Syntax in Spanish-speaking children with Williams syndrome

Journal of Communication Disorders 60 (2016) 51–61

Contents lists available at ScienceDirect

Journal of Communication Disorders

Syntax in Spanish-speaking children with Williams syndrome Antonio Benítez-Burracoa,* , Elena Garayzábalb , Fernando Cuetosc a Departamento de Filología Española y sus Didácticas, Facultad de Humanidades, Universidad de Huelva, Avda. de las Fuerzas Armadas s/ n., 21071 Huelva, Spain b Departamento de Lingüística General, Facultad de Filosofía y Letras, Universidad Autónoma de Madrid, 28049 Madrid, Spain c Departamento de Psicología, Facultad de Psicología, Universidad de Oviedo, Plaza Feijoo, s/n., 33003 Oviedo, Spain

A R T I C L E I N F O

A B S T R A C T

Article history: Received 30 April 2014 Received in revised form 3 March 2016 Accepted 3 March 2016 Available online 5 March 2016

The syntactic skills of Spanish-speaking children with Williams syndrome (WS) were assessed in different areas (phrase structure, recursion, and bound anaphora). Children were compared to typically-developing peers matched either in chronological age (CA-TD) or in verbal age (VA-TD). In all tasks children with WS performed significantly worse than CA-TD children, but similarly to VA-TD children. However, significant differences were observed in specific domains, particularly regarding sentences with cross-serial dependencies. At the same time, children with WS were less sensitive to syntactic constraints and exhibited a poorer knowledge of some functional words (specifically, of nonreflexive pronouns). A processing bottleneck or a computational constraint may account for this outcome. ã 2016 Elsevier Inc. All rights reserved.

Keywords: Williams syndrome Spanish Syntax Phrase rules Recursion Binding Control Computation Chomsky hierarchy

1. Introduction Williams–Beuren syndrome (henceforth, WS) is a developmental disorder caused by a microdeletion in one copy of the chromosome 7, affecting roughly two dozen genes (Korenberg et al., 2008). The hemicygosis of these genes gives rise to diverse socio-affective, cognitive, and physical impairments (Bellugi, Korenberg & Klima, 2001; Tassabehji, 2003), although a one-to-one correlation between genes and abnormal traits cannot be easily drawn (Tassabehji, 2003). Language deficits and strengths have been a major concern regarding the cognitive profile of people with WS. First reports suggested that their language was substantially preserved throughout development, unlike other cognitive capacities (e.g. visuospatial cognition or social abilities) (see for instance Bellugi, Marks, Bihrle & Sabo, 1988; or Bellugi, Wang & Jernigan, 1994). This unusual pattern of dissociation between cognition and language (and even between aspects of language) was then claimed to support a modular organization of the mind. Nonetheless, recent, fine-grained analyses of language deficits in WS suggest that most aspects of language knowledge, language processing, and language use are delayed or perhaps impaired in people with WS (see Karmiloff-Smith & Mills, 2006; Martens, Wilson & Reutens, 2008; Mervis & Becerra, 2007 for reviews). Because pragmatic problems are probably explained by the socio-affective profile of the disorder (see Laws & Bishop, 2004; Stojanovik, 2006), it is problems with structural components of language (i.e. morphology, syntax, etc.) that

* Corresponding author at: Departamento de Filología Española y sus didácticas, Facultad de Humanidades, Campus de “El Carmen” Universidad de Huelva, Avda. de las Fuerzas Armadas s/n, 21071 Huelva, Spain. E-mail addresses: [email protected] (A. Benítez-Burraco), [email protected] (E. Garayzábal), [email protected] (F. Cuetos). http://dx.doi.org/10.1016/j.jcomdis.2016.03.001 0021-9924/ ã 2016 Elsevier Inc. All rights reserved.

52

A. Benítez-Burraco et al. / Journal of Communication Disorders 60 (2016) 51–61

have gained the interest of researchers over the years. In some domains of language people with WS outscore people affected by other developmental disorders. However, they don’t perform above their mental age peers (Karmiloff-Smith, 2008; Karmiloff-Smith & Mills, 2006). At the same time, their command of language is achieved in spite of deeper cognitive deficits. Importantly, language abilities improve as the child grows, pointing to cognitive compensatory mechanisms that seem to be active during development. For example, it has been claimed that working memory helps her to compensate semantic, syntactic, and pragmatic deficits (Karmiloff-Smith et al., 1998; Mervis & Becerra, 2007). Ultimately, in WS we observe “a mixture of delay, deviance, and asynchronies across the developing system” (Karmiloff-Smith & Mills, 2006: 587). In order to characterize the linguistic profile of people with WS confidently it is important to (i) analyze their language abilities at different stages of development, looking for differences with the typically- developing (henceforth, TD) population (Karmiloff-Smith, 2008; Karmiloff-Smith & Mills, 2006; Martens et al., 2008; Mervis & Becerra, 2007); (ii) find the core deficit (if any) that may explain the observed problems in the language domain (and perhaps in other domains of cognition); and (iii) collect more cross-linguistic data that help us to examine the peaks and valleys of WS language from the broader perspective of languages that are typologically different to English. Syntax abilities in WS are controversial. First descriptions of the condition (e.g. Bellugi, Bihrle, Neville, Jernigan & Doherty, 1992; Bellugi et al., 1988, 1994; Clahsen & Almazan, 1998) reported that ungrammaticalities were of lexical nature and that people with WS performed almost at ceiling in several syntactic indicators, both in comprehension tasks (passive sentences, negation, conditionals) and in production tasks (conditionals, nested dependencies, recursion), and also in tasks evaluating their metalinguistic knowledge. This uneven linguistic profile was then claimed to support a distinction between a computational device for language processing and a storage mechanism for lexical representations (see Ullman, 2001), that could be selectively impaired. However, recent studies suggest that syntax in WS is not spared (e.g. Grant, Valian & KarmiloffSmith, 2002; Joffe & Varlokosta, 2007; Karmiloff-Smith et al., 1998; see also Mervis & Becerra, 2007; Udwin & Yule, 1991). However, it is still disputed whether it is just delayed or it is really deviant compared to TD population. According to several studies (e.g. Karmiloff-Smith, 2008; Karmiloff-Smith & Mills, 2006; Martens et al., 2008; Mervis & Becerra, 2007 for review) the WS syntax differs both chronologically and qualitatively from the TD syntax. According to others (e.g. Mervis, Morris, Bertrand, & Robinson, 1999) children with WS show syntactic abilities that are at the expected level for their mental age. A related concern is whether the observed deficits are interpretable in terms of a selective breakdown of specific components of syntactic knowledge. For example, concerning syntactic dependences (e.g. passives, and binding of reflexive and non-reflexive pronouns) Perovic and Wexler (2007, 2010), Perovic and Wexler (2007) found that children with WS performed better on passives with agentive verbs, similarly to younger TD children (Hirsch & Wexler, 2006; Maratsos, Fox, Becker & Chalkley, 1985). At the same time, they had mastered the binding principle between a reflexive and its antecedent. Perovic and Wexler concluded that the capacity for forming A-chains is impaired in WS, whereas the knowledge of the binding principles involved in anaphora is spared. We believe that this controversy around language abilities (and particularly, syntactic knowledge) in WS may benefit from a bottom-up approach. On the one hand, current neurolinguistics urges us to distil language into computational and representational primitives that are computable by the brain in real time in order to overcome the interface problems for the study of language (see Poeppel & Embick, 2005 for details). Related to this, we should expect that deficits in specific, highlevel capacities result from the impairment of low-level, more generalized processes (at the same time, low level deficits may affect several cognitive abilities simultaneously). As pointed out by Karmiloff-Smith (2009: 58), “early, more general lowlevel processing deficits [can] affect several domains but to differing degrees and at different developmental times”. Finally, we wish emphasize that similar problems with grammar are observed in different language disorders of diverse aetiology. For example, difficulties with verbal morphology are experienced by children with Down syndrome (Eadie, Fey, Douglas, & Parsons, 2002), specific language impairment (SLI) (Polite & Leonard, 2006), speech-sound disorder (SSD) (Mortimer & Rachew, 2010), autism (Roberts, Rice, & Tager-Flusberg, 2004), and WS (Clahsen & Almazan, 1998; Thomas et al., 2001). It has been claimed that the most vulnerable aspects of cognition rely on less resilient neural network and that these networks are more sensitive to damage or to ontogenetic disturbances because of their evolutionary novelty (Toro et al., 2010). All these aspects and concerns, when considered together, surely explain the observed deficits in the domain of the WS syntax. This paper assesses in detail the syntactic abilities of a sample of Spanish-speaking participants with WS. In doing so, our aim is twofold: First, we want to contribute to the ongoing discussion around the underlying deficit(s) in WS that emerge(s) in the form of syntactic problems. Following Fitch and Hauser (2004) we will resort to the hierarchy of formal languages known as the Chomsky Hierarchy (Chomsky, 1969), which has succeeded in capturing the computational constraints on syntactic processing in non-human primates. This Hierarchy classifies logically possible languages into sets of nested regions corresponding to patterns describable by means of grammars and with smaller regions captured by increasingly less powerful machinery. Chomsky Hierarchy has emerged as a good tool for comparative studies, because it makes use of computational primitives (non-specific to language), it avoids “theory-internal” debates and it enables cross-domain, crossstages, and cross-species comparisons. Moreover, it has been claimed that the most computationally demanding tasks according to the Hierarchy (i.e. mildly context-sensitive computations, such as binding and control) were achieved latter in our evolutionary lineage (Balari, Benítez-Burraco, Longa, & Lorenzo, 2013). Accordingly, they rely on an enhanced computational device which resulted from gaining working memory space and that allowed us to manipulate words well beyond a strictly lineal computational regime (see Balari & Lorenzo, 2013 for details). If recently evolved neural networks are really the most sensitive to damage, as noted above, we expect that the most computationally-demanding tasks are

A. Benítez-Burraco et al. / Journal of Communication Disorders 60 (2016) 51–61

53

preferably impaired in language disorders, and specifically, in WS. Accordingly, we have tested different syntactic tasks involving different computational demands. Second, we want to provide the scientific community with syntactic data from a WS population acquiring a language different to English. At present, we can only rely on few studies dealing with small samples (e.g. Bartke (2004) and SchanerWolles (2004) for German, or Stavrakaki (2003) on Wh-questions for Greek-speaking children), a limited set of longitudinal studies of one or two subjects (e.g. Capirci et al. (1996) for Italian, or Levy (2002) for Hebrew-speaking children), and a couple of broad analyses of cognitive and linguistic dysfunctions in larger populations (e.g. Volterra, Capirci, Pezzini, Sabbadini & Vicari, 1996, for Italian children with WS). That said, we wish note that this is not properly a cross-linguistic study, because although Spanish syntax differs from English syntax in some important aspects (e.g. the pronominal subject can be omitted in the former), the linguistic structures we have tested are similar in both languages. Likewise we wish stress that this is not a longitudinal study: although we provide with data from children, they are of limited usefulness for disentangling the ontogeny of language in WS. 2. Material and methods 2.1. Participants Three groups of participants took part in this study. Thirteen children with WS (six boys and seven girls, mean age 10;7, range 7;3–15;6) with previously confirmed diagnosis by positive fluorescence in situ hybridization (FISH) to the elastin gene deletion in chromosome 7 conformed the target group (hereafter, WS). Thirteen TD children (six boys and seven girls, mean age 6;4, range 5;4–7;4), matched to the WS group in verbal mental age, composed the first control group (hereafter, VA-TD). The matching between the WS and the VA-TD groups was performed according to the short form of the Wechsler Intelligence Scale for Children-Fourth Edition (WISC-IV) and also to a lexical decision task (Table 1). The short version of the WISC-IV provides high levels of reliability and validity when applied to research purposes (King & King, 1982; Blyler, Gold, Iannone, & Buchanan, 2000). This reliability has been specifically attested for the Spanish version of the test when used for tracking cognitive changes over time (Demsky, Gass, Edwards & Golden, 1988). This abridged version of the WISC-IV test consists of four subtests: Coding, Picture Completion, Similarities, and Digit Span, one of each represents main indexes of the WISC-IV: the Processing Speed Index (PSI), the Perceptual Reasoning Index (PRI), the Verbal Comprehension Index (VCI), and the Working Memory Index (WMI), respectively. No significant differences were found between the WS group and the VA-TD group taking into account the direct scores in WMI and VCI, the subtests that are more closely related to verbal behavior (see Table 1). With regards to the lexical decision task, a total of 16 words and 16 non-words (primitive and morphologically derived) randomly mixed were presented orally and the participant had to say whether it was a real word or not. The frequency band employed was low, between 3 and 19 occurrences per million (with a mean value of 9.8). No significant differences were found between both groups, who performed equally well in this task. At the same time, no ceiling effect was observed either (see Table 1). Finally, other eight TD children (four boys and four girls, mean age 10;4 with a range of 8;0– 12;5) comprised the second control group that matched the WS group in chronological age (hereafter, CA-TD). This CA-TD group was further used as a control group to prove the validity of the tasks (the VA-TD group performed low on several subtasks; see Tables 2–4). All participants were monolingual native speakers of Spanish with no hearing deficits. For all groups, their parents provided written informed consent for their participation in the study. Overall, we wish acknowledge that the small size and the large age range of our WS group may be a limitation of our study (although this is a common problem due to the rarity of the syndrome). Moreover, the level of error we have accepted to probe the matching of the WS and the VA-TD groups may be too high according to some authors (e.g. Frick 1995) and that this circumstance might affect the conclusions of our study. 2.2. Tasks Following our main hypothesis, we assessed aspects of syntactic knowledge with different computational demands according to Chomsky Hierarchy: metalinguistic knowledge of phrase structure (phrase rules, auxiliary markers, subcategory constraints), context-free tasks (production of recursive structures with noun complement or relative clauses); and mildly context-sensitive tasks (comprehension of control principles and of binding principles involving reflexive or nonreflexive pronouns). Table 1 Cognitive and verbal assessment of the three experimental groups. WISC-IV subtest

WS

VA-TD

Sign. (WS/VA-TD)

CA-TD

Sign. (WS/CA-TD)

Digit symbol coding (PSI) Picture completion (PRI) Similarities (VCI) Digit span (WMI) Lexical decision task

18.46 (12.02) 9.46 (5.22) 9.15 (2.58) 8.54 (2.54) 26.15 (3.74)

30.23 (10.06) 14.38 (4.03) 11.85 (4.06) 10.23 (2.98) 29.00 (2.34)

.022 .022 .082 .116 .087

46.37 (9.47) 27.00 (5.50) 26.13 (6.87) 15.75 (1.75) 29.38 (1.18)

.000 .000 .000 .000 .105

54

A. Benítez-Burraco et al. / Journal of Communication Disorders 60 (2016) 51–61

Accordingly, the following six tasks were designed: (1) Phrase structure: 32 items were tested to evaluate judgments on grammaticality of orally presented structures. Participants only had to say whether the sentence was correct or not. Different conditions were assessed: (1a) basic phrase rules (constituent order within the noun phrase or the prepositional phrase) (e.g. Esta mesa es muy ancha ‘This Table is very wide’ vs. *Casa esta es muy bonita ‘House this is very beautiful’) (8 items); (1b) auxiliary markers and basic structure of the inflectional phrase (e.g. Antonio está comiendo ‘Anthony is eating’ vs. *Juan está comió ‘John is ate’) (4 items); (1c) verbal valency (i.e. number of verbal arguments) (e.g. Nieva ‘It snows’ vs. *Ella nieva ‘She snows’, or El hombre estornuda ‘The man sneezes’ vs. *El hombre agarró ‘The man grasped’) (10 items), and 1d) syntactic subcategory constraints within the verbal phrase (e.g. La abuela compró un libro a su nieta ‘The grandma bought her granddaughter a book’ vs. *La niña compró un regalo su madre ‘The girl bought a present her mother’) (10 items). Half of the items were ungrammatical in Spanish; the other half were grammatical in this language. (2) Production of noun complement clauses: 16 items evaluated the participants’ ability to complete a complex sentence from a given drawing and a given input (i.e. the main clause) (e.g. El gato quiere . . . ‘the cat wants . . . ’). Their responses (e.g. que el perro salte ‘that the dog jumps over him’, or jugar con el perro ‘to play with the dog’) were analyzed according to the following criteria: (a) Evidence of recursion (i.e. production of an embedded complementizer phrase; errors at some other structural level [e.g. verbal inflection, use of clitics, etc.] were admitted). (b) Presence of a well-formed recursive structure (i.e. production of a grammatical embedded complementizer phrase). (c) Presence of an embedded complement clause introduced by the complementizer que ‘that’ (errors at some other structural level were admitted). (d) Presence of an embedded non-finite clause (errors at some other structural level were admitted). (e) Presence of other types of embedded clauses. (f) Production of simple sentences. (3) Production of adjunct (relative) clauses: this task comprised 16 items that assessed the participants’ capability to produce complex sentences with relative clauses from a given question and a given picture (e.g. Qué elefante está asustado? ‘Which elephant is frightened?’). As in the previous task, we analyzed their answers (e.g. el que es perseguido por el ratón ‘the one chased by the mouse’ or el elefante al que persigue el ratón ‘the elephant that the mouse chases’) according to different criteria: ?

(a) Evidence of recursion per se (i.e. production of an embedded clause; errors at some other structural level [e.g. verbal inflection, use of clitics, etc.] were admitted). (b) Presence of a well-formed recursive structure (i.e. production of a grammatical embedded clause). (c) Presence of an embedded relative clause (errors at some other structural level were admitted). (d) Production of simple sentences. (4) Comprehension of bound anaphora (binding and control relations): this task assessed whether participants were able to properly bind a determiner phrase (either null or manifest) in the subordinated clause to a referential element (a control noun phrase) in the main clause. Different control verbs (i.e. verbs that determine the binding because of their semantic nature) were considered: type I or subject control verbs (i.e. amenazar ‘to threaten’, anunciar ‘to announce’, asegurar ‘to assure’, decir ‘to say/tell’, prometer ‘to promise’) and type II or object control verbs (i.e. convencer ‘to persuade’, obligar ‘to force’, ordenar ‘to order’, pedir ‘to ask’, rogar ‘to beg’). Moreover, three different anaphoric elements were included: PRO (i.e. the null subject of a non-finite clause), pro (i.e. the null subject of a finite clause), and a non-reflexive pronoun acting as the object of the clause (pronounOBJ). Whereas pronouns are BOUND to a noun phrase in the main clause, both PRO and pro are said to be CONTROLLED by a noun phrase within the main clause. On the whole, 16 items and four different conditions were tested: (a) Control verb I + PRO + pronounOBJ (e.g. La mujeri prometió a la niñaj PROi besarlaj ‘The woman promised the girl to kiss her). (b) Control verb II + PRO + pronounOBJ (e.g. El policíaiordenó al médicojno PROjsoltarloi “The policeman ordered the doctor not to let him go”. (c) Control verb I + pro + pronounOBJ (e.g. El payasoiaseguró al bomberojque proilojmojaría ‘The clown announced the fireman that he would splash him’). (d) Control verb II + pro + pronounOBJ (e.g. El cocheipidió al camiónjrojo que projloigolpease ‘The car asked the red truck to hit it’). Items were presented orally and the participants had to choose a drawing from four given. Two of them displayed the entities mentioned in the sentence (e.g. mujer ‘woman’ and niña ‘girl’ in La mujer prometió a la niña besarla) as the antecedent of either PRO or pro (and the other one as the antecedent of the pronounOBJ). The two remaining pictures contained a semantic distractor (e.g. a boy) acting as the antecedent of either PRO (or pro) or the pronounOBJ in the subordinated clause.

A. Benítez-Burraco et al. / Journal of Communication Disorders 60 (2016) 51–61

55

(5) Comprehension of reflexive and non-reflexive pronouns: this task tested if simple or complex sentences containing either reflexive or non-reflexive pronouns were correctly understood (e.g. El niñoi sei lava ‘The boy washes himself’ vs. El niñoi laj peina ‘The boy brushes her’). Reflexive pronouns are always locally bound (e.g. El niñoi ordenó al hombrej PROj secarsej la cara ‘The boy ordered the man to dry his (the man’s) face’, but *El niñoi ordenó al hombrej PROj secarsei la cara ‘The boy ordered the man to dry his (the boy’s) face’; to convey this meaning the correct sentence is El niñoi ordenó al hombrej secarlei la cara). In the complex sentence condition, the embedded clause was tenseless (i.e with PRO) and two different types of control verbs were used, as in task 6 (e.g. La niñai prometió a la mujerj PROi despertarlaj ‘The girl promised the woman to wake her’ or El policíai ordenó al médicoj PROj no soltarsej ‘The policeman ordered the doctor not to let himself go’). A total of 14 items were employed to test the participants’ knowledge. They were presented orally and the participants had to choose a drawing from four given. In the simple sentence condition (4 items) (e.g. El niñoi sei lava ‘The boy washes himself’ or El niñoi laj peina ‘The boy brushes her’), three of them corresponded to a non-reflexive interpretation of the verbal action (i.e. a boy washing or brushing a girl or another boy, or instead a girl washing or brushing a boy). The fourth drawing corresponded to the reflexive reading of the verbal action (i.e. a boy washing or brushing himself). In the complex sentence condition (10 items), two of them displayed the entities mentioned in the sentence (e.g. mujer ‘woman’ and girl ‘niña’ in La mujeri prometió a la niñaj PROi cepillarlej los dientes ‘The woman promised the girl to brush her (the girl’s) teeth) as the antecedent of PRO (and the other one as the antecedent of the pronounOBJ). The two remaining pictures corresponded to a reflexive reading of the verbal action (i.e. a woman or a girl brushing her own teeth). 2.2.1. Procedure All participants were individually tested over two days. During the first day their cognitive profile was assessed according to the short version of the WISC-IV. We applied the linguistic tests in two separate sessions during the second day, because they were highly demanding for the participants with WS. Because of their limited sustained attention we also avoided distractions that might affect their performance. Target sentences were read aloud by the experimenter, who used a neutral prosody. The same protocol was employed across tasks and across experimental groups. The tasks were always introduced by an example. To check that the child had correctly understood what she was expected to do, first the experimenter read the question in the example and then asked the child to provide an answer. When the child did not understand the procedure, the experimenter read the question for a second time and provided her with the correct answer. Next, the same question was read again to the child who was encouraged to provide an answer by herself. Only then the target sentences were read (only once) and the child’s answers were written down. Depending on the task, participants had to point at the drawing that matched the given input (one among four in tasks 4 and 5), to say whether a given sentence was correct or not (task 1), or to complete orally some incomplete statement according to the information provided visually (tasks 2 and 3). In the first two cases, correct answers were given a score of 1, while incorrect answers were given a score of 0. In the third case, answers were written down and later analyzed according to the different criteria described above. Answers satisfying the criteria were given a score of 1, whereas those not satisfying them were given a score of 0. 3. Results 3.1. General overview The performance of the WS and the control groups was compared using non-parametric statistics (Mann–Whitney U test). Table 2 shows the scores achieved by each group in the different tasks, the standard deviations, and the statistical significance of the differences observed between them. The WS group scored significantly lower than the CA-TD group in all tasks. Children with WS also scored lower than the VA-TD group, although the observed differences were statistically significant only regarding tasks 1 and 5. Overall, the scores achieved by the WS group were more variable than those of the VA-TD group. Differences between groups regarding variability were more prominent in the task assessing the comprehension of bound anaphora (Table 2). Because of the different types of items we included in each task, next we provide a qualitative analysis of the results. Whenever items can be confidently grouped in larger subcategories (e.g. problems with subcategorization in task 1, or problems with binding involving type II control verbs in task 4), we used the Mann–Whitney U test to look for significant differences (see also Tables 3 and 4). Since the CA-TD group scored almost at ceiling in all tasks (and all conditions within tasks), we will focus on the VA-TD group. Table 2 Success in each task. Task

N

WS

VA-TD

Sign. (WS/VA-TD)

CA-TD

Sign. (WS/CA-TD)

Phrase structure Production of complement clauses Production of relative clauses Comprehension of bound anaphora Reflexives vs. non-reflexive

32 16 16 16 14

27.00 (2.45) 10.62 (2.63) 6.69 (4.66) 11.62 (3.62) 9.08 (3.07)

29.30 (2.10) 10.92 (2.06) 10.15 (5.14) 14.23 (1.88) 11.85 (2.34)

.013 .850 .076 .091 .011

30.62 (0.74) 13.50 (0.77) 13.25 (1.54) 15.75 (0.91) 12.87 (0.85)

.001 .005 .002 .007 .001

56

A. Benítez-Burraco et al. / Journal of Communication Disorders 60 (2016) 51–61

Table 3 Success in the noun complement clauses task. Task

WS

VA-TD

Sign. (WS/VA-TD)

CA-TD

Sign. (WS/CA-TD)

Presence of recursion Correct recursive structure Complement clauses with que ‘that’ Complement clauses with non-finite verbs Non-complex clauses

10.61 (2.66) 10.00 (2.65) 7.23 (3.49) 3.38 (1.98) 4.85 (2.61)

10.92 (2.06) 10.92 (2.06) 4.38 (2.47) 6.54 (2.11) 4.31 (1.70)

.835 .375 .055 .001 .0715

13.50 (1.19) 13.37 (1.06) 5.50 (0.92) 8.00 (1.31) 1.63 (1.30)

.005 .001 .045 .000 .005

Table 4 Success in the relative clauses task. Task

WS

VA-TD

Sign. (WS/VA-TD)

CA-TD

Sign. (WS/CA-TD)

Presence of recursion Correct recursive structure Relative clauses Non-complex clauses

7.92 (4.75) 5.92 (4.21) 6.69 (4.66) 4.08 (3.20)

10.31(5.30) 7.38 (4.44) 10.15 (5.14) 3.62 (3.99)

.217 .439 .076 .500

14.00 (1.14) 9.75 (3.15) 13.25 (1.16) 0.88 (0.64)

.002 .050 .002 .007

3.2. Phrase structure, auxiliaries, and subcategory constraints In this task differences between the VA-TD group and the WS group were significant. A qualitative analysis of the answers suggested that syntactic subcategory constraints were the most problematic aspect for children with WS. Apparently, they paid less attention to formal cues (e.g. prepositions) than to semantic cues (i.e. plausibility), quite contrary to what their TD peers did. Children with WS also scored quite low when rules governing the phrase structure within local constituents were violated. Again, it seems that they paid less attention to formal cues than to semantic information (i.e. coherence, plausibility) when judging the acceptability of these sentences (e.g. *Nosotros niños estudiamos inglés ‘We children study English’). Both control groups performed better in this condition. Finally, errors encompassing the use of wrongly inflected auxiliaries were always noticed by children with WS. 3.3. Context-free computation Tasks 2 and 3 aimed to test if recursion is altered in children with WS acquiring Spanish. When we evaluated the presence of noun complement clauses (task 2), significant differences between the WS and the VA-TD groups were not observed (Table 2). Moreover, both groups generated quite the same amount of non-complex sentences (which were residual in the CA-TD group). An in-depth analyses of the results (Table 3) revealed that children with WS favored noun complement clauses with an inflected verb, whereas their VA-TD peers preferred tenseless clauses. When we assessed the production of relative clauses (task 3) no significant differences between the WS group and the VATD group were observed (Table 2). Children with WS produced a greater amount of non-relative subordinated clauses and of simple sentences (Table 4), but they formed recursive structures correctly. To same extent relative clauses were problematic for TD children also (both groups tended to interpret subject-object sentences as subject-subject sentences). 3.4. Mildly context-sensitive computation In tasks evaluating the comprehension of bound anaphora, non-significant (task 4) and significant (task 5) differences between groups were found (Table 2). Regarding task 4 we performed non-parametric analyses of two different composite conditions: sentences with either PRO or pro and type I control verbs (involving cross-serial dependencies and hence, truly mildly context-sensitive computations) and sentences with either PRO or pro and type II control verbs. Children with WS scored significantly lower in items with type I control verbs (WS = 4.69 (1.26), VA-TD = 6.61 (0.92), p = .000). In particular, they used to bind PRO or pro to the proximal noun. For instance, when they were said El policíai amenazó al ladrónj con PROi perseguirloj ‘The policeman threatened the robber to chase him’, most of the children with WS pointed to the picture depicting a thief chasing a policeman. Finite clauses (with pro) did not improve the performance of the WS group, contrary to what was observed in the VA-TD group (in these clauses the link between the anaphora and the antecedent in the main clause is also marked by the verbal inflection: conditional mood for type I control verbs and subjunctive mood for type II control verbs). Overall, it seems that sentences with PRO controlled by subject control verbs are really problematic for children with WS. Additionally, the lexical nature of the anaphoric element (reflexive vs. non-reflexive pronouns) also affected significantly their performance. In task 5 the WS group scored significantly lower in sentences that contain non-reflexive pronouns in an

A. Benítez-Burraco et al. / Journal of Communication Disorders 60 (2016) 51–61

57

object position (WS = 4.85 (2.035), VA-TD = 6.69 (1.437), p = 0.011); in turn, children with WS performed better on sentences with reflexives (WS = 4.23 (1.481), VA-TD = 5.15 (1.144), p = 0.067). To some extent they seemed to (mis) interpret nonreflexive pronouns as reflexives (e.g. lo ‘him’ > se ‘himself’). In other cases, non-reflexives pronouns were mixed up (lo ‘him’ > la ‘her’). 4. Discussion Our results suggest that all aspects of syntax we have evaluated are delayed in children with WS. However, when they are compared to their peers matched on verbal age, some of them seem to be spared, while others are impaired. Overall, mildly context-sensitive computation (specifically, the understanding of bound anaphora involving cross-serial dependencies and non-reflexives) is difficult for them. Furthermore, our results support the view that the intragroup variability among children with WS is not significantly greater than among their TD peers (in the line of Levy and Bechar, 2003). Finally, we have not observed qualitative differences with children acquiring English, but it is true that, because of our core hypothesis, we have not examined the areas where English and Spanish differ (e.g. pronoun dropping). Next, we will discuss in detail the results obtained in the different domains subject to scrutiny. We will finish with a conclusion section in which we will examine the results under the light of our core hypothesis. 4.1. Phrase structure, auxiliaries, and subcategory constraints In task 1 we tested the metalinguistic knowledge of basic phrasal rules in Spanish. Not many studies have evaluated the metalinguistic knowledge of grammatical rules in people with WS. Although we have relied on an off-line picture-pointing task, which explores explicit processing, our results are in line with Karmiloff-Smith et al. (1998), who used an on-line word monitoring task tapping implicit processing. Accordingly, we have also found that children with WS and TD children equally respond to violations of auxiliary structure, and that children with WS seem to be quite insensitive to syntactic constraints which are lexically specified. The former circumstance seems to account for the significant differences between groups observed in the whole task. However, in our opinion this is not a structural problem per se and might be explained if some syntactically-relevant constituents are ignored because of their prosodic weakness (e.g. *[D]el niño estornuda ‘[of] the boy sneezes’) or are wrongly inferred when they are absent (e.g. *El niño come pan [y] jamón ‘The boy eats bread [and] ham’). Similarly, we have not found significant differences between both groups regarding their metalinguistic knowledge of rules governing the phrase structure. Contrary to Capirci et al. (1996) and Volterra et al. (1996), our children did not omit the article in the noun phrase nor wrongly inflect auxiliaries (Capirci et al., 1996; Volterra et al., 1996). However, these studies evaluated the productive abilities of the children, whereas we have tested their metalinguistic knowledge (it should be noted that grammaticality judgments are not self-explanatory in impaired populations because the difficulties they have for understanding this sort of tasks, and plausibly also, because memory shortages and/or attentive deficits). 4.2. Context-free computation Tasks 2 and 3 aimed to test if recursion was altered in children with WS acquiring Spanish. Recursion in WS is a topic of great interest, but it has been generally examined in spontaneous speech samples. Bellugi et al. (1992) found that children with WS performed almost at ceiling in comprehension tests involving sentences with conditionals. Moreover, they were able to produce grammatical complex sentences with MLU values higher than those of their TD peers matched in mental age. These sentences contained conditionals, multiple embeddings, and relative clauses (Bellugi et al., 1988). Most errors involved an incorrect choice of specific lexical items, but were not structural in nature. Similarly, Clahsen and Almazan (1998) found that 15% of the spontaneous speech utterances generated by their children with WS contained embedded sentences, including relative and complement clauses. According to their view, the low scores obtained by their subjects in tests like TROG were explained by the misunderstanding of specific lexical items. In a similar vein, they found that only lexical compounds were problematic for children with WS, whereas they correctly employed regular plurals as non-head elements inside nominal compounds (Clahsen and Almazan, 2001). Notice that according to Kiparsky (1982) phrasal compounds are generated by a recursive mechanism. In contrast, Mervis and Becerra (2007) found that people with WS underscored in tasks involving complex sentences, because they made errors that were both semantic and grammatical in nature: omission of tense markers, omission of auxiliaries, failures in joining a mandatory second clause when relational terms were involved, etc. Furthermore, Zukowski (2001) found that children with WS generally performed as younger TD children (matched on cognitive and verbal abilities) do with regards to relative clauses. According to Volterra et al. (1996) children with WS acquiring Italian (with ages ranging from 4;10 to 15;3) are not ahead of their TD peers matched on mental age regarding the correct use of infinitives in sentences that require finite verbs. Finally, Grant et al. (2002), who assessed the ability for repeating different kinds of relative clauses, found that children with WS performed similarly to younger TD children. Our results give support to previous research suggesting that this aspect of grammar (and this computational regime) is not impaired in children with WS. Similarly to their VA-TD peers, our children with WS preferred to generate non-complex sentences. Obviously, simple sentences are easier to generate and less demanding in terms of computational resources. Nonetheless, when they produced complex sentences with embedded clauses, they performed similarly to the VA-TD group. Our results are in line with previous research conducted with children with WS acquiring English. Hence, according to Grant

58

A. Benítez-Burraco et al. / Journal of Communication Disorders 60 (2016) 51–61

et al. (2002), subject-object relative sentences are the most problematic type for them. Similarly, they don’t add nor omit relative clauses when it is not appropriate. Finally, they also prefer complement clauses with full complementizers and inflected verbs (in Grant et al.’s study the presence of that in a repetition task involving relative clauses improved their performance scores). 4.3. Mildly context-sensitive computation Our data suggest that binding and control are problematic computations for children with WS. Previous research on this area has produced quite opposite results. According to some studies (e.g. Bellugi et al., 1988; Clahsen & Almazan, 1998; Ring & Clahsen, 2005), binding of reflexive and non-reflexive pronouns is spared in WS. In fact, Ring and Clahsen (2005) claimed that their children with WS scored above their mental-age matched peers in tasks involving binding. On the contrary, Perovic and Wexler (2007) concluded that young children with WS (below CA of 12) had problems with personal pronouns. Binding is thought to obey different regulatory principles depending on the nature of the anaphoric element involved. Following Chomsky (1981), reflexive pronouns always refer to an antecedent located within the same clause (Binding Principle A). In turn, non-reflexive pronouns can be syntactically bound to a non-local antecedent (Binding Principle B). It has been frequently noted that very young TD children tend to interpret non-reflexive pronouns as reflexive anaphors (Chien & Wexler, 1990; Wexler & Chien, 1985). This phenomenon is sometimes referred to as the Delay of Principle B Effect (Wexler & Chien, 1985). According to Kapur, Lust, Harbert & Martohardjono (1992) this outcome may also result from the confusion of reflexive and non-reflexive pronouns. Another plausible explanation is that some processing constraint (e.g. less short-term memory or difficulties for integrating grammatical knowledge of different sources) exists at these ages. Specifically, this constraint could prevent younger children from applying already mastered principles to items that are excessively ‘downstream’ in the discourse (Goodluck, 1990). If this hypothesis is correct, we should expect worse scores as the distance between the anaphora and the antecedent increases. This may explain as well what we observe regarding control relations involving pro and a nonarbitrary PRO. Although these relations are usually accounted by a different component of grammar, they resemble binding relations at this level. In fact, younger TD children seem to obey as well a ‘minimal distance principle’ between PRO and its antecedent when computing these structures (Chomsky, 1969). Regarding bound anaphora per se, our children with WS scored lower (but not significantly lower) than their VA-TD peers. Interestingly however, significant differences were observed in sentences containing subject control verbs, which involve real cross-serial dependencies and thus demand mildly context-sensitive computations to be understood. In order to correctly process these sentences one has to bind two different anaphoric elements to their respective antecedents relying on several morphosyntactic cues that are scattered thorough the whole sentence. Clearly, this was a too demanding task for our children with WS, who sometimes performed binding randomly. Interestingly, complex sentences with dependent clauses with pro (which bear the highest number of morphosyntactic cues) did not increase their performance. Overall, these results can be explained if the ability for performing mildly context-sensitive computation is selectively impaired. Another interpretation may be that children with WS are obeying the ‘minimal distance principle’ to a greater extent than the VA-TD children. Lastly, we found that our children with WS were able to understand sentences containing reflexive pronouns, but sometimes they misinterpreted non-reflexive pronouns as reflexive anaphors, as very young TD children also do. Accordingly, we have not found any qualitative difference between the grammar knowledge of both groups. On the contrary, in their analysis of an Italian-speaking girl with WS, Capirci, Sabbadini, & Volterra (1996) found that sometimes she misplaced non-reflexive pronouns in the sentence and omitted reflexive pronouns. But because we have relied on a comprehension task, we cannot conclude that similar problems with the correct use of pronominal items are expected in children with WS acquiring Spanish, a language that is typologically similar to Italian. Overall, we regard some aspects of binding and control processing to be delayed in children with WS. Also the knowledge of specific anaphoric items is delayed in them. It is probable that some processing constraint may contribute to their difficulties with long-distance binding and with sentences containing complex anaphora (particularly, those involving two anaphoric items and two antecedents). 5. Conclusions In the last years, the ongoing research on WS has tried to establish whether some core syntactic operation or principle is selectively impaired in this clinical condition. As we argued in the introduction, we believe that this debate may benefit from a bottom-up approach that searches for basic cognitive processes, or computational or representational primitives that may be selectively altered in people with WS. Because semantic knowledge is a relative strength in WS (Mervis & Becerra, 2007), we hypothesized that syntax deficits in children with WS may result from the selective dysfunction of the computational system of language. In order to test our hypothesis we have relied in the Chomsky Hierarchy, a promising cognitive tool for comparative studies (Fitch & Hauser, 2004). Specifically, we expected that context-sensitive computations were impaired in them, because they demand more working memory resources and rely on neural devices that are more vulnerable because of their evolutionary novelty. Accordingly, we have found more problems with binding and control involving cross-dependencies than with recursion. At the same time, although recursion per se is not impaired in them, our children with WS favored simpler structures, plausibly because they are less demanding in terms of working memory. Finally, we have observed that they have difficulties for

A. Benítez-Burraco et al. / Journal of Communication Disorders 60 (2016) 51–61

59

computing several morphosyntactic cues simultaneously, as it is also the case with younger TD children. Overall, the syntactic knowledge of these children with WS acquiring Spanish seems to be delayed. Plausibly, this delay is due to the same memory constraints observed in young TD children. However, because binding relations involving cross-serial dependencies are the most problematic item for them, we cannot rule out the possibility that they have no access to the highest regime of the computational device for language (i.e. a mildly context-sensitive grammar according to the Chomsky Hierarchy). That said, we wish add two notes of caution in order to under-emphasize the theoretical implications of these conclusions. First, children with WS are distinguished by their relatively spared verbal working memory (Vicari, Bellucci & Carlesimo 2005). However, we find no real contradiction with our results (and our interpretation of the results). In the Chomsky Hierarchy the memory resources available to each type of computational regime (what we have labelled as ‘working memory’) do not necessarily correlate with particular psychological models of ‘working memory’ (e.g. Baddeley’s, 2007). The latter are full-fledged performance models. On the contrary, the computational regimes posited by the Hierarchy merely establish some limiting conditions that performance models must respect when exploring the psychological nature of our abstract memory component (see Balari, Benítez-Burraco, Camps, Longa, & Lorenzo, in press for details). Of course, this is an empirical question that should be examined experimentally. Related to this, we think it would be interesting as well to see whether the pattern of delays (or even impairment) that we have attested in the domain of language is also observed in other cognitive domains. According to our view, the computational system of language is functionally unspecific (i.e. not specific to language), this meaning that it interfaces with different devices and renders different outputs (endowed with similar formal properties) (see Balari et al., 2013; Balari & Lorenzo, 2013 for details). The ultimate reason is that neural devices perform basic types of computations that are recruited for language and for other cognitive functions (Poeppel & Embick, 2005). Accordingly, we expect that children with WS are unable to process cross-serial dependencies in other different domains (like vision or motor behavior). This issue deserves to be explored experimentally too. Second, we wish acknowledge that the theory of formal languages, and particularly, the Chomsky Hierarchy, may be not the best tool for studying cognitive processes. According to Berwick, Beckers, Okanoya & Bolhius (2012:2) it is both “too weak and too strong” to properly capture natural languages. On the one hand, different components of language fall in different ranges within the Hierarchy. Moreover, it does not uniquely characterize human language. Additionally, it does not say enough about the cognitive devices used to master language. Finally, it takes the notion of syntax too literally: in natural languages semantics is involved too (see Berwick et al., 2012; Heinz & Idsardi, 2013; and Heinz, in press for a detailed discussion). The first two concerns do not really go against our interpretation of the results: we have focused on a specific domain of language (syntax) and we were not concerned with other cognitive systems in other species that are susceptible to be characterized by the Hierarchy. On the contrary, the last two concerns can be problematic for us. As pointed out above, it is not clear which type of ‘real’ memory is involved in performing the tasks we have designed (but this is also an empirical question). Similarly, our tasks cannot be equated to the tasks used for testing the acquisition of artificial grammars (for which the Chomsky Hierarchy has been acknowledged to fail to reflect cognitive processes (Öttl, Jäger, & Kaup, 2015)). Semantic cues always improve the performance of the subjects and more complex formal grammars can be learned (Fedor, Varga, & Szathmáry, 2012). As we have seen, children with WS pay more attention to semantic cues than to formal cues. Consequently, it may well be that they underscore in tasks evaluating formal aspects of language only (incidentally, this would support our view that syntax deficits in WS result from the selective impairment of the capacity for performing mildly context-sensitive computations). But this possibility is also an empirical one that can be tested experimentally. In any case, we believe that this fresh approach to language (and syntactic) deficits in WS can be an illuminating one if the empirical concerns raised above are correctly addressed. In particular, it should help us to identify the putative core, low level deficit that might explain the deficits observed in this condition (in the language domain, but also in other cognitive domains). In doing so, we will be in the line with the shift of perspective that is currently underway in cognitive sciences (e.g. de Waal & Ferrari, 2010). Acknowledgments Preparation of this work was supported in part by funds from the Spanish Ministry of Economy and Competitiveness (grant numbers FFI-2013-43823-P and FFI2014-61888-EXP to ABB). References Balari, S., & Lorenzo, G. (2013). Computational phenotypes: towards an evolutionary developmental biolinguistics. Oxford: Oxford University Press. Balari, S., Benítez-Burraco, A., Longa, V. M., & Lorenzo, G. (2013). The fossils of language: what are they, who has them, how did they evolve? In C. Boeckx, & K. Grohmann (Eds.), The cambridge handbook of biolinguistics (pp. 489–523).Cambridge: Cambridge University Press. Baddeley, A. (2007). Working memory, thought, and action. Oxford: Oxford University Press. Balari, S., Benítez-Burraco, A., Camps, M., Longa, V., & Lorenzo, G. (2016). Festschrift in Honor of Juan Uriagereka. In A. Gallego, & R. Martin (Eds.), My head’s in knots. On Uriagereka’s Generalization and the knot-sentence connection. Bartke, S. (2004). Passives in German children with Williams syndrome. In S. Bartke, & J. Siegmüller (Eds.), Williams syndrome across languages (pp. 345–370). Amsterdam: Benjamins. Bellugi, U., Bihrle, A., Neville, H., Jernigan, T., & Doherty, S. (1992). Language, cognition, and brain organization in a neurodevelopmental disorder. In M. Gunnar, & C. Nelson (Eds.), Developmental behavioural neuroscience (pp. 201–232).Hillsdale, NJ: Erlbaum Press. Bellugi, U., Korenberg, J. R., & Klima, E. S. (2001). Williams syndrome: an exploration of neurocognitive and genetic features. Clinical Neuroscience Research, 1, 217–229.

60

A. Benítez-Burraco et al. / Journal of Communication Disorders 60 (2016) 51–61

Bellugi, U., Marks, S., Bihrle, A., & Sabo, H. (1988). Dissociation between language and cognitive functions in Williams syndrome. In D. Bishop, & K. Mogford (Eds.), Language development in exceptional circumstances (pp. 177–189).London: Churchill Livingstone. Bellugi, U., Wang, P. P., & Jernigan, T. L. (1994). Williams syndrome: An unusual neuropsychological profile. In S. H. Broman, & J. Grafman (Eds.), Atypical cognitive deficits in developmental disorders: implications for brain functions (pp. 23–56).Hillsdale, NJ: Lawrence Erlbaum. Berwick, R. C., Beckers, G., Okanoya, K., & Bolhius, J. J. (2012). A bird’s eye view of human language evolution. Frontiers in Evolutionary Neuroscience, 4, 5. Blyler, C. R., Gold, J. M., Iannone, V. N., & Buchanan, R. W. (2000). Short form of the WAISIII for use with patients with schizophrenia. Schizophrenia Research, 46, 209–215. Capirci, O., Sabbadini, L., & Volterra, V. (1996). Language development in Williams syndrome: a case study. Cognitive Neuropsychology, 13, 1017–1039. Chien, Y.-C., & Wexler, K. (1990). Children’s knowledge of locality conditions in binding as evidence for the modularity of syntax and pragmatics. Language Acquisition, 1, 225–295. Chomsky, C. (1969). The acquisition of syntax in children from 5 to 10. Cambridge, MA: MIT Press. Chomsky, N. (1981). Lectures on government and binding. Dordrecht: Foris. Clahsen, H., & Almazan, M. (1998). Syntax and morphology in Williams syndrome. Cognition, 68, 167–198. Clahsen, H., & Almazan, M. (2001). Compounding and inflection in language impairment: evidence from Williams Syndrome (and SLI). Lingua, 111, 729–757. Demsky, Y., Gass, C., Edwards, W. T., & Golden, C. J. (1988). Optimal short forms of the Spanish WAIS (EIWA). Assessment, 5, 361–364. de Waal, F., & Ferrari, P. F. (2010). Towards a bottom-up perspective on animal and human cognition. Trends in Cognitive Sciences, 14, 201–207. Eadie, P. A., Fey, M. E., Douglas, J. M., & Parsons, C. L. (2002). Profiles of grammatical morphology and sentence imitation in children with specific language impairment and DS. Journal of Speech, Language, and Hearing Research, 45, 720–732. Fedor, A., Varga, M., & Szathmáry, E. (2012). Semantics boosts syntax in artificial grammar learning tasks with recursion. Journal of Experimental Psychology. Learning, Memory, and Cognition, 38, 776–782. Fitch, W. T., & Hauser, M. D. (2004). Computational constraints on syntactic processing in a nonhuman primate. Science, 303, 377–380. Frick, R. W. (1995). Accepting the null hypothesis. Memory & Cognition, 23, 132–138. Goodluck, H. (1990). Knowledge integration in processing and acquisition: Comments on Grimshaw and Rosen. In L. Frazier, & J. de Villiers (Eds.), Language processing and language acquisition (pp. 369–382).Boston, MA: Kluwer Academic Publishers. Grant, Valian, V., & Karmiloff-Smith, A. (2002). A study of relative clauses in Williams syndrome. Journal of Child Language, 29, 403–416. Heinz, J. (2016). Computational theories of learning and developmental psycholinguistics. In J. Lidz, W. Snyder, & J. Pater (Eds.), The oxford handbook of developmental linguisticsCambridge: Cambridge University Press in press. Heinz, J., & Idsardi, W. (2013). What complexity differences reveal about domains in language. Topics in Cognitive Science, 5, 111–131. Hirsch, C., & Wexler, K. (2006). Children’s passives and their resulting interpretation. In K. U. Deen, J. Nomura, B. Schulz, & B. D. Schwartz (Eds.), The Proceedings of the Inaugural Conference on Generative Approaches to Language Acquisition—North America, Honolulu, HI (pp. 125–136).. Joffe, V., & Varlokosta, S. (2007). Patterns of syntactic development in children with Williams syndrome and Down’s syndrome: evidence from passives and Wh-questions. Clinical Linguistics and Phonetics, 21, 705–727. Kapur, S., Lust, B., Harbert, W., & Martohardjono, G. (1992). Universal grammar and learnability theory: the case of binding domains and the subset principle. In E. Reuland, & W. Abraham (Eds.), Knowledge and language: (Vol. 3. pp. 185–216).Dordrecht: Kluwer Issues in Representation and Acquisition. Karmiloff-Smith, A. (2008). Research into Williams syndrome: the state of the art. In C. A. Nelson, & M. Luciana (Eds.), Handbook of developmental cognitive neuroscience (pp. 691–700).Cambridge, MA: MIT Press. Karmiloff-Smith, A. (2009). Nativism versus neuroconstructivism: rethinking the study of developmental disorders. Developmental Psychology, 45, 56–63. Karmiloff-Smith, A., & Mills, D. L. (2006). Williams syndrome. In K. Brown (Ed.), Oxford: Elsevier. Karmiloff-Smith, A., Tyler, L. K., Voice, K., Sims, K., Udwin, O., Howlin, P., & Davies, M. (1998). Linguistic dissociations in Williams syndrome: evaluating receptive syntax in on-line and off-line tasks. Neuropsychologia, 36, 343–351. Kiparsky, P. (1982). From cyclic phonology to lexical phonology. In H. van der Hulst, & N. Smith (Eds.), The structure of phonological representations (Part I) (pp. 131–175).Dordrecht: Foris. Korenberg, J. R., Dai, L., Bellugi, U., Jarvinen-Pasley, A., Mills, D. L., Galaburda, A., Reiss, A. L., & Pober, B. R. (2008). Deletion of 7q11.23 genes and Williams syndrome. In C. J. Epstein, R. P. Erickson, & A. Wynshaw-Boris (Eds.), Inborn errors of development. The molecular basis of clinical disorders of morphogenesis (pp. 1544–1552).New York, NY: Oxford University Press. Laws, G., & Bishop, D. (2004). Pragmatic language impairment and social deficits in Williams syndrome: a comparison with Down’s syndrome and specific language impairment. International Journal of Language Communication Disorders, 39, 45–64. Levy, Y., (2002) Longitudinal study of language acquisition in two children with Williams syndrome. in: Skarabela, B., Fish, S., & Do, A.H.-J., (eds.), BUCLD 26: Proceedings of the 26th Annual Boston University Conference on Language Development (pp. 348–358). Ithaca, NY: Cascadilla. Levy, J., & Bechar, T. (2003). Cognitive, lexical and morpho-syntactic profiles of Israeli children with Williams syndrome. Cortex, 39, 255–271. Martens, M. A., Wilson, S. J., & Reutens, D. C. (2008). Williams syndrome: a critical review of the cognitive, behavioral, and neuroanatomical phenotype. Journal of Child Psychology and Psychiatry, and Allied Disciplines, 49, 576–608. Mervis, C. B., & Becerra, A. M. (2007). Language and communicative development in Williams syndrome. Mental Retardation and Developmental Disabilities Research Reviews, 13, 3–15. Maratsos, M., Fox, D. E. C., Becker, J. A., & Chalkley, M. A. (1985). Semantic restrictions on children’s passives. Cognition, 19, 167–191. Mervis, C. B., Morris, C. A., Bertrand, J., & Robinson, B. F. (1999). Williams syndrome: Findings from an integrated program of research. In H. Tager-Flusberg (Ed.), Neurodevelopmental disorders (pp. 65–110).Cambridge, MA: MIT Press. Mortimer, J., & Rachew, S. (2010). A longitudinal investigation of morpho-syntax in children with Speech Sound Disorders. Journal of Communication Disorders, 43, 61–76. Öttl, B., Jäger, G., & Kaup, B. (2015). Does formal complexity reflect cognitive complexity? Investigating aspects of the Chomsky Hierarchy in an artificial language learning study. PLoS One, 10, e0123059. Perovic, A., & Wexler, K. (2007). Complex grammar in Williams syndrome. Clinical Linguistics & Phonetics, 21, 729–745. Poeppel, D., & Embick, D. (2005). Defining the relation between linguis-tics and neuroscience. In A. Cutler (Ed.), Twenty-first century psycho-linguistics: four cornerstones (pp. 103–120).Hillsdale: Lawrence Erlbaum. Polite, E. J., & Leonard, L. B. (2006). Finite verb morphology and phonological length in the speech of children with specific language impairment. Clinical Linguistics and Phonetics, 20, 751–760. Ring, M., & Clahsen, H. (2005). Distinct patterns of language impairment in Down’s syndrome and Williams syndrome: the case of syntactic chains. Journal of Neurolinguistics, 18, 479–501. Roberts, J. A., Rice, M. L., & Tager-Flusberg, H. (2004). Tense marking in children with autism. Applied Psycholinguistics, 25, 429–448. Schaner-Wolles, C. (2004). Domain-general or domain-specific cognitive capacities? Language acquisition in Williams syndrome and Down syndrome. In S. Bartke, & J. Siegmüller (Eds.), Williams syndrome across languages (pp. 93–124).Amsterdam: Benjamins. Stavrakaki, S. (2003). Differences in sentence comprehension tasks between children with SLI and Williams syndrome: evidence from Greek. Paper presented at Generative Approaches to Language Acquisition 2003, Utrecht. Stojanovik, V. (2006). Social interaction deficits and conversational inadequacy in Williams syndrome. Journal of Neurolinguistics, 19, 157–173. Tassabehji, M. (2003). Williams–Beuren syndrome: a challenge for genotype-phenotype correlations. Human Molecular GeneticsR229–R237 Special No. 2. Toro, R., Konyukh, M., Delorme, R., Leblond, C., Chaste, P., Fauchereau, F., et al. (2010). Key role for gene dosage and synaptic homeostasis in autism spectrum disorders. Trends in Genetics, 26, 363–372. Thomas, M. S. C., Grant, J., Barham, Z., Gsodl, M., Laing, E., Lakusta, L., Tyler, L. K., Grice, S., Paterson, S., & Karmiloff-Smith, A. (2001). Past tense formation in Williams syndrome. Language and Cognitive Processes, 16, 142–176.

A. Benítez-Burraco et al. / Journal of Communication Disorders 60 (2016) 51–61

61

Udwin, O., & Yule, W. (1991). A cognitive and behavioral phenotype in Williams syndrome. Journal of Clinical and Experimental Neuropsychology, 13, 232–244. Ullman, M. T. (2001). The declarative/procedural model of lexicon and grammar. Journal of Psycholinguist Research, 30, 37–69. Vicari, S., Bellucci, S., & Carlesimo, G. A. (2005). Visual and spatial long-term memory: differential pattern of impairments in Williams and Down syndromes. Developmental Medicine and Child Neurology, 47, 305–311. Volterra, V., Capirci, O., Pezzini, G., Sabbadini, L., & Vicari, S. (1996). Linguistic abilities in Italian children with Williams syndrome. Cortex, 32, 663–677. Wexler, K., & Chien, Y.-C. (1985). The development of lexical anaphors and pronouns. Papers and Reports on Child Language Development, 24, 138–149. Zukowski, A. (2001). Uncovering grammatical competence in children with Williams syndrome. Doctoral Dissertation. Boston University.