Premium
Automated lexical and acoustic analysis of young and older healthy adults
Author(s) -
Cho Sunghye,
Nevler Naomi,
Shellikeri Sanjana,
Parjane Natalia,
Ryant Neville,
Ash Sharon,
Irwin David J,
Cieri Christopher,
Liberman Mark Y,
Grossman Murray
Publication year - 2020
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1002/alz.038284
Subject(s) - noun , concreteness , age of acquisition , psychology , duration (music) , ambiguity , linguistics , computer science , connected speech , speech recognition , audiology , natural language processing , cognition , cognitive psychology , medicine , art , philosophy , literature , neuroscience
Background Language use is affected by factors such as an individual’s age and sex. An understanding of these factors is essential to studies of speech in neurodegenerative disease. While age and sex have received considerable attention in the literature, results are mixed. Also, very few studies have considered both lexical and acoustic features at the same time, which leaves a major gap in our understanding of the effect of age and sex on language use. In this study, we analyze both lexical and acoustic features from 1‐minute speech samples using novel, objective, reproducible, fully automated methods. Method We examined digitized Cookie Theft picture descriptions produced by 37 older (52‐89y, mean=68y) and 76 young (18‐22y, mean=20y) participants. Using modern natural language processing and automatic speech recognition tools, we automatically annotated part‐of‐speech categories of all tokens and rated nouns and verbs for five lexical features, including word frequency, familiarity, concreteness, age of acquisition and semantic ambiguity. We automatically segmented each sample into speech and silent pause segments; extracted acoustic features such as total speech time, mean speech segment duration, and mean pause duration; and measured pitch percentiles from all speakers. Result Older speakers produced significantly more interjections ( p =0.023), pronouns ( p <0.001), and verbs ( p =0.004), and fewer conjunctions ( p =0.013), determiners ( p =0.009), nouns ( p =0.049), and prepositions ( p =0.002) than young participants. Older speakers’ nouns and verbs were more familiar (nouns: p =0.008, verbs: p <0.001), more frequent (verbs only: p =0.002), and less ambiguous (nouns: p =0.049, verbs: p =0.057) compared to those of young speakers. Older speakers produced shorter utterances ( p =0.001) with a lower type/token ratio for nouns ( p =0.016) than young participants. Also, older participants produced shorter speech segments ( p =0.017) and longer pauses ( p <0.001) with increased total speech time ( p <0.001) and total number of words ( p =0.002). Lastly, we observed interactions of age and sex in pitch ranges (F(1,109)=4.37, p =0.039), the number of conjunctions (F(1,109)=9.52, p =0.003) and tense‐inflected verbs (F(1,109)=4.02, p =0.047). Conclusion These results suggest that older speakers’ lexical content is less diverse and they use shorter utterances than young participants. The findings show that lexical and acoustic characteristics of semi‐structured speech samples can be examined using automated methods.