looking for exemplar effects
play

Looking for exemplar effects: testing the comprehension and memory - PowerPoint PPT Presentation

Looking for exemplar effects: testing the comprehension and memory ry representations of f r' r'duced words in in Dutch le learners of f French LISA MORANO, LOUIS TEN BOSCH, MIRJAM ERNESTUS (2019) 1 LILAS MAUBORGNE EXEMPLAR THEORY IN


  1. Looking for exemplar effects: testing the comprehension and memory ry representations of f r' r'duced words in in Dutch le learners of f French LISA MORANO, LOUIS TEN BOSCH, MIRJAM ERNESTUS (2019) 1 LILAS MAUBORGNE EXEMPLAR THEORY IN LANGUAGE AND SPEECH SCIENCE 01.07.2020

  2. Table of contents Introduction 1. Method 2. Results 3. General Discussion 4. Conclusion 5. 2

  3. Introduction  Most experiments investigating exemplars  Native listeners (L1)  Almost never second language learners (L2)  Reason for an investigation on exemplars in L2:  If exemplars important in speech comprehension → findings for L1 listeners should be the same for L2 listeners  Provide information about which acoustic details are stored in exemplars  Are exemplars true representation of the acoustic signal?  Are they affected by the listener’s knowledge? 3

  4. Introduction  L1 phonological filter:  L2 learners perform well in a phoneme categorization task but not always in lexical decision tasks  L2 listeners can perform simple low-level tasks in phonetic mode  As soon as linguistic processing is needed their L1 phonological filter keeps them from processing the stimuli like L1 listeners  Exemplars formed before phonological filter applies: L2 exemplars can encode L1 irrelevant variation  Exemplars formed after phonological filter applies: L2 exemplars encode less L1 irrelevant variation 4

  5. Introduction  Existing exemplar research in L2 provides evidence that L2 listeners can store exemplars  Main questions:  Are e L2 L2 intermediate lea earners ab able le to to en encode, in n the form orm of of exempla lars, fin fine lin inguis istic de details s that ar are e no not t rele elevant to to their L1 L1 an and to to use use them for or spe speech com omprehension?  Can Can exemplar effects s in L2 L2 listeners al also so be be fou ound whe hen manipulating reg egular pho phonetic vari ariation ins nstead of of indexical vari ariation? 5

  6. Introduction  Red eduction = the weakening or deletion of phonemes or even whole syllables  Occurs in informal speech  Most other experiments: replaced one allophone with another allophone  Listeners could have stored several abstract representations rather than different exemplars  Categorial variation: more difficult to attest for the role of exemplars  Reduction: can’t be stored abstractly  May result in an infinite number of realization  All activate same abstract pronunciation variant of the word 6

  7. Introduction  Red eduction = the weakening or deletion of phonemes or even whole syllables  Here: Reduction phenomenon of phrase-medial high vowel devoicing in French  “la cité ” ( the city )  /i/: more or less devoiced → voicing fails to be re-established in time after devoiced consonant /s/  Phrase-medial high vowel devoicing has never been reported in Dutch  Part of sound pattern of French but not of Dutch 7

  8. Method  Lexical decision task in French for Dutch intermediate French learners  Experimental words: High vowel + voiceless consonant  Pronunciation match: prime and target both reduced or not reduced  Pronunciation mismatch: prime is reduced, target is not or vice-versa  Expectations:  Faster reaction to a target when it matches with prime  L2 participants show exemplar effect  Are able to store phonetic information that doesn’t happen in L1 8

  9. Method: Participants  120 Dutch students  Studied French in high school (4 to 7 years)  18-29 years old  One of the 3 conditions randomly assigned 9

  10. Method: Materials  Ex Experim imental l wor ords ds:  24 most frequent from vocabulary of beginners’ textbook in French for Dutch learners  Bisyllabic  1 st syllable: High vowel: /i/, /y/, /u/ followed by voiceless consonant  “la cité ” ( the city )  “le f u ture” ( the future )  “la pou oubelle ” ( the garbage )  Frequency of occurrence: 0.71 per million words to 107.92 10

  11. Method: Materials  Ex Exis istin ing-word fi fill llers  78 bisyllabic frequent words form the Stimuli same textbook  No restriction  Pseu eudo-word fille fillers Experimental Existing-word Pseudo-word words fillers fillers  102 created  Phonotactically legal syllable added to 1 st syllable of all experimental and existing- word fillers Voiced Devoiced 11

  12. Method: Materials Devoiced Voiced  Stimuli recorded with their definite determiners by first author of the paper  Devoiced stimuli: word’s 1 st syllable = completely devoiced high syllable  Voiced stimuli: word’s 1 st syllable = fully voiced high syllable /i/ /i/ Waveforms and spectrograms of target “le silence” ( the silence ) 12

  13. Method: Materials  Two Tokens for each voicing type: Stimuli  Voiced and devoiced token A:  On average 805 ms long  Voiced and devoiced token B: Experimental Existing-word Pseudo-word  On average 811 ms long words fillers fillers Voiced Devoiced A B A B 13

  14. Method: Materials  Lexical decision task:  Two blocks of 132 trials Block x2  One block =  12 experimental words  Experimental words repeated Experimental Real-word fillers Pseudo-word words x12 x36 fillers x48  Variant match = both prime and target have (de)voiced vowels  Variant mismatch = prime is devoiced and target is voiced or vice-versa Match Mismatch  Prime-targets separated by 7-97 trials  Remainder of the trials: Prime = voiced Prime = devoiced Prime = voiced Prime = devoiced  36 bisyllabic real-word fillers Target = voiced Target = devoiced Target = devoiced Target = voiced  48 bisyllabic pseudo-word fillers 14

  15. Method: Materials  Lexical decision task:  Practice trials:  6 real-word fillers  6 pseudo-word fillers  Same for all participants  Similar in frequency of occurrence and phonological structure to stimuli of experiment  5 pseudo-randomization of trials  Four different stimulus lists in each  Each experimental word tested for each of four possible matching and mismatching combinations 15

  16. Method: Materials  Con Condit ition AB: B:  Co Condit itions AA an and BB BB:  AA: only tokens A  Different tokens for the primes and the targets  BB: only tokens B  Match: prime A and target B are different  Match: same token and same duration recordings  Primes and targets:  Same pronunciation variant  Different durations Average absolute temporal differences 16 between primes and targets per condition

  17. Method: Materials Stimuli Experimental Existing-word Pseudo-word words fillers fillers AB AA BB Voiced Devoiced A B A B 17

  18. Method: Procedure  Participants:  Listened to the words  Push a yes or no button  Say as fast as possible whether the words were real in French or not  Participants did not need to know the meaning of the words  Experiment lasted less than 30 minutes “la cité ” 18

  19. Results  Accuracies:  Linear mixed effects model for logistic regression  Dependent variable = probability of a correct response  Reaction times = RTs  Measured from noun onset  Mixed effects regression models  Before analysis: all RTs and stimulus log-transformed  Dependent variable for linear mixed effect model = log-transformed RT  Crossed random effect factors = item and participant 19

  20. Results  Predictors of interest:  Voic oicin ing: categorical predictor indicating whether the first vowel of the stimulus was voiced or devoiced  Con Condit ition: AB, AA or BB  Tok oken: A or B  Repetitio ion match: whether the prime and target of the experimental word were of the same pronunciation variant 20

  21. Results  Control predictors:  Log Log Sti Stimulus du duration  Tria ial nu number: position of the trial in the experiment  Control for learning or fatigue effects  Prim rime/target Di Distance (in number of intervening trial)  Log Log RT to to the pr previous tria ial  Control local speed effects  Log Log RT on on the pr prim ime 21

  22. Results: Accuracy data  Participant’s accuracy: relatively high  Overall: 83.92%  Pseudo-word fillers: 85.52%  Real-word fillers: 75.70%  Lower accuracy on experimental words  Experimental words less frequent than real-word fillers 22

  23. Results: Accuracy data Fir irst occ occurrences:  Sensitivity to devoicing manipulation on primes only:  Participants more accurate on the voiced than the devoiced tokens A  Simple effect of Voicing → Participants sensitive to devoicing for tokens A  No statistical difference for tokens B  Significant interaction between Voicing and Token → Participants not sensible to devoicing for tokens B  Significant random slope of Voicing on Item  Voicing significantly larger for some items than others 23

  24. Results: Accuracy data Second occ occurrences:  Participants sensitive to devoicing manipulation (for tokens A)  Investigate whether they were more accurate on matching than on mismatching targets  For targets with correctly answered primes:  No effect of Repetition match on accuracy  Not as main effect  Not in interaction with Condition or Token 24

Recommend


More recommend