Development of a measure of function word use in narrative discourse: core lexicon analysis in aphasia

Abstract

Background: Although discourse-level assessments contribute to predicting real-world performance in persons with aphasia (PWA), the use of discourse measures is uncommon in clinical settings due to resource-heavy procedures. Moreover, assessing function word use in discourse requires the arduous procedure of defining grammatical categories for each word in language transcripts. Aims: The purpose of this exploratory study was twofold: (1) to develop core function word lists as a clinicianfriendly means of evaluating function word use in discourse; and (2) to examine the ability of the core function word measure to differentiate PWA from cognitively healthy adults and persons with fluent aphasia from nonfluent aphasia. Methods & Procedures: The 25 most commonly used function words (core function words) were extracted from narrative language samples from 470 cognitively healthy adults, which were divided into seven age groups (20s, 30s, 40s, 50s, 60s, 70s and 80s). The percent agreement of core function words for 11 PWA (fluent aphasia = 5; non-fluent aphasia = 6) and 11 age-and education-matched controls were then calculated. Percent agreement for the core function words produced was compared between the controls and the PWA group, and between participants with fluent aphasia and non-fluent aphasia. Outcomes & Results: The results indicated that PWA produced fewer core function words from the lists than the control group, and that core function word use was strongly correlated with aphasia severity. Persons with non-fluent aphasia produced fewer core function words than those with fluent aphasia, although this could be a confound of aphasia classification from the use of the Western Aphasia Battery (WAB)-Revised. Conclusions & Implications: Core function word lists consisting of a limited number of items for quantifying function word use in discourse remain in a nascent stage of development. However, the findings are consistent with previous studies analysing the total production of function words in language samples produced by PWA. Therefore, core function words may potentially serve as a clinician-friendly manner of quantifying function words produced in discourse.

Applying core lexicon analysis in patients with anomic aphasia: Based on Mandarin AphasiaBank

“…High‐frequency words can be retrieved more easily than low‐frequency words at the discourse level. Furthermore, the use of core lexicon measures has been created with computational language analysis programs, which can reduce analysis errors (Dalton et al., 2020a; 2020b; Kim et al., 2019, 2021; Kim & Wright, 2020a, 2020b). Core word lists of nouns and verbs were successfully extracted in preliminary analysis and applied to further analyses.…”

“…To generate the core words through core lexicon analysis based on the normative data, the FREQ program associated with CLAN was applied, which was produced within the transcripts along with the frequency information. For the discourse task, the top 25 of the most high‐frequency words were used to create the core word lists for each group according to the cutoff of previous studies (Dalton & Richardson, 2015; Kim et al., 2021).…”

“…As a result, frequency problems were transferred to all‐or‐none questions. The percent agreement was calculated by adding the numerator throughout the list, then multiplying by four times as 100 was the full mark (Kim et al., 2021).…”

“…The further analysis from Kim et al. confirmed that core word use was strongly correlated with aphasia severity, and patients with non‐fluent aphasia produced fewer core words than patients with fluent aphasia (Kim et al., 2021).…”

Language Sample Analysis With TalkBank: An Update and Review

“…This command produces a spreadsheet showing how many and specifically which core lexicon words were used in a language sample or set of language samples. These results can be used to assess typical language usage (Dalton and Richardson, 2015;Kim et al, 2019). A recent study compared automated and manual CoreLex scoring and found them to be highly correlated, with automated scoring again requiring a small fraction of the time that it takes to train scorers and score manually 1 .…”