Underspecified Phonological Features for Lexical Access

The FUL (featurally underspecified lexicon) system of automatic speech recognition is based on the representation of words in the lexicon with underspecified distinctive features. The speech signal is converted from the waveform into an online spectral representation made up of LPC formants and a few parameters describing the overall spectral shape. These spectral parameters are converted into distinctive phonological features by simple logical decisions. The phonological features, in turn, are compared with all entries in the lexicon. No classification into phone(me) segments, syllables, or spectral templates is used for the selection of words from the lexicon. Comparison of signal features with those stored in the lexicon uses a ternary system of matching, no-mismatching, and mismatching features. Matching features increase the scoring for potential word candidates, no-mismatching features do not exclude candidates and only mismatching features lead to the rejection of word candidates. Activated (i.e., not rejected) word candidates are expanded to include word hypotheses, even without further acoustic evidence, and are used in the phonological and syntactic parsing that operates in parallel with the acoustic front-end. Only after the activation of a word from the lexicon as a possible candidate additional information about it, like prosodic, morphological, syntactic, and, ideally, semantic information becomes available.