Reproducible Speech Research With the Artificial Intelligence-Ready PERCEPT Corpora

BACKGROUND

Publicly available speech corpora facilitate reproducible research by providing open-access data for participants who have consented/assented to data sharing among different research teams. Such corpora can also support clinical education, including perceptual training and training in the use of speech analysis tools.

PURPOSE

In this research note, we introduce the PERCEPT (Perceptual Error Rating for the Clinical Evaluation of Phonetic Targets) corpora, PERCEPT-R (Rhotics) and PERCEPT-GFTA (Goldman-Fristoe Test of Articulation), which together contain over 36 hr of speech audio (> 125,000 syllable, word, and phrase utterances) from children, adolescents, and young adults aged 6-24 years with speech sound disorder (primarily residual speech sound disorders impacting /ɹ/) and age-matched peers. We highlight PhonBank as the repository for the corpora and demonstrate use of the associated speech analysis software, Phon, to query PERCEPT-R. A worked example of research with PERCEPT-R, suitable for clinical education and research training, is included as an appendix. Support for end users and information/descriptive statistics for future releases of the PERCEPT corpora can be found in a dedicated Slack channel. Finally, we discuss the potential for PERCEPT corpora to support the training of artificial intelligence clinical speech technology appropriate for use with children with speech sound disorders, the development of which has historically been constrained by the limited representation of either children or individuals with speech impairments in publicly available training corpora.

CONCLUSIONS

We demonstrate the use of PERCEPT corpora, PhonBank, and Phon for clinical training and research questions appropriate to child citation speech. Increased use of these tools has the potential to enhance reproducibility in the study of speech development and disorders.

Exploring word production in three-year-old monolingual French-speaking children

There is a general need for more knowledge on the development of French phonology, and little information is currently available for typically developing French-speaking three-year-old children. This study took place in Belgium and explores the accuracy of speech production of 34 typically developing French-speaking children using a picture naming task. Measures of speech accuracy revealed lower performance than previously seen in the literature. We investigated speech accuracy across different phonological contexts in light of characteristics of target words that are known to have an influence on speech production, namely the condition of production (spontaneous vs. imitated), the length of the word (in number of syllables), syllable complexity (singleton vs. cluster) and positional complexity (onset vs. coda). Results indicate that the accuracy of words produced spontaneously did not differ from imitated words. The presence of consonant clusters in the target word was associated with lower performance on measures of Percentage of Consonants Correct and Whole Word Proximity for both 1- and 4-syllable words. Singleton codas were produced less accurately than onsets in 1-syllable words. Word-internal singleton codas were produced less accurately than final codas. In our sample, 1-syllable words showed surprisingly low levels of performance which we can explain by an over-representation of phonologically complex properties in the target words used in the present study. These results highlight the importance of assessing various aspects of phonological complexity in French speech tasks in order to detect developmental errors in typically developing children and, ultimately, help identify children with speech sound disorders.

Automated phonological analysis and treatment target selection using AutoPATT

Automated analyses of speech samples can offer improved accuracy and timesaving advantages that streamline clinical assessment for children with a suspected speech sound disorder. In this paper, we introduce AutoPATT, an automated tool for clinical analysis of speech samples. This free, open-source tool was developed as a plug-in for Phon and follows the procedures of the Phonological Analysis and Treatment Target Selection protocol, including extraction of a phonetic inventory, phonemic inventory with corresponding minimal pairs, and initial consonant cluster inventory. AutoPATT also provides suggestions for complex treatment targets using evidence-based guidelines. Automated analyses and target suggestions were compared to manual analyses of 25 speech samples from children with phonological disorder. Results indicate that AutoPATT inventory analyses are more accurate than manual analyses. However, treatment targets generated by AutoPATT should be viewed as suggestions and not used to substitute necessary clinical judgement in the target selection process.

Language and Learner Specific Influences on the Emergence of Consonantal Place and Manner Features

This article focuses on the emergence of consonantal place and manner feature categories in the speech of first language learners. Starting with an overview of current representational approaches to phonology, we take the position that only models that allow for the emergence of phonological categories at all levels of phonological representation (from sub-segmental properties of speech sounds all the way to word forms represented within the child's lexicon) can account for the data. We begin with a cross-linguistic survey of the acquisition of rhotic consonants. We show that the types of substitutions affecting different rhotics cross-linguistically can be predicted from two main observations: the phonetic characteristics of these rhotics and the larger system of categories displayed by each language. We then turn to a peculiar pattern of labial substitution for coronal continuants in the speech of a German learner. Building on previous literature on the topic, we attribute the emergence of this pattern to distributional properties of the child's developing lexicon. Together, these observations suggest that our understanding of phonological emergence must involve a consideration of multiple, potentially interacting levels of phonetic and phonological representation.

The development of preverbs in Northern East Cree: A longitudinal case study

This study is based on naturalistic speech samples produced by one child learning Cree as her first language (2;01-4;03) and presents the first investigation into the development of preverbs in the language. Preverbs are an optional class of morpheme which precede the lexical verb stem, dividing into grammatical, lexical and directional (deictic) subclasses. Of nine preverb types in the child's inventory, 47/48 tokens are grammatical. We argue that these appear early because they exhibit phonological transparency (do not alternate in form) and positional predictability (are restricted to preverb position). Lexical and directional preverbs, however, alternate in form and may appear in either preverb position or within the lexical stem. Furthermore, the child first began to use preverbs with the grammatically simpler independent (default) inflection (2;04), and 7 months later with the more syntactically restrictive conjunct inflection (2;11). She also used each preverb with one inflection type only, even where a choice of inflections was available, and she never produced more than one preverb per verb complex although multiple preverbs are common. In sum, grammatical complexity appears to be a driving factor in determining the child's development pattern.

Markedness and implicational relationships in phonological development: A cross-linguistic investigation

PURPOSE

The complexity approach to speech disorders, based on the theoretical notion of phonological markedness, has been gaining interest over the last decade. In a nutshell, this approach suggests that the acquisition of phonologically marked units (e.g. complex onsets) implies the acquisition of less marked ones (e.g. singleton onsets). However, because the notion of markedness is, itself, subject to controversies, we need to constrain what types of implications can be generalised among language learners, within and across languages.

METHOD

We report on longitudinal data from one phonologically-disordered and five typically-developing children documented across four different languages (English, French, German, Portuguese), using data from the PhonBank database (https://phonbank.talkbank.org). Using the Phon software program (https://www.phon.ca), we systematically analysed each longitudinal study for consonants in singleton onsets and codas as well as in onset clusters.

RESULT

The implicational relationships supported by our study involve units of similar types (e.g. relations between different segmental categories), while relationships that involve different types of units or processes cannot be generalised across learners.

CONCLUSION

A better understanding of implicational relationships makes the complexity approach more predictive of developmental patterns of phonology and related phonological disorders.

There is no phonology without abstract categories: A commentary on

Ambridge's proposal cannot account for the most basic observations about phonological patterns in human languages. Outside of the earliest stages of phonological production by toddlers, the phonological systems of speakers/learners exhibit internal behaviours that point to the representation and processing of inter-related units ranging in size from sub-segmental (featural) and segmental to larger prosodic units ranging from the syllable all the way to the phonological utterance. Only theories which embrace these units can make proper predictions about phonological behaviours.

Questioning the role of lexical contrastiveness in phonological development: Converging evidence from perception and production studies

In this article, we address relations between lexical and phonological development, with an emphasis on the notion of phonological contrast. We begin with an overview of the literature on word learning and on infant speech perception. Among other results, we report on studies showing that toddlers' perceptual abilities do not correlate with the development of phonological contrasts within their lexicons. We then engage in a systematic comparison between the lexical development of two child learners of English and their acquisition of consonants in syllable onsets. We establish a developmental timeline for each child's onset consonant system, which we compare to the types of phonological contrasts that are present in their expressive vocabularies at each relevant milestone. Like the earlier studies, ours also fails to return tangible parallels between the two areas of development. The data instead suggest that patterns of phonological development are best described in terms of the segmental categories they involve, in relative independence from measures of contrastiveness within the learners' lexicons.

Fostering human rights through TalkBank

In accord with articles 19 and 27 of the Universal Declaration of Human Rights, people with speech and language disorders have the right to receive maximal benefit from academic research on speech and language acquisition and disorders. To evaluate the diverse nature of speech and language disorders, this research must have access to large datasets, as well as to refined tools for the systematic analysis of these datasets. The TalkBank system addresses this need by providing researchers with thousands of hours of open-access database archives of digital audio, video and transcript files documenting typical and disordered language use in dozens of languages and cultures. In this paper, we review the TalkBank system, with an emphasis on the AphasiaBank, PhonBank and FluencyBank databases. We describe how specialised assessment tools can be used to study issues in speech and language acquisition and disorders recorded within these databases. We then provide illustrations of how assessments support the needs of researchers, clinicians, developers, and educators, whose combined work contributes solutions for people with speech, language and language learning disorders worldwide.

Analyzing Clinical Phonological Data Using Phon

In this article, we describe how Phon, an open-source software program for the transcription and analysis of phonological data, can be applied to facilitate clinical phonological analyses. We begin with a summary of the types of analyses that are frequently used in the assessment and management of speech sound disorders. We then discuss challenges inherent to the transcription and analysis of clinical phonological data. For each challenge, we discuss solutions currently available within Phon and offer an outlook on future methodological and technical developments in the area of clinical phonology. This article includes a step-by-step introduction to Phon suitable for readers who lack previous experience with the software. We conclude with a discussion of data sharing and its vital role in advancing research and intervention practices in the area of speech development and disorders.

[The phonological profile of children with speech sound disorders]

Concern over reduced intelligibility in a child is a common reason for consulting a speech-language pathologist. Nevertheless, there is little published research on speech sound disorders among French-speaking children. As a result, speech-language pathologists must rely on their clinical expertise or on research involving children who speak other languages, often the English language. The aim of this article is to contribute to the literature on French-speaking children with reduced intelligibility. In this study we describe the phonological profiles of seven children with a speech sound disorder. We use the Phon phonological assessment tool, which provides a rapid and precise analysis of phonological productions. Three distinct phonological profiles are identified on the basis of the severity of the disorder.

Using PhonBank and Phon in studies of phonological development and disorders

The goal of this paper is to present an overview of new tools that can be used to further our understanding of phonological development and disorders. We begin with a summary of the field of child phonology with a focus on databases and methods of analysis and then move to a description of PhonBank, a shared database for the study of phonology, and Phon, a specialised software system capable of performing various types of phonological analyses based on both phonetic transcriptions and acoustic analyses of speech productions. We provide a detailed example of using PhonBank and Phon to examine the use of velar fronting using longitudinal data from one child with typical development and three children with phonological disorder. We conclude with an emphasis on data sharing and its central relevance to further advances in our field.

Debunking the Trochaic Bias Myth: Evidence from Phonological Development

n/a

Relations between segmental and prosodic structure in first language acquisition

In this paper, I discuss a number of relations that take place between melodic content and higher prosodic structure in first language phonological development. I explore acquisition patterns found in data on the acquisition of Québec French. Starting with the observation that prosodic structure and, more specifically, stressed syllables, play a central role in phonological acquisition, I hypothesize that the inter-relations between prosodic and segmental structure posited by formal models of phonological organisation should be witnessed within and across developmental stages. I support this hypothesis through two findings from the French data. First, complex onsets emerge in stressed syllables before unstressed ones. Second, different types of consonants (placeless versus place-specified) emerge in word-final position at different stages. From these observations, I argue that the phenomena observed in these data are best captured in an analysis based on constituent structure and relationships between feature specification and prosodic constituency, which are governed by universal markedness.

Uninflected structure in familial language impairment: evidence from French

We present the results of 20 French subjects with familial language impairment (FLI) on a linguistic battery task, with an emphasis on verb production. The results show strong qualitative differences between the verb production of FLI subjects and that of controls. Language-specific factors do not seem to determine the production of verbs in French FLI individuals. Rather, verb frequency and the inflectional status (uninflected vs. inflected) of the form seem to be determining factors in correct/incorrect production of a verb in a sentence context. The phonetic structure of French inflection provides additional arguments against the hypothesis of a processing deficit in FLI subjects. French tense morphemes are stressed and salient, and should therefore be produced without problems, according to the processing hypothesis. We found evidence contrary to this postulate. We therefore submit that the morphological deficit hypothesis is supported by the French data.

Superoxide-mediated decomposition of biological S-nitrosothiols

Incubation of S-nitrosocysteine or S-nitrosoglutathione (5-100 M) in the presence of a generator of superoxide (xanthine/xanthine oxidase) resulted in a time-dependent decomposition of S-nitrosothiols and accumulation of nitrite/nitrate in reaction mixtures. Quantitatively, the amounts of nitrite/nitrate represented >90% of nitrosonium equivalent of S-nitrosothiols degraded during the incubation. The reaction rates were unaffected by the presence catalase (1 unit/ml). Kinetic analysis showed that the degradation of S-nitrosothiols in the presence of superoxide proceeded at second order rate constants of 76,900 M-1 s-1 (S-nitrosocysteine) and 12,800 M-1 s-1 (S-nitrosoglutathione), respectively, with a stoichiometric ratio of 1 mol of S-nitrosothiol per 2 mol of superoxide. The findings provide the evidence for the involvement of superoxide in the metabolism of S-nitrosothiols. Furthermore, substantially slower reaction rates of superoxide with S-nitrosothiols relative to the reaction rate with NO are consistent with the contention that the transient formation of S-nitrosothiols in biological systems may protect NO from its rapid destruction by superoxide, thus enabling these compounds to serve as carriers or buffers of NO.

Credit where credit's due. How to borrow money effectively

Neither a borrower or a lender be, says the old proverb. However, used effectively, credit can help you to make the most of your money - so long as you are careful!

A home of your own? A guide to getting a mortgage

High mortgage rates are regularly in the news at the moment, and many people are wary of deciding to buy property. However, with property prices depressed, there may be some bargains to be found, so the idea may be worth considering.

[Fulminating refractory hypoxemia. Apropos of 3 cases including 1 case of tuberculosis]

The authors report 3 cases of refractory fulminant hypoxemia. The etiology was various, neoplastic lymphangitis, influenza and tuberculosis. The authors emphasize the role of tuberculosis in the onset of certain diffuse interstitial fibroses.