The Effect of Semantic Interference of Babble Maskers on the Perception of AzBio Sentences

Abstract

Speech in noise testing is conducted to evaluate speech perception in everyday listening conditions. Speech in noise testing involves speech recognition in the presence of noise or talker babble. A babble masker contains semantic (meaningful) information that may affect the results of the speech perception test. The goal of this study was to determine the effect of semantic masker content on speech perception. Semantic interference is a phenomenon in which semantic content of the background babble interferes with intelligibility of the target speech. In the present study, AzBio sentences were used to measure speech perception ability. The maskers for this study included 4-talker babble and 10-talker babble presented forward and reversed. Semantic interference was calculated as the AzBio score for the reversed babble condition minus the score for the forward babble condition. It was hypothesized that semantic interference would be greater for the 4-talker babble than for the 10-talker babble listening condition. Poorer speech perception ability was found in the presence of forward babble than in reversed babble for each talker condition. Semantic interference for the 4-talker babble condition was 19.69 percentage points, and it was greater than the semantic interference for the 10-talker condition, which was 9.88 percentage points. The difference in semantic interference between these conditions was 9.81 percentage points, which was statistically significant (p < 0.01). The secondary purpose of this study was to investigate the relationship between pure-tone threshold averages vs. AzBio performances for the four listening conditions. It was hypothesized that a weak and statistically non-significant relationship would be found between these two variables. There was no statistically significant relationship found between bilateral pure-tone averages and AzBio scores. It was concluded that it is more difficult to perceive speech when there is semantic content in the background babble than when there is not. It was also concluded that speech recognition in noise (SRN) ability cannot be predicted by pure-tone thresholds. This study can be applied clinically, as speech in noise testing can provide audiologists with information about a patient's hearing ability that is unavailable from the standard pure-tone threshold testing. Detecting the amount of semantic interference and understanding how it may impact a patient’s hearing allows for more useful interpretations of test results and provides for proper counseling and recommendations for hearing disorders.