On versus tu and vous: Pronouns with indefinite reference - CiteSeerX

24 downloads 1675 Views 234KB Size Report
indefinite subject pronouns in synchronous (i.e. real-time, live) electronic French discourse. ... First, real-time or live chat is primarily labeled and referred to.
ARTICLE IN PRESS Available online at www.sciencedirect.com

Language Sciences xxx (2008) xxx–xxx www.elsevier.com/locate/langsci

On versus tu and vous: Pronouns with indefinite reference in synchronous electronic French discourse Lawrence Williams *, Re´mi A. van Compernolle Department of Foreign Languages and Literatures, University of North Texas, P.O. Box 311127, Denton, TX 76203-1127, USA Received 8 August 2007; received in revised form 25 November 2007; accepted 26 November 2007

Abstract This article examines the variable use of the French pronouns on, tu, and vous with generic-indefinite reference in synchronous computer-mediated communication (CMC). The primary objective is to compare the use of these pronouns in CMC, a written/typed form of discourse, and in conversational spoken French as it has been analyzed over the past few decades. A VARBRUL analysis produced results comparable to those reported in previous studies in which such an approach was used. Overall, the results suggest that the use of on versus tu/vous with indefinite reference is influenced by affirmation/negation, syntactic frame (generalizations versus implicatives), discursive-pragmatic effect (situational insertions versus morals/truisms), and type of event (specific versus non-specific). Ó 2007 Elsevier Ltd. All rights reserved. Keywords: Computer-mediated communication; French; Pronouns; Sociolinguistics

1. Introduction The French subject pronoun on possesses a wide range of potential meanings and referents, especially in spoken discourse. Its versatility, while offering many opportunities for a variety of uses and discursive-pragmatic flexibility, can also create ambiguity, whether this is done expressly or unintentionally.1 As one of the very few lexical items to move into a restricted or closed category, from common noun (from a form of the Latin homo, meaning man, then to om in Old French) to pronoun, on has never been entirely semantically stable, which might explain its continued ability to adapt itself to the needs of the shifting pronoun paradigm in French (see Rickard, 1993, pp. 23, 49, 68; Ayres-Bennett, 1996, pp. 28, 105). In grammars and textbooks, the pronoun on is usually presented first as an equivalent of the generic English pronoun one. Some textbooks explain the versatility of on, but the scope is often limited to a discussion of its alternation with the subject pronoun nous (i.e. an equivalent of the English pronoun we) in everyday, conversational French (Blondeau, *

Corresponding author. E-mail addresses: [email protected], [email protected] (L. Williams). 1 For a thorough description of the various referential values of the pronoun on, see Jisa and Viguie´ (2005) or a comprehensive dictionary such as the Tre´sor de la langue francßaise. An historical overview of indefinite French pronoun use is provided in Coveney (2003). 0388-0001/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.langsci.2007.11.001

Please cite this article in press as: Williams, L., van Compernolle, R.A., On versus tu and vous: Pronouns with indefinite reference ..., Lang. Sci. (2008), doi:10.1016/j.langsci.2007.11.001

ARTICLE IN PRESS 2

L. Williams, R.A. van Compernolle / Language Sciences xxx (2008) xxx–xxx

2001; Coveney, 2000; Fonseca-Greber and Waugh, 2003). Even when textbooks, for example, go beyond the interchangeable use of on and nous, they rarely offer an account of the pronoun paradigm that reflects anything other than literary/formal French. Coveney (2003) notes one specific example of a French textbook for English-speaking learners in which ‘the following mild criticism [is offered regarding the generic-indefinite use of tu and vous]: Vous should not be used in this general sense (Price 1993: 207)’ (p. 168). Nonetheless, on can alternate with tu, vous, and ils with indefinite reference, and in addition to alternating with nous as a definite pronoun, it can also be used (as a definite pronoun) to replace or alternate with all other subject pronouns, even though these cases are certainly less frequent and more stylistically marked (Peeters, 2006). Example 1 demonstrates a use of indefinite tu, a case in which indefinite on could also have been used. (Incidentally, singular vous is, of course, possible instead of tu, but the complexities of tu vs. vous are beyond the scope of this article.) This example is an excerpt from a discussion in this chat room about some of the social aspects of on-line communication. In Example 1, Titigre uses a direct object pronoun with definite reference in the first clause, but in the second clause the use of tu is clearly indefinite since BiiJoujou is not being singled out as the only person who can make friends on line. Titigre is obviously offering an opinion based on personal experiences and observations. (No non-traditional linguistic forms were altered in these examples. They were reproduced as they appear in the corpus.) Example 1. hTitigrei BiiJoujou je te rassure, tu peux te faire de belles amitie´s et partager des moments avec des gens en ligne [‘BiiJoujou I assure you, you can establish good friendships and spend time with people on line’] Example 2 shows tu as a definite or non-generic subject pronoun, which is made clear by the use of nominal address (i.e. Lisette) to introduce the clause and the presence of the disjunctive pronoun toi in phrase-final position. In this case, tu could not alternate with on since this is an instance of definite reference. Example 2. hCDroleiLisette t’aimes le rap ou pas toi? [‘Lisette do you like rap or not?’] The present study analyzes the distribution and variation of on and tu/vous when they are used as generic or indefinite subject pronouns in synchronous (i.e. real-time, live) electronic French discourse. All tokens of on and tu/vous coded for our analysis are those that can be used interchangeably since our primary objective is to understand how and in which contexts they alternate. Although this type of communication—live chat—is not absolutely synchronous, we refer to it as such in spite of Garcia and Jacobs’s (1999) labeling of it as quasisynchronous, and we do so for two reasons. First, real-time or live chat is primarily labeled and referred to as being synchronous in the literature since it resembles, in many ways, real-time speech, which is usually in real-time and more or less spontaneous. Second, the only text-based chat software that allows participants to view letters and words as they are being typed (i.e. truly synchronous) is ICQ, which is primarily a tool for sending instant messages (i.e. in a private, restricted, limited group), whereas most studies of live chat are based on public communication spaces where many-to-many interaction occurs. Although voice chat, which is similar to telephone conferencing via the Internet, would fit into the category of truly synchronous discourse proposed by Garcia and Jacobs, this is a relatively new technology that has not been as widely studied as textbased forms of chat. Nonetheless, we suspect that as other new technologies are developed and the field of computer-mediated communication matures, the distinction made by Garcia and Jacobs will eventually be incorporated into analyses, descriptions, and typologies of electronic discourse. Our analysis of the distribution and variable use of generic-indefinite on, tu, and vous in synchronous electronic French discourse aims to compare our findings with those of researchers who have studied spoken discourse recorded during sociolinguistic interviews within the variationist tradition developed by Labov (1966, 1972, 1994) in addition to studies operating within different frameworks. Although studies using the Labovian model were limited in the early years to phonological variation, the scope has since been widened to include studies of lexical, morphological, and syntactic variation, a move which has not been entirely uncontroversial. Some of the earliest criticism can be found in Lavandera (1978), a well known and often-cited article in which Lavandera, while recognizing the value of variationist research that investigates non-phonological linguistic variation, cautions that additional methodological considerations must be addressed and that data for any such analyses should not necessarily and automatically be given the same status as those used to study phonological variation ‘because they need further interpretation; they do not in themselves constitute a definitive analysis’ (p. 171; see also Wolfram, 1991). Please cite this article in press as: Williams, L., van Compernolle, R.A., On versus tu and vous: Pronouns with indefinite reference ..., Lang. Sci. (2008), doi:10.1016/j.langsci.2007.11.001

ARTICLE IN PRESS L. Williams, R.A. van Compernolle / Language Sciences xxx (2008) xxx–xxx

3

One such methodological consideration, as Milroy (1987) noted, involves obtaining the ‘full range of realizations associated with [non-phonological variables]’ (p. 44; italics in original). In light of this critique, we have limited our analysis to instances where on and tu/vous can be used interchangeably for generalizing, a practice described elsewhere in the literature, thereby defining a strict range of realizations (i.e. two variants) associated with this variable. Nonetheless, we acknowledge that on can vary with other pronouns, most notably ils, for indefinite reference (Sankoff and Laberge, 1978), although this use of indefinite pronouns (i.e. on versus ils) functions to defocus the referent, whereas on and tu/vous can alternate when making generalizations (Ashby, 1992, p. 137–139). Thus, our analysis considers the choice of on versus tu/vous for expressing one type of indefinite meaning (i.e. generalizations) that can only be expressed by one or the other of these pronouns. The closed-class nature of this variable is similar to other well known grammatical variables in modern French, such as variable ne deletion (Armstrong, 2002; Ashby, 2001; Coveney, 1996; Hansen and Malderez, 2004), the use of on versus nous for (definite) first-person plural reference (Blondeau, 2001; Coveney, 2000; Fonseca-Greber and Waugh, 2003), and yes/no interrogatives structures (Coveney, 1996; Elsig and Poplack, 2006), among others, that have been the object of variationist analyses. Since our data do not come from sociolinguistic interviews, and since we have no verifiable biographical or demographic information about the people who produced the data we have collected, this study is limited to a comparison of the distribution and use of indefinite on, tu, and vous in spoken and electronic discourse, with some focus on the influence of discursive-pragmatic factors on variation. Our aim is to understand in what ways and to what extent indefinite pronoun use in synchronous chat reflects patterns found in previously reported studies of spoken French. As such, we expect that, given the seemingly spontaneous and informal nature of real-time, many-to-many chat, the use of tu and vous will be favored in cases where the indefinite use of these pronouns can vary with on, the older and more formal variant. In line with Bell’s (2001) revised framework of language style as audience design, we also hypothesize that regardless of a noticeable preference for either on or tu/vous, participants will establish a pattern or system that can serve as the norm for the group. General principle two of Bell’s framework states that ‘[s]tyle derives its meaning from the association of linguistic features with particular social groups’ (2001, p. 142). Although participants in a particular chat room might not constitute a traditional social group, they should, at some point, individually and collectively begin to produce discourse that is expected in this context or discourse that patterns itself after that of other participants. Before sending their first message, new participants may or may not spend time reading messages sent by other people, and it may take longer for some new participants to notice and appropriate the norms of the group. However, as is the case in most communication environments, participants can learn from the reactions of others which aspects, types, or elements of discourse are unexpected, inappropriate, or nonconforming. It is not clear if people who communicate on line could be viewed as members of a community of practice (Lave and Wenger, 1991), a social network (Paolillo, 2001), a speech community (Labov, 1972; Gumperz, 1972), a virtual community (Rheingold, 2000), or some combination thereof. It is our view that most on-line communication environments could be analyzed using any of these different frameworks, depending on the focus and purpose of the study. Given the type of analysis undertaken in the present study, we consider the participants to be part of a speech community since, within this framework, according to Davies (2005), linguistic data is ‘the prime locus of interest’ (p. 559). Bell’s general principle three states that ‘[s]peakers design their style primarily for and in response to their audience’ (p. 143), and his fourth principle refers specifically to the focus of our analysis: ‘Audience design does not refer only to style-shift. Within a language, it involves features such as choice of personal pronouns or address terms (Brown and Gilman 1960, Ervin-Tripp, 1972), politeness strategies (Brown and Levinson, 1987), use of pragmatic particles (Holmes, 1995), as well as quantitative style-shift (Coupland, 1980,1984)’ (p. 144). Our analysis focuses on the choice of personal pronouns, and we expect a clear pattern of indefinite pronoun use and variation to emerge, and such a pattern may or may not align with previous research on everyday, conversational spoken French. Although our analysis is somewhat limited by the lack of biographical information from the chat participants, there is a strong advantage to using this type of data since it is most likely representative of naturally occurring discourse in such an environment. There is some evidence that discourse produced by informants or Please cite this article in press as: Williams, L., van Compernolle, R.A., On versus tu and vous: Pronouns with indefinite reference ..., Lang. Sci. (2008), doi:10.1016/j.langsci.2007.11.001

ARTICLE IN PRESS 4

L. Williams, R.A. van Compernolle / Language Sciences xxx (2008) xxx–xxx

participants in other types of settings can be influenced by the interviewers or researchers themselves, the types of elicitation techniques they use, or the role(s) they play during observations or interactions when data is being collected (see Schilling-Estes, 1998; Bailey and Tillery, 1999; Wertheim, 2006), which is not an issue for this study since the researchers were not present in the virtual spaces where the communication was taking place. While we compare our findings to those of researchers who have studied spoken French, we also wish to establish a baseline for replication and comparison with future analyses of synchronous electronic French discourse. 2. Review of previous research The use of on versus tu/vous for indefinite reference has been the object of study of several articles over the past 30 years or so. We have selected four well known studies for review, each of which has served as a point of comparison and a basis for research questions and hypotheses to be explored in the present study: Laberge and Sankoff (1980), Ashby (1992), Coveney (2003), and Fonseca-Greber and Waugh (2003). We focus our review first on overall frequencies of on versus tu/vous use as pronouns with indefinite reference and social factors constraining this variable. We then discuss the various linguistic and discursive factors that contribute to the variation, around which our analysis is centered. 2.1. Frequencies and social factors In the earliest large-scale study of indefinite French pronoun use, focusing specifically on tu and vous versus on based on a corpus of interviews with 120 speakers of Montreal French, Laberge and Sankoff (1980) noted that although on has been traditionally viewed as the indefinite par excellence, it has lost ground to the pronouns tu and vous as indefinites that are used almost interchangeably with on and occur just as often as or, in some cases, much more frequently than on. ‘Tu and vous have taken up the challenge, and are now locked in combat with on in a bout for indefinite champion, a title on thought it had locked up’ (p. 271). Although some participants (5.8%) used tu and vous categorically, and others (15.0%) used on in all instances of indefinite reference, the majority (78.3%) alternated between the two variants, and one participant did not use any indefinite pronouns at all. In total, on was used in 47.9% of cases where tu/vous could have been used. Among participants who used both tu/vous and on, the use of on was much more likely for female participants under the age of 40, although age was not found to be a significant factor overall. It appears that younger men, having moved on into the traditional nous slot, have filled the on slot with tu and vous, creating a potential ambiguity with ‘‘personal” tu and vous. For younger women, however, the potential ambiguity is with the two uses of on . . . The fact that younger men show decreasing input probabilities for on as an indefinite indicates a change in progress among this segment of the population. It does not, however suggest that younger male speakers have difficulty understanding women and older men when they use on . . . or that these people cannot in turn correctly decode uses of indefinite tu or vous. Sankoff, 1980, p. 92. Findings by Thibault (1991), Blondeau (2001, 2007), and Blondeau et al. (2002), using additional Montre´al corpora confirm the widespread use and continued increase of generic second-person pronouns that was revealed by Laberge and Sankoff. Incidentally, although not a variationist study, the use of indefinite pronouns was also identified in Quebec City by Deshaies (1985), who demonstrated that this phenomenon is not limited to Montre´al. In a study similar to that of Laberge and Sankoff (1980), Ashby (1992) explored the variable use of on versus tu/vous with indefinite reference in a corpus of spoken French. His corpus came from interviews recorded in Tours (France) in 1976 and included eight men and eight women of various social situations and two age groups. The interviews were conducted in places familiar to the informants (e.g. home or workplace). Ashby acknowledged that his presence may have had an effect on the interviewees, describing the speech as ‘spontaneous, yet probably somewhat guarded, conversational style’ (p. 141). Although Ashby identified three types of indefinite on use (as well as its definite use, which varies with nous), his analysis focused on tokens of on that could co-vary with tu/vous in generalizations. Please cite this article in press as: Williams, L., van Compernolle, R.A., On versus tu and vous: Pronouns with indefinite reference ..., Lang. Sci. (2008), doi:10.1016/j.langsci.2007.11.001

ARTICLE IN PRESS L. Williams, R.A. van Compernolle / Language Sciences xxx (2008) xxx–xxx

5

Ashby reported that on was used 57.4% of the time for general-indefinite reference where tu/vous would have been a viable option, yet he noted that rates varied greatly from speaker to speaker (11% to 100%). Although both tu and vous occurred in Ashby’s corpus, vous was overwhelmingly preferred. Ashby attributed this preference for vous over tu as the indefinite second-person form to the fact that this was also the form used as a definite address pronoun between his informants and himself. This imbalance between tokens of tu and vous led Ashby to count the few tokens of indefinite tu with tokens of indefinite vous. Ashby found that younger speakers seemed to favor on, while older speakers tended to use vous more often. This finding does not support the assumption that on is the older, and therefore more established, genericindefinite pronoun. However, a VARBRUL analysis found age to be a statistically non-significant factor in his data, which is what Laberge and Sankoff (1980) had found as well. On the other hand, the data suggest that women and speakers belonging to the higher class tended to favor on, which led Ashby to consider the variability of on versus tu/vous use as a ‘linguistic feature subtly signaling a social distinction, rather than an ongoing change’ (pp. 154–155). He concluded that on and tu/vous are not free variants; ‘rather, they are variants of a sociolinguistic variable whose probability of occurrence is constrained by a complex of linguistic, discursive, and sociolinguistic factors’ (p. 156). Like Laberge and Sankoff (1980) and Ashby (1992), Coveney (2003) explored on and tu/vous with indefinite reference in a corpus of spoken French, with data from Picardy (northern France), in order to determine to what extent tu and vous have become acceptable substitutes for or variants of indefinite on. Coveney noted that ‘[i]n contrast with indefinite on, which is considered standard and is acceptable in all registers . . . the status of indefinite tu/vous is less clear-cut’ (p. 167), and very few grammars and dictionaries mention this usage of tu/vous. His data were produced by 30 speakers, each of whom participated in a ‘one-to-one conversational interview’ (p. 174) with a fieldworker, constituting a corpus of 17 h of spoken French. The informants were employees at several colonies de vacances (summer camps). Although these colonies de vacances provided an ‘ideal context in which to collect adults’ speech close to the vernacular’ (p. 173), only a limited number of older speakers—three women in their fifties—and speakers from working-class backgrounds—five, all under 23 years of age—were available for interviews. Following Ball (2000), Coveney investigated the suggestion that tu and vous are replacing on for indefinite reference because on ‘has become ‘‘overloaded”, and indefinite tu is used by some speakers to relieve it of one of its functions’ (p. 166), in addition to the hypothesis that indefinite on is associated with more formal discourse. This point will be re-introduced below in the analysis of the data used for the present study. Before exploring the distribution of indefinite pronouns, Coveney (2003) provided a brief explanation of the ‘three main values for on’ (p. 174): (1) on in variation with nous, (2) on in variation with indefinite tu/vous, and (3) on in variation with indefinite ils (p. 174). In addition, he recognized that a certain number of examples were ambiguous, that is, it was not perfectly clear whether on or tu/vous was being used for definite or indefinite reference. Ambiguous tokens could not, therefore, be included in the quantification of the variable. Coveney found that, although there were ambiguous examples of tu/vous, occurrences of on tended to be much more difficult to categorize as definite or indefinite. Nine hundred sixty-four instances (64.1%) of on were counted as ambiguous while only nine tokens (35.9%) of tu/vous fit into this category, which ‘in itself seems to support the hypothesis that the use of indefinite tu/vous is motivated partly by a wish to avoid ambiguity’ (p. 174). Coveney also noted that ‘[f]or clear-cut cases of on, it is often the presence of another pronoun that makes the reference unambiguous’ (p. 175). Coveney (2003) identified 1031 occurrences of general-indefinite pronoun usage and reported that on was used 52.3% of the time, while tu/vous accounted for 47.7% of indefinite tokens. These frequencies matched results reported in Ashby (1992) for Tours, as well as those reported in Laberge (1977) for Montre´al (see also Laberge and Sankoff, 1980). Coveney focused on interpersonal variation in order to ‘test . . . a number of hypotheses that may help to account for the use of indefinite tu/vous’ (p. 176). Individual scores for indefinite on use varied greatly in his corpus, ranging from 8.2% to 100%. Similar results have been reported by Laberge and Sankoff (1980) and Ashby (1992) (also referenced in Coveney (2003, p. 176)Coveney, 2003, p. 176). Coveney reported that speakers aged over 30 years tended ‘to use indefinite tu/vous a good deal more than their younger counterparts’ (p. 176), whereas the younger speakers showed a preference for indefinite on. He also noted that these results do not, however, ‘reflect a change back to the older variant on, nor even a straightforward pattern of age-grading without change’ (p. 177). Please cite this article in press as: Williams, L., van Compernolle, R.A., On versus tu and vous: Pronouns with indefinite reference ..., Lang. Sci. (2008), doi:10.1016/j.langsci.2007.11.001

ARTICLE IN PRESS 6

L. Williams, R.A. van Compernolle / Language Sciences xxx (2008) xxx–xxx

In a study also focusing on spoken French, Fonseca-Greber and Waugh (2003) analyzed selected pronouns—primarily tu, vous, nous, and on—from a corpus of 194,000 words in order to demonstrate a rather drastic divergence between (formal, literary) written French and ‘ordinary’ (p. 226) spoken French. The data in their corpus were produced during ‘15 conversations [with] 27 educated, middle-class speakers, 10 over 40, 17 under 40, 11 men and 16 women’ (p. 228). Their participants included ‘family members and friends [who were recorded] in natural settings’ (p. 228), some of whom were from France (14 speakers, 117,000 words) and others from Switzerland (13 speakers, 77,000 words). These two data sets—each originally its own corpus— were combined into a single corpus for the analysis since ‘[a] preliminary investigation of each of the corpora with respect to the issues discussed here showed that the everyday spoken language represented in the corpora was basically the same’ (p. 228). Fonseca-Greber and Waugh, following Offord’s (1990) recommendation, recognized that the varieties of French used in Belgium and Switzerland are essentially the same as that of metropolitan France. Indeed, the minor differences that do exist ‘are usually considered to be lexical or phonological, rather than morpho-syntactic or grammatico-semantic and this accorded with what our preliminary investigation showed’ (p. 228). Before focusing on the use of indefinite-generic pronouns, Fonseca-Greber and Waugh (2003) offered an overview of the replacement of nous by on (when on means we in English). According to their data, this is a change that has been completed in Switzerland (zero occurrences of nous), and even in the part of the corpus from France only 13 residual utterances—‘1% of the total first-person plural tokens with on’ (p. 229)—of nous were found, ‘and the bulk of these came from older, more conservative speakers, who are schoolteachers’ (p. 229). They then demonstrated the resulting shift in the paradigm of the various meanings of on: indefinite, 100 tokens (5.7%); vague, which could be interpreted as indefinite or personal, 314 tokens (18%); and personal, 1335 tokens (76.3%). These results indicate that as on-personal has replaced nous, a need for the indefinite use of on to be displaced or transferred, namely to the second-person pronouns tu and vous, has occurred, most likely gradually. Fonseca-Greber and Waugh (2003) thus found that tu and vous now produce the majority (69.3%) of indefinite meanings in everyday conversational European French. (Since only 1.3% of the tokens were vous, these were combined with the 68% of tu tokens, a practice used elsewhere in the literature.) These results indicated that, compared to tu/vous-indefinite, the indefinite use of on is at only 30.7%, which actually includes vague uses of on. ‘If we tally all 1350 tokens of indefinite meaning in the corpus, we find that on does not account for even a third of all indefinite meanings, even by giving it the benefit of the doubt and according it all of the vague tokens of on as well, so as not to overstate the extent of the change’ (p. 234). Results reported by researchers in the studies summarized above clearly indicate that tu and vous have become acceptable substitutes for indefinite on in both Canadian and European everyday conversational French. (Overall frequencies for indefinite on use when compared to tu/vous as reported in previous studies are provided in Table 1.) Laberge and Sankoff (1980), Ashby (1992), and Coveney (2003) have all reported rates of indefinite on use at or around 50% when compared to tu/vous. Fonseca-Greber and Waugh (2003), however, demonstrated that tu/vous was overwhelmingly preferred in their informal conversations with informants (nearly 70%). We may conclude that Fonseca-Greber and Waugh’s (2003) results suggest a change in progress, as their data represent a more recent sampling of everyday conversational French relative to Laberge and Sankoff (1980), who conducted interviews in the early 1970s, Ashby (1992), whose data were collected in the mid 1970s, and Coveney (2003), whose informants were recorded in the mid 1980s. Although the approaches and aims are rather diverse, these studies all have many points in common, especially regarding methodology of the type of data collected. Recorded interviews provided the data for each

Table 1 Rates of indefinite on use (versus indefinite T/V) reported in previous studies Study

Data

Laberge and Sankoff (1980) Ashby (1992) Coveney (2003) Fonseca-Greber and Waugh (2003)

Spoken Spoken Spoken Spoken

% indefinite on Montre´al French (early 1970s) French from Tours, France (mid 1970s) French from Picardy, France (mid 1980s) French from France and Switzerland (late 1990s)

47.9 57.4 52.3 30.7

Please cite this article in press as: Williams, L., van Compernolle, R.A., On versus tu and vous: Pronouns with indefinite reference ..., Lang. Sci. (2008), doi:10.1016/j.langsci.2007.11.001

ARTICLE IN PRESS L. Williams, R.A. van Compernolle / Language Sciences xxx (2008) xxx–xxx

7

corpus, and in all cases the interviewers knew the gender, (approximate or actual) age, and (approximate or actual) social class of the participants. Such is not the case for the present study, which derived all its data from network-mediated communication where users are able to remain more or less anonymous since information from and about these types of participants is not easily verifiable. In addition, the data for the present study are comprised written discourse, but not the formal, literary kind that Fonseca-Greber and Waugh (2003) referred to and used as a baseline for comparison with everyday spoken discourse. The present study attempts to answer the same overarching questions that provided the underpinning for the work on indefinite pronoun use summarized above. However, instead of focusing on a single variety of French spoken in any particular region or country, our data is a geo-social mix of Internet users who communicate in French. Although there exist many differences between this relatively new form of communication and more traditional types of discourse (e.g. speech and writing), one of our working hypotheses is that French-language synchronous chat, which we analyze in this study, shares many sociolinguistic and pragmatic features of discourse with everyday conversational French. Thus, given the apparent rise in the use of tu/vous for indefinite reference in everyday speech, it seems reasonable to presume that this increase should be reflected in our data as well. Since our data come from a non-traditional form of written (i.e. typed) discourse produced in an environment offering a great deal of anonymity, we cannot take into consideration many of the social factors shown to be significant in previous studies. Our analysis will, therefore, focus first on a comparison of rates of indefinite on versus tu/vous use between spoken French and synchronous chat. We will then explore the influence of a number of linguistic and discursive factors that contribute to the variation, which are summarized in the following section.

2.2. Linguistic and discursive factors Laberge and Sankoff (1980) created two discursive-pragmatic categories for classifying utterance containing indefinite referents: situational insertion and the formation of truisms and morals. (Examples from the corpus used for the current study are provided in Section 3.) Although these two categories may indeed seem very similar in some ways, Laberge and Sankoff proposed a clear distinction. Situational insertions are ‘utterances . . . involving an indefinite person, and they all have the effect of locating this person in a potentially repeatable activity or context. Anyone’s experience may constitute the basis for generalization, though most often it is the speaker’s’ (pp. 280–281). These types of utterances are based on personal experience, but they refer to events and actions that most people are familiar with, such as traveling, dealing with a health-related problem, studying for a test, driving a car, and so forth. The use of indefinite pronouns has less to do with specific contexts and more to do with the speaker’s and the interlocutor’s ability to imagine himself or herself in a given situation, regardless of the specific details of the event or action. Similar to situational insertions, morals and truisms necessarily invoke a certain type of event or action. However, ‘they also generally possess a strong judgmental or evaluative connotation. This is particularly clear when the utterance is a saying or proverb overtly borrowed from the oral tradition of the community’ (p. 282). Laberge and Sankoff’s (1980) analysis showed that a large part of indefinite pronoun tokens were ‘in the more general ‘‘situational insertion” category (82.3% as opposed to 17.7% ‘morals’)’ (p. 283). In utterances determined to be formulations of morals or truisms, they found that 65.2% used indefinite on, almost twice the rate of indefinite tu/vous use (34.8%). However, the distribution of indefinite tu/vous versus indefinite on was almost equal in utterances classified as situational insertions: 54% for tu/vous and 46% for on. Ashby (1992) also examined discursive effect of generalization on generic-indefinite on and vous use. However, Ashby did not use Laberge and Sankoff’s (1980) categorization of truisms and morals, since he considered ‘existentials involving on and vous . . . to have a specific discourse function, namely to introduce new referents into the flow of discourse’ (p. 147). Ashby therefore provided four types of discourse function for indefinites: referent introduction, evaluation, situational insertion, and exposition. The new category for expositions includes tokens that would have been situational insertions for Laberge and Sankoff. However, Ashby ‘included in the ‘‘situational insertion” type only those tokens which involve a generalization clearly based on the speaker’s personal experience. Reference to situations based on the experience of another party or on common knowledge [were] classified as ‘‘exposition”’ (p. 151). A VARBRUL analysis determined that on was Please cite this article in press as: Williams, L., van Compernolle, R.A., On versus tu and vous: Pronouns with indefinite reference ..., Lang. Sci. (2008), doi:10.1016/j.langsci.2007.11.001

ARTICLE IN PRESS 8

L. Williams, R.A. van Compernolle / Language Sciences xxx (2008) xxx–xxx

favored in evaluations (91%) (similar to Laberge and Sankoff’s truisms and morals category) and expositions (74%), while it was disfavored with referent introduction (19%) and situational insertions (41%). Syntax, too, appears to influence the distribution of pronouns with indefinite reference. Laberge and Sankoff (1980) indentified three syntactic frames: implicative constructions, presentative-headed constructions, and generalizations with indefinite pronoun use. Implicative constructions are those that express cause and effect. ‘They consist of two sentences, the first of which (the protasis) sets up a supposition and the second (the apodosis) states the implications. Conditionals with if are common examples of implicative constructions’ (p. 277). Constructions headed by a presentative notify interlocutors that what is about to be said contains ‘a generally admitted truth, or a personal opinion that speakers hope are shared, if not universally, at least by their interlocutors’ (p. 277). Some common examples of French presentative-headed constructions are the following: il me semble ‘it seems to me’, je pense ‘I think’, disons ‘let’s say’, and so forth. ‘Indefinite referents in these cases occur not in the presentative clause itself, but rather in the following sentence’ (p. 278). Laberge and Sankoff (1980) reported that, when indefinite uses of on and tu/vous were combined, implicatives made up 38% of situational insertions, yet this rose to 46% in the formulation of morals/truisms. There was also an increase of presentative-headed constructions, from 10% in situational insertions to 21% in the formulation of morals/truisms. The higher percentages of implicatives and presentative-headed constructions in the formulation of morals/truisms resulted, necessarily, in a decrease of indefinite pronoun use in the broad category generalizations, from 52% to 33%. In sum, their analysis demonstrated that ‘on is very much preferred in presentative-headed constructions (65.8% cases versus 34.2% of tu and vous), and dispreferred in implicatives’ (p. 285). In addition, the distribution in the discursive-pragmatic categories showed a striking preference for the use of on in the formation of morals/truisms. Ashby (1992) also explored generic-indefinite on and vous use in several syntactic frames. In presentative heads, for example, on was favored (91%), as Laberge and Sankoff (1980) had found in Montre´al, whereas on was disfavored in the majority of conditional structures (see Ashby, 1992, p. 145). Ashby also analyzed three measures of transitivity, based on Hopper and Thompson’s (1980) model: kinesis (i.e. higher versus lower), mode (i.e. specific versus non-specific event), and affirmation (i.e. affirmative versus negative utterance). (Examples from the corpus used for the current study are provided in Section 3.) According to Ashby, ‘Hopper and Thompson (1980) describe transitivity as a global property of a clauses [sic] that is determined by a complex of semantic, syntactic, and pragmatic features. Clauses with high transitivity tend to occur in foregrounded portions of discourse, those with low transitivity tend to occur in backgrounded portions’ (Ashby, 1992, p. 151). Within Hopper and Thompson’s model, high transitivity is indicated by specific or ‘‘real” events, by affirmative (rather than negative) verbs, and by verbs that are more ‘‘active” than ‘‘stative” (see Ashby, 1992, pp. 151–153). According to Ashby, ‘most generalizations of the type involving indefinite-generic on and vous are based on non-specific situations or events. But when the generalization is based on a specific event, . . . [tu or] vous [have] a much higher probability of occurrence than on . . .’ (p. 152). Ashby uses the example of a person’s recollection of taking the French baccalaure´at exam to demonstrate a specific event, as opposed to taking exams in general (i.e. non-specific). Affirmative clauses tend to have higher transitivity, too, and they often refer to more specific events, while negative utterances are associated with less specific events. In Ashby’s (1992) analysis, VARBRUL weights for transitivity measures produced significant results only when comparing specific events (vous favored) versus non-specific events (on favored). Although the results for kinesis and affirmation were not statistically significant, Ashby argues that the distribution of indefinite pronouns was formed in the right direction, as clauses with lower kinesis and negative utterances were associated with the use of on, while indefinite on was used less frequently in clauses having higher kinesis and affirmative clauses. Coveney’s (2003) analysis focused primarily on co-variance of on  tu/vous with other features of discourse, rather than focusing on the various discursive-pragmatic and syntactic factors investigated by Laberge and Sankoff (1980) and Ashby (1992). Coveney (2003) explored, for example, instances of ambiguous on in order to determine whether informants with higher rates of indefinite on also produced more instances of ambiguous on, and, on the other hand, if informants showing a preference for indefinite tu/vous had lower rates of ambiguous on (i.e. indefinite tu/vous use might be motivated by a desire to avoid ambiguity or confusion). He found that speakers with high rates of ambiguous on (30 or more per 5000 words) ‘also have very high frequencies for Please cite this article in press as: Williams, L., van Compernolle, R.A., On versus tu and vous: Pronouns with indefinite reference ..., Lang. Sci. (2008), doi:10.1016/j.langsci.2007.11.001

ARTICLE IN PRESS L. Williams, R.A. van Compernolle / Language Sciences xxx (2008) xxx–xxx

9

Table 2 Linguistic constraints on the on  tu/vous variable Factor

on favored

tu/vous favored

Syntactic frame Discursive effect Transitivity

Generalizations Morals and truisms (evaluation) Non-specific event Negative sentence High rate of ambiguous on Ne retention Definite second-person avoided with interlocutor

Implicatives Situational insertions Specific event Affirmative sentence Low rate of ambiguous on Ne deletion Definite second-person used with interlocutor

Ambiguity Linguistic conservatism Ease with address

indefinite on (at least 80%)’ (p. 179), whereas speakers ‘with the lowest rates for ambiguous on . . . tend to use tu/vous’ (p. 179). Although only moderate, the correlation between the preferred indefinite pronoun and ambiguous on supports the hypothesis that ‘the more an individual uses indefinite tu/vous (in preference to on), the fewer instances of ambiguous on they will produce in their discourse’ (p. 179). Another factor investigated by Coveney (2003) was ‘the ‘‘conservatism” and/or formality of the speech of the various informants’ (p. 179). Working under the hypothesis that indefinite on is associated with more formal discourse (i.e. indefinite tu/vous are the less formal variants), Coveney examined the correlation between indefinite on use and another variable normally associated with formality: the retention of the negative particle ne, which he had previously analyzed for the same corpus (see Coveney, 1996). His analysis suggests that, to some extent at least, ‘there is some positive correlation between ne retention and on use’ (p. 180). He noted, however, that this positive correlation was not strong and did not ‘indicate an entirely straightforward relationship between these two variables’ (p. 179) since ‘the (on  tu/vous) variable is subject to some influence that does not come into play with a variable such as (ne)’ (p. 180). Coveney (2003) also investigated the use of address pronouns in order to determine the extent to which the use of definite tu/vous influences the use of indefinite tu/vous, that is, ‘the more frequently an informant uses tu/vous for address . . . the more s/he will also use that pronoun for indefinite reference’ (p. 184). In his corpus, speakers who used definite tu/vous at a fairly high rate ‘also used indefinite tu/vous more often than on’ (p. 185). However, informants who had low rates of definite tu/vous (i.e. they did not often address the fieldworker) produced higher rates of indefinite on. Coveney thus found that ‘it is reasonable to conclude that these speakers who strongly preferred indefinite on rather than tu/vous during the interview did so because they also avoided using address pronouns on this occasion’ (p. 185). He added, however, that ‘it is entirely possible that these individuals [who avoided using address pronouns and preferred indefinite on to tu/vous] do actually use indefinite tu/vous in their everyday language’ (p. 186). He also noted that a number of other factors ‘undoubtedly affect the frequency of use of both definite and indefinite tu/vous, including conversational structure for the former and topic for the latter’ (p. 186). Another determining factor suggested by Coveney is the informant’s ‘ease with his or her choice of address pronoun’ (p. 187). In sum, a number of linguistic and discursive factors contribute to variation in the use of on versus tu/vous for indefinite reference, which are presented in Table 2. While Coveney’s (2003) analysis of ambiguous on tokens, co-variance with ne retention/deletion, and ease of address provides an important contribution to our understanding of the on  tu/vous variable, the nature of our data is such that these types of inter-individual correlations cannot be performed. Instead, we have centered our analysis around the influence of syntactic frame, discursive effect, and transitivity. In line with Laberge and Sankoff (1980) and Ashby (1992), we expect that on will be preferred in generalizations, during the telling of morals and truisms, with references to non-specific events, and in negative utterances. Conversely, we anticipate that tu/vous will be favored with implicatives and situational insertions, in references to specific events, and in affirmative clauses. 3. Method The data sets presented in Table 3 were randomly selected for this analysis from a larger corpus including several additional channels. These four data sets come from public chat channels on different networks, and Please cite this article in press as: Williams, L., van Compernolle, R.A., On versus tu and vous: Pronouns with indefinite reference ..., Lang. Sci. (2008), doi:10.1016/j.langsci.2007.11.001

ARTICLE IN PRESS 10

L. Williams, R.A. van Compernolle / Language Sciences xxx (2008) xxx–xxx

Table 3 Composition of the corpus Data set #

Turns

Words

Emoticons

Total (words + emoticons)

1 2 3 4 Total

2591 13,826 5454 10,025 31,896

14,216 71,656 22,746 46,255 154,873

226 561 899 2096 3782

14,442 72,217 23,645 48,351 158,655

the data were collected at regular intervals from each channel over a 2-year period in order to lessen any influence that a particular participant or participants may have had on the overall distribution of on, tu, and vous if the data had been collected during a shorter period of time and from fewer sources. Days of the week and times of day were rotated for the data collection process in order to increase the probability that the corpus is representative of synchronous French chat. It is important to note that, although we have no evidence that any particular speaker produced a disproportionate number of tokens of one or more pronouns, it is impossible to know the true identity or name of any participants since the use of pseudonyms is common, and participants can change their pseudonym as often as they wish, unbeknownst to others (Bechar-Israeli, 1995; Paolillo, 1999). Nonetheless, we are confident that the corpus is representative enough to provide an overview of this type of discourse given the rather wide-ranging period of time during which data were collected in addition to the size of the corpus and its comparability to another analysis of synchronous electronic French discourse (Williams and van Compernolle, 2007). In addition, the number of indefinite tokens of on, tu, and vous per data set happens to correspond proportionally to the number of words per data set in relation to the entire corpus. This means that, for example, Data Set 3, with 22,746 words is approximately 15% of the corpus, and this same data set provided around 15% of the indefinite tokens. Such is also the case for the other data sets. Each data set represents a chat channel described on its network or site as being intended for general discussion or any/all topics and open to the general public. No channels had a specific theme or topic of discussion, and none required membership information or fees. In spite of this, there were many lengthy discussions about specific topics in these rooms, but at other times topic shifts occurred in rapid succession, and in many instances there was evidence of several simultaneous conversational topics/strands, which is not uncommon in synchronous (i.e. real-time) electronic discourse (Herring, 1999). Although chat discourse is a written form of language, a number of modifications occur that are either similar to spoken discourse or unique to electronic discourse. Therefore, establishing the number of words in any corpus of electronic discourse can become rather complicated since a grouping of a few letters can represent two, three, or even more words that would normally be counted individually in a corpus of written discourse. One such example is tlm, the initial letters representing three French words, tout le monde, translated into English as everyone or everybody. In this case, there is a clear one-to-one relationship between the abbreviation and the full expression since each letter of the shortened form represents one word. Other terms such as bjr (an abbreviated form of bonjour) and lut (an aphetic form of salut) are also relatively easy to count because they represent only one word. However, something such as che´pa represents either three or four words, depending on how it is analyzed. This term is a representation of the spoken French form je sais pas (which would, in a more formal, emphatic, and/or marked context, have an extra negative particle—ne—before the verb sais) when je and sais are pronounced together very quickly to produce che´, and then pa (the pronounced segment of pas) is added as if this unit had become lexicalized for the chat participant. Since this is most likely a strategy used to increase efficiency, thereby reducing the time delay between mentally preparing the message, typing it, then finally sending it, such terms are quite common in many forms of electronic discourse, and we have therefore decided to count only the words that are represented in such abbreviated forms. In this case specifically, the pre-verbal negative particle ne is not represented anywhere in che´pa, and since this particle can also be deleted in spoken French (and also in traditional written French, which is less common), we have not added it to the expression che´pa during the counting process. Another issue related to establishing the number of words in a corpus of any type of discourse, whether spoken, written, or electronic, involves compound forms and fixed expressions. One example that illustrates this problem well is peut eˆtre vs. peut-eˆtre. The former is a combination of a modal verb plus a main verb Please cite this article in press as: Williams, L., van Compernolle, R.A., On versus tu and vous: Pronouns with indefinite reference ..., Lang. Sci. (2008), doi:10.1016/j.langsci.2007.11.001

ARTICLE IN PRESS L. Williams, R.A. van Compernolle / Language Sciences xxx (2008) xxx–xxx

11

(translated in English as can be), and the latter is a compound form, joined by a hyphen, meaning perhaps or maybe in English. In this and similar cases, we decided to count the former as two words and the latter as one word. For instances of a misspelling (e.g. a hyphen-related error) or an abbreviated form (e.g. pe representing peut-eˆtre), meaning was used to determine, for counting purposes, what the accurate or full traditional form would be. Fixed expressions pose a similar potential problem; however, since at least one or more words in any fixed expression can exist and function independently, we chose to count each word in any fixed expression individually. Emoticons such as characters representing smiley faces, sad faces, kisses, and other non-lexical expressions of emotion were counted separately. These signs and symbols have become an important part of electronic discourse since they allow participants to replace some of the missing contextual and prosodic clues that would often be present in viva voce communication, and they are an especially efficient way of conveying elements of the message that would otherwise be too inefficient to express in full written form. In our corpus, approximately 10% of all turns have an emoticon, which demonstrates their utility. We have therefore included them in an effort not only to provide a complete description of the corpus, but also to recognize their importance and encourage authors of future studies of electronic discourse to do the same. An analysis of these emoticons is beyond the scope of the present study; however, we feel strongly that it is important to note that they play an integral part in electronic communication, so much so that many turns actually consist solely of an emoticon. Data collection involved a variety of procedures and techniques since there exist different types of interfaces for accessing public chat environments. Almost half of the data (Data Sets 3 and 4) in our corpus comes from chat channels that require a specific type of software (widely available as freeware or shareware) developed for accessing Internet Relay Chat (IRC) networks and servers (e.g. Epiknet, Dalnet, IRCnet, and so forth2) hosting any number of chat areas or rooms that are not accessible using a Web browser as an interface.3 Data collection is relatively easy for this type of chat environment since the software can, in most cases, save the transcript (or log) from one or more rooms as a text file that can be retrieved at any time and used in a word processing or concordance program. Some IRC servers also have an accompanying Web site where anyone can download transcripts, view participation rates and other statistics, and find more information about the server and its administration. The other parts (i.e. just over half) of the data in our corpus (Data Sets 1 and 2) come from chat rooms hosted on networks that allow the participant to use a Web browser as an interface, which means that no special software is required. However, the data collection can be more complicated since it may or may not be possible to select/highlight, copy, then paste (into word processing software) the text of the chat displayed on the screen. In some cases, a screen capture has to be taken at regular intervals as the new messages eventually replace all the previous messages in the viewable area (i.e. on the screen). For both types of data, each token of tu, vous, and on was reviewed individually in order to establish the overall distribution. Tu, vous, and related forms were first coded for grammatical function (i.e. subject vs. non-subject), and then all tokens of on, tu, and vous were coded for type of reference (i.e. specific/definite vs. generic/indefinite). In order to code for a VARBRUL analysis using Goldvarb X (Sankoff et al., 2005), tokens of indefinite tu and vous (hereafter referred to as tu/vous since they were combined for comparison with on) were labeled according to the following factor groups, with the first of these representing the dependent variable and the others, the independent variables: (1) generic-indefinite vs. definite on or tu/vous; (2) syntactic frame (i.e. generalization or implicative); (3) discursive/pragmatic effect (i.e. moral/truism or situational insertion; (4) mode: specific or non-specific event (as a measure of transitivity); and (5) affirmation or negation (as a second measure of transitivity). Although Ashby also included kinesis as a measure of transitivity, we did not code our data for this type of variation since it was impossible in many cases to make a clear distinction between‘more active, kinetic verbs [and] more stative, less kinetic verbal locutions involving avoir. . .’ (Ashby, 1992, p. 152). The factor groups for syntactic frame and discursive-pragmatic effect are based on the work of Laberge and Sankoff (1980), and the factor groups for transitivity (mode and affirmation) are based on the analysis by 2

Information about Internet Relay Chat can be found at the following URL: http://www.irchelp.org/. Readers who would like more information regarding the terminology of new technologies should consult Webopedia (http:// www.webopedia.com/), TechWeb’s TechEncyclopedia (http://www.techweb.com/encyclopedia/), or any similar on-line or printed resource. 3

Please cite this article in press as: Williams, L., van Compernolle, R.A., On versus tu and vous: Pronouns with indefinite reference ..., Lang. Sci. (2008), doi:10.1016/j.langsci.2007.11.001

ARTICLE IN PRESS 12

L. Williams, R.A. van Compernolle / Language Sciences xxx (2008) xxx–xxx

Ashby (1992). Although Ashby had expanded the previous coding scheme and analysis offered by Laberge and Sankoff, we were unable to replicate Ashby’s categories exactly due to the smaller number of indefinite on, tu, and vous tokens in our corpus, which is approximately the same size as that of Fonseca-Greber and Waugh (2003). Examples 3–8 are turns that are representative of those we coded for each category. No modifications were made to the non-traditional French orthography in any of these examples, and no examples are provided for affirmation and negation since the absence or presence of a negative particle (e.g. pas, rien, personne) was the only criterion used for coding each token for this measure of transitivity. Example 3. Syntactic frame: generalization hThe´Oi on peut tj espe´rer [‘one can always hope’] Example 4. Syntactic frame: implicative hBadrai kan tu voi une dizaine de ops [‘when you see ten or so chat channel operators/monitors’] ... hBadrai ta care´ment peur de dire un mot lol [‘you are totally afraid to say anything lol’] Example 5. Discursive-pragmatic effect: moral or truism hFlonFlon22i qd on rentre sur un chan on dis bonsoir avt de prendre en pv [‘when one enters a chat channel one says ‘good evening’ before going into private chat’] Example 6. Discursive-pragmatic effect: situational insertion hQuiKii Tu prends ca une fois et tu est rassasie´ pour la vie [‘you have that once and you are set for life’] Example 7. Transitivity: non-specific event hDoLoSoi quand on voyage seule fo faire attention [‘when one travels alone one has to be careful’] Example 8. Transitivity: specific event hTaDapi quand tu vas a Pau tu manges super bien [‘when you go to Pau you eat really well’] In the remainder of this article, we compare rates of on versus tu/vous use for indefinite reference with those reported in previous studies. We then present the results of a VARBRUL analysis, which considered the factors highlighted above in examples 3–8 as well as affirmation/negation. In concluding, we discuss the shifting French pronoun paradigm, taking into account new or different emerging systems of address and reference in computer-mediated communication, and we provide possible directions for future research. 4. Results and analysis Table 4 shows the distribution of all (i.e. definite and indefinite) forms of tu, vous-singular (V-sg), vous-plural (V-pl), and on. The tokens of on are all subjects since that is the only function on can have in French. The overwhelming preference for tu (5423 tokens or 98.02%) compared to V-sg (52 tokens or 1.08%) matches the proportional distribution of tu (which is singular in French) and V-sg found in the analysis of synchronous electronic French by Williams and van Compernolle (2007), which reported tu use as high as 98.53% and 99.02% when contrasted with V-sg in non-emphatic and non-ludic stretches of discourse. (Emphatic and ludic stretches of discourse include role-playing or imitating part of a radio or television advertisement or imitating someone such as a butler or a member of the nobility). Since most of the non-subject V-sg tokens in our corpus co-occurred in the same turns with subject V-sg tokens, the rate of tu compared to V-sg in our data would increase to just over 99% if only subject and imperative forms were used to calculate the percentage. Given this almost categorical preference for tu over V-sg, our analysis collapses indefinite tu and vous into the same category for comparison with indefinite on. Collapsing the second-person pronouns has been done in the previous studies on generic on, tu, and vous use in French since each corpus has produced a majority of either tu or vous tokens. It is worth noting that the four tokens of indefinite V-sg were produced by participants who also—at least initially—used V-sg as a form of address. This is one of the few pieces of the analysis where specific tokens/ Please cite this article in press as: Williams, L., van Compernolle, R.A., On versus tu and vous: Pronouns with indefinite reference ..., Lang. Sci. (2008), doi:10.1016/j.langsci.2007.11.001

ARTICLE IN PRESS L. Williams, R.A. van Compernolle / Language Sciences xxx (2008) xxx–xxx

13

Table 4 Distribution of all definite and indefinite tokens of tu, V-sg, V-pl, and on Pronoun

No. of tokens

Percentage

tu (subject) tu form (non-subject) tu form of imperative verb Total tu forms

2716 2118 589 5423

50.08% 39.06% 10.86% 100.00%

V-sg (subject) V-sg form (non-subject) V-sg form of imperative verb Total V-sg forms

30 21 1 52

57.69% 40.39% 1.92% 100.00%

V-pl (subject) V-pl form (non-subject) V-pl form of imperative verb Total V-pl forms Total on forms Combined total

156 225 232 613 553 6641

25.45% 36.70% 37.85% 100.00% 100.00% 100.00%

turns can be linked to specific participants since many chatters only appear for a few seconds because they decide to go elsewhere or take a break, and others have computer-related problems resulting in disconnection from the network, or any number of other situations that might occur. These four tokens of indefinite V-sg happened to be produced relatively soon after the participants had entered the room, at the same time when these people were still using V-sg to address individuals in the room. All four made the switch to tu as a form of address rather quickly, but none happened to produce subsequent tokens of indefinite on, tu, or vous, which would have allowed us to determine if their use of pronouns with indefinite reference had some relationship to their preference for one address pronoun over another. Of the 553 forms of on found in our corpus, 29.7% (164) were generic-indefinite, and of these indefinite tokens, 116 were of the type of on that can alternate with indefinite tu/vous, as shown in Table 5, which also includes the figures for indefinite tu/vous tokens. All tokens with an ambiguous type of reference and selected discourse particles or fixed expressions including on, tu, or vous were excluded from the analysis, following Coveney (2003) (174-5, 181), since certain terms and expressions use on, tu, or vous categorically and do not alternate (e.g. tu vois ‘you see’ or tu sais ‘you know’). As mentioned earlier, Coveney identified 964 tokens of on (64.1% of indefinite and possibly indefinite on) as ambiguous for definite versus indefinite reference and only nine tokens of tu/vous (35.9% of indefinite and possibly indefinite) as ambiguous. In the present study, the percentages of ambiguous tokens of on (7.9%) and T/V (3.7%) are much lower than those found by Coveney; however, the ratio (i.e. twice as many ambiguous on tokens) is approximately the same (see Table 5). Given the reduced amount of prosodic extra-linguistic features and information available to (inter)locutors, it seems logical to posit that in a text-based communication environment, participants take extra care to ensure that there Table 5 Distribution of definite, indefinite, and ambiguous tokens of on, tu (T), and vous (V) Pronoun

No. of tokens

Percentage

T/V-definite T/V-indefinite (on) T/V-ambiguous on-definite on-indefinite (other than T/V) on-indefinite (T/V) on-ambiguous T/V-indefinite (on) T/V-ambiguous on-indefinite (all) on-ambiguous

2899 (2716 T; 183 V) 78 (74 T; 4 V) 3 375 48 116 14 78 3 164 14

97.3% 2.6% 0.1% 67.8% 8.7% 21.0% 2.5% 96.3% 3.7% 92.1% 7.9%

Please cite this article in press as: Williams, L., van Compernolle, R.A., On versus tu and vous: Pronouns with indefinite reference ..., Lang. Sci. (2008), doi:10.1016/j.langsci.2007.11.001

ARTICLE IN PRESS 14

L. Williams, R.A. van Compernolle / Language Sciences xxx (2008) xxx–xxx

could be little ambiguity or possibility for misinterpretation, at least regarding definite versus indefinite reference. This supposition is only speculative and certainly deserves treatment in future research since it is clear that even in some text-based communication environments, participants create new ways of expressing and interpreting prosodic features and extra-linguistic information. It is already obvious on this first level of analysis, based on the figures in Table 5, that our percentages are approximately the inverse of those reported by Fonseca-Greber and Waugh (2003), who found that the use of indefinite tu/vous was at 69.3%, with indefinite on at just over 30% in conversational European French. Their findings were similar to those of Deshaies (1985), who found indefinite tu/vous produced by adolescents in Quebec City at a rate of over 75%. However, our data are, overall, aligned more closely with those of Ashby (1992) and Coveney (2003), who reported that indefinite tu/vous was used 43% and 47.7% of the time, respectively, compared to indefinite on. The data reported by Laberge and Sankoff (1980) showed an approximate 50–50% split between on and tu/vous. These percentages of the general distribution of indefinite on and tu/vous need to be interpreted within the context of a wider view of the distribution of second-person pronouns since, as Laberge and Sankoff note, ‘[i]t is, of course, the case that the distinction between tu and vous as terms of address is interactionally significant in French’ (p. 287). For instance, although our data align with those of Ashby (1992) at the first level of analysis (i.e. overall distribution of indefinite on vs. tu/vous), participants used tu as a form of address (with each other) over 98% of the time in our chat corpus, whereas all of Ashby’s interviewees used vous when addressing him, and this is most likely why he only found 10 tokens of indefinite-generic tu in his data. He notes that ‘[i]n the case of the interviews [with three specific speakers], other discourse participants (peers of the speakers) were present, which may have triggered the sporadic use of indefinite tu instead of vous . . .’ (p. 143). Laberge and Sankoff, however, found more variation or instability when comparing the second-person pronoun used to address the interviewer and the one used for indefinite reference. Of the 60 women, 54 addressed the interviewer as vous; of the 60 men, only 25 did. (This includes nine people who alternate between tu and vous.) Moreover, the unanswered question of the relationship between the form of address used to one’s interlocutor and the form of the indefinite used is a complicated one. Certainly there is no one-to-one relationship. Fifteen people who restricted themselves to vous in addressing the interviewer never used vous as an indefinite; a further 19 such speakers used both tu and vous as indefinites. One interesting fact, however, was that none of the 28 speakers who used only tu in addressing the interviewer used only vous as an indefinite (a few of them used both). pp. 287–288 The apparent lack of consistency and stability led Laberge and Sankoff to conclude that all three forms were indeed viable options since they ‘function quite equivalently in the most semantically empty contexts, where even on is so impersonal as to be serving more as a dummy element than anything else’ (p. 288). Nonetheless, the majority of interviewees who used a specific second-person pronoun as a form of address used the same one for indefinite reference. In Fonseca-Greber and Waugh’s (2003) study, tu was used as the secondperson pronoun to address interviewers, and they found that tu was also the preferred indefinite form, at 68% of all indefinite pronouns. It is noteworthy that only 1.3% of all indefinite pronoun use reported by Fonseca-Greber and Waugh included tokens of vous (which can be plural), and the authors of the study attribute this to the presence of more than one interviewee in many contexts when data was collected. The data from the present study suggest that, within this second level of analysis, our data are most closely aligned with those of Fonseca-Greber and Waugh since in both cases the use of tu as an address pronoun was virtually categorical. Likewise, Coveney (2003) found high rates of tu as a personal address pronoun, and his data suggest that, similar to the findings of Fonseca-Greber and Waugh and the current study, the form of address pronoun is the same used for indefinite reference. There exists, thus, a clear trend for the dominant second-person address pronoun to be the same one used for indefinite reference among on, tu, and vous. Given the typically younger demographics of chat participants (see Pierozak (2003)), we had expected to find a disproportionately higher number of tokens for indefinite tu/vous compared to indefinite on since this was the case in the analysis by Fonseca-Greber and Waugh (2003), who had interviewed family members and close friends. Of course it is impossible to ignore findings by Laberge and Sankoff (1980) and Coveney (2003) who found some evidence that indefinite tu and vous are not necessarily always preferred by younger speakers simply because on is considered the more standard and older form. One explanation for our relatively high Please cite this article in press as: Williams, L., van Compernolle, R.A., On versus tu and vous: Pronouns with indefinite reference ..., Lang. Sci. (2008), doi:10.1016/j.langsci.2007.11.001

ARTICLE IN PRESS L. Williams, R.A. van Compernolle / Language Sciences xxx (2008) xxx–xxx

15

percentage of indefinite on might be that tu as a personal address pronoun in synchronous French chat is simply overloaded functionally, which means that indefinite on has been collectively adopted as a clear alternative. The comparison of tu and on from two corpora of approximately the same size in Table 6 illustrates this point. In our corpus, the overall proportional use of tu (83.2%) compared to on, is much higher than the rate found in the Fonseca-Greber and Waugh corpus (59.7%). For our third level of analysis, Goldvarb X was used to run a VARBRUL analysis for a comparison of our results with those of Laberge and Sankoff (1980) and Ashby (1992). Overall, the data suggest that indefinite on will be used instead of indefinite tu/vous with a probability of 0.63 (on a scale of 0.00–1.00). A step-up/stepdown analysis was then done in order to test for statistical significance and to determine if there was any interaction between or among any factors. The results of the syntactic frame analysis are displayed in Table 7. Laberge and Sankoff (1980) had presentative head as an additional category in their analysis of the syntactic frame, but since we found only one example of a presentative head (with indefinite on), this was excluded from the coding so that we could perform a VARBRUL analysis, which would not have been possible if presentative head had been a knockout factor with zero tokens (in the tu/vous column). The score produced by Goldvarb X was found to be significant, and the probability of the application of the rule is quite favorable (i.e. that on will be used) for generalizations, and the opposite is the case for implicatives. These results are similar to the totals of Laberge and Sankoff (1980), who found that, overall, on is more commonly used for generalizations and tu/vous is more often used for implicatives. It should be noted that although their data suggested that on was preferred for generalizations, the difference was only slight, with 51.1% (1075 tokens) for on and 48.9% (1029 tokens) for tu/vous, which is almost a 50–50% split, whereas our data produced a 71–29% division in favor of on. As mentioned above, we believe that this increase in indefinite on use is due to the functional overload of tu as a personal address pronoun, which results in its decreased capacity to serve also as an indefinite pronoun. As for the implicatives category, our percentage of on (36.5%) is very similar to the percentage of on reported by Laberge and Sankoff (41.6%). The overall distribution of indefinite on, tu, and vous in Laberge and Sankoff’s study and in ours appears to be quite similar, which is also the case for the results of their VARBRUL analysis. Although we were not able to include the category presentative head, both studies found that implicative structures disfavor indefinite on and that generalizations favor indefinite on. The results of the coding for discursive-pragmatic effect are provided in Table 8. The Goldvarb score was also found to be significant for this factor group, and the score indicates that the application of the rule (i.e. the use of indefinite on) is probable at .65 on a scale of 0–1. These results also demonstrate the same patterns found by Laberge and Sankoff (1980) since the use of indefinite on is also heavily favored in their data for Morals and Truisms (65.25% on, 34.75% tu/vous). Although our percentage of indefinite on is slightly higher, at 73.7%, this can again be explained by the fact that on use in our data was so much more common that tu/vous. In the category situational insertion, Table 6 Comparative distribution of tu and on Pronoun

Present study

Fonseca-Greber and Waugh (2003)

tu on

83.4% (2716) 16.6% (539)

59.7% (2126) 40.3% (1435)

Table 7 Syntactic frame Syntactic frame

No. of on/total

% of on

Goldvarb score

Generalizations Implicatives

93/131 23/63

71.0% 36.5%

.63 .24

Rangea a

.39

The range is the difference between the lowest and the highest VARBRUL score in each factor group.

Please cite this article in press as: Williams, L., van Compernolle, R.A., On versus tu and vous: Pronouns with indefinite reference ..., Lang. Sci. (2008), doi:10.1016/j.langsci.2007.11.001

ARTICLE IN PRESS 16

L. Williams, R.A. van Compernolle / Language Sciences xxx (2008) xxx–xxx

Table 8 Discursive-pragmatic effect Function

No. of on/total

% of on

Goldvarb score

Morals and truisms Situational insertion

70/95 46/99

73.7% 46.5%

.65 .35

Range

.30

Table 9 Transitivity: affirmation Phrase type

No. of on/total

% of on

Goldvarb score

Negative Affirmative

19/21 97/173

90.5% 56.1%

.87 .44

Range

.43

indefinite on was found to be used 45.6% of the time by Laberge and Sankoff, which is a difference of less than one percent compared to our data (46.5%). Given the similarity of the distributions in the variables of this factor group, it is not surprising that the VARBRUL results are nearly identical. The VARBRUL scores reported by Laberge and Sankoff are 0.63 for morals and truisms and 0.37 for situational insertion. The results for affirmatives vs. negatives are presented in Table 9. These Goldvarb scores also turned out to be significant, and the score of .87 produced for affirmation (one measure of transitivity), a category proposed by Ashby (1992) based on the work of Hopper and Thompson (1980), indicates that there is a very high probability that the rule will be applied (i.e. that indefinite on will be used) in negative structures. It is worth noting that only three examples of two-particle (i.e. formal and/or emphatic) negation were found in our data among the tokens of indefinite on, tu, and vous, and they were collapsed with the single-particle (i.e. less formal and more typical of everyday conversation) tokens of negation.4 The examples of two-particle negation in our data all co-occurred with indefinite on, which could be expected since on has traditionally been considered to be an older and more formal indefinite pronoun than tu or vous just like two-particle negation. The VARBRUL score indicating a very high probability that the rule will be applied to negative structures (i.e. that indefinite on will be used) is likely due to the fact that negation co-occurred so often with generalizations and morals/truisms, two factors which already had very high rates of indefinite on use (71.0% and 73.7%, respectively). We found no evidence of on having inherent or latent properties that would cause it to co-occur in all contexts with negation, and no such evidence has been reported in the literature. The co-occurrence with generalizations and morals/truisms can only mean that in these types of utterances, participants happened to use high rates of verbal negation, a claim that is supported by Ashby (1992, p. 152)), even though his factor group with affirmatives and negatives was not selected by VARBRUL as statistically significant. When compared to our results, Ashby’s distribution itself (i.e. using raw counts) indicates that the patterns are similar enough to suggest that Affirmation is an important factor group, even if its level of importance or effect on statistical significance is unclear. Ashby also found, as did we, that affirmatives comprise the great majority of all indefinite tokens of on, tu, and vous (89% in the present study and 88% in Ashby, 1992). This additional comparison allows us to be confident in our results even though we have far fewer tokens than would have been ideal. The results of the category mode/type of event (as one measure of transitivity), another category proposed by Ashby (1992) based on the work of Hopper and Thompson (1980), are presented in Table 10. Since the specific event factor group was almost categorical, we did not include it in the final VARBRUL analysis. Nonetheless, it is clear that indefinite on is preferred in contexts with reference to a non-specific event,

4

In ‘‘good” French, based on what is often referred to as le bon usage, double-particle negation has a negative particle both before and after the verb, in most cases, and single-particle negation has only a post-verb negative particle, in most cases.

Please cite this article in press as: Williams, L., van Compernolle, R.A., On versus tu and vous: Pronouns with indefinite reference ..., Lang. Sci. (2008), doi:10.1016/j.langsci.2007.11.001

ARTICLE IN PRESS L. Williams, R.A. van Compernolle / Language Sciences xxx (2008) xxx–xxx

17

Table 10 Transitivity: mode Type of event

No. of on/total

% of on

Specific Non-specific

1/26 115/168

3.9% 68.5%

a result also reported by Ashby (1992), who found that specific events prefer tu/vous (which was actually only vous in his data), whereas non-specific events prefer on. Overall, Ashby also found, as did we, that more indefinite uses of on, tu, and vous occur when reference is made to non-specific events (87% in the present study and 88% in Ashby (1992)). If the results from both measures of transitivity are combined, it becomes clear that our results are parallel to those reported by Ashby (1992). Both types of low-transitivity features (i.e. negation and non-specific events) favor indefinite on in Ashby’s study and ours. In addition, the results of both studies indicate that more uses of indefinite pronouns on, tu, and vous overall occur in affirmatives and references to non-specific events. The only real difference between the two studies is not related specifically to indefinite pronoun use, but rather the very high rates of tu as a personal address pronoun in our study and, in Ashby’s, the categorical use of vous as the personal address pronoun. The range of each set of VARBRUL scores allows us to determine which factor groups have the most influence on the use of on and tu/vous as indefinite pronouns. From highest to lowest in strength they are: (1) transitivity: affirmation (with a range of .43); (2) syntactic frame (with a range of .39); and (3) discursive-pragmatic effect (with a range of .30). Although we did not include the factor group transitivity: mode (i.e. specific vs. non-specific event) in the final VARBRUL analysis, a preliminary calculation indicated that it produced a range of .57, which means that transitivity-mode would be the factor group with the most influence on the choice of on vs. tu/vous. 5. Conclusion Although the current study relies on data from synchronous electronic (written) discourse instead of data from sociolinguistic interviews, as was the case in previous work on generic-indefinite on, tu, and vous, the results have turned out to be very similar. In spite of finding a higher proportion of on to tu/vous, which might be expected in a formal communication environment in French, the results of the VARBRUL analysis suggest that on, tu, and vous are being used as indefinite pronouns very much in the same ways that were identified in previous studies going back to the 1970s in conversational spoken French. Given the extremely high rate of tu overall as a personal pronoun, it no longer seems surprising to discover that indefinite on has assumed more of the indefinite reference workload. The higher use of indefinite on than expected might also be due to the fact that participants have no face-to-face contact in synchronous chat, which makes them less willing or less apt to use tu or vous with indefinite reference since the more traditional indefinite on is available and tu/vous with specific-definite reference could be misinterpreted as indefinite tu/vous and vice versa; however, this is only speculation based on one of the main differences between interviews and synchronous chat (i.e. lack of visual contact and cues). Our findings suggest, therefore, that real-time communication in this environment reflects at least some of the patterns and features found in everyday spoken French in spite of the fact that electronic communication, in this case, is written/typed. There is other evidence that real-time chat reflects certain aspects of (informal, everyday, conversational) spoken French, such as a low retention rate of double-particle negation, a high rate of discourse particle use, and orthographic variation used to reproduce or represent words and phrases phonetically (e.g. typing the letter g to represent j’ai [‘I have’] since the French pronunciation of the letter g is the same as the pronunciation of j’ai). Even if orthographic variation can be used to represent everyday spoken French, this is not always the case since there are many different types of abbreviated and otherwise truncated forms of words and phrases used for a variety of reasons, an analysis of which is beyond the scope of the present article. Nonetheless, it is clear that synchronous chat is a mix of written, spoken, and electronic discourses, and we have only been able to treat the indefinite use of a limited number of pronouns in one type of electronic communication environment. Please cite this article in press as: Williams, L., van Compernolle, R.A., On versus tu and vous: Pronouns with indefinite reference ..., Lang. Sci. (2008), doi:10.1016/j.langsci.2007.11.001

ARTICLE IN PRESS 18

L. Williams, R.A. van Compernolle / Language Sciences xxx (2008) xxx–xxx

Further research is needed to understand how the French pronoun paradigm might be distributed and operate differently in asynchronous or other types of synchronous computer-mediated communication. This area of inquiry could prove especially fruitful given the many different possibilities and combinations of participants in relation to audience. For example, synchronous chat is often considered many-to-many communication since all participants can read all messages as they are posted. However, the use of a screen name (i.e. on-line ID) as a form of address is quite common in order to select a specific interlocutor even though some or all of the other people in the room may read the message and respond to it. In other words, even when a specific addressee has been selected, all messages are available to the entire audience, except when users decide to use the private message function, an area in which private messages can be exchanged while the chatters are still in the public part of the room. A similar phenomenon exists in discussion fora, where participants often receive responses from many users even when only one specific interlocutor has been selected. Private communication related to communication in discussion fora can take place by e-mail, and references to such communication can subsequently appear in the public space, which creates a mixed audience that might not have had access to an entire communicational exchange. In chat sessions where a moderator and an invited guest are present, the moderator and/or the guest can decide whose questions to answer, but as the guest answers questions from individual participants the answers are available to (and often addressed to) everyone, which is yet another example of the differences between traditional written communication and text-based chat. In addition to studying other environments of computer-mediated communication, larger corpora are needed, and more needs to be known about the people who chat since more information about participants would allow analyses to investigate additional levels of complexity that are certainly missing in studies where little is known about the people who produced the discourse. In the present study, for example, we did not have the opportunity to see if a particular participant used both indefinite on and indefinite tu/vous with any consistency, given the high rate of entrances, exits, changes of pseudonyms, and brief periods of participation in many cases. No patterns for individual participants were able to be established since proportionally fewer tokens were produced per the number of words in the corpus. Since we were not able to ask specific questions—as would be the case for interviews—it was not possible to lead participants to topics (e.g. giving advice, explaining specific events or situations) that might have produced more pronouns with indefinite reference; instead, our analysis is based on conversation including a variety of topics that followed a path defined only by the participants. This leads us to believe that in this type of electronic communication environment, especially in chat rooms with no specific suggested theme or topic, the relatively low number of tokens of pronouns with indefinite reference is typical. Additional research including an analysis of a larger and similar corpus might confirm or reject this notion of what is typical in synchronous chat. References Armstrong, N., 2002. Variable deletion of French ne: a cross-stylistic perspective. Language Sciences 24, 153–173. Ashby, W., 2001. Un Nouveau Regard sur la Chute du ne en Francßais Parle´ Tourangeau: S’agit-il d’un Changement en Cours?. Journal of French Language Studies 11 1–22. Ashby, W., 1992. The variable use of on versus tu/vous for indefinite reference in spoken French. Journal of French Language Studies 2, 135–157. Ayres-Bennett, W., 1996. A History of the French Language through Texts. Routledge, London. Bailey, G., Tillery, J., 1999. The Rutledge effect: the impact of interviewers on survey results in linguistics. American Speech 74, 389–402. Ball, R., 2000. Colloquial French Grammar: A Practical Guide. Blackwell, Oxford. Bechar-Israeli, H., 1995. From hBoneheadi to hcLoNehEAdi: Nicknames, play, and identity on internet relay chat. Journal of Computer Mediated Communication 1, np. Available at: . Bell, A., 2001. Back in style: reworking audience design. In: Eckert, P., Rickford, J. (Eds.), Style and Sociolinguistic Variation. Cambridge University Press, Cambridge, pp. 139–169. Blondeau, H., 2001. Real-time changes in the paradigm of personal pronouns in Montreal French. Journal of Sociolinguistics 5, 453–474. Blondeau, H., March 2007. De la mouvance sociolinguistique du francßais parle´: Montre´al en temps re´el. Paper presented at Quebec at the Dawn of the New Millennium: Between Tradition and Modernity. Denton, Texas. Blondeau, H., Sankoff, G., Charity, A., 2002. Parcours individuels et changements linguistiques en cours dans la communaute´ francophone montre´alaise. Revue que´be´coise de Linguistique 31, 13–38. Brown, R., Gilman, A., 1960. The pronouns of power and solidarity. In: Sebeok, T. (Ed.), Style in Language. MIT Press, Cambridge, MA, pp. 253–276. Brown, P., Levinson, S., 1987. Politeness: Some Universals in Language Usage, second ed. Cambridge University Press, Cambridge.

Please cite this article in press as: Williams, L., van Compernolle, R.A., On versus tu and vous: Pronouns with indefinite reference ..., Lang. Sci. (2008), doi:10.1016/j.langsci.2007.11.001

ARTICLE IN PRESS L. Williams, R.A. van Compernolle / Language Sciences xxx (2008) xxx–xxx

19

Coupland, N., 1980. Style-shifting in a Cardiff work-setting. Language in Society 9, 1–12. Coupland, N., 1984. Accommodation at work: some phonological data and their implications. International Journal of the Sociology of Language 46, 49–70. Coveney, A., 2003. ‘Anything you can do tu can do better’: tu and vous as substitutes for indefinite on in French. Journal of Sociolinguistics 7, 164–191. Coveney, A., 2000. Vestiges of nous and the 1st person plural verb in informal spoken French. Language Sciences 22, 447–481. Coveney, A., 1996. Variability in Spoken French: A Sociolinguistic Study of Interrogation and Negation. Elm Bank, Exeter, England. Davies, B., 2005. Communities of practice: legitimacy not choice. Journal of Sociolinguistics 9, 557–581. Deshaies, D., 1985. Re´fe´rences personnelles et types de discours en situation d’entrevue. In: Le´on, P., Perron, P. (Eds.). Le Dialogue, Didier, Ottawa, pp. 77–91. Elsig, M., Poplack, S., 2006. Transplanted dialects and language change: question formation in Que´bec. U. Penn Working Papers in Linguistics vol. 12(2), pp. 77–90. Ervin-Tripp, S., 1972. On sociolinguistic rules: alternation and co-occurrence. In: Gumperz, J., Hymes, D. (Eds.), Directions in Sociolinguistics: the Ethnography of Communication. Holt, Rinehart and Winston, New York, pp. 213–250. Fonseca-Greber, B., Waugh, L., 2003. On the radical difference between the subject personal pronouns in written and spoken European French. In: Leistyna, P., Meyer, C. (Eds.), Corpus Analysis: Language Structure and Language Use. Rodopi, Amsterdam, pp. 225– 240. Garcia, A., Jacobs, J., 1999. The eyes of the beholder: understanding the turn-taking system in quasi-synchronous computer-mediated communication. Research on Language and Social Interaction 32, 337–367. Gumperz, J., 1972. Introduction. In: Gumperz, J., Hymes, D. (Eds.), Directions in Sociolinguistics. Holt, Rinehart and Winston, New York, pp. 1–25. ´ tude en Temps Re´el. Langage & Socie´te´ 107, 5–30. Hansen, A., Malderez, I., 2004. Le ne de Ne´gation en Re´gion Parisienne: Une E Herring, S., 1999. Interactional coherence in CMC. Journal of Computer Mediated Communication 4, np. Available at: . Holmes, J., 1995. Women, Men and Politeness. Longman, London. Hopper, P., Thompson, S., 1980. Transitivity in grammar and discourse. Language 56, 251–299. Jisa, H., Viguie´, A., 2005. Developmental perspectives on the role of French on in written and spoken expository texts. Journal of Pragmatics 37, 125–142. ´ tude de la variation des pronoms sujet de´finis et inde´finis dans le francßais parle´ a` Montre´al. Unpublished Ph.D. Laberge, S., 1977. E dissertation, Universite´ de Montre´al. Labov, W., 1966. The Social Stratification of English in New York City. Centre for Applied Linguistics, Washington, D.C. Labov, W., 1972. Sociolinguistic Patterns. University of Pennsylvania Press, Philadelphia. Labov, W., 1994. Principles of Linguistic Change: Internal Factors. Blackwell, Oxford. Laberge, S., Sankoff, G., 1980. Anything you can do. In: Sankoff, G. (Ed.), The Social Life of Language. University of Pennsylvania Press, Philadelphia, pp. 271–293. Lavandera, B., 1978. Where does the sociolinguistic variable stop? Language in Society 7, 171–182. Lave, J., Wenger, E., 1991. Situated Learning: Legitimate Peripheral Participation. Cambridge University Press, Cambridge. Milroy, L., 1987. Language and Social Networks, second ed. Blackwell, Oxford. Offord, M., 1990. Varieties of Contemporary French. Macmillan, London. Paolillo, J., 1999. The virtual speech community: social network and language variation on IRC. Journal of Computer Mediated Communication 4, np. Available at: . Paolillo, J., 2001. Language Variation on Internet Relay Chat: A Social Network Approach. Journal of Sociolinguistics 5, 180–213. Peeters, B., 2006. Nous on vous tu(e): La guerre (pacifique) des pronoms personnels. Zeitschrift fu¨r romanische Philologie 122, 201–220. Pierozak, I., 2003. Le francßais tchate´. Une e´tude en trois dimensions – sociolinguistique, syntaxique et graphique – d’usages IRC. Unpublished Ph.D. Dissertation, Universite´ d’Aix-Marseille I. Price, G., 1993. L.S.R. Byrne and E.L. Churchill’s A Comprehensive French Grammar, fourth ed. Blackwell, Oxford. Rheingold, H., 2000. The Virtual Community: Homesteading on the Electronic Frontier. MIT Press, Cambridge, MA. Rickard, P., 1993. A History of the French Language, second ed. Routledge, London. Sankoff, D., Tagliamonte, S., Smith, E., 2005. Goldvarb X: A Variable Rule Application for Macintosh and Windows. University of Toronto, Department of Linguistics. Sankoff, D., Laberge, S., 1978. The linguistic market and the statistical explanation of variability. In: Sankoff, D. (Ed.), Linguistic Variation: Models and Methods. Academic Press, New York, pp. 239–250. Sankoff, G., 1980. The Social Life of Language. University of Pennsylvania Press, Philadelphia. Schilling-Estes, N., 1998. Investigating ‘‘self-conscious” speech: the performance register in Ockracoke English. Language in Society 27, 53–83. Thibault, P., 1991. La langue en mouvement: simplification, re´gularisation, restructuration. LINX 25, 79–92. Wertheim, S., 2006. Cleaning up for company: using participant roles to understand fieldworker effect. Language in Society 35, 707–727. Williams, L., van Compernolle, R.A., 2007. Second-person pronoun use in on-line French-language chat environments. French Review 80, 804–820. Wolfram, W., 1991. The linguistic variable: fact and fantasy. American Speech 66, 22–32.

Please cite this article in press as: Williams, L., van Compernolle, R.A., On versus tu and vous: Pronouns with indefinite reference ..., Lang. Sci. (2008), doi:10.1016/j.langsci.2007.11.001