The Production of Consonant Clusters

Implications for Phonology and Sound Change

Daniel Recasens

• Publisher: De Gruyter Mouton
• Year: 2018
• eBook ISBN: 9783110565720
• Edition: 1
• Topic: Languages & Linguistics
• Subtopic: Linguistics
• Index: Languages & Linguistics

The book analyzes the articulatory motivation of several adaptation processes (place assimilations, blending, coarticulation) involving consecutive consonants in heterosyllabic consonant sequences within the framework of the degree of articulatory constraint model of coarticulation. It also shows that the homorganic relationship between two heterosyllabic consonants contributes to the implementation of manner assimilations, while heterorganicity as well as sonorancy and voicing in the syllable-onset C2 are key factors in the weakening of the syllable-coda C1. Experimental and descriptive evidence is provided with production, phonological and sound change data from several languages, and more especifically with tongue-to-palate contact and lingual configuration data for Catalan consonant sequences. The book also reviews critically research on the c-center effect in tautosyllabic consonant sequences which has been carried out during the last thirty years.

1Introduction

The phonetic realization of consonant clusters presents a challenge for studies on speech production. Differently from segmental sequences composed of vowels and consonants, the realization of consonant sequences offers a large variety of production scenarios which speakers may resolve through an array of articulatory mechanisms. These mechanisms are conditioned by the need to implement the place and manner of articulation and voicing characteristics for the consecutive consonants and may result in highly antagonistic situations. For example, in a sequence like /ɲr/ which may be found in the Catalan string bany rus “Russian bath”, where / ɲ/ is an alveolopalatal nasal and /r/ is an alveolar trill, the need to anticipate the tongue body lowering and backing movement for the trill conflicts necessarily with the raised and fronted tongue body configuration for the preceding alveolopalatal nasal. Articulatory data show that speakers often solve this antagonistic situation by depalatalizing C1 rather than by palatalizing C2. There is no conflict in other consonant sequences where the lingual activity for C1 is more compatible with that for C2 as, for example, in the case of /nʃ/ where the tongue body raising and fronting movement for /ʃ/ is freely anticipated during the preceding alveolar nasal since the production of /n/ does not require a very precise lingual configuration partly due to the low manner of articulation-dependent production demands involved (see section 4.3.1.2).

A good reason for investigating the production mechanisms of consonant clusters is in order to be able to account for why certain consonant-to-consonant adaptations in place and manner of articulation take place more often than others. A production-motivated rationale for why these articulatory adaptations occur the way they do appears to be more acceptable than other approaches based on phonological principles such as the strength hierarchy, which runs opposite to the sonority hierarchy and proceeds in the progression stops (consonants with maximal strength) > fricatives > nasals > laterals > rhotics (consonants with minimal strength). According to this view, a CC syllable contact is favoured the higher the strength value of onset C2 relative to that of coda C1, and assimilation causes the stronger consonant to become less strong. Therefore, it is predicted that /ln/ and /rn/ should assimilate progressively into [ll] and [rr], respectively, while /nl/ and /nr/ ought to assimilate regressively into the same phonetic outcomes [ll] and [rr] (Maiden, 1995: 72, Vennemann, 1988: 50–51). A problem with this and similar hierarchies is that strength values are not independently motivated but are derived from the outcomes of the phonological and sound change processes themselves, which may lead to different hierarchies, especially among sonorant consonants, depending on which process is subject to analysis (see for a similar objection Cho, 1999: 205–211). Thus, for example, the relatively high strength value for nasals in the scale referred to above does not hold when we take into consideration the trend for /n/ rather than /t/ or /s/ to assimilate in place of articulation to the following syllable-onset consonant (chapter 4). Another problem with manner categories such as stop, fricative, nasal, lateral and rhotic is that they are far too general for making valid predictions about the range of adaptability processes which may apply to consonant sequences such as the ones analyzed in the present study. For instance, under the label rhotic we find manner classes as different as taps, trills and approximants, which operate very differently as coarticulation and assimilation triggers and targets, and a similar objection may be made regarding differences in adaptability between stops and nasals depending on the primary articulator and thus whether they are coronal or dorsal.

A similar criticism may be raised with respect to other phonological models based on features or gestures which consider place and manner of articulation characteristics independently of each other or attribute more relevance to gestural goals than to manner of articulation demands. Throughout this book we generally render consonantal assimilations dependent on the articulatory and aerodynamic requirements involved in the production of the two successive consonants in consonant clusters. The rationale behind this approach is that, in a good number of cases, place and manner of articulation properties account jointly for whether specific assimilatory processes apply or not.

A large number of recent studies on the realization of word-initial and word-final consonant sequences have tested several predictions about gestural overlap and articulatory cohesion at the syllable level formulated by Articulatory Phonology (Browman & Goldstein, 1992). Within this and other frameworks, much attention has been paid to the mechanisms of interarticulatory coordination for successive consonants produced with independent articulatory structures which may overlap in time, i.e., lips, jaw, tongue, velum and larynx, and tongue front and tongue body as well. The present book reviews citically the existing literature on the production of these two-consonant cluster structures, and provides some new insigths about the articulatory organization of heterosyllabic sequences which are (nearly)-homorganic and thus composed of consonants sharing essentially the same primary articulator. Moreover, it does not only deal with the place adaptation mechanisms between the two consonants in the cluster, but also with how the homorganic or heterorganic relationship between C1 and C2 as well as other articulatory and aerodynamic characteristics influence the implementation of manner assimilations (see section 2.4.3 for details) and the weakening of syllable-final consonants. These research topics are relevant in so far as they have been paid little attention in the literature. The study of manner assimilations will not deal with voicing adaptations or segmental insertions which may result from changes in glottal or articulatory timing between the consonants in succession.

A main goal of the present investigation is to show that attention to phonetic detail may contribute to unveil the causes of specific intersegmental adaptation processes. Sometimes what appears to be a categorical assimilation to the ear is not intended as such by the speaker. Even though in Chapter 5 we treat as categorical several apparently complete place and manner assimilations occurring in /sC/ and /rC/ sequences with a voiced or sonorant C2, we are aware that they often are not. Experimental research reveals that speakers may produce these consonant sequences through several C1 phonetic realizations which have been generated through different degrees of articulatory reduction and regressive adaptation. We believe that to the extent that these production mechanisms operate on certain coda consonants and consonant sequences rather than others, they ought to be modelled and accounted for by theories of phonology and sound change.

The book is structured as follows. Chapter 2 provides general information about the production characteristics of consonant sequences and consonant-to-consonant adaptation processes, and formulates a set of theoretical predictions to be tested in the following chapters. It deals with how articulatory gestures are organized in consonant clusters also as a function of syllable position, prosodic and lexical factors, as well as with the articulatory strategies which speakers use when producing the two consonants in succession, i.e., coarticulation, assimilation and blending.

Chapter 3 tests with articulatory data for Catalan consonant sequences a set of predictions made by the DAC model of coarticulation about place adaptation in (nearly)-homorganic clusters across a word boundary. This chapter has a much larger scope than previous publications (Recasens & Pallarès, 2001a, Recasens, 2006) in so far as it reports data for a larger subset of consonant combinations produced by a higher number of speakers from three different dialects and pays special consideration to the production strategies used by the individual subjects. Moreover, ample articulatory evidence is drawn in support of the predictions of the model with production data gathered not only by means of electropalatography (EPG) and acoustic analysis as in other publications but of ultrasound as well. The methodology for analyzing changes in closure or constriction location over time for consonants on the EPG record also differs from previous accounts. Descriptive and experimental data for other languages taken from literature sources are also included.

Chapter 4 reviews production data on the degree of gestural overlap in heterorganic consonant sequences composed of consonants of different place and manner characteristics, and evaluates the implications of these coproduction patterns for blending and regressive place assimilation processes. New ultrasound data are presented on changes in tongue configuration for sequences of overlapping consonants such as velars and dentals/alveolars. The chapter is also concerned with the c-center effect in syllable-onset and syllable-coda clusters.

Chapter 5 investigates, mostly for heterosyllabic consonant sequences, those phonetic factors (e.g., presence of voicing in C2, homorganic and heterorganic relationship between C1 and C2) which are directly involved in the implementation of regressive and progressive manner assimilations and in the weakening of consonants appearing preferably in syllable-final position. Among the weakening processes subject to analysis we have not only lenition and deletion but also rhotacism, vocalization and aspiration.

An aspect worth being mentioned is the crosslinguistic scope of the book. In chapters 2, 3 and 4, production data are presented for consonants and phonetic and phonological processes from not only the Romance, Germanic and Slavic languages but from languages spoken in Asia, Africa and Australia as well. Chapter 5 on manner assimilations in consonant sequences is centered mostly but not only in the Romance languages.

Some terminological concepts need clarification. Throughout the book we distinguish among the following tongue regions: tip and blade at the alveolar zone; front dorsum or predorsum at the palatal zone; back dorsum or postdorsum at the velar zone; back tongue body at the pharyngeal zone. The term ‘tongue front’ is also used to refer to either the tip and blade or else to the tongue tip, blade and front dorsum.

The following abbreviations are used throughout the book: Am. (American), Ar. (Arabic), Cast. (Castilian), Cat. (Catalan), Celt. (Celtic), coll. (colloquial), dial. (dialectal), E. (Eastern), EMA (electromagnetic midsagittal articulometry), Eng. (English), EPG (electropalatograhy), fem. (femenin), Fr. (French), Francoprov. (Francoprovençal), fut. (future tense), Ger. (Germanic), IE (Indo-European), It. (Italian), masc. (masculin), N. (Northern), Occ. (Occitan), pl. (plural), Port. (Portuguese), S. (Southern), Sard. (Sardinian), sing. (singular), Sp. (Spanish), Sw. (Swedish),W. (Western). An index of the languages and dialects is appended at the end of the book.

I would like to thank professor Aditi Lahiri for making this publication possible, the help provided by Clara Rodríguez with the ultrasound data analysis and the comments raised by two anonymous reviewers on a previous version of the manuscript. The research was funded by projects FFI2013-40579-P of the Spanish Ministry of Economy and Competitiveness and 2014SGR61 and ICREA Academia of the Catalan Government.

2Production constraints and articulatory adaptation mechanisms

Consonant sequences differ in interarticulatory organization and gestural overlap depending on whether they are tautosyllabic and thus placed within the same onset or coda syllable, or heterosyllabic and thus split by a syllable boundary which may coincide with a word boundary. Mostly in the case of heterosyllabic consonant sequences, interarticulatory organization and gestural overlap are also very much influenced by whether the two consonants are produced with the same or different articulators and are thus, homorganic or heterorganic. Sections 2.1 and 2.2 look into how the tautosyllabicity/heterosyllabicity and homorganicity/heterorganicity dichotomies shape gestural organization in consonant clusters. As discussed in sections 2.3, 2.4 and 2.5, gestural overlap may give rise to three articulatory adaptation processes, two of which appear to be phonetic (coarticulation, blending) and a third one is phonological (assimilation). The three latter sections deal with how these mechanisms of articulatory adaptation operate spatially and temporally on consonant sequences depending mostly on segmental composition. Section 2.6 is concerned with differences in gestural overlap as a function of prosodic, morphological and lexical factors as well as of speech rate. Sections 2.7 and 2.8 are central to the book. Section 2.7 introduces the basic principles of coarticulatory and assimilatory behaviour in consonant sequences formulated by the degree of articulatory constraint (DAC) model of coarticulation, which will be applied to the study of homorganic clusters in chapter 3 and of manner assimilation and weakening of coda consonants in chapter 5. Section 2.8 accounts for how differences in the direction of the articulatory adaptation effects in heterosyllabic consonant sequences may give rise to regressive and progressive assimilatory processes.

2.1Tautosyllabic consonant sequences

Articulatory data reveal that syllable position influences interarticulatory timing in tautosyllabic consonant sequences, which happen to be more stable if placed syllable initially than syllable finally. An analogous situation holds for single consonants next to a vowel such as /m/ for the production of which, relative to the end of lip raising for the oral closure, velum lowering offset has been found to occur earlier in syllable-final position than syllable initially (as, for example, in seam ore vs see more; Krakow, 1999). Also, during the production of palatalized and velarized/pharyngealized consonants, tongue dorsum movement may precede the activation of the primary labial or lingual articulator provided that the consonant occurs syllable finally but not syllable initially. Thus, in the case of palatalized /p/ in Russian, the tongue dorsum may start moving upwards and frontwards before the labial closing gesture begins if the consonant appears in syllable-coda position while this anticipatory action is not at work if the consonant in question occurs in syllable-onset position (Kochetov, 2006). Kinematic data for American English dark /l/ exhibit comparable syllable-position-dependent differences in interarticulatory organization: tongue dorsum lowering and backing usually precedes the tongue tip raising gesture during the production of dark /l/ syllable finally, as in the word peel,while the two actions take place simultaneously or else tongue tip raising precedes tongue dorsum activation for a clearer variety of /l/ syllable initially, as in the word leap (Sproat & Fujimura, 1993).

Articulatory Phonology has shown that syllable-position-dependent differences in interarticulatory organization result from the way the consonants in sequence are coupled to the syllable vowel nucleus (Browman & Goldstein, 1988, Hoole & Pouplier, 2015). In syllable-onset position, the temporal midpoint of consonant clusters of different degrees of complexity (c-center) is hypothesized to be timed invariantly to the vowel; therefore, adding more consonants should not cause an increase of this temporal interval but the consonant adjacent to the vowel to shift towards the vowel (and the vowel to shorten) and the leftmost consonant to move away from it. In clusters occurring in syllable coda, on the other hand, the first consonant is hypothesized to be coupled with the preceding vowel nucleus and the following consonants each with the immediately preceding consonant but not with the vowel; consequently, adding consonants ought to yield an increase in the temporal distance from the c-center to the vowel but not in the distance between the postvocalic consonant with respect to the vowel itself. In addition, the successive consonants of the two cluster types are expected to differ regarding degree of overlap, which should be less in onset position than in coda position in agreement with the notion that consonantal gestures are more prominent and more precisely articulated in the former position than in the latter. Moreover, as a general rule, the lesser the degree of gestural overlap between the two consonants in onset position, the lesser the c-center effect and the further away C1 stays from the vowel.

It has also been claimed that the mode of articulatory organization of syllable-onset clusters just referred to could serve to preserve the phonemic information for C1 whose acoustic cues risk being masked by C2 in specific segmental combinations. An obvious scenario would be that of sequences where gestural overlap is prone to occur in the front-to-back order rather than in the back-to-front order. Thus, for example, in the case of the sequence /kt/ the burst for the (more posterior) velar stop will be inaudible if the tongue front closure for the following (more anterior) dentoalveolar stop is already formed during C1, while in the case of /tk/ closure formation for the dorsovelar stop before C1 is released should not prevent the b...

Table of contents

• Cover
• Title Page
• Copyright
• Table of contents
• 1 Introduction
• 2 Production constraints and articulatory adaptation mechanisms
• 3 Homorganic clusters
• 4 Heterorganic articulators
• 5 Manner assimilation and weakening
• 6 Recapitulation and discussion
• References
• Index of languages and dialects