Investigation of the effects of semantic neighbours in aphasia: a facilitated naming study

ABSTRACT

Background

It is well established that word retrieval can be improved in people with aphasia. However, there has been little research regarding the influence of specific word properties on the success of such treatment.

Aims

This study aimed to better understand the mechanisms supporting naming treatment effects in aphasia, by exploring effects of word-specific semantic neighbourhood variables (based on featural overlap or on association strength) on the outcomes of a facilitation task.

Methods And Procedures

Two individuals, one with primarily lexical-semantic difficulties (SJS) and one with primarily lexical difficulties (DEH) participated. Their picture naming performance was assessed before and after a facilitation task in which each target word was repeated in the presence of the corresponding picture.

Outcomes And Results

Both participants showed improved naming following the facilitation task. However, for DEH, inhibitory effects of words with many semantic neighbours were enhanced by the facilitation task. For SJS, in contrast, targets with a strongly associated word in the lexicon were less likely to result in a semantic error compared to those with an associate of weaker association strength.

Conclusions

It is hypothesised that individuals like DEH, with lexical retrieval impairments, may show increased sensitivity to neighbourhood density, whereas individuals like SJS, with an impairment of the links between semantics to lemmas, may show sensitivity to other neighbourhood measures that are more likely to be encoded at the semantic level.

KEYWORDS:

• Aphasia
• treatment
• facilitated naming
• semantic neighbourhood density
• association strength

Acknowledgments

We thank DEH and SJS for their participation in this study. We thank David Howard and Wendy Best for helfpul discussions when the experiment was being planned. We thank Serje Robidoux for very useful statistics advice. An earlier version of this paper was written as part of the first author’s PhD thesis, funded by an International Macquarie University Research Excellence Scholarship (iMQRES). During the preparation of this paper, Solène Hameau was funded by the Australian Research Council Centre of Excellence in Cognition and its Disorders (CCD) [CE110001021]. Lyndsey Nickels was funded by an Australian Research Council Future Fellowship [FT120100102].

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. In a previous version of this study which formed part of the first author’s PhD thesis (Hameau, 2016), we also examined seven additional semantic neighbourhood measures. We did not include these analyses here as these measures were all moderately or highly correlated to at least one of the four measures reported here. These measures were: Mean similarity of neighbours (the average of the proportion of shared features between the target and all its neighbours); Number of higher frequency neighbours, based on lemma frequency; Word-form frequency of the closest semantic neighbour; Lemma frequency of the closest semantic neighbour; Number of associates; Word-form and Lemma frequency of the first associate.

2. We replicated the analyses performed with these association-based measures using another database (The University of South Florida Word Association, Rhyme and Word Fragment Norms: Nelson et al., 2004; we had originally chosen the Edinburgh database as being perhaps more culturally similar to our Australian population). Four items from our set were missing from these norms (ostrich, peacock, pineapple, skis). Association strength for the 192 overlapping items ranged from 9–88, average=31.61, SD=15.98). As for the Edinburgh database, on average, about 18% of the first associates were also coordinates of the target. Although the two measures were only moderately correlated (r=.55, p<.001), significant effects of StrengthAsso were maintained (or even stronger) in these additional analyses using the Nelson et al. (2004) database (see Table 6).

3. While some theories state that frequency affects phonological retrieval but not lemma retrieval (e.g., Levelt et al., 1999) it has also been hypothesised that frequency effects arise both at the level of phonological retrieval and at the level of lemma retrieval (see Kittredge, Dell, Verkuilen, & Schwartz, 2008, for a review).

Additional information

Funding

This work was supported by the Centre of Excellence in Cognition and its Disorders, Australian Research Council [CE110001021];Australian Research Council Future Fellowship to Lyndsey Nickels [FT120100102].