“I don’t know”: a usage-based approach to familiar collocations in non-fluent aphasia

ABSTRACT

Background: Familiar collocations (e.g., “it’s alright”) are an important part of everyday conversation. Such word combinations are often retained in speakers with Broca’s aphasia. However, only few investigations have studied the forms and functions of familiar collocations available to speakers with Broca’s aphasia.

Aims: We first apply a frequency-based perspective to word combinations produced by speakers with Broca’s aphasia and their conversation partners (CPs), and compare the frequency characteristics of word combinations in dyadic and non-dyadic speech. Second, we investigate the conversational functions of one prominent familiar collocation, “I don’t know” (IDK).

Methods & Procedures: In the first analysis, speech samples from interactions of nine dyads (each a speaker with Broca’s aphasia and their CP) were examined. Non-dyadic samples were selected from 39 speakers with Broca’s aphasia from AphasiaBank (MacWhinney et al., 2011). The Frequency in Language Analysis Tool (FLAT; Zimmerer & Wibrow, 2015) was used to estimate collocation strength (the degree of association between words in a combination) of well-formed bigrams (two-word combinations) and trigrams (three-word combinations). The second analysis presents a qualitative investigation of uses of IDK in dyadic exchanges.

Outcomes & Results: Analysis 1 revealed that residual trigrams in Broca’s aphasia were more strongly collocated in comparison to language produced by CPs. There was no difference in frequency-based profiles between dyadic and non-dyadic aphasic speech. Analysis 2 indicated that speakers with Broca’s aphasia and CPs used IDK to achieve a variety of communicative functions. However, patterns specific to each participant group were found.

Conclusions: These findings highlight that frequency-based analysis is useful in explaining residual, grammatically well-formed word combinations in Broca’s aphasia. This study provides evidence that IDK can aid turn construction in aphasia.

KEYWORDS:

• Broca’s aphasia
• frequency-based
• collocations
• formulaic expressions

Acknowledgements

SB and MC acquired the naturalistic data. CB analysed the data, under the supervision of SB, RV and VZ. GB was rater 2 in the IDK analysis. CB wrote the first draft of the manuscript. SB, RV, VZ, MC and GB contributed to further drafts. We thank all participants who provided everyday conversational data, and the AphasiaBank team for access to semi-structured interview data. We also thank student researchers who transcribed some of the video samples, and two anonymous reviewers for their valuable comments.

Disclosure of interest

The authors have no financial conflicts of interest.

Supplementary material

Supplemental data for this article can be accessed here.

Notes

1. Dyad 2 in Best et al.

2. The FLAT 1.1 employs additive smoothing by which 1 is added to every unigram, bigram and trigram frequency count in order to avoid a frequency value of 0 to enable calculation of t-scores.

3. Unless the token occurred at the beginning or end of a sample or was followed by a long pause.

4. Due to frequencies below 5 in some cells, the chi-square test could not be applied.

5. Conventions for Cramer’s V with df = 3: .06 = small effect, .17 = medium effect, .29 = large effect; Cramer’s V with df = 4: .05 = small effect, .15 = medium effect, .25 = large effect (Goss-Sampson, 2018).

Additional information

Funding

This work was supported by an Economic and Social Research Council Training Grant [1475581] awarded to the first author.