Comprehension of Reversible Relative Clauses in Specifically Language Impaired and Normally Developing Greek Children
Pilot study: Comprehension of Relative Clauses in German Children with SLI
Term Paper 2010 10 Pages
1. Experimental hypothesis method and results of Stavrakaki´s study (2001)
Referring to authors like van der Lely and Harris (1990) Stavrakaki (2001) points out that specific language impaired (SLI) children`s performance in expressive and comprehensive language skills differ from language and chronological age counterparts. Because of that reason she also claims that SLI can´t be treated as a delay in the language acquisition process. Stavrakaki (2001) constructed this study to prove this hypothesis and to test the comprehension of reversible relative clauses in Greek SLI children. In her current study she wanted to examine wether SLI children´s comprehen- sion of reversible relative clauses differs from that of their chronological and language age peers. If there occurs any difference she also wants to find in what way comprehension of SLI children dif- fers especially concerning their competence in syntactic and morphological knowledge.
Furthermore Stavrakaki (2001) characterizes the human parser as follows. First the parser has some local preference which means long-distance dependencies are harder to parse than short ones. Sec- ond filled gaps are avoided. The parser prefers empty gaps over processing a full NP. Third there is also a language specific property of Greek. The phonological properties of the language like case marking will guide the parser through the sentence and enable him to undertake theta role assign- ment. In conclusion the violation of the preferences of the parser will direct to difficulties in proc- essing these kinds of sentences. But language competence e.g. knowledge of case marking will re- duce the parsers problems.
The participants are divided in three groups. First group contains eight SLI children who in mean 7.38 years. Each SLI child was examined with a Verbal IQ Test for Greek children (Stavrakaki & Tsimpli, 1999) and on base of the raised data matched with two language-matched children. These 16 language-matched children represented the second group. Their mean age was 4.1 years. The third group contained eight age matched children whose mean age was 7.37 years. Stavrakaki (2001) used seven sentence types in her study (see table 1 p. 3). Four of the used sen- tences are comparable with the german relative constructions. Object head-subject gap (O-S), object head object gap (O-O), subject head-subject gap (S-S), subject head-object gap (S-O) rank to these sentences. The other three (object head with clitic pronoun with and without case marking and ob- ject head with preverbal subject) are special Greek construction with a filled gap in the relative clause.
The comprehension of the semantically reversible sentences were examined in an acting out task with toys. Whereby Stavrakaki (2001) used two toys which were compatible with the head of the restrictive relative.
Abbildung in dieser Leseprobe nicht enthalten
The results for all children corroborate the theory of the human parser Stavrakaki (2001) made. Whereas the SLI children`n competence was more effected by the parsers preference. Concerning the point that the parser avoids long distances all children had difficulties in interpreting S-O sentences. Nevertheless the age matched and even the language matched controls showed better performance in this sentence construction than the SLI children. Furthermore only children with SLI had real difficulties with the filled gap sentences. The control groups showed full grammatical competence in using case morphology to interpret O-O sentences with clitic and case marking. Moreover the presence of the filled gap did not effect their performance.
After interpreting the results Stavrakaki (2001) made following conclusions:
First SLI children did reach the performance of the language matched controls only in structures with low processing demands (like O-O sentence). Second parser preference violations effected SLI children´s performance much more than typical developed ones because SLI children are not able to use grammatical knowledge to overcome those difficulties. And third the performance of SLI chil- dren is qualitatively different from typical developing children which corroborates van der Lely´s and Harris´ (1990) assumption.
2. German pilot study
Before the pilot study is presented there are to distinguish some special remarks on german relative clauses. In general in German is existing free word order because of rich morphology. The relative clauses can occur centre embedded or right branching. German relative sentences are always con- taining relative pronouns which are referring to a head of the main clause whereas the relative pro- noun is always marked for case but only overt marked in masculine gender. The subject in german clauses is marked for nominative and the object is marked for accusative case. So the parser can assign theta roles because of case marking. In case of feminine and neuter gender there is some ambiguity in the object position. The parser can solve the ambiguity by using grammar knowledge in case there is a masculine subject which has to be overtly marked by nominative case or accusative case in object position (see example 1).
Abbildung in dieser Leseprobe nicht enthalten
If there is no masculine subject or object the theta role assignment must be done by world knowledge ( e.g. if there is an animate actor doing something to an inanimate object) or context knowledge.
The hypothesis for the pilot study will be based on the results of Stavrakaki (2001) who found that SLI could not use knowledge of grammar to interpret relative clauses and undertake theta role as- signment. We suppose that any default structure(e.g SOV) would occur during interpretation of ob- ject- object and subject-object relative clauses. Furthermore we will also use neuter gender actors to be able to make forced predictions about SLI because for normal developing children sentences containing nouns in neuter gender might be harder than all masculine sentences. For SLI children we expect no difference between these two conditions. Because of the same behavior of neuter and feminine in case marking in accusative and nominative (concerning non overt case marking) we decided to use only neuter gender animals.
The participants of our study will be also divided in three groups. First group will contain SLI chil- dren who will be about 5 to 7 years old. They will be tested by thePatholinguistische Diagnostik (Kauschke & Siegmüller, 2002) to determine mean age in language abilities. The second group will contain the language matched children based on the results of the diagnostics for the SLI children. There will be the double count of language matched children comparing to the SLI group to get best compliance in language abilities. The third group will be composed of age matched children who will be the same count as the SLI group. To avoid that they have problems with the length of the sentences in case of a decrease of memory. we also would use a memory test likeMottier-Test (Mottier, 1951).