It hit me yesterday when I was watching an animation movie "the painting" at Goethe Institute, that the human word recognition processing is not an isolated, simple problem, but is a much integrated brain process. In this movie, there is an imagined world in every painting on the wall, and within that world, when a painter did not finish the work, there are three classes of people. The name of the three classes are presented to the audience, without the written forms: All-done, Halfy, and Sketchy(characters in the painting that are all done, half done, and only a sketch). It took me a few minutes and a few instances of usage before I realized that these spoken words correspond to those three written forms that are listed above. The first time the character said "i'm a halfy", I had no idea what the sound 'halfy' corresponds to in English, and how to spell it. Similarly, when they say "she's an all-done", I had no idea that the sound is 'all done', because syntactically and semantically, I had not heard the sound sequence 'all-done' to be used as a noun and i have no experience with a noun in such a syntactic position that maps to any English word I know that sounds like 'all done'. Therefore, I did not successfully complete the task of word recognition even though I, in fact, know these words. What does this tell us? Well, first of all, i think recognition of word in an embedded syntactic structure and semantic/pragmatic context is probably different from recognizing a word in isolation. In this case, in order to map the sound to the form 'all done', a good amount of syntactic, semantic and pragmatic information is needed (also many of those information may not be presented linguistically, for instance, in this movie, visual information will help decoding the visual difference between an 'all-done' and a 'sketchy'). Second of all, it seems like to map the sound 'all done' to the abstract label of 'all done' is not a simple task that is relevant to whether the speech signal is varied or distorted. Imagine you hear a British person, an Indian person and a standard American person saying this. American English sounds the most familiar to me, and it was presented in a movie, kind of a newscaster sound, so it's very standard. I still can't effectively map the sound of 'all done' to any abstract label (in fact, the label of 'all done', or any other label, did not appear to me as a possible candidate for this sound at all). in other words, variation or no variation in speech signal, I simply cannot recognize this word effectively without the corresponding information of syntax, semantics, and pragmatics. You can say that syntax is blocking me from activating the label of 'all done' that it is not even activated slightly as a possible candidate. In this case, it seems that even in the case of exemplar theory, we will have to consider the possibility that in word recognition of the brain, not only the phonetic details and the social information is stored simultaneously, but also syntactic, semantic, and pragmatic information. In other words, it is stored in a truly holistic fashion that if one or more of these aspects (or constraints) are not matched, no matter how unvaried the speech signal is, you still cannot activate and access that label even slightly.
0 Comments
Leave a Reply. |
NEWSLOG
|