Grammatical Abilities of Greek-Speaking Children with Autism

Abstract

This study investigates pronoun reference and verbs with nonactive morphology in high-functioning Greek-speaking children with Autism Spectrum Disorders (ASD). It is motivated by problems with reflexive pronouns demonstrated by English-speaking children with ASD and the fact that reflexivity is also expressed via nonactive (reflexive) verbs in Greek. Twenty 5- to 8-year-old children with ASD and 20 vocabulary-matched typically developing controls of the same age range completed a sentence-picture matching, an elicitation, and a judgment task. Children with ASD did not differ from controls in interpreting reflexive and strong pronouns but were less accurate in the comprehension of clitics and omitted clitics in their production. The findings render clitics a vulnerable domain for autism in Greek, and potentially for other languages with clitics, and suggest that this could be a consequence of difficulties in the syntax-pragmatics or the syntax-phonology interface. The two groups did not differ in the comprehension of nonactive morphology but were less accurate in passive than reflexive verbs. We argue that this is likely to stem from the linguistic representation associated with each type of verb, rather than their input frequency.

ACKNOWLEDGMENTS

We would like to thank the clinics in Athens, Messolongi, and Pyrgos for allowing us to recruit the children with ASD; public schools 19 and 60, and public kindergarten 42 in Patras, for allowing us to recruit TD children; and the children and their parents for taking part in the study. Earlier parts of this work were presented at the 7th European Group for Child Language Disorders Conference (Eucldis) (Chalkidiki, May 2011), at the University of Reading (July 2011), the 3rd Conference for Autism Spectrum Impairments (Messolongi, October 2011), the 36th Boston University Conference on Language Development (November 2011), the Workshop on Pronoun Binding and Nonactive Morphology in Greek (Technological Educational Institute of Patras, November 2011), the Linguistics Meeting at the University of Ioannina (May 2012), and the Patras Goes Psycho Meeting at the University of Patras (June 2012). We would like to thank the audiences of these events for their comments and suggestions. Thanks also to Dimitra Bafa for collecting the data and to Sofi Miliopoulou for drawing the pictures. Last but not least, we would like to thank three Language Acquisition reviewers for their insightful comments. Errors or omissions are our own responsibility, of course.

Notes

¹A reviewer comments that him can be interpreted as coreferential with Bill in the last conjunct of (4) only if it is replaced by Bill in its first occurance, unlike what is argued to be the case in the related literature (see Conroy et al. 2009 for a recent survey). It seems to us that even then, coreference is impossible in Greek.

²See Spathas (2010) for additional differences between Greek and English reflexive pronouns.

³Three children with ASD initially recruited for the study were not included in this article because their Raven's score was below 80. The phrase “ … 5- to 8-year-old children … ” is misleading, since only 1 of the 20 children of each group was over 8 years old. Mean ages are a more accurate manner to characterize the age of the participants. In any case, the actual ages of the ASD children were 5;06, 5;08, 5;09, 6;01, 6;02, 6;06, 6;08 (3), 6;09, 6;10, 6;11, 7;01, 7;02 (2), 7;03, 7;07 (2), 7;11, 8;11, and the ages of TD children were: 5;06, 5;08, 5;09, 6;01, 6;02, 6;06, 6;08 (3), 6;09, 6;10, 6;11, 7;01, 7;02 (2), 7;03, 7;07 (2), 7;11, and 8;00.

⁴A reviewer comments that the same verbs should have been used in all three conditions, and moreover, they should be verbs that do not express reflexivity with nonactive morphology. The first suggestion is fulfilled to a large extent, since we used the same verbs in all three conditions, except for three verbs in the reflexive pronouns condition, which were new. As to the second suggestion, it cannot be considered, because it was our intention to compare reflexivity as expressed with reflexive pronouns and reflexivity as expressed via verbal morphology, precisely because of the problems encountered with English reflexive pronouns. After considering the results, it emerges that the findings are not challenged by any of the comments. Binding of reflexive pronouns is almost at ceiling, while the conditions testing binding of strong pronouns and clitics, which could potentially be an issue, since performance of ASD and TD children on clitics is different, do indeed include the same verbs.

⁵A reviewer expressed doubts as to whether children could remember the names of the persons. Our experience was that they remembered them extremely accurately, and, moreover, even from their first occurrence. This was also true for the control group of the college students we tested; nevertheless, we routinely repeated the names in every single slide. It should also be pointed out that if children could not remember the names, this would presumably affect all conditions/sentences, contrary to fact.

⁶The accuracy of two children with ASD in the comprehension of clitics was at 50%, which is 2 standard deviations below the mean of the group (88.3%). Therefore, they could be classified as outliers. To investigate whether the significant difference between the children with ASD and the two control groups was caused by these two children, we excluded their data and reran the same analyses. These analyses showed also a significant difference between the children with

ASD and the two control groups (both comparisons: p = 0.021), indicating that the difference on clitics was not caused by these two children.

⁷The binding results alone, along with a less-extensive discussion than the current one, are included in Terzi et al. (2012). This work does not include an adult control group, nor the follow-up experiments on clitics that are reported in the following sections of the present article.

⁸In Reinhart & Reuland (1993) an NP is defined as +R if and only if it carries full specification for phi-features (number, gender, person).

⁹After the article was submitted for publication, one of the reviewers pointed out to us a new study of Greek-speaking children with SLI reporting children with SLI who did worse on binding of reflexive than on binding of clitic pronouns (Varlokosta & Nerantzini 2012). This is the opposite pattern from the one attested in the study by Stavrakaki & Lely (2010) for children with SLI and the findings in the present study for high-functioning children with ASD. This contrast is very likely to be due to the different methodologies used. Varlokosta & Nerantzini (2012) used a binary picture-selection task that does not include the condition on which the children with SLI in Stavrakaki & Lely and the ASD children in our study committed the majority of errors on clitics. Therefore, the studies are not directly comparable. If such a pattern is indeed established for children with SLI, it would be interesting to see whether problems with reflexives hold across various types of atypical language and why, as the reviewer suggested.

¹⁰The weaker statistical difference between children with ASD and TD children on feminine subjects as compared to masculine subjects is likely to be due to the large range on responses in feminine subjects for both groups and the ceiling effect on masculine subjects for the TD children.

¹¹A reviewer commented that the verbs we tested in this condition do not have only a passive interpretation, hence, they cannot be considered as passives, because most of them make perfect reciprocals if used in the plural. This is indeed true, however, we tested them in the singular, where no such ambiguity arises. It would admittedly be extremely interesting to investigate these verbs in the plural as passives, as well as reciprocals, on a par with Conditions 4 and 6 of the current experiment and compare the results. Obviously, such a task could not be undertaken within the confines set by the initial aims of this work.

¹²A reviewer comments that not all native speakers accept some of the passive sentences employed in the task, in particular, those containing the verbs ‘pinch,’ ‘feed,’ and ‘kiss.’ This observation may be able to explain low performance of children on ‘kiss,’ but not on ‘kick.’ Moreover, performance on the first two verbs that do not make felicitous passives according to the reviewer was not particularly low.

¹³Views that attribute early appearance of passives to their high occurrence in the input children receive are held by Demuth (1989), and subsequent work, for Sesotho. However, a recent study of Sesotho passives found that Sesotho-speaking children in fact perform no better than English-speaking children, although the frequency of passives in Sesotho is much higher than in English (Crawford 2012).

¹⁴Such views of reflexives contrast with views such as of Embick (2004), who considers reflexives like unaccusatives in the sense that they both involve derived subjects.

¹⁵A reviewer comments that passives are often better in the past tense in Greek, referring to observations due to Zombolou (2004). This may be true, but the passive condition that employed reflexive verbs, i.e., Condition 6, also used verbs in nonpast, yet, performance was much higher than on the passives of Condition 4.