Presentation is loading. Please wait.

Presentation is loading. Please wait.

6th International Conference on Language Variation in Europe

Similar presentations


Presentation on theme: "6th International Conference on Language Variation in Europe"— Presentation transcript:

1 6th International Conference on Language Variation in Europe
(Non-)Influence of nationality labels on the perception of final devoicing Marten Juskan

2 of fish and kangaroos variable: short high front vowels ('fish and chips' stereotype) task: matching natural vowels to resynthesised ones (continuum from Au-like to NZ-like) conditions: label at top of answer sheet (either 'New Zealand' or 'Australian') presence of a stuffed toy in the room (either 'kiwi' or 'kangaroo') results: subjects in 'Australian'/'kangaroo' condition chose more Australian-like vowels subjects in 'NZ' condition chose more NZ-like vowels (cf. Hay et al and Hay and Drager 2010) 2

3 INDEXING framework: exemplar theory acoustic signal RP [ɹɪtn]
GA [ɻɪɾn] L2 [Rɪtn] INDEXING cognitive process long-term memory GA [ɻɪɾn] male ... L2 [Rɪtn] female ... RP [ɹɪtn] female ... 3

4 primary research question
Will subjects perceive a German accent when they expect one, regardless of the acoustic material? 4

5 test design variable: final devoicing (considered a salient feature of a German accent in English) task: forced decision task (minimal pairs) Conditions: 1. English- or German-speaking subject 2. label at the top of the screen (either 'English' or 'German') 5

6 hypotheses English-speaking subjects will make very few mistakes
participants in the 'German speaker' condition will have a higher error rate than those in the 'English speaker' condition German-speaking subjects will make more mistakes overall error rates in the 'German' or 'English' speaker condition will not differ significantly 6

7 test stimuli minimal pair list, recorded by a native speaker of RP
words differentiated by /b/-/p/, /d/-/t/, or /g/-/k/ respectively 72 items, 36 test tokens, 36 dummy tokens dummy tokens: opposition at beginning or in the middle of a word medal-metal, gum-come, bin-pin... test tokens: opposition at the end of a word nod-not, league-leak, robe-rope... 7

8 artificial final devoicing
editing of test tokens using Praat cut off final plosive mix final plosives to get an ambiguous sound reattach this sound to each of the two bases individual mixture for every word pair fairly natural sounding test items no other part of the word was altered (cf. Kraljic et al. 2008) 8

9 test layout GERMAN robe rope 9

10 participants 33 in total (20 females, 13 males)
18 native speakers of English (9 in 'English speaker' and 9 in 'German speaker' condition) 15 native speakers of German (8 in 'English speaker' and 7 in 'German speaker' condition) → 2291 data points in total 10

11 results reaction times
'English' speaker 'German' speaker L1 English ms ms p < 0.01 L1 German ms ms p = 0.05 native speakers of English have more exemplars indexed to 'English' native speakers of German have more exemplars indexed to 'German' → more variation to work through? 11

12 results regression tree for error rates 12 1.8% 12.9% 3.3% 23.9% 31.4%
13.9% regression tree for error rates 12

13 error rates (only test tokens)
results error rates (only test tokens) 'English' speaker 'German' speaker L1 English 2.8 % 4.7 % p = 0.20 L1 German 23.6 % 24.2 % p = 0.87 no significant difference between the different speaker conditions no gender effect → no exemplar priming effect? abstract submission deadline 13

14 results G-E group: no linear development
G-G group: no linear development E-E group: no linear development E-G group: training effect! (cor = 0.17, p < 0.01) 14

15 conclusion English-speaking subjects will make very few mistakes → confirmed participants in the 'German speaker' condition will have a higher error rate than those in the 'English speaker' condition → confirmed? German-speaking subjects will make more mistakes overall → confirmed error rates in the 'German' or 'English' speaker condition will not differ significantly → confirmed 15

16 conclusion subjects were more ready to perceive final devoicing when they expected it → an exemplar priming effect could be identified BUT: the effect was small (difference of 8% [p < 0.01] in the first half of the test – problem of salience/conscious awareness?) the effect disappeared in the course of the test → expectations create an initial perceptual bias wrong expectations can be corrected by 'objective' acoustic material 16

17 bibliography Hay, Jennifer, Aaron Nolan and Katie Drager (2006) “From Fush to Feesh: Exemplar Priming in Speech Perception“. The Linguistic Review 23: Hay, Jennifer and Katie Drager (2010) “Stuffed Toys and Speech Perception”. Linguistics 48: 865–892. Kraljic, Tanya, Susan E. Brennan and Arthur G. Samuel (2008) “Accommodating Variation: Dialects, Idiolects, and Speech Processing“. Cognition 107: Niedzielski, Nancy (1999) “The Effect of Social Information on the Perception of Sociolinguistic Variables“. Journal of Language and Social Psychology 18: Pierrehumbert, Janet (2006) “The next toolkit”. Journal of Phonetics 34: 17


Download ppt "6th International Conference on Language Variation in Europe"

Similar presentations


Ads by Google