Fundamental frequency (F0) has been widely studied and used in the context of speaker discrimination and forensic voice comparison casework, but most previous studies focused on long-term F0 statistics. Lexical tone, the linguistically structured and dynamic aspects of F0, has received much less research attention. A main methodological issue lies on how tonal F0 should be parameterized for the best speaker discrimination performance. This paper compares the speaker discriminatory performance of three approaches with lexical tone modeling: discrete cosine transform (DCT), polynomial curve fitting, and quantitative target approximation (qTA). Results show that using parameters based on DCT and polynomials led to similarly promising performance, whereas those based on qTA generally yielded relatively poor performance. Implications modeling surface tonal F0 and the underlying articulatory processes for speaker discrimination are discussed.

Developmental Language Disorder (DLD) has been explained as either a deficit deriving from an abstract representational deficit or as emerging from difficulties in acquiring and coordinating multiple interacting cues guiding learning. These competing explanations are often difficult to decide between when tested on European languages. This paper reports an experimental study of relative clause (RC) production in Cantonese-speaking children with and without DLD, which enabled us to test multiple developmental predictions derived from one prominent theory - emergentism. Children with DLD (N = 22; aged 6;6-9;7) were compared with age-matched typically-developing peers (N = 23) and language-matched, typically-developing children (N = 21; aged 4;7-7;6) on a sentence repetition task. Results showed that children\'s production across multiple RC types was influenced by structural frequency, general semantic complexity, and the linear order of constituents, with the DLD group performing worse than their age-matched and language-matched peers. The results are consistent with the emergentist explanation of DLD.

UNASSIGNED: This study examined the early efficacy of a new theory-driven principle of grammar intervention, graduated input type variation (GITV).
UNASSIGNED: Three Cantonese-speaking children, aged between 4;01 and 5;10, with oral language difficulties participated in this single baseline within-participant single case experimental study. The children received a total of 300 teaching episodes of the target serial verb construction via focused stimulation and recast over 10 30- to 45-minute sessions. The 30 exemplars of the target included low type variation of the verbs in each of the first five sessions, followed by high type variation in the remaining sessions.
UNASSIGNED: Visual analysis revealed that all children improved their performance in the target construction but not the control vocabulary in the probes, suggesting a treatment effect. Maintenance of treatment effects was also observed in all children. Positive results in across-behaviour generalisation to the untrained construction were observed in all children. Generalisation to other less structured linguistic contexts and to the narrative retell discourse context was minimal and observed in one child only.
UNASSIGNED: Preliminary evidence suggested early efficacy of GITV as a principle for grammar intervention. Modifications in the research methodology are recommended for future studies involving children with developmental language disorder.

Procedural circuit Deficit Hypothesis (PDH) of Developmental Language Disorder (DLD) predicts problems with learning and retention of grammar. Twenty 7- to 9-year-old Cantonese-speaking children with DLD and their typically developing (TD) age peers participated in a syntactic priming task that was given in two sessions one week apart. Production of Indirect Object Relative Clause (IORC) was tested using a probe test before and after the priming task, and one week later. The study involved two cycles of learning and retention, and two levels of prior knowledge. Bayesian linear mixed effects modelling was used for data analysis. Children with DLD learned, and possibly retained, IORC less well than TD children after age, working memory and general grammatical knowledge were controlled for. No interaction effects were significant, meaning that cycle and prior knowledge affected both groups similarly in learning and retention. Results were discussed in relation to PDH and the Complementary Learning Systems Theory.

OBJECTIVE: Atypicalities in the prosodic aspects of speech are commonly considered in clinical assessments of autism. While there is an increasing number of studies using objective measures to assess prosodic deficits, such studies have primarily focused on the intonational and rhythmic aspects of prosody. Little is known about prosodic deficits that are reflected at the segmental level, despite the strong connection between prosody and segmental realization. This study examines the nature of sibilant-vowel coarticulation among male adult native speakers of Cantonese with autism and those without.
METHODS: Fifteen Cantonese-speaking autistic (ASD) adults (mean age = 25 years) and 23 neuro-typical (NT) adults (mean age = 20 years) participated. Each participant read aloud 42 syllables with a sibilant onset in carrier phrase. Spectral means and variance, skewness and kurtosis were measured, and regressed by vocalic rounding (rounded vs. unrounded), cohort (ASD vs. NT), sibilant duration, and articulation rate.
RESULTS: While neurotypical participants exhibit sibilant-vowel coarticulation that are sensitive to variation in sibilant duration, autistic participants show no sensitivity to segmental temporal changes.
CONCLUSIONS: These findings point to the potential for atypicalities in prosody-segment interaction as an important characteristic of autistic speech.

Most languages of the world use lexical tones to contrast words. Thus, understanding how individuals process tones when learning new words is fundamental for a better understanding of the mechanisms underlying word learning. The current study asked how tonal information is integrated during word learning. We investigated whether variability in tonal information during learning can interfere with the learning of new words and whether this is language and age dependent. Cantonese- and French-learning 30-month-olds (N = 97) and Cantonese- and French-speaking adults (N = 50) were tested with an eye-tracking task on their ability to learn phonetically different pairs of novel words in two learning conditions: a 1-tone condition in which each object was named with a single label and a 3-tone condition in which each object was named with three different labels varying in tone. We predicted learning in all groups in the 1-tone condition. For the 3-tone condition, because tones are part of the phonological system of Cantonese but not of French, we expected the Cantonese groups to either fail (toddlers) or show lower performance than in the 1-tone condition (adults), whereas the French groups might show less sensitivity to this manipulation. The results show that all participants learned in the 1-tone condition and were sensitive to tone variation to some extent. Learning in the 3-tone condition was impeded in both groups of toddlers. We argue that tonal interference in word learning likely comes from the phonological level in the Cantonese groups and from the acoustic level in the French groups.

This study established psycholinguistic norms in Cantonese for a set of 1286 colored pictures sourced from several picture databases, including 750 colored line drawings from MultiPic (Duñabeitia et al., 2018) and 536 photographs selected for McRae et al. (2005) concepts. The pictures underwent rigorous normalization processes. We provided picture characteristics including name and concept agreement, familiarity, visual complexity, and frequency of modal responses. Through correlational analyses, we observed strong interrelationships among these variables. We also compared the current Cantonese norming to other languages and demonstrated similarity and variations among different languages. Additionally, we embraced the multilingual diversity within the current sample, and found that higher Cantonese proficiency but lower non-native language proficiency were associated with better spoken picture naming. Last but not least, we validated the predictive power of normed variables calculated from typed responses to spoken picture naming, and the consistency between typed and spoken responses. The present norming provides a timely and valuable alternative for researchers in the field of psycholinguistics, especially those studying Cantonese production and lexical retrieval. All raw data, analysis scripts, and final norming results are available online as psycholinguistic norms for Cantonese in the following link at the Open Science Framework: https://osf.io/dz9j6/?view_only=a452d8a56c92430b9dedf21ac26b1bc1 .

The Simon Fraser University Speech Error Database (SFUSED) is a multi-purpose database of speech errors based in audio recordings. The motivation for SFUSED Cantonese, a component of this database, is to create a linguistically rich data set for exploring language production processes in Cantonese, an under-studied language. We describe in detail the methods used to collect, analyze, and explore the database, including details of team workflows, time budgets, data quality, and explicit linguistic and processing assumptions. In addition to showing how to use the database, this account supports future research with a template for investigating additional under-studied languages, and it gives fresh perspective on the benefits and drawbacks of collecting speech error data from spontaneous speech. All of the data and supporting materials are available as open access data sets.

BACKGROUND: Sentiment analysis is a significant yet difficult task in natural language processing. The linguistic peculiarities of Cantonese, including its high similarity with Standard Chinese, its grammatical and lexical uniqueness, and its colloquialism and multilingualism, make it different from other languages and pose additional challenges to sentiment analysis. Recent advances in models such as ChatGPT offer potential viable solutions.
OBJECTIVE: This study investigated the efficacy of GPT-3.5 and GPT-4 in Cantonese sentiment analysis in the context of web-based counseling and compared their performance with other mainstream methods, including lexicon-based methods and machine learning approaches.
METHODS: We analyzed transcripts from a web-based, text-based counseling service in Hong Kong, including a total of 131 individual counseling sessions and 6169 messages between counselors and help-seekers. First, a codebook was developed for human annotation. A simple prompt (\"Is the sentiment of this Cantonese text positive, neutral, or negative? Respond with the sentiment label only.\") was then given to GPT-3.5 and GPT-4 to label each message\'s sentiment. GPT-3.5 and GPT-4\'s performance was compared with a lexicon-based method and 3 state-of-the-art models, including linear regression, support vector machines, and long short-term memory neural networks.
RESULTS: Our findings revealed ChatGPT\'s remarkable accuracy in sentiment classification, with GPT-3.5 and GPT-4, respectively, achieving 92.1% (5682/6169) and 95.3% (5880/6169) accuracy in identifying positive, neutral, and negative sentiment, thereby outperforming the traditional lexicon-based method, which had an accuracy of 37.2% (2295/6169), and the 3 machine learning models, which had accuracies ranging from 66% (4072/6169) to 70.9% (4374/6169).
CONCLUSIONS: Among many text analysis techniques, ChatGPT demonstrates superior accuracy and emerges as a promising tool for Cantonese sentiment analysis. This study also highlights ChatGPT\'s applicability in real-world scenarios, such as monitoring the quality of text-based counseling services and detecting message-level sentiments in vivo. The insights derived from this study pave the way for further exploration into the capabilities of ChatGPT in the context of underresourced languages and specialized domains like psychotherapy and natural language processing.

Stuttering is characterised by disruptions in speech fluency that normally emerges between the ages of 2 to 5 when children start to formulate sentences. Current stuttering identification in children is largely based on speech disfluency criteria (>3% stuttering-like disfluencies, SLDs) developed for monolingual English-speaking children. Research in a Western language context shows that application of the criteria for monolingual to bilingual children may result in false positive diagnosis of stuttering. The applicability of these criteria to children speaking languages typologically distinct from English remains unclear. This preliminary study focused on bilingual Cantonese-English-speaking children, aiming to explore the manifestations of the speech disfluencies in Cantonese (a syllable-timed language) and English (a stress-timed language) while accounting for language dominance/proficiency and speaking task. Nineteen typically fluent Cantonese-English bilingual preschoolers were recruited for this study and their speech samples were collected across different speaking tasks (i.e. conversation and narration), and languages (i.e. Cantonese and English). The types and frequency of speech disfluencies were compared across both languages and the speaking tasks. The results showed that between 21-68% of children showed higher than 3% SLDs across different languages and speaking tasks. Linear mixed-effect analysis revealed that the prevalence of SLDs is higher in English (less dominant language) than Cantonese (more dominant language), and the prevalence is also higher in narration than conversation. These findings suggest the need for tailored stuttering identification criteria for bilingual children speaking diverse languages and emphasise the importance of considering language dominance/proficiency and speaking task when assessing stuttering in bilingual populations.