JUNE 02 2013

Describing alternative articulations of the Spanish trill /r/ by

ultrasound technology 
Ahmed Rivera-Campos; Suzanne Boyce

Proc. Mtgs. Acoust. 19, 060251 (2013)

https://doi.org/10.1121/1.4801059

CrossMark

 
View Export
Online Citation

23 November 2023 09:44:37

A. Rivera-Campos and S. Boyce

Proceedings of Meetings on Acoustics

Volume 19, 2013 http://acousticalsociety.org/

ICA 2013 Montreal

Montreal, Canada
2 - 7 June 2013

Speech Communication
Session 5aSCb: Production and Perception II: The Speech Segment (Poster Session)

23 November 2023 09:44:37

5aSCb7. Describing alternative articulations of the Spanish trill /r/ by ultrasound
technology
Ahmed Rivera-Campos* and Suzanne Boyce​

​ *Corresponding author's address: Communication Sciences and Disorders, University of Cincinnati, Cincinnati, Ohio 45220,
riveraam@mail.uc.edu
The Spanish trill /r/ is typically described as having a single realization-that is, as the result of aerodynamic forces operating on a tongue
tip/blade constriction placed midsagittally in the dental, alveolar or postalveolar place of articulation, in such a way that air channels along the
sides of the tongue open and close for multiple cycles of vibration. As with American English /ɹ/, this sound is acquired late by typically
developing children and is frequently an element in articulatory disorders. As with American English / ɹ /, perceptually equivalent "correct" trill
/r/'s may be realized differently by different speakers. Knowledge of these alternate "correct" realizations would clearly be helpful to clinicians
and learners of Spanish. In this preliminary study, we report data from ultrasound images of individuals who speak different dialects of
Spanish. Preliminary data suggests there are at least two different articulatory postures used when producing the Spanish trill /r/, one of which
involves lateralization. These articulatory differences do not affect what native listeners categorize as perceptually correct Spanish trills.

Published by the Acoustical Society of America through the American Institute of Physics

© 2013 Acoustical Society of America [DOI: 10.1121/1.4801059]

Received 22 Jan 2013; published 2 Jun 2013
Proceedings of Meetings on Acoustics, Vol. 19, 060251 (2013) Page 1
A. Rivera-Campos and S. Boyce

Background
Most varieties of Spanish have two rhotic sounds, a tap and a trill, which contrast in intervocalic position. This
characteristic is shared by the related languages Catalan and Portuguese. Both trill and tap involve complex
combinations of articulatory gestures (oral and pharyngeal) that must be precisely coordinated with aerodynamic
conditions (Lewis, 2004; Recasens and Pallarès, 1999; Solé, 2002). Some scholars have suggested that the trill is a
longer version of a rhotic alveolar tap, but Ladefoged and Maddieson (1996) found that the requirements for a trill
are more complex than those for a tap. Ladefoged and Maddieson (1999) define the rhotic tap as a single movement
of an active articulator, the tongue tip, making a brief contact with the palate, while the rhotic trill involves the
vibration of the tongue tip against the palate by aerodynamic means. In their comparison of the tap and trill
phonemes of Catalan, Recasens and Pallarès (1999) reported that the trill is produced by a lower tongue predorsum
(in the context of /i/), more jaw opening (in the context of /a/) and a more retracted apicoalveolar closure when
compared to what it is used for the tap. If the trill and tap of Spanish follow the Catalan pattern, speakers of Spanish
may use a more retracted apicoalveilar constriction in order to produce the trill when compared to the constriction
used for producing the tap.
One well known key aspect that differentiates a trill versus a tap is the aerodynamic conditions required to
produce the fast vibratory movements of the tongue tip that characterizes a trill The vibratory movements result
from the Bernoulli effect engendered when the tongue tip is raised toward the alveolar ridge but is lax enough to
move freely and sufficient air pressure is generated through the gap between the alveolar ridge and the tongue tip to
cause the tongue tip to vibrate. These conditions require precise positioning of the tongue tip, precise calibration of
muscle tone, and precise timing in order to generate sufficient airflow (Solé, 2002). The length of the trill is a
function of the pressure differential across the gap; the vibration maintains itself until the differences in pressure are
normalized. By studying oropharyngeal pressure and airflow, Solé (2002) found that voiced trills in particular have
extremely sensitive aerodynamic requirements that if not met will produce an apical stop or a voiceless trill (i.e.

23 November 2023 09:44:37

drop in oropharyngeal pressure between 2.5-3.5 cm H2o). This makes voiced trills highly sensitive to changes in
oropharyngeal pressure when compared to voiceless trills which require a drop of 5 cm H20 in order to become
disrupted.
Knowledge of alternative ways of creating a Spanish apical trill may be of benefit to professionals who work
with Spanish speakers who have articulation difficulties or for people who are learning Spanish as a second
language. By studying this alternate ways of production, individuals with articulation difficulties might experiment
with their articulators in order to judge or “feel” which production of the trill provides to them with a more native or
perceptually correct Spanish trill. Additionally, we can acquire information on whether Spanish speaking
individuals consistently use an alternative articulatory posture or switch between them when using speech depending
on the context where the trill appears.

Methods

Eight speakers of different Spanish dialects were recorded producing several isolated words as well as producing
the Spanish trill in isolation. The Spanish dialects ranged from Colombia, Cuba, Mexico, Puerto Rico and Spain.
The productions of the participants were recorded using an Aloka SD-1000 ultrasound scanner in the mid-sagittal
and coronal planes using a 5MHz probe located under the jaw midway between the point of the chin and the angle of
the neck. The subjects were instructed to hold their heads still but were not restrained. The subjects held the probe
themselves perpendicularly to the coronal and midsagittal planes, and but were continuously monitored to prevent
the probes from moving out of position. It is important to note that the surface of the tongue appears as a bright
white line in both coronal and sagittal views in ultrasound imaging. Also, tongue tip is not visible in sagittal view
due to the tongue tip being located behind the teeth, only the blade, dorsum and root of the togue can be capture.
Audio and recordings were collected simultaneously during the vocal productions using a Sony VCR System and
high quality microphone and saved in .avi format. Participants read a series of words that contained the Spanish trill
/r/ in different vowel context at syllable onset and at word beginning. They were also asked to produce the trill in
isolation. Clips of perceptually correct (as judged by the authors) trill /r/ productions were extracted from the .avi
video stream and formant frequencies from the middle portion of the trill were measured using the The Wavesurfer
acoustical analysis computer program (http://www.speech.kth.se/wavesurfer/).

Proceedings of Meetings on Acoustics, Vol. 19, 060251 (2013) Page 2

A. Rivera-Campos and S. Boyce

Results
The majority of the participants showed evidence of using the more widely known tongue configuration to
produce the trill; that is lowering of the predorsum, retraction (moving towards the back wall of the pharynx) and
lowering of the postdorsum and creating an apicoalveolar constriction to make the tongue tip vibrate when the air
passes. An alternative tongue configuration was observed in the ultrasound video files for two of the participants
(Table 1). Pharyngeal constriction or dorsopharyngeal constriction was noted on all the productions of the trill,
even when the trills were being voiced or when their productions were unvoiced during. Formant measurements (F1
and F2) in word production for the trills with vowel context /i/ and /a/ from four selected Spanish dialects are shown
in Table 2. These measurements are taken from one participant from each of the four selected Spanish dialects.

TABLE 1. Frequency of participant tongue configuration when producing Spanish trill by Spanish dialect.
Dialect Number of participants Standard tongue Alternative tongue
configuration configuration
Colombia 2 X X
Cuba 1 X
Mexico 1 X
Puerto Rico 3 XXX
Madrid, Spain 1 X

TABLE 2. F1 and F2 formant measurements for Spanish trill when produced in words with the vowel /a/
context (/rasa/) and vowel /i/ (/riso/). One participant from each Spanish dialect.
Dialect Vowel F2 F3 Vowel F2 F3
Colombia a 1417 2389 i 1579 2470
Madrid, Spain a 1660 2551 i 1660 2693

23 November 2023 09:44:37

Cuba a 1255 2430 i 1417 2551
Puerto Rico a 1458 2065 i 1660 2227

As mentioned earlier, the majority of the participants used a tongue configuration that sets the tip of the tongue
into vibration against the palate. Interestingly, however, two of the participants showed what appears to be an
alternate way to produce the trill sound. Ultrasound imaging shows that an apicoalveolar constriction is not being
used. Instead, Coronal imaging video showed vibration on the right side of the tongue, while the left side of the
tongue remained stable (Figure 2 and Figure 4). This observation indicates that the two participants appear to be
using the lateral side of the tongue during trill production. For Columbian Dialect Participant 1, video showed the
right side of the tongue in vibratory movement during trill production. In addition, as Figure 1 shows, a very low
tongue blade while the tongue’s dorsum is high up was also observed for the same participant. Although the number
of taps, or contact points, was of not measured during this study, perceptual judgment suggested that the taps were
coming faster for lateral trill than for the alveolar trill –that is, the rhotic sound was being “rolled” at a faster rate
when compared to the other trill productions of the other participants who showed an alveolar trill. This might
suggests that the lateral side of the tongue is making more brief contacts at a faster rate.

FIGURE 1. Still picture from mid-sagittal view video of tongue formation for trill when saying /rasa/. Notice the low
positioning of the tongue blade and high dorsum. The participant was a speaker of Columbian dialect of Spanish.

Proceedings of Meetings on Acoustics, Vol. 19, 060251 (2013) Page 3

A. Rivera-Campos and S. Boyce

FIGURE 2. Still picture from coronal view video of trill production when saying /rasa/. Notice that the right side of the tongue
is lower than the left side. This suggests right side of tongue vibrates while left side remains stable. The participant was a
speaker of Columbian dialect.

Mid-sagittal imaging of trill production for the speaker of the Spanish dialect from Madrid shows what appears
to be a typical or standard tongue formation for producing the trill (Figure 3). When imaging changes to coronal
view, the video demonstrates how the left side of tongue is higher and stable, as if bracing the teeth, while the left
side vibrates during the trill production (Figure 4). Across all participants, there was no apparent effect of vowel
context on tongue configuration during trill productions. Participants who use the alternate (lateral) and alveolar
tongue configuration maintained the same configuration across the different vowel contexts and productions.

23 November 2023 09:44:37

FIGURE 3. Mid-sagittal view of tongue formation for trill when saying /rasa/. Participant was a speaker of the Spanish dialect
from Madrid, Spain.

FIGURE 4. Still picture of the coronal view video of trill production when producing the word /raθa/. Dotted red line reflects
what it is seen in the dynamics of the video. Left side of tongue suggests being stable, as if bracing the teeth while the right side
suggest vibratory area. Participant was a speaker of the Spanish dialect from Madrid, Spain.

Proceedings of Meetings on Acoustics, Vol. 19, 060251 (2013) Page 4

A. Rivera-Campos and S. Boyce

Conclusions

The data demonstrate the use of an alternate way to produce the Spanish trill that is perceptually correct to the
listener. This alternate tongue configuration uses one side of the tongue as the point of constriction where the air
will rush through to create the aerodynamic conditions to make that side of the tongue vibrate while the other side
stays stable. Limited data has been published regarding the use of the lateral sides of the tongue for production of
trill sounds that are usually articulated using apicoalveolar constrictions. A statement in Wikipedia (2013) about trill
consonant mentions the use of lateral trills as a possible way to imitate bird calls but no information is provided
regarding the use of lateral trills as an alternative way to produce the Spanish /r/. Ladefoge (1996) mentions the use
of lateral flaps in several languages (i.e. KiChaka) but there is no mention of Spanish using lateral flaps. It is not
uncommon for a language to have multiple ways to produce a rhotic sound. It has been found that American
speakers of English can create the rhotic sound /ɹ/ using different tongue formations; retroflex, bunched with tongue
tip up and bunched with tongue tip down (Tiede, Boyce, Holland and Chou, 2004). The data in this study suggest
that an alternate way of producing the Spanish /r/ —that can also be perceived as an allophone of /r/ by native
speakers of Spanish—is by vibrating one side of the tongue or using a lateral flap. The use of an alternate way to
produce the trill in connected speech for Spanish opens ground for more research in order to study if native speakers
of Spanish alternate between tongue forms when speaking or if this alternate way releases the speaker from some of
the constrictive parameters that must be made in other to produce a trill using the tip of the tongue. It has been
reported that in Catalan the apicoalveolar tongue configuration for producing the trill, when compared to the tap, the
tongue’s body is more constraint (a more controlled tongue configuration is required) when coarticulated with
vowels /a/ and /i/ that if changed does not result in the production of the trill (Recasens and Pallarès, 1999)
Although none of the participants in this study alternated from one tongue configuration to another during their trill
productions, the use of the alternate (lateral) tongue configuration might release the speaker from the more
controlled apicoalveolar tongue configuration that might be needed during trill production in Spanish.
Additionally, more research is needed in order to study how this alternate way of producing the Spanish trill differs

23 November 2023 09:44:37

from the tongue formation that is usually used. Studying its similarities and differences to the “standard” tongue
formation can help in creating strategies for ameliorating articulation deficits in children who are resistant to
traditional speech therapy or for individuals who are learning Spanish as a second language.

ACKNOWLEDGEMENTS
This research project would not have been possible without the support of many people. The authors wish to
express their gratitude to the participants of this study for giving part of their time to provide their speech samples.
We would also like to thank Sue Schmidlin, supervisor at the University of Cincinnati’s speech clinic, for
facilitating the use of the ultrasound equipment and the clinic facilities and to Dr. Nuria López-Ortega for helping
with participant recruitment.

REFERENCES
Ladefoged, M., Maddieson, I., (1996). The Sounds of the World’s Languages (Blackwell, Oxford), Chap.7, pp. 217-232
Lewis, A. (2004). “Coarticulatory effects on Spanish trill productions” In proceedings of the 2003 Texas Linguistics Society
Conference, ed. Augustine Agwuele et al., Somerville, MA: Cascadilla Proceedings Project, 116-127.
Martínez, E. (1997) “El mecanismo de producción de la vibrante apical multiple” Estudios de Fonética Experimental, 8, 85-97
Recasens and Pallarès (1999). “ A study of /ɾ/ and /r/ in the light of the DAC coarticulation model” Journal of Phonetics, 27, 143-
169.
Sole, M. J. (2002). “Aerodynamic characteristics of trills and phonological patterning” Journal of Phonetics. 30, 655-688.
Trill Consonant. (2013, January 22) In Wikipedia, the free Enciclopedia. Retrieved from
http://en.wikipedia.org/wiki/Trill_consonant
Tiede, M., Boyce, S. E., Holland, C., Chou, A., (2004). “A new taxonomy of American English /r/ using MRI and ultrasound”
Journal of the Acoustical Society of America 115 (5), 2633–2634.

Proceedings of Meetings on Acoustics, Vol. 19, 060251 (2013) Page 5