Subjects were shown the terms of simple sentences in sequence (e.g., “A sparrow / is not / a vehicle”) and manually indicated whether the sentence was true or false. When the sentence form was affirmative (i.e., “X is a Y”), false sentences produced scalp potentials that were significantly more negative than those for true sentences, in the region of about 250 to 450 msec following presentation of the sentence object. In contrast, when the sentence form was negative (i.e., “X is not a Y”), it was the true statements that were associated with the ERP negativity. Since both the false‐affirmative and the true‐negative sentences consist of “mismatched” subject and object terms (e.g., sparrow / vehicle), it was concluded that the negativity in the potentials reflected a semantic mismatch between terms at a preliminary stage of sentence comprehension, rather than the falseness of the sentence taken as a whole. Similarities between the present effects of semantic mismatches and the N400 associated with incongruous sentences (Kutas & Hillyard, 1980) are discussed. The pattern of response latencies and of ERPs taken together supported a model of sentence comprehension in which negatives are dealt with only after the proposition to be negated is understood.

No abstract

No abstract

No abstract

This paper discusses the use of the Hidden Markcv Model (HMM) in phonetic recognition. In particular, we present improvements that deal with the problems of modeling the effect of phonetic context and the problem of robust pdf estimation. The effect of phonetic context is taken into account by conditioning the probability density functions (pdfs) of the acoustic parameters on the adjacent phonemes, only to the extent that there are sufficient tokens of the phoneme in that context.This partial conditioning is achieved by combining the conditioned and unconditioned pdfs models with weights that depend on the confidence in each pdf estimate. This combination is shown to result in better performance than either model by itself. We also show that it is possible to obtain the computational advantages of using discrete probability densities without the usual requirement for large amounts of training data.