Phonemic Similarity Research Articles

Effective Phoneme Decoding With Hyperbolic Neural Networks for High-Performance Speech BCIs.

Speech brain-computer interfaces (speech BCIs), which convert brain signals into spoken words or sentences, have demonstrated great potential for high-performance BCI communication. Phonemes are the basic pronunciation units. For monosyllabic languages such as Chinese Mandarin, where a word usually contains less than three phonemes, accurate decoding of phonemes plays a vital role. We found that in the neural representation space, phonemes with similar pronunciations are often inseparable, leading to confusion in phoneme classification. We mapped the neural signals of phoneme pronunciation into a hyperbolic space for a more distinct phoneme representation. Critically, we proposed a hyperbolic hierarchical clustering approach to specifically learn a phoneme-level structure to guide the representation. We found such representation facilitated greater distance between similar phonemes, effectively reducing confusion. In the phoneme decoding task, our approach demonstrated an average accuracy of 75.21% for 21 phonemes and outperformed existing methods across different experimental days. Our approach showed high accuracy in phoneme classification. By learning the phoneme-level neural structure, the representations of neural signals were more discriminative and interpretable. Our approach can potentially facilitate high-performance speech BCIs for Chinese and other monosyllabic languages.

An Acoustic Study of the Stop Consonants in Lettered-Words Produced by Chinese Mandarin Speakers

This paper presents key findings from an acoustic study on Chinese Mandarin Speakers’ realization of the stop consonants in lettered-words and the stop consonants in Chinese phrases. The research was conducted through voice recording and analysis of data collected from Chinese Mandarin speakers. Recordings of 2000 overall tokens were collected and analyzed using Praat phonetic software. The mean VOT values obtained from both lettered-words and Chinese phrases were compared and statistically tested using the Independent Samples t-test. The research findings show that first, the stop consonants in lettered-words produced by Chinese Mandarin speakers are all voiceless stop consonants, distinguished as being [+aspirated] and [-aspirated]. Second, the VOT values of [p] [t] [k] in lettered-words are significantly lower than their counterparts in Chinese phrases. Third, the stop consonants in lettered-words have the same stop pattern as those in Chinese phrases. The research presents proof that where there is phonemic similarity (but phonetic dissimilarity) across Chinese Mandarin and English, L1 phonetic properties are strong for Chinese Mandarin speakers to produce lettered-words in L1 environment, which further confirms former scholars’ conviction of mother tongue interference in L2 learning. Similar phonemes in L1 and L2 are realized identically to L1 sounds, and there is significant interference from the speaker’s native language on phonetic properties produced in L2. In addition, the research presents implication for future research that might explore other acoustic features of lettered-words in Chinese Mandarin.

Global versus phonemic similarity: Evidence in support of multi-level representation

There is long-standing debate about the extent to which children cognitively represent words in terms of global properties or phonological segments, yet few studies have investigated how children's sensitivity to phonemic versus global similarity changes over time. The current study uses a mispronunciation-reconstruction task to measure both types of sensitivity within a cross-sectional (N = 90, aged 3;2 to 5;7) and longitudinal sample (N = 23, aged 3;2 to 5;1). The results show that children's sensitivity to phonemes increases over the first two years of school but does not reach adult levels. The findings indicate that global similarity relations remain important throughout development and support the idea of multi-level representation.

Integration of Phonotactic Features for Language Identification on Code-Switched Speech

In this paper, phoneme sequences are used as language information to perform code-switched language identification (LID). With the one-pass recognition system, the spoken sounds are converted into phonetically arranged sequences of sounds. The acoustic models are robust enough to handle multiple languages when emulating multiple hidden Markov models (HMMs). To determine the phoneme similarity among our target languages, we reported two methods of phoneme mapping. Statistical phoneme-based bigram language models (LM) are integrated into speech decoding to eliminate possible phone mismatches. The supervised support vector machine (SVM) is used to learn to recognize the phonetic information of mixed-language speech based on recognized phone sequences. As the back-end decision is taken by an SVM, the likelihood scores of segments with monolingual phone occurrence are used to classify language identity. The speech corpus was tested on Sepedi and English languages that are often mixed. Our system is evaluated by measuring both the ASR performance and the LID performance separately. The systems have obtained a promising ASR accuracy with data-driven phone merging approach modelled using 16 Gaussian mixtures per state. In code-switched speech and monolingual speech segments respectively, the proposed systems achieved an acceptable ASR and LID accuracy.

A computational investigation of inventive spelling and the “Lesen durch Schreiben” method

In primary schools, Lesen durch Schreiben (LdS; “reading through writing”, known internationally as inventive spelling ) is a prevalent didactic method of reading and spelling instruction. In LdS, pupils learn writing through prolonged inventive spelling, meaning that only phonological but not orthographic spelling errors are corrected. Rigorous studies of the effectiveness of LdS are scarce and have delivered inconsistent results, casting doubt on the suitability of LdS for primary school instruction. Empirical investigations of writing acquisition methods are time-consuming, costly, and are plagued by methodological evaluation difficulties, such as separating method effects from other instruction-related variables. In this work, we developed a computational framework (based on recurrent neural networks) for reading and writing acquisition. This framework enables us to extract and systematically investigate some core principles of writing acquisition methods. Focusing on two German corpora, we compared the behavior of learning agents trained using the LdS regime against agents trained using a classical, primer-based regime. Experimental results revealed that our LdS agents performed significantly worse than our primer agents in writing tasks and, to a lesser extent, in reading tasks. Our results show that the stereotypical spelling mistakes of children exposed to LdS can be replicated with neural network models. These mistakes arise naturally during writing acquisition for all learning agents but are either suppressed or reinforced depending on the learning regime. We examined the learned, internal representations of both agents and found deviations in the LdS agent that may have induced the amplified confusion of similar phonemes. While we focused on two German corpora, similar results can be expected for alphabetic languages with similar graphene-phoneme regularities. In sum, LdS does not exhibit benefits over standard instruction in our simulations. However, we urge caution in drawing immediate conclusions for human learners. Instead, our work presents a modest step towards the construction of a computational framework for writing and reading instructional methods that may inspire future research.

Learning to Read Interacts with Children’s Spoken Language Fluency

ABSTRACT Until at least the end of adolescence, children articulate speech differently than adults. While this discrepancy is often attributed to the maturation of the speech motor system, we sought to demonstrate that the development of spoken language fluency is shaped by complex interactions across motor and cognitive domains. In this study, we specifically tested for a relationship between reading proficiency and coarticulatory organization, a fundamental correlate of spoken language fluency, used for both reading aloud and conversational speech. We conducted reading assessments and ultrasound-based kinematic measurements of intersegmental coarticulation in a group of 32 German children. In German, a language which supports rather consistent grapheme-to-phoneme relationships, reading aloud uses similar phoneme to speech motor gesture correspondences as well as coarticulatory mechanisms as conversational speech. Using general additive modeling we found that better readers exhibited lower degrees of intersegmental coarticulation than poorer readers. This study therefore provides evidence that reading proficiency interacts with coarticulatory patterns in beginning readers. It suggests that in addition to maturational factors, interactions between speech motor ability and other co-developing skills must be considered to fully account for spoken language fluency.

Dissimilar Phonemes Create a Contextual Interference Effect During a Nonword Repetition Task.

Purpose: The contextual interference effect is a motor learning phenomenon where conditions that decrease overall learning during practice enhance overall learning with new tasks. In the limb literature, this effect is observed when different practice conditions are used (e.g., blocked vs. random practice schedules). In speech motor learning, contextual interference effects are mixed. The differences observed during speech motor learning may be due to the stimuli used. We hypothesized that dissimilar phonemes might create interference during speech motor learning, such that training accuracy would decrease. However, generalization accuracy would increase compared to practice with nonwords containing similar phonemes. Method: Thirty young adults with typical speech and hearing participated in a motor learning study using a cross-over design. Participants engaged in nonword repetition training followed by an immediate retention and transfer task with two sets of nonwords: nonwords with similar phonemes and nonwords with dissimilar phonemes. Percent consonants correct were calculated to examine the effects of the two different types of nonwords based on the stage of skill acquisition. Results: A contextual interference effect was observed in this study using nonwords that varied in phonemic similarity. Nonwords with similar phonemes were produced with greater accuracy during the training stage of skill acquisition, and nonwords with dissimilar phonemes were produced with greater accuracy during the transfer stage. Conclusion: The proposed hypothesis for this study was met – practicing nonwords with dissimilar phonemes resulted in greater accuracy in the transfer phase of this experiment. Results indicate that phonemic dissimilarity produced contextual interference and influenced speech motor learning. These results indicate that the linguistic properties of stimuli must be factored into speech motor learning. Future research should explore if other linguistic variables interact with variables of motor learning to enhance speech practice and generalization outcomes.

АВСТРІАЦИЗМИ У МОВІ СУЧАСНОЇ АВСТРІЙСЬКОЇ ПРЕСИ


 
 
 
 The proposed study is aimed at confirming the hypothesis that there exist two opposing trends in media discourse. On the one hand, there is a tendency to globalization; on the other hand, linguocultural communities are quite determined to preserve their culture and identity. To prove this hypothesis we analyzed a corpus of newspaper texts published over several years. Namely, we studied 1483 Austriacisms recorded by the dictionary Variantenwörterbuch des Deutschen. The objective was to discover the frequency in the use of certain Austriacisms and their Teutonic equivalents in Austrian newspaper Die Presse to identify convergent or divergent processes in the development trends of the German language in Austrian media discourse. The research showed that only 453 lexical units dominated in newspaper articles; it made 30% of 1483 codified Austriacisms. We found that 71 lexemes showed tendency to the parallel use in forms of Austriacisms and Teutonisms which makes less than 5 % of the total number of the lexical units. Such terms have predominantly similar pronunciation hence we can draw a conclusion that such phonemic similarity facilitates equal use of these Austriacisms and Teutonisms in newspapers and stipulates their convergence. These lexical units are not marked by any particular ethnocultural specificity. Furthermore, the analysis shows that the word stock denoting Austrian culture, traditions, and realia of daily life.
 
 
 

Perception of Unfamiliar English Phonemes by Native Mandarin Speakers

In second language acquisition, speech sounds, or phonemes, not present in a learner’s native language often pose an extra challenge for speech production. When hearing one of these unfamiliar phonemes, the learner either maps it to a similar native phoneme, perceives it as a completely foreign sound, or does not perceive it as speech at all. In the first case, the learner is unable to perceive a difference between the unfamiliar phoneme and the native phoneme to which it is mapped. This mapping difficulty potentially creates problems for the learner during word recognition. The present research investigated the extent to which English phonemes absent from the Mandarin phonological inventory impact processing of native Mandarin speakers in an auditory lexical decision task. Results of this research will expand the understanding of second language perception, especially within the context of auditory lexical decision tasks. A list of ten phonemes—/ɪ/, /æ/, /ʊ/, /ɛ/, /v/, /z/, /ʒ/, /ɵ/, /ð/, /ʤ/—present in the English phonological inventory but absent from that of Mandarin were identified as unfamiliar to native Mandarin speakers. Data from the Massive Auditory Lexical Decision (MALD) database, in which participants decided whether recorded utterances were English words or made-up words, were utilized. The effects of the proportion of unfamiliar phonemes, proportion of unfamiliar vowels, and proportion of unfamiliar consonants on reaction time, representative of processing difficulty, were then calculated using statistical techniques. It was found that the proportion of all unfamiliar phonemes in an utterance had no significant effect on the reaction time of the native Mandarin speakers. However, when the list of unfamiliar phonemes was divided into vowels and consonants, a greater proportion of unfamiliar vowels was noticed to increase reaction time, while a greater proportion of unfamiliar consonants was found to decrease reaction time. Further research in this area is required to determine a concrete explanation for these results. Interestingly, when the same analysis was performed on the data of native English speakers, similar results were observed. This may reflect a common language processing mechanism in second language learners and native speakers.

An Analysis of the Short Utterance Problem for Speaker Characterization

Speaker characterization has always been conditioned by the length of the evaluated utterances. Despite performing well with large amounts of audio, significant degradations in performance are obtained when short utterances are considered. In this work we present an analysis of the short utterance problem providing an alternative point of view. From our perspective the performance in the evaluation of short utterances is highly influenced by the phonetic similarity between enrollment and test utterances. Both enrollment and test should contain similar phonemes to properly discriminate, being degraded otherwise. In this study we also interpret short utterances as incomplete long utterances where some acoustic units are either unbalanced or just missing. These missing units are responsible for the speaker representations to be unreliable. These unreliable representations are biased with respect to the reference counterparts, obtained from long utterances. These undesired shifts increase the intra-speaker variability, causing a significant loss of performance. According to our experiments, short utterances (3–60 s) can perform as accurate as if long utterances were involved by just reassuring the phonetic distributions. This analysis is determined by the current embedding extraction approach, based on the accumulation of local short-time information. Thus it is applicable to most of the state-of-the-art embeddings, including traditional i-vectors and Deep Neural Network (DNN) xvectors.

Learning Difficulties in Listening Comprehension

This paper aims at describing the student’s listening skill and their learning difficulties in listening comprehension. The method used was a descriptive method. The instruments were pre-TOEFL test and questionnaire. The population was the second year students of State University of Makassar. The results of this research are: the students have poor score in their listening comprehension and the students' learning difficulties in listening comprehension are: problem in interpretation, the lack of vocabulary, difficulty to concentrate, problem with lecturer's accent, confusion in similar phonemes, the speed of the speech problem in accent, problem in reduction and there are no repetitions and in learning media, the noise from students themselves, unclear sound, air conditioner problems and seat position. Based on the result, the writer concludes that the students' ability in listening comprehension is still poor, the teacher should motivate them to develop the habit of writing daily listening so that they can improve their skill.

Phonemic Comparison of English and Punjabi

English and Punjabi are languages which do not belong to the same families of languages. English is one of the West Germanic languages whereas; Punjabi is a part of the Indo-Aryan family. Punjabi is spoken by various nations on the globe, especially Pakistan and its province Punjab as well as in Indian Panjab. Both English and Punjabi manifest themselves through various dialects on the basis of diversified geographical areas. English is used as the first language by 379,007,140 speakers and further 753,359,540 speakers use it as a second language in more than 104 nations. So, the total speakers of English around the globe are 1,132,366,680 (Ethnologue, 2019). The importance of Punjabi cannot be denied being the 10th most widely used language on the globe (Ghai & Singh, 2013). According to Ethnologue (2019), the total number of Punjabi speakers is 125,326,840. In Pakistan, it is the language of the majority of the people residing in the most populous province of Pakistan, Punjab. It is among twenty-two languages that have obtained official status. Unfortunately, no considerable work has been done on its phonology. This study is an attempt to describe the phonemic differences between English and Punjabi by using the theoretical framework of the Levenshtein algorithm. The index of differences and similarities is determined through the inventories of both languages. The inventories are used as data in this research paper. The Levenshtein algorithm (Levenshtein, 1965) is used to analyse the inventories to calculate the ratio of differences and similarities. The outcome of the current research shows that both English and Punjabi have a phonemic similarity level of 56.25% whereas the index of difference is 43.75%.

Asykāliyatu Nuṭqil Ḥurūf Al ‘Arabiyyah Laday Ṭalabati Madrasati Tarbiyatil Qur’an: Tahlīl Taqābulī bayna Al-Lugah Al ‘Arabiyyah wa Al-Lugah Al Indūnīsiyyah fī Mustawā Al Fūnīm

Abstract Indonesian learners of Arabic frequently encounter problems when learning Arabic language and its phonetics. The difficulties arise for there are differences between Arabic and Indonesian sound system. This article aimed to explore the learners’ difficulties in learning the Arabic sounds. Some learners of TPQ Al Muhajirin Sidayu Gresik were participated in this study. To conduct this research, the researcher used descriptive and contrastive approach and to collect the data by using a reading test of Arabic phonemes and interview were employed. This study revealed that the learners faced difficulties in pronouncing the Arabic consonants in which were influenced by their Indonesian language. To overcome those problems, the learners used the following strategies, i.e. pronouncing the same phonemes as in ح[ħ] and خ[x], pronouncing similar phonemes as in [s] and ش [ʃ], pronouncing phonemes by changing the phonemes in order to make it easier to utter as in [q] to [s]. This study offers a recommendation for overcoming learners’ problems in pronouncing the Arabic sounds. Keywords : Arabic phonology, Arabic teaching, contrastive analysis, pronunciation. أشكالية النطق الحروف العربية لدى طلبة مدرسة تربية القرآن: التحليل التقابلي بين اللغة العربية واللغة الإندونيسية في مستوى الفونيم كمال يوسف، سري إنداه أيو دمايانتي الجامعة الإسلامية الحكومية سونان أمبيل kamalinev@gmail.com , sriindahatudamayanti@gmail.com ملخص إن المتعلمون الإندونيسيون يوجهوا المشاكل كثــيرا عند تعلم اللغة العربية و بالخصوص فى مجال علم الصوتيات ( phonetics )، و تظهر الصعوبات لوجود الاختلافات بين نظام الصوت العربي و الإندونيسي. فيهدف هذه المقالة إلى استكشاف صعوبات و مشكلات المتعلمين فى تعلم الأصوات العربية، و اشتركت الطلبة من المدرسة لتعلم القرآن المهجرين سيدايو، جريسيك، إندونيسيا فى هذا البحث العلمي. لإجراء هذا البحث، استخدم الباحث المنهج الوصفىي التقابلي، و عقد الاختبار لقراءة الصوتيات العربية و المقابلة لأجل جمع البيانات. كشفت هذه الدراسة على أن المتعلمين الإندونيسين واجهوا الصعوبات فى نطق الحروف العربية المتأثرة باللغة الإندونيسية. استخدم المتعلمون هذه الاستراتيجيات للتغلب على الصعوبات، وهي نطق نفس الصوتيات كالصوت ح و خ، و نطق الأصوات المماثلة كالصوت س و ش، ونطق الأصوات عن طريق تغيــير الصوتيات فتكون أسهل فى النطق كالصوت ث تغيــيره بالصوت س. هذا البحث يقدم الاقتراحات والتوصيات فى التغلب على صعوبات لدي المتعلمين الإندونيسين فى تعلم الأصوات العربية ونطقها الصحيحة. الكلمات المفتاحية: علم الأصوات العربية، تعليم العربية، المنهج التقابلي، النطق

KEKERABATAN BAHASA JAWA DAN BAHASA GAYO: KAJIAN LINGUISTIK HISTORIS KOMPARATIF

This study aims to reveal the kinship relationship of Javanese and Gayo languages ​​seen from some vocabulary similarities, as well as estimate the time separation with the run of error. The method used is descriptive qualitative and quantitative with the techniques refer to note. This research data is a vocabulary in BJ and BG obtained from various written sources and informant interview. From 200 words Swadesh to BJ and BG there are 13% ie 25 words pairs of relatives. These data provide evidence that BJ and BG are not close relatives because the core vocabulary equation is less than 36% or part of sub-grouping. But the language of both comes from the same pilum or stock. Both languages ​​have identical word pairs that are all similar phonemes of 16 relatives couples; 12 relatives couples have a phonemic correspondence; 9 relatives couples who are phonetically similar; and 13 relatives couples who have a different phoneme. The time of split of both languages ​​was 3,393 thousands of years ago. It is thought that this second language was a single language about 3.4 thousands years ago and is thought to have begun to separate from a proto language about the 5th century BC. It is estimated that the age of both languages ​​is a single language on 4,712 + 335 years ago and at 5,047+335 years ago.

Phonological Feature Repetition Suppression in the Left Inferior Frontal Gyrus.

Models of speech production posit a role for the motor system, predominantly the posterior inferior frontal gyrus, in encoding complex phonological representations for speech production, at the phonemic, syllable, and word levels [Roelofs, A. A dorsal-pathway account of aphasic language production: The WEAVER++/ARC model. Cortex, 59(Suppl. C), 33-48, 2014; Hickok, G. Computational neuroanatomy of speech production. Nature Reviews Neuroscience, 13, 135-145, 2012; Guenther, F. H. Cortical interactions underlying the production of speech sounds. Journal of Communication Disorders, 39, 350-365, 2006]. However, phonological theory posits subphonemic units of representation, namely phonological features [Chomsky, N., & Halle, M. The sound pattern of English, 1968; Jakobson, R., Fant, G., & Halle, M. Preliminaries to speech analysis. The distinctive features and their correlates. Cambridge, MA: MIT Press, 1951], that specify independent articulatory parameters of speech sounds, such as place and manner of articulation. Therefore, motor brain systems may also incorporate phonological features into speech production planning units. Here, we add support for such a role with an fMRI experiment of word sequence production using a phonemic similarity manipulation. We adapted and modified the experimental paradigm of Oppenheim and Dell [Oppenheim, G. M., & Dell, G. S. Inner speech slips exhibit lexical bias, but not the phonemic similarity effect. Cognition, 106, 528-537, 2008; Oppenheim, G. M., & Dell, G. S. Motor movement matters: The flexible abstractness of inner speech. Memory & Cognition, 38, 1147-1160, 2010]. Participants silently articulated words cued by sequential visual presentation that varied in degree of phonological feature overlap in consonant onset position: high overlap (two shared phonological features; e.g., /r/ and /l/) or low overlap (one shared phonological feature, e.g., /r/ and /b/). We found a significant repetition suppression effect in the left posterior inferior frontal gyrus, with increased activation for phonologically dissimilar words compared with similar words. These results suggest that phonemes, particularly phonological features, are part of the planning units of the motor speech system.

The influence of the native language on phonological preparation in spoken word production in a second language

Abstract Three experiments investigated the phonological preparation unit in planning English spoken words, comparing English monolinguals, native Chinese and Japanese-speakers who spoke English as their second language. All three groups named pictures in English, and the names could either share the same initial phoneme, mora, or syllable, or had no systematic commonality. A phoneme preparation effect was shown among English monolinguals but not among the two bilingual groups, suggesting that the phoneme is the phonological preparation unit for English monolinguals, but not for the two bilingual groups. All three groups showed mora and syllable preparation effects, but further analysis and a follow-up experiment suggested that Chinese-English bilinguals may treat morae as open syllables. English monolinguals showed similar phoneme and mora preparation effect sizes, possibly as a result of flexibility. Together, the selection of phonological preparation could be flexible, influenced by both the nature of the target language and speakers’ language experiences.

Investigating Oral Reading Miscues Produced by Students with Down Syndrome: A Descriptive Study

Objective: Individuals with Down syndrome (DS) exhibit difficulties with oral language and reading development. Research is needed to understand the language and reading connections of students with DS to develop effective interventions that ensure academic success in this population. The purpose of this present study was to investigate the oral language and reading processes of six students (mean age: 16; 3) with DS by analysing the oral miscues produced while reading a storybook. Method: Using a miscue analysis approach, students with DS independently read a storybook out loud and then were asked to retell the story. Oral reading miscues were analysed within the context of the sentence and the entire text to determine semantic and syntactic acceptability, meaning change, correction and graphic and phonemic similarity. In addition, thematic analysis was incorporated to provide an in-depth analysis of the miscue data. Results: Based on the miscue analysis data, students exhibited higher meaning construction scores as compared to grammatical relations scores (63.67% vs. 46%) and higher graphic similarity scores as compared to phonemic similarity scores (66.34% vs. 60.67%). Using a thematic analysis procedure, students with DS rarely selfcorrected miscues, exhibited relative strength with meaning but difficulty with syntax and relied on visual support while reading. Conclusion: The present study adds to the existing body of literature about the language and reading profile of students with DS by including analysis of their oral reading miscues and provides implications on intervention techniques that may be beneficial with this population.

Does linguistic comprehension support the decoding skills of struggling readers?

This study investigated the contribution of linguistic comprehension to the decoding skills of struggling readers. Participants were 36 children aged between eight and 12 years, all below average in decoding but differing in linguistic comprehension. The children read passages from the Neale Analysis of Reading Ability and their first 25 miscues were categorised into syntactic, semantic, phonemic, and graphophonic similarity. Children were first grouped in terms of higher and lower linguistic comprehension levels. Analysis of miscues showed no differences between the groups in miscue similarity. They were then grouped according to pseudoword reading skill. There were significant differences between the groups for all miscue types except semantic similarity. An analysis of miscues using multiple regression showed that, after taking account of age, pseudoword reading was the best predictor of quality of miscues. In addition, linguistic comprehension contributed to syntactic similarity of miscues over and above decoding.

A Novel Investigation of Generalized Motor Program Theory: Syllable Stress as a Motor-Class Variable.

This experiment evaluated syllable-stress position as a motor class directed by a syllable-sized generalized motor program. Reaction times were predicted to be slower for stimuli with untrained stress patterns outside the trained motor class. Furthermore, reaction times were predicted to be stable for untrained stimuli within the same motor class regardless of phonetic similarity. Twenty-three young adults with typical speech and hearing participated in a motor-learning study comprised of motor-class training and a judgment task. Reaction times derived from the judgment task were evaluated using a within-subject repeated-measures design to evaluate the effects of syllable-stress position on stimulus type. Reaction times were not significantly different across proposed syllable-stress motor-class boundaries. However, reaction times for the stimuli within the same motor class were significantly different. To be specific, slower reaction times were associated with untrained stimuli that were phonetically similar to the trained stimuli. The proposed hypotheses for syllable stress as a motor class were not met. Results indicate that multiple stimulus features, including syllable stress and phoneme similarity, may be encoded into memory during motor learning. Future research should evaluate how phonetic similarity of stimuli may influence motor-learning outcomes. https://doi.org/10.23641/asha.5116837.

Exploiting acoustic similarities between Tamil and Indian English in the development of an HMM‐based bilingual synthesiser

In this study, an efficient hidden Markov model (HMM)-based bilingual speech synthesiser for the Indian language Tamil and Indian English is developed. Initially, phone mapping approach is tried to synthesise English text using Tamil corpus alone by mapping English phonemes to the perceptually similar phonemes in Tamil, and is found that the approach is language-dependent and requires a large dictionary for Indian pronunciation. Therefore, given the speech data for both languages, the straight-forward approach to develop a bilingual synthesiser is to build separate synthesiser for each language and combine them or by merging the perceptually similar phonemes. These approaches introduce language-switching/influence in the synthesised speech. To minimise switching and influence, HMM-based bilingual synthesisers are developed by merging acoustically similar phonemes, derived based on model parameters and likelihood Gaussian using various distance metrics. The performance of these synthesisers are evaluated based on mean opinion score (MOS), language-switching and language-influence. Results reveal that the set of phonemes derived using product-of-likelihood Gaussians in the likelihood space is the optimum set of phonemes that can be merged and the system developed by merging these phonemes outperforms the rest with an MOS of 3.66. Furthermore, only 8% and 23% of the sentences have language-switching/influence, respectively.