Anda di halaman 1dari 30

Journal of Sociolinguistics 18/5, 2014: 604633

Stereotypes versus experience: Indexing


regional identity in Bolivian Valley Spanish1
Anna M. Babel
The Ohio State University, U.S.A.
Stereotypes of highland and lowland identity categories in the Santa Cruz
valleys of Bolivia are linked to phonetic variation in the Spanish discourse
marker pues. Highlanders are believed to say [pwes] or [ps] while lowlanders
are believed to say [pweh] or [pwe]. However, these beliefs erase two types of
differences. First, they erase a distinction among highlanders. While
highlanders from the Potos-Oruro area pronounce pues as they are
believed to, highlanders from Cochabamba do not. Secondly, these beliefs
erase intra-speaker variation. Highlanders sometimes use the lowland
variants of pues, and vice versa. When people use atypical pronunciations,
they invoke an indexical field linked to the group associated with that
variant. This indexical field references stereotypical ideas about groups even
when they are inaccurate. Specifically, highland variants index pushy or
aggressive stances, which are associated with people from Cochabamba,
even though people from Cochabamba do not use highland variants of pues.
Los estereotipos de las categoras de identidad de alte~
no y terrabajense en
los valles cruce~
nos de Bolivia se vinculan con la variaci
on fonetica en el
marcador de discurso pues. Se cree que los alte~
nos dicen [pwes] o [ps] mientras
que los terrabajenses dicen [pweh] o [pwe]. Sin embargo, estas creencias
ignoran dos clases de diferencias. Primero, ignoran una distinci
on entre los
alte~
nos de manera que los alte~
nos de la zona de Potos y Oruro pronuncian
pues seg
un lo que se cree, mientras que en los alte~
nos de Cochabamba este no
es el caso y no lo dicen como se cree. Segundo, estas creencias ignoran la
variaci
on intra-hablante. Los alte~
nos suelen usar las variantes terrabajense
de pues de vez en cuando, y viceversa. Cuando la gente utiliza las formas
atpicas, invocan un campo indexical (indexical field) que se vincula con el
grupo asociado con esa variante. El campo indexical se refiere a ideas
estereotpicas sobre los grupos, a
un cuando estas son inexactas.
Especficamente, las variantes alte~
nas se usan para crear actitudes
agresivas o enf
aticas, las cuales se asocian con la gente de Cochabamba a
pesar de que ellos no utilizan las variantes alte~
nas de pues. [Spanish]

KEYWORDS: Awareness, stereotypes, erasure, indexical fields,


Andean Spanish, pues

2014 John Wiley & Sons Ltd

STEREOTYPES VERSUS EXPERIENCE

605

INTRODUCTION
In this article, I describe a mismatch between speakers ideas about a highly
stereotyped linguistic variable and their experience with language in a small
town in the Santa Cruz valleys of central Bolivia. The stereotyped variable that
I examine, the discourse marker pues, is a highly salient index of regional
identity. In this area, people hold strong ideologies that contrast two social
groups, highlanders and lowlanders. People say that highlanders pronounce
pues as [ps] or [pwes], and they consider this to be especially typical of people
from neighboring Cochabamba. However, in natural production data, people
from Cochabamba do not appear to follow this pattern. There is an apparent
discrepancy between peoples ideas about language and their experience with
language.
I structure my discussion using three sets of evidence. First, I discuss the
explicit ideologies that people from the town in which I do my research hold
about language use and social groups. Second, I use quantitative analysis of a
corpus of naturally produced speech that I recorded in a variety of settings in
the community to establish how speakers use language in daily interaction.
Finally, using close analysis of transcripts, I examine the way that speakers use
variants of pues that are highly atypical for their group.
This study participates in a body of research that characterizes linguistic
usage not only as reflecting meaning, but as producing it through the links
between sociolinguistic features and the social categories that they participate
in (Bucholtz and Hall 2008; Eckert 2012; Irvine and Gal 2000). The ideas that
speakers hold about language do not always line up with their actual
experience. However, these ideas do influence the way that they use language
as a symbolic resource.
This work has two main theoretical contributions. First, it contributes to
research on awareness and control. Research in speech perception and
sociophonetics demonstrates that the ideas that people hold about social
groups can influence the way they perceive the speech signal. This article
furthers this research by examining how beliefs about social structures can
shape the way that people produce linguistic signs. Secondly, it contributes to
work on indexical reference and indexical fields. When there is a mismatch
between peoples beliefs about a highly stereotyped variable and their
experience with that variable, indexical fields are constructed based on
peoples ideas about language rather than their experience with language.

LITERATURE REVIEW
Awareness and control
Most existing work on the topic of awareness and control of sociolinguistic
variables takes Labovs (1972) classic distinction between markers, indicators,
2014 John Wiley & Sons Ltd

606

BABEL

and stereotypes as its starting point. Labov suggests that there is a distinction
between linguistic variables that vary across social groups but which people
are unaware of (indicators), variables that show stylistic variation (markers),
and variables that people are aware of and can discuss explicitly (stereotypes).
The variable that I examine in this article is clearly a stereotype. Labov notes
that stereotypes are often inaccurate, as I find them to be in my data.
In sociophonetics and speech perception, it has been demonstrated that the
social categories that people are presented with can affect the way that they
perceive the speech sounds that they hear. When stereotypes conflict with
what people actually hear, they appear to block peoples ability to perceive
fine-grained phonetic details (Niedzielski 1999). Conversely, people may
perceive linguistic variables even when they are absent because of
stereotypes that they hold about categories of speakers (Munson 2007;
Rubin 1992). It has also been argued that mismatches between peoples
expectations of typical speech and what they actually hear prompts slower and
less accurate processing of linguistic variables (McGowan 2011; Sumner et al.
2013).
The ideas that people hold about language and speakers also affect their
evaluation of speakers who use highly stereotyped variables. The perception of
regional accents affects listeners evaluation of speakers who use regionallyindexed variables, particularly when they speak in a way that listeners
consider unexpected or atypical for their group (Campbell-Kibler 2008;
Carmichael to appear; Squires to appear).
As the aforementioned studies demonstrate, people come to the task of using
and understanding language with pre-existing expectations about the way
that members of social groups use language. However, a persistent question is
how these expectations are constructed, including knowledge gained from
experience and from stereotypes (Drager and Kirtley to appear; Hay, Warren
and Drager 2006; Johnson 2006; McGowan to appear).
These perception studies demonstrate that peoples experience with
language use, when tested in carefully controlled experimental settings, can
be overpowered by the stereotypes that they hold. These studies complement
another body of literature that considers the explicit stereotypes that people
hold about language varieties. This literature can be grouped broadly under
the labels language ideologies (e.g. Coupland and Jaworski 2004; Irvine and
Gal 2000; Silverstein 1979; Woolard 1998), language attitudes (Garrett 2001,
2010; Garrett, Coupland and Williams 2003) and folk linguistics (Niedzielski
and Preston 2003; Preston 1996, to appear). These approaches study the ideas
that people hold about language. However, unlike perception research, this
area of scholarship focuses on data gathered mainly from peoples explicit
opinions about language.
The confluence of these two types of approaches has demonstrated that there
are multiple dimensions and aspects to awareness and control. We must
approach the study of awareness and control from multiple angles and
2014 John Wiley & Sons Ltd

STEREOTYPES VERSUS EXPERIENCE

607

understand it to include multiple levels or dimensions. In particular, awareness


and control are embedded in systems of power that influence social perceptions
of language (Babel to appear-b; Choksi and Meek to appear; Zimman to
appear). The interconnections between language and social systems require an
integrated, socially informed approach to awareness and control.

Indexicality and erasure


Two key concepts that I use in this article to link social systems to language
are indexicality and erasure. Erasure is one of a group of language ideological
processes that structure the ideas that people hold about language (Irvine
and Gal 2000). Erasure simplifies relationships by ignoring or denying
real-world complexity in the process of creating an abstract system of
reference. Indexicality is the process by which ideas about social structures
and values are linked to particular linguistic signs or variables. A linguistic
sign is an index of the ideas that people hold about social groups, linking
the structural elements of language to the larger social context in which it
is used. Sociolinguistic variables can incorporate or generate reference not
only to a single meaning, but to a hierarchy or field of related meanings
(Eckert 2008; Silverstein 2003). Because sociolinguistic variables have
an array of possible meanings, they must be interpreted both in terms of
large-scale tendencies and in terms of particular speakers and contexts of
use (Kiesling 1998: 69). Interpreting the indexical field relies on context,
and varies both in terms of who uses a variable and in terms of who hears it
(Babel to appear-a).

Pues as a discourse marker in Spanish


Pues is generally described as a discourse marker. Most commonly, pues has
been described as having functions relating to causation, coordination, and
contrast (Travis 2005: 227286). Pues can be used to mark hesitation or
consideration when used at the beginning of a sentence, to mark an answer
and also serves for stance-taking (Portoles 1989; Serrano 1997). Serrano also
considers pues to have a significant component of assertion and an affective
charge (significado emotivo). In Colombia, pues has been described as reinforcing
an illocutive act (e.g. Anda pues Go ahead, go on, hurry up) (Grajales Alzate
2011).
The greater part of the literature on pues in the Andes has considered its
status as a contact feature (Calvo Perez 2000; Olbertz 2013; Zavala 2001).
There is no clear parallel in Quechua to support these claims; however, pues in
the Andes is both more frequent and subtly different from pues in other parts of
the world. Andean pues shares some features that have been reported in the
broader Spanish literature, such as being used for stance-taking and for
assertion. It may connote emphasis and contribute to the logical structure of a
conversation or argument (Pfander et al. 2009: 126130). However, it is
2014 John Wiley & Sons Ltd

608

BABEL

generally not used at the beginning of sentences or to mark hesitation. Pues in


the Andes is used to structure and organize relationships between elements of
discourse, a function that can imply and encompass stance-taking, assertion,
and emphasis.

Vowel and consonant variation in Spanish


In my data, pues has a number of phonetic variants:

fully realized [pwes];


final consonant reduction or elision ([pwe] or [pweh]);
diphthong elision ([ps]); or
both diphthong and consonant elision ([p]).

These variants represent the results of two types of processes: (1) syllable-final
consonant reduction and elision; and (2) vowel elision.
There is an extensive literature on the elision of syllable-final /s/ in dialects of
Spanish. As discussed in Brown (2009a, 2009b), a number of factors have
been shown to influence /s/-reduction and elision, such as word length (Terrell
1979), prosodic stress (Alba 1982), and most importantly, the following
phonological context (Lipski 1984). Brown (2009a) confirms that the
following phonological context is the most important conditioning factor for
the reduction of /s/, while also demonstrating that word frequency plays a role
in its realization. There is an interaction between the position of /s/ within the
word and frequency as factors in the reduction or elision of this phoneme
(Brown 2004; Brown and Torres Cacoullos 2002, 2003; File-Muriel 2009;
Minnick-Fox 2006). Terrell (1979: 32) notes that in Puerto Rican Spanish,
pues is unique among monosyllabic words in having an exceptionally high
frequency of /s/ elision.
Vowel reduction and elision is a much less-studied phenomenon in Spanish.
Unstressed vowel reduction has been documented in central and northern
Mexico (Lope Blanch 1983; Serrano 2006) and the Andean highlands
(Delforge 2008). While Lope Blanch finds no social factors affecting unstressed
vowel reduction, Serrano suggests that men tend to use more extreme forms of
reduction (i.e. elision) and that they shift styles, avoiding elision when they
read word lists. Serrano (2006: 15) also mentions pues as an environment in
which vowels are often reduced or elided.
In the Andes, Delforge (2008:107) notes that unstressed vowel reduction is
most frequent when associated with a following /s/. Delforge concurs with
previous studies that the most common environment for devoicing is between
voiceless consonants. Lipski (1990: 3, 13) discusses vowel reduction in
highland Ecuador, concluding that vowel reduction usually occurs in contact
with /s/ and is generally most frequent in the final syllable and in the
environment of /e/. Lipski (1990:1), too, mentions pues as a prototypical, even
stereotypical, environment for vowel elision.2
2014 John Wiley & Sons Ltd

STEREOTYPES VERSUS EXPERIENCE

609

While both vowel and consonant reduction and elision have been the focus
of study in research on Spanish, to my knowledge there are no existing studies
that specifically examine reduction in the lexical item pues.

THE SANTA CRUZ VALLEY REGION


My field site is located in central Bolivia in the transition zone between the
Andes mountains and the Eastern Bolivian lowland plains. The local dialect of
Spanish has many phonetic features in common with lowland dialects of
Spanish, such as syllable-final consonant reduction and elision. At the same
time, there is considerable grammatical and lexical influence from Quechua
due to extensive contact influence.
I refer to the town in which I work by the pseudonym Iscamayo.3 Iscamayo
is a relatively large town of about 10,000 residents and is a regionally
important population center. In Figure 1, the circle marks the approximate
location of Iscamayo, halfway between the cities of Cochabamba and Santa
Cruz. The cities of Cochabamba and Santa Cruz are located in provinces of the
same name. Iscamayo is positioned precisely on the border between the
provinces of Cochabamba and Santa Cruz. In Iscamayo, everything to the west
of the Cochabamba border is considered a highland area, even though western
Santa Cruz and eastern Cochabamba are geographically similar. This contrasts
with discourses in other parts of Bolivia, especially the highland altiplano
region, where Cochabamba is not considered part of the highlands, but rather
part of the valley region. In Iscamayo, those who are considered vallunos
people from the valleys were born and raised either in the town of Iscamayo
or in smaller towns surrounding the Iscamayo valley. This regional
identification is part of a broader identification with the Santa Cruz valleys
that forms a common link between towns that lie within the Andean foothills
in the western part of the Santa Cruz province.
Iscamayo is a migration center because of its productive agricultural
economy. The town is linked to the cities of Cochabamba and Santa Cruz by
major roads, economic activity, cultural ties, and relative geographical
proximity. Passenger buses travel two or three times a day to Santa Cruz (a
journey of six to seven hours) and three times a week to Cochabamba (11
hours). It is only via these two cities that Iscamayo is connected to other major
cities, such as La Paz, Potos, and Oruro. The latter cities are considered
relatively remote and people travel to them only occasionally. As can be seen
in Figure 1, Potos and Oruro are located at some distance to the south and
west of Iscamayo. For these reasons, in discussions of highlanders the group
of reference for people from Iscamayo is generally Cochabamba rather than
Potos or Oruro. A small number of residents of Iscamayo come from Potos
and Oruro. They tend to be relatively well-educated individuals teachers,
nurses, doctors, and some merchants.
2014 John Wiley & Sons Ltd

610

BABEL

Figure 1: Map of Bolivia (based on United Nations Map No. 3875 Rev. 3 August
2004). Circle indicates approximate location of Iscamayo

I have carried out long-term ethnographic fieldwork in this area. My


connection with the community dates to 2002, when I arrived there as a
Peace Corps volunteer. Over the subsequent years, I married into a local family
and continue to return for long visits on an annual basis. I began doing
sociolinguistic research in the area in 2004. My involvement with the
community allowed me to gather a range of data, including unselfconscious
speech from speakers who might ordinarily resist being recorded. In the field,
I make recordings in WAV format with a hand-held solid-state recorder. I also
keep field notes and collect ethnographic data through participant-observation
activities.
2014 John Wiley & Sons Ltd

STEREOTYPES VERSUS EXPERIENCE

611

My recordings are field recordings. Like any other data source, this brings
with it strengths and weaknesses. The strength of recordings made in the field
is that it is possible to collect relaxed, natural data in conversational settings.
The weaknesses of field recordings are that their sound quality is often poor
because of ambient noise and speakers moving around, and it is virtually
impossible to control the demographics of the participants in the recordings.
Likewise, I cannot generally control the participation structure or the relative
quantity of speech from a particular participant.
The recorded data that I discuss in the following sections come from
conversations, interviews, and meetings that I recorded over several trips to
the field between 20082012. They also include observations from my field
notes and participant-observation activities. I discuss the particular
methodologies for each data set in the corresponding section below.

DATA I: LANGUAGE IDEOLOGIES REGARDING HIGHLANDERS


This section consists of an overview of language ideologies regarding the
use of pues as an index of regional identity in Iscamayo. The main portion
of these data come from July 2012, when I carried out thirty interviews as
part of a perception experiment, though they also include comments and
observations from prior fieldwork periods. The interviews were intended to
guide the respondents to focus on regional language stereotypes, as well as
to learn more about their understandings of typical language use among
people of different social categories. The responses I received were similar to
responses that I heard in language ideologies interviews that I carried out
on other occasions.
Three relevant points emerged from the language ideologies interviews.
First, people associated /s/-retention with people from the highlands and
specifically with people from Cochabamba. Secondly, people associated the use
of [pwes] and [ps] with people from the highlands and specifically with people
from Cochabamba. And thirdly, people characterized people from Cochabamba
as aggressive, rude, and pushy.

Highlanders whistle when they talk


Consultants of all backgrounds commented that people from the highlands
made the s whistle or whistle when they talk. This observation was
repeatedly linked specifically to people from Cochabamba. These comments
were consistent across speakers. Alejandro named a town on the border of the
Cochabamba province and told me that from that point on, con puro S hablan
They speak with all S-es. Remedios told me that she heard un sonido sssss en
Cochabamba, y en Santa Cruz no es as An sssss sound in Cochabamba, and in
Santa Cruz its not like that. Alba told me that En Cochabamba la S mas le
aumenta In Cochabamba they add S more, an idea that was echoed by Liliana,
2014 John Wiley & Sons Ltd

612

BABEL

who said that En Cochabamba usa con S las palabras In Cochabamba [they] use S
with words. Likewise, Berta told me that En Cochabamba la S lo incluyen en todo
In Cochabamba they include the S in everything. Mario contrasted this
pronunciation with Santa Cruz. He told me that Quechua-speakers [i.e.
highlanders] mas pronuncian la S, J en Santa Cruz pronounce the S more, in
Santa Cruz [they say] J.4

Ps, ps in the highlands, Pueee-e-e-eh in the lowlands


Pues emerged again and again as one of the most salient markers of regional
identification in the course of my fieldwork in the community. Consultants
commented that highlanders emphasize pues more and that they generally
use this discourse marker frequently. Performances of highland speakers
speech included frequent use of [pwes] and [ps]. People from the lowlands, on
the other hand, were characterized as using an extended [pwe] or [pweh] with
a long diphthong and rising prosody. This performance included consistent /s/
reduction or elision.
Many of the same consultants who mentioned /s/ also mentioned pues.
Liliana told me that people from Cochabamba said Claro pueSSS o s pueS
Naturally or of course. She drew out the [s] sound as she imitated these
phrases. Isabel gold me that En Cochabamba dicen puro pueSSS In Cochabamba
theyre always going pueSSS, emphasizing the final sibilant. Other consultants
contrasted this with Santa Cruz. Inocencia told me that in Cochabamba people
said pues [pwes], while in Santa Cruz people said pueh [pweh]. Similarly,
Rosala imitated people from Santa Cruzs pronunciation of pues [pweh] and
contrasted it with Cochabamba, where she said it was pronounced [pwesssss].
Jhesica told me that in Cochabamba people said ya pues [pwes] okay, while in
Santa Cruz people used la jota, orthographic J (see Note 2). In his interview,
Mario agreed, saying that in Santa Cruz people said pueh [pweh].
This consistency in the responses that I obtained is all the more striking
because I never mentioned nor asked about /s/ variation or the pues discourse
marker in the interviews. The perception experiment was designed to elicit
attitudes about non-standard vowel height, a common Quechua contact
feature. For this reason, /s/ was not manipulated in the stimuli, nor was it
intended to be an object of analysis. In the interview questions, I never made
any mention of /s/ or pues, although in follow-up questions I would sometimes
ask about tone, accent, or words that are typical of regional or language
groups to elicit a more specific answer from a participant. All the interviews in
July 2012 were carried out individually, so consultants were not responding to
what other people said. Nevertheless, 22 of the 30 respondents mentioned the
pronunciation of /s/. Fifteen participants mentioned the lexical item pues. Only
four respondents volunteered neither /s/ nor pues as an important marker of
regional distinction.

2014 John Wiley & Sons Ltd

STEREOTYPES VERSUS EXPERIENCE

613

Two consultants also mentioned Potos as a distinct category. One, himself a


Quechua-speaking migrant from Cochabamba, said that people from Oruro
speak silbando con S whistling with S. The other, from Potos, did not mention
any difference in the use of pues or the use of /s/. Rather, she commented that
people from the Potos-Oruro area no lo pronuncian correctamente dont
prounounce [Spanish] correctly and use the discourse marker i (borrowed
from Quechua).

Aggressive, rude, and ill-bred


My interviewees held negative ideologies about highlanders in general and
people from Cochabamba in particular. In these comments, highlanders were
framed as aggressive, rude, and pushy. Because my consultants were reluctant
to express baldly negative statements during recorded interviews, I look not
just to the July 2012 interviews but to my broader experience in the field to
support this claim.
In his interview with me, Octavio imitated the speech of a person from
Cochabamba. Wawa anda trae tantarme i! Go bring me bread, child! he said,
using a mixture of Quechua and Spanish. When I asked him to explain the
differences that he perceived, he mentioned the Quechua deference suffix ri in
Cochabamba Spanish and the use of affectionate terms such as the Quechua
loanword waway my child, but added, sin embargo ya despues te echan palo but
they beat you with a stick afterwards anyway. This contrast was projected to
linguistic practice in a comment by Tom
as, the son of highland migrants. He told
me that En Santa Cruz mas lo alargan las palabras . . . en Cochabamba mas secos, mas
directos son In Santa Cruz they draw their words out more . . . in Cochabamba
they are drier, more direct. In this context, I understand seco dry to mean
abrupt having a conversational style with little politeness or mitigation.
These attitudes were also evident in interactions that I observed in previous
fieldwork periods. In 2008, I interviewed a schoolteacher, Carmen, who told
me that she didnt like immigrants, meaning highlanders. Continuing on the
same topic, she told me that she didnt want her children to marry people from
Cochabamba because they were malos mean. When I asked her what she
meant, she elaborated on this idea. No son como nuestra gente, te hablan muy
lindo pero si le niegas alg
un favorcito ya no te miran Theyre not like our people,
they talk to you very nicely but if you ever deny them some little favor they
wont even look at you, she told me. Another interviewee, Sandra, told me
that she didnt like the Quechua language because Me da rabia por lo que, por las
personas que son tan ignorantes, no? It ticks me off because of, because of the
people who are so ignorant [rude], right?
In daily interactions, I also observed negative attitudes towards highlanders.
A woman who complained about her neighbors unfriendly and inconsiderate
conduct was met with the reply Colla es pues Its that shes a highlander. After
a party that was put on by a family from the Cochabamba area got out of
2014 John Wiley & Sons Ltd

614

BABEL

hand, a consultant who was a native of the Santa Cruz valleys characterized
the people who attended as agresivos, maleducados y malcriadotes aggressive,
rude, and terribly ill-bred, adding no son como nuestra gente theyre not like our
people. I was told more than once that I was lucky that my mother-in-law,
who people described as kind and good-natured, was from Santa Cruz rather
than from Cochabamba. When I asked why, I was told that highlanders were
demanding and critical and would not shy away from verbally and physically
abusing a daughter-in-law in their household. I heard highlanders in general
and migrants from Cochabamba in particular blamed for everything from
water shortages to local petty thefts and disputes between neighbors. I was
strenuously advised not to rent my house in Bolivia to a family from the
highlands while in absentia in the United States, on the presumption that they
would destroy the property and resist paying rent.
Even people who were themselves migrants from other parts of the
highlands were hostile towards people from Cochabamba. Sara, originally
from Potos, told me that En Cochabamba mas habladores son People from
Cochabamba are more malicious-gossips. This echoed a conversation I had
some years before, speaking with Lorenza, also a migrant from the highlands,
who used nearly identical words in describing people from Cochabamba as
malicious gossips. She added Unas viboras son Theyre snakes. A teacher from
Potos commented to me that children from the Potos area were well-behaved,
obedient, and respectful, while children from the valleys which in this
context I understood to include both Santa Cruz and Cochabamba were
rambunctious, difficult to discipline, and talked back to teachers.
When people made positive comments about people from Cochabamba, they
emphasized their hard work and their ambition to advance in the world. In
separate interviews, Marina and Silvio told me that all the big businessmen in
Santa Cruz were originally from Cochabamba. Marina added that highlanders,
who she identified as people from Cochabamba, a ellos les gusta trabajar,
y ahorrarse y tener algo they like to work, and save, and have something.5 She
contrasted this with lowlanders, who she said were happy even when they had
nothing at all. This attitude was summed up in a common saying: Colla burro,
camba flojo. This exchange of insults contrasts highlanders, whom people called
burros donkeys meaning they were hard-working but stubborn and rather
stupid and lowlanders, whom people called flojos lazy.

Discussion: Language ideologies


In interviews, consultants repeatedly associated highlanders with s-retention
and lowlanders with s-reduction. Overwhelmingly, my consultants associated
people from Cochabamba with s-retention, especially in the context of pues.
These perceptions were very consistent across participants, regardless of the
age, gender, place of origin, identification as a language speaker, or other social
attributes of the interviewees.
2014 John Wiley & Sons Ltd

STEREOTYPES VERSUS EXPERIENCE

615

My consultants expressed negative attitudes towards highlanders in general


and in particular towards people from Cochabamba. Whether positively or
negatively framed, all interviewees agreed that people from Cochabamba were
hard-working but aggressive and badly behaved. These negative attitudes were
not limited to interviewees from the Santa Cruz valleys, but were also expressed
by migrants from the Potos-Oruro area.

DATA II: QUANTITATIVE ANALYSIS OF PUES VARIATION


In this section, I describe results from a quantitative study in which I examined
the distribution of pues in my corpus of recorded interaction in the Iscamayo
area. Using categories that emerged from my interviews, I examined pues
variation as a function of place of origin, identification as a Quechua-speaker,
and length of residence in the Iscamayo area.
Following language ideologies I found in the interviews I carried out, locals
include people who were born and raised in Iscamayo or in the local valleys
surrounding Iscamayo (the latter are referred to as vallunos). Highlanders, on
the other hand, were born and raised in the Cochabamba province or in the
Potos-Oruro region. One town that has special status is Carapar
(a pseudonym), which is something of a sister city to Iscamayo, just over
the Cochabamba border. In interviews, Carapar came up frequently as a place
where people talked like highlanders, though this was not attached to the same
negative ideologies that people associated with immigrants from the highlands.
I added Carapar as a separate highland category due to its special relationship
with Iscamayo. Based on the language ideologies I found in interviews,
I expected to find a clear split in the pronunciation of pues between
highlanders, including Cochabamba, Potos-Oruro, and Carapar, and locals,
including people from Iscamayo and the local valleys.

Methods
The data that are presented in this section come from a data set that was
collected in Iscamayo during an 11-month fieldwork period in 2008. During
this period, I made recordings of 16 conversations, 16 interviews, and 16
community meetings. The second 10 minutes of each recording were
transcribed for a separate project (Babel 2010). I selected a subset of
recordings based on the presence of non-standard orthographic transcription
of pues.6 This subset comprised 26 10-minute recordings, in which
approximately 56 speakers participated. Twenty-nine of these speakers were
Valluno (from the valleys surrounding Iscamayo). Thirteen were born and
raised in Iscamayo, six were from Cochabamba, four were from Carapar and
four were from the Potos-Oruro area.
A trained research assistant used the transcriptions to listen through the
original recordings. He clipped the part of the WAV file that contained each
2014 John Wiley & Sons Ltd

616

BABEL

instance of pues and provided a transcription of the token in IPA. When


possible, he used WaveSurfer (Beskow and Sj
olander 2011) to generate a
spectrogram, which he used to check his auditory analysis. Because the
recordings were field recordings, not collected in a lab or a sound booth, they
were often noisy. When the spectrogram was not useful, he relied on auditory
analysis alone. I checked each of the transcriptions and discussed any
questions with the research assistant. When we did not agree on a positive
identification of a particular token, it was discarded. In all, a total of 359
tokens were used in the analysis. Of these tokens, 132 were produced by
Vallunos, 78 by people from Iscamayo, 55 by people from Potos-Oruro, 53 by
people from Cochabamba, and 26 by people from Carapar. Fifteen tokens of
pues were produced by speakers whose origins I was unable to determine. The
latter tokens were used for the linguistic analysis but not for the social
analysis.

Regression analysis
In order to understand patterns of phonetic reduction of pues and their
correlations with social groups, I used mixed effects logistic regression to
examine the social and linguistic factors that influenced vowel and consonant
elision in my corpus. The predictive factors were analyzed collectively within
type (social, linguistic). I analyzed the data for four dependent variables in four
separate logistic regression models:

Consonant elision: [pwe], [pweh], [p]


Vowel elision: [ps], [p]
Both consonant and vowel elision: [p]
Vowel elision only: [ps]

The five independent variables were grouped into social factors and
linguistic factors. The three social factors were place of origin, length of
residence in Iscamayo, and self-identification as a Quechua speaker. The two
linguistic factors were sentence position and following sound. The models
included all of the fixed effects in Table 1 as well as a random effect of speaker
with random intercepts.
While my consultants mentioned only consonant vs. vowel elision, I found
in transcriptions and recordings that my local consultants sometimes elided
both the vowel and the consonant in a single production to form another pues
variant, [p], which was simply a bilabial stop with a release. Because [ps] was
the variant most often mentioned as a typical highland variant, I examined
that variant separately.
As shown in Table 2, for consonant elision (yielding [pwe], [pweh], or [p]),
people from Potos-Oruro (14%) and Carapar (62%) were significantly less
likely to delete the consonant than the Valluno comparison group (87%).
Neither people originally from Iscamayo (95%) nor Cochabamba (94%) were
2014 John Wiley & Sons Ltd

STEREOTYPES VERSUS EXPERIENCE

617

Table 1: Fixed effects for regression analysis social and linguistic factors
Social factors

Place of origin (Iscamayo, Valleys,


Carapar, Cochabamba, Potos-Oruro)
Long-term resident (Yes/No)
Self-identifies as Quechua speaker
(Yes/No)

Linguistic factors

Sentence position (initial,*


medial, final)
Following sound (pause,**
vowel, /s/, consonant
other than /s/)

*Discarded because there were only four cases. This discourse marker typically appears in
medial or final position in this dialect.
**A pause was defined as a >0.5 second space between pues and the following sound.

significantly different from the Valluno comparison group for this data set;
indeed, their rates of elision were virtually identical. Neither length of residence
nor self-identification as a Quechua speaker was found to be significant.
For vowel elision (yielding [ps] or [p]), the results were similar. People from
Potos-Oruro (65%) were significantly more likely to delete the vowel than the
Valluno comparison group (34%). People from Iscamayo (36%), Carapar
(29%), and Cochabamba (33%) did not vary significantly in their usage from
the Valluno group.
Further investigation showed that the [ps] variant was indeed significantly
more likely to be used by people from Potos-Oruro (44%), while people from
Iscamayo never used this variant (0%). However, no significant difference was
found between Valluno (4%), Cochabamba (2%), Carapar (8%), and Iscamayo
(0%). For elision of both the consonant and the vowel [p], once again there was
no significant difference between Valluno (52%) and Cochabamba (47%),
Carapar (37%), or Iscamayo (61%), but people from Potos-Oruro almost
never elided both the consonant and the vowel (2%). The latter finding was
significant and once again there were no significant differences between any of
the other factors. When the table was rotated so that Potos-Oruro was the
reference level, significant differences were observed between Potos-Oruro and
Cochabamba for both vowel elision only ([ps], p<.001) and vowel elision in
general ([p], [ps], p=0.003).
As shown in Table 3, among the linguistic factors, the only significant
variable affecting consonant elision was a following /s/, which strongly
disfavored elision as compared to a following pause (p<0.001). There were no
significant linguistic factors affecting vowel elision. I also examined the data for
interactions between social and linguistic factors and found no significant
interactions.

Cluster analysis
The regression analysis that I carried out showed a clear distinction between
Potos-Oruro and all other groups. Next, I used the statistical software package
2014 John Wiley & Sons Ltd

618

Table 2: Results of regression analysis for social factors

Social factor

Level

Number of
pues tokens

% likelihood
of consonant
elision
(p-value)
(N/A)
(.083)
(.000)*
(.003)
(.272)

% likelihood
of vowel
elision
(p-value)
34
36
65
29
33

(N/A)
(.577)
(.008)*
(.735)
(.959)

% likelihood
of vowel
elision only
(p-value)
4
0
44
8
2

(N/A)
(.998)
(.000)*
(.416)
(.432)

% likelihood
of both consonant
and vowel elision
(p-value)

Valluno
Iscamayo
Potos-Oruro
Carapar
Cochabamba

132
78
55
26
53

87
95
14
62
94

52
61
2
37
47

(N/A)
(.246)
(.001)*
(.315)
(.746)

Long-term
resident

Yes
No

308
36

76 (N/A)
79 (.818)

30 (N/A)
48 (.082)

0 (N/A)
0 (.397)

9 (N/A)
68 (.000)*

Self-identified
Quechua
speaker

Yes
No

149
195

75 (.617)
80 (N/A)

35 (.483)
42 (N/A)

0 (.135)
0 (N/A)

25 (.161)
38 (N/A)

*Statistically significant at p < 0.01.

BABEL

2014 John Wiley & Sons Ltd

Place of origin

STEREOTYPES VERSUS EXPERIENCE

619

Table 3: Results of regression analysis for linguistic factors

Linguistic
factor

Level

Number of
pues tokens

% likelihood
of consonant
elision (p-value)

% likelihood
of vowel
elision
(p-value)

Position

Phrase-final
Phrase-medial

256
100

68 (N/A)
72 (.573)

35 (N/A)
31 (.604)

Following
sound

Consonant not /s/


Pause
/s/
Vowel

123
140
23
70

81
81
35
74

38
26
43
27

(.886)
(N/A)
(.000)*
(.275)

(.051)
(N/A)
(.099)
(.844)

*Statistically significant at p < 0.01.

R to carry out an unsupervised cluster analysis. The purpose of this analysis


was to analyze patterns of vowel and consonant elision in pues with reference
to place of origin groups. The technique used was agglomerative clustering,
which is carried out using the hclust() function in R. This technique allows us
to easily visualize the groupings within the data set based on statistical
similarities between the data points for each group (see Baayen 2008:
118164).
The dendograms in Figures 2 and 3 were produced to visualize the
relationship between different groups of origin according to their tendency to
elide or weaken the consonant (lowland-indexed [pwe] or [pweh]) or to elide
the vowel (highland-indexed [ps]). For both measures, Cochabamba, Iscamayo,
and Valley speakers were closely related, while Potos-Oruro stood apart.
In these dendograms, the y axis indicates the magnitude of the
difference between the place of origin groups. These results support and
illustrate the regression analysis. According to groupings from this data set,
Potos-Oruro always stands as a separate category. Cochabamba, Iscamayo,
and the local valleys, on the other hand, are closely related, though the
configurations of this relationship vary somewhat depending on the factor
selected.

Discussion: Quantitative factors


In their metalinguistic commentary, my consultants clearly grouped
Cochabamba together with Potos-Oruro and Carapar as regions that make
the s whistle. Indeed, this was the most prominent feature associated with
Cochabamba speakers in comparison with local speakers. In contrast, the
quantitative results show that it is Potos-Oruro speakers and to some extent
those from Carapar who show a clear difference from Valley speakers in the
pronunciation of pues. The results differentiating Potos-Oruro from Iscamayo
2014 John Wiley & Sons Ltd

Iscamayo

Cochabamba

Valluno

Carapari

0.6
0.4
0.0

0.2

PotosiOruro

0.8

BABEL
Distance between groups

620

hclust (*, complete)

Figure 2: Grouping by consonant-deletion rates [pwe], [pweh]

Valluno

Cochabamba

Carapari

Iscamayo

PotosiOruro

Distance between groups


0.0
0.2
0.4
0.6

and Valley speakers are robust. Potos-Oruro speakers also stand apart from
Cochabamba speakers, despite metalinguistic commentary linking the two
groups as highlanders. Table 4 summarizes the differences between the
qualitative analysis of language ideologies and the statistical analysis of pues
realizations.
The fact that people hold such strong ideologies linking people from
Cochabamba to the use of /s/ in pues should not be dismissed out of hand,
however. One possible explanation is that permanent, long-term residents of
Iscamayo who were originally from Cochabamba have assimilated to the Santa
Cruz pronunciation, at least in this aspect of their speech. The fact that length
of residence was not selected as a significant factor may be due to the fact that

hclust (*, complete)

Figure 3: Grouping by vowel-deletion-only rates [ps]


2014 John Wiley & Sons Ltd

STEREOTYPES VERSUS EXPERIENCE

621

Table 4: Summary of groupings in qualitative versus quantitative analysis


a. Language ideologies
Highland (pwes], [ps])

Lowland ([pwe], [pweh])

Potos-Oruro
Carapar
Cochabamba

Iscamayo
Valleys

b. Statistical groupings
Group 1 ([pwes], [ps])

Group 2 ([pwe], [pweh])

Potos-Oruro
Carapar [pwes] only

Iscamayo
Valleys
Cochabamba

the great majority of my data come from people who are long-term residents of
the Santa Cruz valleys.
If there were evidence to support this hypothesis, there would still be a
mismatch between the way that my consultants perceive people from
Cochabamba and their personal experience with people from Cochabamba.
Length of residence was not a factor that emerged from the interviews as being
important in the way people spoke. Quite to the contrary, on several occasions
people emphasized to me that people spoke in the manner of the region where
they grew up, no matter where they currently lived. It is important to
remember, too, that the direction of migration is overwhelmingly from the
highlands to the lowlands. Very few of my consultants from the Santa Cruz
valleys spent long periods of time in Cochabamba or other regions of highland
Bolivia. Therefore, their experience with and representations of highland
speech come largely from people who have migrated to and settled in the
Iscamayo area.

DATA III: INTRA-INDIVIDUAL VARIATION ATYPICAL


PRONUNCIATIONS OF PUES
Ideologies that link social groups to particular pues variants erase internal
differentiation among highland groups. However, this is not the only type of
variation that they erase. They also erase the variation that occurs in the
speech of a given individual. In this section, I discuss speakers who employ
atypical pronunciations of the discourse marker pues. These are forms that
are very unusual for these speakers and for their social groups. Through
these examples, I turn to a third perspective on this data the use of pues in
2014 John Wiley & Sons Ltd

622

BABEL

natural conversation as a link between social groups and linguistic


stereotypes.

Methods
The data presented in this section are a subset of the data presented in the
previous section, collected during my 2008 fieldwork. The groups that I focus
on are people from the local valleys surrounding Iscamayo (vallunos) and
highlanders from Potos-Oruro, each of which displayed strong but not
categorical tendencies for the use of these variables.
Because each of these groups has a strong tendency to use their own regional
variant, the total number of examples from which I sampled is relatively small.
Since I am unable to discuss them all individually for reasons of conciseness,
I chose representative examples for detailed discussion. For the Potos-Oruro
group, there were six cases in which people used pues with consonant reduction
or deletion ([pwe] or [pweh]). Three examples are discussed below. For people
from the local valleys, there were 11 cases in which people used a form without
final consonant reduction ([pwes] or [ps]) when not followed by an /s/. I have
chosen four of these cases to examine in detail in the following section. In addition
to being representative examples, these cases were selected because they were
used by speakers who I knew well due to a long-standing relationship and who
I recorded on more than one occasion. Based on my familiarity with these
speakers and their attitudes and habits of speech, I felt that I could make a
confident assessment of their stances in particular situations. Additionally, since
they were well-represented in the recordings, I had a set of pues tokens on which
to base my analysis of their patterns of speech.
I analyze each of these cases in which an atypical pues occurred in detail, in
context, as part of a transcript and in the context of the conversation. I also
present the total distribution of realizations of pues for each speaker and include
ethnographic information and background on the speaker and the topic. While
regional groups have a particular profile in terms of the way that they use pues,
individual speakers may more or less closely resemble their group. In Table 5,
I give the distribution of pues forms for the four speakers whose speech
I examine closely in this section.

Potos-Oruro
Sara and Pedro, a married couple, were both teachers in the local school
district. Pedro was one of the most educated people in town and had served as
the director of the school district. Sara taught Quechua and Spanish at various
levels in the local schools. Both were originally from small towns in the PotosOruro highlands.
Sara used the [p] variant of pues twice. The two instances of Saras use of this
form occurred one after the other as she discussed language mixture in
Iscamayo.
2014 John Wiley & Sons Ltd

STEREOTYPES VERSUS EXPERIENCE

623

Table 5: Phonetic realizations of pues by individual speakers (P-O = Potos-Oruro; V


= Valluno)

Speaker
(place of
origin)

Vowel
elision
[ps]

Neither
[pwes]

Consonant
elision
[pwe],
[pweh]

Both
[p]

Total

Sara (P-O) 13
Pedro (P-O) 2
Prima (V)
3
Marina (V)
0

72
20
10
0

2
7
4
2

11
70
13
14

1
1
14
8

6
10
47
57

2
0
9
4

11
0
30
29

18
10
30
14

100
100
100
100

Transcript 1 (S = Sara, A = Author, P = Pedro)


1.

S:

2.

3.

4.

5.
6.
7.
8.
9.

A:
S:
P:

Ellos piensan que es todo


Quechua.
Mejor dicho, que es
castellano. No est
an
hablando Quechua.
Ahora le vas a preguntar
aqu que est
an hablando
ellos, le escuchas?
Eso est
a en castellano,
no es en Quechua,
le dices.
Ellos te van a decir
que es castellano.
Pero no es pues [p].
No es
No es pues [p].
No es

1.

S:

They think its all Quechua.

2.

I mean, Spanish. Theyre not


speaking Quechua.

3.

Now, if you ask them what


they are speaking, do you
hear them?
This is in Spanish, not in
Quechua, you tell them.

4.

5.
6.
7.
8.
9.

They will say its in Spanish.

A:
S:
P:

But its not pues [p].


Its not.
Its not pues [p].
Its not.

In this transcript, Sara claimed that people from Iscamayo think theyre
speaking pure Spanish, when in fact they mix Spanish with Quechua. As she
asserted Its not [pure Spanish], she added an emphatic [p].
The lone instance of consonant reduction in her husbands speech also
occurred in this context, as we discussed the mixture of Spanish and Quechua
in the local area.

2014 John Wiley & Sons Ltd

624

BABEL

Transcript 2 (A = Author, P = Pedro, S = Sara)


1. A: Y en, en Quechua Usted
ha notado que se mezcla
tambien? Con castellano?
2. P: Si. Tambien mezclan.
3. S: Mezclan, [grave].
4. P: En Quechua mismo,
hablan muchas palabras
mezclado con el castellano
pues [pweh], no?
5.
Ese es el objectivo de la
Reforma Educativa es
pues [pwes], eh, hablar
bien el
6. S: Puro Quechua
7. P: Quechua, limpio, o sea
hablar puro.

1. A: And, in Quechua have you found


that people mix too? With
Spanish?
2. P: Yes. They mix too.
3. S: They mix, [terribly].
4. P: Even in Quechua, they talk with
many words mixed with Spanish
pues [pweh], no?
5.

That is the goal of the Educational


Reform, its pues [pwes], um, to
speak better

6. S:
7. P:

Pure Quechua
Clean Quechua, that is, to speak
purely.

In these transcripts, Pedro and Sara discussed the mixture of Quechua with
Spanish, which they evaluated negatively. While discussing the local mixture of
Spanish with Quechua, Pedro used a variant of pues with consonant elision
[pweh] immediately after the word Spanish. In the following sentence, speaking
of the Educational Reform act, he used the unreduced [pwes] that was typical of
his speech. This phonetic alternation mirrored the contrast that he presented
between clean or pure language and the mixed-up language of the local valleys.
Sara and Pedro used variants with consonant elision variants that were rare
both for them personally and for their social group as they talked about people
or places that were associated with Santa Cruz or the Santa Cruz valleys, both
areas in which pues with consonant elision is common. Other examples of
consonant reduction and elision used by speakers from the Potos-Oruro region
shared this tendency.

Vallunos
Marina and Prima were speakers from the local valleys who used pues without
final consonant reduction or elision. Both were women I knew well and had a
social relationship with. Both women worked with their husbands in
agriculture and had relatively little formal education.
I recorded Marina several times, including a conversation with her family,
an interview about her experience as a local politician, and a language
ideologies interview, as well as some community meetings in which she was a
participant. Marina was in her thirties at the time of these recordings. In the
conversation represented in Transcript 3, she discussed her ideas about the
differences between highlanders and lowlanders.
2014 John Wiley & Sons Ltd

STEREOTYPES VERSUS EXPERIENCE

625

Transcript 3 (M = Marina, A = Author)


1. M: Porque los, cambas tienen 1. M:
cabalito para el da,
tienen que tener.
2. A: Mhm
2. A:
3. M: Mientras los collas en
3. M:
cambio, es, pues
[pweh], a ellos les
gusta trabajar, y
ahorrarse y tener
algo pues [pwes],
no ve? Trabajar,
pues [p]. Y los
cambas en
cuanto no tengan
nada. Est
an siguen
felices.

Because the, lowlanders have just


enough for the day, they need to
have.
Mm-hmm
While the highlanders on the
other hand, its, pues [pweh],
they like to work, and save and
have something pues [pwes],
right? Work, pues [p]. And the
lowlanders even when they
have nothing. Theyre still happy.

In this discussion of differences between highlanders and lowlanders, Marina


voiced common stereotypes lowlanders are lazy, while highlanders are hardworking. In turn 3 of the transcript, Marina used three pues variants in a row,
perhaps an allusion to the perception that highlanders use the discourse
marker more. One of these variants, following her main idea a ellos les gusta
trabajar y ahorrarse y tener algo they like to work and save and have something
was unreduced [pwes]. In the same turn, she used a grammatical construction
that is typical of Quechua-dominant speakers, estan siguen felices theyre still
happy. This construction is highly ungrammatical in most Spanish varieties,
and is not common among Spanish-dominant speakers in this area. This use of
a highland pues variant and a highland-linked grammatical structure in a
discussion of highland groups mirrors the way that highlanders from PotosOruro used lowland variants when discussing topics related to the lowlands.
This is not the only way that Marina used pues with final consonant
retention, however. In Transcript 4, in a recording made on a
different occasion, Marina used highland-indexed pues in order to express
emphasis or exasperation. In this recording, she was trying to comb nits from
her daughters hair and was having trouble getting the right angle and light. It
was an informal occasion, and several family members were gathered around.
This exchange occurred between Marina and her young daughter, C.
Transcript 4 (C = Marinas daughter, M = Marina)
1. C:

Ay, no ponga su pie.

2014 John Wiley & Sons Ltd

1. C:

Hey, dont put your foot [there].

626

2. M: Es muy abajo pues


2. M:
[pweh], no veo!
3. C: Su pie, pues [pwej] . . .
3. C:
Est
a su pie muy abajo?
4. M: Tu cabeza, pues [pwes]. 4. M:

BABEL

Its too low pues [pweh],


I cant see!
Your foot, pues [pwej] . . . Is your
foot too low?
Your head, pues [pwes].

In this exchange, Marina was exasperated by her daughters wiggling and


uncooperative attitude, and she had just lost her temper and raised her voice at
her (turn 2). The little girl continued to protest in a whiny tone of voice. Pues,
in this exchange, was used as part of a construction of emphasis, as Marina
responded to her daughters confusion about what was too low (her mothers
foot or her own head). In turn 4, using more affectionate tone of voice, but still
a bit shortly, she explained that its the childs head, not her own foot, that was
too low. It was on this repetition that Marina used [pwes]. This response, along
with her tone of voice and body language, communicated an assertive and
rather impatient stance in response to her daughters protests.
This assertive stance contrasts with the ambivalent stance that Marina
expressed when I asked her about her election to the town council. In Transcript
5, Marina discussed her reluctance to participate in local politics given her lack of
formal education. She eventually agreed because, she told me, she liked to help
people and she felt that a person like herself would do a better job than a rich
person who didnt understand the needs of the community. In Transcript 5, she
told me about having been talked into running for the position.
Transcript 5 (M = Marina, A = Author)
1. M: Yo, pues [pweh], casi no quera 1. M: I, pues [pweh], really
aceptar. Aqu me vinieron, a
didnt want to accept.
pedir, digamos, la comunidad,
They came to me, you
la, unas cuantas personas de
know, the community
la comunidad, me pidieron.
asked me, the, a few people
from the community, they
asked me.
2. A: Mhm
2. A: Mm-hmm
...
...
3. M: As yo acepte estar en la lista.
3. M: So I agreed to be on the
Pas
o un tiempo, ya, me
ballot. Then some time
dijeron, pues [p], no hay mas
passed, then, they told me
opcion, que tienes que entrar.
pues [p], theres no other
Yo no quera, casi no me
alternative, you have to
animaba, porque no se,
take office. I didnt want to,
pues [p].
I didnt feel like it, because
I dont know, pues [p].

A similar pattern can be seen in Primas uses of pues. I recorded two interviews
with Prima, one on the topic of language ideologies and one of the topic of
2014 John Wiley & Sons Ltd

STEREOTYPES VERSUS EXPERIENCE

627

cooking practices. In Transcript 6, taken from the cooking practices interview,


we discussed the scarcity of firewood. Then I asked Prima how she thought the
mountains where she went to gather firewood looked, expecting her to say that
they were bare or dead-looking due to over-collection of firewood. Instead, she
responded that they looked fine to her. After a five-second pause while I wrote
down her responses, I returned to one of her earlier statements, confirming that
her experience was that there was a scarcity of dead wood for firewood.
In her response, Prima responded to my implicit challenge of her answer by
strongly re-asserting her original position, first using pues with diphthong
elision only ([ps], turn 2) and then, as she modified her answer, pues with
consonant elision.
Transcript 6 (A = Author, P = Prima)
1. A: Pero dice que no hay le~
na ya? 1. A: But you say theres no
firewood anymore?
2. P: Ya no hay le~
na ya pues [ps]. 2. P: Theres no firewood anymore
pues [ps].
3.
Hay le~
na, pero muy lejos ya
3.
There is firewood, but its very
pues [pweh].
far away now pues [pweh].

I understood Primas use of [ps] to indicate emphasis or perhaps annoyance


at being asked to repeat her answer a strong, assertive response. The use of
[ps] indexed a more authoritative, less deferential stance to my repeated
question. The discursive effect of shifting back to [pweh] was to move from a
strong assertion to a more polite explanation of her answer.
The language ideologies interview that I conducted with Prima and her
husband returned to the topic of English again and again. Prima emphasized to
me how much she had wanted her children to study the language and perhaps
even become English teachers. The following excerpts are drawn from several
minutes of recording.
Transcript 7 (J = Primas husband, P = Prima)
1. J:

Ojal
a, pues [pwe], que pudieran 1. J: I wish, pues [pwe], that
aprender ellos el ingles.. . .
they could learn English.. . .
2. P: Por lo menos que entiendan,
2. P: At least understand [it],
pues [p], que fuerza que hablen,
pues [p], it doesnt matter
porque hace mucha falta
if they speak, because its
[no ve].. . .
so necessary [you know]. . . .
3. P: Porque ellas [mis hijas] tambien 3. P: Because they [my daughters]
han estudiado ingles pues
also studied English pues
[pwe].. . .
[pwe].. . .

2014 John Wiley & Sons Ltd

628

4. P: Hay, pues [pwe], [una


oportunidad] pa salir y, ser,
profesora de ingles estan,
no ve?

BABEL

4. P: There is, pues [pwe], [an


opportunity] to graduate
and be, an English teacher,
isnt there?

In these short excerpts, Prima and her husband constructed aspirational


stances towards higher education and towards English, a language that stands
for social status and upward mobility to many Bolivians. In doing so, they often
used pues as they expressed their sincere desire to motivate their children to study
English. In this conversation, Prima and her husband used the local versions of
pues ([pwe] and [p]). In contrast to Primas assertive stances in Transcript 6, in
this conversation she did not use the highland-indexed pues variants.

Discussion: Atypical uses of pues


People from Potos-Oruro used lowland pues variants to index topics associated
with the lowlands (Transcripts 1, 2). Likewise, Vallunos used highland pues
variants in order to index topics associated with the highlands (Transcript 3).
However, Vallunos also used pues in making strong assertions or projecting
firm stances, even when they were not explicitly discussing highlanders
(Transcripts 4, 6). This is perfectly in line with the semantic and pragmatic
functions of pues as a discourse marker, which include the construction of
assertive stances (e.g. Zavala 2001) or conveying emphasis (e.g. Pf
ander et al.
2009:126130). However, it contrasts with the local phonetic variant of pues
that the same speakers used when constructing ambivalent or aspirational
stances (Transcripts 5, 7).
Beyond the status of pues as a discourse marker, there is meaning attached
to the phonetic variants that speakers used in these transcripts. Speakers use
pronunciations of pues that are unusual for them to index places and people
who are associated with that variant. When people from the valleys construct
particularly assertive stances, they use highland varieties of pues.
These atypical pronunciations of pues are anything but accidental. Rather, they
are revealing of the way that people think about language as a symbolic resource.
Speakers draw on and produce the indexical field surrounding pues as they make
these references, linking linguistic form to geographical region and to stereotypes
about people from that region. The patterns of use of pues with and without
reduction correspond to different levels of indexical reference. The speakers most
common or unmarked variant of pues alternates with socially meaningful
instances in which they use variants that index aspects of another social group.

CONCLUSION
In this article, I have examined phonetic variation in a discourse marker, pues, that
is a highly stereotyped index of regional identification. People in the Santa Cruz
2014 John Wiley & Sons Ltd

STEREOTYPES VERSUS EXPERIENCE

629

valleys of Central Bolivia believe that highlanders pronounce pues as [ps] or [pwes],
while lowlanders pronounce pues as [pwe] or [pweh]. However, these stereotypes
erase an internal distinction between highlanders. Natural production data shows
that highlanders from the Potos-Oruro area do generally pronounce pues as they
are believed to. However, immigrants from the Cochabamba valleys, the most
commonly mentioned group in interview data, differ significantly from people
from Potos-Oruro and closely mirror the pronunciation of speakers from the Santa
Cruz valleys. Peoples beliefs about the way that regional identity groups use the
discourse marker pues do not match up with the way that members of these groups
use the discourse marker in my corpus.
The stereotype of the highland [ps] or [pwes] pronunciation also erases
differences within individual speakers. It is true that people from Potos-Oruro
have a strong tendency to maintain the consonant and elide the diphthong,
while people from the Santa Cruz valleys tend to elide or reduce the consonant.
However, these tendencies are not categorical. In cases where people from
Potos-Oruro use the lowland-associated consonant reduction or elision, they
are referring to people, places, or practices associated with Santa Cruz. When
people from the Santa Cruz valleys use the highland variants, they refer to
people from the highlands, but also project assertive stances that reference
stereotypes of highlanders as pushy, bossy, or aggressive. Atypical uses of pues
are not used by chance or at random; they are meaningful acts that are
imbedded in a particular social system.
This study contributes to the study of indexical reference and indexical fields by
examining the links that people believe exist between particular phonetic variants
and social groups. These beliefs shape the way that people use and interpret
language. However, beliefs about language are not necessarily faithful reflections
of experience with language. Previous studies have shown that the ideas that
people hold about social groups can influence the way that they perceive
linguistic variables. The present research shows that they can also influence the
way that people use language in natural contexts to refer to social groups and the
qualities that are associated with them. The data I have presented here show that
people construct higher levels of indexical reference based on their stereotypes
about language use rather than their experience with speakers.
Most studies of awareness and control have focused either on speech
perception, using experimental methods, or on explicit stereotypes that speakers
hold. This work bridges experimental studies of speech perception and qualitative
investigations of attitudes and stereotypes. Previous research has shown that
awareness and control of linguistic variables works on many levels and in many
dimensions. This study connects three aspects of awareness and control:

what people say about language when they are discussing stereotypes;
what people do when they produce language in natural contexts; and
what people are referring to when they use variables in ways that are
unusual for their group.

2014 John Wiley & Sons Ltd

630

BABEL

Through an examination of these three aspects of language use, this study


offers an evaluation of the relationship between beliefs about highly
stereotyped variables and experience with language use in natural contexts.
Not all highlanders say ps, but the fact that people believe they do affects the
way that people understand and use this variable.

NOTES
1. I gratefully acknowledge the help that I received in developing this article. As
always, thanks go first and foremost to my consultants in Iscamayo for their
great patience and generosity in sharing their lives with me. My very sincere
thanks to two anonymous reviewers for their comments on an initial version of
the manuscript, and to the editors for their vote of confidence on the initial
submission. I am indebted to Kathryn Campbell-Kibler and Kevin McGowan for
giving me detailed comments on a later draft. In the quantitative data section,
Steven Naber of the OSU Statistical Consulting Service assisted me with the
regression analysis, Lauren Squires helped me to think through my approach to
the data, and Kevin McGowan produced Figures 1 and 2. My very, very
appreciative thanks to Shukri Zanika, my research assistant, for the many, many
pues tokens that he painstakingly clipped, coded, and analyzed, and especially for
his willingness to disagree with me. Thanks to Karen Lopez Alonso for her
assistance with the Spanish abstract. Parts of this research were funded by an
NSF Graduate Research Fellowship, the Rackham Graduate School of the
University of Michigan, the College of Arts and Sciences of the Ohio State
University, and the Freiburg Institute for Advanced Studies.
2. It is not entirely clear from these articles whether pues would (or would always)
count as an unstressed vowel. As a monosyllabic word, it would seem that the
vowel might generally be considered to be stressed. Regardless, it is identified as
a prototypical environment for vowel reduction by these authors.
3. As mandated by the IRB, and following my practice in other published articles,
I use pseudonyms for places that are not major cities and for people who are not
public figures.
4. Orthographic J is pronounced as [h] or [x] in Latin American Spanish.
5. This conversation is quoted at greater length in Transcript 3 (p. 625).
6. The original transcriptions were orthographic in nature and were not meant to
be phonetically accurate. As it turned out, my impressionistic transcriptions had
a low degree of phonetic accuracy, so the sample is perhaps a better (in the sense
of more random) one than it might have been.

REFERENCES
Alba, Orlando. 1982. Funci
on del acento en el proceso de elisi
on se la /s/ en la Rep
ublica
Dominicana. Paper presented to the El espa~
nol del Caribe: Ponencias del VI Simposio de
Dialectologa, Santiago, Dominican Republic.
Baayen, R Harald. 2008. Analyzing Linguistic Data: A Practical Introduction to Statistics Using R.
New York/Cambridge, U.K.: Cambridge University Press.
2014 John Wiley & Sons Ltd

STEREOTYPES VERSUS EXPERIENCE

631

Babel, Anna M. 2010. Contact and contrast in Valley Spanish. Unpublished PhD dissertation.
Ann Arbor, Michigan: University of Michigan.
Babel, Anna M. To appear-a. The role of context in interpreting linguistic variables. Boletn de
Filologa de la Universidad de Chile XLXIX.
Babel, Anna M. To appear-b. Silence as control: Shame and self-consciousness in
sociolinguistic positioning. In Anna M. Babel (ed.) Awareness and Control in Sociolinguistic
Research. Cambridge, U.K.: Cambridge University Press.
Beskow, Jonas and Kares Sj
olander. 2011. WaveSurfer, version 1.8.8. Software available at
http://sourceforge.net/projects/wavesurfer/
Brown, Earl K. 2009a. The relative importance of lexical frequency in syllable- and word-final
/s/ reduction in Cali, Colombia. In Joseph Collentine, Maryellen Garcia, Barbara Lafford and
Francisco Marcos Marin (eds.) Selected Proceedings of the 11th Hispanic Linguistics
Symposium. Somerville, Massachusetts: Cascadilla Proceedings Project. 165178.
Brown, Earl K. 2009b. A Usage-Based Account of Syllable- and Word-Final /s/ Reduction in Four
Dialects of Spanish. Munich, Germany: LINCOM.
Brown, Esther L. 2004. The reduction of syllable-initial /s/ in the Spanish of New Mexico and
southern Colorado: A usage-based approach. Unpublished PhD dissertation. Albuquerque,
New Mexico: University of New Mexico.
Brown, Esther L. and Rena Torres Cacoullos. 2002. Que le vamoh aher? Taking the syllable out of
Spanish /s/ reduction. University of Pennsylvania Working Papers in Linguistics 8: Article 3.
Brown, Esther L. and Rena Torres Cacoullos. 2003. Spanish /s/: A different story from
~ ez-Cede~
beginning (initial) to end (final). In Rafael N
un
no, Luis L
opez and Richard Cameron
(eds.) A Romance Perspective on Language Knowledge and Use: Selected Papers from the 31st
Linguistic Symposium on Romance Languages (LSRL). Chicago, Illinois: John Benjamins
Publishing Company. 2138.
Bucholtz, Mary and Kira Hall. 2008. All of the above: New coalitions in sociocultural
linguistics. Journal of Sociolinguistics 12: 401431.
Calvo Perez, Julio. 2000. Partculas en el castellano andino. In Julio Calvo Perez (ed.) Teora y
practica del contacto: El espa~nol en America en el candelero. Madrid, Spain: Iberoamericana /
Vervuert. 73112.
Campbell-Kibler, Kathryn. 2008. Ill be the judge of that: Diversity in social perceptions of
(ING). Language in Society 37: 637659.
Carmichael, Katie. To appear. Place-linked expectations and listener awareness of regional
accents. In Anna M. Babel (ed.) Awareness and Control in Sociolinguistic Research. Cambridge,
U.K.: Cambridge University Press.
Choksi, Nishaant and Barbra A. Meek. To appear. The limits of salience: Orthographic practice
and the enfigurement of minority languages. In Anna M. Babel (ed.) Awareness and Control
in Sociolinguistic Research. Cambridge, U.K.: Cambridge University Press.
Coupland, Nikolas and Adam Jaworski. 2004. Sociolinguistic perspectives on metalanguage:
Reflexivity, evaluation, and ideology. In Adam Jaworski, Nikolas Coupland and Dariusz
Galasisnski (eds.) Metalanguage: Social and Ideological Perspectives. Berlin, Germany: Mouton
de Gruyter. 1552.
Delforge, Ann Marie. 2008. Unstressed vowel reduction in Andean Spanish. In Laura
Colantoni and Jeffrey Steele (eds.) Selected Proceedings of the 3rd Conference on Approaches to
Spanish Phonology. Somerville, Massachusetts: Cascadilla Proceedings Project. 107124.
Drager, Katie and Joelle Kirtley. To appear. Awareness, salience, and stereotypes in exemplarbased models of speech production and perception. In Anna M. Babel (ed.) Awareness and
Control in Sociolinguistic Research. Cambridge, U.K.: Cambridge University Press.
2014 John Wiley & Sons Ltd

632

BABEL

Eckert, Penelope. 2008. Variation and the indexical field. Journal of Sociolinguistics 12:
453476.
Eckert, Penelope. 2012. Three waves of variation study: The emergence of meaning in the
study of sociolinguistic variation. Annual Review of Anthropology 41: 87100.
File-Muriel, Richard J. 2009. The role of lexical frequency in the weakening of syllable-final
lexical /s/ in the Spanish of Barranquilla, Colombia. Hispania 92: 348360.
Garrett, Peter. 2001. Language attitudes and sociolinguistics. Journal of Sociolinguistics 5:
626631.
Garrett, Peter. 2010. Attitudes to Language: Cambridge, U.K.: Cambridge University Press.
Garrett, Peter, Nikolas Coupland and Angie Williams. 2003. Investigating Language Attitudes:
Social Meanings of Dialect, Ethnicity and Performance. Cardiff, U.K.: University of Wales Press.
Grajales Alzate, R
obinson. 2011. Functions of the discourse marker pues in the speech of
Medelln, Colombia. Forma y Funcion 24: 2545.
Hay, Jennifer, Paul Warren and Katie Drager. 2006. Factors influencing speech perception in
the context of a merger-in-progress. Journal of Phonetics 34: 458484.
Irvine, Judith T. and Susan Gal. 2000. Language ideology and linguistic differentiation. In
Paul V. Kroskrity (ed.) Regimes of Language. Santa Fe, New Mexico: School of American
Research Press. 3583.
Johnson, Keith. 2006. Resonance in an exemplar-based lexicon: The emergence of social
identity and phonology. Journal of Phonetics 34: 485499.
Kiesling, Scott Fabius. 1998. Mens identities and sociolinguistic variation: The case of
fraternity men. Journal of Sociolinguistics 2: 6999.
Labov, William. 1972. Sociolinguistic Patterns. Philadelphia, Pennsylvania: University of
Pennsylvania Press.
Lipski, John M. 1984. On the weakening of /s/ in Latin American Spanish. Zeitschrift f
ur
Dialektologie und Linguistik 51: 3143.
Lipski, John M. 1990. Aspects of Ecuadorian vowel reduction. Hispanic Linguistics 4: 119.
Lope Blanch, Juan M. 1983. Analisis gramatical del discurso. Mexico City, Mexico: Universidad
Nacional Aut
onoma de Mexico.
McGowan, Kevin B. 2011. The role of socioindexical expectation in speech perception.
Unpublished PhD dissertation. Ann Arbor, Michigan: University of Michigan.
McGowan, Kevin B. To appear. Sounding Chinese and listening Chinese: Imitation, perception,
and awareness of non-native phonology. In Anna M. Babel (ed.) Awareness and Control in
Sociolinguistic Research. Cambridge, U.K.: Cambridge University Press.
Minnick-Fox, Michelle. 2006. Usage-Based Effects in Latin American Spanish Syllable-Final /s/
Lenition. Philadelphia, Pennsylvania: University of Pennsylvania.
Munson, Ben. 2007. The acoustic correlates of perceived sexual orientation, perceived
masculinity, and perceived femininity. Language and Speech 50: 125142.
Niedzielski, Nancy. 1999. The effect of social information on the perception of sociolinguistic
variables. Journal of Language and Social Psychology 18: 6285.
Niedzielski, Nancy A. and Dennis Richard Preston. 2003. Folk Linguistics. Berlin, Germany/
New York: Walter de Gruyter.
Olbertz, Hella. 2013. Pues en el espa~
nol rural de la sierra ecuatoriana: Interferencia del
quichua? In Christine Felbeck, Andre Klump and Johannes Kramer (eds.) America Romana:
Perspektiven transarealer Vernetzungen. Frankfurt, Germany: Peter Lang. 179204.
Pf
ander, Stefan, Juan Ennis, Mario Soto Rodrigues and Espa~
na Villegas Pinto. 2009. Gramatica
Mestiza: Presencia del quechua en el castellano boliviano. La Paz, Bolivia: Academia Boliviana
de la Lengua and Editorial Signo.
2014 John Wiley & Sons Ltd

STEREOTYPES VERSUS EXPERIENCE

633

Portoles, Jose. 1989. El conector argumentativo pues. Dicenda: Cuadernos de Filologa Hispanica
8: 117133.
Preston, Dennis. 1996. Whaddayaknow: The modes of folk linguistic awareness. Language
Awareness 5: 4074.
Preston, Dennis. To appear. Whaddayaknow now? In Anna M. Babel (ed.) Awareness and
Control in Sociolinguistic Research. Cambridge, U.K.: Cambridge University Press.
Rubin, Donald L. 1992. Nonlanguage factors affecting undergraduates judgments of
nonnative English-speaking teaching assistants. Research in Higher Education 33: 511531.
Serrano, Julio. 2006. En torno a las vocales caedizas del espa~
nol mexicano: Una aproximaci
on
socioling
ustica. Los lderes lingusticos. Estudios de variacion y cambio. 3: 759.
Serrano, Mara Jose. 1997. Marcadores discursivos en espa~
nol: Acerca de la verdad y pues.
Boletn de Filologa de la Universidad de Chile 36: 265286.
Silverstein, Michael. 1979. Language structure and linguistic ideology. In Paul R. Clyne,
William F. Hanks and Carol L. Hofbauer (eds.) The Elements: A Parasession on Linguistic Units
and Levels. Chicago, Illinois: Chicago Linguistic Society. 193195.
Silverstein, Michael. 2003. Indexical order and the dialectics of sociolinguistic life. Language
and Communication 23: 193229.
Squires, Lauren. To appear. Processing grammatical differences: Perceiving versus noticing. In
Anna M. Babel (ed.) Awareness and Control in Sociolinguistic Research. Cambridge, U.K.:
Cambridge University Press.
Sumner, Meghan, Seung Kyung Kim, Ed King and Kevin B. McGowan. 2013. The socially
weighted encoding of spoken words: A dual-route approach to speech perception. Frontiers
in Psychology 4. doi: 10.3389/fpsyg.2013.01015
Terrell, Tracy D. 1979. Final /s/ in Cuban Spanish. Hispania 62: 599612.
Travis, Catherine E. 2005. Discourse Markers in Colombian Spanish: A Study in Polysemy. Berlin,
Germany/New York: Walter de Gruyter.
Woolard, Kathryn. 1998. Language ideology as a field of inquiry. In Bambi B. Schieffelin,
Kathryn A. Woolard and Paul V. Kroskrity (eds.) Language Ideologies: Practice and Theory.
New York: Oxford University Press. 347.
Zavala, Virginia. 2001. Borrowing evidential functions from Quechua. Pragmatics 33:
9991023.
Zimman, Lal. To appear. Sociolinguistic agency and the gendered voice: Metalinguistic
negotiations of vocal masculinization among female-to-male transgender speakers. In Anna
M. Babel (ed.) Awareness and Control in Sociolinguistic Research. Cambridge, U.K.: Cambridge
University Press.

Address correspondence to:


Anna M. Babel
The Ohio State University Spanish and Portuguese
298 Hagerty Hall
1775 College Rd
Columbus, Ohio 43210
U.S.A.
babel.6@osu.edu
2014 John Wiley & Sons Ltd

Anda mungkin juga menyukai