CoChat
Sign Up
Cited by 376Phonetics and Phonology ResearchNeurobiology of Language and BilingualismReading and Literacy Development

Cross-language evidence for three factors in speech perception

Open access
Werker, Janet F., Logan, John S.·1985·Perception & Psychophysics
DOI 10.3758/bf03207136
Open this paper in CoChat — free accountOpen PDF

AI snapshot

Written from the paper’s full text.

TL;DR

Three experiments with English and Hindi listeners tested whether speech perception involves one, two, or three processing factors. The results support a three-factor model—auditory, phonetic, and phonemic processing—whose use depends on task conditions such as interstimulus interval and practice.

Key findings

  • Across three AX experiments, English and Hindi listeners showed patterns consistent with three distinct processing strategies: phonemic, phonetic, and auditory; single-factor psychoacoustic and dual-factor models were judged inadequate.
  • Experiment 1 (30 English adults; within-subjects; 250/500/1,500-msec ISI): significant pairing effect [F(2,58)=48.19, p<.001], no main effect of ISI, and no interaction. Tukey tests showed PI vs. NI did not differ at 250 or 500 msec, but both differed from DIF (p<.05); at 1,500 msec all three pairing types differed (p<.05). Complex position/order interactions indicated the within-subjects design was not appropriate.
  • Experiment 2 (30 English adults; between-subjects; 10 per ISI; 480 trials): significant effects of ISI [F(2,147)=3.79, p<.05], pairing [F(2,294)=399.81, p<.001], and their interaction [F(4,294)=14.64, p<.001]. All three pairing types differed in all ISIs, less pronounced at 1,500 msec (p<.01 at 250 and 500 msec; p<.05 at 1,500 msec). RT showed a pairing effect [F(2,54)=7.20, p<.002] and interactions.
  • Experiment 3 (4 native Hindi speakers; 1,500-msec ISI; one block): significant pairing effect [F(2,12)=380.79, p<.001]; PI and NI were equally high (XPI=99.16, XNI=94.16) and DIF low (XDIF=9.8), matching a phonetic pattern for English listeners but phonemic for Hindi listeners; no auditory level was evident.

Methods

  • Design: same-different (AX) discrimination with natural Hindi retroflex/dental syllables (4 retroflex, 4 dental; approximately 400 msec each); P(Same) for physically identical (PI), name-identical (NI), and different (DIF) pairings; reaction time also measured in Experiment 2.
  • Samples: Experiment 1 used 30 unilingual English adults (15 male, 15 female) tested within-subjects across 250/500/1,500-msec ISI; Experiment 2 used 30 unilingual English adults between-subjects, 10 per ISI, with 480 trials (5 blocks of 96); Experiment 3 used 4 native Hindi speakers at 1,500-msec ISI with one block.
  • Analysis: ANOVAs on P(Same) and RT, with Tukey post hoc comparisons; Experiment 1 also included position and order analyses; Experiment 3 used a one-way ANOVA.

Limitations

  • Stated: The experiments do not explain the derivation of phonetic or phonemic processing; it is unclear whether phonetic perception reflects a specific linguistic processor or second-order auditory factors, and whether phonemic processing is based on modified innate universal phonetic sensitivities or learned auditory-based categories. Further research with different procedures, populations, and stimuli is needed.
  • Stated/evident: Experiment 1’s blocked within-subjects design produced complex interactions and was judged not appropriate for assessing the three factors; subjects could not easily switch processing strategies across ISI blocks.
  • Evident: Experiment 3 used only 4 Hindi speakers, one block, and only the 1,500-msec ISI, so it cannot test alternative processing levels among Hindi listeners; Experiment 2’s second-order interaction was not significant (p=.13) though post hoc tests were still interpreted (footnote 3).

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.3758/bf03207136

Cites

34 of 34 references verified0 citation errors · 0 not checkable

Statistics

Claims the paper makes

0 of 22 cited sources checked · 0 supported · 0 flagged

  • Findingown data

    In Experiment 1, at the 1,500-msec ISI, the means for all three pairing types were significantly different from each other.

    “In the I ,5OO-msec lSI condition, the means for all three types of pairings were significantly different from each other (p < .05).”
    Quote in the paperSubstantial claim
  • Findingown data

    In Experiment 1, there was a significant main effect for type of pairing (PI, NI, DIF) on P(Same) responses.

    “A significant main effect for type of pairing [F(2,58) = 48.19,P < .001]was obtained in this analysis.”
    Quote in the paperSubstantial claim
  • Findingown data

    In Experiment 1, P(Same) to NI and PI pairings differed significantly from DIF pairings at 250 and 500 msec ISIs.

    “The means of the P(Same) to the NI and PI pairings, as compared with the DIF pairings, did, however, differ significantly at these ISIs (p < .05).”
    Quote in the paperSubstantial claim
  • Findingnone given

    In Experiment 1, the effect of ISI was negligible.

    “The effect ofISI was negligible.”
    Quote in the paperNot a substantial claim
  • Findingown data

    In Experiment 1, performance in an ISI block was better if that block had been preceded by a 250-msec condition.

    “In most cases, performance in an lSI block was better if that block had been preceded by a 250-msec condition.”
    Quote in the paperSubstantial claim
  • May combine two claims. 19
  • The quote only partly states this claim. 4
  • Classified as probably not a substantial claim. 4

  • May combine two claims.

    Findingown data

    In Experiment 1, there was no main effect for ISI and no significant interaction between type of pairing and ISI.

    “There was no main effect for lSI, and no significant interaction.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    In Experiment 2, in all three ISI conditions there was a significant difference between P(Same) for each type of pairing, but the effect was less pronounced at the longest ISI.

    “Multiple comparisons using the Tukey method revealed that in all three lSI conditions there was a significant difference between the P(Same) response for each type of pairing, but that the effect was less pronounced at the longest lSI (p < .01 at 250 and 500 msec; p < .05 at 1,500 msec).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Although the second-order interaction between ISI, type of pairing, and position did not reach significance in Experiment 2, the data pattern suggests P(Same) responses to the three pairing types tended to separate at a dissimilar rate.

    “Although the second-order interaction between lSI, type of pairing, and position did not reach significance (p = .13),the data pattern suggests that the P(Same) responses to the three pairing types did tend to separate at a dissimilar rate at the”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    In Experiment 2 practice analysis, there were significant main effects of type of pairing and of the interaction between type of pairing and position.

    “There were two significant main effects: type of pairing [F(2,54) = 171.05,P < .001]and the other between type of pairing and position [F(8,216) = 11.25,P < .001].”
    Quote in the paperNot a substantial claim

  • The quote only partly states this claim.

    Findingown data

    In Experiment 2, there was a significant interaction between type of response and type of pairing [F(2,54) = 55.46, p < .001].

    “one between type of response and type of pairing [F(2,54) = 55.46,p < .001]”
    Quote in the paperSubstantial claim

  • The quote only partly states this claim.

    Findingown data

    In Experiment 2, there was a significant interaction among ISI, type of response, and type of pairing [F(4,54) = 5.97, P < .001].

    “the other among lSI, type of response, and type of pairing [F(4,54) = 5.97, P < .001]”
    Quote in the paperSubstantial claim

  • The quote only partly states this claim.

    Findingown data

    In the present experiment, the effect of nonnative (phonetic) discrimination in the 500-msec condition for DIF pairings is not evident until close to 300 trials.

    “In the present experiment, this effect is not evident until close to 300 trials.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    In Experiment 2, in the 1,500-msec ISI condition, P(Same) responses to the three pairing types were not significantly different until the third block of trials, at which time all three (PI, NI, and DIF) were significantly different.

    “Finally, in accordance with our initial predictions, the P(Same) responses to the three pairing types in the 1,5OO-msec lSI condition were not significantly different until the third block of trials, at which time all three (PI, NI, and DIF) were significantly different.”
    Quote in the paperSubstantial claim

  • The quote only partly states this claim.

    Findingown data

    In Experiment 2, there was a significant main effect for type of pairing [F(2,54) = 7.20, P < .002].

    “nificant main effect for type of pairing [F(2,54) = 7.20, P < .002]”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    In Experiment 2, in the 500-msec ISI condition, P(Same) responses to PI pairings were significantly greater than those to NI and DIF pairings, but NI and DIF pairings did not diverge significantly until the third block of trials.

    “In the 5OO-msec condition, P(Same) responses to PI pairings were significantly greater than those to NI and DIF pairings, but NI and DIF pairings did not diverge significantly until the third block of trials.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    RT to NI and DIF pairings is virtually identical in Block 1 and much slower than that to PI pairings; RTs to NI and DIF begin to spread apart in Block 2, foreshadowing the P(Same) pattern evident in Block 3.

    “The RT to NI and DIF pairings is virtually identical in Block 1 and much slower than that to PI pairings; RTs to NI and DlF begin to spread apart in Block 2, foreshadowing the P(Same) pattern evident in Block 3.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Premisecitation

    In dual-factor models, the acoustic waveform is stored in both an auditory and a phonetic code, but the auditory code decays rapidly relative to the acoustic code.

    “In dual-factor models, it is proposed that the acoustic waveform is stored in both an auditory and a phonetic code, but that the auditory code decays rapidly relative to the acoustic code (Fujisaki & Kawashima, 1969, 1970;Pisoni, 1973).”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites Fujisaki & Kawashima, 1969, 1970

      Not checked in the free run.

    • cites Pisoni, 1973 source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationreasoning

    When feedback is not provided and the experimenter is interested solely in determining how ISI affects level of processing, a blocked, within-subjects design may not be optimal.

    “However, when feedback is not provided and the experimenter is interested solelyin determining how lSI affects the level of processing, a blocked, within-subjects design may not be optimal.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    English-speaking adults and English-learning infants aged 10-12 months cannot discriminate the non-English Hindi and Thompson distinctions.

    “Initial results suggested that Englishlearning infants aged 6-8 months can discriminate these non-English distinctions as well as native Hindi and Thompson speakers, but that English-speaking adults (Werker, Gilbert, Humphrey, & Tees, 1981) and Englishlearning infants aged 10-12 months (Werker & Tees, 1984a) cannot.”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites Werker, Gilbert, Humphrey, & Tees, 1981 source

      Not checked in the free run.

    • cites Werker & Tees, 1984a

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    At shorter ISIs, subjects showed evidence of within-category auditory-level discriminations, whereas at longer ISIs only phonetic categorization was evident.

    “At the shorter ISIs, subjects showed evidence of within-category auditory-level discriminations, whereas at longer ISIs only phonetic categorization was evident.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Pisoni (1973) source

      Not checked in the free run.

  • May combine two claims.

    Interpretationown data

    The Experiment 1 data pattern suggests subjects used a phonetic processing strategy in the two shortest ISI conditions and possibly both phonetic and auditory strategies in the 1,500-msec condition.

    “This data pattern suggests that subjects were using a phonetic processing strategy in the two shortest lSI conditions, and were possibly using both a phonetic and an auditory strategy in the 1,500-rnsec condition.”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: If stimuli are not discriminated, then P(Same) responses should be identical across the pairing types.

  • Classified as probably not a substantial claim.

    Interpretationreasoning

    The complex interactions from Experiment 1 may have masked important data patterns.

    “The complex interactions that resulted from the withinsubjects design used in Experiment 1 may have masked important data patterns.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisereasoning

    A considerable body of research suggests that under some testing conditions subjects discriminate speech and speech-like stimuli according to phonetic category boundaries, but under other conditions they are sensitive to auditory information and can demonstrate more continuous discrimination functions.

    “a considerable body of research does suggest that under some testing conditions subjects discriminate speech and speech-like stimuli according to phonetic category boundaries, but that, under other conditions, they are sensitive to the auditory information and can demonstrate more continuous discrimination functions.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    The Experiment 1 position effect suggests practice may enhance discrimination performance for all pairing types, with a greater effect for DIF than NI pairings.

    “This suggests that practice may enhance discrimination performance for all types of pairings, but that the effect may be greater for DIF than for NI pairings.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    Although English speakers cannot discriminate the non-English distinctions under most testing conditions, they can differentiate the syllables according to Hindi and Thompson phonetic categories under some testing conditions.

    “Subsequent experiments have shown that although English speakers cannot discriminate these distinctions under most testing conditions (Tees & Werker, 1984), they can differentiate the syllables according to Hindi and Thompson phonetic categories under some testing conditions (Werker & Tees, 1984b).”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites Tees & Werker, 1984 source

      Not checked in the free run.

    • cites Werker & Tees, 1984b

      Not checked in the free run.

  • May combine two claims.

    Interpretationown data

    The data from all three experiments together indicate the presence of a phonetic level as well as phonemic and auditory levels, and suggest that dual-factor models may also be inadequate to explain the data pattern.

    “However, the data from all three experiments together, indicating the presence of a phonetic level as well a phonemic and auditory levels, suggest that dual-factor models may also be inadequate to explain the data pattern.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Interpretationreasoning

    The P(Same) data reported by Pisoni and Tash differ markedly from our results, presumably because Pisoni and Tash gave their subjects explicit NI instructions.

    “The P(Same) data reported by Pisoni and Tash differ markedly from our results, presumably because Pisoni and Tash gave their subjects explicit NI instructions.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Pisoni and Tash (1974) source

      Not checked in the free run.

  • May combine two claims.

    Interpretationcitation

    Our work suggests that it is more difficult, but still possible, to learn to differentiate nonnative place of articulation distinctions.

    “Our work suggests that this is more difficult, but still possible, in the case of nonnative place of articulation distinctions.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Pisoni, Aslin, Perey, and Hennessy's (1982) source

      Not checked in the free run.

  • May combine two claims.

    Interpretationown data

    These experiments provide evidence for three processing strategies, and show that phonemic perception is clearly distinct from auditory perception.

    “These experiments provide evidence for three processing strategies, and show that phonemic perception is clearly distinct from auditory perception.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    In the first two blocks in the 500-msec condition, subjects show both auditory and phonemic processing, since the P(Same) responses are very high for the PI pairings but not for the other pairing types.

    “In the first two blocks in the 5OO-msec condition, it appears that subjects show both auditory andphonemic processing, since the P(Same) responses are very high for the PI pairings but not for the other pairing types.”
    Quote in the paperSubstantial claim
25 of 25

  • The authors state this as an assumption.

    If stimuli are not discriminated, then P(Same) responses should be identical across the pairing types.

    “Presumably, if stimuli are notdiscriminated, then the P(Same) responses should be identical across the pairing types.'”

  • The paper relies on this without support.

    The Hindi retroflex/dental distinction does not have phonemic significance in English, and both categories of consonants are typically perceived as the English alveolar phone [t].

    “However, this distinction does not have phonemic significance in English, and both categories of consonants are typically perceived as the English alveolar phone [t].”

  • The paper relies on this without support.

    The proportion same [P(Same)] measure gauges the perceived similarity of the three types of pairings.

    “We used a proportion same [P(Same)] measure to gauge the perceived similarity of the three types of pairings.”

  • The paper relies on this without support.

    There is likely to be more within-category variability for the Hindi retroflex/dental tokens than for the Thompson velar/uvular distinctions.

    “since there was likely to be more within-category variability for these tokens than for the Thompson velar/uvular distinctions”

  • The authors state this as an assumption.

    Hindi-speaking subjects would also have access to an auditory processing level.

    “It was already known from Experiment 2 that Englishspeaking subjects have access to an auditory processing level, and it can be assumed that Hindi-speaking subjects would also have access to this level”

  • The authors state this as an assumption.

    The chance level would vary depending upon processing strategy.

    “For the same reasons, responses were not compared with chance, since the chance level would vary depending upon processing strategy.”
6 of 6

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Adult performance on non-native contrasts can be increased by techniques that minimize memory effects, extensive training, and use of contrasts not related to native-language categories.

    “Adult performance on non-native contrasts could be increased by a number of factors: (i) the use of techniques that minimize the effects of memory (33, 34), (ii) extensive training (35, 36), and (iii) the use of contrasts, such as Zulu clicks, that are not related to native-language categories (37, 38).”

    From A new view of language acquisition (2000) · cites this paper as [33]

Statistics recomputed

Every reported test result, recomputed from its test statistic and degrees of freedom. A result shows as not adding up only once an independent check has confirmed it.

  • ConsistentF(2, 58) = 48.19reported p < .001 · recomputed p < .001

    “F(2,58) = 48.19,p < .001”

  • ConsistentF(2, 54) = 58.58reported p < .001 · recomputed p < .001

    “F(2,54) = 58.58,p < .001”

  • ConsistentF(2, 27) = 7.78reported p < .005 · recomputed p = .002

    “F(2,27) = 7.78, p < .005”

  • ConsistentF(4, 54) = 3.63reported p < .05 · recomputed p = .011

    “F(4,54) = 3.63, p < .05”

  • ConsistentF(2, 48) = 47.13reported p < .001 · recomputed p < .001

    “F(2,48) = 47.13,p < .001”

  • ConsistentF(20, 96) = 2.97reported p < .001 · recomputed p < .001

    “F(20,96) = 2.97, p < .001”

  • ConsistentF(2, 147) = 3.79reported p < .05 · recomputed p = .025

    “F(2,147) = 3.79, p < .05”

  • ConsistentF(2, 294) = 399.81reported p < .001 · recomputed p < .001

    “F(2,294) = 399.81,p < .001”

  • ConsistentF(4, 294) = 14.64reported p < .001 · recomputed p < .001

    “F(4,294) = 14.64,p < .001”

  • ConsistentF(2, 54) = 171.05reported p < .001 · recomputed p < .001

    “F(2,54) = 171.05,p < .001”

  • ConsistentF(8, 216) = 11.25reported p < .001 · recomputed p < .001

    “F(8,216) = 11.25,p < .001”

  • ConsistentF(2, 54) = 7.2reported p < .002 · recomputed p = .002

    “F(2,54) = 7.20, p < .002”

  • ConsistentF(2, 54) = 55.46reported p < .001 · recomputed p < .001

    “F(2,54) = 55.46,p < .001”

  • ConsistentF(4, 54) = 5.97reported p < .001 · recomputed p < .001

    “F(4,54) = 5.97, p < .001”

  • ConsistentF(2, 12) = 380.79reported p < .001 · recomputed p < .001

    “F(2,12) = 380.79,p < .001”

15 of 15

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Experiment 2 ANOVA degrees of freedom conflict with the sample size

    Major

    The reported degrees of freedom, F(2,147) and F(2,294), imply 150 subjects in the stated 3 × 3 mixed ANOVA. Experiment 2 reports only 30 subjects, ten in each ISI group, so these conclusion-bearing test statistics cannot come from that analysis as described.

  • Citation year differs from its reference entry

    Minor

    The text cites the Pisoni, Carrell, and Gans study as published in 1982, but its reference entry gives 1983.

2 of 2

Cites

Flagged first, then by citations

Research works printed but not linked: 2

  1. Fujisaki, H., &Kawashima, T. (1969). Some experiments on speech perception and a model for the perceptual mechanism.Annual Report of the Engineering Research Institute (Tokyo),28, 67–73.
  2. Posner, M. (1978).Chronomemc explorations of mind. Hillsdale, NJ: Erlbaum.

Cited by

Sorted by citations

Related

Authors
Werker, J. F., & Logan, J. S.
Affiliations
Dalhousie University (CA)
ORCID
Werker, Janet F.
Venue
Perception & Psychophysics
Publisher
Springer Science and Business Media LLC
Published
1985-01
Volume
37
Issue
1
Pages
35-44
ISSN
  • 0031-5117 (ISSN-L)
  • 1532-5962
DOI
10.3758/bf03207136
OpenAlex
W2026070756
Semantic Scholar
ca7ec19ba9c8f1b59ce7e8124520a584e4cf1edf
Cited by
376
References
32
Citations by year
2026: 9 · 2025: 5 · 2024: 9 · 2023: 29 · 2022: 18 · 2021: 15 · 2020: 9 · 2019: 7 · 2018: 14 · 2017: 20 · 2016: 14 · 2015: 8 · 2014: 24 · 2013: 13 · 2012: 9
FWCI
5.64
Type
journal article
Language
en
Open access
Licence
www.springer.com/tdm (tdm)
Locations
  • Perception & Psychophysics (publishedVersion)
  • PubMed (publishedVersion)
Funders
Eunice Kennedy Shriver National Institute of Child Health and Human Development (HD12420)
Topics
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Phonetics and Phonology Research
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Neurobiology of Language and Bilingualism
  • Social Sciences › Psychology › Developmental and Educational Psychology › Reading and Literacy Development
Keywords
speech perception, cross-language speech perception, three-factor model, psychoacoustic model, phonetic processing, phonemic processing, auditory processing, interstimulus interval
MeSH terms
Female, Humans, Language, Male, Models, Psychological, Phonetics, Psychoacoustics, Speech Perception

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-10.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-10.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review