TY  - THES
T1  - Shared representations in language processing and verbal short-term memory: The case of grammatical gender
A1  - Schweppe,Judith
Y1  - 2006/11/23
N2  - The present study investigates short-term memory for syntactically structured verbal information. Five experiments have demonstrated that grammatical gender contributes substantially to short-term recall of sentences and short texts. 
This study is based on models that assume that comprehension, production, and retention of verbal materials share the same representations (language-based models of working memory, e.g., Martin & Saffran, 1997). Findings that lexical features (such as phonological similarity and word frequency) influence recall of word lists are regarded as evidence for language-based models. Yet other models that do not assume such a strong relationship between language processing and verbal short-term memory can also account for these findings. Moreover, several linguistic features (such as syntactic or propositional aspects) are relevant only when processing sentences. Hence, their contribution to short-term memory tasks should be addressed best when using sentences instead of word lists. 
The main basis of this thesis is an elaborated version of a language-based model of verbal retention that particular addresses short-term memory for sentences (Rummer, 2003). Its basic assumption is that those types of information are retained that are potentially required for language comprehension or production. They are assumed to be retained as long as they are potentially required. 
The present study investigates the model';s assumption by means of a type of representation that has not been investigated in a short-term memory context so far: grammatical gender. However, based on its role in language processing, grammatical gender should also play a role in verbal retention and in particular in retention of syntactically structured materials. For instance, grammatical gender information is used for determining the right antecedent for a pronoun. As nouns and pronouns are often separated by several words or even sentences, access to gender information is required for a rather long period of time, similar to conceptual information. 
In order to test for an influence of grammatical gender information on verbal retention, I have used a modified version of Potter and Lombardi';s (1990) intrusion paradigm. The intrusion paradigm combines sentence recall and word list retention. In half of the cases, the word lists include a "lure word';, which is semantically related to a target word in the sentence and which is a better fit than the target word. One investigates whether and how often the target word is substituted by the lure word and whether the intrusion rate increases due to presentation of the lure word on the distractor list (e.g., sentence: "Der Chauffeur passte auf das Fahrzeug auf, während er auf seinen Chef wartete." /English gloss: "The driver watched over the automobile while he was waiting for his boss.';; lure word: "Auto"/';car';). As a modification, one can construct the sentences so that two lure words exist that are a better fit than the target word, one of them being gender congruent and the other one being gender incongruent. In the above example "Auto';/';car'; (gender congruent) and "Wagen';/';vehicle'; (gender incongruent) are potential lure words. If gender information has an influence on sentence retention, there should be more intrusions by the gender congruent than by the gender incongruent lure word.
Such an error pattern (termed a gender congruency effect) could indeed be observed in the five experiments of the present study. The gender congruency effect occurred independently of the modality of sentence and list presentation (Exp. 1 & 5: auditory presentation, Exp. 2, 3 & 4: visual presentation) and independently of whether sentence recall immediately followed sentence presentation (Exp. 1, 2 & 4) or was briefly delayed (Exp. 3). In addition, the gender congruency effect was not influenced by the morphophonological form of the target word';s determiner. In other words, the effect could even be observed when there was no discrepancy between the target determiner and the gender incongruent lure word. As a further result, the effect also occurred under conditions in which the critical sentence was followed by one more sentence (Exp. 5). This finding suggests that gender information has an influence on retention for a longer period of time. This is what was predicted on the basis of the model described above. Yet this finding contrasts with the widespread assumption that surface information is forgotten beyond sentence boundaries (e.g., Jarvella, 1971). In sum, the findings support the assumption that features relevant in language processing are also relevant in verbal short-term memory.
KW  - Arbeitsgedächtnis
KW  - Genus
KW  - Kurzzeitgedächtnis
CY  - Saarbrücken
PB  - Universitäts- und Landesbibliothek
AD  - Postfach 151141, 66041 Saarbrücken
UR  - http://scidok.sulb.uni-saarland.de/volltexte/2006/851
ER  -