TISK 1.0: An easy-to-use Python implementation of the time-invariant string kernel model of spoken word recognition

You, Heejo; Magnuson, James S.

doi:10.3758/s13428-017-1012-5

Cited by 16 publications

(13 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There is, however, some evidence in favor of a more flexible phoneme‐order encoding from studies showing that words that can be generated by adding or deleting a phoneme in the target word also influence target word recognition (Dufour & Frauenfelder, ; Luce & Pisoni, ; Vitevitch & Luce, ; see also Allopenna et al, ; Connine, Blasko, & Titone, ). Evidence for such flexibility is in line with more recent accounts of spoken word recognition, such as the TISK model of Hannagan, Magnuson, and Grainger (; see You & Magnuson, , for a more recent implementation). Such flexibility is achieved in TISK via open diphone units that represent ordered sequences of contiguous and non‐contiguous phonemes.…”

Section: Introductionsupporting

confidence: 79%

Phoneme‐Order Encoding During Spoken Word Recognition: A Priming Investigation

Dufour

Grainger

2019

Cognitive Science

View full text Add to dashboard Cite

In three experiments, we examined priming effects where primes were formed by transposing the first and last phoneme of tri‐phonemic target words (e.g., /byt/ as a prime for /tyb/). Auditory lexical decisions were found not to be sensitive to this transposed‐phoneme priming manipulation in long‐term priming (Experiment 1), with primes and targets presented in two separated blocks of stimuli and with unrelated primes used as control condition (/mul/‐/tyb/), while a long‐term repetition priming effect was observed (/tyb/‐/tyb/). However, a clear transposed‐phoneme priming effect was found in two short‐term priming experiments (Experiments 2 and 3), with primes and targets presented in close temporal succession. The transposed‐phoneme priming effect was found when unrelated prime‐target pairs (/mul/‐/tyb/) were used as control and more important when prime‐target pairs sharing the medial vowel (/pys/‐/tyb/) served as control condition, thus indicating that the effect is not due to vocalic overlap. Finally, in Experiment 3, a transposed‐phoneme priming effect was found when primes sharing the medial vowel plus one consonant in an incorrect position with the targets (/byl/‐/tyb/) served as control condition, and this condition did not differ significantly from the vowel‐only condition. Altogether, these results provide further evidence for a role for position‐independent phonemes in spoken word recognition, such that a phoneme at a given position in a word also provides evidence for the presence of words that contain that phoneme at a different position.

show abstract

Section: Introductionsupporting

confidence: 79%

Phoneme‐Order Encoding During Spoken Word Recognition: A Priming Investigation

Dufour

Grainger

2019

Cognitive Science

View full text Add to dashboard Cite

show abstract

“…the rime) contribute to spoken word recognition. Note that the observation that the rime contributes to pre-lexical processing is fully compatible with a recent model of spoken word recognition, the TISK model (Hannagan et al, 2013;You & Magnuson, 2018), which incorporates a set of diphone units that code the order of phonemes, and a set of positionindependent phoneme units. Although there is recent evidence for the existence of positionindependent phoneme units coming from studies showing transposed-phoneme effects at least when consonants are transposed (e.g.…”

Section: Discussionsupporting

confidence: 53%

Phonological abstraction before lexical access: New evidence from rime priming

Dufour

Fléchard

Nguyen

2022

Quarterly Journal of Experimental Psychology

View full text Add to dashboard Cite

In this study, we focused on the rime priming effect and examined whether the frequency of rimes influences the size of the effect. Using the lexical decision task, we observed that the rime priming effect interacted with rime frequency with a stronger priming effect when the target words contained low-frequency rimes, in comparison to target words containing high-frequency rimes. Moreover, for both target words with low- and high-frequency rimes, a priming effect of similar magnitude was observed whether the primes and targets were pronounced by the same speaker or by different speakers. All together, these findings provide additional support for all the studies that plead for the existence of prelexical units and, in a more general way, for phonological abstraction prior to lexical access.

show abstract

“…Decades after the discovery of the lack of invariance problem —the absence of invariant cues to speech sounds (e.g., Joos, 1948; Liberman et al, 1952; Peterson & Barney, 1952)—speech science offers limited explanations of how humans achieve phonetic constancy despite the many‐to‐many mapping between acoustics and percepts. Computational models of HSR have provided little insight, since most current models sidestep the vagaries of the signal and use idealized, abstract elements such as phonetic features (McClelland & Elman, 1986), phonemes (Hannagan, Magnuson, & Grainger, 2013; You & Magnuson, 2018), or human phoneme confusion probabilities (Norris & McQueen, 2008) rather than real speech as input. Such assumptions can ultimately complicate rather than simplify problems (Magnuson, 2008), as the details they bypass may contain constraints essential to the mechanisms underlying human performance.…”

Section: Discussionmentioning

confidence: 99%

EARSHOT: A Minimal Neural Network Model of Incremental Human Speech Recognition

et al. 2020

Self Cite

View full text Add to dashboard Cite

Despite the lack of invariance problem (the many-to-many mapping between acoustics and percepts), human listeners experience phonetic constancy and typically perceive what a speaker intends. Most models of human speech recognition (HSR) have side-stepped this problem, working with abstract, idealized inputs and deferring the challenge of working with real speech. In contrast, carefully engineered deep learning networks allow robust, real-world automatic speech recognition (ASR). However, the complexities of deep learning architectures and training regimens make it difficult to use them to provide direct insights into mechanisms that may support HSR. In this brief article, we report preliminary results from a two-layer network that borrows one element from ASR, long short-term memory nodes, which provide dynamic memory for a range of temporal spans. This allows the model to learn to map real speech from multiple talkers to semantic targets with high accuracy, with human-like timecourse of lexical access and phonological competition. Internal representations emerge that resemble phonetically organized responses in human superior temporal gyrus, suggesting that the model develops a distributed phonological code despite no explicit training on phonetic or phonemic targets. The ability to work with real speech is a major advance for cognitive models of HSR.

show abstract

TISK 1.0: An easy-to-use Python implementation of the time-invariant string kernel model of spoken word recognition

Cited by 16 publications

References 22 publications

Phoneme‐Order Encoding During Spoken Word Recognition: A Priming Investigation

Phoneme‐Order Encoding During Spoken Word Recognition: A Priming Investigation

Phonological abstraction before lexical access: New evidence from rime priming

EARSHOT: A Minimal Neural Network Model of Incremental Human Speech Recognition

Contact Info

Product

Resources

About