Cues to lexical choice: Discriminating place and voice

Abstract

Two experiments used the gating paradigm to investigate the manner in which acoustic-phonetic information is mapped onto the lexical level during the processes oflexical access and selection. The first experiment tested word identification across successive 25-msec gates of monosyllables contrasting in word-final voicing and showed a continuous uptake of durational cues. The second experiment expanded upon earlier research into the uptake of partial cues in the spectral domain and revealed strong effects at vowel closure on the choice between word candidates terminating in different places of articulation. The results were interpreted as suggesting a contrast between symmetric and asymmetric decision processes, with phonological structure being the potential source of asymmetries in the lexical interpretation of acoustic cues.The process of spoken language comprehension begins with the projection of the speech input onto mental representations of lexical form. The purpose of the research reported here is to explore the properties of the acousticphonetic decision space within which the listener conducts this process of lexical access and selection. In earlier research (Warren & Marslen-Wilson, 1987), we showed that the fine-grained detail of variation in the speech signal is continuously projected onto the lexical level. Listeners do not need to wait until the end of a segment, as conventionally defined, in order to guide and constrain lexical choice. In particular, they seem to be able to exploit online the temporal overlap, in the speech signal, of acoustic cues to distinct speech segments (see Fowler, 1984).In our first study (Warren & Marslen-Wilson, 1987), we looked at the consequences for lexical choice of temporally overlapping cues in the spectral domain. To do this, we used a gating task in which listeners heard successively larger fragments of the initial consonant cluster and vowel of words such as scoot or scoop, in which the final consonant differed in place of articulation, and of words like crown or crowd, in which the final consonant differed in manner of articulation. The subjects in this task were able to use coarticulatory changes in the formant structure of the vowel to help them identify the correct word before the final consonant was heard.We thank Marie Jefsioutine for her help in designing and constructing the stimuli and in running the experiments. We also thank Aditi Lahiri and Uli Frauenfelder for their comments on the manuscript. This research was supported by a programme grant from the Medical Research Council. Address reprint requests to Dr. Paul Warren, University of Cambridge, Department of Experimental Psychology, Downing St., Cambridge CB2 3EB, England.The research reported here expands on the previous study in two ways: by looking at a more differentiated set of contrasts in place of articulation, and by examining a different kind of partial cue-namely, variations in relative duration for vowels preceding voiced and unvoiced stops. This research sheds light not only on the kinds...

The temporal dynamics of ambiguity resolution: Evidence from spoken-word recognition

“…More often, however, the evidence appears partial, and information must be accumulated over longer stretches in order to discriminate between lexical candidates with any degree of certainty. This fits in with numerous previous demonstrations of sensitivity to coarticulation in the perception of spoken words (e.g., Warren & Marslen-Wilson, 1987, 1988.…”

Self-Organizing Dynamics of Lexical Access in Normals and Aphasics

“…A great deal of research has been conducted exploring the influence of sound structure variations on the processes of word recognition and lexical access (cf. Kessinger, 1998;Connine, Titone, Deelman, & Blasko, 1997;Utman, 1997;Zwisterlood, 1996;Andruski, Blumstein, & Burton, 1994;Warren & Marslen-Wilson, 1987, 1988Streeter & Nigro, 1979). On balance, these results have shown that indeed listeners are sensitive to phonological as well as within phonetic category differences, and importantly, that these differences affect both word recognition and lexical access processes.…”

“…And yet, listeners seem to harness the variability in the speech stream and are able to rely on the fine acoustic details intrinsic in the variation of speech in the process of word recognition (Warren & Marslen-Wilson, 1987, 1988; see also Marslen-Wilson, 1978, 1989. In particular, listeners monitor the acoustic signal continuously, not waiting until the end of a segment in order to guide or constrain their lexical choice.…”

Auditory word identification in dyslexic and normally achieving readers

“…Therefore, these two words sound quite similar until late in the vowel. For a listener to know that the final consonant is a voiceless oral stop (at any place of articulation) already shows sensitivity to a good deal of acoustic phonetic information (Warren & Marslen-Wilson, 1988). Yet an incorrect response is not given partial credit, so to speak, for the strong resemblance of /kaet/ to /kaep/, as opposed to /kaen/ or /kaemp/, which would have substantially nasalized and lengthened vowels.…”