Fachbereich Sprachwissenschaft
 Department of Linguistics
Logo der Universität Konstanz
University of Konstanz Department of Linguistics LingLab
Participation in Experiments
Talks and Publications
Qualification Works
Language Processing
Language Processing

To most people understanding one's own language may appear too obvious to consider it as a scientific research topic. Computer scientists are likely to think otherwise. Machines do not understand human language unless it can be successfully transferred into a formal meta-language. Aphasic patients who are not able to comprehend and/or produce language normally would also see the problem.

One of the central goals of psycholinguistic research is to explore the process of language comprehension and to explain it with the help of theoretical models.

Structural linguistics describes and explains the structure of language (or of an individual language) with formally explicit methods and generalizations. Psycholinguistics is a distinct discipline as it studies how language structures (as defined by structural linguistics) develop in the mind/brain of the speaker/hearer in the actual process of listening or reading.

An example from German:

1a. Ich kann ihn nicht leiden.
I can him not stand
"I cannot stand him"

1b. Ich kann ihn nicht leiden sehen.
Ican him not suffer see
"I cannot see him suffer"

(1a) looks as if it was a part of (1b). But this is only true in the trivial sense of phonematic and graphematic overlap. The two sentences differ in structure and in meaning: (1a) could be described in a simple formal language as NOT (CAN (STAND (I, him))); while (1b) could be described as NOT(CAN (SEE (I (SUFFER (he))))). There is evidence that the linguistic analysis and interpretation of sentences starts as the soon as the reader perceives the first word of a sentence. Thus, the structure and meaning of (1a) may well be assigned as (1b) unfolds, with the consequence that part of the structure has to be discarded and reorganized as the sentence continues.

Another German example is given in (2):

2a. Ich bin froh, dass Peter das Auto endlich gekauft hat.
I am glad that Peter the car finally bought has.
"I'm glad that Peter finally bought the car"

2b. Ich bin froh, dass Peter das Auto endlich gekauft wird.
I am glad that Peter the car finally bought was.
"I'm glad that the car will finally be bought for Peter"

Up to the participle gekauft both sentences are indistinguishable. They are, however, structurally ambiguous. This ambiguity is not resolved prior to the appearance of the auxiliary verb (hat versus wird). In (2a) Peter is nominative and das Auto is accusative. Despite morphological identity, case assignment in (2b) is different. Here Peter must be a dative (as shown by dem Peter, dem Mann etc.) and das Auto must be a nominative.

An investigation of the human parser

Language comprehension is a highly automatized and almost reflex-like process. This is quite obvious with frequent words. The total number of words in a language is finite. They just have to be called from our mental lexicon. But it is also a fact that the number of (possible) sentences is infinite, and that therefore sentences (except for formulaic expressions) cannot be memorized or learned like words are learned. Therefore there must be a highly efficient cognitive system for the computation of sentences and their meaning.

It is a major goal of psycholinguistics to characterize this computational system ("parsing mechanism"), and to generalize it to the extent that efficient processing can be explained in languages with different structural properties.

At this point, a major typological aspect comes into play. The languages of the world differ drastically in their word order. This is especially evident when we consider the positioning of the verb. English insists on strict VO-order (ate an apple) while Japanese follows strict OV-order (ringo-o tabeta = apple-ACC ate).German, on the other hand, seems to use both structures, but the underlying structure must be OV, and the finite verb isundergoes leftward movement.

The question is, whether the postulation of argument structure depends solely on the verbal information or whether argument structure can also be predicted efficiently from the arguments themselves.

It is a major and so far only superficially answered question what influence word order has on language comprehension in different languages. As a baseline, one can be certain that there is no language with "dysfunctional" word-order. Such languages would not have been robust enough in communication and would long ago have been erased in competition with "better" languages.

Investigation of language comprehension and the development of a universally valid model of human sentence parsing are desiderata of fundamental research in the cognitive sciences. This endeavour is part of an international research cluster which is located in several laboratories around the world.






Letzte Aktualisierung: 15.01.2015        © 2006-15 : FB Sprachwissenschaft