Morpho-Phonetic Effects in Speech Production: Modeling the Acoustic Duration of English Derived Words With Linear Discriminative Learning

Frontiers in Psychology ◽

10.3389/fpsyg.2021.678712 ◽

2021 ◽

Vol 12 ◽

Author(s):

Simon David Stein ◽

Ingo Plag

Keyword(s):

Speech Production ◽

Morphological Structure ◽

Computational Approach ◽

Natural Speech ◽

Discriminative Learning ◽

Promising Alternative ◽

Learning Network ◽

Speech Data ◽

Production Modeling

Recent evidence for the influence of morphological structure on the phonetic output goes unexplained by established models of speech production and by theories of the morphology-phonology interaction. Linear discriminative learning (LDL) is a recent computational approach in which such effects can be expected. We predict the acoustic duration of 4,530 English derivative tokens with the morphological functions DIS, NESS, LESS, ATION, and IZE in natural speech data by using predictors derived from a linear discriminative learning network. We find that the network is accurate in learning speech production and comprehension, and that the measures derived from it are successful in predicting duration. For example, words are lengthened when the semantic support of the word's predicted articulatory path is stronger. Importantly, differences between morphological categories emerge naturally from the network, even when no morphological information is provided. The results imply that morphological effects on duration can be explained without postulating theoretical units like the morpheme, and they provide further evidence that LDL is a promising alternative for modeling speech production.

Download Full-text

Durational Differences of Word-Final /s/ Emerge From the Lexicon: Modelling Morpho-Phonetic Effects in Pseudowords With Linear Discriminative Learning

Frontiers in Psychology ◽

10.3389/fpsyg.2021.680889 ◽

2021 ◽

Vol 12 ◽

Author(s):

Dominic Schmitz ◽

Ingo Plag ◽

Dinah Baer-Henney ◽

Simon David Stein

Keyword(s):

Speech Production ◽

Discriminative Learning ◽

Production Data ◽

Real Word ◽

Theoretical Level ◽

Learning Network ◽

Recent Approach ◽

Production Study ◽

Different Types

Recent research has shown that seemingly identical suffixes such as word-final /s/ in English show systematic differences in their phonetic realisations. Most recently, durational differences between different types of /s/ have been found to also hold for pseudowords: the duration of /s/ is longest in non-morphemic contexts, shorter with suffixes, and shortest in clitics. At the theoretical level such systematic differences are unexpected and unaccounted for in current theories of speech production. Following a recent approach, we implemented a linear discriminative learning network trained on real word data in order to predict the duration of word-final non-morphemic and plural /s/ in pseudowords using production data by a previous production study. It is demonstrated that the duration of word-final /s/ in pseudowords can be predicted by LDL networks trained on real word data. That is, duration of word-final /s/ in pseudowords can be predicted based on their relations to the lexicon.

Download Full-text

Environmental and Linguistic Typology of Whistled Languages

Annual Review of Linguistics ◽

10.1146/annurev-linguistics-011619-030444 ◽

2021 ◽

Vol 7 (1) ◽

pp. 493-510

Author(s):

Julien Meyer

Keyword(s):

Speech Production ◽

Point Of View ◽

Natural Speech ◽

Language Diversity ◽

Human Language ◽

Language Family ◽

Long Distance ◽

Linguistic Typology ◽

Language Structure

Whistled forms of languages are distributed worldwide and survive only in some of the most remote villages on the planet. They are not limited to a given continent, language family, or language structure, but they have been detected only sporadically by researchers and travelers, partly because they can be taken for nonlinguistic phenomena, such as simple signaling. Whistled speech consists of speaking while whistling to communicate at a long distance. The result is a melody that imitates modal speech and that remains intelligible for the interlocutors. This review proposes a typology of this special, little-known, natural speech type and takes socio-environmental and linguistic aspects into consideration. The amazing potential of this phenomenon to provide an alternative point of view into language diversity and speech offers a unique occasion to revisit human language with original insights embracing the adaptive flexibility that characterizes speech production and perception.

Download Full-text

Word-initial [h]-drop variation in Nmbo

Asia-Pacific Language Variation ◽

10.1075/aplv.20002.kas ◽

2020 ◽

Vol 6 (2) ◽

pp. 250-277

Author(s):

Eri Kashima

Keyword(s):

Papua New Guinea ◽

Intergenerational Transmission ◽

New Guinea ◽

Natural Speech ◽

Speech Community ◽

Speech Corpus ◽

Language Ecology ◽

Speech Data ◽

The Village

Abstract This paper presents a natural speech corpus-based study of word-initial [h]-drop from the Nmbo speech community of southern Papua New Guinea. It is a speech community within a traditional egalitarian multilingual language ecology sustained by a practice of virilocal exogamy, and there is strong intergenerational transmission of local vernacular languages. This study investigates the propensity of word-initial [h]-drop in nouns, based on Nmbo speech data of Kerake tribe people. The results from the Nmbo Sociolinguistic Corpus shows clear age-conditioned variation, with younger speakers showing a higher propensity for [h]-drop. Nmbo speakers residing both within and outside their Nmbo villages of origin appear to be partaking in the innovative [h]-drop. The origin of the [h]-drop appears to be from the village with a more multilingual profile, as would be predicted by the notion of a multilingual feature pool (Cheshire, Kerswill, Fox, & Torgersen, 2011, Mufwene 2001).

Download Full-text

Phonetic effects of morphology and context: Modeling the duration of word-final S in English with naïve discriminative learning

Journal of Linguistics ◽

10.1017/s0022226719000203 ◽

2019 ◽

pp. 1-39 ◽

Cited By ~ 5

Author(s):

FABIAN TOMASCHEK ◽

INGO PLAG ◽

MIRJAM ERNESTUS ◽

R. HARALD BAAYEN

Keyword(s):

Lexical Processing ◽

Discriminative Learning ◽

Prosodic Structure ◽

Context Modeling ◽

Learning Network ◽

Theoretical Morphology ◽

English Vowels ◽

Third Person ◽

Lexical Phonology ◽

Segment Duration

Recent research on the acoustic realization of affixes has revealed differences between phonologically homophonous affixes, e.g. the different kinds of final [s] and [z] in English (Plag, Homann & Kunter 2017, Zimmermann 2016a). Such results are unexpected and unaccounted for in widely accepted post-Bloomfieldian item-and-arrangement models (Hockett 1954), which separate lexical and post-lexical phonology, and in models which interpret phonetic effects as consequences of different prosodic structure. This paper demonstrates that the differences in duration of English final S as a function of the morphological function it expresses (non-morphemic, plural, third person singular, genitive, genitive plural, cliticizedhas, and cliticizedis) can be approximated by considering the support for these morphological functions from the words’ sublexical and collocational properties. We estimated this support using naïve discriminative learning and replicated previous results for English vowels (Tucker, Sims & Baayen 2019), indicating that segment duration is lengthened under higher functional certainty but shortened under functional uncertainty. We discuss the implications of these results, obtained with a wide learning network that eschews representations for morphemes and exponents, for models in theoretical morphology as well as for models of lexical processing.

Download Full-text

How German children use intonation to signal information status in narrative discourse

Journal of Child Language ◽

10.1017/s0305000913000251 ◽

2013 ◽

Vol 41 (5) ◽

pp. 1015-1061 ◽

Cited By ~ 5

Author(s):

LAURA E. DE RUITER

Keyword(s):

Narrative Discourse ◽

Pitch Accent ◽

Natural Speech ◽

Control Group ◽

Adult Control ◽

Target Model ◽

Information Status ◽

Speech Data ◽

Adults And Children ◽

German Children

ABSTRACTRecent research on adult German suggests that speakers use particular pitch accent types to signal the information status of discourse referents. This study investigates to what extent German five- and seven-year-olds have acquired this mapping. Semi-natural speech data was obtained from a picture-elicited narration task in which the information status was systematically varied. Surprisingly, data from an adult control group were inconsistent with the claim of a clear status-accent mapping, and demonstrated that adult scripted speech cannot be taken as a target model. However, compared with adults' unscripted speech productions, children were indeed adult-like in their information status marking. Both child groups accented new referents, but tended to deaccent given referents. Accessible referents (whose first mentions were less recent) were mostly realized like new referents. Differences between adults and children emerged in the use of intonation to structure narrations, suggesting that some functions of intonation may be acquired only later.

Download Full-text

What connectionist models learn: Learning and representation in connectionist networks

Behavioral and Brain Sciences ◽

10.1017/s0140525x00079760 ◽

1990 ◽

Vol 13 (3) ◽

pp. 471-489 ◽

Cited By ~ 107

Author(s):

Stephen José Hanson ◽

David J. Burr

Keyword(s):

Computational Approach ◽

Complex Interaction ◽

Connectionist Models ◽

Promising Alternative ◽

Rule Based ◽

Symbol Processing ◽

Connectionist Networks ◽

Connectionist Approach ◽

The Brain ◽

Explicit Rule

AbstractConnectionist models provide a promising alternative to the traditional computational approach that has for several decades dominated cognitive science and artificial intelligence, although the nature of connectionist models and their relation to symbol processing remains controversial. Connectionist models can be characterized by three general computational features: distinct layers of interconnected units, recursive rules for updating the strengths of the connections during learning, and “simple” homogeneous computing elements. Using just these three features one can construct surprisingly elegant and powerful models of memory, perception, motor control, categorization, and reasoning. What makes the connectionist approach unique is not its variety of representational possibilities (including “distributed representations”) or its departure from explicit rule-based models, or even its preoccupation with the brain metaphor. Rather, it is that connectionist models can be used to explore systematically the complex interaction between learning and representation, as we try to demonstrate through the analysis of several large networks.

Download Full-text