Selection
المؤلف:
Paul Warren
المصدر:
Introducing Psycholinguistics
الجزء والصفحة:
P128
2025-11-06
23
Selection
If multiple word candidates have been contacted on the basis of the initial sounds of an input word, then some kind of selection process must choose between them so that a word can be recognised. The illustration in Figure 8.1 suggests that as further bottom-up information becomes available, i.e. as more of the word is heard, then the appropriate word will eventually emerge as the only one that still matches the input. For each word we can identify a uniqueness point, i.e. a point in the word where it no longer overlaps with other words in the initial cohort. ep ending on the speaker’s pronunciation, for captain / kæptən / this point could be the second vowel sound /ə/, where this word becomes distinct from captive /kæptiv/. Theoretical uniqueness points can be determined by searching through pronunciation-based dictionaries or lexical databases. These can be confirmed through gating experiments see above, which indicate that the actual recognition points of words correlate highly with these unique ness points. There has been some debate about the mechanism that results in the decision that a particular word has been heard. In our illustration this might require that a word’s activation passes some threshold level. Alternatively, there may need to be some difference between the relative probabilities of the words in the cohort expressed as some measure of the differences between the curves in Figure 8.1.
But what about nonwords – how are these recognised’ That is, how do we know when we hear slundle that it is not a word of our language Clearly there is no stored representation in our mental lexicon for a word that does not exist. It is claimed, though, that the recognition of nonsense words follows the same process as the recognition of real words. This is a process that is optimally efficient, in that decisions are made as rapidly as possible. It makes sense that the same process should apply to real and nonsense words, because we do not know before we hear a word whether it will be one that we know or not, so we are not able to deploy different strategies in advance of hearing it. In the case of a nonword input, a set of words matching the initial input is activated, and the activation levels of the members of this set change as more of the input is heard and processed, until there is no remaining strong candidate word. This will be at the point in the nonsense word where it diverges from known words. This is known as the deviation point.
The significance of the deviation point was first demonstrated in an experiment in which participants were required to press a response but ton whenever they heard a nonsense word in a list of stimulus words. It was found that response times were reasonably constant at around 450 msec when measured from the deviation point (Marslen-Wilson, 1980). Other studies, however, have cast a shadow of doubt over that result, since they have found some variation in the response times from the deviation point, variation which depends both on the nature of the task that participants are doing and also on the word-likeness of the material after the deviation point (Goodman & Huttenlocher, 1988; Taft Hambly, 1986). It seems that listeners continue to monitor the input after the deviation point, so that a more word-like ending, such as a regular affix like se or - will slow down the no’ responses in the lexical decision task. Such continued monitoring of the input is of practical value in ordinary speech comprehension situations, since we are able to recognise words even under noisy conditions e.g. at parties or with machinery operating in the background. We can also recognise words when sounds within them are mispronounced. As we will see below, this is facilitated by the use of information from the utterance context in which the word is heard.
الاكثر قراءة في Linguistics fields
اخر الاخبار
اخبار العتبة العباسية المقدسة
