PP1604 - Can Monosyllabic Word Recognition in Quiet Scores Predict Speech Recognition in Noise Ability: An Evaluation of the Guidelines from the National Academies

Abstract:
Monosyllabic word test results in quiet were compared to speech recognition in noise (SRN) test results to determine if the two tests measure the same construct (convergent validity). Results from the Northwestern University (NU-6) CNC and Maryland CNC tests were compared to the Hearing in Noise Test (HINT) and BKB-SIN SRN test results. No statistically significant relationships were found except for NU-6 vs. HINT results (r = 0.55, p = 0.01); however, better performance on the NU-6 test was related to poorer performance on the SRN test; implying that SRN ability cannot be predicted from monosyllabic word tests in quiet.

Summary:
The purpose of this study was to determine if monosyllabic words in quiet and speech recognition in noise (SRN) tests measured the same construct. A poor relationship between the two test types was predicted.

Rationale:
The National Academies of Sciences, Engineering, and Medicine (NASEM, 2021) recommended using monosyllabic word recognition tests in quiet as a measure of functional hearing ability. This recommendation implies monosyllabic word recognition tests in quiet may be used to infer speech recognition in noise (SRN) ability, but is not supported by previous investigations (Wilson, 2011; Torkildsen et al, 2019). Of 1601 participants with excellent word recognition in quiet (NU-6) scores, Wilson (2011) found more than 70% of participants scored below normal on the Words-in-Noise test (Wilson et al., 2007). Furthermore, Torkildsen et al. (2019) reported poor relationships between speech perception scores using monosyllabic words in quiet vs. SRN thresholds using the Norwegian Hearing in Noise Test (NHINT; Myhrum and Moen, 2008).

The present study tested NASEM’s presumption of monosyllabic word test results in quiet may be used as a predictor of SRN ability. Their recommendation was based on the committee's “state of knowledge” and lacked evidential support (NASEM, 2021). Additionally, the recommendation is not supported by previous investigations (Wilson, 2011; Torkildsen et al, 2019).

Methods:
Twenty-nine native English speakers aged 19-26 years (mean=20.86, SD = 2.26) with normal pure-tone thresholds (< 25 dB HL, 0.25 - 8.0 kHz) participated in this study; all were female but one. Maryland CNC words were spoken by a male talker and presented at 65 dBA. NU-6 words were spoken by a female talker and presented at 65 dBA (Torkildsen et al, 2019). All lists contained 50 words and were used to evaluate monosyllabic word recognition in quiet ability. The same participants were evaluated for SRN ability using the BKB-SIN in four-talker babble and the HINT Noise Front condition in steady-state speech-shaped noise presented at 65 dBA.  Both SRN noise tests used Americanized versions of the Bamford-Kowal-Bench corpus (BKB; Bench, Kowal, & Bamford, 1979). All test stimuli were delivered binaurally using supra-aural headphones in a sound-treated booth; test conditions were randomized.

Results:
Mean performances for the NU-6 and Maryland CNC were 96.7% (SD = 2.6) and 95.4% (SD = 3.4), respectively. The average performances for the HINT and BKB-SIN were –0.5 dB SNR (SD = 1.4) and –0.8 dB SNR-50 (SD = 1.5), respectively. Spearman rho statistic was used due to the non-normal distribution of the data. No significant relationships were found between Monosyllabic words vs. SRN results, except a moderate but statistically significant positive relationship between NU-6 vs. HINT results (r = 0.38, p = 0.03); indicating better NU-6 test results are associated with poorer HINT thresholds.

Conclusions:
NASEM’s presumption that monosyllabic word recognition tests in quiet can predict SRN ability was not supported by results of this study. It is recommended to test SRN ability directly as opposed to making an inference from the results of monosyllabic word tests in quiet.