Lexical bias in word recognition by cochlear implant listeners

Malleability of the Lexical Bias Effect for Acoustically Degraded Speech

OBJECTIVES

Lexical bias is a phenomenon wherein impoverished speech signals tend to be perceived in line with the word context in which they are heard. Previous research demonstrated that lexical bias may guide processing when the acoustic signal is degraded, as in the case of cochlear implant (CI) users. The goal of the present study was twofold: (1) replicate previous lab-based work demonstrating a lexical bias for acoustically degraded speech using online research methods, and (2) characterize the malleability of the lexical bias effect following a period of auditory training. We hypothesized that structured experience via auditory training would minimize reliance on lexical context during phonetic categorization for degraded speech, resulting in a reduced lexical bias.

DESIGN

In experiment 1, CI users and normal hearing (NH) listeners categorized along 2 /b/-/g/ continua (BAP-GAP; BACK-GACK). NH listeners heard each continuum in a clear and eight-channel noise-vocoded format, while CI users categorized for clear speech. In experiment 2, a separate group of NH listeners completed a same/different auditory discrimination training task with feedback and then completed phonetic categorization for eight-channel noise-vocoded /b/-/g/ continua.

RESULTS

In experiment 1, we observed a lexical bias effect in both CI users and NH listeners such that listeners more consistently categorized speech continua in line with the lexical context. In NH listeners, an enhanced lexical bias effect was observed for the eight-channel noise-vocoded speech condition, while both CI users and the clear speech condition showed a relatively weaker lexical bias. In experiment 2, structured training altered phonetic categorization and reliance on lexical context. Namely, the magnitude of the lexical bias effect decreased following a short period of auditory training relative to untrained listeners.

CONCLUSIONS

Findings from experiment 1 replicate and extend previous work, suggesting that web-based methods may provide alternative routes for testing phonetic categorization in NH and hearing-impaired listeners. Moreover, findings from experiment 2 suggest that lexical bias is not a static phenomenon; rather, experience via auditory training can dynamically alter reliance on lexical context for speech categorization. These findings extend theoretical models of speech processing in terms of how top-down information is weighted for listeners adapting to acoustically degraded speech. Finally, these findings hold clinical implications for tracking changes in phonetic categorization and reliance on lexical context throughout the CI adaptation process.

Advance Contextual Clues Alleviate Listening Effort During Sentence Repair in Listeners With Hearing Aids

PURPOSE

When words are misperceived, listeners can rely on later context to repair an auditory perception, at the cost of increased effort. The current study examines whether the effort to repair a missing word in a sentence is alleviated when the listener has some advance knowledge of what to expect in the sentence.

METHOD

Sixteen adults with hearing aids and 17 with typical hearing heard sentences with a missing word that was followed by context sufficient to infer what the word was. They repeated the sentences with the missing words repaired. Sentences were preceded by visual text on the screen showing either "XXXX" (unprimed) or a priming word previewing the word that would be masked in the auditory signal. Along with intelligibility measures, pupillometry was used as an index of listening effort over the course of each trial to measure how priming influenced the effort needed to mentally repair a missing word.

RESULTS

When listeners were primed for the word that would need to be repaired in an upcoming sentence, listening effort was reduced, as indicated by pupil size returning more quickly toward baseline after the sentence was heard. Priming reduced the lingering cost of mental repair in both listener groups. For the group with hearing loss, priming also reduced the prevalence of errors on target words and words other than the target word in the sentence, suggesting that priming preserves the cognitive resources needed to process the whole sentence.

CONCLUSION

These results suggest that listeners with typical hearing and with hearing loss can benefit from priming (advance cueing) during speech recognition, to accurately repair speech and to process the speech less effortfully.

The time course of cognitive effort during disrupted speech

Listeners often find themselves in scenarios where speech is disrupted, misperceived, or otherwise difficult to recognise. In these situations, many individuals report exerting additional effort to understand speech, even when repairing speech may be difficult or impossible. This investigation aimed to characterise cognitive efforts across time during both sentence listening and a post-sentence retention interval by observing the pupillary response of participants with normal to borderline-normal hearing in response to two interrupted speech conditions: sentences interrupted by gaps of silence or bursts of noise. The pupillary response serves as a measure of the cumulative resources devoted to task completion. Both interruption conditions resulted in significantly greater levels of pupil dilation than the uninterrupted speech condition. Just prior to the end of a sentence, trials periodically interrupted by bursts of noise elicited greater pupil dilation than the silent-interrupted condition. Compared to the uninterrupted condition, both interruption conditions resulted in increased dilation after sentence end but before repetition, possibly reflecting sustained processing demands. Understanding pupil dilation as a marker of cognitive effort is important for clinicians and researchers when assessing the additional effort exerted by listeners with hearing loss who may use cochlear implants or hearing aids. Even when successful perceptual repair is unlikely, listeners may continue to exert increased effort when processing misperceived speech, which could cause them to miss upcoming speech or may contribute to heightened listening fatigue.

Lexical effects on talker discrimination in adult cochlear implant usersa)

The lexical and phonological content of an utterance impacts the processing of talker-specific details in normal-hearing (NH) listeners. Adult cochlear implant (CI) users demonstrate difficulties in talker discrimination, particularly for same-gender talker pairs, which may alter the reliance on lexical information in talker discrimination. The current study examined the effect of lexical content on talker discrimination in 24 adult CI users. In a remote AX talker discrimination task, word pairs-produced either by the same talker (ST) or different talkers with the same (DT-SG) or mixed genders (DT-MG)-were either lexically easy (high frequency, low neighborhood density) or lexically hard (low frequency, high neighborhood density). The task was completed in quiet and multi-talker babble (MTB). Results showed an effect of lexical difficulty on talker discrimination, for same-gender talker pairs in both quiet and MTB. CI users showed greater sensitivity in quiet as well as less response bias in both quiet and MTB for lexically easy words compared to lexically hard words. These results suggest that CI users make use of lexical content in same-gender talker discrimination, providing evidence for the contribution of linguistic information to the processing of degraded talker information by adult CI users.

Phonological discrimination and contrast detection in pupillometry

Introduction

The perception of phonemes is guided by both low-level acoustic cues and high-level linguistic context. However, differentiating between these two types of processing can be challenging. In this study, we explore the utility of pupillometry as a tool to investigate both low- and high-level processing of phonological stimuli, with a particular focus on its ability to capture novelty detection and cognitive processing during speech perception.

Methods

Pupillometric traces were recorded from a sample of 22 Danish-speaking adults, with self-reported normal hearing, while performing two phonological-contrast perception tasks: a nonword discrimination task, which included minimal-pair combinations specific to the Danish language, and a nonword detection task involving the detection of phonologically modified words within sentences. The study explored the perception of contrasts in both unprocessed speech and degraded speech input, processed with a vocoder.

Results

No difference in peak pupil dilation was observed when the contrast occurred between two isolated nonwords in the nonword discrimination task. For unprocessed speech, higher peak pupil dilations were measured when phonologically modified words were detected within a sentence compared to sentences without the nonwords. For vocoded speech, higher peak pupil dilation was observed for sentence stimuli, but not for the isolated nonwords, although performance decreased similarly for both tasks.

Conclusion

Our findings demonstrate the complexity of pupil dynamics in the presence of acoustic and phonological manipulation. Pupil responses seemed to reflect higher-level cognitive and lexical processing related to phonological perception rather than low-level perception of acoustic cues. However, the incorporation of multiple talkers in the stimuli, coupled with the relatively low task complexity, may have affected the pupil dilation.

The Role of Word Content, Sentence Information, and Vocoding for Voice Cue Perception

PURPOSE

For voice perception, two voice cues, the fundamental frequency (fo) and/or vocal tract length (VTL), seem to largely contribute to identification of voices and speaker characteristics. Acoustic content related to these voice cues is altered in cochlear implant transmitted speech, rendering voice perception difficult for the implant user. In everyday listening, there could be some facilitation from top-down compensatory mechanisms such as from use of linguistic content. Recently, we have shown a lexical content benefit on just-noticeable differences (JNDs) in VTL perception, which was not affected by vocoding. Whether this observed benefit relates to lexicality or phonemic content and whether additional sentence information can affect voice cue perception as well were investigated in this study.

METHOD

This study examined lexical benefit on VTL perception, by comparing words, time-reversed words, and nonwords, to investigate the contribution of lexical (words vs. nonwords) or phonetic (nonwords vs. reversed words) information. In addition, we investigated the effect of amount of speech (auditory) information on fo and VTL voice cue perception, by comparing words to sentences. In both experiments, nonvocoded and vocoded auditory stimuli were presented.

RESULTS

The outcomes showed a replication of the detrimental effect reversed words have on VTL perception. Smaller JNDs were shown for stimuli containing lexical and/or phonemic information. Experiment 2 showed a benefit in processing full sentences compared to single words in both fo and VTL perception. In both experiments, there was an effect of vocoding, which only interacted with sentence information for fo.

CONCLUSIONS

In addition to previous findings suggesting a lexical benefit, the current results show, more specifically, that lexical and phonemic information improves VTL perception. fo and VTL perception benefits from more sentence information compared to words. These results indicate that cochlear implant users may be able to partially compensate for voice cue perception difficulties by relying on the linguistic content and rich acoustic cues of everyday speech.

SUPPLEMENTAL MATERIAL

https://doi.org/10.23641/asha.23796405.

Identifying Listeners Whose Speech Intelligibility Depends on a Quiet Extra Moment After a Sentence

PURPOSE

An extra moment after a sentence is spoken may be important for listeners with hearing loss to mentally repair misperceptions during listening. The current audiologic test battery cannot distinguish between a listener who repaired a misperception versus a listener who heard the speech accurately with no need for repair. This study aims to develop a behavioral method to identify individuals who are at risk for relying on a quiet moment after a sentence.

METHOD

Forty-three individuals with hearing loss (32 cochlear implant users, 11 hearing aid users) heard sentences that were followed by either 2 s of silence or 2 s of babble noise. Both high- and low-context sentences were used in the task.

RESULTS

Some individuals showed notable benefit in accuracy scores (particularly for high-context sentences) when given an extra moment of silent time following the sentence. This benefit was highly variable across individuals and sometimes absent altogether. However, the group-level patterns of results were mainly explained by the use of context and successful perception of the words preceding sentence-final words.

CONCLUSIONS

These results suggest that some but not all individuals improve their speech recognition score by relying on a quiet moment after a sentence, and that this fragility of speech recognition cannot be assessed using one isolated utterance at a time. Reliance on a quiet moment to repair perceptions would potentially impede the perception of an upcoming utterance, making continuous communication in real-world scenarios difficult especially for individuals with hearing loss. The methods used in this study-along with some simple modifications if necessary-could potentially identify patients with hearing loss who retroactively repair mistakes by using clinically feasible methods that can ultimately lead to better patient-centered hearing health care.

SUPPLEMENTAL MATERIAL

https://doi.org/10.23641/asha.21644801.

Effects of stimulus naturalness and contralateral interferers on lexical bias in consonant identification

Lexical bias is the tendency to perceive an ambiguous speech sound as a phoneme completing a word; more ambiguity typically causes greater reliance on lexical knowledge. A speech sound ambiguous between /g/ and /k/ is more likely to be perceived as /g/ before /ɪft/ and as /k/ before /ɪs/. The magnitude of this difference-the Ganong shift-increases when high cognitive load limits available processing resources. The effects of stimulus naturalness and informational masking on Ganong shifts and reaction times were explored. Tokens between /gɪ/ and /kɪ/ were generated using morphing software, from which two continua were created ("giss"-"kiss" and "gift"-"kift"). In experiment 1, Ganong shifts were considerably larger for sine- than noise-vocoded versions of these continua, presumably because the spectral sparsity and unnatural timbre of the former increased cognitive load. In experiment 2, noise-vocoded stimuli were presented alone or accompanied by contralateral interferers with constant within-band amplitude envelope, or within-band envelope variation that was the same or different across bands. The latter, with its implied spectro-temporal variation, was predicted to cause the greatest cognitive load. Reaction-time measures matched this prediction; Ganong shifts showed some evidence of greater lexical bias for frequency-varying interferers, but were influenced by context effects and diminished over time.

Accuracy and cue use in word segmentation for cochlear-implant listeners and normal-hearing listeners presented vocoded speech

Cochlear-implant (CI) listeners experience signal degradation, which leads to poorer speech perception than normal-hearing (NH) listeners. In the present study, difficulty with word segmentation, the process of perceptually parsing the speech stream into separate words, is considered as a possible contributor to this decrease in performance. CI listeners were compared to a group of NH listeners (presented with unprocessed speech and eight-channel noise-vocoded speech) in their ability to segment phrases with word segmentation ambiguities (e.g., "an iceman" vs "a nice man"). The results showed that CI listeners and NH listeners were worse at segmenting words when hearing processed speech than NH listeners were when presented with unprocessed speech. When viewed at a broad level, all of the groups used cues to word segmentation in similar ways. Detailed analyses, however, indicated that the two processed speech groups weighted top-down knowledge cues to word boundaries more and weighted acoustic cues to word boundaries less relative to NH listeners presented with unprocessed speech.

Listening Effort Is Not the Same as Speech Intelligibility Score

Listening effort is a valuable and important notion to measure because it is among the primary complaints of people with hearing loss. It is tempting and intuitive to accept speech intelligibility scores as a proxy for listening effort, but this link is likely oversimplified and lacks actionable explanatory power. This study was conducted to explain the mechanisms of listening effort that are not captured by intelligibility scores, using sentence-repetition tasks where specific kinds of mistakes were prospectively planned or analyzed retrospectively. Effort measured as changes in pupil size among 20 listeners with normal hearing and 19 listeners with cochlear implants. Experiment 1 demonstrates that mental correction of misperceived words increases effort even when responses are correct. Experiment 2 shows that for incorrect responses, listening effort is not a function of the proportion of words correct but is rather driven by the types of errors, position of errors within a sentence, and the need to resolve ambiguity, reflecting how easily the listener can make sense of a perception. A simple taxonomy of error types is provided that is both intuitive and consistent with data from these two experiments. The diversity of errors in these experiments implies that speech perception tasks can be designed prospectively to elicit the mistakes that are more closely linked with effort. Although mental corrective action and number of mistakes can scale together in many experiments, it is possible to dissociate them to advance toward a more explanatory (rather than correlational) account of listening effort.