Auditory cortex is susceptible to lexical influence as revealed by informational vs. energetic masking of speech categorization

Abstract

Speech perception requires the grouping of acoustic information into meaningful phonetic units via the process of categorical perception (CP). Environmental masking influences speech perception and CP. However, it remains unclear at which stage of processing (encoding, decision, or both) masking affects listeners' categorization of speech signals. The purpose of this study was to determine whether linguistic interference influences the early acoustic-phonetic conversion process inherent to CP. To this end, we measured source level, event related brain potentials (ERPs) from auditory cortex (AC) and inferior frontal gyrus (IFG) as listeners rapidly categorized speech sounds along a /da/ to /ga/ continuum presented in three listening conditions: quiet, and in the presence of forward (informational masker) and time-reversed (energetic masker) 2-talker babble noise. Maskers were matched in overall SNR and spectral content and thus varied only in their degree of linguistic interference (i.e., informational masking). We hypothesized a differential effect of informational versus energetic masking on behavioral and neural categorization responses, where we predicted increased activation of frontal regions when disambiguating speech from noise, especially during lexical-informational maskers. We found (1) informational masking weakens behavioral speech phoneme identification above and beyond energetic masking; (2) low-level AC activity not only codes speech categories but is susceptible to higher-order lexical interference; (3) identifying speech amidst noise recruits a cross hemispheric circuit (AC_left → IFG_right) whose engagement varies according to task difficulty. These findings provide corroborating evidence for top-down influences on the early acoustic-phonetic analysis of speech through a coordinated interplay between frontotemporal brain areas.

Keywords: Auditory event-related potentials (ERPs); Categorical perception (CP); Lexical effects; Speech-in-noise (SIN) processing.

Figure 1:

Acoustic spectrograms for the /da/ to /ga/ speech continuum. Tokens were spaced over five equidistant steps by varying the F2 formant using STRAIGHT (Kawahara et al., 2008) based on original speech materials described in Nath and Beauchamp (2012). Effective token duration was 350 ms (zoomed here for clarity). Speech was presented at 74 dB SPL.

Figure 2:

Behavioral speech categorization is differentially hindered by informational (INF) vs. energetic (ENG) acoustic interference. (A) Perceptual psychometric functions for phoneme identification amidst various masking noise. The clean curve shows an abrupt shift in category responses indicative of strong CP. Masking linearizes identification slopes indicative of more continuous perception but this effect various with masker type; CP is weaker under informational vs. energetic interference. (B) Reaction times for speech identification. Listeners are slower at labeling speech in noise overall but are faster under energetic compared to informational masking. (C) Location of the perceptual boundary varies little with masker type. (D) Psychometric slopes are shallower for informational vs. energetic masking indicating speech categorization is hindered by concurrent linguistic competition (INF masker) above and beyond acoustic interference alone (ENG masker). Error bars here and throughout = ±1 s.e.m.

Figure 3:

Cortical speech-ERPs. Channel-level (A) scalp topographies and (B) time-domain waveforms across noise conditions. Responses represent time-locked activity to the continuum’s speech stimuli (pooled across tokens) to illustrate noise effects. Noise reduces ERP amplitudes and prolongs latencies (e.g., Cz) with more apparent changes in morphology in frontal electrodes, particularly over right hemisphere (e.g., F6).

Figure 4:

Cluster-based permutation results jointly comparing the strength of categorical processing between prototype (green) and boundary (orange) tokens across masker types and ROIs. Shading = segments where ERPs show category-level coding (i.e., Tk15 ≠ Tk3, p<0.05, corrected). (A) For clean speech, early right-lateralized IFG activity shows categorical organization. (B) During energetic masking, categorical speech processing emerges during both early and late stages in AC and IFG. (C) During informational masking, speech categories are distinguishable in early bilateral IFG and left AC, with additional later clusters in left AC and right IFG.

Figure 5:

Speech-ERP source waveforms as a function of masker extracted from bilateral auditory cortex (AC) and inferior frontal gyrus (IFG). (A) Left IFG, (B) Right IFG, (C) Left AC, and (D) Right AC. Shaded regions denote time segments where responses differ between masker type (running t-test, p<0.05) (Guthrie and Buchwald, 1991). (E-F) Source waveform amplitudes between the informational and energetic masking conditions in IFG and AC. AC and IFG amplitudes were extracted from and early (~200 ms) and late (~400 ms) time windows, respectively (see bars along abscissa, panels A-D). Neural responses are weaker when categorizing speech during informational vs. energetic masking indicating linguistic interference hinders the acoustic-phonetic conversion process of CP.

Figure 6:

Functional connectivity (Granger causality) reveals directed neural signaling from left AC to right IFG differentiates speech categorization under different forms of noise. (A) The spectrographic map shows frequency-specific AC_left→IFG_right connectivity across frequency and time of the epoch window. Dark shading, clusters where connectivity strength differed between noise conditions (F-test; p<0.05). (B-C) Connectivity strength in the low gamma band (35–40 Hz; 100–300 ms) is higher when classifying clean speech and weakens during energetic masking. Connectivity is intermediate for informational masking. **p< 0.01

Figure 7:

Pearson’s correlation showing the relationship between masker-related changes in left inferior frontal gyrus and behavioral reaction times. ΔIFG reflects the change in source amplitude within left IFG between informational and energetic masking (i.e., ΔIFG = IFG_I − IFG_E). Similarly, ΔRT reflects the change in behavior (i.e., ΔRT = RT_I − RT_E). Positive ΔIFG and ΔRT indicate larger neural responses and slower decision speeds during informational relative to energetic loads. The negative association indicates listeners who were more susceptible to lexical influences at the neural level also show less benefit between I and E behaviorally. I=informational masking; E=energetic masking.