Publications

Abstract

This study tests whether potential differences in the perceptual robustness of speech sounds influence continuous-speech processes. Two phoneme-monitoring experiments examined place assimilation in Korean. In Experiment 1, Koreans monitored for targets which were either labials (/p,m/) or alveolars (/t,n/), and which were either unassimilated or assimilated to a following /k/ in two-word utterances. Listeners detected unaltered (unassimilated) labials faster and more accurately than assimilated labials; there was no such advantage for unaltered alveolars. In Experiment 2, labial–velar differences were tested using conditions in which /k/ and /p/ were illegally assimilated to a following /t/. Unassimilated sounds were detected faster than illegally assimilated sounds, but this difference tended to be larger for /k/ than for /p/. These place-dependent asymmetries suggest that differences in the perceptual robustness of segments play a role in shaping phonological patterns.

Abstract

Listeners heard an ambiguous /f-s/ in nonword contexts where only one of /f/ or /s/ was legal (e.g., frul/*srul or *fnud/snud). In later categorisation of a phonetic continuum from /f/ to /s/, their category boundaries had shifted; hearing -rul led to expanded /f/ categories, -nud expanded /s/. Thus phonotactic sequence information alone induces perceptual retuning of phoneme category boundaries; lexical access is not required.

Abstract

A Bayesian model of continuous speech recognition is presented. It is based on Shortlist ( D. Norris, 1994; D. Norris, J. M. McQueen, A. Cutler, & S. Butterfield, 1997) and shares many of its key assumptions: parallel competitive evaluation of multiple lexical hypotheses, phonologically abstract prelexical and lexical representations, a feedforward architecture with no online feedback, and a lexical segmentation algorithm based on the viability of chunks of the input as possible words. Shortlist B is radically different from its predecessor in two respects. First, whereas Shortlist was a connectionist model based on interactive-activation principles, Shortlist B is based on Bayesian principles. Second, the input to Shortlist B is no longer a sequence of discrete phonemes; it is a sequence of multiple phoneme probabilities over 3 time slices per segment, derived from the performance of listeners in a large-scale gating study. Simulations are presented showing that the model can account for key findings: data on the segmentation of continuous speech, word frequency effects, the effects of mispronunciations on word recognition, and evidence on lexical involvement in phonemic decision making. The success of Shortlist B suggests that listeners make optimal Bayesian decisions during spoken-word recognition.

Abstract

Dutch listeners' looks to printed words were tracked while they listened to instructions to click with their mouse on one of them. When presented with targets from word pairs where the first two syllables were segmentally identical but differed in stress location, listeners used stress information to recognize the target before segmental information disambiguated the words. Furthermore, the amount of lexical competition was influenced by the presence or absence of word-initial stress.

Abstract

Segmental as well as suprasegmental information is used by Dutch listeners to recognize words. The time-course of the effect of suprasegmental stress information on spoken-word recognition was investigated in a previous study, in which we tracked Dutch listeners' looks to arrays of four printed words as they listened to spoken sentences. Each target was displayed along with a competitor that did not differ segmentally in its first two syllables but differed in stress placement (e.g., 'CENtimeter' and 'sentiMENT'). The listeners' eye-movements showed that stress information is used to recognize the target before distinct segmental information is available. Here, we examine the role of durational information in this effect. Two experiments showed that initial-syllable duration, as a cue to lexical stress, is not interpreted dependent on the speaking rate of the preceding carrier sentence. This still held when other stress cues like pitch and amplitude were removed. Rather, the speaking rate of the preceding carrier affected the speed of word recognition globally, even though the rate of the target itself was not altered. Stress information modulated lexical competition, but did so independently of the rate of the preceding carrier, even if duration was the only stress cue present.

Abstract

Prosodic influences on phonetic realizations of four Dutch consonants (/t d s z/) were examined. Sentences were constructed containing these consonants in word-initial position; the factors lexical stress, phrasal accent and prosodic boundary were manipulated between sentences. Eleven Dutch speakers read these sentences aloud. The patterns found in acoustic measurements of these utterances (e.g., voice onset time (VOT), consonant duration, voicing during closure, spectral center of gravity, burst energy) indicate that the low-level phonetic implementation of all four consonants is modulated by prosodic structure. Boundary effects on domain-initial segments were observed in stressed and unstressed syllables, extending previous findings which have been on stressed syllables alone. Three aspects of the data are highlighted. First, shorter VOTs were found for /t/ in prosodically stronger locations (stressed, accented and domain-initial), as opposed to longer VOTs in these positions in English. This suggests that prosodically driven phonetic realization is bounded by language-specific constraints on how phonetic features are specified with phonetic content: Shortened VOT in Dutch reflects enhancement of the phonetic feature {−spread glottis}, while lengthened VOT in English reflects enhancement of {+spread glottis}. Prosodic strengthening therefore appears to operate primarily at the phonetic level, such that prosodically driven enhancement of phonological contrast is determined by phonetic implementation of these (language-specific) phonetic features. Second, an accent effect was observed in stressed and unstressed syllables, and was independent of prosodic boundary size. The domain of accentuation in Dutch is thus larger than the foot. Third, within a prosodic category consisting of those utterances with a boundary tone but no pause, tokens with syntactically defined Phonological Phrase boundaries could be differentiated from the other tokens. This syntactic influence on prosodic phrasing implies the existence of an intermediate-level phrase in the prosodic hierarchy of Dutch.

Abstract

We conducted four experiments to investigate the specificity of perceptual adjustments made to unusual speech sounds. Dutch listeners heard a female talker produce an ambiguous fricative [?] (between [f] and [s]) in [f]- or [s]-biased lexical contexts. Listeners with [f]-biased exposure (e.g., [witlo?]; from witlof, “chicory”; witlos is meaningless) subsequently categorized more sounds on an [εf]–[εs] continuum as [f] than did listeners with [s]-biased exposure. This occurred when the continuum was based on the exposure talker's speech (Experiment 1), and when the same test fricatives appeared after vowels spoken by novel female and male talkers (Experiments 1 and 2). When the continuum was made entirely from a novel talker's speech, there was no exposure effect (Experiment 3) unless fricatives from that talker had been spliced into the exposure talker's speech during exposure (Experiment 4). We conclude that perceptual learning about idiosyncratic speech is applied at a segmental level and is, under these exposure conditions, talker specific.

Abstract

In a phonological priming experiment using spoken Dutch words, Dutch listeners were taught varying expectancies and relatedness relations about the phonological form of target words, given particular primes. They learned to expect that, after a particular prime, if the target was a word, it would be from a specific phonological category. The expectancy either involved phonological overlap (e.g., honk-vonk, “base-spark”; expected related) or did not (e.g., nest-galm, “nest-boom”; expected unrelated, where the learned expectation after hearing nest was a word rhyming in -alm). Targets were occasionally inconsistent with expectations. In these inconsistent expectancy trials, targets were either unrelated (e.g., honk-mest, “base-manure”; unexpected unrelated), where the listener was expecting a related target, or related (e.g., nest-pest, “nest-plague”; unexpected related), where the listener was expecting an unrelated target. Participant expectations and phonological relatedness were thus manipulated factorially for three types of phonological overlap (rhyme, one onset phoneme, and three onset phonemes) at three interstimulus intervals (ISIs; 50, 500, and 2,000 msec). Lexical decisions to targets revealed evidence of expectancy-based strategies for all three types of overlap (e.g., faster responses to expected than to unexpected targets, irrespective of phonological relatedness) and evidence of automatic phonological processes, but only for the rhyme and three-phoneme onset overlap conditions and, most strongly, at the shortest ISI (e.g., faster responses to related than to unrelated targets, irrespective of expectations). Although phonological priming thus has both automatic and strategic components, it is possible to cleave them apart.

Abstract

Although researchers studying human speech recognition (HSR) and automatic speech recognition (ASR) share a common interest in how information processing systems (human or machine) recognize spoken language, there is little communication between the two disciplines. We suggest that this lack of communication follows largely from the fact that research in these related fields has focused on the mechanics of how speech can be recognized. In Marr's (1982) terms, emphasis has been on the algorithmic and implementational levels rather than on the computational level. In this article, we provide a computational-level analysis of the task of speech recognition, which reveals the close parallels between research concerned with HSR and ASR. We illustrate this relation by presenting a new computational model of human spoken-word recognition, built using techniques from the field of ASR that, in contrast to current existing models of HSR, recognizes words from real speech input.

Abstract

We report detailed analyses of a very large database on timing of speech perception collected by Smits et al. (Smits, R., Warner, N., McQueen, J.M., Cutler, A., 2003. Unfolding of phonetic information over time: A database of Dutch diphone perception. J. Acoust. Soc. Am. 113, 563–574). Eighteen listeners heard all possible diphones of Dutch, gated in portions of varying size and presented without background noise. The present report analyzes listeners’ responses across gates in terms of phonological features (voicing, place, and manner for consonants; height, backness, and length for vowels). The resulting patterns for feature perception differ from patterns reported when speech is presented in noise. The data are also analyzed for effects of stress and of phonological context (neighboring vowel vs. consonant); effects of these factors are observed to be surprisingly limited. Finally, statistical effects, such as overall phoneme frequency and transitional probabilities, along with response biases, are examined; these too exercise only limited effects on response patterns. The results suggest highly accurate speech perception on the basis of acoustic information alone.

Abstract

Simulations explored the inability of the TRACE model of spoken-word recognition to model the effects on human listening of acoustic-phonetic mismatches in word forms. The source of TRACE's failure lay not in its interactive connectivity, not in the presence of interword competition, and not in the use of phonemic representations, but in the need for continuously optimised interpretation of the input. When an analogue of TRACE was allowed to cycle to asymptote on every slice of input, an acceptable simulation of the subcategorical mismatch data was achieved. Even then, however, the simulation was not as close as that produced by the Merge model.

Abstract

Lexical knowledge influences how human listeners make decisions about speech sounds. Positive lexical effects (faster responses to target sounds in words than in nonwords) are robust across several laboratory tasks, while negative effects (slower responses to targets in more word-like nonwords than in less word-like nonwords) have been found in phonetic decision tasks but not phoneme monitoring tasks. The present experiments tested whether negative lexical effects are therefore a task-specific consequence of the forced choice required in phonetic decision. We compared phoneme monitoring and phonetic decision performance using the same Dutch materials in each task. In both experiments there were positive lexical effects, but no negative lexical effects. We observe that in all studies showing negative lexical effects, the materials were made by cross-splicing, which meant that they contained perceptual evidence supporting the lexically-consistent phonemes. Lexical knowledge seems to influence phonemic decision-making only when there is evidence for the lexically-consistent phoneme in the speech signal.

Abstract

We briefly describe the Merge model of phonemic decision-making, and, in the light of general arguments about the possible role of feedback in spoken-word recognition, defend Merge's feedforward structure. Merge not only accounts adequately for the data, without invoking feedback connections, but does so in a parsimonious manner.

Abstract

The central thesis of the target article was that feedback is never necessary in spoken word recognition. The commentaries present no new data and no new theoretical arguments which lead us to revise this position. In this response we begin by clarifying some terminological issues which have lead to a number of significant misunderstandings. We provide some new arguments to support our case that the feedforward model Merge is indeed more parsimonious than the interactive alternatives, and that it provides a more convincing account of the data than alternative models. Finally, we extend the arguments to deal with new issues raised by the commentators such as infant speech perception and neural architecture.

Abstract

Top-down feedback does not benefit speech recognition; on the contrary, it can hinder it. No experimental data imply that feedback loops are required for speech recognition. Feedback is accordingly unnecessary and spoken word recognition is modular. To defend this thesis, we analyse lexical involvement in phonemic decision making. TRACE (McClelland & Elman 1986), a model with feedback from the lexicon to prelexical processes, is unable to account for all the available data on phonemic decision making. The modular Race model (Cutler & Norris 1979) is likewise challenged by some recent results, however. We therefore present a new modular model of phonemic decision making, the Merge model. In Merge, information flows from prelexical processes to the lexicon without feedback. Because phonemic decisions are based on the merging of prelexical and lexical information, Merge correctly predicts lexical involvement in phonemic decisions in both words and nonwords. Computer simulations show how Merge is able to account for the data through a process of competition between lexical hypotheses. We discuss the issue of feedback in other areas of language processing and conclude that modular models are particularly well suited to the problems and constraints of speech recognition.

Abstract

Two word-spotting experiments are reported that examine whether the Possible-Word Constraint (PWC) [1] is a language-specific or language-universal strategy for the segmentation of continuous speech. The PWC disfavours parses which leave an impossible residue between the end of a candidate word and a known boundary. The experiments examined cases where the residue was either a CV syllable with a lax vowel, or a CVC syllable with a schwa. Although neither syllable context is a possible word in English, word-spotting in both contexts was easier than with a context consisting of a single consonant. The PWC appears to be language-universal rather than language-specific.

Abstract

Simulations explored the inability of the TRACE model of spoken-word recognition to model the effects on human listening of subcategorical mismatch in word forms. The source of TRACE's failure lay not in interactive connectivity, not in the presence of inter-word competition, and not in the use of phonemic representations, but in the need for continuously optimised interpretation of the input. When an analogue of TRACE was allowed to cycle to asymptote on every slice of input, an acceptable simulation of the subcategorical mismatch data was achieved. Even then, however, the simulation was not as close as that produced by the Merge model, which has inter-word competition, phonemic representations and continuous optimisation (but no interactive connectivity).