Semantic Predictability of the QuickSIN Sentences: A Cloze Probability Analysis

Alyssa Smith

Co-Presenters: Iyad Ghanim

College: College of Health Professions and Human Services

Department: Communication Sci & Disorders

Abstract:

Purpose: Difficulty hearing in adverse listening environments is a common patient report in the audiology clinic. Supplemental tests are commonly administered to assess a patient’s ability to perceive speech-in-noise (SIN). In this study, we evaluated the 72 items of the QuickSIN, a common SIN test, particularly focusing on evaluating the semantic context in the sentences. The premise of this work is that semantic context may not only help facilitate processing of sentences presented in noise but also differs substantially between patients based on individual factors. Though the impact of semantic context may be slight on a per-sentence basis, the imbalance in processing effort across the sentences may potentially impact the reliability of QuickSIN results.Objective: To evaluate the semantic predictability of the sentences used in the QuickSIN, a clinical assessment tool of speech in noise perception, with the aims of evaluating the consistency of the stimuli used to assess SIN deficits.Methods: 86 college-aged participants completed a cloze task procedure in which they viewed the QuickSIN sentence stems without the final word on a computer screen and supplied the most-likely ending. Upon completion, participants received a gift card. The procedures in this study were approved by the Kean University IRB (# FY2025-111).For each sentence, unique responses were tallied and the frequency of the most-popular supplied ending was calculated (the “cloze rate”) using an Excel formula. We identified sentences with a high cloze proportion, defined as .67 as per Block and Baldwin (2010).Results: Tallied responses showed inconsistencies in the semantic predictability of the QuickSIN sentences, with 8 predictably ending sentences and 64 unpredictably ending sentences, about half of which did not match the QuickSIN endings.Conclusions: Results indicate inconsistencies in the QuickSIN stimuli set. Because speakers process linguistic context differently, these results suggest that an auditory AIN assessment should be normed for semantic predictability to accurately assess auditory rather than linguistic processing.Significance: Our results indicate that the QuickSIN sentences vary in their degree of predictability. These findings warrant continued investigation and standardization of the semantic context imbued in the QuickSIN and like SIN assessments.