1. Rhotic Neutralization versus Contrast Maintenance in the Spanish of the United States
Sara Zahler
March 10, 2012
GURT 2012

2. Spanish Rhotics [ɾ] [r]
Alveolar tap that consists of one occlusion. Similar to the intervocalic sound in the English word better.
[r]
Alveolar trill that consists of multiple occlusions.

3. Distribution of [ɾ] and [r]
The [ɾ] and [r] are only contrastive in intervocalic position.
Perro vs. pero; [pé.ro] vs. [pé.ɾo]
This distinction does not have a high functional value (Hammond 2000).
Only [r] can occur word-initially, causing [ɾ] to be the only Spanish consonant that cannot occur in this context.
Rota [ró.ta]
When compound words are formed in which word-initial [r] is preceded by a vowel, the [r] is maintained.
Puerto Rico vs. puertorriqueño; [pwéɾ.to.rí.ko] vs. [pweɾ.to.ri.ké.ɲo]

4. Distribution of [ɾ] and [r]
In post-consonant syllable-initial position, the [r] is pronounced.
Honra [ón.ra]
In post-consonant syllable-internal position, the [ɾ] is realized.
Prensa [pɾén.sa]
In syllable-final position, both word-internal and word-final, [ɾ] is more frequent, though [r] is possible as a dialectal or emphatic variation.
Parte [páɾ.te]; hablar [a.ßláɾ] vs. [a.ßlár]
However, when the rhotic is in word-final position but followed by a vowel, it is only realized as [ɾ].
Hablar así [a.ßlá.ɾa.sí] (Lipski 1990: )
The [ɾ] demonstrates an allophonic variation different from that which characterizes the [r] (Canfield 1981)

5. Issues with the phonological study of the Spanish rhotics
Traditionally, the [ɾ] and [r] were given the status of separate phonemes (Navarro Tomás 1977; Stockwell and Bowen 1965; Dalbor 1969,1980).
More recently, phonological theory contends that there is only one Spanish rhotic /ɾ/ with [r] as an allophone (Harris 1983; Nuñez Cedeño 1989).
This study will consider these sounds as two separate phonemes: /ɾ/ and /r/.

6. Issues with the phonetic study of the Spanish rhotics
Lack of studies on the actual realization of these two sounds in most dialects of Spanish (Hammond 2000) (Calero Fernández 1993; Molina Martos 1991)
Discrepancy between what is described in Dialectology and what is actually produced.
In much of the literature, only one variant is ascribed to a particular region. However, many studies have shown that the articulation of the /r/ is generally not as clear or as geographically demarcated as described in the literature.
For example, Willis (2007: 37-45) and Willis (2006:124)

7. Issues with the phonetic study of rhotics
Many investigations only use impressionistic data or do not describe how they obtained/measured their results:
Medina-Rivera (1999)
Vásquez Carranza (2006)
Hammond (2000)

8. Hammond 2000
Contends that there is no longer a distinction between /ɾ/ and /r/ in the natural speech of Spanish speakers.
5 of 180 participants produced /r/ as [r], with 9 tokens of [r] out of 756 in total.
Claims that a distinction between /ɾ/ and /r/ is no longer necessary since it would be difficult to find a list of 30 minimal pairs like perro/pero.

9. Hammond 2000
The excerpt that his participants read only contained examples of word-initial /r/.
The results were largely impressionistic. Acoustic analyses were only carried out on 4 participants with heritage from dialects where the /r/ is claimed to be articulated as the standard [r].
Duration was not considered, which can be a contrastive element between two otherwise similar sounds.

10. Importance of studying duration
Willis and Bradley 2008
Found that when considering number of occlusions, the /r/ (1.2 closures) and the /ɾ/ do not contrast.
However, when considering duration, the /r/ was nearly three times as long as the /ɾ/.
Consequently, in Dominican Spanish, these two sounds are still contrastive intervocalically.
Bradley and Willis 2008
Found same distinction in Veracruz Mexican Spanish

11. Goal of this study
To investigate the claim of rhotic neutralization in U.S. Spanish by acoustically analyzing native speakers’ production of the tap and trill in intervocalic contexts.
Both number of occlusions and duration are taken into account as contrastive features.

12. Research Questions Do the /r/ and /ɾ/ still contrast in U.S. Spanish?
If so, what causes the contrast?
Number of occlusions? Produced as a trill, [r].
Duration? Produced as a tap, [ɾ], with a longer duration.
Some other feature? A non-normative variant.
Lastly, if the trill [r] is still produced, is it done so variably?
If so, what are the factors that constrain this variation?

13. Methodology
17 University of Florida students that are bilingual in English and Spanish.
Spanish is the language at home
Have been in the U.S. since they were four years old or younger
Sociolinguistic survey
Sociolinguistic interview
Reading task

14. Methodology
All tokens of /r/ were extracted from the interviews and coded for various linguistic and social factors (341).
Every 10th token of /ɾ/ was extracted and coded for the same factors (335).
Each token of /r/ and /ɾ/ were acoustically analyzed using Praat (Boersma and Weenink 2012) for duration and number of occlusions.
Statistical analyses using GoldvarbX (Sankoff et al. 2005) were performed on each context to see which factors affected the choice of one variant over the other.

15. Analysis
(H) I e rr o

16. Analysis
T i e n e c a rr o

17. Analysis
r u s a p o r qu e

18. Analysis
F ui m o p a (ra) f i l a d

19. Results /r/ (trill context)
Average duration
Avg. # of occlusions
% trills
% taps
% other
# of tokens
Intervocalic
71 ms
1.798 (119)
53%
24%
23%
153
Post-consonant
Syllable-initial
73 ms
2.667 (3)
75% 0%
25% 4
Word-initial vowel
79 ms
1.764 (72)
46%
27% 97
Word-initial consonant
81 ms
1.568 (51)
28%
35%
37% 82
Word-initial pause
63 ms
2.000 (1)
20%
80% 5
All /r/
75 ms
1.745 (246)
45%
341

20. Results /ɾ/ (tap context)
Average duration
Avg. # of occlusions
# taps
# of tokens
Intervocalic within word
26 ms
0.924 (132)
96%
132
Intervocalic between words
28 ms
0.955 (22)
95% 22
Syllable-final
31 ms
0.842 (57)
44% 84
Phrase final
53 ms
1.000 (9)
70% 10
Post-consonant
Syllable-internal
21 ms
0.948 (77)
84% 87
All /ɾ/
27 ms
0.924 (297)
83%
335
Mention why these taps have an average number of occlusions that’s lower than

21. Intervocalic /r/ and /ɾ/ compared
Phoneme
# of occurrences
Average duration
Avg. # of occlusions
/r/ all
153
71 ms
1.798
/r/ - [r] 81
77 ms
2.185
/r/ - [ɾ] 36
57 ms
1.000
/ɾ/ all
132
26 ms
0.924
/ɾ/ - [ɾ]
128
· An unpaired t-test was performed with R and the difference between the /r/ intervocalic tap and the /ɾ/ context intervocalic tap was statistically significant with a p-value = 7.461e-15
· The mean difference between their duration is 32 ms.

22. Factors conditioning the use of [r]

23. Factors conditioning the use of [ɾ]

24. Discussion
The trill is still produced with high frequency (45% overall, 53% in intervocalic contexts).
In intervocalic contexts, where the /r/ still contrasts with /ɾ/, the /r/ is produced as a tap [ɾ] only 24% of the time.
However, when produced as a tap [ɾ], this sound has a mean duration of 57 ms as compared to when the /ɾ/ tap-context is produced as [ɾ], which has a duration of 26 ms. This difference in duration is extremely statistically significant.

25. Discussion
The [r] is produced least in word-initial position following a consonant (44%). This position is non-contrastive.
Statistical analyses show that word-position is the only significant factor affecting the use of the [r] compared to the [ɾ] in the trill context, /r/. This variant occurs the most in intervocalic contexts.
Statistical analyses also show that word-position is the only significant factor affecting the use of the [ɾ] compared to other variants in the tap context. The [ɾ] occurs more in intervocalic contexts, presumably to contrast with the /r/.

26. Conclusions
Do the /r/ and /ɾ/ still contrast in intervocalic position in U.S. Spanish?
If so, what causes the contrast?
Number of occlusions? Produced as a trill, [r].
Duration? Produced as a tap ,[ɾ], with a longer duration.
Some other feature? A non-normative variant.
Yes. The /r/ and /ɾ/ do still contrast. They contrast primarily through number of occlusions (53% produced as [r]), however 23% of the data is produced as non-normative variants which contrast with the /ɾ/.
24% of the data is produced as a [ɾ], however this tap still contrasts with the [ɾ] in the tap context durationally, with a statistically significant difference in duration between the two allophones.

27. Conclusions
Lastly, if the trill [r] is still produced, is it done so variably? If so, what are the factors that constrain this variation?
Word-position is the statistically significant factor affecting the use of [r] compared to the [ɾ] in the trill context /r/. The [r] occurs most in intervocalic position.
Consequently, this study provides statistical and empirical evidence showing that the /r/ and the /ɾ/ still are separate phonemes, at least in intervocalic position.
In word-initial position, the [r] may not belong to the /r/ phoneme but rather be an allophone of the /ɾ/, due to its statistically significant lower occurrence in word-initial position.

28. New proposed phonemic distribution?
/r/  [r]
[ɾ:]
Any number of non-normative variants such as [ř], [x], [R],
etc.
-intervocalic position, represented orthographically by –rr-
/ɾ/  [ɾ]
[r]
[ø]
Any number of assimilated variants (for example patte for parte)
-intervocalic position –r-, word-initial, syllable final, consonant cluster, etc.
So we see here that the trill sound [r] belongs to both the trill phoneme /r/ and the tap phoneme /ɾ/
Similar to the Spanish nasals where /n/ and /m/ are separate phonemes (rama vs. rana), but [m] is also an allophone of /n/ when it is followed by a bilabial consonant.
En Valencia [em.ba.lén.sja]

29. Works Cited
Bradley, Travis G. and Erik W. Willis. Rhotic Contrast and Neutralization in Veracruz Mexican Spanish. Paper at the 38th Linguistic Symposium on Romance Languages, Urbana-Champaign, IL, April 2008
Boersma, Paul & Weenink, David (2012). Praat: doing phonetics by computer [Computer program]. Version , retrieved 5 March 2012 from
Calero Fernández, María Angeles Estudio sociolingüístico del habla de Toledo: Segmentos fonológicos -/s/ y -/y/. Lleida, España: Pagès Editors
Canfield, D. Lincoln Spanish pronunciation in the Americas. Chicago: University of Chicago Press.
Dalbor, John B Spanish pronunciation: Theory and practice. New York: Holt, Rinehart and Winston.
Dalbor, John B Spanish pronunciation: Theory and practice. New York: Holt, Rinehart and Winston.
Hammond, R., “The multiple vibrant liquid in U.S. Spanish.” In A. Roca’s (ed.) Research on Spanish in the U.S. Somerville, MA: Cascadilla Press,

30. Works Cited
Harris, James W Syllable structure and stress in Spanish: A nonlinear analysis. Cambridge, MA: MIT Press.
Lipski, J “Spanish taps and trills: phonological structure of an isolated opposition.” Folia Linguistica 24,
Medina-Rivera, Antonio Variación fonológica y estilística en el español de Puerto Rico. Hispania
Molina Martos, Isabel Estudio sociolingüístico de la ciudad de Toledo. Ph.D. dissertation, Universidad Complutense de Madrid.
Navarro Tomás, Tomás (19th ed.) Manual de pronunciación española. Madrid: Consejo Superior de Investigaciones Científicas.
Núñez Cedeño, Rafael La /r/, único fonema vibrante en español. Anuario de Lingüística Hispánica 5:
Sankoff, David, Tagliamonte, Sali A., & Smith, Eric. (2005). "Goldvarb X: A Multivariate Analysis Application." Dept. of Linguistics, Univ. of Toronto.

31. Works Cited
Stockwell, Robert P. and Bowen, J. Donald The Sounds of English and Spanish. Chicago: University of Chicago Press.
Vásquez Carranza, Luz Marina On the phonetic realization and distribution of Costa Rican rhotics. Filología y Lingüística 32 (2)
Willis, Erik Trill variation in Dominican Spanish: an acoustic examination and comparative analysis. In Nuria Segarra and Almeida Jacqueline Toribio (eds.), Selected proceedings of the 9th Hispanic Linguistic Symposium, Somerville, MA: Cascadilla Proceedings Project.
Willis, Erik An acoustic study of the “pre-aspirated trill” in narrative Cibaeño Dominican Spanish. Journal of the International Phonetic Association 37(1)
Willis, Erik and Bradley, Travis Contrast maintenance of taps and trills in dominican Spanish: Data and analysis. In Laura Colantoni and Jeffrey Steele (eds.) Selected Proceedings of the 3rd Conference on Laboratory Approaches to Spanish Phonology, Somerville, MA: Cascadilla Proceedings Project.

32. Acknowledgements
Gillian Lord, associate professor at the University of Florida
Department of Spanish and Portuguese, University of Florida
Delano Lamy, PhD student at the University of Florida