Page 1

What is and what is not an

articulatory gesture in

speech production:

The case of lateral, rhotic and

(alveolo)palatal consonants

Daniel Recasens

daniel.recasens@uab.es

Universitát Autònoma de Barcelona

Gradus

Revista Brasileira de Fonologia de Laboratório

Vol. 1, nº 1

Dezembro de 2016

https://gradusjournal.com/index.php/gradus/article/view/2

Bibtex: @article{recasens2016what, author = {Daniel Recasens},

issn = {2526-2718}, journal = {Gradus}, month = {dec}, number

= {1}, pages = {23–42}, title = {What is and what is not an artic-

ulatory gesture in speech production: The case of lateral, rhotic and

(alveolo)palatal consonants}, volume = {1}, year = {2016}}

Este texto pode ser livremente copiado, sob os termos da licença

Creative Commons Atribuição-NãoComercial 4.0 Interna-

cional (CC BY-NC 4.0).

https://creativecommons.org/

licenses/by-nc/4.0/deed.pt_BR

Gradus 1 (1)

24

Abstract

Articulatory data are provided showing that, in lan-

guages in which they have phonemic status, (alve-

olo)palatal consonants, dark /l/ and the trill /r/ are

articulated with a single lingual gesture instead of two

independent tongue front and tongue body gestures. They

are therefore simple, not complex segments. It is argued

that tongue body lowering and retraction for dark /l/ and

the trill /r/ is associated with manner of articulation re-

quirements and with requirements on the implementation

of the darkness percept in the case of dark /l/, and that

tongue body raising and fronting for (alveolo)palatals

results naturally from the contraction of the genioglossus

muscle. These consonant units resemble truly complex

palatalized and velarized or pharyngealized dentoalveo-

lars regarding lingual configuration and kinematics, as

well as coarticulatory effects and phonological and sound

change processes. Contrary to some views, the study also

contends that clear /l/ and the tap /ɾ/ are not complex

segments but consonants articulated with a more or less

neutral tongue body configuration which is subject to

considerable vowel coarticulation.

Keywords: segmental complexity, (alveolo)palatal

consonants, alveolar lateral and alveolar trill.

Resumen

Datos articulatorios muestran que, en lenguas en las

cuales las consonantes alveolo-palatales tienen rango

fonológico, la lateral velarizada /l/ y la vibrante ápico-

alveolar /r/ son producidas mediante un único gesto

lingual en vez de dos gestos linguales, uno anterior y otro

dorsal. Así pues se trata de segmentos simples y no de seg-

mentos complejos. El artículo argumenta lo siguiente: el

descenso y la retracción del dorso de la lengua durante la

producción de la lateral velarizada y de la vibrante están

asociados con sus correspondientes requisitos de modo

de articulación, y también con la implementación de la

velarización en el caso de /l/ velarizada; la elevación y

anteriorización del dorso lingual durante la producción

de las consonantes alveolo-palatales son consecuencia

natural de la contracción del músculo geniogloso. Estas

dos consonantes se asemejan a las consonantes dentoal-

veolares complejas ya sean palatalizadas o velarizadas

o faringealizadas, tanto por lo que respecta a la con-

figuración y cinemática linguales como a los efectos

Gradus 1 (1)

25

coarticulatorios y a los procesos fonológicos y de cambio

fonético. En desacuerdo con otros puntos de vista, el

presente estudio también postula que la /l/ no velarizada

y la rótica /ɾ/ no son segmentos complejos sino conso-

nantes articuladas con una configuración lingual más o

menos neutra, que está sujeta a efectos de coarticulación

vocálica considerables.

Palabras clave: segmentos complejos; consonantes

alveolo-palatales; lateral alveolar y vibrante alveolar.

Introduction

The goal of this paper is to argue for an appropriate use

of the concept of articulatory gesture. According to Articu-

latory Phonology, articulatory gestures are phonological

primitives which characterize high-level phonological

units in place of distinctive features.1 Examples of artic-

1. Browman and Goldstein,

“Articulatory phonology: an

overview” (1992).

ulatory gestures are lip and tongue dorsum closing for

bilabial and velar stop consonants, respectively. In recent

times, gestural status has been attributed to the activity

of tongue body regions which are not involved directly in

the formation of the primary closure or constriction for

consonants, as in the case of tongue body retraction for

dark /l/ and the alveolar trill /r/ and tongue body raising

and fronting for alveolopalatal and palatal consonants

such as /ɲ/ and even /j/. According to this view, these

tongue body actions correspond to secondary lingual

gestures which are activated independently from the

primary tongue front gesture whether it involves the

tongue tip, blade or predorsum. The issue is whether, in

parallel to pharyngealized and palatalized dentoalveolars

in languages like Arabic and Russian, respectively, those

consonantal units should be treated as complex segments

endowed with two simultaneous lingual gestures or as

simple segments specified for a single tongue gesture.

The claim that the tongue body actions referred to

above for dark /l/, the trill and (alveolo)palatals should

qualify as secondary gestures is supposed to come from

different sources of evidence. Supporting evidence has

been claimed to derive from specific articulatory events

such as considerable tongue-to-palate contact and the

presence of a /j/-like component at consonant release

in the case of (alveolo)palatal consonants, and tongue

body lowering/backing prior to tongue tip raising during

a preceding front vowel in the case of dark /l/ and the

trill /r/. Another aspect in support of the independent

existence of a tongue body gesture for these consonants

Gradus 1 (1)

26

appears to be their low degree of articulatory variability

whether measured across contextual conditions (contex-

tual variability) or across tokens (random variability),

which may be indicative that the tongue region in ques-

tion is subject to active control by the speaker. According

to earlier proposals, additional evidence is to be sought

in specific sound changes and phonological processes

such as the insertion of the glides [w] before dark /l/

and [j] before or after an (alveolo)palatal consonant, the

argument being that gestural decomposition has taken

place in this case.2

2. Operstein, Consonant

structure and prevocalization

(2010).

In contrast with this line of thought, we believe that

these (co)articulatory characteristics as well as related

sound change and phonological processes should not

be taken in support of the existence of an independent

tongue body gesture. More specifically, we argue that

the parallel behaviour between the consonants under

study and true complex segments such as palatalized and

pharyngealized dentoalveolars follows simply from the

fact that two sets of consonants are articulated similarly.

Thus, for example, the presence of a relatively long tem-

poral lag between the activity of the tongue body and the

tongue front at the release of an alveolopalatal consonant,

which may give rise to an off-glide through segmental

categorization of the CV acoustic transitions, depends

on the relative independence between the two articula-

tory structures during given productions of the simple,

non-complex consonant of interest. Similar instances of

on-glide or off-glide insertion may be triggered by simple

consonants of other places of articulation.3 Moreover,

3. Albano, O gesto e suas

bordas: esboço de fonologia

acústico-articulatória do português

brasileiro (2001).

we cannot accept the view that phonological units may

be specified as simple or complex independently of the

available phonetic data since, among other reasons, this

approach places phonology outside the real world and

does not contribute to explaining the causes of sound

change.

The existence of a tongue body gesture has also been

assigned to consonants which show no clear tongue body

activation. Thus, in Proctor’s view,4 clear /l/ and the

4. Proctor, “Gestural charac-

terization of a phonological class:

the liquids” (2009).

alveolar tap /ɾ/ in Spanish also ought to be specified as

complex, given that the tongue body is less sensitive to

contextual coarticulation during their production than

during that of other dentoalveolars. Likewise, according

to Operstein,5 the fact that all dentals and alveolars may

5. Operstein, Consonant

structure and prevocalization

(2010).

trigger prevocalization or on-gliding proves that these

consonants must be endowed with a tongue body vowel

gesture independently of whether they involve some

lowering and backing of the tongue body or not. Several

Gradus 1 (1)

27

arguments will be provided later against this claim.

The following sections will deal with the issue of seg-

mental complexity in (alveolo)palatals,6, dark and clear

6. (Alveolo)palatal consonants,

p. 27.

/l/,7 and the trill /r/ and the tap /ɾ/.8 Reference will

7. The alveolar lateral, p. 32.

8. The alveolar trill, p. 36.

be made to articulatory and acoustic data for these con-

sonants as well as for true complex consonants such as

palatalized and pharyngealized dentoalveolars. The last

section9 draws some general conclusions about segmental

9. Discussion and conclusions,

p. 38.

complexity based on the production data reported in the

preceding sections.

(Alveolo)palatal consonants

A two-gestural status has been assigned to consonants

such as /ɲ/, /ʎ/ and /j/ labeled often as palatal in the

phonetics literature.10 These consonantal productions

10. Keating, “Coronal places

of articulation” (1991); Keating,

“Phonetic representation of

palatalization versus fronting”

(1993); Lahiri and Evers,

“Palatalization and coronality”

(1991); Calabrese, Markedness

and economy in a derivational

model of phonology (2008);

Gussenhoven and Jacobs,

Understanding Phonology (2013).

have been treated as complex segments specified for a

Coronal node and a Dorsal node and thus, for two inde-

pendent lingual gestures which are activated more or

less simultaneously. Moreover, this gestural specifica-

tion has been meant to apply not only to alveolopalatal

realizations whose closure or constriction encompasses

the alveolar and palatal zones, but also to purely palatal

productions which are articulated at the hard palate

exclusively. Even palatoalveolar consonants such as the

fricative and affricate sounds of the English words shoe

and catch are assigned two gestures in the above-cited

studies.

The assigment of complex status to (alveolo)palatals

in languages in which these consonants are phonemic

is in disagreement with several articulatory facts. Re-

garding place of articulation, lingual configuration and

linguopalatal contact data strongly suggest that (alve-

olo)palatal and palatoalveolar consonants are produced

with a single articulatory gesture in these languages.11

11. Recasens, “On the

articulatory classification of

(alveolo)palatal consonants”

(2013).

Palatoalveolars are articulated with the tongue blade

generally at the postalveolar zone; alveolopalatals are

implemented with the tongue blade, the tongue predor-

sum or both at a constriction location which includes the

alveolar and front palatal areas as a general rule; purely

palatals are realized with the tongue dorsum at the hard

palate. The exact closure or constriction location of (alve-

olo)palatal consonants may vary depending on factors

such as manner of articulation, speaker and language.

Thus, the lateral /ʎ/, as in the Italian word aglio “onion”,

may be alveolar or alveolopalatal but not purely palatal

and thus more anterior than the nasal /ɲ/ (Italian word

Gradus 1 (1)

28

bagno “bath”) and the oral stop /c/ (Romansh word notg

“night”), which are mostly alveolopalatal but may be

purely palatal as well. On the other hand, syllable-onset

/j/, as in English yes and Spanish raya “line”, is often

purely palatal though may also be articulated with an

alveolopalatal central constriction. In light of these place

of articulation characteristics, it may be ascertained that

tongue dorsum raising behind closure or constriction

location for palatoalveolar and alveolopalatal conso-

nants is not triggered by the separate activation of the

tongue body but by the coupling effects between the pri-

mary laminal or predorsal articulator and more posterior

tongue dorsum regions, such that the latter are automati-

cally raised as the former become involved in closure or

constriction formation.

Additional evidence in support of the simple, non-

complex nature of (alveolo)palatals comes from lingual

movement trajectories associated with these consonants.

Indeed, an analysis of the lingual trajectories for produc-

tions of the sequence /aɲa/ by several Catalan speakers,

starting at V1, continuing during the closure or constric-

tion period and ending at V2, reveals that the spatiotem-

poral characteristics of the alveolopalatal nasal consonant

are not related to gestural complexity but to the physico-

mechanical properties of the tongue blade and dorsum.

EMA (electromagnetic midsagittal articulometry) data

for this VCV sequence show that, while the tongue dor-

sum is activated before the tongue blade before closure

onset and the reverse applies after closure release, the

laminodorsal region travels a somewhat smaller distance

at a slower speed and for a longer time during the latter

period than during the former.12 Moreover, the temporal

12. Recasens and Espinosa,

“Lingual kinematics and coartic-

ulation for alveolopalatal and

velar consonants in Catalan”

(2010).

lag between the tongue blade and tongue dorsum dis-

placement maxima during the production of intervocalic

/ɲ/ is very short (shorter than 10 ms according to EMA

data for one Catalan speaker reported by Recasens and

Romero,13 which confirms the existence of a single artic-

13. Recasens and Romero,

“An EMMA study of segmental

complexity in alveolopalatals and

palatalized alveolars” (1997).

ulator. Electropalatographic (EPG) data collected during

the closure period for the Catalan alveolopalatal nasal

also speaks in favour of a single gesture:14 the degree of

14. Recasens, Fontdevila,

and Pallarès, “A production

and perceptual account of

palatalization” (1995).

tongue contact at closure location increases towards the

front alveolar zone and towards the back palate from clo-

sure onset to about closure midpoint, which appears to be

related to an increase in laminodorsal contact against the

palate as closure is being formed; the closure release, on

the other hand, proceeds from front to back and therefore

occurs at the tongue blade before it does at the tongue

dorsum since the former, less sluggish, tongue region

travels faster than the latter. Moreover, the reason why

Gradus 1 (1)

29

(alveolo)palatal consonants (also palatoalveolars) are

highly resistant to coarticulatory effects in tongue dor-

sum lowering/backing from low and back vowels is not

to be sought in the existence of an independent tongue

dorsum gesture but in the fact that the contraction of

the genioglossus muscle and an increase in tongue-to-

palate contact during the closure or constriction period

constrains the entire body of the tongue considerably.

(a) Sequence /aɲ/.

(b) Sequence /ɲa/.

Figure 1: Ultrasound data for

the sequence /aɲa/ produced by

a Catalan speaker. See fig. 4, p.

42.

Ultrasound data for the /aɲ/ portion of the sequence

/aɲa/ produced by a Catalan speaker, plotted in fig. 1 (a),

reveal that, as shown by the sense of the arrows, the

tongue body moves frontward and upward gradually and

holistically as it travels from V1 onset to the maximum

displacement for the nasal consonant during the closure

period. A slow and holistic tongue lowering/backing

motion from closure offset (thick line) to V2 offset may

also be observed during the /ɲa/ portion of the same

sequence, in fig. 1 (b). There is therefore no sign of

two separate tongue components which correspond to

separate lingual gestures.

A different scenario holds in languages in which alve-

olopalatal consonants can only occur as realizations of a

dentoalveolar + /j/ sequence, mostly in casual speech, as

for example the [ɲ]-like and [ʎ]-like realizations of the

sequences /nj/ and /lj/ in English words such as onion

and scalion. The realizations in question are not phone-

mic and thus alternate with [nj] and [lj] and with other

intermediate realizations exhibiting different degrees of

spatiotemporal proximity between the laminoalveolar

gesture for the alveolar consonant and the dorsopalatal

gesture for the palatal consonant. This is a typical blend-

ing scenario in which two separate lingual gestures yield

a single articulatory outcome exhibiting an intermediate

Gradus 1 (1)

30

closure or constriction location between the closure or

constriction locations for the two consonants in succes-

sion.15 In fact, to the ears of speakers of languages in

15. Browman and Gold-

stein, “Articulatory gestures as

phonological units” (1989).

which [ɲ] and [ʎ] have phonemic status such as Spanish,

Catalan or Czech, those phonetic realizations sound more

like the bigestural palatalized productions of Russian

(see below) than like their own simple, non-complex

consonant productions. In Early Romance, the realiza-

tions [ɲ] and [ʎ] of Latin /nj/ and /lj/ (also [ʃ] of /sj/)

became systematic and gave rise to the phonemic units

/ɲ/ and /ʎ/ (also /ʃ/) which are found in the present-day

Romance languages, as exemplified by /ʎ/ and /ɲ/ in

the Italian words aglio “garlic” and bagno “bath”, which

can be traced back to the Latin cognates /áljo/ ALIU

and /bánjo/ *BANEU < BALNEU. In sum, a given alve-

olopalatal production may correspond at the same time

to a blended realization of the primary gestures for two

successive consonants (as in English) and to a phonemic

unit implemented through a single lingual gesture (as in

the Romance languages).

In the phonologists’ view, evidence for segmental

complexity in (alveolo)palatal consonants may be de-

rived from specific sound change and phonological pro-

cesses such as on-glide and off-glide insertion. On-gliding

triggered by /ʎ/ and /ɲ/ has occurred in Northern Por-

tuguese abeilha ‘Portuguese abelha’ “bee” and in Asturian

Spanish [peˈkejɲo] ‘Spanish pequeño’ “little”. Off-gliding

triggered by [c] may be exemplified with Fassan [cjaf]

derived from Latin CAPUT “head”,16 and also with the

16. Elwert, Die Mundart des

Fassa-Tals, p. 67 (1943).

French word chièvre “goat” derived from Latin CAPRA,

whose [j] must have been appended at the time that

the originary velar stop was realized as [c] and there-

fore before this (alveolo)palatal oral stop shifted to [ʃ]

through the intermediate outcome [tʃ].17 An example

17. Recasens, Coarticulation

and sound change in Romance

(2014).

of a phonological process which is supposed to relect

the complex nature of (alveolo)palatal consonants is the

splitting into [jn], or better into [jN], in which the nasal

[N] is unmarked for place, of preconsonantal word-final

/ɲ/ in Majorcan Catalan ([ˈajmbɔ] /áɲbɔ/ any bo “good

year”). In our opinion, there is no need to assume that

these phonetic realizations are achieved through gestural

dissociation and thus, temporal separation of two exist-

ing gestures. Indeed, electropalatographic and acoustic

data reveal that on-gliding before (alveolo)palatal con-

sonants is achieved through the perceptual integration

as a separate segment /j/ of the VC formant transitions,

which may be longer syllable finally than intervocalically

whenever the alveolar contact maximum is delayed con-

siderably relative to the dorsopalatal contact maximum.18

18. Recasens, Fontdevila,

and Pallarès, “A production

and perceptual account of

palatalization” (1995).

Gradus 1 (1)

31

Therefore, the dissociation of articulatory events does not

have to be identified with gestural dissociation, which

may certainly take place in the case of bigestural con-

sonants such as the bilabial nasal /m/ in syllable- final

position.19

19. Krakow, “Physiological

organization of syllables: a

review” (1999).

It has also been proposed that fronted velars need

to be specified as complex in order to account for the

fact that they may front into (alveolo)palatal stops, as

exemplified by the change /ɡj/ > [ɟ] in the word ghianda

“acorn” in Tuscan. Thus, according to Calabrese,20 this

20. Calabrese, Markedness

and economy in a derivational

model of phonology, p. 311

(2008).

change involves the promotion of the coronal articulator

from secondary in the original fronted velar to primary

in the outcoming (alveolo)palatal stop. This analysis

cannot possibly hold since the tongue front stays in

a low position and inactive during the production of

dorsovelars whether front before front vowels or back

before low or back rounded vowels.

Within the autosegmental phonology framework, the

depalatalization of an (alveolo)palatal into an alveolar

is attributed to the delinking of the Dorsal node in the

complex consonant. This change takes place in Alguerese

Catalan, in which preconsonantal word-final /ɲ/ is re-

alized as an alveolar nasal assimilated in place to the

following consonant ([ˈambɔ] /áɲbɔ/ any bo “good year”).

In our view, the depalatalization in question may be

accounted for through articulatory reduction and thus a

decrease in tongue contact size all over the palate surface

without assuming that a dorsal gesture is present. It may

be considered to be just the opposite of the palatalization

of /n/ into [ɲ], which may take place through articula-

tory strengthening, and thus an increase in tongue contact

in prominent positions (Asturian [ˈɲuðo] ñudo from Latin

NODU “knot”) or whenever the alveolar has an intrin-

sic long duration (Spanish [ˈaɲo] año from Latin ANNU

“year”).

In view of these considerations, it may be concluded

that (alveolo)palatal consonants are implemented through

a single lingual gesture in languages in which they have

phonemic status. Complex segments exhibiting a sec-

ondary tongue dorsum raising and fronting gesture are

only available in palatalized consonants of languages like

Russian, in which palatalization applies extensively to

labials, dentoalveolars and dorsovelars. The independent

status of the primary and secondary gestures in this case

is consistent with the presence of a 20-30 ms lag between

the tongue front and tongue dorsum displacement max-

ima for Russian /nj/ in a word like niet “no”.21 Gestural

21. Recasens and Romero,

“An EMMA study of segmental

complexity in alveolopalatals and

palatalized alveolars” (1997).

independence also accounts for the fact that, in compari-

Gradus 1 (1)

32

son to alveolopalatals, palatalized dentoalveolars show a

more anterior closure or constriction location (which is

nevertheless more retracted than that for non-palatalized

dentoalveolars).

The alveolar lateral

It has been claimed that dark /l/ is specified for two

lingual gestures, a primary apical gesture and a secondary

tongue body gesture.22 This claim is based on produc-

22. Browman and Gold-

stein, “Gestural syllable position

effects in American English”

(1995); Sproat and Fujimura,

“Allophonic variation in English

/l/ and its implications for pho-

netic implementation” (1993);

Gick, “Articulatory correlates of

ambisyllabicity in English glides

and liquids” (2003).

tion data for American English /VlV/ sequences with a

front vowel showing that the tongue body is activated

before the tongue tip in the case of dark /l/ while the two

tongue regions are activated more or less simultaneously,

or else tip activation precedes tongue body activation in

the case of clear /l/. Moreover, it has also been suggested

that, in addition to the consonantal apical raising gesture,

all /l/ varieties have a vocalic tongue body gesture which

is realized through more or less lowering and retraction

depending on the degree of darkness.23 This section criti-

23. Sproat and Fujimura,

“Allophonic variation in En-

glish /l/ and its implications

for phonetic implementation”

(1993); Proctor, “Towards

a gestural characterization of

liquids: evidence from Spanish

and Russian” (2011); Proctor,

“Gestural characterization of a

phonological class: the liquids”

(2009).

cally evaluates the arguments which have been put forth

in support of the complex nature of the alveolar lateral

consonant and, in particular, of its dark variety.

An argument in support of the existence of a tongue

body gesture for /l/ is the fact that the transition from

strongly dark to clear varieties of the consonant proceeds

gradually instead of categorically. This may be taken to

mean that a tongue body gesture is available in all cases,

which may be realized through a more or less lowered

and retracted tongue body position depending on factors

such as syllable position, language and speaker. This

possibility is consistent with the existence of no clearcut

differences in /l/ darkness degree among languages and

among syllable positions within the same language.24

24. Recasens, “A cross-

language acoustic study of initial

and final allophones of /l/”

(2011).

Thus, languages may have a strongly dark variety of /l/

(Russian), a moderately dark /l/ variety (Eastern Catalan)

or a clear variety of the consonant (Italian). Moreover, in

languages of the former group, a strongly dark realization

of /l/ occurs typically in all syllable positions, while

languages of the second groups show a clearer realization

of the alveolar lateral at syllable onset than at syllable

coda.

In any case, this scenario appears to be partly con-

tradicted by the existence of languages showing two

extrinsic allophones of /l/ in onset and coda position,

namely, strongly clear and strongly dark varieties in the

syllable-initial and syllable-final positions, respectively

Gradus 1 (1)

33

(Dutch).

Another piece of evidence in support of the complex

nature of dark and clear /l/ is that the tongue body is

quite resistant to vowel coarticulation in the case of the

two consonant varieties, which could be indicative that

this lingual region is subject to active control by the

speakers,25 and, in the case of dark /l/ and as referred

25. Proctor, “Gestural

characterization of a phonolog-

ical class: the liquids” (2009);

Proctor, “Towards a gestural

characterization of liquids: evi-

dence from Spanish and Russian”

(2011).

to above, that the tongue body precedes the action of

the tongue tip during a preceding front vowel. We would

like to argue that the specific tongue body configuration

for alveolar laterals (whether clearer or darker) is to

a large extent conditioned by laterality requirements.

Indeed, the need to let airlow exit through the mouth

sides may account for why the consonant is apical and at

the same time its production involves more predorsum

lowering and jaw opening than in the case of other den-

toalveolars.26 Therefore, the tongue body lowering and

26. Lindblad and Lundqvist,

“[l] tends to be velarised, apical

as opposed to laminal, and

produced with a low jaw, and

these features are connected”

(2003).

retraction motion as well as a restricted degree of contex-

tual variability associated with these lateral consonants

could be conditioned by these manner of articulation

requirements rather than by the independent activation of

the tongue body.

(a) Sequence /il/.

(b) Sequence /li/.

Figure 2: Ultrasound data for

the sequence /ili/ produced by a

Catalan speaker. See fig. 4, p. 42.

Ultrasound data for several Catalan speakers show that

productions of /l/ which are neither strongly clear nor

strongly dark exhibit some more tongue body retraction

towards the velar or pharyngeal zones than other den-

toalveolars such as /t, d, n/. This tongue body action may

be seen during the /il/ portion of the sequence /ili/ in fig.

2 (a); indeed, as shown by the arrows in the graph, dur-

ing the vowel preceding the alveolar lateral, the tongue

body retracts gradually mostly at its upper region and

thus at the velar zone as the tongue front is being raised.

On the other hand, according to fig. 2 (b), some tongue

Gradus 1 (1)

34

body fronting occurs as the anterior portion of the tongue

front is lowered from closure release (thick line) until

about V2 midpoint and thus during the /li/ portion of the

sequence /ili/.

We would also like to claim that the extra u-shaped

tongue configuration for dark /l/ is associated with the

need to give rise to a darkness acoustic percept by retract-

ing the tongue body and lowering the tongue predorsum

and the jaw considerably. If so, even though dark /l/ is

articulated with a similar tongue configuration to the

pharyngealized dentoalveolar consonants of Arabic, the

latter consonants, but not the former, should be treated

as complex segments involving a secondary tongue body

lowering/backing gesture. Another good reason for this

assumption is that the tongue body lowering/backing

motion may be superimposed on several dentoalveolar

consonants in Arabic, while the darkness characteristic

is only available for /l/ in English or Portuguese. The

fact that dark /l/ and pharyngealized consonants exhibit

a similar overall lingual configuration accounts for sev-

eral common characteristics, which should not be taken

to imply that dark /l/ is a complex segment: all these

consonants may be highly resistant to lingual coarticu-

latory effects exerted by contextual phonetic segments;

the common u-like configuration explains why, as also in

the case of pharyngealized dentoalveolars, tongue body

activation precedes tongue tip raising before closure and,

therefore, much C-to-V anticipatory coarticulation may be

available.27

27. Watson, “The directional-

ity of emphasis spread in Arabic”

(1999); Watson, The phonology

and morphology of Arabic (2002).

Differences in contextual variability and coarticulatory

direction between dark /l/ and clear /l/ run against the

notion that clear /l/ should be treated as a complex seg-

ment. Indeed, dark /l/ is much more vowel coarticulation

resistant than clear /l/, and prevalence of the anticipatory

over the carryover direction of the C-to-V coarticulatory

effects is by no means obvious in the case of the latter

consonantal variety.28 Proctor’s argument that clear /l/

28. Recasens and Farnetani,

“Articulatory and acoustic proper-

ties of different allophones of /l/

in American English, Catalan and

Italian” (1990).

ought to be specified for a tongue body gesture because

it shows less coarticulation than [ð] in Spanish cannot be

valid since, as revealed by ultrasound data for Catalan

and by acoustic data for this language and for Spanish,

the dental approximant is far more sensitive to lingual

coarticulatory effects than other dentals and alveolars,

due to its being realized with an open constriction and

involving no strict manner of articulation requirements.

Presumable evidence for segmental complexity in /l/

also comes from sound change. It has been contended in

this respect that the vocalization of /l/ into [w] if dark

Gradus 1 (1)

35

and into [j] if clear speaks in support of the existence of

a vocalic tongue body gesture;29 thus, the tongue body

29. Proctor, “Towards a

gestural characterization of

liquids: evidence from Spanish

and Russian” (2011).

gesture would take over once the tongue tip gesture is

reduced and apical contact ceases to occur. We think that

the change in question is better accounted for without

assuming the presence of a tongue body gesture. Thus,

regarding dark /l/, reduction of the apical gesture is

prone to take place syllable-finally before consonants

allowing apical contact loss, such as labials and velars,30

30. Browman and Gold-

stein, “Gestural syllable position

effects in American English”

(1995); Lin, Beddor, and Coet-

zee, “Gestural reduction, lexical

frequency, and sound change: a

study of post-vocalic /l/” (2014).

as revealed by vocalization examples like Occitan [ˈawbo]

Latin ALBA “dawn” and [fawˈku] FALCONE “falcon”.31

31. Camproux, Essai de géo-

graphie linguistique du Gévaudan,

p. 316 (1962).

After contact loss has occurred, the /w/-like tongue

configuration may give rise to a /w/-like acoustic real-

ization whose spectrum has an F2 frequency of about

1000 Hzへるつ or less. As to the vocalization of clear /l/ into

[j], apical reduction may account for the syllable-final

cases reported below, but not for those occurring syllable-

initially, which may be attributed more easily to either

acoustic similarity, and thus to the fact that both clear /l/

and /j/ share a high F2 frequency above 1500 Hzへるつ,32 or

32. Ohala, “Phonetic explana-

tion in phonology” (1974).

to the presence of an intermediate alveolopalatal lateral

stage in the development from /Cl/ to [Cj] (i.e., /Cl/ >

[Cʎ] > [Cj]).

(/l/ > [j] syllable finally) Central Italian dialects aitro

ALTERU “another one”, Ligurian suicu SULCU “furrow”,

vuipe VULPE’ “fox”, Emilian aibre ALBARU “popplar”.33

33. Rohlfs, Grammatica stori-

ca della lingua italiana e dei suoi

dialetti (Fonetica), pp. 344–346

(1966).

(/l/ > [j] syllable initially) Tuscan and Central Italy

iore FLORE “lower”, chiave CLAVE “key”, pieno PLENU

“full”.34

34. Rohlfs, Grammatica

storica della lingua italiana e dei

suoi dialetti (Fonetica), pp. 243,

247, 252 (1966).

Another sound change which can be interpreted with-

out reference to an underlying tongue body gesture is

the insertion of an on-glide before the dark alveolar lat-

eral, as in the case of the phonetic variants [awl] ALTU

“high” in Sursilvan Rhaetoromance and aurdeia for aldeia

“village” in Northern Portuguese.35 The phonemic catego-

35. Loriot, “Les caractères

originaux du dialecte du Val

Tujetsch (Tavetsch) dans la

famille des parlers sursilvains”

(1952); Vasconcellos, Esquisse

d’une dialectologie portugaise,

p. 96 (1987).

rization of the on-glide as /w/ in phonetic variants such

as these may occur when the falling VC transitions are

highly prominent and correspond to the lowering/back-

ing activity of the tongue body prior to the raising of the

tongue front during the vowel.

A final issue needs to be raised, namely, how dark

/l/ has originated and why it is prone to occur syllable

finally rather than syllable initially. In parallel to the

change /r/ > [ʀ],36 it may be hypothesized that dark

36. The alveolar trill, p. 36.

realizations of /l/ originated from clearer productions

of the alveolar lateral in syllable-final position, where

consonants are often articulated with less tongue-to-

palate contact than in syllable-onset position. This view

Gradus 1 (1)

36

is consistent with the existence of dialects exhibiting

extrinsic allophones (clear in onset, dark in coda), and

also of other dialects which, in spite of having one of

the two /l/ varieties across the board, show a somewhat

darker realization of the consonant syllable-finally than

syllable-initially (see above).

The alveolar trill

It has also been claimed that the alveolar trill /r/ is

specified for two lingual gestures, since it is produced

with some tongue body lowering and retraction, as when

surrounded by high front vocalic sounds, and is highly

resistant to the coarticulatory effects in lingual configu-

ration exerted by contextual vowels and other phonetic

segments.37 Ultrasound data for the sequence /iri/ pro-

37. Proctor, “Towards a

gestural characterization of

liquids: evidence from Spanish

and Russian” (2011).

duced by a Catalan speaker reveal indeed some tongue

body backing and predorsum lowering as we proceed

from V1 onset to about the trill midpoint (fig. 3 (a)), and

the opposite tongue motion, i.e., tongue body fronting

and tongue predorsum raising, from the last apical con-

tact for the trill (thick line) to V2 offset (fig. 3 (b)).

(a) Sequence /ir/.

(b) Sequence /ri/.

Figure 3: Ultrasound data for

the sequence /iri/ produced by a

Catalan speaker. See fig. 4, p. 42.

Little contextual variability both at the alveolar con-

striction location and in tongue dorsum contact at the

palatal zone has been reported to occur for /r/ in Span-

ish and Catalan,38 and ultrasound data for the latter

38. Recasens, “Lingual

Coarticulation” (1999).

language also reveal considerable resistance to tongue

body coarticulatory effects during the production of the

alveolar trill.

There are good reasons to believe that the tongue

configuration and kinematic characteristics of the alve-

olar trill are not associated with the presence of an in-

Gradus 1 (1)

37

dependent tongue body gesture. It has been shown in

this respect that the overall tongue shape and the small

changes in lingual configuration induced by the adjacent

vowels are mainly conditioned by the aerodynamic and

articulatory mechanisms required for apical trilling.39

39. Solé, “Aerodynamic

characteristics of trills and

phonological patterning” (2002).

As pointed out for dark /l/, the fact that the tongue

body configuration for /r/ is similar to that for pharyn-

gealized dentoalvelars does not mean that the trill is a

complex segment. In many respects, this lingual config-

uration also resembles that of retrolex alveolars, which,

to our knowledge, have not been characterized as com-

plex in the phonetics and phonology literature: /r/ and

retrolex consonants share a retracted tongue tip clo-

sure or constriction (postalveolar for /r/, postalveolar or

prepalatal for retrolexes) and a lowered predorsum and a

retracted tongue body position.40

40. Narayanan, Byrd, and

Kaun, “Geometry, kinematics,

and acoustics of Tamil liquid

consonants” (1999); Scobbie,

Punnoose, and Khattab,

“Articulating five liquids: a single

speaker ultrasound study of

Malayalam” (2013).

As attested by data on sound change and phonological

processes, the alveolar trill also parallels both pharyngeal-

ized dentoalveolars and retrolex consonants in that it

favours anticipatory C-to-V and C- to-C effects in clusters

(see Bhat, “Retrolexion: an areal feature” (1973) and

Hamann, “The phonetics and phonology of retrolexes”

(2003) for retrolex consonants, and Recasens, Coartic-

ulation and sound change in Romance (2014), Watson,

“The directionality of emphasis spread in Arabic” (1999)

and Watson, The phonology and morphology of Arabic

(2002) for /r/ and pharyngealized dentoalveolars). Thus,

among other regressive changes, /r/ may cause a preced-

ing front vowel to lower and back (Auvergnat [faˈrado]

FERRATA “bucket”, Ardennes [lo:r] LARDU “lard”),41

41. Camproux, Essai de géo-

graphie linguistique du Gévaudan,

p. 96 (1962); Bruneau, Étude

phonétique des patois d’Ardennes,

p. 233 (1913).

and a preceding dental stop to become alveolar as in the

case of the heterosyllabic Catalan sequence /tr/ (set rams

“seven branches”).

Proctor’s assumption that the alveolar tap [ɾ] exhibits

a dorsal gesture cannot hold since it lies on the finding

that this consonant shows less coarticulation than the

dental approximant [ð], which is highly sensitive to

coarticulatory effects from vowels.42 Ultrasound and elec-

42. The alveolar lateral, p. 32.

tropalatographic (EPG) data for Catalan VCV sequences

collected by ourselves reveal no differences in tongue

body lowering and retraction or in coarticulation between

the alveolar tap and other dentoalveolar consonants such

as /t, d, n/, and acoustic data also show a low degree of

coarticulatory resistance for this consonant as a function

of /i/ vs /a/, which is consistent with its being extremely

short.43

43. Recasens and Pallarès,

“A study of /ɾ/ and /r/ in the

light of the ‘DAC’ coarticulation

model” (1999).

In parallel to the thoughts about the genesis of dark

/l/,44 a word needs to be said about the relationship

44. The alveolar lateral, p. 32.

Gradus 1 (1)

38

between apical /r/ and the uvular rhotic /ʀ/, which may

be found in all word positions in present-day French and

Portuguese dialects. The uvularization of the apical rhotic

is prone to have originated through apical contact loss

and simultaneous tongue dorsum raising in syllable-final

position.45 This hypothesis is in line with the existence

45. Straka, “Contribution à

l’histoire de la consonne R en

français” (1965).

of instances of double articulated rhotics exhibiting a

primary apical constriction and a dorsouvular or dor-

sopharyngeal constriction in utterance-final position in

Canadian French.46

46. Morin, “From apical [r] to

uvular [ʀ]: what the apico-uvular

r in Montreal French reveals

about abrupt sound changes”

(2013).

Discussion and conclusions

Articulatory data for phonemic (alveolo)palatal conso-

nants and the alveolars dark /l/ and the trill /r/ reveal

that there is no apparent reason why in languages in

which palatalization and pharyngealization/velarization

is not distinctive, these consonants should be considered

complex segments composed of a primary tongue front

gesture and a secondary tongue body gesture which are

activated independently by the speaker. Spatiotemporal

lingual characteristics, degrees of coarticulatory resis-

tance and sound change and phonological processes

associated with these consonants, which have been taken

as evidence for segmental complexity, may be attributed

to other factors. These simple consonants parallel truly

complex palatalized and pharyngealized dentoalveolars

in that are produced with a similar tongue body config-

uration. A contrast has also been established between

the blended realizations [ɲ] and [ʎ] of the phonemic

sequences /nj/ and /lj/ in English or Early Romance and

the phonemic units /ɲ/ and /ʎ/ in Spanish or Czech.

Regarding tongue configuration and lingual coarticula-

tion, the presence of tongue body lowering and retraction

and a more or less pronounced dorsal constriction at the

velar or pharyngeal zones appears to be associated with

constraints imposed by trilling in the case of /r/ and with

requirements on laterality and on the implementation

of the darkness percept in the case of dark /l/. On the

other hand, the contraction of the genioglossus muscle is

responsible for the high and anterior position of the body

of the tongue, and for the high degree of tongue body

coarticulatory resistance, in the case of (alveolo)palatals

and palatoalveolars. Clear /l/ and the tap /ɾ/ do not

differ much from other dentoalveolar consonants both

regarding tongue body position and coarticulatory re-

sistance, though laterality demands may cause clear /l/

Gradus 1 (1)

39

to be articulated with some more tongue predorsum

lowering and tongue body retraction.

Specific sound changes like on-gliding and off-gliding

are not generated from complex consonants through

gestural dissociation, but are related to other factors such

as an increase of the temporal lag between tongue front

and tongue dorsum activation during the vowel preced-

ing the consonant or immediately after closure release

during the following vowel. The kinematic characteristics

of all consonants under analysis are conditioned to a

large extent by the physico-mechanical properties of the

tongue front and the tongue body, as well as by the target

configuration that these lingual regions must adopt for a

successful consonant realization.

In sum, we believe that complex consonants involv-

ing a tongue front gesture and a tongue dorsum gesture

should include only palatalized and velarized/pharyn-

gealized consonants of languages where palatalization

and velarization/ pharyngealization may apply to several

consonants differing in primary place of articulation, but

not to consonants from other languages which resemble

the ones above regarding articulatory configuration and

coarticulatory behavior due to their specific manner of

articulation requirements.

Acknowledgments

This research was supported by project FFI2013-

40579-P of the Spanish Ministry of Economy and Com-

petitiveness, by the ICREA Academia program, and by

project 2014SGR61 of the Catalan Government.

References

Albano, Eleonora Cavalcante (2001). O gesto e suas bordas: esboço de fonologia acústico-

articulatória do português brasileiro. Campinas: Mercado de Letras.

Bhat, Darbhe N. S. (1973). “Retrolexion: an areal feature”. In: Working Papers on

Language Universals 13, pp. 27–67.

Browman, Catherine P. and Louis Goldstein (1989). “Articulatory gestures as phono-

logical units”. In: Phonology 6.2, pp. 201–251.

Browman, Catherine P. and Louis Goldstein (1992). “Articulatory phonology: an

overview”. In: Phonetica 49.3–4, pp. 155–180.

Browman, Catherine P. and Louis Goldstein (1995). “Gestural syllable position effects

in American English”. In: Producing speech: contemporary issues (for Katherine Saford

Harris). Ed. by Fredericka Bell-Berti and Larry J. Raphael. New York: Woodbury,

pp. 19–33.

Gradus 1 (1)

40

Bruneau, Charles (1913). Étude phonétique des patois d’Ardennes. Paris: Honoré Cham-

pion.

Calabrese, Andrea (2008). Markedness and economy in a derivational model of phonology.

Berlin: De Gruyter Mouton.

Camproux, Charles (1962). Essai de géographie linguistique du Gévaudan. Paris: Presses

Universitaires de France.

Elwert, Theodor W. (1943). Die Mundart des Fassa-Tals. Heidelberg: Carl Winter.

Gick, Bryan (2003). “Articulatory correlates of ambisyllabicity in English glides and

liquids”. In: Papers in Laboratory Phonology VI: constraints on phonetic interpretation. Ed.

by John Local, Richard Ogden, and Ros Temple. Cambridge: Cambridge University

Press, pp. 222–236.

Gussenhoven, Carlos and Heike Jacobs (2013). Understanding Phonology. New York:

Routledge.

Hamann, Silke (2003). “The phonetics and phonology of retrolexes”. PhD thesis.

Utrecht University.

Keating, Patricia (1991). “Coronal places of articulation”. In: Phonetics and Phonolo-

gy. Vol. 2: The special status of coronals: internal and external evidence. Ed. by Carole

Paradis and Jean François Prunet. San Diego: Academic Press, pp. 29–48.

Keating, Patricia (1993). “Phonetic representation of palatalization versus fronting”. In:

UCLA Working Papers in Phonetics 85, pp. 6–21.

Krakow, Rena (1999). “Physiological organization of syllables: a review”. In: Journal of

Phonetics 27.1, pp. 23–54.

Lahiri, Aditi and Vicent Evers (1991). “Palatalization and coronality”. In: Phonetics and

Phonology. Vol. 2: The special status of coronals: internal and external evidence. Ed. by

Carole Paradis and Jean François Prunet. San Diego: Academic Press, pp. 79–100.

Lin, Susan, Patrice Speeter Beddor, and Andries W. Coetzee (2014). “Gestural reduc-

tion, lexical frequency, and sound change: a study of post-vocalic /l/”. In: Laboratory

Phonology 5.1, pp. 9–36.

Lindblad, Per and Sture Lundqvist (2003). “[l] tends to be velarised, apical as opposed

to laminal, and produced with a low jaw, and these features are connected”. In:

Proceedings of the 15th International Congress of the Phonetic Sciences. Ed. by Maria Josep

Solé, Daniel Recasens, and Joaquín Romero. Barcelona: Causal, pp. 1899–1902.

Loriot, Robert (1952). “Les caractères originaux du dialecte du Val Tujetsch (Tavetsch)

dans la famille des parlers sursilvains”. In: Mélanges de linguistique et de littérature

romanes oferts à Mario Roques. Vol. 3. Paris: Didier, pp. 111–138.

Morin, Yves Charles (2013). “From apical [r] to uvular [ʀ]: what the apico-uvular

r in Montreal French reveals about abrupt sound changes”. In: Studies in phonetics,

phonology and sound change in Romance. Ed. by Fernando Sánchez Miret and Daniel

Recasens. München: Lincoln Europa, pp. 65–94.

Narayanan, Shrikanth S., Dani Byrd, and Abigail Kaun (1999). “Geometry, kinematics,

and acoustics of Tamil liquid consonants”. In: Journal of the Acoustical Society of

America 106.4, pp. 1993–2007.

Ohala, John J. (1974). “Phonetic explanation in phonology”. In: Papers from the Parases-

sion on Natural Phonology. Ed. by Anthony Bruck, Robert A. Fox, and Michael W. La

Gally. Chicago: Chicago Linguistic Society, pp. 251–274.

Operstein, Natalie (2010). Consonant structure and prevocalization. Amsterdam: John

Benjamins.

Paradis, Carole and Jean François Prunet, eds. (1991). Phonetics and Phonology. Vol. 2:

The special status of coronals: internal and external evidence. San Diego: Academic Press.

Gradus 1 (1)

41

Proctor, Michael (2009). “Gestural characterization of a phonological class: the

liquids”. PhD thesis. Yale University.

Proctor, Michael (2011). “Towards a gestural characterization of liquids: evidence

from Spanish and Russian”. In: Laboratory Phonology 2.2, pp. 451–485.

Recasens, Daniel (1999). “Lingual Coarticulation”. In: Coarticulation. Ed. by William J.

Hardcastle and Nigel Hewlett. Cambridge: Cambridge University Press, pp. 80–

104.

Recasens, Daniel (2011). “A cross-language acoustic study of initial and final allophones

of /l/”. In: Speech Communication 54.3, pp. 368–383.

Recasens, Daniel (2013). “On the articulatory classification of (alveolo)palatal conso-

nants”. In: Journal of the International Phonetic Association 43.1, pp. 1–22.

Recasens, Daniel (2014). Coarticulation and sound change in Romance. Amsterdam: John

Benjamins.

Recasens, Daniel and Aina Espinosa (2010). “Lingual kinematics and coarticulation for

alveolopalatal and velar consonants in Catalan”. In: Journal of the Acoustical Society of

America 127.5, pp. 3154–3165.

Recasens, Daniel and Edda Farnetani (1990). “Articulatory and acoustic properties of

different allophones of /l/ in American English, Catalan and Italian”. In: Proceedings of

the First International Conference on Spoken Language Processing. International Speech

Communication Association, pp. 961–964.

Recasens, Daniel, Jordi Fontdevila, and Maria Dolors Pallarès (1995). “A produc-

tion and perceptual account of palatalization”. In: Phonology and phonetic evidence:

papers in Laboratory Phonology IV. Ed. by Bruce Connell and Amalia Arvaniti.

Cambridge: Cambridge University Press, pp. 265–281.

Recasens, Daniel and Maria Dolors Pallarès (1999). “A study of /ɾ/ and /r/ in the

light of the ‘DAC’ coarticulation model”. In: Journal of Phonetics 27.2, pp. 143–169.

Recasens, Daniel and Joaquín Romero (1997). “An EMMA study of segmental complex-

ity in alveolopalatals and palatalized alveolars”. In: Phonetica 54.1, pp. 43–58.

Rohlfs, Gerhard (1966). Grammatica storica della lingua italiana e dei suoi dialetti (Foneti-

ca). Turin: Einaudi.

Scobbie, James M., Reenu Punnoose, and Ghada Khattab (2013). “Articulating five

liquids: a single speaker ultrasound study of Malayalam”. In: Rhotics: new data and

perspectives. Ed. by Lorenzo Spreafico and Alessandro Vietti. Bolzano: Bozen-Bolzano

University Press, pp. 99–124.

Solé, Maria Josep (2002). “Aerodynamic characteristics of trills and phonological

patterning”. In: Journal of Phonetics 30.4, pp. 655–688.

Sproat, Richard and Osamu Fujimura (1993). “Allophonic variation in English /l/ and

its implications for phonetic implementation”. In: Journal of Phonetics 21, pp. 291–311.

Straka, Georges (1965). “Contribution à l’histoire de la consonne R en français”. In:

Neuphilologische Mitteillungen 66.4, pp. 572–606.

Vasconcellos, José Leite de (1987). Esquisse d’une dialectologie portugaise. 3rd ed.

Lisboa: Instituto Nacional de Investigação Científica.

Watson, Janet C. E. (1999). “The directionality of emphasis spread in Arabic”. In:

Linguistic Inquiry 30.2, pp. 289–300.

Watson, Janet C. E. (2002). The phonology and morphology of Arabic. Oxford: Oxford

University Press.

Gradus 1 (1)

42

Figure 4: Lingual configuration data for the sequences /aɲa/, /ili/ and /iri/ produced by a Catalan speaker,

collected with ultrasound. (Left graphs) Lingual splines from V1 onset to V1 offset (discontinuous lines) and

from closure onset to about closure offset (thick lines). (Right graphs) Lingual splines at about closure offset

(thick line) and from V2 onset to V2 offset (discontinuous lines). The temporal evolution of the splines is

indicated by the orientation of the arrows. The front of the mouth is on the right edge of the graph.