Large language models (LLMs) are ubiquitous in modern day natural language processing. However, previous work has shown degraded LLM performance for under-represented English dialects. We analyze the effects of typifying “standard” American English language questions as non-”standard” dialectal variants on multiple choice question answering tasks and find up to a 20% reduction in accuracy. Additionally, we investigate the grammatical basis of under-performance in non-”standard” English questions. We find that individual grammatical rules have varied effects on performance, but some are more consequential than others: three specific grammar rules (existential “it”, zero copula, and y’all) can explain the majority of performance degradation observed in multiple dialects. We call for future work to investigate bias mitigation methods focused on individual, high-impact grammatical structures.

Word frequencies are integral in linguistic studies, showing strong correlations with speakers’ cognitive abilities and other important linguistic parameters including the Age of Acquisition (AoA). However, the formulation of credible Korean word frequency norms has been obstructed by the lack of expansive speech data and a reliable part-ofspeech (POS) tagger. In this study, we unveil Korean word frequency norms (KoFREN), derived from large-scale spontaneous speech corpora (41 million words) that include a balanced representation of gender and age. We employed a machine learning-powered POS tagger, showcasing accuracy on par with human annotators. Our frequency norms correlate significantly with external studies’ lexical decision time (LDT) and AoA measures. KoFREN also aligns with English counterparts sourced from SUBTLEX_US - an English word frequency measure that has been frequently used in the literature. KoFREN is poised to facilitate research in spontaneous Contemporary Korean and can be utilized in many fields, including clinical studies of Korean patients.

In this study, we rigorously evaluated eight machine learning and deep learning classifiers for identifying Alzheimer’s Disease (AD) patients using crosslinguistic acoustic features automatically extracted from one-minute oral picture descriptions produced by speakers of American English, Korean, and Mandarin Chinese. We employed eGeMAPSv2 and ComParE feature sets on segmented and non-segmented audio data. The Multilayer Perceptron model showed the highest performance, achieving an accuracy of 83.54% and an AUC of 0.8 on the ComParE features extracted from non-segmented picture description data. Our findings suggest that classifiers trained with acoustic features extracted from one-minute picture description data in multiple languages are highly promising as a quick, language-universal, large-scale, remote screening tool for AD. However, the dataset included predominantly English-speaking participants, indicating the need for more balanced multilingual datasets in future research.