Research

The sounds we experience in our everyday communication can vary greatly depending on the environment. Paradoxically, increasing the sound level may lead to worse speech perception, especially in noise. This decline is known as “rollover.” It is unclear how rollover is experienced differentially across aging groups, as well as for those with and without hearing loss. In addition, there is mounting evidence that listening effort plays an important role in challenging listening conditions and can be directly quantified with objective measures such as pupil dilation. To investigate the interacting effects of temporality, sound level, background noise, and effort on perception, we assessed word identification at multiple levels and in quiet and in noise, in younger (≤25 years) and older (>50 years) listeners without or with mild-to-moderate hearing loss. Pupillometry was used to assess listening effort during these measurements. Behavioral results demonstrated that rollover occurred for both temporal and non-temporal speech contrasts. Rollover occurred in older normal-hearing and older hearing-impaired listeners when speech was presented in background noise. Pupillometry results demonstrated that listening effort was modulated by sound level and that greater listening effort was expended primarily in the presence of background noise. The effect of increased sound level on listening effort was exacerbated by age and hearing loss, with the greatest effect in older listeners with hearing loss. The age- and hearing-dependent effects of rollover highlight the potential negative impact of amplification of high sound levels and therefore has implications for effective treatment of age-related hearing loss.

Abstract

Purpose: Advances in automatic speech recognition (ASR) and natural language processing have the potential to meaningfully impact language assessment. Despite the increasing availability of these technologies, little is known about their performance in routine assessment tasks such as language sample analysis (LSA), which is commonly used in the identification of developmental language disorder (DLD). This study examined the performance of ASR in transcribing child language samples, as well as the concurrent validity and diagnostic accuracy of metrics derived from an automated LSA system.

Method: We used an existing corpus of narrative samples from 66 children with DLD and 285 with typical language (TL), which included audio recordings and transcripts. We conducted automated LSA using the proprietary software LSA+, which integrates ASR and natural language processing. We evaluated transcription accuracy by comparing ASR transcripts to those included in the corpus. We then examined associations between language ability metrics from ASR and manual transcripts, as well as the diagnostic accuracy of each, by using LSA+ as a common means of analysis for all transcripts.

Results: ASR produced less accurate transcripts for children with DLD than for those with TL and for younger children. Despite differences in transcription accuracy, automated LSA measures exhibited positive correlations with their manual counterparts, ranging from r = .79 to .98. Manual LSA measures in isolation exhibited better diagnostic accuracy than that of their automated counterparts. However, composite models with multiple automated metrics yielded the highest accuracy when including mean length of utterance, number of different words, and total number of words, in addition to ASR-derived mean length of speech segment and pauses per word.

Discussion: Full automation of LSA demonstrates the potential to improve the efficiency and robustness of language assessment, particularly when utilizing composite measures. Despite its advantages, automated LSA may result in inaccurate estimates of language ability if used without consideration of its limitations. Continued examination of ASR and related technologies is needed to better understand their efficacy across languages, ages, and types of communication disorders.

Abstract: Given that the U.S. K-12 immigrant and multilingual learner (ML) populations are growing faster than professional development opportunities for in-service educators to address these students’ academic, language, and socio-emotional needs, this chapter explores the grassroots development process of a post-master's certificate program to equip in-service professional school counselors (PSCs) with the specialized knowledge and skills necessary to effectively serve this growing student population. Here, we lay out a roadmap of the process involved in creating and implementing the program, from the initial cross-disciplinary collaboration and administrative logistics to the curriculum development and delivery stages. We also shed light on the impact and lessons learned from the first initial cohorts of participants. Finally, we offer insights and recommendations for future improvements and adaptations to the program, as well as describe how this program exemplifies the need for more interdisciplinary collaborations in meeting the evolving needs of immigrant and ML student populations.

Aging and hearing loss are associated with difficulty understanding fast speech. Here, we investigated how peripheral hearing sensitivity, neural speech representation, and cognitive processing contribute to this difficulty. For this purpose, we collected electrophysiological, cognitive, and behavioral assessments from 30 younger normal-hearing adults, 32 older normal-hearing adults, and 26 older hearing-impaired adults. Subcortical envelope-following responses to monosyllabic words and cortical envelope tracking to continuous speech were recorded at 0% and 40% time compression. Cognitive performance was assessed using the NIH Toolbox. Time-compressed sentence recognition was measured using an adaptive approach to determine the time compression ratio that corresponded to 50% correct performance. Younger participants showed stronger subcortical envelope encoding than older participants, and phase locking decreased with time compression across groups. Cortical envelope tracking was greater in older adults, with and without hearing loss, than in younger participants, and increased in the time-compression condition. Younger participants had higher processing speed scores than older participants and achieved 50% correct performance with higher time compression ratios than older participants. Audiometric thresholds, cortical envelope tracking to natural-rate speech, and processing speed significantly predicted time-compressed speech recognition performance. These results suggest that while both peripheral hearing sensitivity and processing speed are predictive of time-compressed speech recognition, cortical mechanisms appear to contribute to time-compressed speech recognition primarily when temporal cues are faithfully preserved in the speech signal.

This study investigates to what extent various factors affect the stop production of North Korean refugee speakers living in Seoul, South Korea, from the perspective of second dialect acquisition (SDA). The VOT and F0 values of 6,495 stops were measured from twenty-two North Korean speakers (F: 16, M: 6). The factors examined were age of arrival, length of residence (LoR), and three different aspects of social adaptation (identification, assimilation, and language attitudes). Our findings showed that the North Korean speakers with stronger South Korean identity and longer LoR produced more Seoul standard South Korean-like VOT patterns. Moreover, the speakers with positive language attitudes toward South Korea and longer LoR produced more South Korean-like F0 patterns. The results suggest that understanding various dimensions of adaptation is important to explain North Korean refugees’ SDA.

Abstract: Elicited production studies have suggested that children up to age five sometimes misuse singular definite descriptions (the N) in contexts where the referent is not mutually known or uniquely identifiable in the domain of reference. These errors have been variously attributed to children’s developing ability to reason about the perspectives of others, an incorrect semantics for the, or pragmatic immaturity. To assess these claims, we first examine children’s spontaneous use of the—its frequency, communicative outcomes, and adult-likeness—then turn to a controlled elicited production task to diagnose whether children can succeed with the requirements of definiteness. We find that in natural speech, children use the at comparable rates to their mothers overall, with their distribution pattern across syntactic environments largely mirroring their mothers’ (Corpus Study 1), and with rare communication breakdowns (Corpus Study 2). We further find that 2- to 4- year-old children’s production of the and a are highly guessable, just like that of their caregivers, suggesting that children’s article choices show substantial similarities to adults’ (Experiment 1). Finally, in a new elicitation task, we find that most 3- to 4-year-olds succeed in producing definite descriptions; they only use the when the referent is mutually known or uniquely identifiable in the context, although some differences between child and adult usage remain (Experiment 2). These findings suggest that while preschoolers make errors sometimes, they are, by and large, sensitive to the conditions that license singular definite descriptions.

Addressing multilingual learners’ literacy development requires understanding cross-language relationships between their first and second languages. The current study investigated how Spanish language and literacy skills relate to English language and literacy skills in 106 Spanish-speaking multilingual learners in kindergarten through second grade. Using an achievement test in both Spanish and English, multilevel models revealed three key findings. First, there was partial support for the interdependence continuum: Spanish decoding significantly predicted English decoding (b = 0.73, p < .001), but Spanish reading comprehension and oral language did not predict their English counterparts. Second, Spanish decoding and oral language showed no significant cross-domain relationships with English reading comprehension. Third, English oral language did not moderate Spanish-to-English relationships, but English oral language was a consistent and significant predictor of English reading comprehension, while Spanish skills did not uniquely predict English reading comprehension beyond English oral language. Findings suggest the importance of L1 word-level skills instruction to support multilingual learners’ English literacy development in early elementary grades.

Picture books, which feature the display of both text and images, have been widely used in second language (L2) learning and teaching. During picture book reading, learners need to split their attention between text and images. This behavior is believed to be crucial in cross-modal integration of information that supports comprehension. Despite its importance, integrative saccades, which refer to rapid eye movements between information presented in different modalities, have received little attention in previous L2 multimodal input research. Moreover, studies that capture attention-based learning processes often overlook the ongoing learning processes as a mediator in the treatment-outcome relationship (Jia, Ruirui, and Bronson Hui. 2025. “Modeling Relationships Between Learning Conditions, Processes, and Outcomes: An Introduction to Mediation Analysis in SLA Research.” Studies in Second Language Acquisition 47 (3): 912–47). To fill these gaps, the present brief report reanalyzed the eye-tracking data from Tytko, Tetiana, Bronson Hui, and Nick Pandža (2026. “The Effect of Multimodal Input on L2 Learners’ Reading Comprehension: A Pre-registered, Eye-Tracking Study.” Modern Language Journal 110 (1): 210–39), which examined the effect of different combinations of multimodal input reading conditions: reading-only (RO) versus reading with images (RI) versus reading with images and audio (RIA) on comprehension. With their data, we aim to explore the extent to which the effect of reading conditions on reading comprehension was mediated by the number of integrative saccades. To address this intermediary relationship, we conducted multilevel mediation analysis. It was found that, through integrative saccades, the RIA condition led to better reading comprehension than the RI condition, confirming the hypothesized mechanism that the inclusion of the audio in the RIA condition allows L2 learners to go back and forth between text and images more frequently, and this increased integration of verbal and pictorial information, in turn, led to better reading comprehension.

Hearing loss and related impairments are among the most prevalent service-related disabilities affecting military Veterans. Yet, many of the cognitive and psychosocial implications related to auditory functioning remain insufficiently understood. Growing evidence suggests that hearing aid usage can alleviate cognitive symptoms and reduce psychological, social, and emotional burdens associated with hearing loss. The present study examined associations between hearing status, overall perceptions of health, and neurocognitive functioning in combat deployed Veterans. More specifically, this study assessed whether hearing status contributes to impaired neurocognitive function and, if hearing aid usage may serve as a protective factor for cognitive symptoms and neuropsychological performance.

As humans age, circadian/diurnal preferences typically shift to earlier times of day. Recent research identifies the fatigue associated with listening effort as a fundamental concern among listeners with hearing impairment. The impact of time-of-day on perceiving speech under challenging listening conditions is, however, not well established for older listeners with hearing loss. A study was therefore initiated to measure time-of-day effects. Experienced cochlear-implant users with postlingual onsets of hearing loss completed several effortful perceptual tasks either in the morning, mid-afternoon, or at both timepoints. Sentence pairs were presented in quiet listening conditions and listeners were cued either before or after presentation, to repeat the first, second, or both sentences. Speech-in-noise (+10 SNR) scores for single sentences were also recorded. Following this, information was collected regarding time-of-day preferences, sleep habits, and fatigue, along with standardized measures of visual attention, processing speed, and working memory. Preliminary data do not currently bear out participants' intuitions regarding performance in the morning versus afternoon. Although memory load and target uncertainty reduce scores, performance across timepoints appears similar, irrespective of time-of-day-preference. Additional data for speech-in-noise will be reported.