Exposure overrides the effect

of attention control in accent categorisation

​

WFLL Research Seminars, 12^th March 2025 �

Marc Barnard�m.e.barnard@qmul.ac.uk�marcbarnard.co.uk�marcbarnard.bsky.social�

Second Dialect Acquisition (SDA)

• When people move regions, they often exhibit second dialect acquisition (SDA).
• SDA is highly variable between individuals and speakers rarely adopt an entire new dialect, rather acquiring specific new variants.

​

• I explore three types of changes after migration:
• Production
• Perception:
• Sociolinguistic processing
• Sociolinguistic identity

​

​

2

Project overview

(Dossey et al., 2020)

PhD project consists of two studies:

• Medium-term longitudinal study of 26 undergraduate students from the North West of England who have moved to the South for university
• Sociolinguistic interviews

​

• Online study (259 participants) in two parts:
• Accent categorisation test
• Sociolinguistic self-concept experiment

​

3

Attention, exposure and salience in accent categorisation

Marc Barnard�m.e.barnard@qmul.ac.uk�marcbarnard.co.uk�

4

Background – sociophonetic perception

• We know that listeners associate acoustic input with social categories.
• And that this is bidirectional, so phonetic information can trigger social associations and vice versa. (Bouavichith et al., 2019; Squires, 2013)

​

• Work has also shown that adult listeners are able categorise unfamiliar speakers by regional with reasonable accuracy. (Clopper & Pisoni, 2004)
• Listeners make judgements in real time. (Montgomery & Moore, 2018)
• Sometimes using individual phonetic variants. (Ruch, 2018)

​

• What are underlying representations of these social/linguistic associations?
• How do we develop social categories from phonetic information?

5

Background – Exemplar Theory

• Exemplar theory often used to explain social/phonetic links.
• Each memory stored with contextual information
• Memory traces stored in ‘clouds’ based on similarity to previously encountered speech.

​

• Exposure plays an important role in exemplar theory, as perception is dependent on having encountered similar tokens before.
• Regional mobility improves accent categorisation ability (Clopper & Pisoni, 2006).

​

​

6

Background – Salience and social weighting

• Salience:
• More canonical forms are more easily accessed despite being more rarely encountered �(Sumner et al. 2014)
• More salient variants may be more robustly encoded. (Drager & Kirtley 2016)

​

​

• Dual route approach (Sumner et al. 2014): social weighting
• Mediated by attention: If speakers pay more attention to stimuli because it is noticeable, memories are more robust and more quickly accessed.

​

​

​

[t]

[ø]

7

Background – Attention control

• Not everyone has the same ability to control their attention.
• Domain-general cognitive factors correlate with individual differences in linguistic perception and production (e.g., Yu & Zellou, 2019).
• Attention control: the ability to direct attention towards a relevant object and inhibit attention to irrelevant input.
• Can facilitate perceptual learning (Mora & Darcy, 2023), phonological discrimination (Ou & Law, 2017), and lexical retrieval (Lev-Ari & Peperkamp, 2013)
• Little work on attention in sociophonetic processing
• Levon & Buchstaller (2015) highlight as potential factor.

​

• Could attentional differences play a role in accent categorisation?

​

8

Research questions

• RQ1: Do individual differences in attention control mediate the perception of socially variable phonetic forms?

​

• RQ2: Does exposure influence the effect of attention control?
• Do individuals with higher attention control develop even more robust representations of sociolinguistic variation after increased exposure?
• Or does experience override any attention control effects?

​

• RQ3: Does an individual’s attention control ability affect their sensitivity to variables of differing social salience?
• Are socially salient forms perceived more easily by all participants regardless of attention control?

9

Participants and variables

• Online experiment hosted on Gorilla.

​

• Three groups of participants: Northern non-mobile (88); Southern non-mobile (86); Northern mobile (85).

10

Methods

• Forced-choice single-word accent categorisation task inspired by Ruch (2018).
• Stimuli (total of 704 stim items)
• Each participant heard 176 tokens
• 6 words of each variable (BATH; STRUT; lateral; VNP)
• 5 words of each distractor (TRAP; DRESS; FLEECE; KIT; LOT)
• Each word 4x (1NF; 1NM; 1SF; 1SM)

​

• Participants had 3 seconds to respond to the stimuli by categorising as ‘Southern English’ or ‘Northern English’.

​

• Discrimination performance calculated using d-prime. This calculates a score for sensitivity while taking account of bias.

​

​

​

11

Methods – Attention tests

• Traditional attention control tests poor for individual differences research because the effect is so robust.

​

​

• Three adapted ‘squared’ tests aim to resolve this. (Burgoyne et al., 2023)
• Stroop squared
• Flanker squared
• Simon squared

​

• Attention score = mean of all three tasks

​

​

​

​

12

Results –

D-prime

by region

p<0.05

n.s.

p<0.01

Mean: 0.73

Mean: 0.974

Mean: 0.94

13

Results –

D-prime

by variable

All significantly different from ‘no regional feature’ at (p<0.001) – �pairwise t-tests (Bonferroni correction)

p<0.001

n.s.

p<0.001

n.s.

14

Results –

Attention score

by region

n.s.

n.s.

No significant difference in mean attention scores between groups

15

Discriminability model

• Difficult to model d’ with within-participant factors.

​

• Instead, Response (whether a participant selected ‘Northern English’ or ‘Southern English’) was the DV, and ‘Voice region’ – the correct answer – was added as a fixed effect in interaction with other predictors. In a logistic mixed effects model employing a probit link function, this also models discrimination ability.

​

• Random intercepts: Participant and Stimulus item. �Random slopes: Variable type by Participant; Attention by Stimulus item

​

• Fixed effects: Voice region x Region group x Attention (***);�Voice region x Variable type x Region group (***)

16

p<0.001

n.s.

p<0.001

Voice region x Region group x Attention

17

By-variable models

• Same structure as main model with ‘Variable Type’ removed.

​

p<0.001

p=0.004

p<0.001

p=0.001

p=0.018

est.=0.22

est.=0.17

est.=-0.17

est.=0.18

est.=-0.13

est.=0.12

est.=0.10

18

Discussion – Attention control

​

• Result: Non-mobile listeners’ ability to correctly associate social and linguistic information is dependent on their attention control ability.

​

• Previous work has suggested that individual differences in attention control ability could play a role in sociolinguistic awareness. (Levon & Buchstaller, 2015)
• Evidence that this is the case, testing attention control directly.
• Could have implications for who picks up on e.g. changes in progress

​

• Noticing, encoding or retrieval?

​

​

​

​

19

Discussion – Attention and Exposure

​

• Result: Mobile listeners were equally adept at categorising regional accent stimuli regardless of their attention control score.

​

• Second dialect acquisition: attention control does not appear to cause improved learning – in-person exposure leads to improved categorisation ability.
• Media exposure appears to be insufficient. Similar to perceptual processing (Barnard et al., under review; Smith et al., 2014)

​

• Exemplar Theory: Exposure appears to override any ‘attention-weighting’ effects.
• If listeners are surrounded by variation, attention less important in encoding.

​

​

​

​

​

20

Discussion – Attention and Salience

​

• Result: Attention effects patterned similarly across most variables but were stronger and more significant among salient variables: BATH and STRUT.

• Attention control could play a role on the top-down influence of stereotypes on sociophonetic processing.
• Drager & Kirtley (2016) argue that explicit social stereotypes influence processing in exemplar-based models of perception.

​

​

• Role of stereotypes undertheorised in exemplar models.

​

21

Summary

22

Thank you

Marc Barnard�m.e.barnard@qmul.ac.uk�marcbarnard.co.uk�@MarcEBarnard

23

References

• Barnard, M., Kunkel, S., Chong, A. & Lamarque, R. (under review)/ Pupil response as a measure of accent processing effort. Under review at Language and Speech.
• Bouavichith et al. (2019). Perceptual Influence of Social and Linguistic Priming are Bidirectional. Proceedings of the 19th International Congress of Phonetic Sciences.
• Burgoyne et al. (2023). Nature and measurement of attention control. Journal of experimental psychology, 152(8).
• Clopper, C. & Pisoni, D. (2004). Homebodies and army brats: Some effects of early linguistic experience and residential history on dialect categorization. Language Variation and Change, 16(1).
• Clopper, C. & Pisoni, D., (2006). Effects of region of origin and geographic mobility on perceptual dialect categorization. Language Variation and Change, 18(2).
• Drager, K. & Kirtley, M. J. (2016). Awareness, Salience, and Stereotypes in Exemplar-Based Models of Speech Production and Perception. In Babel, A., Awareness and Control in Sociolinguistic Research. Cambridge: Cambridge University Press.
• Lev-Ari, S. & Peperkamp, S. (2013). Low inhibitory skill leads to non-native perception and production in bilinguals’ native language. Journal of Phonetics, 41(5).
• Levon, E. & Buchstaller, I. (2015). Perception, cognition, and linguistic structure: The effect of linguistic modularity and cognitive style on sociolinguistic processing. Language Variation and Change, 27(3).
• Montgomery, C. & Moore, E. (2018). Evaluating S(c)illy voices: The effects of salience, stereotypes,and co-present language variables on real-time reactionsto regional speech. Language, 94(3).

24

References

• Mora, J. & Darcy, I. (2023). Individual differences in attention control and the processing of phonological contrasts in a second language. Phonetica 80(3-4).
• Ou, J. & Law, S-P. (2017). Cognitive basis of individual differences in speech perception, production and representations: The role of domain general attentional switching. Attention, Perception, & Psychophysics 79(3).
• Smith, R, et al. (2014). Cross-Accent Intelligibility of Speech in Noise: Long-Term Familiarity and Short-Term Familiarization. Quarterly Journal of Experimental Psychology, 67(3).
• Sumner, M, et al. (2014). The socially weighted encoding of spoken words: a dual-route approach to speech perception. Frontiers in Psychology 4.
• Ruch, H. (2018). The Role of Acoustic Distance and Sociolinguistic Knowledge in Dialect Identification. Frontiers in Psychology 9.
• Yu, A. & Zellou, G. (2019). Individual Differences in Language Processing: Phonology. Annual Review of Linguistics 5(1).

​

25

Probability of responding with ‘Southern English’

p<0.001

p<0.001

p=0.039

p<0.01

Voice region x Region group x Variable type

26