find [-].’ (Cyrino 1997, 148) The author, however, relativizes the importance of this example, saying that in this case the specificity of the antecedent cannot be categorically confirmed, as there is a reading of the killer of the crime, whoever he is. And even if it were in fact specific, this structure could be considered VP ellipsis, which does not have any animacy restrictions. In fact, it is an ambiguous example between both constructions, because the only element internal to the VP is the direct object and it is elided, so it is difficult to say whether NO or VP ellipsis happens. This is the second question addressed in this article, the apparent confusion between the two phenomena, which I will summarise more fully below. Thus, returning to Cyrino’s work, besides the animacy feature conditioning the null object, which roughly states that this construction is inanimate in BP, the feature of specificity would also have some effect on it for the author, through her observation that the NO can be animate if this feature is combined with the nonspecific feature. Likewise, BP´s NO could be inanimate and specific (see (15), although she did not provide an example of this latter type: (15) Eu não comi o chocolatei hoje, eu comi __i ontem. ‘I didn´t eat the chocolate today, I ate [-] yesterday.’ (ValverdeHübner 2012, 44) Therefore, the role of specificity is not clear in this author’s works, in the sense that she does not systematize its performance in the BP’s NO. Given the example (15), one cannot say that only nonspecific antecedents appear with the NO and likewise, as has been shown, saying that it is always inanimate in BP is not sufficient. It seems to be some combination of these two features that allow the NO in BP and, even so, it seems that other features are involved. In this logical combination, Cyrino, Duarte and Kato (2000) propose a hierarchy of referentiality to account for the deletion of the object, and also the subject, in several languages. In this hierarchy, the referentiality of arguments is determined based only on their features of animacy and specificity:

Null Object features in Brazilian Portuguese

136

(16) Referentiality Hierarchy non-arguments

propositions

[-animate]

[+animate] [-human] [+human]

3rdp

3rdp

3rdp

3rdp 3rdp - 2ndp- 1stp

[-specific] [+specific] [-referential] [+referential] (CYRINO, DUARTE, KATO, 2000) Thus, arguments with the animate feature would be in the highest position in the hierarchy, tending to be filled as they were referentials. This would explain the fact that the personal pronouns eu (I) and tu (you) in BP are always filled. Non-arguments would be in the lowest position. The specificity feature interacts with every other feature in the sense that the more specific, the greater the probability for completion. Therefore, a language that has the internal option of variants filled or null operates influenced by the referential status of the antecedent which, in turn, is defined by considering the animacy and specificity features. This proposal seems to be more interesting because it captures the occurrences of NO considering the combination of these features, explaining, for example, inanimate and specific NOs by the performance of specificity toward referentiality. However, problems regarding the treatment of deletion of the object only in terms of these two features, those relevant to this hierarchy, remain, because, as has been shown, there are occurrences of animate and specific NOs. This notion of hierarchy, however, will be useful, although in another form, for the analysis of the phenomenon in BP which I will present later. I will now examine the literature of BP´s NO with regard to relations with the construction of VP ellipsis.

2.2. NO x VP Ellipsis The second question which I have listed concerning the literature on BP´s NO addresses the relation between this construction and VP ellipsis, given the challenges to define the boundaries between these two phenomena based on the available literature for this language. One of the statements found regarding this topic (see Cyrino, 1997) is that animate and specific NO would only be possible in the VP ellipsis structure, as in

Mirna Sodré Valverde-Hübner

137

(6), repeated here as (17) (see also (14)), classified as VP ellipsis thanks to the adjunct no parquinho elided along with the direct object in the clause in which the deletion happens: (17) A mãe do Marquinhos sempre leva elei no parquinhoj, mas seu pai não costuma levar__ij . ‘Marquinho´s mother always takes him to the playground, but his dad does not usually take [-]. (Valverde-Hübner 2012, 14) However, as has already been showed, there are NO examples which contradict this statement, which occur with an animate and specific antecedent, as in (5), here repeated as (18). In this case the construction is certainly NO thanks to the overt adjunct assim que acorda in the clause with the deletion: (18) A Maria escova o cachorroi antes de dormir e a Joana escova __i assim que acorda. ‘Mary brushes the dog before she goes to bed and Joana brushes [-] as soon as she awakes.’ (Valverde-Hübner 2012, 14) Another common statement that also connects the two constructions is that both of them display strict and sloppy readings, as shown in (19), a NO example, and in (20), a VP ellipsis example: (19) Maria trouxe a mochilai para a escola e João trouxe __i para o curso de inglês. ‘Mary brought her backpack to school and John brought [-] to the English class.’ (Valverde-Hübner 2012, 90) (20) A Ana já tinha lido o livro à irmã, mas a Paula não tinha __. ‘Ana had already read the book, but Paula hadn´t [-].’ (Cyrino and Matos 2002, 4) In a first reading, in construction (19), a NO one, John brought Mary´s backpack to the class, in the same way that Paula, in the VP ellipsis construction (see 20), read the same book that Ana did, i.e., there is the possibility of strict reading in both. Likewise, in (19), John could have brought his own backpack to school, and Paula could have read her own book, which constitutes the sloppy reading. A third similarity between the two constructions would have to do with their licensing, since both would result from the verb raising to INFL (see

138

Null Object features in Brazilian Portuguese

Matos 1992 and Cyrino 1997). Furthermore, for Cyrino (1997), both phenomena are explained as a result of ellipse in PF (phonological form) and reconstruction5 in LF (logical form). So many similarities lead the author to assert in Cyrino (2006) that NO is actually a case of NP ellipsis. Furthermore, the same author had already stated in Cyrino (1997, 61), noticing similarities between the two phenomena, that: “VP ellipsis (...) is a process which merges with the null object, since the verb is left without its complement, without lexical restriction”6. Differences between NO and VP ellipsis are also presented in the literature in an attempt to distinguish the two constructions. Cyrino and Matos (2002), for example, indicate some of these differences, the first being the quantity of VP elided elements, only the direct object in the NO (see (21a)) and various elements, including arguments and adjuncts, in VP ellipsis (see (21b)): (21) a. Ela trouxe o computador para a Universidade e ele trouxe [-] para o escritório. ‘She brought the computer to the University and he brought [-] to the office’ (Cyrino and Matos 2002, 5) b. Quando a Ana pôs os óculos na mesa, a Maria também pôs [VP-]. ‘When Ana placed the glasses on the table, Maria did [-] too.’ (Cyrino and Matos 2002, 6) Furthermore, the authors assert that there is a verbal identity needed for VP ellipsis (see (21b), here repeated as (22a)) and this is not so for NO (see (22b)): (22) a. *Quando a Ana colocou os óculos na mesa, a Maria também pôs __. ‘When Ana laid the glasses on the table, Maria put [-] too.’

5

The author reproduces the definition of reconstruction given in Fiengo and May (1994), for whom it is, under condition of syntactic identity, a copy of a structure, in this case, the antecedent of the NO, at the LF level, without reflections in FF. 6 In the original: “a elipse de VP (...) é um processo que se confunde com o objeto nulo, visto que o verbo é deixado sem seu complemento, sem restrição lexical”.

Mirna Sodré Valverde-Hübner

139

b. Olhou para a fotografia daquele homem. Reconheceu [-] imediatamente: era o João. ‘(He/she) looked at the picture of that man. (He/she) recognised [-] immediately: it was John.’ (Cyrino and Matos 2002, 6) However, if VP ellipsis is the omission of all internal arguments and adjuncts of the VP with verbal identity, examples like (23) remain ambiguous between the two constructions because, as (24) shows, clearly a case of NO because of the overt indirect object, the NO can also appear with this identity: (23) O João leu esse livro e a Ana também leu __. ‘John read this book and Ana also did [-].’ (Cyrino and Matos 2002, 5) (24) Ela trouxe o computador para a Universidade e ele trouxe __ para o escritório. ‘She brought her computer to the University and he brought [-] to the office.’ (Cyrino and Matos 2002, 5) Therefore, verbal identity on its own could not be used to differentiate VP ellipsis from NO. Another difference indicated by Cyrino and Matos (2002) is that there is the possibility of displaying the NO by means of a pronoun (see (25a)), but not the VP ellipsis (see (25b)): (25) a. Ela tirou o anel do dedo e guardou-o no cofre. ‘She took the ring from her finger and put [-] in a safe.’ (Cyrino and Matos 2002, 6) b. A Ana pensa muito nos amigos, enquanto o Pedro não (*o) pensa. ‘Anna is thoughtful to her friends while Peter isn´t (*it)’. (Cyrino and Matos 2002, 7) However, in my research, a VP ellipsis example like (26), with the ellipsis being shown by a pronoun, was classified as grammatical by 70% of my informers:

140

Null Object features in Brazilian Portuguese

(26) a. Nenhum de nós tinha votado nesse candidato nem admitíamos que alguém o tivesse ‘None of us had voted for that candidate or admitted that someone had it’. (Cyrino and Matos 2002, 7) Thus, there does not seems to be enough instruments in the literature to safely differentiate the two constructions. Moreover, in regard to the NO, it has been shown that animacy on its own or in combination with specificity are not enough to characterize the construction. There are still clauses with only the direct object deleted (see (7)), which would be ambiguous in terms of NO and VP ellipsis (see Goldberg 2005 and Cyrino and Matos 2002), and finding no support for a suitable classification in literature on BP. I will now present my analysis for these issues, hoping to cast some light on them.

3. Analysis In order to deal with the two issues raised in this article, namely, the relevance of animacy in the characterization of BP´s NO and the difficulties of establishing a suitable distinction between NO and VP ellipsis in that language, I will use the work of Goldberg (2005), which presents a diagnosis for VP ellipsis in natural languages and ways to differentiate it from an analysis in terms of null arguments, such as NO, even using semantic features such as animacy to do this. The author analyses Hebrew, for instance, as a language which displays both well-defined NO and VP ellipsis constructions, resulting, as in BP, in the issue of classifying ambiguous sentences, that is, with only the direct object deleted (see (7)), and the author states that they are distinct through the animacy feature. In this language, Goldberg noted a strong animacy restriction on the null object, the same assigned to BP in the literature, in the sense that Hebrew’s NOs are inanimate. Goldberg tested this condition in many kinds of constructions (coordinated sentences, coordinated verb phrases, question-answer pairs and adjacent sentences uttered by the same speaker) and with many kinds of antecedents (singular, plural, feminine, masculine, count nouns and mass nouns, definite and indefinite), and confirmed her hypothesis in all these cases. As a way of ensuring this animacy restriction in the Hebrew’s NO, the author provides NO´s examples (27) – (30), so classified because the rest

Mirna Sodré Valverde-Hübner

141

of the VP is overt, with animate antecedents to make sure that the ungrammaticality is presented, which, in fact, it is: (27) *Šmu’el hošiv et ha-yeladot al ha-mita, ve-Dina hilbiša besimlot. Shmuel sit[Past3Msg] ACC the-girls on the-bed and-Dina dress[Past3Fsg] in-dresses ‘Shmuel sat the girls on the bed, and Dina dressed (them) in dresses.’ (28) *Rina hisi’a et Gil ha-’ira ve-horida le-yad ha-bayit. Rina drive[Past3Fsg] ACC Gil the-town and-drop[Past3Fsg] tonear the-house ‘Rina drove Gil to town and dropped (him) near his home.’ (29) Q: Eyfo ha-’iš še-’amad po lifney rega? where the-man that-stand[Past3Msg] here before moment ‘Where (is) the man who stood here a moment ago?’ A: *Miryam hovila la-misrad. Miryam lead[Past3Fsg] to.the-office ‘Miryam led (him) to the office.’ (30) *Hine ha-yeladot šeli. Šošana hisi’a le-Tel-’Aviv etmol. here the-girls of.me Shoshana drive[Past3Fsg] to-Tel-Aviv yesterday ‘Here (are) my daughters. Shoshana drove (them) to Tel-Aviv yesterday. (Goldberg 2005, 49) Moving these examples to BP, I believe that the equivalent of (28) and possibly the equivalent of (29) are grammatical in this language, with an animate antecedent. This could indicate that the animacy restriction would not be so decisive in BP in comparison with Hebrew. Returning to the process followed by Goldberg (2005) in establishing the distinction between NO and VP ellipsis in Hebrew, the author then shows structures which are clearly VP ellipsis (with more than one internal argument in the VP elided) with animate antecedents and notices that the restriction no longer remains, as shown in the Hebrew example below:

142

Null Object features in Brazilian Portuguese

(31) Q: (Ha-’im) Miryam hisi’a et Dvora la-makolet? Q Miryam drive[Past3Fsg] ACC Dvora to.the-grocery.store ‘(Did) Miryam [drive Dvora to the grocery store]?’ A: Ken, hi hisi’a. yes she drive[Past3Fsg] ‘Yes, she drove [Dvora to the grocery store].’ (Goldberg 2005, 53) The author’s conclusion is that Hebrew has null object only with inanimate antecedents. With animate direct objects, only VP ellipsis could be employed. In cases of ambiguity, therefore, where there is only one internal argument to the VP to be elided, with an animate antecedent, the construction would be a VP ellipsis, excluding the alternative of being NO in these cases, because VP ellipsis, according to Goldberg, has no animacy restrictions (see also Cyrino, 1997). Applying this same process to BP, I intend to confirm my impression that in this language the animacy restriction is not that categorical. I began by testing whether there are structural restrictions regarding the occurrence of NO in those structure types listed by Goldberg (coordinated sentences, coordinated verb phrases, question-answer pairs and adjacent sentences uttered by the same speaker), respectively (32), (33), (34), (35), and with various types of DPs (defined ((32) – (34)), mass nouns (35), count nouns ((32) – (34)), feminine ((33b), (34)), masculine ((32a), (33a), (35)), singular ((34), (35)), plural ((32) – (33)) and indefinite (35a), testing, at the same time, the animacy restriction (examples (b)): (32) a. João levou os presentes para o quarto e Maria desembrulhou __ na sala. ‘John took the gifts to the bedroom and Mary unwrapped [-] in the living room.’ b. ??João acordou as meninas às 07h e Maria arrumou __ às 08h. ‘John woke up the girls at 7 a.m. and Mary got [-] ready at 8 a.m.’ (33) a. João comprou os cadernos e deixou __ no quarto da Maria. ‘John bought the notebooks and left [-] in Mary´s room.’ b. ??João beijou as filhas e colocou __ no carro. ‘John kissed his daughters and put [-] in the car.’

Mirna Sodré Valverde-Hübner

143

(34) a. P: Onde está minha carteira? R: Coloquei __ na gaveta. ‘Q: Where is my wallet? A: I put [-] in the drawer.’ b. P: Onde está a Maria? R: Levei __ para casa. ‘Q: Where is Mary? A: I drove [-] home’ (35) a. Tem dinheiro aqui. Não tinha visto __ ontem. ‘There is some money here. I didn´t see [-] yesterday.’ b. ??O Pedro não está aqui. A Maria deve ter levado __ para a escola. ‘Peter isn´t here. Mary must have driven [-] to school.’ (Valverde-Hübner 2012, 91) Given these and many other examples of animate NO that have appeared in Valverde-Hübner (2012), from several authors, I concluded that the BP productive strategy for NO does not have an animacy restriction as strong and as categorical as Hebrew, although, notably, animacy is one of the relevant features of this construction. Thus, I conclude that the ambiguity between NO and the VP ellipsis cannot be undone in those classic examples only by animacy, and requires this to be associated with other important features to make a more well-established distinction. Because of this difficulty in characterizing BP´s NO only through the animacy feature, with the consequent inability to differentiate it from VP ellipsis based only on that feature, my hypothesis is that there is a hierarchy of features acting in BP NOs to license them, in addition to animacy and specificity, as seen in the hierarchy proposed by Cyrino, Duarte and Kato (2000), also considering the imperfective aspect feature and possibly others. The example below shows the contrast between a perfective sentence in (36a) and an imperfective one in (36b) in NO constructions: (36) a. Ah, eu usei a tua blusa hoje, tenho que lavar ela / *__. ‘Oh, I wore your blouse today, I have to wash it/ *__.’ (Casagrande 2010, 2)

144

Null Object features in Brazilian Portuguese

b. Quando eu usava suas blusas eu sempre lavava __. ‘When I used your blouses, I always washed [-].’ (ValverdeHübner 2012, 92) In these examples, the imperfective aspect seemed to favor the NO. In my research in Valverde-Hübner (2012), I also collected lots of examples of NO with this feature (see (37) – (39)), along with data from other authors (see 8): (37) - A Bia roubou o buquê da Sofia e não queria devolver __. - Mas também, Maria, como é que você dá um buquê para uma menina e não dá __ para a outra? ‘- Bia has stolen Sophie´s bouquet and didn´t want to give [-] back to her. - Of course, Mary, how can you give a bouquet to a girl and not give [-] to the other one?’ (38) Alguém quer mais bolo? Eu vou parar de servir __, eu também não quero mais__. ‘Does anyone want more cake? I´m going to stop offering [-], I also don´t want [-] anymore.’ (39) P: - Você acha que eu coloco a lasanha mais tempo no fogo? R: - Deixa eu ver __, amiga. ... Acho que pode colocar __. Pode deixar que eu vou pôr __. É sempre melhor sobrar __ do que faltar __. ‘Q: - Do you think I should leave the lasagna longer in the oven? A: - Let me see [-], my friend.... I think you can put [-], I´ll put []. It´s always better to have more [-] than to have less [-].’ (Valverde-Hübner 2012, 86) In my preliminary proposal, I assume that the imperfective feature is part of the set of features which characterize BP´s NO, due to their importance and repeated occurrence in the data, although it occupies a lower position than animacy, because this feature has also not proved to be categorical on its own. Nevertheless, it could explain the NO construction in (5), with an animate antecedent, repeated here as (40):

Mirna Sodré Valverde-Hübner

145

(40) A Maria escova o cachorroi antes de dormir e a Joana escova __i assim que acorda. ‘Mary brushes the dog before she goes to bed and Joana brushes [-] as soon as she awakes.’ (Valverde-Hübner 2012, 14) This example (40) is a case of an animate and specific NO, which should be blocked according to all the BP literature; however, it occurs due to the influence of imperfectivity in my analysis. I assume that these features (animacy, specificity and perfectivity) follow a hierarchy because the animacy feature seems to occupy a higher position, since, in addition to appearing in all recent work on the NO in BP, it in fact plays a more decisive role. Thus, in (41), we have a sentence where, despite the imperfective aspect, animacy and specificity are hampering the occurrence of NO. In contrast, in (42), we have an example of inanimate NO, specific and with perfective aspect which is perfectly possible: (41) ??O João respeitaria a Mariai se tivesse conhecido __i. ‘John would have respected Mary if he had known [-].’ (Valverde-Hübner 2012, 95) (42) João descascou a banana, mas Pedro não comeu __. (Cyrino 1997, 146) ‘John peeled the banana, but Peter did not eat [-].’ Thus, we are not claiming, given the construction in (42), which displays a specific antecedent and a perfective aspect, that all the three features pointed out as relevant – animacy, specificity and perfectivity – must be absent for NO to occur, although this is the ideal situation for its occurrence. By absence of these features, I mean a construction with an inanimate and nonspecific antecedent and with perfective aspect. I´m just stating that these three features, initially, would act in a hierarchical relationship in BP’s NO occurrences. However, if these features are all displayed, on the other hand, the NO would be barred: (43) ??Ontem o João beijou a Maria na escola, e o Pedro beijou __ hoje no shopping. ‘Yesterday John kissed Mary at school, and Peter kissed [-] at the mall today’.

146

Null Object features in Brazilian Portuguese

(44) *O Marcos cumprimentou a Joana, mas não chamou __ para a festa. ‘Mark greeted Joana, but he did not invite [-] to the party.’ (45) *A Bia chorou ontem assim que eu coloquei __ no berço. ‘Bia cried as soon as I put [-] in the cot.’ (46) *A Maria chamou o João e mandou __ ir pra escola. ‘Mary called John and ordered [-] to go to school.’ (Valverde-Hübner 2012, 96) As a preliminary proposal, I have not mapped the position of all these features in the hierarchy in an exact manner, I have only specified that animacy would be the most important one, occupying a higher position than the others. I must also note the distinct nature of the features that make up the hierarchy, one of them being typical of the antecedent noun – animacy, and other, broader in terms of the computation of the DP – specificity – and the latter one, still wider – perfectivity, with it being related to the whole structure. Perhaps the relevance of these features in the hierarchy is proportional to the closeness of its relationship with the retrieved noun. This could be addressed in a future work. Applying these findings to Goldberg’s proposal (2005), in order to distinguish the NO from VP ellipsis, and in front of ungrammatical constructions with an animate antecedent, which is specific and has the perfective aspect, along with another overt element internal to the VP, to mark the NO strategy, as is shown in (43-46), it is expected that in the construction of VP ellipsis, which has no semantic restrictions, this configuration of features could appear, which also seems to be confirmed, as the examples (47) and (48) show: (47) O João beijou a Mariai na escolaj, e o Pedro também beijou __ij. ‘John kissed Mary at school, and Peter also did [-].’ (48) A Maria encontrou a Joanai ontemj no shoppingk, e a Júlia também encontrou __ijk. ‘Mary met Joana yesterday at the mall, and Julia also did [-].’ (Valverde-Hübner 2012, 96) This leads to a possible analysis of those ambiguous sentences, with only one argument internal to VP and with that element being deleted, the direct object, as in (49):

Mirna Sodré Valverde-Hübner

147

(49) O João esbofeteou a Mariai, mas o Marcos não esbofeteou __i. ‘Mary slapped John, but Mark didn´t slap [-].’ (Valverde-Hübner 2012, 97) Because of the animate and specific object which is elided, Maria, and as it is a perfective construction, I would argue that this is a case of VP ellipsis, since the NO, which has restrictions due to these features, would be barred in this case. Thus, the ambiguous sentences could be classified as NO or VP ellipsis according to the combination of features displayed.

4. Conclusion Throughout this article, two issues have been raised in the literature concerning BP´s NO which needed to be clarified: i) the predominant role of animacy as a NO feature in this language and ii) the difficulty of drawing a suitable distinction between NOs and VP ellipsis constructions in BP. To solve the former, I have argued that, in addition to animacy and specificity, imperfectivity can also be important in the characterization of BP’s NOs, according to a hierarchy in which animacy occupies the highest position. Imperfectivity proved to be relevant as it could account for some occurrences of NO with an animate and specific antecedent, as shown in (5). As regards the second issue, I conclude that the aspects to distinguishing the two constructions listed in the literature also do not account fully characterize this, because the topics listed, such as verbal identity and the use of a pronoun to recuperate the elided elements, were neither categorical nor useful with regard to an ambiguous construction where there is a direct object as the only elided element internal to the VP. Given this, the mechanisms of Goldberg (2005) for distinguishing the NO from VP ellipsis in a language that has the two constructions and where the animacy feature seems to characterize the NO, as is the case in Hebrew and supposedly in BP, were used. In my analysis for BP, however, I showed that the animacy restriction for BP´s NO was not as categorical as in Hebrew, which led me to search for more features of this construction. In this search, I was led to a hierarchy of features, as proposed in Valverde-Hübner (2012), in which imperfectivity appears as a BP NO feature, albeit in a lower position than animacy. Thus, I could conclude that in an ambiguous construction which might be NO or VP ellipsis, with the direct object as the only argument elided, and which displays animacy,

148

Null Object features in Brazilian Portuguese

specificity and perfectivity features, VP ellipsis would occur, not NO (see (43) – (46)). I do not exclude, however, the possibility of other features acting in this NO hierarchy in BP, which could help to clarify the combination of these features even more, especially when some of them are present and others absent in the NO occurrence, which might be the object of future work.

References Casagrande, Sabrina. 2010. “A correlação entre aspecto e objeto no PB: uma análise sintático-aquisicionista”. PhD diss., Universidade de Campinas. Corrêa, Vilma Reche. 1992. “O Objeto Direto Nulo no Português do Brasil”. MA Diss, Universidade Estadual de Campinas. Cummins, Sarah and Yves Roberge. 2004. “Null objects in French and English”. In Contemporary Approaches to Romance Linguistics: Selected papers from the 33rd Linguistic Symposium on Romance Languages (LSRL), edited by J. Auger, J. C. Clements, and B. Vance, 121-138. John Benjamins, Amsterdam. Accessed January, 10, 2011. http://homes.chass.utoronto. ca/~roberge/ pdf / null_objects.pdf. Cyrino, Sônia M. L.1997. “O objeto nulo no Português do Brasil: um estudo sintático-diacrônico”. Londrina: Ed. UEL. http://www.unicamp.br/iel/site/docentes/cyrino/Publications.htm Cyrino, Sônia M. L.. 2006. “Algumas questões sobre a elipse de VP e objeto nulo em PB e PE”. In Teoria e análise lingüísticas: novas trilhas, edited by M. Guedes, R. Berlinck, and C. Murakawa, 53-79. Accessed July, 15, 2011. http://www.unicamp.br/iel/site/docentes/cyrino/araquararaTXT.pdf Cyrino, Sônia M. L., Maria E. L. Duarte, and Mary Kato. 2000. “Visible subjects and invisible clitics in Brazilian Portuguese”. In Brazilian Portuguese and the Null Subject Parameter, edited by Mary Kato and Esmeralda Vailati Negrão, 55-73. Frankfurt and Madrid, VervuertIberoamericana. Cyrino, Sônia M. L., and Gabriela Matos. 2002. “VP ellipsis in European and Brazilian Portuguese – a comparative analysis”. Journal of Portuguese Linguistics 1(2): 177-196. Accessed March, 13, 2010. http://www.unicamp.br/iel/site/docentes/cyrino/Cyrino%20and%20 Matos%20VPel%20in%20JPL%20com%20citacao.pdf Duarte, Maria E. L.. 1989. “Clítico acusativo, pronome lexical e categoria vazia no português do Brasil”. In Fotografias Sociolingüísticas, edited

Mirna Sodré Valverde-Hübner

149

by Fernando Tarallo, 19-34. Campinas: Universidade Estadual de Campinas. Farrell, Patrick. 1990. “Null Objects in Brazilian Portuguese”. Natural Language and Linguistic Theory 8: 325-46. Fiengo, Robert and Robert May. 1994. “Indices and Identity”. Cambridge: MIT Press. Galves, Charlote. 1989. “O Objeto Nulo no Português Brasileiro: Percurso de uma Pesquisa”. Cadernos de Estudos Lingüísticos, 17: 65-90. Goldberg, Lotus. 2005. “Verb-Stranding VP Ellipsis: A Cross-Linguistic Study”. PhD diss., McGill University. Huang, C. T. James. 1991. “Remarks on the Status of the Null Object”. In Principles and Parameters in Comparative Grammar, edited by R. Freidin, 56-76. Cambridge: MIT Press. Kato, Mary. 1993. “Recontando a História das Relativas em uma Perspectiva Paramétrica”. In Português Brasileiro – uma Viagem Diacrônica, edited by Roberts, and Kato, 223-261. Campinas: Editora da UNICAMP. Lopes, Ruth Vasconcellos. 2006. “Traços semânticos na aquisição da linguagem”. Letras de Hoje. 41:161-178. Matos, Gabriela. 1992. “Construções de Elipse de Predicado em Português – SV Nulo e Despojamento”. PhD diss., Universidade de Lisboa. Omena, Nelize Pires. 1978. “Pronome Pessoal de Terceira Pessoa: suas Formas Variantes em Função Acusativa”, MA diss., Pontifícia Universidade Católica do Rio de Janeiro. Raposo, Eduardo. 1986. “On the Null Object in European Portuguese”. In Studies in Romance Linguistics, edited by Jaeggli, O and Carmen Silva-Corvalán, 373-390. Dordrecht: Foris Publications. Rizzi, Luigi. 1986. “Null Objects and the Theory of pro”. Linguistic Inquiry 17: 501-558. Valverde-Hübner, Mirna Sodré. 2012. “Em busca de uma caracterização para o objeto nulo no Português Brasileiro”. MA Diss, Universidade de Brasília.

SUBJECT-VERB INVERSIONIN DECLARATIVEEXCLAMATIVE SENTENCES IN THE PORTUGUESE LANGUAGE RITA VALADAS PEREIRA UNIVERSIDADE DE LISBOA

Abstract In this paper, I will analyze the semantic and pragmatic properties of the Subject-Verb Inversion (SVI) that occurs in Portuguese sentences such as Comprou a Joana a camisola para não a usar!1 or Gosto eu de morangos e tu não me dás nenhum!2. In these kinds of sentences, the canonic order of a simple declarative sentence – SVO – changes to VSO, and the resultant meaning is not simplistic, as that change gives a simultaneous modal and assertive content to the sentence. It is argued that this syntactic inversion influences the semantics of the sentence. With the use of SVO canonic order, the sentence would be a mere description of the world, but with the VSO order, and with a specific exclamatory intonation, the sentence acquires a subjective feature which reveals an implicit comment, i.e. the expression of a speaker’s attitude. At the same time, new information is introduced about what is being commented on. Therefore, SVI modalizes the speech and gives these sentences properties that are both exclamative and declarative. 1

For all examples given, a gloss with direct word correspondence will be provided, followed by a possible translation with the aim of conveying and explaining the Portuguese meaning of the sentence. For this particular sentence, we have: Bought Joana the shirt for not to wear it. ‘Joana bought the shirt, but now she doesn’t wear it.’ (Joana should wear the shirt, since she bought it!) 2 Like I strawberries and you are eating them all! ‘I like strawberries, but you are eating them all alone!’ (That is not right. You should offer me a strawberry, since I like them so much.)

Rita Valadas Pereira

151

Keywords: Subject-Verb Inversion, Declarative-exclamative sentence, Discursive modalization, Assertiveness.

1. Introduction Subject-Verb Inversion (SVI) has been considerably analyzed in grammar literature regarding the Portuguese language, and is mainly considered as a syntactic process of change (or movement) in the basic sentence structure. Authors like Ambar (1992), Costa (2004) and others have analyzed and schematized several constructions in which Portuguese syntax allows or demands the change of Subject and Verb order in a sentence. This shows that, most of the time, the syntactic structure of specific sentences such as interrogative forms or structures with preposed constituents, among many others, is particularly characterized by this change. The aim of the present work is to analyze and to problematize, mostly semantically and pragmatically (although without ignoring the syntactic point of view), a specific context of Subject-Verb Inversion (SVI) in European Portuguese. The analysis will focus particularly on the meaning of this kind of construction, in order to explain that its form (VSO) is a consequence of particular semantic and pragmatic features in the sentence (such as the speaker’s intentions or meaningful nuances). I will analyse sentences such as (1), mainly from created data, taking into account the importance of linguistic intuition as a variable. Notice that these kinds of structures have not yet been studied in literature (which has been more focused on non-assertive, less simple and well-marked SVI contexts). (1) a. Comi eu o chocolate e afinal havia bolo! Ate I the chocolate and after-all there-was cake! ‘I ate the chocolate, but there was cake after all!’ (I should not have eaten the chocolate, because there was cake, which I prefer). b. Comprou a Joana a camisola para agora não a usar! Bought the Joana the shirt for now not it to-wear! ‘Joana bought the shirt, but now she does not wear it!’ (This should not be the case. She should wear it, since she bought it!) In the above, the Verb-Subject-Object (VSO) order seems to be clearly influenced by matters of semantics and pragmatics, since the SubjectVerb-Object (SVO) order that would be expected in the corresponding

152

Subject-Verb Inversion in Declarative-Exclamative Sentences

simple declarative sentence is here altered, giving it a new (mainly exclamatory) pragmatic status. The analysis of these SVI cases will lead to a questioning of several concepts concerning the clauses it integrates, with regard to the sentence type or speech act, the basic order concept, verb types and verb tenses used, the type of subject, and the role of meaning systems (such as reference, definiteness, or modality) in the syntactic structure. This will demonstrate the interaction between syntax and semantics justifying the inversion studied here. After a brief presentation of other contexts in which Portuguese allows a Subject-Verb Inversion, this work will present a more profound analysis of this specific case of SVI. It will be shown that the inverted order is actually due to semantic and pragmatic issues that determine the type of verbs, subjects, and clausal contexts in which this process can be observed.

2. Other contexts of Subject-Verb Inversion in Portuguese Regarding the inversion process, it is important to first highlight that the basic order of Portuguese sentences is the one less marked in tone or intentionality. Therefore, the one easily accepted by any native Portuguese speaker as the canonic order for a simple declarative sentence is SubjectVerb-Object (SVO)3. The inversion analyzed here will have this canonic order as its primary basis, but it will focus only on the subject4 and verb of a sentence, without any change of the position of the object: the object will remain, in most cases, in the third position. In Portuguese, there is a group of clausal contexts where the SVO order is naturally altered to VSO without resulting in any 3

Notice that, as Ambar (1992, 22) states, the combination of elements S(ubject) – V(erb) – O(bject) may always produce, mathematically and abstractly, six possible orders: SVO, VSO, SOV, OSV, OVS and VOS. The basic order will be given by one of these six hypotheses, the one that happens without the need for any special syntactic, semantic, phonetical, morphological or pragmatic elements to be legitimized by a speaker. 4 “Subject” is here considered by taking into account the considerations of Brito and Duarte (1980), as a syntactic function that is applied to a sentence’s constituent (typically they are nominal phrases or clauses) due to semantic-pragmatic issues. The subject of a sentence will be the argument of the predication which, being mandatory, is semantically related to the predicator; it is “’the starting point’ for the state of affairs expressed by a certain predication” (Brito e Duarte 1980, 225 – author’s translation). Consequently, the subject determines the verbal agreement of the sentence, and it is the preferential controller for inter or intra-clausal coreferentiality.

Rita Valadas Pereira

153

ungrammaticality, and this inversion is even mandatory sometimes. This group, which has been substantially described (cf. Ambar 1992), implies constructions such as interrogative structures, question-answer contexts, parentheticals, contexts of focused preposed constituents, infinitive structures, Subjunctive structures, participle structures, and ergative/unaccusative or even passive constructions. The syntactic characteristics of all of these justify choosing the VSO order instead of the canonic SVO order. Although the analysis or revision of these contexts does not fall within the scope of this paper, there are five cases that can be pointed out in which the inversion is not only justified by syntactic reasons, but it is also justified by semantic-pragmatic properties, bringing them closer to the contexts analyzed in this study. According to Ambar’s (1992) scheme, these are cases of closed interrogatives or yes/no questions (cf. 2), of a type of dependent infinitive structure (cf. 3), independent Subjunctive structures (cf. 4a), subordinated Subjunctive structures without the connector “se” [if] (cf. 4b/4c), correction of a previous statement contexts (cf. 5), and contexts of preposed contrastive focus (cf. 6)5, which have been commented on by Costa and Martins (2010). (2) a. Irá o Pedro ao cinema? Will-go the Pedro to-the cinema? ‘Will Pedro go to the cinema?’ b. Terias tu tomado outra atitude?5 Would-have you taken another attitude? ‘Would you have done something else?’ (3) Dizeres-me tu a verdade!6 Telling-me you the truth! ‘As if you are ever going to tell me the truth!’ (4) a. Digam eles a verdade! Tell they the truth! ‘Let them tell the truth!’

5

In these cases, there is an object movement, since it is preposed to the verb, causing an OVS order. However, what remains relevant to this study is that the verb still comes before the subject. 5 Examples from Ambar (1992, 64-65) 6 Examples from Ambar (1992, 89)

154

Subject-Verb Inversion in Declarative-Exclamative Sentences

b. Tivesse eu tempo e havias de ver como tudo corria bem. Had I time, and [you] would-see how everything would-go well. ‘Had I the time, and you would see how everything would work out.’ c. Estudasse o Pedro um pouco mais e verias como ele é brilhante.7 Had-studied the Pedro a little more and [you] would-see how he is brilliant. ‘Had Pedro studied a little harder and you would see how brilliant he is.’ (5) A: Ninguém comeu a sopa. No-one ate the soup. B: Comeu o João a sopa.8 Ate the João the soup. ‘João ate the soup.’ (6) a. Isso queria o escritor. That wanted the writer. ‘It was that what the writer wished.’ b.Uma melancia inteira me comeu aquele bruto.9 A watermelon whole me ate that beast. ‘That beast ate all of my watermelon.’ In all of these cases, the inversion is not determined by a syntactic element like the wh-constituent, as happens in other SVI cases. Instead, there are specific semantic-pragmatic features that justify the canonic SVO order not being applied, such as the verb tenses or verb moods used (Future, Conditional or Subjunctive), and, of course, the order inversion. The sentences of (2), with Future or Conditional and inversion, are not mere questions, because the speaker seems to have doubts concerning a state of affairs, and s/he is not only asking for information. As Ambar (1992) points out, these questions express a speaker’s attitude because the verb tense and mood, as well as the verb’s position, indicate that “the speaker does not wait for a categorical yes/no answer from the listener: instead, he expresses doubt about a particular state of affairs, seeking the confirmation or disconfirmation of his hypothesis” (Ambar 1992, 66 – 7

Examples from de Ambar (1992, 102, 106) Examples from Costa (2004, 124) 9 Examples from Costa and Martins (2010) 8

Rita Valadas Pereira

155

author’s translation). The pre-subject verb position will have a modality that overcomes its most basic and broader meaning, and the speaker will be forced to select verb tenses with broader modal features that assist the construction of the discursive context intended. If the question were a simple request for information, the SV order would be perfectly acceptable – O Pedro irá ao cinema? [‘Will Pedro go to the cinema?’], or A Joana gostaria de ir ao cinema? [‘Would Joana like to go to the cinema?’] –, notwithstanding the fact that these verb tenses are not the most usual in a native speaker’s day-to-day routine: the most usual would be O Pedro vai ao cinema? [‘Is Pedro going to the cinema?’] and A Joana gostava de ir ao cinema? [‘Would Joana like to go to the cinema?’]10. However, if a discursive intention, like doubt, is grammaticalized, modalizing the question, it will be more common for the native speaker to choose the VSO order. The same attitude’s grammaticalization can be observed in sentences (3) and (4). Ambar (1992) shows that they are expressive speech acts (according to Searle’s (1969) terminology), because they contain a null modal operator (which is not produced phonetically, and being implicit), which explains why these sentences are modalized: they express the speaker’s intention. In (3), placing the subject after the verb leads to an implied evaluation of the proposition’s truth value: there is a certain discredit concerning the truthfulness of the proposition, taking into account that there is at least one possible world where the proposition is not considered true or truthful. So, a comment such as “it is not possible that p” or “I do not agree that p” becomes implicit. In (4), the modality which is generally associated with the Subjunctive mood is connected with the SVI, and this creates sentences that are not simple assertive predications: in (4a) there is an imperative intention, as clear as the conditional intention of (4b). In (4b), the consideration of a possible world (the implicit value of se [if]) is shown not only by using the Subjunctive, but also through the inversion of the verb and subject order, so that, when the conjunction is explicit, the SVO order is accepted again (Se eu tivesse tempo, havias de ver como tudo corria bem [‘If I had the time, you would see how everything would work out.’]). The same happens with the example (4c).

10

In Portuguese, there are two different constructions that can be translated to “would Joana like to go”: ‘A Joana gostaria de ir’ and ‘A Joana gostava de ir’. The former corresponds to the regular present conditional in English, but the second, which is called the ‘pretérito imperfeito’ (past imperfect), does not really exist in English. In Portuguese, both these forms have the same meaning in this kind of interrogative sentence.

156

Subject-Verb Inversion in Declarative-Exclamative Sentences

On the other hand, in sentences (5) and (6) the inversion semantically and pragmatically marks an attitude of opposition or contradiction regarding an expectation. That expectation may be presented in an explicit proposition, as is the case in (5), or may be implicit, as is the case in (6). These are cases of informational focus that add new information and express an attitude at the same time. To sum up, these five cases are examples of Portuguese contexts in which VSO order might be used to form sentences where the semantics includes issues of not only reference and predication, but also discursive modalization, sharing features with those described in this work.

3. Subject-Verb Inversion in declarative-exclamative sentences: semantic-pragmatic properties This paper’s object of study comprehends a type of SVI structures that has not yet been deeply analyzed in the grammar literature (cf. Martins 2010). They do not directly relate (at least not in their syntactic structure) to the contexts described before (cf. section 1), except in what concerns the studied semantic-pragmatic process of inversion. These are sentences such as the paradigmatic examples from (7) to (13): they all have an SVI which endows them with an exclamatory and modalized character, and so they become a kind of a speaker’s comment. (7) a. Comi eu o bolo e (afinal) o jantar estava pronto! Ate I the cake and (after-all) the dinner was ready! ‘I ate the cake but (after all) the dinner was ready!’ (I should not have eaten the cake because that spoiled my appetite for dinner.) b. Comi eu o bolo quando o jantar estava pronto! Ate I the cake when the dinner was ready! ‘I ate the cake when after all the dinner was ready!’ c. Comeu a Rita o bolo e (afinal) o jantar está pronto! Ate the Rita the cake and (after-all) the dinner is ready! ‘Rita ate the cake but (after all) the dinner is ready!’

Rita Valadas Pereira

157

(8) a. Expliquei-lhe eu tantas vezes o caminho e (ainda assim) ela perdeu-se! Explained-her I so-many times the way and (even so) she got-lost! ‘I explained the way to her so many times but, even so, she got lost!’ (She was not supposed to get lost, as I explained the way to her so many times.) b. Disse-me a Joana o nome da rua onde mora e eu esqueci-me! Told-me the Joana the name of-the street where [she] lives and I forgot [it]! ‘Joana told me the name of the street where she lives, but I still forgot it!’ (I should not have forgotten the street name, as Joana told it to me so many times.) c. Almoço eu todos os dias no mesmo restaurante e hoje está fechado! Lunch I every-day at the-same restaurant and today [it] is closed! ‘I lunch every day at the same restaurant but today it is closed!’ (It should not be closed, as I go there every day.) d. Pediu o rapaz um café e o empregado trouxe-lhe um sumo de laranja! Asked-for the boy a coffee and the waiter brought-him an orangejuice! ‘The boy asked for a coffee, but the waiter brought him an orange juice!’ (The waiter should have brought the coffee, since that is what the boy asked for.) (9) a. Gosto eu de morangos e tu estás a comê-los todos! Like I of strawberries and you are eating-them all! ‘I like strawberries, but you are eating them all alone!’ (That is not right. You should offer me a strawberry, since I like them so much.) b. Gosto eu de morangos e tu não me dás nenhum! Like I strawberries and you not me give any! ‘I like strawberries, but you are not offering me any!’ (10) a. Queria eu falar contigo e tu foste-te embora!/? Wanted I to-talk with-you and you left!/? ‘I wanted to talk to you, but you left!/?’ (That is not right. You should have stayed to talk to me.)

158

Subject-Verb Inversion in Declarative-Exclamative Sentences

b. Fui eu até Lisboa para tu não estares lá!/? Went I to Lisbon for you not to-be there!/? ‘I went to Lisbon, but you were not there!/?’ (You should have been in Lisbon, since I went there.) c. Comprei-te eu a camisola amarela e tu agora preferes uma azul!/? Bought-you I the shirt yellow and you now prefer a blue [one]!/? ‘I bought you the yellow shirt but now you prefer a blue one!/?’ (I got annoyed, because you should prefer the shirt I bought you, and not another one.) (11) a. Está um belo dia de sol e estou eu aqui na biblioteca a estudar! [It]-is a beautiful day of sun and am I here in-the library studying! ‘It is a beautiful sunny day, and yet here I am in the library studying!’ (I feel sad being in the library, because I would prefer to be somewhere else enjoying the beautiful day.) b. Estou eu em casa enquanto os outros estão na praia! Am I at home while the others are at-the beach! ‘I am at home while the others are at the beach!’ (I feel sad because I would prefer to be at the beach with the others, not at home.) c. Convidou-me o Diogo para irmos passear e eu hoje tenho que trabalhar! Asked-me the Diogo to go for-a-walk and I today have to work! ‘Diogo asked me to go for a walk with him, but today I have to work!’ (I regret it, since I would prefer to go for a walk with Diogo, but I cannot because of work.) d. Comprei eu um carro velho quando o carro mais novo estava em promoção! Bought I a car old when the car newer was on sale! ‘I bought an old car, but the newer was on sale!’ (I should have bought the newer one; I regret buying the old one.) (12) a. Estudei eu tanto para só sair metade da matéria no teste! Studied I so-much for only to-get-out half of-the topics in-the exam! ‘I studied so hard, but only half of the topics were in the exam!’

Rita Valadas Pereira

159

b. Fui eu a Paris para o Museu do Louvre estar fechado! Went I to Paris for the Museum of Louvre to-be closed! ‘I went to Paris, but the Louvre Museum was closed!’ (I feel sad because I wanted the museum to be open so that I could visit it.) .

c. Deu-lhe a mãe um casaco novo para ele o perder no dia seguinte! Gave-him the mother a jacket new for him it to-lose on-the day next! ‘His mother gave him a new jacket, but he lost it the next day!’ (He should not have lost a new jacket!) (13) a. Vais tu passear e eu estudar! Will you go-for-a-walk and I [will go] to-study! ‘You are going for a walk, but I have to go study!’ (I would prefer to go for a walk with you instead of having to study.) b. Comeste tu um belo jantar e eu tive que comer fast food! Ate you a great dinner and I had to eat fast food! ‘You ate a great dinner, but I had to eat fast food!’ (That is unpleasant, because I would prefer to eat a great dinner instead of fast food.) As we can observe, in these examples, the aforementioned inversion is present, but there is no explicit or specific syntactic factor (such as a verb tense or mood, for example) to define it. The SVI seems to justify itself in the resultant pragmatic and semantic properties: the codification of the meaning of these sentences resorts to their illocutionary value – because they have a somewhat assertive-declarative character –, and it also alludes to the speaker’s cognitive state when s/he produces the sentence. These sentences are not only informational, but they also represent the speaker’s attitude. Thus, they are neither simple declarations or assertions, nor just comments or presentations of a speaker’s attitude; their semantics adjoins the concepts of Modality, or more specifically of Modalization, and of Assertiveness. The SVI allows them to be descriptions of the world and also comments on the facts described. The partially modal character of sentences with this kind of SVI is the most relevant aspect in their interpretation. Their pragmatic features justify the recovering of Palmer’s theoretical frame (1986, 16), where modality is taken as “the grammaticalization of the speaker’s (subjective) attitudes and opinions,” because these sentences are in fact several types of comments: a lamentation, a complaint, an opinion, or even judgment values, depending

160

Subject-Verb Inversion in Declarative-Exclamative Sentences

on the sentence or context that they enfold. This means that the SVI process we are studying apparently attributes some modal properties to the sentences it forms, since it is this specific subject-verb order inversion that grammaticalizes the speaker’s attitude when facing a state of affairs. It could, therefore, be considered that there is some kind of implicit modality11 in these sentences, since the speaker’s attitude is not produced objectively, grammatically, and lexically (as would happen if there was a verb or a modal adverb). Their interpretation is clearly dependent on the pragmatic context in which they are produced as well as on the listener’s ability to understand the speaker’s cognitive state (which exists, although is not expressed). However, this might not be such a pure covert modality as Bhatt (1999, 15) presents it, because what happens here is a discursive modalization process that is perfectly perceivable. The SVI is not a typically modal construction such as a modal adverb, but it is a syntactic strategy that represents the speaker’s evaluation of the proposition; in other words, it lexicalizes the fact that the sentence is not a declarative or informational assertion, but is in particular a comment on a certain situation. This discursive modalization process has an implicative reading, since there is an implicit comment which clearly involves modality, and which could be evaluated according to its evaluative, deontic, or epistemic features. This type of modality of the implicit comment defines and depends on the intention the speaker grammaticalizes through it. In contexts where sentences are criticisms or value judgments (cf. (7), (8), (9), (10)) the modality will seem deontic, since the criticism seems to presuppose an evaluation of the subject concerning the “correct/incorrect” parameter; but in cases where they constitute a lamentation, a complaint or an opinion (cf. (11), (12), (13)), the sentences are closer to what Rescher (1968) called “evaluative modality”, because their interpretation is easily determined by the parameter “being good that p/ being bad that p”. Since in both cases the evaluation is extremely subjective, the concept of deonticity followed here is based on Palmer (1986) and Kratzer (1991), as a rigorous consideration of specific moral or behavior rules, but it is applied in the context where the speaker defines “should not” as non-compliance with specific moral or attitudinal rules inside his/her own set of moral values, and inside the set of accessible worlds in the conversation. However, it is hard to assertively attribute a type of modality to these sentences, as their intention can shift according to the pragmatic context in 11

Here the concept of “covert modality” has been adopted, as proposed by Bhatt (1999, 15): “Covert Modality is modality which we interpret but which is not associated with any lexical item in the structure that we are interpreting.”

Rita Valadas Pereira

161

which they occur. This means that the same sentence could be, in one context, a lamentation (closer to evaluative modality), and, in another context, a criticism (more deontic). Let us analyze the sentence (10c), once again: (10) c. Comprei-te eu a camisola amarela e tu agora preferes uma azul!/? Bought-you I the shirt yellow and you now prefer a blue [one]!/? ‘I bought you the yellow shirt but now you prefer a blue one!/?’ The implicit comment of this sentence is pragmatically ambiguous, since it can have an evaluative character, as a kind of a complaint which is paraphrasable by “é desagradável que p”[it is unpleasant that p], as well as it can have a more deontic and critical aspect, taking into consideration the behavioural rules of the world considered by the speaker to be disrespected, and so is paraphrasable as “não deves p” [you should not p]. Therefore, if p is the proposition preferir uma camisola azul [to prefer a blue shirt], then the implicit comment of this sentence can be paraphrasable either by É desagradável que prefiras uma camisola azul quando eu te comprei a camisola amarela [‘It is unpleasant that you prefer the blue shirt when I bought you a yellow one’] or by Não deves preferir uma camisola azul quando eu te dei a camisola amarela [‘You should not prefer the blue shirt since I offered you a yellow one’]. Obviously, some of these examples are pragmatically more prone to a certain type of interpretation (or modality). This is the case of examples (11a) or (11b) that would not make much sense as deontic criticisms or judgment values. It would be strange if someone criticized oneself (thinking about not following some rules) by being in the library studying on a beautiful sunny day, or by staying at home while the others were at the beach. Therefore, the modality of the implicit comments in the analyzed sentences can either be evaluative or deontic, depending on the linguistic and pragmatic contexts in which they are uttered. The concepts of implicature (or implicative reading), inference, or presupposition, are gathered here with the concept of covert modality, because the interpretation of these sentences, with no more than a change in the syntactic order, will depend on conversational interaction, in order for the listener to understand (or presuppose) what type of attitude the speaker has towards the proposition. This will enable an understanding of what is semantically and pragmatically implied by the syntactic order switch to VSO. All the pragmatic elements that contextualize the conversation

162

Subject-Verb Inversion in Declarative-Exclamative Sentences

contribute to that happening, such as common knowledge, the listener’s interpretative ability, and the modal basis of the uttered sentence12. At this point, it seems important to remember the concept of “ordering source”, also presented by Kratzer (1991). This relates to one of the most important features of this type of SVI structures, namely its contrastive character. In fact, these constructions always involve an element (or a proposition) that contrasts with another (proposition). In the sentence Comi eu o bolo e afinal o jantar estava pronto (7a), the true critique (or lamentation) the speaker makes is expressed in the proposition comi eu o bolo [ate I the cake] – in which there is an inversion –, and that attitude exists because of the other proposition – e afinal o jantar estava pronto [and after all the dinner was ready] –, with which it actually contrasts. This idea of contrast between two propositions demands firstly that both participants in the conversation – speaker and listener – assume both propositions as realities that can be contrasted according to one or several values and under diverse conditions. As such, the concept of “ordering source” presented by Kratzer (as a gradual consideration of the possible accessible worlds in a conversation, in relation to approaching an ideal13) shows that this contrast is mutually interpreted (giving inversion a meaning) because both participants in the discussion agree (directly or indirectly) with a certain ideal world, according to the comment made. In sentences that express a lamentation, such as the ones in (11) and (12), inversion is the result of an evaluation, generally critical and negative, of the situation presented by the proposition – to be at the library studying (11a), to stay at home (11b), to have to work (11c), to buy an old car (11d), the test being about only half of the studied matter (12a), the Louvre Museum being closed when the subject went to Paris (12b), or having lost the new jacket his mother gave him (12c). That situation is compared to an implicit situation that would be ideal – taking a stroll or enjoying a sunny day (11a), being at the beach with friends (11b), taking a walk with Diogo (11c), buying a new car (11d), the test being about all the studied topics (12a), the Louvre being open (12b), and not losing the new jacket (12c). For that to happen, the explicit circumstances are taken in consideration – a beautiful sunny day, friends at the beach, a new car on sale, the test with only half the studied topics, the closed Louvre, and the 12

According to Kratzer’s (1991) concept, “modal basis” is here understood as the group of propositions that can constitute the epistemic or evidential informational basis of a certain evaluation, of a certain comment. 13 “For each world, the second conversational background induces an ordering on the set of worlds accessible from that world. It functions as the ordering source” (Kratzer 1991, 644 – highlighted by the author)

Rita Valadas Pereira

163

brand new jacket offered by the mother. To interpret correctly the intention behind the inversion, we then need the ideal world to be common knowledge, and both the speaker and the listener to consider the described situations as ideal, or, at least, for the listener to know how the speaker considers them to be. Only in this way can there be a common evaluation, and the inversion might really be able to imply an attitude of disapproval as well as a comment with the meaning of an adverb like infelizmente [unfortunately]. The same happens in other examples. The sentence (9b)14, for example, will only be interpreted correctly if both the discursive participants consider an ideal world as that in which the person eating strawberries offers one to someone who is watching him/her and who really likes them, with it being unpleasant or even incorrect if that does not occur. In sentences (13)15, the contrastive character is even stronger since the speaker regrets the situation precisely because it contrasts with another: s/he firstly presents and emphasizes (through SVI) the situation that is the positive pole of contrast, so that s/he can then present the proposition for which s/he is truly sorry. What can be noticed in these contexts is the relevance of shared social conventions in the ordering source taken into consideration; the negative idea towards fast food is a good example of a conventional and social idea that, even if the listener does not agree with, must be recognized by him/her. Only if the listener recognizes it can we know which idea is shared by the speaker in the utterance of (13b). And even if it is not acknowledged a priori that the speaker has a negative idea about eating fast food, the listener will have that information the moment the sentence is uttered with SVI. This ends up showing the speaker’s opinion on eating fast food but also about fast food itself. Furthermore, this latter idea has to be shared as a social convention, and the listener has to acknowledge it too, so that s/he can understand the attitude behind the subject-verb inversion. As I said before, the semantics of these clauses with subject-verb inversion is not confined to its modalizing feature. They are not simple expressive speech acts – mere lamentations, mere criticisms, etc. Actually, if we follow Panzeri’s (2003) definition of assertion as a proposition that adds new information to the speech, increasing the common knowledge between participants in a conversation, it can be assumed that this type of

14 15

(9b) Gosto eu de morangos e tu não me dás nenhum! (13) a. Vais tu passear e eu estudar! b. Comeste tu um belo jantar e eu tive que comer fast food! c. Teve a Maria 20 no teste e eu tive apenas 10!

164

Subject-Verb Inversion in Declarative-Exclamative Sentences

construction also has a stronger assertive character, as it too presents the facts that are commented upon. The already debated presence of a contrastive clause is the most revealing factor of this assertiveness; it presents a situation that ends up being the modal basis, or the reason, of the discursive modalization that characterizes the clause containing the inversion. This means that the sentences from (7) to (13) are not only expressions of a critical or remorseful attitude by the speaker, but they are also ways of giving new information: the speaker uses them not only to inform the listener about what happened and what originated this attitude, but also as a comment on it. S/he presents a true fact of the world, opposing it to another that is generally true in his/her personal experience and that is the target of the evaluation (which is mostly negative). This is the reason why the inverted sentence would be strange by itself, without a proposition (implicit or explicit) with which it would contrast: it seems somehow incomplete just to say Comi eu o bolo! [Ate I the cake!]. We feel the need for something more (either contextually or linguistically) to understand the sentence’s meaning. Given this, it seems appropriate to state that we are dealing with sentences that are simultaneously representative speech acts (in Searle’s (1969) terminology), i.e., assertions or assertive statements, and expressive speech acts, that is, comments (expressions of a speaker’s attitudes) that modalize the speech. Their meaning gathers assertiveness (they introduce new information in speech and they are even analyzable through its truth value) and modality or discursive modalization (whence we try to understand if it is a case of epistemic, deontic, or some other kind of modality). The proof that this feature is a result of inversion rests on the fact that the same sentences, those of examples (7) to (13), would not be interpreted as comments, and would only be interpreted as assertions or declarations, if they did not have the inversion commented on here (and, of course, the necessary intonation). Some examples of that can be observed in (14): (14) a. Eu comi o bolo e/mas afinal o jantar está pronto. ‘I ate the cake and/but after all dinner is ready.’ b. Eu expliquei-lhe tantas vezes o caminho e/mas (ainda assim) ela perdeu-se. ‘I explained the way to her so many times and/but (nevertheless) she still got lost.’

Rita Valadas Pereira

165

c. A Joana disse-me o nome da rua onde mora e/mas eu esquecime. ‘Joana told me the name of the street where she lives and/but I forgot it.’ d. Eu almoço todos os dias no mesmo restaurante e/mas hoje está fechado. ‘I lunch every day at the same restaurant and/but today it is closed.’ e. O rapaz pediu um café e/mas o empregado trouxe-lhe um sumo de laranja. ‘The boy asked for a coffee and/but the waiter brought him an orange juice.’ f. Eu gosto de morangos e/mas tu não me dás nenhum. ‘I like strawberries and/but you are not offering me any.’ h. Está um belo dia de sol e/mas eu estou aqui na biblioteca a estudar. ‘It is a beautiful sunny day and/but here I am in the library studying.’ i. Eu estou em casa enquanto os outros estão na praia. ‘I am at home while the others are at the beach.’ j. O Diogo convidou-me para irmos passear e/mas eu hoje tenho que trabalhar. Diogo asked me to go for a walk with him and/but today I have to work. k. Eu estudei tanto mas só saíu metade da matéria no teste. ‘I studied so hard, but only half of the topics were appeared in the exam!’ l. Tu comeste um belo jantar e/mas eu tive que comer fast food. ‘You ate a great dinner and/but I had to eat fast food.’ m. Eu queria falar contigo e/mas tu foste-te embora. ‘I wanted to talk with you and/but you left.’

166

Subject-Verb Inversion in Declarative-Exclamative Sentences

Here, the use of the dot instead of the exclamation mark used in examples (7) to (13) shows that the exclamation mark in SVI sentences represents the way they are prosodically and pragmatically produced – expressing the speaker’s attitude. On the other hand, when there is no inversion, sentences are interpreted as statements, used in a context in which someone would just narrate a situation, for example. The possibility of the use of mas [but] can be noticed in a sentence with no inversion (cf. 14), in opposition to the ungrammaticality or peculiarity of the use of an adversative connector in a sentence with SVI (cf. 15): (15) a. ?/* Comi eu o bolo mas o jantar está pronto! Ate I the cake but the dinner is ready! b. ?/* Expliquei-lhe eu tantas vezes o caminho mas ela perdeu-se! Explained her I so many times the way but she got lost! c. ?/* Pediu o rapaz um café mas o empregado trouxe-lhe um sumo de laranja! Asked the boy for a coffee but the waiter brought him an orange juice! d. ?/* Gosto eu de morangos mas tu estás a comê-los todos! Like I strawberries but you are eating them all! e. ?/* Estou eu em casa mas os outros estão na praia! Am I at home but the others are at the beach! f. ?/* Fui eu a Paris mas o Museu do Louvre estava fechado! Went I to Paris but the Louvre Museum was closed! g. ?/* Vais tu passear mas eu estudar! Will you go for a walk but I will study! h. ?/* Queria eu falar contigo mas tu foste-te embora!/? Wanted I to talk with you but you left!/? These sentences are less familiar, since their acceptability is more restricted. The strangeness of there being a connector in a sentence with inversion proves that the SVI process is itself the linguistic strategy that specifically establishes the contrastive, and yet factive, feature of this

Rita Valadas Pereira

167

sentence’s semantics. This process provides an adversative value to the sentence it forms, competing with the presence of a lexical operator with the same value, to give rise to a repetition or redundancy that would sound unfamiliar to the speaker. In this way, the “e” [and] that generally links the two clauses is an indicator of its assertiveness: it seems to be both a copulative and adversative conjunction, because it conjugates the situation of two propositions (as stated before, these propositions have a truth value and describe states or affairs of the world), and it also may be used to contrast those same situations. After all, this contrast is the essence of the speaker’s propositional attitude in a sentence with this type of inversion. The conjugation of copulative and adversative features that characterize it is thus further evidence that the semantics of sentences with this SVI enfolds assertiveness and discursive modalization. It can also be observed that the verb tenses and moods employed are those typically used in assertive contexts (the “reality moods”), and not those used for modal contexts such as the Subjunctive, the Conditional, or even the Future. The sentences from (16) are odd, or, at least, the contexts in which they are used or their usual pragmatic roles are not relevant for this paper: (16) a. ?? Coma eu o bolo e o jantar esteja pronto! Eat I the cake and the dinner is ready! b. ??Comeria eu o bolo e o jantar estaria pronto! Would-eat I the cake and the dinner would-be ready! c. ?? Comerei eu o bolo e o jantar estará pronto! Will-eat I the cake and the dinner will-be ready! The use of verb tenses or moods referring to possible worlds (worlds that do not necessarily have a referent in the “real world”) seems to be unacceptable because these sentences are actually comments on facts or situations from the real world (a world that the sentence itself presents us, as seen above), and not regarding any hypothesis or possibility; so much so that we need to display a temporal coherence (almost narrative) among the situations denoted. For instance, it would not make sense that the sentence (8a) was the other way around – ??/#Perdeu-se ela e eu expliqueilhe tantas vezes o caminho!16– because there is a clear temporal and causal 16

Got-lost she and I explained-her so-many times the way

Subject-Verb Inversion in Declarative-Exclamative Sentences

168

line between the two propositions which implies that the action of explicar tantas vezes o caminho [explaining the way so many times] is prior to ela perder-se [her getting lost]; otherwise, the critique uttered would be pragmatically odd. Furthermore, the properties of the subjects found in these post-verbal position contexts also point to the need for a specific reference that shows that these sentences really have a situational referent in the real world. Although it does not create total ungrammaticality, in can be seen in (17) that an undefined or non-specific subject creates semantic unfamiliarity, if all the assertive and modalizing properties already described here are considered. In fact, the intention of these sentences to comment, criticize, opine, or regret does not seem justifiable if the subject is not defined or mentioned. It obviously does not make much sense to comment or criticize an action done by an unknown entity, or without a specified referent in the situational-pragmatic context involving the speaker and the sentence with SVI17. (17) a. ?/# Comeu um homem o bolo e afinal o jantar está pronto! Ate a man the cake and after-all the dinner is ready! ?/# ‘A man ate the cake but the dinner is ready!’ b. ?/# Almoça uma senhora todos os dias no mesmo restaurante e hoje está fechado! Lunches a lady every-day at the-same restaurant and today [it] is closed! ?/# ‘A lady lunches every day at the same restaurant but today it is closed!’ c. ?/# Gosta uma rapariga de morangos, e tu estás a comê-los todos! Likes a girl strawberries and you are eating-them all! ?/# ‘A girl likes strawberries, but you are eating them all alone!’

??/#

‘She got lost, but I explained the way to her so many times!’ The cases of (10) may be exceptions to this rule, because it is easier to accept an undefined NP – e.g. Queria uma moça falar contigo e tu foste-te embora?!; Foi um rapaz até Lisboa para tu não estares lá?!. It seems to me that, here, the undefined NP used does not codify a generic value or an unknown entity; instead, it is a non revelation of the subject’s identity by the speaker, although s/he knows who the subject is. The reference is, therefore, not lexically revealed. 17

Rita Valadas Pereira

169

d. ?/# Está um belo dia de sol e está um jovem na biblioteca a estudar! [It] is a beautiful day of sun and is a young-man in-the library is studying! ?/# ‘It is a beautiful sunny day, but a young man is in the library studying!’ e. ?/# Estudou um aluno tanto para só sair metade da matéria no teste! Studied a student so-much for only appear half of-the syllabus in -the test! ?/# ‘A student studied so hard, but only half of the syllabus was in the test.’ The only contexts of this kind in which this undefined nominal phrase and subject may be accepted are those where the pronouns “we” or “I” have a generic interpretation (although this works only with expressions like “um homem” [a man] or “uma pessoa” [a person]18). In those cases, the reference would be specific instead of abstract or undefined. The use of the phrase “uma pessoa” with a generic value is very common, but it is not undefined (because it generally relates to the pronoun “eu” [I]) in a context with inversion like Tem uma pessoa tanto dinheiro para não ter onde o gastar!19. Furthermore, as the subject of these sentences with SVI is mostly pronominal and deictic, this increases and shows the need for a situational reference.

4. Exclamative or declarative Sentence Type? The semantic-pragmatic properties of sentences with SVI, studied in the previous section, show the issue concerning the type of sentence, a consequence of being both assertions and comments. The confluence between the concepts of assertiveness and modality or modalization makes their behavior somewhat hybrid, concerning their exclamatory or declarative character, because they have properties of both types of sentence. This is the reason why I call them “declarative-exclamative” sentences. 18

It should be remember that in Portuguese these expressions may express the generic values of “we” or “I”, depending on the linguistic context in which they are used. 19 Has a person so-much money for not having where it to-spend ‘Someone with so much money and nowhere to spend it on!’

170

Subject-Verb Inversion in Declarative-Exclamative Sentences

An exclamative sentence is traditionally defined as a sentence that corresponds to an expressive illocutionary act, according to Searle’s (1969) speech act typology, and Mateus et al. (2003, 74) describe it as the illocutionary act that aims to “express the psychological state of affairs of the speaker specified in the content of the utterance”. Therefore, the sentences studied in this work clearly have exclamatory properties, since the primary intention, i.e. the justification for its use, is the expression of the speaker’s attitude. What determines the subject and verb’s positions is precisely that modalizing feature of the utterance. However, some elements of the declarative clause’s definition may also coincide with these SVI sentences’ properties, namely their truth value. In a traditional theoretical framework, declarative sentences are considered as those whose interpretation pass through an evaluation of their truth value, knowing that, within a referential semantics perspective, their proposition is only true if their predication is certified in the real world, that is, if the state of affairs presented can be verified. Following all that was described in section 2 of this article, it can easily be perceived that a sentence such as Comi eu o bolo e afinal o jantar estava pronto (7a) has declarative features, as it can be analyzed through these same concepts of truth value, reference and assertiveness. It seems then possible to state that the sentences analyzed in this paper have properties typically associated with exclamative sentences; they are comments, expressive speech acts that demonstrate the speaker’s attitude; they depend on the context and on the ability of the listener to interpret the implicit attitude; and they are clearly produced with an exclamatory intonation. However, they also have declarative features: they describe the world; they introduce new information into the speech; and they have an available truth value, thus being closer to assertive speech acts. While their assertiveness constitutes their own declarative properties, their role and their pragmatic features establish their exclamatory characteristics. The definition of “declarative-exclamative” sentences is, then, closer to Andueza’s (2011) analysis of exclamative sentences, since the author considers that they are also assertive sentences because they express the speaker’s attitude at the same time as describing the world. Nevertheless, the declarative (assertive) properties of SVI sentences studied here justify that they are not exclusively considered as exclamatives: they are somehow closer to modalized declaratives. Indeed, their exclamatory tone and their dependence on conversational context bring them closer to pure exclamative sentences; however, the fact that they present these commented situations, and that they do not depend so much on a presupposition or on an inference by the listener (as happens with Que

Rita Valadas Pereira

171

lindo dia! [What a beautiful day!], which is a typical exclamative), justifies their more hybrid pragmatic status, and therefore the notion of “declarative-exclamative” sentences.

5. Conclusions In this paper, a special case of subject-verb inversion (SVI) from contemporary European Portuguese was analyzed, and was characterized as having a very specific semantic-pragmatic role in the sentences in which it occurs. The semantic and pragmatic analysis of sentences such as Comi eu o bolo e afinal o jantar estava pronto!20 or Gosto eu de morangos e tu não me dás nenhum!21, has led to the understanding that the choice of using VSO order instead of the canonic SVO order turns the sentence’s meaning into neither a simple predication, nor a simple grammaticalization of an attitude: actually, it is at the same time an assertion and a comment. Its semantics does not limit itself to an implicit modal comment, neither in an implicative reading, nor even in a description of the world: these are sentences in which discursive modalization is represented by an assertion, since the subject of the enunciation presents the facts, and, at the same time, comments on them. The construction of this modal assertiveness or assertive modalization involves meaning systems such as propositionality, reference (which is often deictic, and thus solved by the situational context), and time. But it mainly involves semantic-pragmatic systems of interpretation such as inference, presupposition, and implicature; these resort to semantic elements such as modal basis or ordering source, or to pragmatic elements such as common knowledge, factual circumstances or evidence, social conventions, or just the interpretative competence and the mutual knowledge of the participants in a conversation. All the pragmatic elements that surround the production of sentences with SVI allow us to interpret the assertive and modal usage of this syntactic strategy, and to assume that these sentences are both expressive and assertive speech acts. In order to understand what is mentioned about a proposition p, the listener must evaluate its predication and its pragmatic features – if only one of them is chosen, it will definitely lose one of the values that SVI gives to the sentence it integrates.

20 21

Cf. (7a) for translation. Cf. (9b) for translation.

172

Subject-Verb Inversion in Declarative-Exclamative Sentences

To conclude a study that aims mostly to understand the interaction between syntax and semantics in qualifying this process of sentence construction in Portuguese, I will also present sentences such as (18) and (19). (18) a. E tenho eu ainda tanto trabalho para fazer! And have I still so-much work to do! ‘I still have so much work to do!’ (when I wanted to go on vacations so badly!) b. E fui eu ver aquele mau filme! And went I to-watch that bad movie! ‘And I went to watch that bad movie!’ (when it would have been so much better to watch that Oscar winning movie!) c. E é o Pedro tão inteligente e aplicado! And is the Pedro so intelligent and methodical! ‘Pedro is so intelligent and studious!’ (and he had that awfully low grade!) (19) a. Quer ela ser escritora! Wants she to-be [a] writer! ‘And still she wants to be a writer!’ b. (E) vai o João viajar! (And) will the João travel! ‘And still João is travelling!’ c. E diz a Maria que é rica! And says the Maria that [she] is rich! ‘And, in spite of it, Maria still says that she is rich!’ d. E pensava eu que ia chegar a horas a casa! And thought I that [I] would get on time home! ‘I even thought that I would get home on time! These sentences are interesting because they present the inversion studied in this paper, and they seem to have some modalizing and exclamatory intention; but there is also an ellipsis of the contrastive proposition (the one that would represent the modal basis). That proposition, despite not being explicit, is still the cause of this inversion,

Rita Valadas Pereira

173

and therefore the element which captures the essence of the attitude expressed in the clause. In these cases, it is substituted by an initial “e” [and], which is a discursive operator that allows its recovery, so that the attitude expressed in the sentences (and consequently their real meaning) can only be interpreted contextually. This means that, as these sentences do not present all the modal basis, their intention can only be understood (and fully interpreted) if the listener has total access to the situational context in which they are uttered. The work of interpretation goes from analyzing these circumstances to inferring which contrastive proposition is not expressed, and what type of comment is implicit there. Thus, it is not only the comment that is implicit here, as happens with the examples (7) to (13) analyzed throughout this work, since it is also a part of the situational context that forms the basis of this comment. A listener can only understand it if s/he has access, live, to contexts like (20) and (21), where the inversion creates a contrast between what is said and the circumstances surrounding the production of the utterance. (20) (SITUATION: The speaker is chatting to a friend and comments that it is July already, and all his/her friends are on vacation, which gives him the desire to be on vacation too and to go to the beach with his friends): a. E tenho eu ainda tanto trabalho para fazer! 22 (SITUATION: The speaker has just left the movie theatre, where s/he watched a movie that s/he did not like, and his/her friend tells him/her that that evening another movie theatre in town was offering tickets to a session of an Oscar winning movie): b. E fui eu ver aquele mau filme!23 (SITUATION: The speaker is telling someone that he just learnt that Pedro, his/her close friend and acquaintance of the listener, had a bad grade in a work that he had put a lot of effort into): c. E é o Pedro (tão) inteligente e aplicado!24

22

See (18a) for translation. See (18b) for translation. 24 See (18c) for translation. 23

174

Subject-Verb Inversion in Declarative-Exclamative Sentences

(21) (SITUATION: The speaker finishes reading some texts written by his/her friend and s/he verifies that they had a lot of ungrammatical constructions and that they are not cohesive and coherent in the development of the ideas): a. Quer ela ser escritora!25 (SITUATION: The speaker ends up knowing that João, his/her friend and also acquaintance of the listener, is afraid of flying but has still booked a trip to China.) b. (E) vai o João viajar!26 (SITUATION: The speaker has just learned, because the listener told him/her, that Maria, his/her acquaintance, had to ask the bank for a loan to pay her college fees): c. E diz a Maria que é rica!27 (SITUATION: Coming from a party, the speaker, who had promised his/her parents that s/he would be home at 11p.m., looks at his/her watch and notices that it is already 11.55p.m., and s/he remembers that s/he was late due to an interruption in the subway traffic): d. E pensava eu que ia chegar a horas a casa!28 These last cases exemplify and reinforce the conclusion that the subject-verb inversion studied in this paper is more than a syntactic process: it is also a semantic-pragmatic process that specifically marks the speaker’s attitude.

References Âmbar, Maria Manuela. 1992. Para uma sintaxe da inversão sujeito-verbo em português. Lisbon: Colibri. Andueza, Patricia. 2011. “Rhetorical Exclamatives in Spanish.” PhD diss., Graduate School of the Ohio State University.

25

See (19a) for translation. See (19b) for translation. 27 See (19c) for translation. 28 See (19d) for translation. 26

Rita Valadas Pereira

175

Bhatt, Rajesh. 1999. “Covert Modality in Non-Finite Contexts”, PhD diss., University of Pennsylvania. Brito, Ana Maria, and Inês Duarte. 1980. “Condições sobre posposição do sujeito em português”. Boletim de Filologia, tome XXVII. Lisbon. Costa, João, and Ana Maria Martins. 2010. “O Foco do desentendimento”. Manuscrito de comunicação no XXI Encontro Nacional da Associação Portuguesa de Linguística. Porto. Costa, João. 2004. Subject Positions and Interfaces: The Case of European Portuguese. Berlin, New York: Mouton de Gruyter. Elliot, Dale E. 1974. “Toward a grammar of exclamatives”. Foundations of Language 11:231-246. Gutiérrez-Rexach, Javier. 1996. “The Semantics of Exclamatives”. In UCLA Working Papers in Linguistics 1996, edited by Edward Garrett, and Felicia Lee, 146-162. Department of Linguistics, University of California Los Angeles. Hengeveld, Kees. 2004. “Illocution, Mood and Modality.” In Geert Booij., Christian Lehmann, and Joachim Mugdan, Morphology: An International Handbook of Inflection and Word Formation, vol. 2. Berlin: Walter de Gruyter. Kratzer, Angelika. 1991. “Modality”. In Semantics, edited by A. von Stechow e D. Wunderlich. Berlin: Walter de Gruyter. Martins, Ana Maria. 2010. “Relatório do seminário de linguística comparada: Tópicos de gramática do Português numa perspetiva comparativa, apresentado a provas públicas para obtenção do título académico de agregado no ramo de Linguística (Linguística Geral)”. Lisbon University. Mateus, Maria Helena Mira, et. al. 2003. Gramática da Língua Portuguesa. Lisbon: Caminho. Palmer, Frank R. 1986. Mood and Modality. Cambridge: Cambridge University Press. Panzeri, Francesco. 2003. “In the (indicative or subjunctive) mood”. In M. Weisgerber, Proceedings of sinn und bedeutung 7. University of Konstanz. Rescher, N. 1968. Topics in Philosophical Logic. Dordrecht: D. Reidel Pub. Comp. Searle, John R. 1969. Speech acts: an essay in the philosophy of language. Cambridge: University Press. Von Wright, Georg Henrik. 1951. An essay in modal logic. Amsterdam: North Holland.

TERMINOLOGY ISSUES IN SPANISH AND PORTUGUESE PHONETICS AND PHONOLOGY: VIBRANTES OR RHOTICS? IIRIS RENNICKE UNIVERSITY OF HELSINKI CENTRO DE LINGUÍSTICA DA UNIVERSIDADE DO PORTO

Abstract The term vibrante has been used in Spanish and Portuguese linguistics for both phonetic (description of the articulatory mechanism) and phonological (the grouping of ‘r-sounds’) reference. The terms vibrante múltiple/múltipla and vibrante simple(s) are the general terms used in reference to a trill and a tap, respectively. In this paper I will argue that a trill and a tap differ in the duration of their closure phase(s) and articulatory mechanism and therefore cannot be grouped together under the same manner of articulation, vibrante. In addition, the term vibración/vibração is considered unsuitable for the tap since by definition vibration is more likely to comprise several oscillations. I will also propose that due to the vast allophony of the two rhotic phonemes in Spanish and Portuguese, a term covering several forms of articulation is needed: conso(n)antes róticas is suggested for this purpose. Keywords: Terminology, Phonetics, Phonology, Vibrante, Rhotics.

1. Introduction The aim of this paper is to discuss the phonetic and phonological terminology pertaining to the rhotics, or r-sounds of Spanish and Portuguese. Traditionally, the historical phonemes in the two languages, the alveolar trill /r/ and the alveolar tap /‫ݐ‬/, have been generally dubbed vibrante múltiple and vibrante simple in Spanish and vibrante múltipla and vibrante simples in Portuguese. The term vibrante, having the literal

Iiris Rennicke

177

meaning of ‘vibrating’, has been used to describe the trill and the tap under the same manner of articulation, ‘vibration’, whereas in e.g. the English language version of the International Phonetic Alphabet (International Phonetic Association 2013) the two sounds are considered to have clearly different manners of articulation. The false similarity created through terminology between the two sounds has also been the foundation for the phonological term vibrantes, grouping the trill and tap in the same phonological class of sounds. However, as the gradient allophony of these phonemes moves the pronunciation away from full articulatory contact to fricatives and approximants, the term vibrantes becomes even less descriptive of these rhotic sounds. The article begins with a short introduction to the current terminology in Spanish and Portuguese linguistics and to rhotics as a class. Next, the acoustic and articulatory differences between the alveolar tap and trill will be examined, and the suitability of the term vibrante to denominate both of them will be assessed based on dictionary definitions. This is followed by an overview of rhotic allophony in Spanish and Portuguese and an analysis of its consequences for the definition of rhotics as a class in these languages and the phonological terminology pertaining to rhotics. The article concludes with terminology suggestions that are left open for debate.

1.1. Terminology referring to rhotics Among the European languages, the multiple – simple dichotomy seems to be exclusive to Iberian Romance linguistics. The Spanish phonetician Quilis (1999, 307-359), in his chapter on liquids, includes a subchapter on líquidas vibrantes, which includes the ‘simple’ and ‘multiple’. As an example of a Spanish American author, Obediente (2007, 130-133) groups the trill and the tap under the same manner of articulation – vibrantes. He does refer to the English IPA terminology concerning the trill, but then goes on to claim that r-type sounds with only one apical movement are usually included in the vibrantes category. No comment is made in relation to the distinction of the manner of articulation in the IPA. An exception from this tradition is Massone (1988) who translates the IPA terms trill and tap as vibrante and percusiva (‘percussive’), respectively, and states that her study of Buenos Aires Spanish supports the view of considering the trill and the tap to have different forms of articulation. Based on her acoustic analysis and perceptional tests,

178 Terminology Issues in Spanish and Portuguese Phonetics and Phonology

Massone states that Spanish distinguishes between a tap, trill and approximant. As for European Portuguese, the same multiple – simple terminology is used, for example, by Andrade and Viana (1996, 139) in an introductory linguistics textbook. According to the authors, in the articulation of a vibrante the active articulator can perform only one tap (batimento) or multiple taps. Although the terms vibrante múltipla and vibrante simples are still in use in Brazilian linguistics, the terms vibrante and tap/tepe, respectively, have come to substitute them in some publications. For example, Cristófaro-Silva (2007, 34) cites two different manners of articulation: vibrante (múltipla) and tepe (ou vibrante simples), and uses only vibrante and tepe in subsequent tables. The author thus mixes old and new terminology by maintaining vibrante but reserving it only for the trill, creating a distinction between the two manners of articulation. Martínez Celdrán and Fernández Planas (2007, 156-157) argue that the Spanish alveolar trill and tap differ in their manner of articulation, but that in spite of these articulatory differences the two sounds are phonetically very similar and, from this point of view, belong to the same sound class. However, the authors have decided to call the section pertaining to rsounds róticas (‘rhotics’) because, in addition to the traditional vibrantes, approximant and fricative variants are discussed. Here lies the central question of the present article: can both the trill and the tap be called vibrantes, and if so, can that name be used for the Spanish and Portuguese rhotics as a phonological class, taking into consideration their vast allophony?

1.2. Rhotics as a class Rhotics is the term most commonly used for r-sounds in English. Unlike most natural classes referred to in phonetic theory, the sounds covered by the term rhotic have no articulatory or auditory property in common. The IPA symbols for rhotics are different versions of the letter ‘r’, which represents the element that the members of this class have in common: they tend to be written with the letter ‘r’. There are trills, taps, fricatives, and approximants in this class, and both coronal and dorsal places of articulation. Dorsal rhotics are universally quite rare, and occur mostly in Western European languages. Therefore neither the manner nor place of articulation unifies the class of rhotics (Ladefoged and Maddieson, 1996, 217), which could well be considered an argument against the existence of such a class.

Iiris Rennicke

179

According to Ladefoged and Maddieson (1996, 216), some of the rhotics included in the IPA are the following: r Voiced dental or alveolar trill ‫ ݐ‬Voiced dental or alveolar tap or flap ‫ ݋‬Voiced dental or alveolar approximant ‫ ݏ‬Voiced post-alveolar flap ‫ ݍ‬Voiced post-alveolar approximant ‫ ݒ‬Voiced uvular trill ‫ ݓ‬Voiced uvular approximant ‫ ݌‬Voiced dental or alveolar lateral flap Phonologically, rhotics tend to behave in similar ways: they are often the only consonants allowed as second members of a cluster in the syllable onset, or as first members of clusters in coda position. Frequently they share this privileged position with lateral approximants and/or nasals. Rhotics are likely to have syllabic variants or to merge with contiguous vowels; postvocalic (R)’s1 tend to become vowels or disappear altogether. Also, vowels before (R)’s tend to strengthen or to be ‘coloured’ in their quality by the following (R). Rhotics of one type often alternate with other rhotics; diachronic change can be seen to affect the phonetic characteristics of rhotics without leading to their loss as a contrastive element, giving rise to wide crosslinguistic variation. In other words, this type of change is phonetically superficial and not phonologically meaningful. In addition, the allophony of (R) within a single language reinforces the correspondence of heterogeneous phonetic variants of the same phoneme (a single phoneme with superficial allophones). Additional proof of the existence of a rhotic class is that sociolinguistic conditioning of variation in (R) is common within a language, and developmental data often show that during language acquisition the realization of (R) varies greatly. (Ladefoged and Maddieson 1996, 216-217; Lindau 1985, 157; Scobbie 2006, 337) An argument against the idea of a rhotic class could be the difficulty of choosing the appropriate abstract feature for (R), and then explaining why the phonetic realization of this abstract (R) is both systematic (considering the tendencies within a specific language) and arbitrary (considering the variation within a language as well as crosslinguistic variation). Furthermore, in terms of mental representations, the problem is to explain 1

(R) is the symbol I use when referring to an abstract rhotic, following the title of “(R) as a variable” by Scobbie (2006) in order to avoid confusion with the Spanish and Portuguese alveolar trill phoneme, /r/. However, in the literature /r/ is commonly used in the abstract sense, e.g. by Ladefoged and Maddieson (1996).

180 Terminology Issues in Spanish and Portuguese Phonetics and Phonology

how an abstract category (R) is formed from articulatorily and acoustically diverse input. (Scobbie 2006, 338) A lowered third formant was suggested as a unifying property for rhotics, but this does not apply to uvular and dental trills, which have a relatively high third formant. Instead, Lindau (1985) argues for unity through family resemblance, or step-by-step connections between the members. In this way, each member resembles some other member with respect to some property, but it is not the same property that constitutes the resemblance for all members of the class. Ladefoged and Maddieson (1996, 245) state that these family resemblances explain the synchronic variation and diachronic change in rhotics. Lindau comes to the conclusion that a class of sounds may not always be defined in terms of a single phonetic correlate; instead, the motivation for forming a class must be sought in the phonological behaviour of the sounds. This seems coherent with Lass’ (1984, 82) definition of a natural class: “class that we would expect to find behaving as a unit”. Lindau argues that the relation between phonological and phonetic classes is more complex than the one-to-one relation that is generally assumed. Figure 1 reproduces the Wittgensteinian family resemblance pattern of rhotics proposed by Lindau (1985, 167).

Figure 1: Parameter relations among r-sounds (Lindau 1985, 167) The parameters relating rhotics are numbered as follows: 1 = pulse pattern (trill); 2 = closure duration; 3 = presence of formants (sonorant); 4= presence of noise; 5 = distribution of spectral energy (place of articulation).

Iiris Rennicke

181

The relations in Figure 1 are explained by Ladefoged and Maddieson (1996, 244-245): acoustically, a trill seems like a series of taps, and the tap is often an allophone of the trill. Frication and trilling may co-occur, leading to fricative variants, or the open phase of a trill can be prolonged into an approximant. The similar pulsing patterns in apical and uvular trills could explain the diachronic change in some European languages. Once established as an uvular, the trill often weakens, creating free variation between uvular trills, fricatives, and approximants. Scobbie (2006, 338) also defends rhotics as a class. He states that the label (R) seems to be applied to oral lingual sonorant consonants unless they are specifically palatal, lateral, or labial. Therefore, he sees (R) as somewhat of an ‘elsewhere’ category, and that (R) could be considered a rather meaningless label without the phonological, sociolinguistic, diachronic, and acquisitional justification for the establishment of the category. In Scobbie’s opinion, from the point of view of discussing the phenomenon of rhoticity, “/r/ or (R) are labels as good as anything else, so long as we recognize the descriptive and theoretical limitations arising from the use of this convenient international phonetic alphabet or orthographic symbol”. In addition to the view of rhotics being a class not defined by a unifying feature, but by phonological behaviour and family resemblance, one further important definition could be suggested. This class is always defined by a specific language variety: a sound that is rhotic in one language might not be rhotic in another language. This is especially true of back fricatives such as [x Ȥ ‫ ݓ‬h], which are not prototypical rhotics made with the tip or blade of the tongue (Ladefoged and Maddieson 1996, 215). A case in point is the difference in Spanish and Portuguese, since the former in general treats these sounds as allophones of /x/, whereas Portuguese treats them as rhotics due to the retraction process and consequent weakening of the original alveolar trill (r ĺ ‫ ݒ‬ĺ ‫ ~ ݓ‬Ȥ ~ x ~ h). However, in Puerto Rican Spanish an interesting phenomenon emerged most probably in the 19th century: the /r/ phoneme began to have a uvular fricative or trill allophone (Lipski 2007, 355-356). This could potentially mean that the opposition between /x/ and /r/ is neutralized in minimal pairs such as jamón – Ramón and cojo – corro. In light of the arguments presented so far, rhotics seem to be more of a language-specific class unified by allophony and phonological behaviour, as opposed to something that can be defined in terms of features. The name ‘rhotics’ itself is quite arbitrary and points only to the fact that the sounds included in this class tend to be written with a particular character in orthographic systems derived from the Greco-Roman tradition – the

182 Terminology Issues in Spanish and Portuguese Phonetics and Phonology

letter ‘r’ or its Greek equivalent rho (Ladefoged and Maddieson 1996, 215). Nevertheless, rhotics can be considered a valid name for the class of r-sounds for the time being, since no unifying phonetic property can be determined. In addition to describing r-sounds on a more general level, it is a useful concept in the description of the r-sounds of a particular language since r-sounds more often than not are prone to have allophones of different manners and places of articulation.

2. Vibrante in phonetics This section concerns the phonetic differences and similarities between the tap and the trill and whether they permit the grouping of the two sounds under the same manner of articulation. The duration of the closure phases, the articulatory mechanism and the definition of vibration will be discussed.

2.1. Duration of closure phases It was stated above that from an acoustic point of view, a trill resembles a series of taps. This is also the underlying claim behind grouping the tap and the trill under the same concept of vibrante in the languages under examination. However, scholars do not seem to agree on whether the duration of the closure phases of these two sounds is actually the same. Some views on the matter will be presented in this section. Blecua Falgueras’ (2001) detailed study on Peninsular Spanish rhotics reveals that, in coda position (section 3.2.3.1.), the duration of the observed variants was not directly dependent on the number of components2 of which they consisted: a two-component rhotic was 50% (and not 100%) longer than a one-component rhotic, and a threecomponent rhotic was 35% (and not 50%) longer than a two-component rhotic. The author concludes that the duration of variants is not deductible from the number of components because the duration of the components decreases as the number of components increases. In this way, she states, the duration of a one-component rhotic is not the same as that of a closure phase in a two-component rhotic because the former would then be too short. Catford (1977, 130) also argues, from a slightly different perspective, that the duration of the closure phases of trills and taps is not the same. He 2

Blecua Falgueras uses ‘component’ to denote any phase of a rhotic in the spectrogram image, be it the closure phase or the open phase.

Iiris Rennicke

183

argues that the frequency of alveolar and uvular trills is of the order of 30 cycles per second, and that this is much higher than the maximum rate at which one can produce a series of taps (5-6 per second) due to the difference in manner of articulation. He concludes that any idea that a trill is a rapid series of taps, or that a tap is just an ultra-short trill is “quite wrong”. Martínez Celdrán and Fernández Planas (2007, 157), based on the results in Martínez Celdrán and Rallo (1995, 190), question these measurements by stating that the closure phase of a tap has an average duration of 20.8 ms, and that of a trill, 15.78 ms. This would mean that, by simply calculating how many of these ‘fit’ into one second, the Spanish tap would have 20 cycles per second, and the trill 25 cycles per second. Based on these calculations, the authors claim that the 5 cps difference is not significant, and that the duration of closure phases is essentially the same. However, this comparison seems questionable because it is obvious that Catford’s measurement of 5-6 taps per second was calculated in terms of actual serial articulation (“the maximum rate at which one can produce a series of [‫]ݐ‬-flaps”), and not by multiplying the tap’s average duration in milliseconds. In addition, Martínez Celdrán and Rallo’s report of a 5 ms difference in closure phase duration is in itself a proof that the durations are indeed slightly different. Lindau (1985, 166) calculated the duration of an apical tap and that of a closure phase in an apical trill, based on speakers of Swedish, Spanish, Hausa, Degema, Edo, and Kalabari. She states that the average for both is approximately 20 ms, and that ‘a trill can be regarded as a series of taps’. However, in light of the other results presented here, it is likely that taps indeed have a longer duration than the closure phase of a trill, and therefore the two sounds have a slight acoustical difference.

2.2. Articulatory mechanism Another aspect essential to analysing the inclusion of the tap and the trill under the term vibrante is whether or not they share a common manner of articulation. As has been stated above, the position of the International Phonetic Association is that the two sounds differ in this aspect. According to Catford (1977, 130), a tap is produced with a single ballistic and momentary gesture, whereas the trill is a maintained, prolongable posture. Barry (1997, 35-36, 40-41) states that the literature is in good agreement about the physical basis for trilling, namely that a sufficient airstream is required to make a flexible articulator vibrate when

184 Terminology Issues in Spanish and Portuguese Phonetics and Phonology

it is in close proximity to another surface of the vocal tract. Trilling is similar to vocal-fold vibrations in the sense that it is the product of aerodynamic-myoelastic processes. The tap, on the other hand, has been described as a “ballistic” movement, which distinguishes it fundamentally in terms of articulatory control from the careful positioning gesture needed for trilling. “Ballistic”, in terms of muscular activity, means that a brief and predetermined force is exerted on the articulating organ, and it moves for some time, often greatly exceeding the duration of muscular contraction. In practice, a short contraction of the superior longitudinal muscle of the tongue raises the tip rapidly. Once the momentary contraction ceases, the tip returns to the position determined by the underlying vocalic configuration. Barry draws the conclusion that a simple reduction of the time available for trilling does not turn the articulation into a tap. However, in terms of diachronic change, the author sees no reason not to assume that the tap derives from the trill by means of articulatory reinterpretation due to perceptual similarity. Martínez Celdrán and Fernández Planas (2007, 150-156) argue that whereas the trill is produced by Bernoulli’s principle, the tap is produced with a single voluntary elevation of the tongue tip, and that the tap is a lax sound compared to the tense trill. Electropalatographic evidence also shows that the Spanish tap is articulated in the alveolar region, and the trill in the postalveolar region. In addition, the tap is more susceptible to coarticulatory effects from adjacent vowels than the trill since the latter has a more precise articulatory mechanism than the former. The authors admit that the articulatory mechanisms of these two sounds are very different, but choose to continue placing both sounds under the same articulatory term vibrante. Considering the articulatory descriptions in this section, there seems to be no foundation for this decision.

2.3. The definition of vibrar In order to establish whether the term vibrante is adequate for both the tap and the trill, it may be useful to verify the definition of vibrar (‘to vibrate’) in Spanish and Portuguese language dictionaries. This comparison of dictionary definitions was inspired by Martínez Celdrán and Rallo’s (1995) article in which the suitability of the term vibrante is discussed on the basis of several arguments, including a dictionary reference. All citations are my translations. In the Diccionario de la lengua española (Real Academia Española, 2013), the mechanical definition of vibrar is “of an elastic object: to oscillate alternately around its equilibrium position”. Similarly, the

Iiris Rennicke

185

Diccionario de uso del español by María Moliner (Jarraud Milbeau et al. 2007) states that vibrar can mean that an object or a part of it moves “between two close positions, with a rapid and alternate movement from one position to the other”. To Martínez Celdrán and Rallo (1995, 191), this definition does not say that the movement is repeated, although they admit this information is implicit. Therefore, the Spanish definitions seem to suggest that vibration consists of alternate repetitive movements. In the Michaelis Moderno Dicionário da Língua Portuguesa (Editora Melhoramentos, 2012), intransitive vibrar is defined as “to produce sound or harmonies resulting from vibrations”. Another definition can be found in Infopédia (Porto Editora, 2013): “to cause or be the object of rapid and repeated movements to and fro or up and down”. These definitions are clear in stating that the number of oscillations involved in vibration is more than one. It should be remembered that dictionaries are always made by individuals and are thus subjective and partial in their definitions. However, after reviewing the definitions in the four dictionaries cited, we might conclude that vibrante is not the optimal name for an articulation that consists of only one tap of the tongue tip.

3. Vibrante in phonology After considering the suitability of vibrante as a name for the tap, I shall now examine the allophony of the two rhotic phonemes in Spanish and Portuguese. An interesting aspect to observe is the non-trill allophony of /r/. If trill realisations are scarce in these languages, we are faced with the absurd situation that neither /‫ݐ‬/ nor /r/ can be considered essentially “vibrating”. The consequence of this would be that vibrante is an unsuitable name for phonological reference to the rhotics in these languages. As Martínez Celdrán and Fernández Planas (2007, 157) conclude, rhotics is a more convenient name when several manners of articulation are taken into consideration.

3.1. The allophony of rhotics in Spanish The allophony of rhotics in Spanish is much richer than generally thought: acoustic analysis reveals that, in almost all phonological contexts, approximant and fricative variants are abundant and may even be more frequent than taps and trills. Peninsular, Costa Rican and Puerto Rican Spanish will be used to exemplify this variation.

186 Terminology Issues in Spanish and Portuguese Phonetics and Phonology

The /r/ phoneme can have both fricative and approximant variants in Peninsular Spanish. Approximant trill variants are mentioned by Navarro Tomás (1961, 122-123) and Blecua Falgueras (2001, Table 43). /r/ also has a fricative allophone, often referred to as “assibilated” in Hispanic linguistics, and symbolised by [Ĝ]. Navarro Tomás (1961, 124) describes it as a non-standard pronunciation that is produced with a slight rounding of the apical opening. In addition to approximant trills and fricative trills, Blecua Falgueras (2001) also reports one-tap realizations of the trill. References to an approximant variant of intervocalic /‫ݐ‬/ in Peninsular Spanish can be found in the literature (Navarro Tomás 1961, 117-118; Gili 1921, 274). In Blecua Falgueras’ (2001) reading corpus, most tokens of intervocalic /‫ݐ‬/ were pronounced as alveolar approximants, proving that even in a reading task approximant allophones are dominant and most likely do not belong only to spontaneous speech. Approximant and fricative variants of /‫ݐ‬/ can also be observed in consonant clusters of the form /C‫ݐ‬/. Malmberg (1965, 39) explains that the articulatory transition from a consonant to a tap or from a tap to a consonant is rather complex, and for this reason many languages tend to dispose of these series, either by introducing a vocalic element between them or reducing the series in one way or another. This last solution could be seen as the motivation behind the tap’s variants in contact with a consonant in Spanish. In Blecua Falgueras’ (2001) study of Peninsular Spanish, the tap in consonant clusters was realised mostly as an approximant. Especially in /t‫ݐ‬/ and /d‫ݐ‬/, the tap can be realised as the socalled “assibilated” rhotic in the Álava, Navarra, Rioja and Aragón regions of Spain (Navarro Tomás 1961, 120). Syllable coda is the position where Spanish rhotics are known to neutralize. Therefore, in addition to tap variants, we find trills, but also alveolar approximants and fricatives in Peninsular Spanish (Navarro Tomás 1961, Blecua Falgueras 2001). Liquids are prone to neutralisation in this context in Andalusian Spanish, resulting in [‫]ݐ‬, [‫]݋‬, [l], [‫ ;]݌‬in addition, [h] and elision are frequent in word-final coda in this variety (Penny 2000). An interesting variation of rhotics can be observed in Costa Rican Spanish. In addition to alveolar rhotics, references to retroflex approximants can be found in the literature. Quesada Pacheco and Vargas Vargas (2010, 168-171) found alveolar and retroflex approximants as variants of /r/: in both word-initial and intervocalic position, approximant allophones are most frequent in Costa Rican Spanish. Calvo Shahid (1995) finds that the alveolar trill and a “voiced alveolar assibilated affricate” are the most common allophones of /r/ after pause or another consonant; in

Iiris Rennicke

187

intervocalic position, the most frequent realization is the retroflex approximant, followed by [r]. To summarize, /r/ has not only alveolar trill variants in this variety, but also approximant (alveolar and retroflex) and fricative variants. As for the tap phoneme in Costa Rican Spanish, Calvo Shahid (1995) found that 26% of intervocalic /‫ݐ‬/’s were alveolar fricatives. Similarly, in the consonant clusters /t‫ݐ‬/ and /d‫ݐ‬/ the rhotic is often a fricative. In wordmedial coda, Costa Rican Spanish speakers can produce the rhotic as a tap, voiced alveolar fricative, alveolar approximant or retroflex approximant; in word-final coda, voiceless and voiced alveolar approximants are frequent (Quesada Pacheco and Vargas Vargas 2010; Calvo Shahid 1995). In at least one variety of Spanish, that of Puerto Rico, /r/ has retracted much in the same way as in Portuguese (see 3.2.): as mentioned in section 1.2., the trill phoneme of this variety retracted to a uvular place of articulation, and then weakened into a fricative. This brief overview on rhotic variation in Spanish demonstrates that the reality of rhotics in Spanish is certainly not limited to alveolar taps and trills. Thus a more comprehensive term than vibrantes for phonological reference is in order.

3.2. The allophony of rhotics in Portuguese Rhotic variation in Portuguese is more complex in nature than in most varieties of Spanish since it encompasses not only several manners of articulation, but also several places of articulation: alveolar, retroflex, velar/uvular and glottal. The retraction of the alveolar trill /r/ is likely to have begun in the late 19th century in European Portuguese. At the beginning of the 20th century, Gonçalves Viana (1903) described the uvular trill [‫ ]ݒ‬as a common variant at the time, but still considered erroneous. Half a century later, Barbosa (1965) argues that the alveolar trill is already disappearing from Lisbon Portuguese, and only used by the older generation. At this stage, he describes the uvular trill as the most common variant in Lisbon, and the velar fricative [x] as an emerging pronunciation among the younger generation. Recently, voiced and voiceless uvular fricatives [‫ ]ݓ‬and [Ȥ] are described as common allophones in Lisbon by Cruz-Ferreira (1995), Mateus and d’Andrade (2000), and Jesus and Shadle (2005). A detailed study on the European Portuguese /r/ by Rennicke and Martins (2012) revealed that there are at least five different allophones: [r] (present in the speech of 11% of all informants), [‫( ]ݒ‬11%), [‫( ]ݓ‬76%), [Ȥ] (24%), and [x] (16%). An individual informant in the corpus never used more than three

188 Terminology Issues in Spanish and Portuguese Phonetics and Phonology

different variants, and the three most common combinations of rhotics were [‫( ]ݓ‬47% of all informants), [Ȥ]+[‫( ]ݓ‬11%), and [x]+[‫( ]ݓ‬7%). The overwhelming dominance of uvular fricatives in the data confirms the information provided by the authors mentioned above: trills, be they alveolar or uvular, seem to be disappearing from European Portuguese. Jesus and Shadle (2005) also report a tapped alveolar fricative [‫ ]ࡢݐ‬as an allophone of both /r/ and /‫ݐ‬/. In this articulation, there is a constriction of the vocal tract for a very short time, which results in a low-amplitude burst of frication noise. In Brazilian Portuguese, the retraction process of /r/ has gone even further and, through debuccalization3, resulted in glottal fricatives [h ܸ] (Cristófaro-Silva 2007, 140-143). The glottal fricative is the most common allophone on a national level in Brazilian Portuguese (Callou, Leite and Moraes 2002, 544), but alveolar and uvular trills still exist in some varieties. The retraction and weakening process has also affected coda rhotics: in addition to [‫ ]ݐ‬and [r], dialects in Central and Southern Brazil have alveolar and retroflex approximant allophones [‫]ݍ ݋‬. In the rest of the country, coda rhotics are usually back fricatives [x ܵ Ȥ ‫ ݓ‬h ܸ], and more rarely, uvular trills [‫( ]ݒ‬Cristófaro-Silva 2007, Noll 2008, Callou, Leite, and Morais 1996). In most Brazilian dialects, the tap [‫ ]ݐ‬occurs only as the second member of syllable-initial consonant clusters and intervocalically. In short, the trill articulation of /r/ has been gradually replaced by fricatives in both European and Brazilian Portuguese, and is no longer used by the majority of speakers. These same back fricatives, alongside alveolar and retroflex approximants, also occur as allophones of /‫ݐ‬/ in coda in Brazilian Portuguese. Thus the rhotics of Portuguese are produced in several places and with several different manners of articulation, and are insufficiently described by the term vibrantes. The symbol of choice for most Portuguese authors to represent the historical /r/ phoneme has been the uvular trill, /‫ݒ‬/ (e.g. Mateus and d’Andrade 2000; Jesus and Shadle 2005). This practice seems out of date considering the dominance of uvular fricatives in European Portuguese, but also inconsistent from the point of view that the alveolar trill [r] (from which the uvular trill derives) is still used by some speakers. This problem has been solved in Brazilian linguistics with the abstract terms vibrante forte or r-forte (‘strong r’, meaning historical /r/) and vibrante branda/fraca and r-fraco/r-brando (‘weak r’, meaning /‫ݐ‬/) (e.g. Monaretto, Quednau and Da Hora 2010). This solution has the advantage, in the case 3

Weakening process that turns an oral segment into a non-oral (glottal) segment, [h]. Also referred to as de-oralization and de-articulation. (Lass 1984, 179, Honeybone 2008).

Iiris Rennicke

189

of r-forte and r-fraco/r-brando, of not specifying a manner or place of articulation for either phoneme; however, there seems to be no phonetic or phonological basis for the “strength” and “weakness” of the two phonemes respectively. The symbol adopted for historical /r/ in Brazilian linguistics varies from author to author: abstract symbols are used by e.g. CristófaroSilva (2007, 142) (/Rࡄ /) and Callou, Leite and Moraes (2002, 543) (R, without slashes to indicate phoneme status). On the other hand, authors like Barbosa and Albano (2004, 228) prefer /r/.

4. Discussion In this paper I have presented arguments against the traditional terminology vibrante simple(s) for the tap and vibrante multiple/múltipla for the trill in light of some fundamental differences between these sounds. Most studies cited show that the closure phase of a tap is longer than a closure phase of a trill, which means that from an acoustic point of view, a trill is not exactly a series of taps. It is also widely acknowledged that the two sounds differ in their manner of articulation, and Martínez Celdrán and Fernández Planas (2007) present evidence that the Peninsular Spanish tap is more prone to effects of coarticulation, and articulated in a slightly different place than the trill. In addition, the intuition that vibration is repeated oscillation and therefore not an apt name for a single-movement articulation is supported by some dictionary definitions. In terms of phonological reference, an overview on rhotic variation in Spanish and Portuguese indicates that both phonemes have a rich allophony which is not captured by the term vibrantes. This is especially true of Portuguese: it seems that most speakers in Portugal and Brazil never use the trill variants [r] or [‫]ݒ‬. Martínez Celdrán and Rallo (1995) are some of the few authors who have discussed the terminology related to rhotics. Their article concerns Spanish and defends the terms vibrante simple and vibrante multiple, arguing that due to the perceptual similarity and phonological behaviour of taps and trills, they must belong to the same class, and therefore carry the same name. All citations from the original article in Spanish are my translations. The authors claim (p. 185) that, because the Spanish tap and trill can be neutralized in all positions except intervocalic, they are “related” and therefore cannot be considered separate manners of articulation – their differences can be explained by the difference in tension during articulation. The fact that the tap and the trill can be neutralized in most positions proves to the authors that the two sounds have “a unique basis of

190 Terminology Issues in Spanish and Portuguese Phonetics and Phonology

comparison in the system”. The authors pose the following question: if that basis is not the feature vibrante, what is it? To Martínez Celdrán and Rallo, neutralization is not the only proof of unity between the tap and the trill: the fact that in languages with only one rhotic phoneme (French is given as an example) the two sounds can be allophones of the same phoneme is considered evidence of them sharing the same vibrante feature. In addition, the authors give examples of historical changes from the tap to the trill and vice versa as proof of interchangeability from one sound to another “without a logical explanation”. Martínez Celdrán and Rallo then carry out a psychoacoustic test in which the perceptibility of trills created by combining taps in sequence and taps extracted from trill closure phases is assessed. 94.37% of the manipulated words are correctly identified by the test subjects, proving that perceptually the closure phases of these sounds are very much alike. From this the authors conclude that the articulatory differences between these sounds are irrelevant if acoustically and perceptually they produce a “physical reality” that is equivalent to the “sensation” (p. 191). By claiming that the tap and the trill must both be vibrantes because there is no other feature they could possibly have in common, the authors seem to confuse the concepts of manner of articulation and class. As cited in section 1.2, according to Lindau (1985) the relation between phonological and phonetic classes may be more complex than the direct correspondence generally assumed. If rhotics, with all their different manners and places of articulation, are accepted as a valid class, then there is no need to establish a common feature for the tap and the trill in order to make them fit into the same class. It would be impossible to pose the same kind of question in relation to, for example, Brazilian Portuguese: what would be the common feature between /‫ݐ‬/ and the most common variant of /r/, a back fricative, in a minimal pair such as caro – carro [‫ޖ‬kar‫– ]ݜ‬ [‫ޖ‬kah‫ ?]ݜ‬If no feature is shared by the two phonemes (in this case neither manner nor place of articulation), must we decide that the two sounds have absolutely nothing in common? In this case, would the weakening of coda /‫ݐ‬/ to [h] (carta [‫ޖ‬kaht‫ )]ܣ‬be purely coincidental? The most obvious solution is to define rhotics as suggested in section 1.2: as a class of sounds that may or may not have anything in common articulatorily or acoustically, but behave in a similar manner phonologically, and are often in allophonic relation to each other due to diachronic change and weakening processes. The fact that taps and trills can be allophones of the same phoneme in some languages is hardly proof of their “relatedness” and feature sharing any more than all the other allophonic relations explained in section 3. As

Iiris Rennicke

191

has been seen, in Spanish and Portuguese rhotics several allophonic relations can be observed between trills, taps, fricatives and approximants. They are clearly related but may not share any feature. The historical changes from taps to trills and vice versa can be linked to the perceptual similarity between the two sounds. Barry (1997), already cited in section 2.2, argues that the tap may well derive from a trill by means of articulatory reinterpretation due to perceptual similarity. However, he states that a trill does not turn into a tap if the time for articulation is reduced because the manners of articulation are different. Similarly, it can be argued that a tap does not turn into a trill if the time for articulation is increased because the position of the tongue and the airstream required to produce a trill are different. The results of Martínez Celdrán and Rallo’s (1995) psychoacoustic test prove that the closure phases of these sounds truly are auditorily very similar, which might explain their interchangeability in diachronic developments. However, this does not change the fact that the tap and the trill have different manners of articulation, and that the duration of these auditorily similar closure phases actually seems to be slightly different. What can the allophony of /r/ and /‫ݐ‬/ in Spanish and Portuguese tell us about rhotics as a class? Table 1 condenses the variants described in section 3. Non-rhotic allophones due to coda neutralization of liquids are excluded. “x” in the table indicates that the allophone in question can be found in some variety of the language. Allophones from different phonological contexts are grouped under the corresponding phoneme. Syllable coda appears under neither phoneme since, in this context, variation is common and the establishment of an underlying phoneme is challenging. More allophones may exist, of course, but acoustic and articulatory studies especially on European Portuguese rhotics are scarce.

x x

alveolar trill alveolar fricative trill alveolar approximant trill alveolar tap tapped alveolar fricative alveolar fricative alveolar approximant retroflex approximant uvular trill velar/uvular fricative glottal fricative x

Table 1: Rhotic variation in Spanish and Portuguese globally

x x

x

x x

x x

x

x

x

Coda (wordmedial, final) x

x

Spanish /‫ݐ‬/ (second in cluster, intervocalic)

x

x

/r/ (word-initial, intervocalic)

x

x x

x

x

/r/ (word-initial, intervocalic)

x

Portuguese /‫ݐ‬/ (second in cluster, intervocalic)

Terminology Issues in Spanish and Portuguese Phonetics and Phonology

Phoneme Allophones

192

x

x x

x

x

x x

Coda (wordmedial, final) x

Iiris Rennicke

193

Table 1 demonstrates that the manners and places of articulation of the rhotics in these languages are indeed very varied: we find trills, taps, fricatives, and approximants; alveolar, retroflex, velar, uvular, and glottal places of articulation. It also becomes evident that the two phonemes share some allophones, which is one more proof of class unity through allophony and different stages of diachronic change. There has been discussion over the number of rhotic phonemes in Iberian Romance languages: some scholars argue that the fact that the opposition only exists in intervocalic position means that they derive from the one and same phoneme. The view adopted in this article is that of two underlying phonemes since the focus is on phonetic terminology and on the relations between the different rhotic allophones that exist in Spanish and Portuguese, irrespective of the phoneme(s) from which they are derived. A detailed discussion on the phoneme status of rhotics can be found in e.g. Bonet and Mascaró (1997), and Monaretto, Quednau and Da Hora (2010). Figure 2 gathers the allophones in Table 1 in a network of similarities inspired by Lindau’s (1985, 167) parameter relations (Figure 1).

Figure 2: Network of rhotic variants in Spanish and Portuguese: connections through manner of articulation (line) and place of articulation (dash line, linking variants in a chain)

The three different alveolar trills are similar in their manner of articulation in the sense that they all require a sufficient airstream in order to produce the repetitive pulsing movement of the tongue tip. They differ in the degree of the closure phases, varying from full contact to friction,

194 Terminology Issues in Spanish and Portuguese Phonetics and Phonology

and finally to vowel-like approximant phases. The fricative trill then resembles the non-trilled alveolar fricative and the back fricatives in its manner of articulation. The tapped alveolar fricative seems like a combination of a tap and a fricative since it begins with a closure phase but ends in frication, thus resembling both sounds. By reducing articulatory tension, the alveolar fricative becomes an alveolar approximant, which is very similar perceptually to the retroflex approximant. The individual pulses of the alveolar trill resemble the alveolar tap perceptually, and the uvular trill has the same kind of pulse pattern as the alveolar trill. When the uvular trill is weakened (i.e. when articulatory tension is reduced), fricatives emerge in the same place of articulation. These can, in turn, weaken into glottal fricatives by debuccalization. The family resemblance network illustrated in Figure 2, inspired by the classic descriptions of rhotics as a class by Lindau (1985) and Ladefoged and Maddieson (1996), bears a connection to exemplar-based approaches to phonology (Bybee 2001; Pierrehumbert 2001). According to these authors, actual language use and frequency of lexical items form networklike connections between phonetic alternates, and the members of a phonological category are identified in terms of similarity with the more prototypical member(s). Bybee (2001) claims that redundant phonetic detail is stored and categorised, which produces complex categories that are centred around several prototypes or good exemplars. Phonetic alternates are registered in the lexicon and related to one another by lexical connections that track both phonetic and semantic similarities. Essentially, phonemes are sets of phonetically similar variants, and allophones are seen as salient contextually determined prototypes. Similarly, Pierrehumbert (2001) sees that each sound category is represented in memory by a large cloud of remembered tokens of that category. The phonological system is thus a mapping between points in a phonetic parameter space and the labels of the categorization system. Every encountered token of a category is stored as a separate exemplar, and as a consequence, frequent categories will be represented by more tokens than infrequent categories. In the case of rhotics, analysed from an exemplar-based point of view, it would be assumed that the mental representation of rhotics for a speaker is based on context-related exemplars: for every phonological context, there are more exemplar and less exemplar variants, and this is based on frequency of occurrence in actual language use. The assumption that speakers can have multiple mental representations of rhotics also explains why they are not consistent in rhotic production, but rather exhibit considerable variation. Consequently, I propose that dialectal (and other

Iiris Rennicke

195

kinds of linguistic) variation in rhotics is caused by different frequencies of occurrence of linguistic phenomena (Kretzschmar 2010); it also explains why speakers are able to identify segments as different as trills, fricatives, approximants, and taps as rhotic variants. Figure 2 demonstrates how complex the relations within the class of rhotics can be: let us consider two quite opposite variants such as the alveolar trill [r] and the glottal fricative [h]. They do not share a manner or place of articulation, and indeed in many languages such as English [h] would not be considered rhotic at all, but an allophone of /h/; or, in the case of most Spanish varieties, most probably a variant of /x/. However, in the case of Portuguese, once we establish the historical connection between the two, the link becomes obvious: the alveolar trill has retracted to a uvular trill since the 19th century, then weakened into a fricative in the velar/uvular region, and then weakened still into a glottal fricative. In addition, all these intermediate stages are still used by speakers at different levels of frequency, which enables speakers to have a mental representation of the /r/ phoneme that encompasses a rich variety of alternates.

5. Conclusion I hope to have inspired discussion on phonetic and phonological terminology in the target languages regarding rhotics. The rhotics of Spanish and Portuguese belong to a family resemblance network formed by even more variants in number than those mentioned by Ladefoged and Maddieson (1996, 244-245) and Lindau (1985, 166-167), which makes rhotic variation in these languages an interesting research subject. This variation also discourages the use of vibrantes and favours the use of conso(n)antes róticas as an umbrella term in phonological reference to the class of rhotics. As cited in section 1.1, some authors have reserved the term vibrante solely for the trill, and adopted the loan word tap (sometimes adapted to Brazilian Portuguese pronunciation as tepe [‫ޖ‬t‫ܭ‬pܼ]) or the more descriptive term percusiva (in Spanish) for the tap. These solutions can act as a first step towards streamlining and rationalizing the current terminology.

References Andrade, Amália, and Maria do Céu Viana. 1996. “Fonética.” In Introdução à Linguística Geral e Portuguesa, edited by Isabel Hub

196 Terminology Issues in Spanish and Portuguese Phonetics and Phonology

Faria, Emília Ribeiro Pedro, Inês Duarte, and Carlos M. A., Gouveia, 115-167. Lisboa: Caminho. Barbosa, Jorge Morais. 1965. Études de phonologie portugaise. Lisboa: Junta de Investigações Científicas do Ultramar. Barbosa, Plínio A., and Eleonora C. Albano. 2004. “Illustrations of the IPA: Brazilian Portuguese.” Journal of the International Phonetic Association 34(2): 227-232. Barry, William J. 1997. “Another R-tickle.” Journal of the International Phonetic Association 27: 35-45. Blecua Falgueras, Beatriz. 2001. “Las vibrantes del español: manifestaciones acústicas y procesos fonéticos.” PhD diss., Universidad Autónoma de Barcelona. Bonet, Eulalia, and Joan Mascaró. 1997. “On the representation of contrasting rhotics.” In Issues in the Phonology and Morphology of the Major Iberian Languages, edited by Fernando Martínez-Gil and Alfonso Morales-Front, 103-126. Washington, D.C.: Georgetown University Press. Bybee, Joan. 2001. Phonology and Language Use. Cambridge University Press. Callou, Dinah, Yvonne Leite, and João A. Moraes. 1996. “Variação e diferenciação dialetal: a pronúncia do /R/ no português do Brasil.” In Gramática do Português Falado, vol. VI, edited by Ingedore G. Villaça Koch, 465-493. Campinas: Editora Unicamp. Callou, Dinah, Yvonne Leite, and João A. Moraes. 2002. “Processo(s) de enfraquecimento consonantal no português do Brasil.” In Gramática do Português Falado, vol. VIII, edited by Maria Bernadete M. Abaurre and Angela C. S. Rodrigues, 537-555. Campinas: Editora Unicamp. Calvo Shahid, Annette. 1995. “Variación fonética de /‫ݐ‬/ y /r/ en el habla culta de San José.” Revista de Filología y Lingüística de la Universidad de Costa Rica XXI(1): 115-134. Catford, John C. 1977. Fundamental problems in phonetics. Edinburgh: Edinburgh University Press. Cristófaro-Silva, Thaïs. 2007. Fonética e fonologia do português: roteiro de estudos e guia de exercícios. São Paulo: Contexto. Cruz-Ferreira, Madalena. 1995. “European Portuguese.” Journal of the International Phonetic Association 25(2): 90-94. Editora Melhoramentos. 2012. “Significado de ‘vibrar’”. In Dicionário da Língua Portuguesa, edited by Walter Weiszflog, and Michaelis Moderno. Accessed July 1, 2013.http://michaelis.uol.com.br/modern /portugues/definicao/vibrar %20_1065467.html

Iiris Rennicke

197

Gili, Samuel. 1921. “La r simple en la pronunciación española.” Revista de filología española VIII: 271-280. Gonçalves Viana and Aniceto dos Reis. 1903. Portugais: phonétique et phonologie, morphologie, textes. Leipzig: Teubner. Honeybone, Patrick. 2008. “Lenition, weakening and consonantal strength: tracing concepts through the history of phonology.” In Lenition and Fortition, edited by Joaquim Brandão de Carvalho, Tobias Scheer, and Philippe Ségéral, 9-93. Berlin: Mouton de Gruyter. International Phonetic Association. 2013. “Reproduction of The International Phonetic Alphabet (2005).” Accessed January 7. http://www.langsci.ucl.ac.uk/ipa/fullchart.html Jarraud Milbeau, Annie, Silvia Ramón Jarraud, Fabián Ramón Jarraud, and Helena Ramón Jarraud. 2007. “Vibrar.” In María Moliner, Diccionario de uso del español. Madrid: Gredos. Jesus, Luis M. T., and Christine H. Shadle. 2005. “Acoustic Analysis of European Portuguese uvular [Ȥ, ‫ ]ݓ‬and voiced tapped alveolar [‫]ࡢݐ‬ fricatives.” Journal of the International Phonetic Association. 35(1): 118. Kretzschmar, William. A. 2010. “Language Variation and Complex Systems.” American Speech 85 (3): 263-286. Ladefoged, Peter, and Ian Maddieson. 1996. Sounds of the World’s Languages. Oxford: Blackwell. Lass, Roger. 1984. Phonology: An Introduction to Basic Concepts. Cambridge: Cambridge University Press. Lindau, Mona. 1985. “The Story of /r/.” In Phonetic Linguistics: Essays in Honor of Peter Ladefoged, edited by Victoria A. Fromkin, 157-168. Orlando: Academic Press. Lipski, John M. 2007. El español de América. Madrid: Cátedra. Malmberg, Bertil. 1965. Estudios de fonética hispánica. Madrid: Consejo Superior de Investigaciones Científicas. Martínez Celdrán, Eugenio, and Ana María Fernández Planas. 2007. Manual de fonética española: Articulaciones y sonidos del español. Barcelona: Ariel. Martínez Celdrán, Eugenio, and Lucrecia Rallo. 1995. “[‫ ݐ‬- r]: ¿Dos clases de sonidos?” Estudios de fonética experimental, VII: 179-194. Barcelona: PPU. Massone, María Ignacia. 1988. “Estudio acústico y perceptivo de las consonantes nasales y líquidas del español.” Estudios de fonética experimental, III: 14-34. Barcelona: PPU. Mateus, Maria Helena, and Ernesto D’Andrade. 2000. The Phonology of Portuguese. Oxford: Oxford University Press.

198 Terminology Issues in Spanish and Portuguese Phonetics and Phonology

Monaretto, Valéria N. O., Laura Rosane Quednau, and Dermeval Da Hora. 2010. “As consoantes do português.” In Introdução a estudos de fonologia do português brasileiro, edited by Leda Bisol, 202-235. Porto Alegre: EDIPUCRS. Navarro Tomás, Tomás. 1961. Manual de pronunciación española. Madrid: Consejo Superior de Investigaciones Científicas. Noll, Volker. 2008. O português brasileiro: formação e contrastes. São Paulo: Globo. Obediente, Enrique. 2007. Fonética y fonología. Mérida: Consejo de Publicaciones de la Universidad de los Andes. Penny, Ralph. 2000. Variation and Change in Spanish. Cambridge: Cambridge University Press. Pierrehumbert, Janet B. 2001. “Exemplar dynamics: word frequency, lenition and contrast.” In Frequency and the Emergence of Linguistic Structure, edited by Joan Bybee, and Paul Hopper, 137-157. Amsterdam/ Philadelphia: John Benjamins. Porto Editora. 2013. “Definição de vibrar.” In Infopédia (Dicionário da Língua Portuguesa – com Acordo Ortográfico). Accessed January 16. http://www.infopedia.pt/lingua-portuguesa/vibrar Quesada Pacheco, Miguel Ángel, and Luís Vargas Vargas. 2010. “Rasgos fonéticos del español de Costa Rica”. In El español hablado en América Central: Nível fonético, edited by Miguel Ángel Quesada Pacheco, 155-175. Madrid/Frankfurt: Iberoamericana/Vervuert. Quilis, Antonio. 1999. Tratado de fonología y fonética españolas. Madrid: Gredos. Real Academia Española. 2013. “Vibrar.” In Diccionario de la lengua española (22nd ed.). Accessed January 16. http://lema.rae.es/drae/?val=vibrar Rennicke, Iiris, and Pedro Martins. 2012. “Algumas considerações sobre as realizações fonéticas de /R/ em português europeu.” Paper presented at XVIII Encontro Nacional da Associação Portuguesa de Linguística, Faro, Portugal, October 25-27. Scobbie, James M. 2006. “(R) as a variable.” In Encyclopedia of Language and Linguistics, vol. 10, edited by Keith Brown, 337-344. Oxford: Elsevier.

RELATIONS BETWEEN GRAMMAR AND READING: A STUDY OF ANAPHOR PRONOUNS IN LOWER SECONDARY EDUCATION1 JOANA BATALHA CENTRO DE LINGUÍSTICA DA UNIVERSIDADE NOVA DE LISBOA

Abstract This article will present a research project on linguistics and mother tongue teaching. Based on several studies that have established a relation between linguistic awareness and reading, namely between syntactic awareness and reading comprehension (Bowey 1986, Bentin, Deutsch and Liberman 1990, Costa 1991, Leal and Roazzi 1999, Gaux and Gombert 1999, Nation and Snowling 2000, Mokhtari and Thompson 2006), this study aims to investigate the benefits that explicit teaching of some syntactic structures, including anaphor pronouns, may bring regarding an improvement in reading comprehension in the mother tongue. Two groups of students will be compared in terms of their reading comprehension ability, with only one of the groups receiving explicit teaching of the syntactic structures being studied, based on a discovery-learning method (Hudson 1992), and will be taught strategies to use this grammatical knowledge to solve problems which occur during reading comprehension (Costa et al. 2010). Some teaching activities with anaphor pronouns and the first results from a pre-test with students in lower secondary education will be presented. Keywords: Mother tongue teaching, Reading, Syntactic awareness, Anaphor pronouns.

1

This work was supported by the National Funding Organisation – FCT in the Strategic Project PEst-OE/LIN/UI3213/2011.

200

Relations between Grammar and Reading

1. Teaching grammar: relevant national context In the last decade, there has been a lively debate on the teaching of grammar in Portugal, mainly motivated by successive and often contradicting statements from the Portuguese Ministry of Education regarding the Portuguese language curriculum. In 2001, with the publication of the National Curriculum for Basic Education, the teaching of grammar was considered in the subject Portuguese under the name explicit knowledge of language and this was considered one of the core skills students should learn and develop in their mother tongue. However, in the program for the teaching of Portuguese in effect at the time (since 1991), grammar, called functioning of language, was transversal and peripheral. In 1997, a reference document for teachers (Mother tongue in basic education) had already argued in favor of explicit knowledge of language, understood as the “progressive awareness and systematization of implicit knowledge in the use of language” (Sim-Sim, Duarte and Ferraz 1997, 31), but it was not until 2009, with the approval of a new program for the teaching of Portuguese (Reis 2009), that specific contents and a methodology for teaching grammar were presented2. These inconsistencies have been legitimising divergent teaching practices as far as grammar is concerned. A national study (Duarte, R. 2008) showed that Portuguese teachers gave grammar less importance than reading and writing. It also demonstrated that traditional methods, based on knowledge of definitions, were the most used. Similarly, in her research into the teaching of grammar and writing, Costa, A. L. (2010) identified three distinct modes of teaching grammar in Portuguese schools. Firstly, a traditional approach is predominant, focusing on normative rules and assuming memorization and knowledge of definitions as main strategies. Secondly, alongside the traditional approach, and emerging primarily in reaction to it, there is the functioning of language perspective, which argues that grammar should be taught to support the communicative skills of listening, speaking, reading and writing. Since teaching practices based on this perspective do not consider grammar as an autonomous object of study, and make grammatical contents dependent on oral and written texts, this has legitimised the non-existence of systematic linguistic 2

At the beginning of the 2012-2013 school year, the Ministry of Education approved Metas Curriculares para o Ensino Básico – Português, a new reference document which defined curricular goals from the first to the ninth grade. Although these curricular goals are based on the 2009 Portuguese program, there are some divergences between the two documents as far as the teaching of grammar is concerned, but these will not be considered in the present article.

Joana Batalha

201

study and ultimately resulted in the annulment of grammar itself. Finally, the third mode, supported by different theories of linguistics and psychology, corresponds to a group of perspectives of language teaching known as language awareness or, more specifically, linguistic awareness. According to these, “subjects should be made aware of the processes, strategies and structures involved in situations involving oral interaction, writing and reading, in order to deliberately monitor and control their speech and to act socially, and consciously, according to them” (Costa, A.L. 2010, 36). With the linguistic awareness perspective, not only are the instrumental objectives of teaching grammar reinforced, but also a noninstrumental objective is assumed, which is the teaching of grammar as an autonomous object of study. The coexistence of different modes of teaching grammar, or the nonexistence of grammar teaching, may explain, to a certain extent, the poor results obtained by Portuguese students in tasks that require grammatical knowledge. Reports of national examinations from 2010, for example, demonstrate that the percentage of correct answers in grammar items in the sixth grade is approximately 30% (GAVE 2010). Similarly, in 2011, the worst results obtained by students in the ninth grade were found in a grammar item, with only 11.4% of students responding correctly (Sousa 2011).

2. Relations between linguistic awareness and reading: the role of syntactic awareness In recent decades, several studies have tried to establish relations between explicit knowledge of language – or, at least, between a certain degree of awareness about the formal properties of language, referred to as linguistic awareness – and reading. Although the concept of linguistic awareness is used in different fields of linguistics and psychology with different meanings, this article will adopt Titone’s (1988) perspective, according to which linguistic awareness, allowing the subject to observe language as an object that can be explored, is closer to implicit knowledge and results from cognitive maturation, in this way, differing from explicit knowledge of language characteristics and functions, as a result of formal education or systematic instruction, mostly through grammar teaching. In line with Sim-Sim (1998) and Duarte, I. (2008), linguistic awareness is thus understood as an intermediate stage between implicit and explicit knowledge of language, with the latter corresponding to a superior degree of linguistic knowledge control and awareness.

202

Relations between Grammar and Reading

Works from the 1970s such as Kavanagh and Mattingly (1972) claim that listening and speaking activities are not parallel to the reading activity; reading is a secondary linguistic activity, since it requires the reader’s awareness of the primary activities of listening and speaking. This perspective lead to several studies forming hypotheses regarding the relation between the level of linguistic awareness and the reading domain, with it being implied that the difficulties experienced in reading would be related to linguistic awareness, which changes from individual to individual. In fact, it is widely accepted today that reading and writing are not acquired in a natural and spontaneous way, as oral language. The child acquires language from the automatic activation of mechanisms, innate and biologically predetermined, when exposed to linguistic input in his/her family environment. This means that the child understands language and learns how to speak without having consciously to know the formal language structure and its rules, which does not occur in the case of written language. An extended contact with writing is not a prerequisite for mastering this code. In addition, the differences between oral and written language do not stand only in relation to the sensory modality used (visual for reading and writing and auditory for oral), but also in relation to the type of tasks used, which require, in the case of written language, a higher level of abstraction, elaboration and control than in oral language. Hence, we can understand reading as a formal linguistic activity, as its learning requires that the child develops an explicit awareness of linguistic structures that should be intentionally manipulated (Duarte, I. 2008). Although the direction of the relation between linguistic awareness and reading is not consensual (Tunmer and Hoover 1992; Karanth, Kudva and Vijavan 1995; Ravid and Tolchinsky 2002), current theoretical perspectives are in line with the fact that the process of learning how to read implies that the subject becomes aware of the knowledge he/she possesses of oral language in its different domains (Sim-Sim 1998, Guimarães 2003). According to Ravid and Tolchinsky (2002), the relation between linguistic awareness and literacy occurs in at least three ways: firstly, linguistic awareness requires cognitive control over linguistic forms; secondly, not being a uniform phenomenon, linguistic awareness evolves in terms of explicitness into a reorganization of knowledge, making it more coherent and accessible (this evolution results from the combination of three factors: development, linguistic experience and formal instruction); finally, the level of oral language awareness obtained before going to school is correlated to the level of success in reading and writing

Joana Batalha

203

learning, even though the relation between oral language and literacy is not unidirectional, as development in written language promotes linguistic awareness as well. The aspect of linguistic awareness that first interested researchers, due to the central role played in the initial stages of learning reading, was phonological awareness, which can be viewed as “the ability to identify and manipulate syllables, syllable constituents and segments regardless of meaning” (Duarte 2010, 13). Literature that relates phonological awareness and reading is very abundant. Studies carried out, using mostly tasks that measure phonological sensitivity and phonological awareness tasks, have demonstrated that the relation between phonological awareness and reading is of two kinds: phonological awareness facilitates the learning of the written code and can even be predictive of success in reading; on the other hand, learning reading seems to contribute to the development of certain phonological abilities (Demont and Gombert 1996; Gombert 2003; Guimarães 2003). Other types of linguistic awareness – lexical awareness, syntactic awareness, pragmatic awareness and, more recently, morphological awareness and textual awareness (Correa 2004) – have been investigated, focusing researchers’ attention on the measurement and description of these linguistic abilities, as well as their influence on learning written language. Certain authors have been arguing that, just like phonological awareness, syntactic awareness is decisive in the success of learning to read, although this relation is still not as well established as phonological awareness (Barrera and Maluf 2003; Capovilla, Capovilla and Soares 2004; Cain 2007). Syntactic awareness can be defined as the subject’s ability to reflect on the syntactic aspects of language and to deliberately control the application of grammar rules (Gombert 1990). Syntactic awareness can include the ability to identify syntactic units as sentences and phrases, concordance processes and the local and long distance relations established between phrases (Duarte 2010). Most of the studies that examine relations between syntactic awareness and reading focus on the initial stages of reading, studying populations in the 5-8 age group. Some of these studies tried to relate syntactic awareness to other types of linguistic awareness, particularly phonological awareness, and its contributions to learning how to read. The work developed by Tunmer and colleagues (Tunmer, Nesdale and Wright 1987; Tunmer and Hoover 1992) enabled syntactic awareness, phonological awareness and learning reading to be connected. As claimed by these authors, syntactic context influences reading in two ways: for

204

Relations between Grammar and Reading

children who have not yet automated word recognition, it provides information for word identification when the phonological information is incomplete, that is, there is a combination of two types of linguistic awareness, phonological and syntactic, to decode difficult words (for example, spelling patterns associated with different accents) and new words; in addition, it allows readers to monitor comprehension processes in an effective way. In particular, in the Tunmer et al. (1987) study, where poor, older readers (8-9) were matched with good, younger readers (6-7) according to their reading ability, it was shown that the former had worse performances in two oral measures of syntactic awareness (one of sentence completion and the other of correction), thus concluding that a slow development in syntactic awareness may delay reading development. In another study, Rego and Bryant (1993) sought to clarify the role of phonological awareness and syntactic awareness in learning reading through a longitudinal study of 57 English first year children (aged between 5;3 and 5;8). The authors acknowledged that there is a specific relation between initial performance in syntactic awareness tasks and the further use of context when reading words present spelling difficulties. Phonological awareness was shown to be directly related to the understanding of the alphabetical principle. Hence, each of these types of linguistic awareness facilitates the decoding of different forms: syntactic awareness is crucial in the use of contextual clues, whereas phonological awareness acts mostly upon the sound/grapheme correspondence. Demont and Gombert (1996) also studied the role played by phonological awareness and syntactic awareness in the initial stages of reading, predicting a more relevant effect for phonological awareness in word recognition and syntactic awareness in comprehension, namely through monitoring processes. This was a three year longitudinal study which followed a group of French speaking children from pre-school until the third grade (average age between 5;7 years at the beginning of the study and 8;8 years at its end). Besides confirming the relevance of metalinguistic abilities when learning reading, the results suggest that phonemic measurements are those that most contribute to word recognition and that, concerning syntactic awareness, there is a direct relation between syntactic mastery and comprehension skills, since children must be aware of the way sentences are organized to be able to read and understand them successfully. The study even predicted that some difficulties in reading comprehension may be assigned to specific problems in the explicit use of syntactic knowledge.

Joana Batalha

205

For Portuguese, Rego (1993) investigated the role of syntactic awareness in a study with Brazilian children (average age between 5;6 years and 7 years at each testing moment), with one of the objectives being to confirm if the role played by syntactic awareness would be equally relevant in languages with a less irregular spelling than English. Results indicated that children’s syntactic abilities were a good predictor of their performance not only in reading comprehension tasks, but also in reading meaningless words, which requires a better knowledge of spelling. In a more recent research project, in a study involving 65 primary school students of Brazilian Portuguese speakers (average age 7;3 years), Barrera and Maluf (2003) tried to assess the relations between phonological, lexical and syntactic awareness and learning reading. Results, achieved through the application of phonological, lexical and syntactic awareness tasks and word writing and reading assessment, suggest that the ability to pay attention to syntactic sentence organization may play a facilitating role in the literacy process, mostly in relation to reading. This supports the hypothesis which claims that syntactic awareness would be a predictor of a more successful learning of reading. In turn, Capovilla, Capovilla and Soares (2004) examined correlations between syntactic awareness and phonological awareness, vocabulary, reading and writing in a study that covered 204 first to fourth grade subjects. The authors applied a phonological awareness test using pictures, a syntactic awareness test with different tasks, a silent reading test and a writing subtest using dictation, having gathered evidence favoring the correlation between the two aspects of linguistic awareness tested and reading and writing. The authors concluded that syntactic awareness and reading have a causal and reciprocal relation, that is, as a predictor of subsequent success in reading, syntactic awareness is also developed with the introduction of the writing system. With regard to European Portuguese, although research into syntactic awareness is somewhat limited, of note are the studies developed by SimSim (1997), as well as the recent research by Costa, M. (2010) and Alexandre (2010). These studies focus on the assessment of syntactic awareness in pre-school and/or primary school children and provide relevant data for a characterization of syntactic awareness development in these age groups, even though they do not establish a direct relation with reading. Sim-Sim (1988) is one of the few studies, within the national context, that tried to investigate the relation between linguistic awareness (phonological and syntactic) and reading in its initial stage. The study involved 56 subjects, organized into four groups, two groups

206

Relations between Grammar and Reading

(experimental and control) of Cape Verdean children and two groups (experimental and control) of Portuguese children. To study possible relations between linguistic awareness and reading, a test of metalinguistic abilities (which assessed segmentation ability, as well as the ability to detect, correct and explain ungrammatical sentences) and a reading test (global word recognition, reading of isolated words, simple sentences and small texts whose comprehension required inference) were applied. The results obtained point clearly to a positive relation between both tests, with the relation between segmentation/reading being more direct than that between ungrammatical sentences/reading, as children unable to segment based on the phoneme did not go beyond global word recognition in reading. Most of the studies that correlate syntactic awareness and reading have tried, as we have seen, to determine in what way syntactic awareness may have an effect on initial reading stages, as well as other types of linguistic awareness, such as phonological awareness. Gaux and Gombert (1999) suggest, however, that the contribution of syntax in reading, in particular in comprehension, becomes independent when reading focuses more on sentence and text comprehension than on word identification, which is why studies that investigate the role of syntactic awareness in reading comprehension seek to study more advanced readers, particularly preadolescents, and different units in meaning construction (sentences and texts). Reading is, by definition, extracting meaning from what is read, and thus we cannot talk about successful reading if there is no comprehension. We can understand reading comprehension as the process of simultaneously extracting and building meaning through interaction and involvement with written language, identifying three elements in this process – the reader, the text and the reading activity – that interact among each other and that act in a determined context (Snow 2002, 11). It could be possible to think that, when the child learns how to decode in an effective way, comprehension should follow automatically. In fact, we know that if decoding is difficult, the processing system is overloaded and there will not be cognitive resources left for comprehension (Viana et al. 2010). It is also known that, as they progress in reading, children dedicate less time to decoding and have more time for comprehension. However, there is no evidence to support a causal relation between decoding and comprehension, since successful decoding children may not be necessarily successful comprehenders (Oakhill and Cain 2004). Hence, results are convergent showing that decoding and a certain degree of success in word identification are, in fact, necessary for reading

Joana Batalha

207

comprehension, but that is not enough. Besides decoding, reading requires a control of the processing that lead to comprehension, for it is not automatic (Gombert 2003). Comprehension occurs as the reader builds a mental representation of the text message, according to the processing of different levels of information. So that good levels of comprehension can be obtained, it is necessary that the subject takes possession of the text’s basic information, of its strict meaning. Lexical and syntactic levels provide base information for the construction of this mental textual representation. When the new information extracted from the text interacts with the reader’s knowledge and experience, the reader is able to elaborate an individual text representation (Costa 1992). Comprehension thus involves a construction stage, in which a mental model is built locally and gradually, and an integration stage, that occurs every time new information is added. The reader integrates information conveyed in the text, producing a mental representation named text-base, and he/she also integrates that information in his/her knowledge of the world, deriving a situational model (Kintsch 1998, quoted by Spinillo 2008). A good level of reading comprehension results, thereby, from the interaction of diverse linguistic and cognitive processes, which should take form as intervention fields in the teaching of reading: automatic recognition of words and access to their meaning (lexical processing on the basis of phonological, morphological and semantic operations); understanding of grammatical and semantic relations (syntactic parsing and proposition formation); integration of textual ideas through operations supported by memory; inferences; use of world knowledge; and the use of reading strategies (Costa 1998). In an attempt to provide theoretical information for the understanding of the reading act, different reading models, named metaphorical reading models (Grabe and Stoller 2002), appeared: ascendant models, descendant models and interactive models. These models distinguish themselves, basically, by the way they predict the relation between subject and text, the type of strategies used and the role played by the reading material as a basis for the construction of new knowledge (Costa 1992). Other models, named specific reading models, according to Grabe and Stoller (2002), have attempted to summarize the most recent research into reading comprehension. In conformity with one of these models, the simple view of reading model (Gough and Tunmer 1986, Hoover and Gough 1990, quoted by Adlof, Perfetti and Catts 2011), reading comprehension is conceived as a product of reading words and understanding language, the first involving

208

Relations between Grammar and Reading

decoding and the second involving giving meaning to decoded words. This model assumes that both components are necessary to reading, but they can also be dissociated, since a different level of competence can be achieved in each one of them. Adlof, Perfetti and Catts (2011) distinguish four readers’ profiles from the interaction between reading words and understanding language: good readers and readers with difficulties of three kinds - in decoding, in comprehension or in both. By comparing good and bad readers as far as decoding is concerned with its level of syntactic awareness, Bowey (1986) intended to show not only that good readers possess higher levels of syntactic awareness when compared to bad readers, but also that it is possible to establish an association between syntactic awareness and reading comprehension monitoring. The results, obtained from the application of oral tasks of syntactic awareness assessment and oral reading to fourth and fifth grade children, show that good and bad readers differ significantly in syntactic awareness regarding word decoding. In addition, when reading errors made by subjects are analysed, bad readers make more grammatically unacceptable errors (comprehension), as well as making fewer selfcorrections based on grammatical criteria (comprehension monitoring). Thus, the author obtained evidence to support the hypothesis that syntactic awareness may play a facilitation role in reading comprehension, influencing comprehension monitoring. Bentin, Deutsch and Liberman (1990) also identified a relation between syntactic knowledge and comprehension, based on a study with three groups of children showing different levels in their reading skills: a group of children with reading disorders (average age of 11;6 years), a group of good readers (9;3 years) and a group of poor readers (9;1 years). A group of university students formed the control group. The authors assessed the ability to use syntactic awareness to identify words in an oral task, as well as the ability to detect and correct violations to the syntactic structure. Results showed that, for the studied language, Hebrew, syntactic factors are directly related to reading disorders, since the group of children with disorders, contrary to the rest of the groups, showed little difference when identifying words in grammatical and ungrammatical sentences, and they also showed difficulty in detecting violations in sentences. The ability to detect errors and, mostly, to make corrections, was the main difference found between the good reader group and the poor reader group, which suggest that good readers show more ability to use syntactic knowledge in a productive way. From these and other studies (Yuill and Oakhill 1988; Tunmer 1989; Rego 1991; Roazzi and Carvalho 1997, quoted by Leal and Roazzi 1999)

Joana Batalha

209

that show positive effects of syntactic awareness in reading comprehension, Leal and Roazzi (1999) attempted to understand how and to what level syntactic (and semantic) awareness influences text reading, in particular i) if decoding is enough to increment reading comprehension, ii) if syntactic awareness influences reading development and iii) if the effect of syntactic awareness on sentence comprehension is similar to the effect on text comprehension. The study included a group of 112 subjects with average age of 8;2 years and used a reading and sentence comprehension task, a reading and text comprehension task and syntactic awareness tasks. Results suggest that i) although there should be a minimum level in decoding as a prerequisite for comprehension, decoding is not enough for reading comprehension, ii) there is strong evidence of a relation between text reading comprehension and syntactic and semantic awareness and iii) the effect of syntactic and semantic awareness is stronger on text comprehension than on sentence comprehension. Based on these results, the authors argue in favour of a teaching where skills related to the use of grammatical clues during reading should be developed, as these skills are strongly related to comprehension, as well as recommend the early introduction of text reading and the promotion of analytical activities and the comprehension of significant texts from the first school years. Gaux and Gombert (1999) carried out a study involving 83 subjects with average age of 12;3 years. Seven tasks for assessing syntactic knowledge were applied, taking into account the distinction between implicit and explicit knowledge (repetition, judgment, localization, correction, explanation, replication and class and function identification), and five reading comprehension tasks (a task involving reading aloud and four distinct tasks for reading words and pseudo words). The authors obtained evidence that pre-adolescents are able to pay deliberate attention to syntax. Contributions were more frequent when the task required explicit knowledge, which suggests that this type of knowledge is involved in reading and in access to comprehension. When the participants according to their results in reading comprehension were grouped, it was noticed that pre-adolescents with low comprehension skills showed a deficit in syntactic awareness when compared to good comprehenders, showing poor results mostly in tasks that required explicit knowledge. Moreover, the results highlighted a specific relation between word order inversion and comprehension, which shows the importance of syntactic awareness in syntactic processing. Nation and Snowling (2000), in an attempt to avoid confusing difficulties in decoding with difficulties in reading comprehension, examined syntactic awareness in two groups of children with the same

210

Relations between Grammar and Reading

level of decoding (as well as verbal memory and age), but with different levels of comprehension. To achieve this, they used tasks for word order correction with variable syntactic complexity (with active and passive sentences) and tasks involving semantic ambiguity with reversible and irreversible sentences, in order to compare syntactic awareness in children with reading and comprehension deficit and syntactic awareness in children with normal reading development. The results for the first task showed that all the subjects had more difficulties with passive sentences, especially bad comprehenders, which suggests that performance in word order correction is sensitive to syntactic complexity. In line with the results of the second task, all children showed more difficulties with reversible sentences than irreversible sentences. Children with a comprehension deficit demonstrated, in a clear manner, poorer skills in syntactic awareness, which suggests a developmental delay in syntactic awareness in relation to normal readers. The authors propose that these difficulties in syntactic awareness may be a demonstration of general difficulties in language processing. Mokhtari and Thompson (2006) examined the role of syntactic awareness as a potential factor for reading difficulties among fifth grade readers, obtaining evidence to validate the hypothesis that i) the level of syntactic awareness is associated in a significant way with reading fluency (r=.625) and reading comprehension (r=.816) and ii) low levels of syntactic awareness are associated with reading difficulties. The authors consider that the role of syntactic awareness in fluency and comprehension has been undervalued, unlike decoding and phonemic skills, and that this area needs further investigation. In addition, regarding teaching implications, the authors emphasize the need to focus language teaching on syntactic awareness development in order to influence reading performances of students with different levels and reading experiences. For European Portuguese, Armanda Costa’s studies on reading and syntactic processing have given important clues for teaching concerning the relation between metalinguistic knowledge and reading comprehension. In particular, Costa (1991) claimed that the syntactic level is always used as an auxiliary source for the language processor in comprehension, since parsing is automatic, and the use of the syntactic level, which may be more or less intense according to the presence or not of semantic and pragmatic sources, increases with age and instruction, with grammatical knowledge and with the reader’s experience. The author applied three tests to two groups of subjects, who were seventh and eleventh graders: a reading aloud test, composed of the reading of three texts in which changes causing ungrammaticality were introduced, in order

Joana Batalha

211

to test if degradation affects syntactic level processing; a comprehension test, applied to confirm the reader’s objectives in oral reading compared to comprehension of text contents, as well as to gather indicators about the comprehension level obtained and the degradation effect; finally, a grammar test, with assessment and grammaticality explanation tasks, aiming at collecting data about subjects’ metalinguistic knowledge of the studied syntactic structures. The obtained results in comprehension and grammar tests were positively correlated, confirming the assumption that reading comprehension is related to strategic behavior that makes use of additional sources when there are processing problems, i.e., metalinguistic knowledge. The relation between comprehension and grammatical knowledge was more evident in eleventh grade subjects, who revealed more successful skills in detecting ungrammaticality and, chiefly, in resolving this, as well as better comprehension levels.

3. Research problem and methodology Taking into account the current context of mother tongue teaching, grammar, named explicit knowledge of language, assumes a double dimension, as a study object per se and as a skill to use in oral language, reading and writing. However, according to recent studies (Duarte, R. 2008; Costa, A.L. 2010), teachers do not seem to acknowledge the utility of this skill and a set of divergent teaching practices coexist in schools, which may explain students’ poor results when performing tasks that require grammatical knowledge. On the other hand, if the studies mentioned in the previous section provide important data concerning the role of linguistic awareness (and, in particular, syntactic awareness) in reading, they also show that this relation still needs further research, especially regarding reading comprehension development. Moreover, research in this area in the Portuguese national context is still insufficient, mainly within a perspective of its application to mother tongue teaching. Hence, and assuming that: i) to obtain high levels of reading performance, an extensive and profound knowledge is necessary which, to a large extent, has to be explicit; ii) complex stages of explicit knowledge cannot be achieved without formal instruction, the role of school should, particularly in the mother tongue subject, to promote linguistic awareness in order to

212

Relations between Grammar and Reading

develop children’s implicit knowledge of their language to a level of explicit knowledge; iii) syntactic awareness plays an important role in reading, not only in the initial stages of its learning, but also in later stages, acting at the level of comprehension; iv) the inability to deal with complex syntactic patterns, which require an explicit knowledge of syntactic structures, affects reading comprehension, with subjects with more instruction and grammatical knowledge being those who are more able to solve syntactic problems; it is intended to investigate the relations between explicit knowledge of language and reading in the mother tongue teaching context. In particular, an attempt will be made to find out to what extent explicit work on syntactic structures which cause difficulties can improve reading performance and, by contrast, what kind of activities can be developed in the classroom to reinforce the relation between explicit knowledge and reading. Our research will focus on potentially complex structures that resist automatic processing and that increase processing costs, such as certain types of anaphor pronouns. In line with Costa and Luegi (2009), linguistic complexity may be associated with different factors (lexical, grammatical, discursive, rhetorical, written conventions, among others), with grammatical factors, which relate to the syntactic properties of structures, being those which are more interesting for this study. Having as reference non marked basic structures, according to these authors, complexity indicators can be taken for any syntactic properties that result from the alteration of basic patterns in word order, sentence linking processes, or from the existence of long distance dependencies. Structural ambiguity, which puts the parser in a situation to take decisions about competing structures, is also a factor that contributes to complexity. Given that the objective is to test the possible benefits of explicit knowledge in reading in a teaching context, it is assumed, in accordance with the recommendations of the Guião de Implementação do Programa – Conhecimento Explícito da Língua, that the teacher is entrusted with the goal “of enabling students to be aware of the structures that are naturally more difficult to persuade them to carry out a more attentive reading and to the development of strategies for problem solving, drawing their attention (that is, guiding reading) to structural aspects that cause difficulties” (Costa et. al.. 2010, 19). Thus, following the discovery-learning activity method (Hudson 1992), a set of activities is proposed, which will

Joana Batalha

213

allow students to use grammatical knowledge to solve problems raised by potentially complex structures during reading. According to Costa (2011), there are three principles underlying this type of activity: i) to guide readers in the interpretation of more difficult areas as a result of processing and development; ii) to make readers aware of potential sources of difficulty; and iii) to allow readers to have at their disposal instruments that enable reading to become faster and more efficient. The study that it is intended to carry out will assume characteristics of an experimental study and will include two groups of subjects, one experimental group and one control group, both composed of lower secondary education students. The experimental group will be subject to an explicit teaching of the syntactic structures being studied, which will not take place in the second group. In this way, in assessing the reading performance of both groups, it will be possible to establish a comparison between the experimental group and the control group, in which there will not be any controlled intervention by the research project. The didactic activity will be organized in three distinct activities: a reading activity prior to the explicit teaching, explicit teaching activities of the syntactic structures being studied, and a reading activity subsequent to the explicit teaching. The experimental study will be preceded by a pre-test, with different groups of students, in order to enable materials to be tested and validated. It is important to mention that, for this experimental study, materials for the activities will be designed not only to meet the main aim of the research, i.e. to look for relations between grammar and reading, but also to meet a pedagogical aim, which is that of creating activities that can be used in the classroom by any teacher.

4. Anaphor pronouns and reading: some teaching activities in lower secondary education As was stated previously, reading comprehension is a complex phenomenon, which involves different abilities: to understand a text, a reader has to carry out different operations, from access to the meaning of each word to the construction of a textual representation, including sentence processing and its integration within meaning units, as well as interaction with his/her knowledge of the world and his/her experience as a reader (Costa 1992). In particular, sentence comprehension seems to play an important role in achieving successful comprehension, since it provides, together with the lexical level, base information for the construction of a textual

214

Relations between Grammar and Reading

representation. At this level, the comprehension of anaphor pronouns, namely the ability to identify their antecedents, has been pointed out as one of the factors that separate good from bad comprehenders (Cain and Oakhill 2009). In fact, studies on language processing show that referential constructions, due to a variety of factors that include ambiguity conditions in which several entities appear as potential antecedents, such as the nature of the anaphoric expression, the variation of the distance between the antecedent and the anaphor and the syntactic properties of the material inserted between antecedent and pronoun, may represent complex constructions and, consequently, cause an increase in processing costs (Costa and Luegi 2009). On the other hand, in mother tongue teaching, it is known that the identification of pronoun antecedents continues to be a difficult area in lower secondary education and that it may compromise access to text meaning, especially when referential chains are established (Duarte 2002; Santos 2002). Taking this into account, and as a pre-test for the experimental study that it is intended to carry out, two lower secondary education classes (seventh and ninth grades), forming a total of 46 students, in a classroom situation with their Portuguese teachers, received explicit teaching on some anaphor pronouns. These pronouns, due to their morphosyntactic properties and/or the type of antecedent that they refer to, are potentially complex for students. It should be noted that this was a pre-test from a set of didactic materials, in which testing conditions such as time spent on doing the activities, the students’ school profile or the teachers’ profile (qualifications, experience or teaching practices) were not controlled. Given this, the following objectives were established for this pre-test: i) to carry out initial validation of the didactic materials (suitability for the subject syllabus and the year of schooling; organization, accuracy and functionality of the materials; student response to the proposed activities); ii) to obtain data about the level of difficulty of the structures being studied (anaphor pronouns). In the didactic materials, using interphrasal contexts, work on the morphosyntactic and semantic properties of the anaphor pronouns being studied was proposed, as well as the type of antecedent they refer to (noun or clause) and also the antecedent position regarding the pronoun. After the explicit teaching, a reading activity was suggested, in order to test anaphor pronoun comprehension, through the identification of their

Joana Batalha

215

antecedents. As experimental material, a non-literary text was used, of an informative nature, that was previously manipulated in order to include the anaphor pronouns being tested, although other factors besides the anaphor pronoun and the type and position of the antecedent referred to were not controlled. Thus, there were four conditions created in the text, to form a total of eight anaphor pronouns (involving third person personal pronouns and demonstrative pronouns): i) recovery of a nominal antecedent previously introduced in the text by a personal pronoun marked for gender and number (ele, o) and marked for number (lhe), as well as demonstrative pronouns marked for gender and number, but where there is copresent pronominalization (este, aquele); ii) recovery of a clausal antecedent previously introduced in the text by an invariable demonstrative pronoun (isso); iii) recovery of a nominal antecedent introduced subsequently in the text by a demonstrative pronoun marked for gender and number (estas); iv) recovery of a clausal antecedent introduced subsequently in the text by an invariable demonstrative pronoun (o). To resolve the reading activity, students were expected to use strategies for the identification of pronoun antecedents derived from the explicit teaching they had received. The proposed strategies aimed at guiding students, on the one hand, to verify the morphosyntactic and semantic pronoun compatibility with a potential antecedent and, in addition, to identify the possible positions of the antecedent as regards the pronoun. Two tests for the confirmation of the identified antecedent were also provided: a substitution test and a question/answer test, the answer for which will include the identified antecedent. Although this experiment was a pre-test, from which it is not intended to draw conclusions, it was possible to obtain some important clues for our future work, namely concerning the second established goal – to obtain data about the level of difficulty of the structures being studied (anaphor pronouns).

Relations between Grammar and Reading

216

100 80 60 40 20 0

Figure 1: Percentage of correct answers in the reading activity

An analysis of the 46 student answers to the reading activity items, corresponding to the eight anaphor pronouns whose antecedent students had to identify (figure 1), allows the identification of the pronouns which caused more difficulties for students. Taking into account the conditions created in the text, it can be observed that condition (i), with personal and demonstrative pronouns morphologically marked that recover nominal groups in anaphoric position, was the one that, as expected, obtained a higher percentage of correct answers. To be highlighted, on the one hand, is the high percentage of correct answers with the pronouns este and aquele in the copresent pronominalization structure, since it could be expected that this structure would cause more difficulties; on the other hand, it would be interesting to try and understand better the 60% obtained for the personal pronoun o. Being a morphologically marked pronoun and a pronoun that recovers a nominal group, 40% of incorrect answers is not to be expected. This might be explained by the fact that the correct antecedent was not the later nominal group morphosyntactically and semantically compatible with the pronoun, since another nominal group equally compatible was closer to it. The answers concerning conditions (ii) to (iv) resulted in each case in correct answers lower than 50%, with the lowest percentage obtained where the antecedents are in a cataphoric position. The neuter pronoun form and the type of antecedent recovered (nominal or clausal) also seem to be significant factors for the results obtained. The lower percentage of correct answers, closer to 0%, was found with the unstressed form of the

Joana Batalha

217

demonstrative pronoun o, which, besides being a neuter pronoun, recovers a clausal antecedent in cataphoric position in the text.

5. Conclusion This article has presented a research project on language and mother tongue teaching. It is intended to study the relations between grammar and reading, that is, the benefits that explicit teaching of some syntactic structures, including anaphor pronouns, may bring for an improvement of reading comprehension in the mother tongue. As discussed, in recent years, the context of the teaching of Portuguese as mother tongue has seen contradictory curricular indications that have legitimized divergent teaching practices. However, curriculum documents currently in force appear to have recognized the importance of grammar and have considered the explicit knowledge of language a core skill, alongside the skills of language use. Moreover, several studies in the field of linguistics, cognitive psychology and education have shown that to achieve good reading performance levels, extensive and profound knowledge is necessary which, to a large extent, has to be explicit. As far as is known, in the Portuguese context and from a mother tongue teaching perspective, there is limited or no research focusing on the relation between linguistic awareness/explicit knowledge and reading, in particular concerning reading comprehension. Therefore, by identifying and intervening in areas known to be difficult areas for students, such as anaphor pronouns, it is expected to contribute to a field that needs further research and also scientifically validated didactic materials. In particular, after obtaining results from the experimental study, it is the aim of this author to make the materials used in the research available for teachers.

References Adlof, Suzanne, Charles Perfetti, and Hugh Catts. 2011. “Developmental changes in reading comprehension: implications for assessment and instruction”. In What research has to say about reading instruction (4th edition), edited by S. Jay Samuels, and Alan E. Farstrup, 186-214. Newark, DE: International Reading Association. Alexandre, Rita. 2010. “A tarefa de manipulação na avaliação da consciência sintática em crianças do 1.º ciclo de escolaridade”. In Avaliação da Consciência Linguística. Aspetos fonológicos e sintáticos do português, coordinated by Maria João Freitas, Anabela Gonçalves, and Inês Duarte. Lisboa: Edições Colibri.

218

Relations between Grammar and Reading

Barrera, Sylvia, and Maria Regina Maluf. 2003. “Consciência metalinguística e alfabetização: um estudo com crianças da primeira série do ensino fundamental”. Psicologia: Reflexão e Crítica 16(3): 491-502. Bentin, Shlomo, Avital Deutsch, and Isabelle Liberman. 1990. “Syntactic competence and reading ability in children”. Journal of Experimental Child Psychology 49(1):147-172. Bowey, Judith. 1986. “Syntactic awareness in relation to reading skill and ongoing reading comprehension monitoring”. Journal of Experimental Child Psychology 41(2), 282-299. Cain, Kate. 2007. “Syntactic awareness and reading ability: Is there any evidence for a special relationship?” Applied Psycholinguistics 28(04): 679-694. Cain, Kate, and Jane Oakhill. 2009. “Reading Comprehension Development from 8 to 14 years. The contribution of component skills and processes”. In Beyond Decoding. The Behavioral and Biological Foundations of Reading Comprehension, edited by Richard Wagner, Christopher Schatschneider, and Caroline Phythian-Sence. New York: The Guildford Press. Capovilla, Alessandra, Fernando Capovilla, and Jocelli Soares. 2004. “Consciência sintática no ensino fundamental: correlações com consciência fonológica, vocabulário, leitura e escrita”. Psico-USF 9(1):39-47. Correa, Jane. 2004. “A avaliação da consciência sintática na criança: uma análise metodológica.” Psicologia: Teoria e Pesquisa 20(1):69-75. Costa, Armanda. 1991. “Leitura: Compreensão e Processamento Sintático”. MA Diss, Universidade de Lisboa. —. 1992. “Leitura: conhecimento linguístico e compreensão”. In Maria Raquel Delgado Martins, Dília Pereira, Ana I. Mata, Armanda Costa, Luís Prista, and Inês Duarte. Para a didática do português. Seis estudos de linguística, 75-117. Lisboa: Edições Colibri. —. 1998. “Saber ler e saber ensinar a ler. Do Básico ao Secundário”. In Rui Vieira de Castro, and Maria de Lurdes Sousa, orgs., Linguística e educação, 69-82. Lisboa: APL-Edições Colibri. Costa, Armanda, and Paula Luegi. 2009. “Complexidade linguística e processamento referencial”. In Dermeval da Hora, org., Anais do VI Congresso Internacional da ABRALIN. CD-ROM, 2127-2135. Costa, Ana Luísa. 2010. “Estruturas contrastivas: desenvolvimento do conhecimento explícito e da competência de escrita”. PhD diss., Universidade de Lisboa.

Joana Batalha

219

Costa, Magda. 2010. “A tarefa de reconstituição na avaliação da consciência sintática em crianças de 1.º ciclo de escolaridade”. In Avaliação da Consciência Linguística. Aspetos fonológicos e sintáticos do português, coordinated by Maria João Freitas, Anabela Gonçalves, and Inês Duarte. Lisboa: Edições Colibri. Costa, João, Assunção C. Cabral, Ana Santiago, and Filomena Viegas. 2010. Conhecimento Explícito da Língua. Guião de Implementação do Programa. Direção Geral de Inovação e Desenvolvimento Curricular, Ministério da Educação. Costa, João. 2011. Relações entre consciência e explicitação linguística e o desenvolvimento da competência de leitura. Comunicação apresentada na V Conferência Internacional do Plano Nacional de Leitura na Fundação Calouste Gulbenkian. Demont, Elisabeth, and Jean Émile Gombert. 1996. “Phonological awareness as a predictor of recoding skills and syntactic awareness as a predictor of comprehension skills”. British Journal of Educational Psychology 66(3), 315-332. Duarte, Inês. 2002. “Complexidade sintática: implicações no ensino da língua materna”. In Cristina Mello, Isabel Pereira, Maria Helena Santana, Maria José Carvalho, and Filomena Brito, orgs., II Jornadas científico-pedagógicas de Português. Coimbra: Almedina. —. 2008. O Conhecimento da Língua: Desenvolver a Consciência Linguística. Lisboa: PNEP, Direção Geral de Inovação e Desenvolvimento Curricular, Ministério da Educação. —. 2010. “Sobre o conceito de consciência linguística”. In Avaliação da Consciência Linguística. Aspetos fonológicos e sintáticos do português, coordinated by Maria João Freitas, Anabela Gonçalves, and Inês Duarte. Lisboa: Edições Colibri. Duarte, Regina. coord. 2008. Posição dos docentes relativamente ao ensino da Língua Portuguesa. Direção Geral de Inovação e Desenvolvimento Curricular, Ministério da Educação. Gaux, Christine, and Jean Émile Gombert. 1999. “Implicit and explicit syntactic knowledge and reading in pre-adolescents”. British Journal of Developmental Psychology 17:169-188. GAVE. 2010. Relatório. Provas de aferição de Língua Portuguesa – 2.º ciclo. Gabinete de Avaliação Educacional, Ministério da Educação. Gombert, Jean Émile. 1990. Le développement métalinguistique. Paris: PUF. —. 2003. “Atividades metalinguísticas e aprendizagem da leitura”. In Maria Regina Maluf, org., Metalinguagem e aquisição da escrita:

220

Relations between Grammar and Reading

contribuições da pesquisa para a prática da alfabetização. São Paulo: Casa do Psicólogo. Grabe, William, and Fredricka Stoller. 2002. Teaching and Researching Reading. GB: Pearson Education. Guimarães, Sandra. 2003. “Dificuldades no desenvolvimento da lectoescrita: o papel das habilidades metalinguísticas”. Psicologia: Teoria e Pesquisa 19(1): 33-45. Hudson, Richard. 1992. Teaching grammar. A teacher’s guide to the National Curriculum. Oxford: Blackwell. Karanth, Pratibha, Kudva Asha, and Aparna Vijayan. 1995. “Literacy and linguistic awareness”. In Speech and reading, edited by Beatrice De Gelder, and José Morais. UK: Erlbaum Taylor e Francis Publishers. Kavanagh, James, and Ignatius Mattingly, eds., 1972. Language by Ear and by Eye. Cambridge: The MIT Press. Leal, Telma Ferraz, and Antonio Roazzi. 1999. “Uso de pistas linguísticas na leitura: análise do efeito da consciência sintático-semântica sobre a compreensão de textos”. Revista Portuguesa de Educação 12(2), 77104. Mokhtari, Kouider, and H. Brian Thompson. 2006. “How problems of reading fluency and comprehension are related to difficulties in syntactic awareness skills among fifth graders”. Reading Research and Instruction 46(1): 73-94. Nation, Kate, and Margaret Snowling. 2000. “Factors influencing syntactic awareness skills in normal readers and poor comprehenders”. Applied Psycholinguistics 21(2): 229-241. Oakhill, Jane, and Kate Cain. 2004. “The development of comprehension skills”. In Handbook of Children’s Literacy, edited by Teresinha Nunes, and Peter Bryant, 155-180. Dordrecht: Kluwer Academic Publishers. Ravid, Dorit, and Liliana Tolchinsky. 2002. “Developing linguistic literacy: a comprehensive model”. Journal of child language 29: 419448. Rego, Lucia. 1993. “O papel da consciência sintática na aquisição da língua escrita”. Temas em Psicologia 1:79-87. Rego, Lucia, and Peter Bryant. 1993. “The connection between phonological, syntactic and semantic skills and children’s reading and spelling”. European Journal of Psychology of Education 8(3): 235246. Reis, Carlos coord. 2009. Programa de Português para o Ensino Básico. Direção Geral de Inovação e Desenvolvimento Curricular, Ministério da Educação.

Joana Batalha

221

Santos, Fátima. 2002. “Os pronomes pessoais átonos no português europeu. Descrição de problemas que ocorrem no 3.º ciclo e proposta de atividades didácticas”. MA Diss, Universidade de Lisboa. Sim-Sim, Inês. 1988. “Consciência linguística e nível de leitura: que relação? Ou ler ou não ler... Eis a questão”. Revista Portuguesa de Educação 1(1):95-102. —. 1997. Avaliação da linguagem oral: um contributo para o conhecimento linguístico das crianças portuguesas. Lisboa: Fundação Calouste Gulbenkian. —. 1998. Desenvolvimento da Linguagem. Lisboa: Universidade Aberta. Sim-Sim, Inês, Inês Duarte, and Maria José Ferraz. 1997. A Língua Materna na Educação Básica. Lisboa: Departamento de Educação Básica, Ministério da Educação. Snow, Catherine. ed. 2002. “Reading for understanding: toward a research and development program in reading comprehension”. RAND Reading Study Group. Arlington, VA: Rand Education. Sousa, Hélder coord. 2011. Exames Nacionais. Relatório 2011. Gabinete de Avaliação Educacional, Ministério da Educação. Spinillo, Alina Galvão. 2008. “O leitor e o texto: desenvolvendo a compreensão de textos na sala de aula”. Journal of Psychology, 42:2940. Titone, Renzo. 1988. “A crucial psycholinguistic prerequisite for reading: children’s metalinguistic awareness”. Revista Portuguesa de Educação 1(2): 61-72. Tunmer, William, and Hoover Wesley. 1992. “Cognitive and linguistic factors in learning to read”. In Reading acquisition, edited by Philip Gough, Linnea Ehri, and Rebecca Treiman. Hillsdale, NJ: Erlbaum. Tunmer, William, Andrew R. Nesdale and A. Douglas Wright. 1987. “Syntactic awareness and reading acquisition”. British Journal of Developmental Psychology 5:25-34. Viana, Fernanda Leopoldina, Iolanda Ribeiro, Irene Cadime, Ilda Fernandes, Albertina Ferreira, Catarina Leitão, Susana Gomes, Soraia Mendonça, and Lúcia Pereira. 2010. O Ensino da Compreensão Leitora. Da Teoria à Prática Pedagógica. Um Programa de Intervenção para o 2.º Ciclo do Ensino Básico. Coimbra: Almedina.

CLOSING TEXT

LANGUAGE IN ACTION: EPISTEMOLOGICAL AND METHODOLOGICAL ISSUES1 MARIA ANTÓNIA COUTINHO CENTRO DE LINGUÍSTICA DA UNIVERSIDADE NOVA DE LISBOA FACULDADE DE CIÊNCIAS SOCIAIS E HUMANAS/UNIVERSIDADE NOVA DE LISBOA

Abstract During the middle of the 20th century, linguistics was referenced as a pilot case among human sciences. It obtained this position because of its effort to distinguish itself from other areas of knowledge that shared its curiosity about natural languages and, above all, because of its capacity to, at the same time, delimit and construct its own object of study. The clarity of this delimitation seemed to be based upon the formula that for a long time had been attributed to Ferdinand de Saussure (even though it had already long been known that it did not actually belong to him2): “the true and unique object of linguistics is language studied in and for itself” (Saussure 1959, 232). Saussure’s texts, which were in the meanwhile discovered, clearly demonstrate a different vision – showing in a very unequivocal way the separation between language and discourse: “Langue is created only with a view to discourse; but what separates discourse from the language system, and what allows us to say that a language system enters into action as discourse at any given moment?” (Saussure 2006, 197). The present study speaks about language in action. It will start by presenting the general epistemological principles which characterize the social interactionism framework (shared by Saussure) and it will then 1

This work was supported by the National Funding Organisation – FCT in the Strategic Project PEst-OE/LIN/UI3213/2011. 2 Cf. Tulio de Mauro, 1972.

Maria Antónia Coutinho

225

enunciate the methodological guidelines derived from this, as well as the methodological pertinence regarding the analysis of a singular text. The analytical work will lead to a focus on the importance of language in action as the true object of linguistic analysis.

1. The epistemological framework of Social Interactionism Social interactionism is recognized today as an epistemological current considered by various intellectuals and scientists at the beginning of the 20th century, a current sustained by a number of authors, in an explicit and engaged way. At the centre of the interactionist social framework, there is a project for a science of the human, a reunited science, as an alternative to the imposed discipline inherited by positivist rationalism. If the ability to account for the complexity of human beings can only be hoped for from a (re)united science, as an object of study, such science will also have to be, necessarily, a science that permanently seeks itself and/or reconstitutes itself. As Bronckart explains (2004, 114), the issue concerns:   

a science (permanently) supported by epistemological reflection (susceptible of rethinking models and data); a science of intervention (capable of measuring the validity of knowledge that it produces by the capacity of transformation or of practical operationalization); a natural science, which rejects the opposition between natural sciences and human sciences, according to the adherence to the principles of materialist monism and of psychophysiological parallelism, inherited from Spinoza3.

The contributions of Vygotsky ([1934]1997) and of Voloshinov ([1929] 1973) are fundamental references within the social interactionist framework. Furthermore, it is important to underline the point of view Vygotsky assumes, regarding the origin of conscious thinking (or of higher psychological functions). In a paper where he systematically deals with the epistemologies of Piaget and Vygotsky, Bronckart ([1999]2002,42) stresses that the former extends the Cartesian and Kantian 3

Materialistic monism implies that everything is a unique substance, material in movement, with a physical dimension and a psychic dimension (the former inscribed in space and directly observable, unlike the latter). Psychophysiological parallelism brings these two aspects together, assuming that any entity is endowed with an equivalent complexity, from the physical and psychic point of view.

226

Language in Action

inheritance, assuming cognitive development as a condition for the appearance of language and of collective activities, while the latter assumes the positions of Spinoza, Marx and Engels, defending a double origin – biological and sociosemiotical – of conscious thinking in the human species. Vygotsky’s perspective is determinant in the social interactionist framework: assuming that it is the internalization and the appropriation of signs that makes thinking and conscience possible, he shows, at the same time, that the possibility of thinking is indistinguishable from the process of semiotization. That was also Saussure’s point of view, who clearly rebuked the separation between thought, as something previous, and language, which would be only a reflection of the former: “It is wrong to believe there may be forms (existing in themselves, independently of their use) or ideas (existing in themselves, independently of their representation).” (Saussure 2006,15). Although specialized bibliography retains Bakhtine’s name, when speaking about the modern sense of the notion of textual genres, here we assume Voloshinov to be the author of reference – which is decisive in this and other matters (for example, the question of dialogism). Though not taking into account ethically shocking details of the process of the appropriation of Voloshinov’s ideas by Bakhtine, as reported in Bronckart and Bota (2011), these authors (Bota and Bronckart 2008) had previously showed the difference between the phenomenological epistemology, that is present in Bakhtine’s texts, and social epistemology, associated with the texts of Voloshinov. Focusing, thus, in this last case, it should be stressed that Voloshinov assumes, in an unequivocal, way, the social nature of an enunciation: “In reality, the speech act or, more accurately, its product – the utterance, cannot under any circumstances be considered an individual phenomenon in the precise meaning of the word and cannot be explained in terms of the individual psychological or psychophysiological conditions of the speaker. The utterance is a social phenomenon.” (Voloshinov [1929]1973, 82). This social conception of language is directly related to the notion of textual genres, associated with all the circumstances of social/collective life that the author describes.

2. Methodological issues The epistemological orientation that has been briefly outlined goes hand in hand with a top-dwn methodological option, which Voloshinov explicitly argues for regarding the study of language. (Voloshinov [1929]1973, 95-96):

Maria Antónia Coutinho

227

[…] it follows that the methodologically based order of study of language ought to be: 1. the forms and types of verbal interaction in connection with their concrete conditions; 2. forms of particular utterances, of particular speech performances, as elements of a closely linked interaction – i.e., the genres of speech performance in human behavior and ideological creativity as determined by verbal interaction; 3. a re-examination, on this new basis, of language forms in their usual linguistic presentation.

It can be said that this perspective is an inversion, when considering other working paradigms within linguistics. As shown in the quoted passage, Voloshinov’s perspective assumes that social interaction is a determining factor, and furthermore, it implies that one should take into consideration the specificities of this same interaction: the context or, more accurately, the social activity where it takes place and the specific genre (among the genres in use, in the activity in question). Analysing linguistic forms where they occur only makes sense when you take into account these specificities, as they are a technical resource for the pursuance of social purposes (related to the activity in question and shaped within the selected or required text genre) and of the objectives of the persons involved (who specifically perform the action of language). It can be said that this top-down methodological orientation is found, roughly, in the different perspectives of text and discourse linguistics – even though it is the framework of socio-discursive interactionism (Bronckart 1997) that, repeatedly reclaims it, in an explicit way, as a condition for its work programme. The Socio-Discursive Interactionism (henceforth SDI) is an epistemological current that shares the principles of social interactionism referred above, outlining the work programme to be developed within a logocentric perspective – i.e., which assumes the role of the language as fundamental in the development of superior psychic functions and, in a more general way, in the development of the individuals during their life. Enshrined, in a privileged way, within the SDI framework, this contribution shares the methodological top-down approach that is being referred to– taking into consideration, however, that it can, in practice, combine with linguistic descriptions available in totally distinct frameworks (as will be seen below). On the other hand, because the analysis that follows concerns a single text, the reliability and profitability that can be associated to this procedure has to be explained and justified, in methodological terms,– obviously, it

228

Language in Action

cannot be expected to have the capacity of generalization. The text analysed in the following pages is not meant to be an example – if example is considered in its capacity to illustrate a theoretical approach and/or a sample of a class of phenomenon. Attributing to this kind of example the function of objectification and legitimating, Ouellet (1989) distinguishes it from the good example (or model) of a certain phenomenon, used as a mechanism of categorization and questioning. Following Passeron and Revel (2005), it can be admitted that the questioning associated to a particular case does not necessarily lead to a linear process of categorization (or of quick regularization); the singularity opens the possibility of reconfiguration of the data and of the way they are thought of and stabilized, as shown in Passeron and Revel (2005, 10-11): [...] un cas n’est pas seulement un fait exceptionnel et dont on se contenterait qu’il le reste: il fait problème; il appelle une solution, c’est-àdire l’instauration d’un cadre nouveau du raisonnement, où le sens de l’exception puisse être, sinon défini par rapport aux règles établies auxquelles il déroge, du moins mis en relation avec d’autres cas, réels ou fictifs, susceptibles de redéfinir avec lui une autre formulation de la normalité et de ses exceptions.4

From this point of view, resorting to a singular case cannot be seen as a methodological handicap: in a strong sense; it is about exploring the (scientific) potentialities resulting from the possibility of observation of that particular case, as well as from the availability of validation criteria, not limited to inductive generalization nor to deductive necessity (Passeron and Revel 2005, 37). It is precisely in this last sense that the text under analysis has been utilized as a singular text, which raises specific issues, i.e., which brings up a relevant issue for research in the field in question (in this case, in text and discourse linguistics).

4

[…] a case is not only an exceptional fact about which we would be happy for it to remain so: it is a problem; it demands a solution, i.e., the establishment of a new framework of reasoning where the sense of the exception would be, if not defined according to the established rules it breaks, at least put in relation with other cases, real or fictional, susceptible of redefining with it another formulation of the normality and of its exceptions.

Maria Antónia Coutinho

229

3. The text under analysis The case under analysis is a text belonging to the genre cartoon, by the Portuguese cartoonist Luís Afonso. It was published for the first time in the Portuguese newspaper Público (January 26, 2003)5, and also published afterwards in an anthology (and reproduced in the Annex). The internal structure of this cartoon uses another genre, conversation, as a process of intertextualization (i.e., as a resource for its own process of textual organization)6. In the context of the present work, this analysis will focus on these conversations or fragments of conversations. It can be said that it is a type of business conversations in a pre-historical context – represented, on the basis of this same context, as simple conversations (or which would nowadays be called informal conversations) and using (European) contemporary Portuguese7, without reconstitution efforts or problems of adjustment. The three segments of conversation in Portuguese, along with their translations, are reproduced below: Dá-me as peles que eu dou-te o machado / Give me the skins and I will give you the ax Dá-me as peles que eu dou-te a lança / Give me the skins and I will give you the spear Dá-me as peles que eu dou-te o arco e as flechas / Give me the skins and I will give you the bow and arrows The three speech turns show a parallel structure, with the occurrence of the connective que. However, the interpretation of each of the cases allows us to make different paraphrases in European Portuguese. They are presented below, numbered randomly, and with their translation:

5

It has been also published afterwards in an anthology: Afonso, Luís (2005). Sociedade Recreativa. Lisboa: Dom Quixote, p. 38. 6 On the notion of intertextualization, cf. Miranda 2010; Coutinho and Miranda 2009. 7 It should be emphasized that the cartoon(ist) does not seek (nor has to pretend) any validation of the representation she or he proposes, with its humoristic / social and politic criticism purposes. For a more detailed description of the textual genre of cartoon, cf. Leal, 2011.

230

Language in Action

Dá-me as peles para eu te dar o machado (/a lança / o machado e as peles) o arco e as flechas?

Give me the skins so that I can give you the ax (/the spear/the ax and the skins)/the spear/ the bow and the arrows?

Dá-me as peles pois / porque eu doute o machado (/a lança / o machado e as peles) o arco e as flechas?

Give me the skins for/because I will give you the ax (/the spear/the ax and the skins)the bow and the arrows?

Dá-me as peles e (depois) eu dou-te o machado (/a lança / o machado e as peles) o arco e as flechas?

Give me the skins and (then) I will give you the ax (/the spear/the ax and the skins)/the bow and the arrows?

Quando me deres as peles, eu dou-te o machado (/a lança / o machado e as peles) o arco e as flechas?

When you give me the skins, I will give you the ax. (/the spear/the ax and the skins)/the bow and the arrows?

Se me deres as peles, eu dou-te o machado (/a lança / o machado e as peles) o arco e as flechas?

If you give me the skins, I will give you the ax (/the spear/the ax and the skins)/the bow and the arrows?

As can be seen, the initial formulation, marked by the connective que, allows the convergence of various values: final, in 1; explanatory/causal in 2; copulative or copulative and temporal in 3; temporal in 4; and conditional in 5. The hypothesis is that there is no ambiguity here but rather a specific possibility of the form in its occurrence, capable of simultaneously conveying the referred values. This simultaneity contributes to the suitable representation of the tone of the conversation – which combines subtly honest business and selfish cleverness. But to what extent is this possibility taken into account in the available descriptions of the form in occurrence (que)? In other words, how is the form under discussion described? This question will be answered by sticking to the contributions from the Gramática da Língua Portuguesa (Mateus et al. 2003), following the order of the above mentioned values of que. The occurrence of the connective que as introducing the final clauses is described and does not seem to raise problems: “para que, a fim de que, que (= para que) when the final clause uses the verb in a finite tense; in

Maria Antónia Coutinho

231

this case, the subject noun phrases (NP) of the two sentences are in general non co-referential; (…)”8 (Brito 2003, 717). As far as the occurrence of que as a causal expression is concerned, this author considers that: “[…] as orações explicativas iniciadas por pois, que (explicativo e só usado num registo literário) exprimem nexos de causalidade entre proposições, mas a impossibilidade de inversão mostra que são construções de coordenação: […]”9 (Brito 2003, 713). The opposition between the subordination and the coordination will be commented on below. Right now, it is useful to underline that, although cartoons are an authorial genre, it does not seem that the case under analysis should or must be considered literary; and, according to what was said above, the issue here concerns the representation of the (fragments of the) conversations (a representation that could also occur in the literary domain). Thus, it seems quite doubtful to consider that the explanatory que is restricted to literary usage. The conditional value of que is not usually referred to, when listing the connectives that introduce hypothetic conditional clauses, i.e. se, caso, no caso de, se porventura + V future subjunctive futuro do conjuntivo; contanto que, desde que + V presente subjunctivepresente do conjuntivo; quando + V future subjunctive. It should be noted that a double conditional and temporal value, equivalent to the interpretative paraphrases 4 and 5, corresponds to the last possibility (quando + V future subjunctive ). On the other hand, along with the conditional constructions with adverbial subordination, it is also recognized that the juxtaposition structures or those of phrasal coordination can express the same semantic value (conditional); one of these cases is precisely “uma construção de justaposição, em que uma das orações é uma imperativa seguida de declarativa (por vezes iniciada por que) no futuro ou no presente do indicative […]”10 (Brito 2003, 709-710). It is also important to underline the existence of “casos fronteira entre a coordenação e a subordinação em que há disparidade de valor semântico

8

“para que, a fim de que, que (= para que) when the final clause uses the verb in a finite tense; in this case, the subject noun phrases (NP) of the two sentences are in general non co-referential; [...]” 9 “[…] the explanatory clauses that begin by pois, que (explanatory and only used in a literary context) express causal links between propositions, but the impossibility of inversion shows that these are coordination constructions […]”. 10 “a juxtaposition construction, in which one of the clauses is imperative, followed by a declarative clause (sometimes beginning with que) in the future or present tense”.

232

Language in Action

entre os termos coordenados.”11 (Matos 2003, 551, n. 2; 555). Besides the above mentioned difficulties raised in the distinction between explanatory coordination and causal subordination, it will be useful to remember the semantic equivalence between copulative coordination and conditional subordination, as pointed out by Matos (2003, 555)12. From the above mentioned descriptions, it can be said that the occurrences of que with final, explanatory or causal, conditional, temporal and conditional values have been clarified. Although the possibility of a copulative coordination structure corresponding to a conditional semantic value is recognised, there is nothing that relates the occurrence of que with copulative or temporal-copulative value. Above all, although more than one value in one occurrence can be identified (for example, temporal and conditional), nothing leads to the possibility of a multiplicity of values converging in the occurrence of the same form. Is there a reason to reject the supported hypothesis and opt for an interpretation that selects one and only one of the values listed above?

4. Setting off for the journey The descriptions pointed out are characterized by a praiseworthy attempt at structure discreteness. However, as Matos (2003, 557) clearly observes, “a distinção entre coordenação e subordinação é sobretudo de ordem formal e não semântica”13. Other angles of analysis – more semantic and less formal – will perhaps make it possible to account for the effects of the sense that, although recurring to discrete forms, resists discreteness. From this point of view, instead of ambiguity, the above observed cases allow us to observe a possibility in the language – i.e., an (internal) possibility in which the norm and/or the grammar(s) do not hold nor are considered14. This only means that there is work to be continued. For one simple reason:

11 “borderline cases between the coordination and the subordination, in which there is a disparity in the semantic value between the coordinated terms”. 12 These are examples given by the author: Não comes a sopa e não te levo ao cinema! / Se não comeres a sopa, não te levo ao cinema. (Don’t eat the soup and I won’t take you to the cinema! / If you don’t eat the soup, I will not take you to the cinema). (Matos 2003, 555). 13 The distinction between coordination and subordination is above all of a formal order, not a semantic one. 14 On the Saussurian distinction between internal language and normalized language, see Bronckart 2011.

Maria Antónia Coutinho

233

Langue, or indeed any semiological system, is not a ship in dry dock, but a ship on the open sea. [...] Which is the real ship: one in a covered yard, surrounded by engineers, or a ship at sea? Quite clearly, only the ship at sea may yield information about the nature of a ship, and, moreover, it alone is a ship, an object available for study as a ship. (Saussure, 2006: 202)

References Bota, Cristian and Jean-Paul Bronckart. 2008. “Volochinov et Bakhtine: deux approches radicalement opposées des genres de texte et de leur statut”. Linx, 56: 67-83. Brito, Ana Maria. 2003. “Subordinação adverbial”. In Gramática da Língua Portuguesa, org. by M. H. Mateus et al., 695-728. Lisboa: Caminho. Bronckart, Jean-Paul and Cristian Bota. 2011. Bakhtine démasqué. Histoire d’un menteur, d’une escroquerie et d’un délire collectif. Genève : Droz. Bronckart, Jean-Paul. 1997. Activité langagière, textes et discours. Pour un interactionisme socio- discursif. Paris: Delachaux et Niestlé. —. 1999-2002. “La conscience comme “analyseur” des épistémologies de Vygotski et de Piaget”. In Avec Vygotski, dir. by Y. Clot, 27-53.Paris: La Dispute.. —. 2004. “Commentaires conclusifs. Pour un développement collectif de l’interactionnisme sociodiscursif”. Calidoscópio, 2(2). Accessed 13 October 2009, http://unisinos.br/publicacoes_cientificas/images/stories/pdfs_calidosc opio/20calidosco piov2n2_artigo15.pdf. Coutinho, Maria Antónia and Florencia Miranda. 2009. “To describe textual genres: problems and strategies”. In Genre in a Changing World. Perspetives on Writing, org. by Ch. Bazerman, D. Figueiredo and A. Bonini, 33-55 . Fort Collins, Colorado: The WAC Clearinghouse and Parlor Press. http://wac.colostate.edu/books/genre/. Leal, Leal. 2011. A Organização Textual do Gênero Cartoon: Aspectos Linguísticos e Condicionamentos não Linguísticos. PhD diss., Universidade Nova de Lisboa. http://run.unl.pt/handle/10362/6646 Mateus, Maria Helena Mira et. al. 2003. Gramática da Língua Portuguesa. Lisboa: Caminho. Matos, Gabriela. 2003. “Estruturas de coordenação”. In Gramática da Língua Portuguesa, org. by M. H. Mateus et al., 549-592. Lisboa: Caminho. Mauro, Tulio de. 1972. “Notes”. In Cours de Linguistique Générale, F. de

234

Language in Action

Saussure. Paris: Payot. Miranda, Florencia. 2010. Textos e géneros em diálogo – uma abordagem linguística da intertextualização. Lisboa: FCG/FCT. Ouellet, Pierre. 1989. “ «Par exemple...»: statut cognitif et portée argumentative de l'exemplification dans les sciences du langage”. In Perspectives méthodologiques et épistémologiques dans les sciences du langage , edited by M.-J., 95-114. Berne : Peter Lang. Passeron, Jean-Claude and Jacques Revel. 2005. “Penser pas cas. Raisonner à partir de singularités”. In Penser par cas, dir. by J.-C. Passeron and J. Revel, 9-44. Paris: Editions de l’Ecole de Hautes Etudes en Sciences Sociales. Saussure, Ferdinand de. 1959. Course in General Linguistics. N.Y.: Columbia University Press. —. 2002. Écrits de linguistique générale (edited by S. Bouquet and R. Engler). Paris: Gallimard. —. 2006. Writings in General Linguistics. Oxford: Oxford University Press. Vygotsky, Lev Semenovitch [1934]1997. Pensée et langage. Paris: La Dispute. Voloshinov, Valentin Nikolaevich [1929]1977. Marxism and the Philosophy of Language. Harvard: Harvard University Press and the Academic Press Inc.

Maria Antóniaa Coutinho

Annex

Afonso, Luís.. In Público (Jaanuary 26, 2003)

235