Interpreting the Distribution of Mixed Use of Pronominal Address Terms in Colombia using Twitter

(1)

1

Interpreting the Distribution of Mixed Use

of Pronominal Address Terms in Colombia

using Twitter

A thesis submitted to the Faculty of Humanities in partial fulfilment of the requirements of the degree Master of Arts in the Linguistics of European Languages - Spanish at The University of Amsterdam

Daniel Foster 10847251

Word Count: 20,159 Advisor: S. P. Aalberse March 14, 2016

(2)

2

Acknowledgements

I would like to express my acknowledgments and gratitude to all those who have helped and

supported me along the way while completing this thesis. First and foremost, I want to thank Suzanne Aalberse for agreeing to advise my thesis. Your feedback has been extremely insightful and useful and I have enjoyed discussing the project thoroughly along the way. Furthermore, your constant motivation to make the most out of this thesis has been invaluable and I’ve gotten more out of the experience than I ever thought possible. Also to Otto Zwartjes as my second reader and for your guidance since my first weeks here at the UvA. To Horst Simon at the Freie Universität Berlin and to the Grammar & Cognition Group at the UvA for allowing me to come present my work and offering discussion and insight that greatly aided the project. To all my friends and family back home and all the new friends I’ve met in Amsterdam over the past year and a half who have offered unending support, in person, via messages, or over the phone, throughout my time working on the project. I never would have reached this point without your encouragement. Finally, a special thanks to everyone who found themselves living with me on Grote Bickersstraat last year for making Amsterdam an unforgettable place.

(3)

3

The present study aims to interpret the distribution of pronominal address terms on Twitter in order to understand the ongoing process of change in Colombian Spanish. Twitter provides an exciting new source for address research due to the advent of powerful tools such as streamR (Barberá 2013). Colombia presents a unique situation in that certain areas make use of all three Spanish address terms and their verbal paradigms, usted, tú, and vos, in mixed and, many times, overlapping social situations (Millán 2011: 2). A hierarchal favoring of emotional closeness was found to condition address term choice over other factors. Special attention was paid to mixed forms and mixed use, which revealed a possible predilection for the verbal paradigm of tú.

(6)

6

1. Introduction

1.1 Pronominal address terms in Colombian Spanish

One of the most consistently cited variables used to distinguish Latin American varieties of Spanish from one another and from peninsular varieties is the second person pronominal address term system (Campos 2010, Benavides 2003, Penny 2000). In a Latin American context, the most outstanding facet of this system is voseo, the use of the second person singular pronoun vos along with its verbal paradigm (as opposed to its siblings tuteo and

ustedeo); its journey to Latin America has a complicated history which has resulted in what

one could call an even more complicated present. Historically, Spanish on the peninsula had a two element system of tú for familiarity and vos for distance (these forms had survived from Latin in the second person singular, while the addition of otros made vosotros the option for second person plural). Around the 16th century vos had lost its polite status, and its use was being replaced completely by tú on the peninsula while vuestra merced (later derived into

usted) became the polite form (Uber 1985: 388). This change was completed after

colonization of the Americas had begun, though, and vos survived throughout much of Latin America as a familiar form, especially in those areas with weaker connections to the

peninsula (Benavides 2003). This resulted in a complicated landscape for voseo in Latin America, with some areas (Río de la Plata, Central America (except Panamá), Bolivia, Ecuador, and Chile (with its own idiosyncrasy)) adopting it widely, and others only

regionally to varying degrees (México, Panamá, Colombia, Venezuela, Perú, Cuba, Puerto Rico, and The Dominican Republic) (Benavides 2003: 617). Perhaps in none of these varieties of Spanish, however, is voseo’s place in the second person singular pronominal landscape more complex than in Colombia due to the level of dialectal variation and the resulting diverse set of meanings it experiences (Millán 2011, Diáz Collazos 2015). There are three common formal and informal second person singular pronouns in Colombia: the

informal tú and vos, the sometimes formal and sometimes informal usted, alongside two less common forms of address found in specific cities, the formal vusté and sumercé (in areas such as Boyacá). Some areas in Colombia, such as Antioquia and Valle del Cauca, make use of a tripartite system of usted, tú, and vos, all with diverse and overlapping social boundaries, a fascinating paradigm (Millán 2011: 2). Usted’s occurrence in informal situations is referred to as a somewhat paradoxical polysemy because of its usual formal status (see (1)b.ii) (Diáz Collazos 2015: 12). However, while vos and tú present the most common competition for

(7)

7

informal forms of address in the country, usted is also very much in the mix (Millán 2011). Each pronoun in this tripartite system, of course, has its own context depending on the dialect in question. Campos (2010) describes the situation generally along these lines:

(1)

a. The Atlantic coast region of Colombia is tuteante meaning the 2nd_person

pronominal system consists of tú for familiarity and usted for situations of distance.

b. Bogotá, the capital region, while long considered a mix of voseante and

tuteante has been shown to be more purely tuteante by recent studies

(Placencia 2010), but has its own particular pattern conditioned especially by gender according to Bartens (2004):

i. Women use a normal two element system of tú for familiarity and usted for distance;

ii. Men, though, use a three element system of (usted / tú / usted). The first

usted is used for solidarity and familiarity, mostly with men of the same

age and social status, and sometimes with women to emphasize familiarity. Tú, meanwhile, is used as an intermediate between the familiar usted and the second usted of distance (Campos 2010: 225-6, Placencia 2010: 352).

c. The rest of the country can be considered voseante/tuteante, according to Campos, using both vos and tú for familiarity and using usted for distance. The region from the Pacific to the Cauca River (including Medellín and Cali) was previously considered purely voseante using a two element system of vos for familiarity and usted for distance, but has recently been shown to use tú as well (Campos 2010: 230). Murrillo Fernández (2004) and Millán (2011) comment on tú as having a strong prestige because of its association with the Bogotá dialect, suggesting that the spectrum is something approaching (vos / tú / usted)

vos being the most familiar and usted being the most distant. As Campos notes,

though, “others prefer to describe two juxtaposed bidimensional systems (vos/usted; tú/usted)” (2010: 230).

As one can see, some of the lines here are not so clearly defined, especially with tú and vos in (1)c and the two uses of usted with tú in (1)b.ii. Indeed, perhaps related to this, mixing of the forms has been shown to occur between the same interlocutors (Moreno 2010, Millán 2011,

(8)

8

Murrillo Fernández 2004). Furthermore, Spanish pronoun use presents an interesting additional aspect to such mixing with a separate verb inflection associated with each of the three pronouns in question. Verbs associated with tú and vos carry their own inflectional system and some of their own non-nominative pronouns. Each paradigm is shown below, exemplified with two primary verb classes, a regular verb and an irregular “stem-changing” verb, meaning in tuteo the root vowel goes through diphthongization as compared to the infinitive. Verbs in voseo are all regular except for ‘to be’. Usted, meanwhile, carries the verbal paradigm of the third person singular.

(2) Nominative pronoun Accusative pronoun Dative pronoun Possessive pronoun Prepositional pronoun Comitative pronoun

usted la, lo, se le su usted con usted

tú te te tu ti contigo

vos te te tu vos con vos

Regular Verb Tuteo Irregular Stem Changing Verb Tuteo

Saber Singular Plural

1 sé sabemos

2 sabes saben

3 sabe saben

Regular Verb Voseo Irregular Stem Changing Verb Voseo

Saber Singular Plural

1 sé sabemos

2 sabés saben

3 sabe saben

This means mixing does not only occur on the pronominal level, but also the inflectional level. In Colombia, Díaz Collazos (2015) takes into consideration one case of this inflectional mixing from a 19th century play where we see the tú pronominal form with the vos

inflectional form (tenés instead of tienes):

Tener Singular Plural

1 tengo tenemos

2 tienes tienen

3 tiene tienen

Tener Singular Plural

1 tengo tenemos

2 tenés tienen

(9)

9

(3)

¡Tú tenés razón! You.TU have.VOS reason

‘You’re right!’

(Antioquia, Frutos 1896, 152)

Campos (2010) discusses this as well, outlining some categories of inflectional and pronominal mixing as defined by their relation to voseo (Campos 2010: 227):

(4)

a. Complete Voseo (VV)

i. Verbal inflection of voseo alongside the pronoun vos. Prototypically identifies with the well-known Argentinean voseo: vos tenés. (See other forms of the verb in (4)c)

b. Pronominal Voseo (VT)

i. Verbal inflection of tuteo alongside the pronoun vos. The least common of the three, documented in: “Bolivia, especially the western region, the north of Perú, in rural areas of the coastal and mountainous regions of Ecuador, and in the Argentinean provinces of Santiago del Estero and Tucumán”: vos tienes

c. Verbal Voseo (TV)

i. Verbal inflection of voseo alongside the pronoun tú. Characteristic of the Spanish of Uruguay and Chile, also found in Guatemala, Honduras, and other Central American countries: tú estái(s), tenés or tenís.

The questions, then, are how much does mixing, pronominal or inflectional, occur in Colombia as suggested by Diáz (2015), Millán (2011), Moreno (2010), and Murrillo Fernandez (2004), in what social and dialectal contexts does it occur, and in what manner does it occur?

1.2 Corpus Selection

Mixing, of the pronominal and inflectional varieties, while reported by some of the above-mentioned studies, probably does not occur very frequently in naturally occurring data. To really be able to examine the phenomenon thoroughly, a very large sample of data could be necessary. The gathering of such a large sample of naturally occurring data containing

(10)

10

significant instances of mixed pronoun use, along with social and dialectal context as

suggested, would normally be quite the herculean effort for a typical study on address terms. Luckily, an extremely rich and untapped source of such data has, more or less, recently become accessible for research: Twitter.

When using corpora for research on address terms as found in naturally occurring data, texts are mainly sought that provide solid social and situational context for speech acts and are rich in direct one-on-one interactions between speakers. For diachronically focused studies this has most popularly proven to be plays and letters alongside other sources such as court records and fictional writing. Dramatic plays are a useful source not only for being a good representation of speech at the time of its writing, but also “because [drama] depicts interactions between different speakers of different social classes and different role

relationships towards each other, and because it is interesting to see how skillful dramatists choose to depict such interactions” (Jucker et. al. 2003: 9) Letters, meanwhile, especially those from prominent families who have often kept records of correspondence over time, “have proved a fruitful source for studying interaction, and e.g. a growth in politeness can be detected” (Jucker et. al. 2003: 9). Studies of contemporary language, meanwhile, and for the same contextual reasons, often make use of sources such as television or radio interviews, dramas in the form of movies, television, and plays, or advertisements. Others sources are always being sought as well; Moyna (2015) used song lyrics as a corpus because “tuteo/voseo variation is influenced by the fact that song lyrics straddle the written and oral media, like other cultural products (advertising copy, television scripts)” (Moyna 2015:127).

The key benefit to using Twitter is that it provides a massive source of naturally occurring data that is rich with address terms. I would argue, however, that it also delivers the same social context, by and large, as these previously used sources. If we consider Brown and Gilman (1989), plays such as Shakespeare are a good choice for diachronic address research because:

(5)

a. “dramatic texts provide the best information on colloquial speech of the period;

b. the psychological soliloquies in the tragedies provide the access to inner life that is necessary for a proper test of politeness theory;

c. and the tragedies represent the full range of society in a period of high relevance to politeness theory.” (Brown and Gilman 1989: 159)

(11)

11

Most solidly, Twitter provides much the same information on colloquial speech as mentioned in (5)a. Texting and online chatting, two of the most popular modes of computer-mediated communication, are the closest written approximations to how people speak today; as

McWhorter says in a 2013 TED talk on the matter, “one thing that we see is that texting is not writing at all…What texting is, despite the fact that it involves the brute mechanics of

something that we call writing, is fingered speech… Now we can write the way we talk.” One of the largest outlets of such “fingered speech” is Twitter. As a social network, it is filled with users texting and typing publicly every day, often to communicate directly to friends, family, coworkers, and others. A study by the Meridean Group (2012) on Twitter users in Colombia shows that 20% of tweets in the country are messages used to chat online, as one would with an instant messaging program or via SMS.

Further, the psychological context proposed in (5)b is also apparent on Twitter with close examination. While the identity of a speaker and with whom they are speaking may not always be known on Twitter, speakers do reveal emotions and attitudes in various manners using computer-mediated communication (CMC). Bieswanger (2013) shows a few ways this happens. Emoticons, for one, function “1) as emotion indicators … 2) as indicators of non-emotional meanings… 3) as illocutionary force indicators” (Bieswanger 2013: 472, Danet and Herring 2010: 250). Non-standard orthography and non-standard punctuation also provide information on the expressive intent of a speech act. Multiple punctuation marks or use of capitalization, for example, can connote emphasis, or behavior such as arguments (Bieswanger 2013:473-4, 477). With the advent of powerful tools such as ‘StreamR’ (Barberá 2013), an addition to the statistical programming language R, tweets can be downloaded in real time and filtered in extremely specific ways to examine the cues speakers give to show emotion and attitude in speech acts.

With enough data it is also possible to contextualize the characteristics of many speakers and addressees, as suggested in (5)c, in order to determine the factors conditioning pronoun selection with the use of frequency of pronoun choice appearing in different contexts. R can be used to separate data using, for example, certain nominal terms of address which appear with enough frequency to determine with a degree of certainty the identity of an interlocutor in relation to a speaker. Twitter also provides location data for enough users to show

comprehensive data for different dialectal areas. Furthermore, while it may seem Twitter consists of a narrow range of society, Colombia is the 12th largest user base in the world with six million users, and might contain a broader slice of the populace than suspected. Finally,

(12)

12

different ways of looking at the data can always be used to isolate certain groups. The activity of smart phone users, for example, 57% of which use Twitter in Latin America (comScore 2015), can be seen by separating tweets that use “emojis”, emoticons that are most popularly used with the help of smartphone apps, from tweets that do not.

2. Theoretical framework

2.1 Literature review

2.1.1 T/V distinction and pronoun mixing in European Languages

A massive amount of data might be necessary when searching for a peculiar a phenomenon as that of pronoun mixing. It can be a bit counterintuitive that an interlocutor would use two or more different forms with the same addressee. Going through the archetypal examples of address term theory, though, the mechanics of this phenomenon may be unraveled.

If we consider Brown and Gilman’s famous 1960 study, forms of address were theorized to follow a clear system of power vs. solidarity. One would expect a speaker to be able to make a clear choice of pronoun based on this system in any interaction in order to stay within the realm of politeness. For instance, in an utterance from a soldier to an officer, where on the scale of power the officer is superior, it seems obvious this superiority would demand a more respectful form. By and large, European languages have such a respectful form, termed V by Brown and Gilman (1960) (from the Latin vos), as well as an informal form, termed T (from the Latin tu). This system was typical as far back as in feudal Europe and Simon (2003:89) notes it is found in German courtly literature from that period well into the nineteenth century. Solidarity, meanwhile, is the characteristic of a relationship between two speakers based on shared experience, i.e. does the speaker like or dislike the interlocutor. This can also condition pronoun choice; if speakers have a relationship high in solidarity they will be more likely to trade T, while a low solidarity will favor V (Brown and Gilman 1960: 189-90). The determination of these two factors in choosing a pronoun, though, can also be conditioned by how well the speaker and interlocutor know one another, as well as the politeness begged of the situation. Brown and Gilman (1989) take politeness theory (Brown and Levinson 1987) into account in examining Shakespeare’s plays, as mentioned above. Here they considered the factors of power and distance, a combination of interactive closeness (familiarity) and affect (solidarity) in determining the politeness of a situation which would lead to an

(13)

13

expected pronoun choice. The plays they examine show a strong presence of affect but not of interactive closeness in pronoun choice, showing that in Shakespeare’s time the emotional mood of the speech act took precedence in determining the distance of the situation over familiarity between speaker and interlocutor. This meant, of course, that speakers could switch pronouns with the same interlocutor to express positive or negative affection.

Taking a step back, though, politeness itself has also been shown to cause such switching. A famous pronoun switch in a story from Reynard, a body of medieval European fables about a fox called Reynard clearly shows the role of politeness in pronoun choice. In one of the stories, a bear, Bruin, is trying to get honey from Reynard. In the Middle Dutch version of this body of fables, De Reynaert, the pronoun of distance ghi predominates, but a switch occurs in the story of Bruin when the bear addresses Reynard with the familiar du (Lulofs 1967: 72). According to Lulofs (1967), Bruin switches to the du form in order to curry favor with Reynard; a politeness strategy with the goal of communicating solidarity in order to convince Reynard to part with his honey.

(6)

“Bruun: Reinaert, wat aetstu wat?”

Bruin: Reynard, what eat.T,you.T what?

‘Bruin: Reynard, what are you eating there?’

This meaning was even retained in a later transcription of De Reynaert after du had fallen out of use in Dutch, giving way to gi. In this case, Reynard was addressed as, liefste Reinaert, ‘dearest Reynard’, retaining the polite meaning that the pronoun no longer carried. This still may seem cut and dry; the politeness begged of a situation in order to avoid the face-threatening act, and its determinant factors of power and distance determine the choice of a pronoun. A choice which speakers have no trouble making. When considering what solidarity, familiarity, and power actually mean in the social context of a speech act, though, the situation becomes more complex. Millán (2011) cites Friedrich (1971) who outlined an expansion of the classic paradigm to ten proposed components split into four categories as shown here in example (7) (Hughson 2009: 26, Millán 2011: 26):

(7)

I. “universal” components implied by all acts of speech a. topic of discourse

(14)

14

II. “biological” components which differ from culture to culture a. age

b. generation c. sex

d. kinship status

III. social and group components

a. relative jural/political authority (power) b. group membership

c. dialect

IV. solidarity component

a. Solidary semantic distinguishing between pronouns based on individual emotions

While still under the umbrella of the classic components, this model (developed from an analysis of pronoun use in Russian literature) helps us see more clearly and specifically how the paradigm is involved in pronoun selection. For example, if we consider Mazzon’s (2003) examination of pronouns in Shakespearean English, we see pronoun choice being

manipulated by speakers due to factors such as politeness phenomena, but also emotional mood of the speech (as in IV.a). In Othello Mazzon (2003) shows that the husband-wife relationship between Iago and Emilia mainly consists of V, but “the few switches to T are located at intensely dramatic points” (Mazzon 2003: 228).Mazzon (2003) describes pronoun switches as “exquisitely social devices” and proposes the possibility that the switching can be of a strategic nature (Mazzon 2003: 240-241). Stein (2003) takes into consideration the factor of social class (as in III.b) with regards to the use of T and V in Shakespeare. While he found a somewhat predictable and consistent behavior in the interaction between very upper and lower ends of the social strata, within the upper classes he finds some of the strategic use mentioned by Mazzon (2003) in King Lear where “it is the passions of the nobles and sovereigns that are signaled and contextualized by pronominal contrast” (Stein 2003: 302). Simon (2003) examines three versions of the Nibelungenlied, a German epic written around the year 1200. In this epic we see a pronoun switch in a conversation between Gunther, the king, and his friend Siegfried, a nobleman; the two start by exchanging V, ir, but in two of the manuscripts, Gunther switches T, du, when he begins telling Siegfried a secret, meaning topic of discourse (as in I.b) motivated the switch. (Simon 2003:90):

(15)

15

Daz sol dir frivntlichen tovgen sin gechleit

‘I’m telling you.T.SG this secretly, from friend to friend’ (Nibelunglied, ms. A, stanza 600, line 3)

The characters then mutually exchange T in two of the manuscripts of the Nibelungenlied, while Siegfried continues using V in the third manuscript. The writer of this third version may have seen III.a, power, as being more important in the dialogue. These differences between the manuscripts lead Simon (2003) to conclude that perhaps “the distinction between

du and ir was not an absolute one, but rather one that was subject to manipulation by

speakers” (Simon 2003:93).

With so many factors involved in pronoun choice, the situation is complex when one considers that many of these same factors are involved simultaneously and may each be pulling a speaker towards T or towards V. While Simon (2003) theorizes that after the Middle High German of Nibelunglied, respect actually grammaticalized in German (Simon 2003:99-100) showing a more systematized and rigid T/V distinction that would not be favorable to pronoun mixing, we still see such mixing contemporarily in European languages with less rigid systems and distinctions and this can be due to the entangling of simultaneous factors from (7). A few examples of this can be shown in two papers about T and V in contemporary Dutch. First, Aalberse (2004) examines the contemporary usage of V in Dutch, u, to

investigate claims of its coming disappearance, giving way to T, jij/je. She proposes that, rather than theorized causes such as incomplete language learning or a loss of the scope of applications of V, mixing of the two forms comes about from speakers’ acute social knowledge of each form combined with a clash between solidarity and power. With a university professor, for example, it is possible a “conflict of communication intentions” can arise if the professor asks to be addressed with the informal form, a strategy to communicate solidarity, even though the general social rule due to age (II.a) and difference in position (III.a), is to use the formal form (Aalberse 2004: 65). The same can also occur because of the topic of discourse (I.a); this is shown with an Alcoholics Anonymous ad which uses the informal pronoun to express solidarity with the addressee and an understanding of their problem, but the formal pronoun when discussing official or practical information. In (9) lines a to c, we see the formal V being used with official statements being made introducing the message of Alcoholics Anonymous, however in line c the ad introduces the idea that the organization is on the reader’s side. Logically, if we consider this a change in social context

(16)

16

(I.b), what follows is a switch to T in lines d to e. Here the advertisement switches to a sentiment of empathy and understanding, necessitating an emphasis on solidarity (Aalberse 2004: 64):

(9)

a En u bent er welkom, als u de moeite and you.V be.V there welcome as you.V the effort

wilt nemen om tenminste de eerste keer droog want.V take in.order at.least the first time dry

aan te komen. on to come

And you are welcome here, if you want to take the first step to going dry.

b De volgende keren lukt dat dan ook wel, The next times succeed that then also well al is het niet vanzelf.

if is it not from.self

The next steps after that it will work also, even if it is not automatic

c Maar u hoeft dat dan niet meer in uw but you.V do.V that then not more in your.V

eentje voor elkaar te boksen. own for each.other to fight

But you don’t have to fight on your own anymore

d Ben je goed aan de drank, maar wil je eraf, be.T you.T good on the drink but want.T you.T there.off

dan mag je vreselijk zwak zijn. then may.T you.T extremely weak be

When you’re.T hitting the bottle,

but you want to stop, you’re llowed to be extremely weak.

e Dan hoef je geen mannetjes- of vrouwtjesputter then do.T you.T no male or female.finch

te zijn, die dat zonder hulp van wie dan ook oplost. to be who that without help from who then also solves

Then you.T don’t have.T to be a

strong man or woman who has to solve this problem without help.

In the following example of an ad for a swingers’ club, we see something similar (Aalberse 2004: 65). Here we see T when the ad is referring to the more intimate topics in lines a to b (I.a). Then a shift to V occurs in lines c to f when the topic of discourse changes to refer to official information, such as descriptions of services (lines c to e) and directions to the club (line f). Finally a switch back to T occurs with a friendly encouraging message in line g (a change in I.b) similar to that in example (9):

(10)

a Als je van spanning en erotiek houdt kom dan naar As you.T from suspense and erotic love come.T then to

Part2, de parenclub voor jong en oud. Part2 the swingers.club for young and old

‘If you like suspense and eroticism cometo Part2, the swingers club for young and old.’

(17)

17

b Dit merk je meteen bij binnenkomst waar je this mark.T you.T with.one by in.come where you.T

onder het genot van een drankje aan de bar kan zitten. under the delight from a drink on the bar can.T sit

You’ll notice that immediately right at the bar where you. can

sit while enjoying a drink

c Het zitje bij de open haard biedt u gelegenheid om the seat by the open hearth bids you.V opportunity in.order

lekker weg te dromen. nice way to dream

The seat by the fireplace gives you the opportunity to daydream

d Bij de erotische ruimtes kunt u eventueel by the erotic rooms can.V you.V occasional

video's bekijken. videos watch

In the erotic rooms you can watch any videos you like

e Er is een gezellige dansvloer waar muziek van there is a cozy dancefloor where music from nu en van vroeger wordt gedraaid. (ook op verzoek) now and from earlier becomes turned also on request

There is a nice dance floor with music from today and the past. (also upon request)

f Parenclub Part2 kunt u vinden aan de swingers.club part2 can.V you.V find on the

Bredabaan 43 te Wuustwezel, net over de grens aan de Bredabaan 43 to Wuustwezel just over the border on the N263 Wernhout-Wuustwezel, afslag Loenhout.

N263 Wernhout-Wuustwezel exit Loenhout

You can find Swingers club Part2 in Wuustwezel,

Bredabaan 43, just past N263 to Wernhout-Wuustwezel, Loenhout exit

g Kom eens gezellig langs, Come.T once cozy over

je zult er beslist geen spijt van krijgen! you.T should.T there absolutely no spite from get

Come visit us, you won’t regret it!

Vismans (2016) finds such strategic switching as well, but also finds a case of confusion due to dialect contact (III.c). In the Netherlands it is much more common to exchange T, while in Flanders it is quite normal to exchange V. In the example, an interview with an

anesthesiologist in 2011 about the subject of pain management, the speakers begin the

conversation with the pronoun they are each most comfortable using; V for Kris Vissers (KV) from Flanders and T for Colet van der Ven (CV) from the Netherlands. The speakers both expect reciprocal pronominal treatment, so the incongruity comes up in conversation forcing the speakers to acknowledge the dialectal differences between themselves. They fail to come to a definite resolution, though, and in (11) and (12) we see this cause confusion for CV; she is not sure whether to switch to V to accommodate KV or to continue using T as she felt was more natural at the beginning:

(11)

CV: Nou ben b ... ben jij anesthesioloog. now be.T b be.T you.T anesthesiologist

(18)

18

‘CV: Well y... you.T are.T an anesthesiologist.’

(12) CV: Dat deelt u? Nu word ik de Vlaming. that share.V you.V now become I the Fleming

KV: Ja. yes

CV: Dat deel je? that share.T you.T

‘CV: Do you.V agree.V? Now I’m becoming the Fleming

KV: Yeah

CV: Do you.T agree?’

2.1.2 Mixing of informal pronouns in Colombian Spanish

In Colombia, with a mix of dialects and a pronominal paradigm that blurs social lines even more closely than in the above examples (see (1)) a system exists which is quite the opposite of the rigid grammaticalized distinction seen in contemporary German (Simon 2003). In determining the choice of pronoun based on the discussed factors, there is fertile ground for research. In addition to Campos’ breakdown in (1), Montes Giraldo (1982) provides a convenient breakdown of the distinct geography of these address term systems in the table translated below (Montes Giraldo 1982: 57):

(13)

Superdialect Subdivision Pronoun or form that predominates

Coastal

Caribbean Intimate tuteo Pacific Intimate voseo or voseo

between equals Central or Andean Eastern (Bogotá) Usted Tú Su merced (Boyacá) Western Vos Tú

(19)

19

Moreno (2010), finds some of the same processes at play as were found with switches related to the T/V distinction discussed previously by using radio interviews and cinematic

examples. In the example in (14) from the film Maria Full of Grace we see a switch between the informal usted and vos. Dialectal contact is involved in the example to a degree. Franklin (F), who is from Caldas, an area where voseo predominates general use (Montes Giraldo 1982, Moreno 2010: 1043) starts off using voseo with María (M), who is from outside of Bogotá and uses ustedeo almost exclusively throughout the film (Moreno 2010: 1043). He employs voseo “to express a certain proximity and familiarity”, but switches to ustedeo first at the end of the dialogue below (conoce instead of conocés) when “responding to a personal question” about where he is from (Moreno 2010: 1044). Moreno (2010) does not go into further detail about this switch, other than confirming it is the familiar usted rather than the

usted of distance which F uses later in the dialogue. This familiar usted is probably employed

to communicate a slightly closer and more familiar meaning than the voseo he was using up to line 7, or it could be a form of accommodation to M’s Bogotá dialect.

(14)

F: vos bailás muy bacano

you.VOS dance.VOS very cool

M: sí usted también yes you.UD too

F: no pensé que la gente de por acá bailara tan no thought.1 that the people of for here dance so

bacano yo pensé que eran todos más tiesos cool I thought.1 that were all more rigid M: y usted de dónde es de verdad

and you.UD of where is of truth

F: yo soy de un pueblito que se llama Santa Rosa de Cabal

I am of a village that itself call Santa Rosa de Cabal ahí como a diez minutos de Pereira conoce?

there like to ten minutes of Pereira know.UD

‘F: You.VOS dance.VOS really cool M: Yeah, you.UD too

(20)

20

F: I didn’t think people from around here danced so cool, I thought that they were all more uptight

M: And you.UD, where are.UD you from really?

F: I’m from a small village named Santa Rosa de Cabal, it’s about 10 minutes from Pereira know.UD it?’

Further along in the example we see another interesting mixing of forms by F. He moves to the impersonal use of usted at the beginning of the example below when he is describing the details of a drug trafficking job to M. Then, when he notices her distaste at the idea, he switches abruptly back to voseo near the end with sabés que ‘you.vos know.vos what’. This

could be interpreted as a face-threatening act because an expression that “accompanied by a falling intonation frequently anticipates a negative comment or insult” (Moreno 2010: 1044). His goal in this line of dialogue is to emphasize to M that he was doing her a favor by even bringing this up and that he may be offended by her hesitation:

(15)

F: es un trabajo de mula a mí me parece que es un trabajo bacano

is a job of mule to me myself seem that is a job cool porque usted va lleva lo que tiene que llevar lo entrega because you.UD go.UD carry.UD it that have to carry it deposit.UD

le dan su billete y ya todo sano… yo conozco yourself.UD give your.UD paycheck and already all clean I know

un man| el hombre organiza todo, los papeles, manda a la gente… a guy the man organizes all the papers send to the people

sabés qué olvidémoslo, vamos

know.VOS what forget.us.it go.1PL

‘25 F: it’s a job as a mule, it seems like it’s a cool job to me because you.UD just go.UD, take.UD what you have.UD to take, deliver.UD it, they give you.UD

your.UD pay and that’s it, all clean. I know a guy, the man organizes

everything, the papers, sending the people… know.VOS what, forget it, let’s go’

Millán (2011) also offers insight into the Colombian situation and hypothesized along a similar framework as those analyzing T and V pronoun mixing, hoping to determine from questionnaires and oral interviews what extralinguistic factors condition informal pronoun

(21)

21

use in Colombia and when they cause mixing; in other words, could pronoun use and switching still be considered “exquisitely social” when three distinct forms have similar and overlapping meanings? Millán (2011) examines two specific groups of speakers, those in Antioquia and its capital Medellín and those in Valle del Cauca and its capital Cali. She shows via questionnaire when, where, and with whom certain pronouns are used. She also asks in oral interviews if there are specific situations when a speaker would want to switch pronouns to find out when mixing occurs (Millán 2011:48). Millán (2011) concludes that the profile of the two areas is very similar in pronoun usage, but that some of the discussed extralinguistic factors in determining pronoun choice are involved in different ways for each region. A few results relevant to the present study are as follows. For the relationship of the interlocutor, usted is much more common in non-familial relationships while vos and tú are more common within the family. With social class, tú increases the more social status increases in Medellín, while lower class speakers are shown to mix forms five times more in Cali. With generation, usted is used much more to older speakers while vos to equal and younger speakers. The effect of emotional closeness, finally, shows a strong effect where

usted is used in negative context, vos in a positive one, and tú does not co-occur with

negative emotion. These factors and more are summarized in the below table (Millán 2011: 169-174):

(16)

Factor Cali Medellín

Sex of speaker

1. Usted used more in females than males 2. Tú used more in males than females

3. Vos used 7% more by males vs. females 4. Mixed use higher in females

1. Usted used more in males than females

2. Tú used more in females than males

3. Vos used 2% more by males vs. females 4. Mixed use higher in females

Social Class

1. Tú equal across social strata

2.Usted used most by lower class

3. Use of vos increases with class

1.Tú increases with social status

2.Usted used most by lower class

3. Use of vos increases with class

(22)

22

4. Mixed use 5x higher with lower class

4. No clear effect on mixed use

Place of Interaction

1. Vos and tú used most at home

2.Usted used most in public and at university

3. Mixed use is used more at home

1. Vos and tú used most at home

2.Usted used most in public and at university

3. Mixed use is used more in public

Relationship to Interlocutor

1.Vos and usted used more with family

2. Usted used more with classmates

3. Usted used 66% of the time with strangers 4. Usted used 50% of the time with friends

1.Vos used more with family

2.Vos and usted used more with classmates

3. Usted used 66% of the time with strangers 4. Usted used 47% of the time with friends

Generation

1. Usted used to older interlocutors

2.Vos used to younger interlocutors

3.Usted used with equals

1. Usted used to older interlocutors

2.Vos used to younger interlocutors

3.Vos used with equals Sex of Interlocutor 1. Men use tú more often to

women

1. Men use tú more often to women

Topic of Discourse

1. Vos used more for private topics

2. Usted used for non-private topics

3. Mixed use is more common in private interactions than non-private

1. Vos used more for private topics- to a higher degree than in Cali 2. Usted used for non-private topics

3. Mixed use and tú are more common in private interactions than non-private

Emotional Closeness

1. Usted is used negatively in an emotionally close situation

(23)

23

2. Vos is positive in an emotionally close situation 3. Tú occurs more without

negative emotion involved

2. Vos is positive in an emotionally close situation 3. Tú occurs more without negative emotion involved 4. Mixing occurs slightly more in a positive sense in emotionally close

situations

With these results in mind, Millán (2011) proposes a continuum of pronouns for Medellín and Cali that is even more complex than what Campos reports (see (1)c).

(17)

tú --- usted --- vos --- tú --- usted

Solidarity/Familiarity Distance/Respect She also outlines a number of interesting conclusions relevant to the present study:

 Mixed-use of all three informal pronouns is conditioned by extralinguistic factors

o “sex of the speaker, place of interaction, and relationship with the interlocutor display greater variation in mixed-use in Medellín” (Millán 2011: 180)

o “sex of the speaker, social class, place of interaction, relationship with the interlocutor, generation, and topic of discourse seem to be more influential in Cali” (Millán 2011: 180);

 The use of vos is lower than in previous studies suggesting a possible change in progress (Millán 2011: 177);

 The higher use of tú could also signal change from above when it is taken to account that tú is the preferred informal form in the prestigious Bogotá dialect, tú is associated with higher social classes, and that females, usually leaders in change due to prestige (Labov 2001), use tú more than males with younger addressees (Millán 2011: 181-2);

Millán’s study, however, does not go into detail about specifically which pronouns are mixed and how. She only reports data for mixing when a speaker’s answers in an oral interview

(24)

24

suggests they mix pronouns in the same speech act with the same interlocutor based on certain factors, e.g. anger, but not the manner of mixing and the pronouns mixed in specific. She therefore provides two relevant avenues for future research: “a closer examination of the instances of mixed-use, considering all the possible combinations of the pronouns tú, usted and vos to analyze if there are differences in their uses.” and to “collect naturally occurring data to observe actual use of address form and compare it to the results from the

questionnaires and oral interviews” (Millán 2011: 182-3). The present study hopes to follow both of those avenues and investigate the continuum proposed by Millán (2011) in (17).

2.1.3 Language Internal Effect?

In considering (3) and (4) once more, however, we have yet another factor that is worth investigating as it could be involved in resistance to or shaping of ongoing change. In (3) “an Antioquian woman tries to imitate the Bogotan tuteo to talk with her loved one. She does not achieve producing a plain tuteo because she only knows the endings of vos. The verbal ending is more entrenched in the linguistic system, so the innovative element in T-V is uniquely tú as a lexical form (Rona 1967)” (Diáz Collazos 2015: 8). Rona (1967) hints at the possibility of this appearance of the lexical form alone, out of tune with the verbal paradigm, as being a separate type of phenomenon. He states that voseo or tuteo isn’t truly occurring unless the lexical and verbal forms align, a phenomenon he calls liso y llano, or ‘outright’

voseo/tuteo (Rona 1967: 10). If the proposition that the verbal inflection is actually more

deeply entrenched in the minds of speakers is true, it could affect the manner in which pronouns are mixed in Colombian Spanish. The entrenchment of the verbal paradigms of forms of address in Spanish can be seen, for example, in the Río de la Plata region of Uruguay and Argentina, where voseo is by far the dominant 2nd person singular pronominal paradigm: “the vos verbal paradigm is so strongly internalized in the speech of Plate River speakers that it may feel "uncomfortable" if not difficult to produce tú forms of the verb” (Lipski 1994: 341).

The meanings assigned to the nominative pronoun and verbal inflection have also been shown to diverge in parts of Latin America. Tú in Montevideo, Uruguay is employed as a way of communicating an intermediate level of formality, but many times speakers still employ the vos verbal paradigm (Fontanella de Weinberg 1999, Weyers 2009). Furthermore, in Chile and other parts of Uruguay we see that: “it is not the authentic voseo (vos comís), but the verbal voseo (tú comís). Although the idea of voseo as a vulgar form has started to

(25)

25

disappear [in Chile], just as in Uruguay the pronoun vos is avoided because it is still considered vulgar and disrespectful, especially to the upper classes (Torrejón 1991)”

(Benavides 2003: 620). Perhaps the verbal paradigm is less stigmatized in this case precisely because it occupies that theorized more deeply entrenched space for speakers. If this has occurred in other varieties of Spanish, it should definitely be worth investigating in Colombia, where the address term system is rich for examination. It should be noted that, while inflection of tuteo and voseo are involved in pronoun mixing in Spanish, avoidance of a pronoun due to inflectional economy might not play a role. In Spanish, pro-drop allows for pronoun choice to be shown solely with verbal inflection and mixing can occur when inflection of one paradigm is mixed with the nominative pronoun from another paradigm. Pro-drop allows for the fluidity between the pronoun and the verbal inflection seen in (3) and (4), and means that speakers do not mix these pronouns specifically in order to avoid

inflection. This fluidity does not occur in some other European languages historically, such as English or Dutch, where the pronoun and verb are tightly attached. In languages of that nature, pronoun choice can be motivated by avoidance of non-economical inflection; this is reported by Aalberse (2009: 86) as a motivation for the choice of Middle Dutch gi over du, for example, where gi’s verbal paradigm carried the more economical inflection of the third person -t while du’s verbal paradigm carried the non-economical inflection, -s.

In order to investigate the entrenchment of the verbal paradigm and its effect on mixing in Colombia we have a few avenues. First, examining mixing of pronouns with the same addressee in the same conversation is good way to expose the meanings speakers assign, consciously or not, to each form of address and possibly reveal reasons for any ongoing change or resistance to change. If we consider again the large amount of data sought with the use of Twitter, it becomes clear that gathering data could result in quite a few tweets in the same dialectal zones for which the attitude towards each pronoun as broken down in social context will also have been investigated as discussed. If we look further into these groups of tweets that all exist in an equal dialectal context and examine closely exactly how pronoun mixing occurs, as Millán suggests would be useful, we might be able to see some interesting patterns in use. Given the variety of options associated with the tripartite system as seen in (2) many patterns could, hypothetically, be possible. In considering verbal mixing, then, the entrenchment of verbs of frequency (Aalberse and Stoop, 2015) is important. Said verbs should be less flexible in their use, meaning speakers would be more likely to employ them

(26)

26

alongside their correct pronoun. Low-frequency verbs, however, would appear often discordant with the pronoun if the verbal paradigm is indeed more entrenched in speakers’ minds. This would mean a verb like ser, ‘to be’, would not appear often out of tune with the pronominal form; resistance to change in the verbal paradigm would leave the non-frequent verb in the form a speaker is most used to or comfortable with while the pronoun would more likely represent the speaker’s conscious choice to communicate meaning associated with said pronoun.

Overall, if the verbal paradigm is more entrenched for speakers, verbal mixing discordant with pronouns in a given dialect could favor one verbal paradigm, much as it does for speakers in the Río de la Plata area of Argentina and Uruguay (Lipski 1994: 341). If verbal mixing occurs with non-frequent verbs from both paradigms, it could merit further

investigation into why different speakers might be showing signs of entrenchment for different paradigms in the same dialectal context.

2.1.4 Geography’s Role

As mentioned previously, one of the most popular variables for distinguishing dialects in Latin America has proven to be the use of voseo, and it has also been a popular variable with geographical linguists and dialectologists (Pountain 2003, Penny 2000). Mapping of the use of voseo in Colombia, though, hasn’t been done in any detailed way since Montes Giraldo (1967) (see (18)), and even this map is somewhat inconsistent with Montes’ later work: “Montes Giraldo (1982) expresses that the pacific coast of Colombia is also voseante, though his 1967 map ignores this region” (Diáz Collazos, 2015: 11). Creating an updated geographic description of the variable with empirical data could be very useful for confirming or

informing Millán’s (2011) conclusions about Anqtioquian and Caleña Spanish and comparing to the map previously compiled by Montes to inform about change over the past 50 years.

(27)

27

(18)

General Voseo as an Informal Address Term, voseo generalizado como trato de

confianza, Montes Giraldo 1967

Geography can help us determine which dialectal area a tweet belongs to and help inform predictions as well. The most populous and interesting areas are, of course, the two studied by Millán (2011), Antioquia and Valle del Cauca, along with the capital region Bogotá studied by Moreno (2010). Moreno states “if the variation and contact of dialects are

characteristic of Spanish spoken in Colombia, it’s possible that the fluctuations in the system of address terms reflect these phenomena” (Millán 2011: 1038). She goes on to present the interesting fact that in Bogotá “39% of the population surveyed in the first trimester of 2003 correspond to immigrants from other provinces (Pérez Torres 2003: 6-7)” (Moreno 2010: 1038). This is due to Bogotá’s economic and cultural dominance in the country. As we see in (14) and (15) Moreno (2010) shows that alternation of pronouns can be used strategically and she goes on to conclude that this strategy “can be understood as a variation of style and a form of discursive polyphony that is part of the linguistic “habitus” of the speakers that interact in the Colombian capital” (Moreno 2010:1047). In other words, that dialect contact is

(28)

28

the main source of the mixing. On the other hand, if we consider language contact in a different light, one that could create weakness in a complex system, we might have a different view. If dialect contact is stressing the system in a similar way in Bogotá, this should show up in the way speakers mix forms, using the inflection they are most

comfortable with more often. If dialect contact is causing confusion as we saw in (11) and (12), or accommodation as we saw in (14), the mixing would be expected to have verbal and inflectional agreement.

3. Research questions, hypotheses, and predictions

3.1 Research Questions

3.1.1 Can a hierarchy of extralinguistic factors from Friedrich’s (1971) ten components be shown to condition address term choice in naturally occurring data favoring emotional closeness?

Millán (2011) finds in both Cali and Medellín that “in sum, emotional closeness plays an important role in the selection of one form of address over another; whereas the role of the sex and the social class of the speaker does not seem to be as strong.” If this effect is clear from surveys and questionnaires on address term choice, finding it on Twitter as well could tell us that the speakers on the platform assign similar meanings to speech acts on Twitter and in spoken language. Of the factors in Friedrich (1971), the present study will examine sex, generation, kinship status, group membership, dialect, and solidary emotions. Dialect will also be expected to have a strong effect on the corpus, being that the distinct Bogotá dialect will be examined alongside those Millán (2011) examined, Cali and Medellín.

3.1.1.1 Can a qualitative analysis of select mixed tweets show shifts in meaning as related to a hierarchy of external conditioning of address term choice found on the

platform?

If pronoun choice is conditioned by certain extralinguistic factors more strongly than others as inquired in 3.1.1, analysis of select mixed tweets discovered in the data should show clear shifts or conflicts in meaning related to the strongest factors found to be conditioning

pronoun choice. This would support Millán’s (2011:180) assertion that “The use of more than one pronoun with the same addressee in the same discourse (mixed use) is conditioned by external factors.” Millán (2011) found greatest effect on mixed use related to sex of the speaker, place of interaction, and relationship to speaker in Medellín and the same factors alongside topic of discourse and generation in Cali. In terms of Millán’s study, greater effect

(29)

29

by sex of the speaker in Medellín, for example, means that speakers were more likely, if they were female rather than male as opposed to older vs. younger or angry vs. affectionate, to answer in questionnaires that they would perform a switch in a given set of situations. Examples of such situations included “when they were angry at somebody or to ask for a favor” (Millán 2011: 48). In following Millán’s avenue for research suggesting more specific analysis of mixing, specific shifts will be sought related more to the situations given in her questionnaire than to overall likelihood of mixing by different groups.

3.1.2 Can ongoing change from above favoring tú be supported on the platform as seen by Millán (2011)?

Millán (2011: 181-2) finds that “the higher use of tú could also signal change from above when it is taken to account that tú is the preferred informal form in the prestigious Bogotá dialect, tú is associated with higher social classes, and that females, usually leaders in change due to prestige (Labov 2001), use tú more than males with younger addressees.” This could be supported when examining the frequency of tuteo on the platform as related to social class and the overall frequency of tuteo on the platform in Cali and Medellín.

3.1.2.1 Can a comparison of current usage in the region with past linguistic maps support this ongoing process of change?

In relation to 3.1.2 the frequency of tuteo could show geographical gains, and/or the frequency of voseo geographical losses as compared to Montes Giraldo’s (1967) map of

voseo in Colombia. When considering the overall use on the platform, and the strengths and

weaknesses of this map, further support could be unveiled for the change proposed by Millán (2011).

3.1.3 Can the entrenchment of the verbal paradigm be shown to be a prime mover of mixing?

When inflection is available to communicate address term choice, for voseo and tuteo rather than ustedeo, mixed use could occur in a way that shows the entrenchment of the verbal paradigm affecting how speakers mix forms. Diáz Collazos (2015) and Rona (1967) propose that speakers at times mix verbal forms of one paradigm with the pronoun of another due to the entrenchment of the verbal paradigm in the linguistic system. If the entrenchment of the verbal paradigm is a prime mover of mixing, it could mean that this form of mixing, in which

(30)

30

the verb is discordant with the pronoun, could constitute a separate phenomenon from mixing motivated by shifts in external factors, as sought by 3.1.1.1. Mixing of two types, VT

(pronoun vos with verbal tuteo) and TV (pronoun tú with verbal voseo), are described by Campos (2010), with VT reported as being uncommon in Latin America and not reported as occurring in Colombia.

3.2 Hypotheses

3.2.1 Hierarchy of extralinguistic factors conditioning address term choice

It is hypothesized that in answering research question 3.1.1, a hierarchy of factors

conditioning address term choice will be found on the platform in agreement with Millán (2011), showing that emotional closeness and dialect (due to the inclusion of Bogotá) will have a stronger effect on address term choice than other factors, especially sex and social class.

3.2.1.1 Shifts in meaning or conflicting factors supporting hierarchy of extralinguistic factors in select mixed tweets

It is hypothesized that in answering research question 3.1.1.1, evidence showing shifts or conflicts in the most reactive factors affecting pronoun choice, (e.g. shift from positive to negative emotion, dialect contact, etc.) as revealed by answering 3.1.1 will be shown to cause mixing in select tweets, supporting Millán’s (2011:180) assertion that “the use of more than one pronoun with the same addressee in the same discourse (mixed use) is conditioned by external factors.”

3.2.2 Ongoing change from above favoring tú

It is hypothesized that in answering research question 3.1.2, evidence supporting change from above will come in the form of increased use of tú over vos by speakers in Cali and Medellín and a favoring of the pronoun by middle and upper class speakers overall on the platform.

3.2.2.1 Geographical evidence for ongoing change

It is hypothesized that in answering research question 3.1.2.1, evidence supporting Millán’s (2011) assertion of ongoing change favoring tú will be found in the form of an overall geographical reduction of voseo as compared to Montes Giraldo’s (1967) map of voseo in Colombia and the presence of tuteo in the strongest areas of voseo on the past map.

(31)

31

3.2.3 Entrenchment of the verbal paradigm as a prime mover of mixing

It is hypothesized that in answering research question 3.1.3, evidence will be found

supporting the entrenchment of the verbal paradigm as a prime mover of mixing by showing low frequency verbs from the entrenched paradigm discordant with the pronoun significantly more often than frequent verbs.

3.3 Predictions

3.3.1 Hierarchy of extralinguistic factors conditioning address term choice

This hypothesis would be confirmed if the frequencies in address term choice in situations of emotional closeness and in different dialects differ from the overall corpus significantly more than in other situations in which factors like class and sex are isolated. 3.1.1. If the factors of dialect and emotional closeness do not differ from the overall corpus more so than all other factors, this hypothesis would be rejected.

3.3.1.1 Shifts in meaning or conflicting factors supporting hierarchy of extralinguistic factors in select mixed tweets

This hypothesis would be confirmed if mixed tweets are found that show a clear shift in meaning following the most reactive factors shown on the platform. It would be rejected if no such shifts in mixed tweets are found.

3.3.2 Ongoing change from above favoring tú

This hypothesis would be confirmed if the frequency of tú is favored more strongly than in the overall corpus by users shown to most likely be of a higher social class and if tú is favored overall in Cali and Medellín. In the present study, smart phone users will be isolated as a group; considering the financial burden (up front and month-to-month) of owning a smart phone, these users could likelier be members of a higher social class. If this group does not use tú at a higher rate than in the overall corpus, and vos or usted is favored on the corpus by speakers in Cali and Medellín, the hypothesis would be rejected.

(32)

32

This hypothesis would be confirmed if the use of voseo has receded as compared to the past map of the variable reported by Montes Giraldo (1967) while considering its strengths and weaknesses and the nature of the platform examined in the new maps, and if tuteo is present in the strongest areas of voseo reported by the past map. This hypothesis would be rejected if

voseo appears strongly in the same areas as reported in the past map, and/or in new areas, or

if tuteo does not appear in these areas.

3.3.3 Entrenchment of the verbal paradigm as a prime mover of mixing

This hypothesis would be confirmed if there is a distinction found between the behavior of frequent and non-frequent verbs for one of the verbal paradigms in a given dialectal zone. Frequent verbs would be expected to be mixed with disagreeing pronouns less often than other verbs; e.g. if voseo’s verbal paradigm is more entrenched for speakers in Cali and Medellín, this would mean verbs of low frequency in voseo would show up discordant with the pronoun significantly more often than verbs of low frequency in tuteo (or vice versa if

tuteo is more entrenched). To confirm this, the ratio of non-frequent verbs to frequent verbs

should increase when the verbs appear discordant with the pronoun or with any frequent verbs in the speech act as compared to the ratio of non-frequent to frequent verbs in the overall corpus of non-mixed tweets. If this verbal mixing is confirmed as a prime mover of mixing, it would constitute a separate phenomenon from mixing due to social shifts and conflicts between social factors. Anecdotal evidence from Campos (2010) points to this favoring form TV rather than VT which is not reported in Colombia. To reject this

hypothesis, non-frequent verbs would not be shown to appear significantly more than normal in mixed tweets for either paradigm. Another rejection could come in the form of an overall favoring of a ‘clean switch’ rather than verbal mixing between voseo and tuteo. If mixing is done in the form of a switch with verbs alongside the correct corresponding pronouns, it could connote that the verbal switch is not very difficult for speakers, and that perhaps the verbal paradigm is not any more entrenched than the pronominal paradigm.

4. Methodology

4.1 Corpus

The corpus is composed of tweets downloaded from the publicly available Twitter Streaming Application Program Interface (API) using the R package streamR (Barberá 2013). The

(33)

33

tweets downloaded are bound to an area which includes the entirety of Colombia (using the coordinates SW -80.77,-4.39, NE -59.62, 13.07). Data was collected in bulk with the fact in mind that it could be parsed easily in many different ways as required by the study. The collection encompassed a period of about two months from late July to early September of 2015 and totaled to 11 gigabytes of tweets, parsed down to around six or seven megabytes of final data.

4.2 Parsing the corpus

The most important part of the data collection was parsing the bulk data to fit the needs of the investigation. First and foremost, tweets were parsed to include only those that list Spanish as their selected language and Colombia as their country of origin. Tweets that included location data only on the basis of ‘country’ rather than ‘city’, ‘place of interest’, or ‘administrative zone’ were deleted, as these did not show specific enough locations for the purposes of this investigation (such tweets are placed at the geographic center of the country in question). The next step was to parse tweets into separate categories: vos tweets, tú tweets, usted tweets and mixed use tweets (tú/vos, tú/usted, vos/usted). This was accomplished in a few ways. First, in order to determine the overall use of each of the three pronouns in the corpus, tweets containing each form were parsed using a regular expression. Then, a list of 25 commonly used present tense verbs was parsed alongside tú and vos which, regardless of the inclusion of diacritics, could be identified as being inflected for tuteo or voseo. These verbs (in their infinitive forms) are as follows:

(19)

Tener ‘to have’ Ser ‘to be’ Querer ‘to want’ Decir ‘to say’ Pedir ‘to request’ Volver ‘to return’ Poder ‘to be able’ Dormir ‘to sleep’ Cerrar ‘to close’ Costar ‘to cost/to

find difficult’

Interpreting the Distribution of Mixed Use of Pronominal Address Terms in Colombia using Twitter