Evidence of dialectal divergence in the

vowel system of Chilean Spanish

MAURICIO A. FIGUEROA CANDIA -UNIVERSIDAD DE CONCEPCIÓN

BRANDON M. A. ROGERS-TEXAS TECH UNIVERSITY
DARÍO A. FUENTES GRANDÓN-UNIVERSITY OF BRISTOL
Dialectal divergence
▪While variation can be understood as a tendency of languages to tolerate the existence of competing variants
of the same linguistic variable, change can be viewed as the process whereby a variant which was once
preferred by a speech community gives way to a different one, subsequently becoming more frequent and
more prestigious, normally in a time-frame that occurs across several generations of speakers
▪This can happen for a variety of reasons
▪ Internal factors
▪ External factors
▪ Dialect levelling
▪ Dialect convergence
▪ Dialect divergence
Chile and Chilean Spanish
▪One thing that is important to take into account when studying anything related to Chilean
Spanish linguistics, is just how socially stratified and fragmented it is
▪ This leads to differences in access to services such as housing, public transport, health care, and
education
▪ As a consequence, the language is highly socially stratified as well and has been shown to vary greatly
according to factors such as gender (spirantization Figueroa et al. 2013, Rogers 2016, /s/ variation
Cepeda 1991, 1995, Valencia 1993-1994, vowel production Sadowsky 2012), Age (spirantization Rogers
2016, Rogers and Mirisis 2018), Education (/s/ variation Cepeda 1991, 1995, spirantization Rogers
2016), post-dictatorship vs. pre-dictatorship production of several segments (Sadowsky 2012, 2015),
Socioeconomic stratification (vowels, Sadowksy 2012, /s/ production, Cepeda 1991, 1995)

▪Perceptual studies have shown that speakers are aware of many of these differences (Figueroa
et al. 2013)
 Chilean Spanish variation summarized
▪There is ample consensus that Chilean Spanish constitutes a very distinct dialect of Spanish, distinguishable
from all other variants of the same language (Canfield, 1962; Hualde, 2014)
▪This is true not because its characteristics are all unique to it (although some are, Sadowsky, 2015), but
rather because of the particular combination of these traits and their frequency (Lipski, 1994).
▪Most variation is explained based on social factors and not geographical factors, especially on a phonetic
level
▪For the most part, listeners are aware of the main sociophoneitc dynamics of Chilean Spanish (i.e. lenition,
spirantization, frication, elision etc.)
▪This study concerns itself with the sociolinguistic variation of the vowel system in Chilean Spanish, a
relatively understudied area of Chilean Spanish phonetics and sociophonetics. In particular, methods from
variationist sociolinguistics, acoustic phonetics and statistics are employed to evaluate the hypothesis that
the vowel system of Chilean Spanish is undergoing a process of dialect divergence (Hinskens, Jeffrey, &
Taelfeman, 2000)
 The present study
▪The present study seeks to understand this variation from the perspective of dialectal divergence through an
analysis of production data of the Chilean Spanish vowel system and a number of linguistic and
sociodemographic variables
▪ Provides reference values for Chilean Spanish vowels and to help better understand the role of lexical stress
in vowel production.
 Methodology
▪The data for the current study come from the sociolinguistic interviews of an ongoing corpus that, to date,
comprises speakers from the cities of Santiago, Concepción, and Temuco, three of the main urban centres in
Chile.
▪A subset of 30 speakers, 13 female and 17 male, was used for the current study. All speakers were residents
of the city of Concepción and the surrounding neighborhoods, or poblaciones, of Michaihue, Boca Sur,
Hualpén, Candelaria, Talcahuano, Lomas Coloradas, and Villa San Pedro.
▪All participants were family and friends, and friends of family and friends of one of the study’s authors
▪Linguistic factors analyzed: lexical stress, and vowel category
▪Social factors analyzed: age (18-24, 25-44, 44-54) (Rogers 2016, Rogers and Mirisis 2018), Socioeconomic
stratification (low, high) (Sadowsky 2012, in press)
Methodology
▪Appx 200 vowels per speaker segmented using Praat (Boersma and Weenink 2019) for a total of
6044 tokens after 159 outliers were removed
▪ outliers were 2.5 absolute deviations from the median

▪Vocal fry, devoicing and elided tokens excluded as well as instances of glides
▪Praat script obtained F1 and F2 means for each vowel
▪Measurements were conducted in the inner 60% of the vowels to minimize the effects of
coarticulation
▪Normalized using Nearey 2 method and R (R Core team, 2018) and the vowels package (Kendal
and Thomas, 2018)
 Methodology: Statistical analyses
▪Separate linear mixed models were built to evaluate the effect of the variables analized on the
normalized F1 and F2 values
▪ Lmer function in the lmerTest package in R (Kuznetsova, Brockhoff, & Christensen, 2017)

▪Participant was included as a random factor

▪Following Chappell (2016), a stepwise procedure was used to build the models: first, a null model with the
dependent variable and the random factor was created and then the fixed factors were included and kept only
when they significantly improved the model, as judged by an analysis of variance function (anova) from R.
▪Type-II analysis-of-variance tables for the main effects, interactions and the random factor of each model
were produced via the Anova function from the car package (Fox & Weisberg 2011), and via the ranova
function from the lmerTest package.
Results: Overview of data
Forma Vowel N Non-normalized Normalized (Nearey
nt 2)
mean sd mean sd
F1 /i/ 731 395.33 65.51 0.44 0.07
/e/ 1515 498.77 80.84 0.55 0.07
/a/ 1766 654.97 96.01 0.73 0.08
/o/ 1738 541.77 86.33 0.6 0.08
/u/ 294 443.6 72.54 0.5 0.07
F2 /i/ 731 2220.48 301.66 2.48 0.31
/e/ 1515 1871.59 316.17 2.08 0.3
/a/ 1766 1538.89 229.3 1.71 0.22
/o/ 1738 1196.68 235.09 1.33 0.24
/u/ 294 1058 264.3 1.19 0.3
 Results: Descriptive statistics

Vowel category and Vowel category and

Lexical stress Vowel category and Age
Socioeconomic status
Group
 F1 analyses
▪The best fit mixed-effects model for F1 included vowel category, SES and lexical stress as fixed factors, as
well as their interactions, and participant as a random factor.
− Age group did not improve the model significantly
▪Participants from the lower SES group have lower F1 values than those from the higher group (the vowels
from the lower SES group are more closed)
▪In general, unstressed tokens were more centralized than stressed tokens.
 F2 analyses
▪The best fit mixed-effects model for F2 included vowel category, SES and age group as fixed factors, as well as
their interactions, and participant as a random factor.
▪Significant main effect for vowel category ((χ2(4) = 15012.21, p < 0.001))
▪ /a/ had lower F2 values than /i/ and /e/ and higher values than /o/ and /u/, as anticipated

▪Significant main effect for SES (χ2(1) = 6.29, p < 0.05), seemingly driven by strong F2 differences in /i/
▪Significant interactions for the following:
▪ Vowel category and age group (χ2(8) = 51.24, p < 0.001)
▪ Older participants display higher F2 values (more front vowels), which decreases with each subsequent age group

▪ Vowel category, SES, and age group (χ2(8) = 35.43, p < 0.001)
▪ F2 values of 25-44 year-old participants from higher social groups are higher in comparison to those from less
privileged groups when compared to 18-24 year-old speakers
▪ This suggests fronting from younger speakers of the less privileged social groups
Summary and main observations
▪F1
▪ Unstressed vowels are more centralized (Harmegnies & Poch-Olivé, 1992; Skelton, 1969; Delattre,
1969; Alvord & Rogers, 2014)
▪ SES indicated that speakers from lower SES groups produced lower F1 values, resulting in more closed
productions in the vowel space

▪ F2
▪ Vowel category was only significant linguistic factor. When compared to the front vowels, /a/ had
lower F2 productions, and higher F2 values than the back vowels.
▪ Strongest social factor was observed in the interaction between SES and vowel category, which showed
that /i/ was significantly more fronted in the productions of those in the lower SES group when
compared to the higher SES

▪Considering both acoustic variables together, the younger age group appears to have a more centralized
vowel system than the older two age groups
 Discussion
▪The main motivation for the current study was to examine evidence for dialectal divergence in the Chilean
Spanish vowel system.
▪Linguistic implications: While the results showed some evidence of unstressed tokens being hypoarticulated,
this was mostly confined to F1 values, especially for /a/, /e/ and /o/. However there was no significant effect
for lexical stress with relation to /i/ and /o/. Likewise, there was a lack of a significant main effect for
lexical stress for F2. This suggests a lack of support for a clear systematic expansive effect for lexical stress
on the vowel system of Chilean Spanish.
▪ One possible explanation for this apparent lack of variation compared to previous studies is the fact that
the Chilean Spanish vowel space has been described as being inherently centralized when compared to
that of other dialects (Sadowsky, 2013). Thus, the already ostensibly inherent centralized nature of
Chilean Spanish vowels may create resistance to further reduction.
Discussion
▪Dialectal divergence and age group
▪ Age has been proven to be an important sociolinguistic factor in Chilean Spanish (Sadowsky 2012,
Rogers 2016, Rogers and Mirisis 2018)
▪ Data suggest that as age decreases, /a/ becomes more backed (without taking SES into account)
▪ The cases of /e/ and /u/ are the most interesting. Older speakers showed higher F2 values of /e/ (more
anterior) and younger speakers produced more posterior productions of /e/.
▪ In the case of /u/, older and intermediate age groups displayed lower F2 values than the younger age
group. Younger participants, thus, fronted /u/ more than their older counterparts.
▪ The data preliminarily suggest that the vowel system is centralizing over time.
Discussion
▪Dialectal divergence and SES
▪ Overall, F1 values were lower for all values except /i/ in the productions of the less privileged SES suggesting that this
group produces more closed, or tense, realizations than the upper SES speakers.
▪ /e/ and /i/ showed the most variation of all vowels with relation to SES
▪ Further evidence of dialectal divergence since both SES groups show behavior that is indicative of becoming more different over
time.
▪ Likewise, these differences seem systemic In the sense that all vowels from lower SES speakers seem to be more closed,
and that anterior vowels seem to be more fronted in lower SES group.

▪It is not clear at this juncture if these differences are changes from above or from
below (Labov 2006)
▪ However, the production results of the current study match up with perceptual data reported by Salamanca Gutiérrez
and Valverde San Martín (2009), wherein more fronted and tense productions of vowels were stigmatized by speakers
and associated with groups from traditionally linguistically marginalized groups, such as those from lower
socioeconomic strata.
▪ The fact that speakers are aware of this type of variation is further confirmed when other anecdotal evidence of
linguistic stereotypes in Chilean society is taken into account. While speakers from lower SES groups might say that
speakers from higher SES speak “con una papa en la boca” (i.e., “with a potato in their mouths”), that is, with a more
relaxed and overall less tense articulation, speakers from higher SES groups imitate speakers from lower SES by
articulating instances of /e/ as /i/, although only in fixed stereotypical phrases such as [ˈkʲi.t̪i.ˈpa], from underlying /ˈke
t̪e ˈpa.sa/, “¿qué te pasa?” (i.e., “What’s your problem?”).
 Discussion
▪Sadowsky (2012, 2015) argues that due to some instances of consonantal sociolectal leveling, different
phonemic and phonetic reorganizations are currently taking place in Chilean Spanish. Thus, it is possible
that our results are part of one of these reorganizations
▪In other words, as sociolectal leveling occurs for different consonantal segments, it is possible that the vowel
system is changing as a way to maintain some of the phonetic and sociolinguistic diversity that is being lost
to leveling, perhaps in order to maintain clearly established sociolinguistic boundaries between different
groups.
▪As shown by previous studies on Chilean Spanish, opposite social groups seem to adopt opposite directions
of change in the consonant domain, and do so in a systemic way, creating clear sociolinguistic markers that
speakers associate with different groups.
▪In the case of vowels, the current data indicate that speakers, whether consciously or not, appear to assign
social meaning to the frontness of /i/ and to a lesser extent /e/. More fronted and tense vowels are potential
markers of primarily working-class speech, while the opposite is associated with the Spanish of the upper
echelons of Chilean society.
 Conclusions and future work
▪First evidence of dialect divergence in Chilean Spanish via the vowel system
▪Future perceptual data is needed to confirm if speakers are aware in some fashion of the changes
and variation reported in the current study
▪Finally, while the current study looked at the Spanish of the city of Concepción, subsequent studies must
determine if the variation and divergence documented in this study manifest themselves in a similar,
comparable fashion in different urban and rural areas of the country.
Thank you!
 References
Adank, P., Smits, R., & Van Hout, R. (2004). A comparison ofvowel normalization procedures for language variation research. Journal of the Acoustical Society of America, 116, 3099-3107.

Adimark, (2000). El Nivel Socio Económico Esomar. Manual de Aplicación. Santiago.

Alvord, S. M., & Rogers, B. M. A. (2014). Miami-Cuban Spanish vowels in contact. Sociolinguistic Studies, 8(1), 139-170.

Ashby, M., & Maidment, J. (2005). Introducing Phonetic Science. New York: Cambridge University Press.

Auer, P., & Hinskens, F. (1996). The convergence and divergence of dialects in Europe. New and not so new developments in an old area. Sociolinguistica, 10(1), 1-30.

Bernales, M. (1978). Sobre la palatal africada en el español de la ciudad de Valdivia. RLA. Revista de Lingüística Teórica y Aplicada, 16, 41-51.

Boersma, P., & Weenink, D. (2019). Praat: Doing phonetics by computer (Version 6.0.48) [Computer software]. Retrieved from https://www.fon.hum.uva.nl/praat/

Borland Delorme, K. E. (2004). La variación y distribución alofónica en el habla culta de Santiago de Chile. Onomázein, 10, 103-115.

Borsdorf, A., & Hidalgo, R. (2008). New dimensions ofsocial exclusion in Latin America: From gated communities to gated cities, the case ofSantiago de Chile. Land Use Policy, 25(2), 153-160.

Camus, P. (this volume). Así que vení’ de California… ¿y queda’ tení’?: estudiantes internacionales y su grado de conocimiento del voseo chileno.

Canfield, D. L. (1962). La pronunciación del español en América. Bogotá: Instituto Caro y Cuervo.

Carrasco, L. (1974). Análisis acústico de la secuencia [ř] más vocal en el español de Concepción (Chile). RLA. Revista de Lingüística Teórica y Aplicada, 12, 5-13.

Cepeda, G. (1991). Las consonantes de Valdivia. Valdivia: América Limitada.

Cepeda, G. (1995). Retention and deletion ofword-final /s/ in Valdivian Spanish (Chile). Hispanic Linguistics, 6/7, 329-353.

Cerda-Oñate, K., Fuentes Grandón, D., Soto-Barba, J., & Hamdan Rosales, N. (2015). Variación diatópica, discursiva y fonética de /s/ codal en el habla de profesionales chilenos. Onomázein, 32, 254-274.

Chappell, W. (2016). On the social perception of intervocalic /s/ voicing in Costa Rican Spanish. Language Variation and Change, 28(3), 357-378.

Christen, H. (1998). Convergence and divergence in the Swiss German dialects. Folia Linguistica, 32(1-2), 53-68.

Contreras, D. (2003). Poverty and inequality in a rapid growth economy: Chile 1990-96. Journal of Development Studies, 39(3), 181-200.

Cotter, W. M., & Horesh, U. (2015). Social integration and dialect divergence in coastal Palestine. Journal of Sociolinguistics, 19(4), 460-483.

de Jong, K., Beckman, M. E., & Edwards, J. (1993). The interplay between prosodic structure and coarticulation. Language and speech, 36(2-3), 197-212.

Delattre, P. (1969). An acoustic and articulatory study ofvowel reduction in four languages. International Review of Applied Linguistics, 7, 295–325.

Eaton, K. (2004). Risky business: Decentralization from above in Chile and Uruguay. Comparative Politics, 37, 1-22.

Fabricius, A. H., Watt, D., & Johnson, D. E. (2009). A comparison ofthree speaker-intrinsic vowel formant frequency normalization algorithms for sociophonetics. Language Variation and Change, 21(3), 413-435.