(PP1326) The Effect of Semantic Interference on Speech Perception in Noise Ability

Objectives: The goal of this study was to determine the effect of masker type on speech perception in noise abilities, specifically the effect of semantically meaningful vs. semantically anomalous masker types. Speech perception in noise ability was evaluated using the AzBio test. We hypothesized that better speech perception in noise ability would be found for conditions with semantic content as opposed to those without. 

Rationale: Cherry (1953) described the “cocktail party” problem where the listener experiences difficulty understanding speech in the presence of multiple talkers. Carhart (1969) stated that semantic interference occurs when speech perception is poorer in masker conditions with greater semantic content. Sperry et al (1997) evaluated word recognition performances in a meaningful mulittalker competing message and the same message in reverse to eliminate semantic content. They found 15% better speech recognition performance with the backward masker than with the forward masker. Sperry and colleagues concluded that the presence of semantic content in the forward masker had more of a detrimental effect on speech recognition compared to the masker condition without semantic content. The present study is an investigation of this effect using AzBio sentences in forward and reversed speech maskers.

Design: Thirty native English speakers with normal pure-tone thresholds (≤ 25 dB HL, 0.25 – 4.0 kHz) with a mean age 20.9 years participated in this study. The AzBio test (Spahr et al, 2012) was used to evaluate binaural speech perception in four-talker babble forward, four-talker babble in reverse, conversational masker forward, and conversational masker in reverse. The AzBio sentences and maskers were presented at 65 dBA (0 dB SNR). The AzBio was administered binaurally under supra-aural headphones in a sound-treated booth. All test conditions and sentence lists were randomized. 

Results: A repeated-measures ANOVA was conducted to investigate the effects of masker condition on AzBio scores. This analysis revealed that the main effect was statistically significant (F value is 4.690 for masker condition, p < 0.0001). A post hoc analysis was conducted using one-sample t-tests. A statistically significant poorer performance was found for the four-talker forward condition than the four-talker in reverse condition (22.014 percentage points, p < .0001). A small (3.75 percentage points) but statistically significant improvement in speech perception was found for the conversational masker with as opposed to without semantic content (p = 0.03). Semantic interference was observed for the four-talker babble and the conversational masker conditions. 

Conclusions: A significant main effect for masker condition was observed in the present study. Poorer speech perception was found for the forward masker conditions than for the same masker conditions in reverse. Greater semantic interference was found for the four-talker babble conditions as opposed to the conversational masker conditions. This may be due to the greater semantic content of the four-talker babble masker than the conversational masker. The study results support the hypothesis and are consistent with Sperry et al (1997).