Production and perception of gender agreement in French SLI

Abstract

In a previous work on the production and the perception of gender agreement in determiner phrases constituted by a determiner and a noun, we showed that despite their agreement errors in the production task, children with specific language impairment (SLI) were sensitive to agreement violations in the perception task. In this study, we asked whether a slightly “heavier” determiner phrase including a noun, a determiner and a (pre- or post-nominal) adjective (i) would increase the number of agreement errors in production and (ii) prevent or not sensitivity to agreement violations to show up in perception. To answer these questions, 14 French-speaking children with SLI and three groups of controls participated in an elicited production task and in a categorization task testing sensitivity to gender agreement relations in perception. Our main results showed that children with SLI produced significantly more agreement errors than controls in production. In all groups, agreement errors were relatively more frequent for determiner phrases including a pre-nominal adjective than for determiner phrases including a post-nominal one. Secondly, children with SLI as well as controls were slower and less accurate to categorize determiner phrases when the noun and the determiner/adjective disagreed than when they agreed. We concluded that (i) the number of agreement errors in production varied in relation to the size and structure of the determiner phrase and that (ii) a determiner phrase including a noun, a determiner and an adjective is not “heavy” enough to prevent children with SLI from being sensitive to agreement relations.

Keywords:

• SLI
• gender agreement
• production and perception

Notes

1 According to Tucker, Lambert, and Rigault (1977), French words vary according to the high or low predictive value of the noun ending. Corrêa and Name (2003) showed that Brazilian-Portuguese 2-year-olds identify the gender of the noun on the basis of the morphophonological information provided by the determiner. In our previous work, we showed that for children with SLI as well as for controls, the agreement effect was not affected by the predictive value of the noun ending (Roulet, in press; see also Monnery, 2001, for similar results with 6 to 8-year-olds in French).

2 French determiners are of two sorts: definite articles and possessives taking the word marker -a for feminine (see (a)); indefinite determiners and demonstratives adding a “schwa” so that the consonant gets pronounced similarly to adjectives in the feminine (see (b)).

1. La / ta / sa petite table

   “The-fem / your-fem / his or her-fem little table”

2. Une / cette petite table

   “A-fem / this-fem little table”

3 Emphasis added.

4 In the framework of the Minimalist Program adopted here, there is one unique recursive syntactic operation Merge that either applies on an element taken from the lexicon (External Merge) or an element already merged in the derivation (Internal Merge or Move). Internal Merge requires Agreement between a feature of the element that is moved from its original position and a feature of an element in its landing position (see Chomsky, 2005). An anonymous reviewer asked whether in Chomsky's framework, there exists any theoretical/conceptual link between the number of Merges and Agree. The response is “no”. Please note, however, that the notion of syntactic complexity that is used here comes from our work on language production and perception in SLI and is not therefore the same notion of complexity used in generative theoretical linguistics. In the present study, by increasing the number of nominal elements within the DP, we equally increase the number of Merges and Agrees (see (a)). Obviously, it is not always the case and it is possible to increase the number of Merges without increasing the number of Agrees (see (b)).

1. La toute petite maison

2. La très petite maison

In the first example, there are as many Merges as Agrees whereas in the second example, there are three Merges but only two Agrees due to the fact that “très” (very) is an adverb and does not agree with the noun.

5 According to Giorgio & Lomgobardi (1991), in French and more generally in Romance languages, the noun moves over the adjective: it has to leave its basic position of structural complement of the adjective (from the right side of the adjective) to reach a functional position between the determiner and the adjective.

6 Age is expressed in years and months (years;months).

7 Five per cent of adjectives were produced within relative clauses (for all groups) or omitted (only by some of the TD-4 children).

8 In our previous study, as in the present one, children elicitation of determiner phrases was based on drawings presented on a computer screen. Previously, children were asked a question and should produce determiner phrases answering this question whereas in the present study, they were asked to complete the sentential context given by the experimenter.