 |
 |
| University of
Glasgow
Department of Psychology |
|
|
| University of
Glasgow
Department of Psychology |
Discussion extract
What can account for the different response latencies of English and German speakers are the different weights of grammatical cues (word order, animacy, number, tense, case etc.) within and between both languages. Grammatical cues are necessary to understand a sentence. Understanding a sentence involves the following processes. Grammatical constituents (subject, verb, object etc.) have to be encoded as well as thematic roles (agent, theme), i.e. a syntactic and a semantic representation must be build up. The former one relates to the form of the sentence while the latter relates to the function of its words but they do not correspond necessarily one-to-one. Therefore, a mapping process is performed that results in a grammatical sentence representation. This is called the mapping hypothesis (Marshall et al. 1987). In order to map the syntactic onto the semantic representation (or the form onto the function) the grammatical cues are used. ...
... At this point the question emerges why do differences diminish with the increase of constituent comparisons. It appears that the number of grammatical cues that are available in German but not in English stays the same across all conditions. Importantly, the total number of cues to be processed increases as it shall be shown below. That means, the contribution of the different processing of grammatical cues to response latencies decreases with sentence type (from affirmative to embedded sentences). The ratio as estimated above for English to German is 2 : 8 for affirmative, 3 : 9 for negative and 7 : 13 for embedded sentences, yielding a percentage of availabel cues in English of
25 %, 33 % and 54 %, respectively. See Figure 2 (below) for the ratio of cues in both languages. For negative sentences the increase is due to the negation, which must be encoded additionally and thus is counted to affect the processing time to the same amount as one grammatical cue. For embedded sentences the situation is even more difficult. Not only that the share of additional cues in German is decreased (46 % as opposed to 75 %), moreover, these sentences include a relation between their two parts or phrases as it was pointed out earlier. The share of encoding time due to the relation, cannot be estimated but it can be understood as encoding additionally
(recoding ?) the first phrase into an affirmative marker, as it is represented in the CCM. Of course, the processing of that relation is a time consuming process, which results in an increase of response latency by a certain amount x. Note that the state of affairs for the second phrase is slightly different as with the short sentences. Tense and voice are selected at the fourth (relative) position in both languages, which is due to the reversal of the words “rot sind” as opposed to short sentences (“sind rot”). On the contrary, number agreement between the noun and the verb can be checked only at the fourth position in German as opposed to the third position in English. In the first phrase all cues are equal apart from tense. That can be selected at the fourth position in English and at the fifth position in German. However, the decreased share of cues that are available in German but not in English (in long sentences) results in smaller differences in encoding time and thus in similar response latencies between English and German speaking subjects. These results suggest an advantage for English native speakers over German native speakers with respect to the encoding of affirmative sentences that disappears for embedded sentences. In the first case they benefit from the small number of grammatical cues that they have to process.
Figure 2: Number of cues that must be processed to encode the presented sentence for each language. Note that the difference between English and German is constant across sentence type while the total number increases Hence, the advantage of English is reduced with longer sentences.
(c) Dirk Köster; first posted January 06/01
last modified January 06/01
