Impaired language in Alzheimer’s disease: comparing English and Persian implicates word frequency rather than noun-verb distinction

Introduction: This study contributes to a growing consensus that factors other than an overly broad distinction between nouns and verbs impact language production in neurological conditions like Alzheimer’s disease (AD). Although traditionally, frontal brain damage was linked to verb deficits and posterior damage to noun processing difficulties, patients with AD exhibit impairments in both word classes despite their predominant temporoparietal atrophy. Methods: The study analyzed language samples from 186 English and 54 Persian speakers, including healthy individuals (HC) and people with AD (pwAD), as they described a picture. Transcripts were processed using a natural language processing toolkit to extract part-of-speech tags and calculate word frequencies. Linear regression models examined the effects of language and AD status on verb frequency and also on noun frequency. ROC curves compared the predictive power of the combined noun and verb frequencies versus their individual frequencies in distinguishing individuals with AD from healthy controls. Results: English-speaking pwAD relied significantly more on high-frequency verbs compared to HC. In contrast, Persian-speaking pwAD did not show this pattern, with no significant difference in verb frequency between patients and controls. The proportion of nouns to verbs did not differ significantly between pwAD and HC groups in either language. Regression analyses incorporating word frequency as a single predictor performed as well as or better than models distinguishing nouns and verbs in classifying AD. Discussion: Differences between English and Persian allow us to unconfound lexical frequency and grammatical category distinctions. Our findings indicate that the core deficit in AD language production is difficulty accessing low-frequency words rather than greater difficulty accessing a specific word class.