26 Inclusive and Exclusive Patterning of the English First Person Plural: Evidence from Conversation JOANNE SCHEIBMAN 1 1.1 Introduction and Background* Introduction to the Study In a July 2002 newspaper column entitled ‘“We” is not me or a lot of us,’ Molly Ivins (2002) condemns what she calls the promiscuous uses of we found in recent editorials pertaining to the U.S. stock market’s sluggish performance. Specifically these commentators were reminding readers that since we profited mightily during the financial boom of the 1990s, we must now face up to the realities of investment risk. Ivins crisply rebuts the journalists’ inclusion of all of us as beneficiaries of financial gain with “Who you callin’ ‘we,’ white man?” (Ivins 2002)—a famous punch line from a well-known joke based on The Lone Ranger. In using this expression Ivins points out the presumptuousness of assuming that all United States citizens are at an economic level able to participate in the financial markets. The rhetorical success of Ivins’s argument suggests that American English speakers are cognizant when speakers and writers infelicitously include others in a referential group using we. Expression of inclusion and exclusion in * Work on this project was supported by an Old Dominion University College of Arts and Letters Summer Research Grant, 2002. Language, Culture, and Mind Michel Achard and Suzanne Kemmer (eds.). © 2004, CSLI Publications 377 378/ JOANNE SCHEIBMAN discourse is not solely a referential operation; it has both social and interpersonal consequences (e.g. Duszak 2002b). Many languages lexically or morphologically distinguish inclusive uses of the first person plural pronoun—those that refer to the speaker and the addressee(s), from exclusive uses—those that refer to the speaker and another individual or group who are not addressees. In English, first person plural pronouns function both inclusively and exclusively without a change in form. However, in the study reported on in this paper, American English conversational utterances with we subjects exhibit variation in syntactic patterning, contingent on the referentiality of the first person plural subject. Distributional analyses of conversational utterances indicate that predicates with inclusive subjects more frequently contain modal elements and less frequently occur in the past tense, relative to exclusive tokens. Predicates with inclusive we subjects in the database, then, tend to be propositionally mediated in conversation in ways that predicates with exclusive we subjects are not. In interactive discourse, the mediating properties of these utterances soften violations of negative face that are potentially part of the act of speakers’ referentially subsuming another discourse participant under the first person plural pronoun—an interpersonal negotiation that is not required when we functions exclusively. The theoretical point of view guiding this investigation is usage based. It assumes that in large part communicative and cognitive factors shape grammatical and lexical patterning in language (Hopper 1987, Bybee 2001, Thompson and Hopper 2001, Thompson 2002). In this case, analyses of distributional and functional properties of first person plural conversational utterances suggest that there is an inclusive/exclusive distinction in English that is structurally apparent and pragmatically motivated. Moreover, employing a method that focuses on conversational usage demonstrates an interactional basis for linguistic categories. In the case of we, analyses of conversational utterances reveal a distinction in English that is marked morphologically in other languages. The rest of Section 1 highlights previous work on the expression of inclusion and exclusion from both typological and discourse analytic perspectives. Section 2 provides a brief overview of the conversational database and coding categories used in this study, and Section 3 outlines general distributions of utterances by referentiality, tense, and modality. Sections 4 and 5 present analyses and discussion of inclusive and exclusive utterances respectively. And finally, Section 6 offers summaries and conclusions based on the analytical findings. INCLUSIVE AND EXCLUSIVE PATTERNING / 379 1.2 The Inclusive/Exclusive Distinction as a Typological Feature Haas (1969) concludes that the terms inclusive and exclusive have been in use in linguistics for at least two centuries and probably longer. Distribution of the inclusive/exclusive contrast is areally defined. The feature occurs most frequently in language families of Australia, South America, Mesoamerica, and North America, and least frequently in the language families of Africa and Europe (Nichols 1992). In contrast to this strong geographical correlation, Nichols (1992) finds no association between the inclusive/exclusive opposition and any other typological feature she surveyed (e.g. word order, noun class systems, alienable-inalienable possession, voice oppositions). Though the inclusive/exclusive category is independent of number in pronouns, it often combines with number distinctions such as dual and plural (Helmbrecht 2002: 36). These contrasts are illustrated below for Kuri, an Austronesian language spoken in Indonesia. (1) Kuri (Austronesian) dual plural incl. Neá Nullia excl. Neá-wai Nu úllia (Schmidt 1919 cited in Mühlhäusler and Harré 1990: 171) For languages that make such four-way distinctions, the inclusive dual refers to the speaker and one addressee (‘I’ + ‘you’ SG ); the inclusive plural includes the speaker and more than one addressee (‘I’ + ‘you’ PLURAL); and the exclusive forms refer to the speaker and others who are not directly addressed—one other person in the case of the dual (‘I’ + ‘s/he/it’) and more than one for the plural (‘I’ + ‘they’). While there is considerable variation in the morphological and lexical expression of inclusive and exclusive elements crosslinguistically (Ingram 1978, Jacobsen 1980), in many languages the functional association between the inclusive and the second person (the addressee) is reflected formally. It is not uncommon, for example, for second person morphemes to mark inclusive forms and for first person or third person elements to mark exclusive forms. Benveniste (1971: 203) notes that in Fox (an Algonquian language) the inclusive pronoun is marked with a second person prefix, and the exclusive pronoun contains a first person prefix. In Tok Pisin, a Creole language spoken in Papua New Guinea, the inclusive yumi is formed by combining first person and second person morphemes (Romaine 1992:!2). In English conversation there are distributional similarities between utter- 380/ JOANNE SCHEIBMAN ances with second person singular and inclusive we subjects on the one hand, and between utterances with third person singular (human) and exclusive we subjects on the other. Predicates with both second person singular and inclusive we subjects frequently contain modal elements (e.g. can, gotta) and rarely occur in the past tense. Clauses with third person singular (human) and exclusive we subjects hardly ever contain modal elements and often occur in the past tense (Scheibman 2002). These formal facts are compatible with Benveniste’s proposal that ‘the distinction of the inclusive and exclusive forms is modeled in reality on the relationship we have established between the first and second singular and between the first and third singular, respectively’ (1971: 202). 1.3 Discourse Studies of Inclusive and Exclusive ‘We’ Studies investigating discourse functions of inclusive and exclusive pronouns come from a variety of theoretical perspectives, and they typically highlight issues such as group membership, participant alignment, and positive and negative face needs (e.g. Goffman 1959, Goffman 1981, Mao 1996, Valle 1996, Pyykkö 2002, Skarżyńska 2002). In her introduction to the volume Us and Others Anna Duszak (2002a: 6) observes: Both we and they can be skillfully managed in discourse in order t o construct, redistribute or change the social values of ingroupness and outgroupness. We in particular opens up a number of referential and pragmatic options (esp. the inclusive—exclusive distinction) and enjoys a strong cultural salience across languages and contexts. One domain in which ‘we’ is used to globally manipulate group membership is political discourse. In a study on the use of m y ‘we’ in recent Soviet and Russian speeches, official records, and interviews, Pyykkö (2002) shows how the referents of my change over time depending on the historical and ideological context. During the Soviet era, for example, m y referred to the Communist Party and the people; in contrast, the collective my used by Putin expresses a more traditional sense of Russian community (Pyykkö 2002: 244-46). Since by definition all deictic expressions take as part of their meaning aspects of the discourse context, in political discourse the indexical meaning of ‘we’ draws and reinforces ideological and national boundaries. In written English, authors may establish rapport with their readers by using we inclusively. In an analytical comparison of Stephen J. Gould’s popular and academic writings, Valle (1996) finds that Gould only uses inclusive we in his nonspecialist texts, not in his scholarly work. Valle suggests that Gould’s use of inclusive we in his nontechnical writings func- INCLUSIVE AND EXCLUSIVE PATTERNING / 381 tions as one of several high involvement discourse strategies he uses that contributes to the public popularity of his work. In contrast, Kuo’s (1999: 132-33) analysis of a corpus of scientific texts shows that the most common function of inclusive we in these data is to presuppose shared background knowledge and beliefs on the part of expert readers. Both Valle (1996) and Kuo’s (1999) studies of the use of inclusive we point to its having a general rhetorical function in English academic and popular academic writing—that of encouraging the reader to maintain interest and belief in the integrity of the text and its arguments, and by association, in the author herself. Mao (1996) studied the functions of Chinese first person pronouns (wo ‘I/me’, wo-men ‘we/us’) for Chinese speakers studying at a Midwestern university in the United States. Based on these speakers’ judgments, Mao (1996: 123) suggests that the use of wo-men in discourse ‘echoes and emulates a traditional Chinese ethos that values the communal over the individual’. For example, ‘humble wo-men’ is frequently used by speakers ‘to convey a sense of modesty and/or politeness’ as illustrated in (2) (Mao 1996: 118). In this case, the use of wo-men (as opposed to wo ‘I’) functions as a negative politeness strategy, allowing the speaker to show humility. (2) 1.4 Xian ta zhezhong ren duo nenggan, xiang wo-men zhezhong ren jiu buxing le. ‘How capable people like her are, whereas people like us are just the opposite.’ (Mao 1996: 118) ‘We’: Plural or Singular? In both spoken and written discourse it is quite common to find first person plural pronouns having singular or ambiguous number reference, e.g. the royal we (Mao 1996 for Chinese, Kuo 1999 for English, Pyykkö 2002 for Russian). Benveniste (1971) observes that, referentially, ‘we’ is not a collection of individuated ‘I’s. Only one of the referents of ‘we’ has speaker status (Margolin 1996). Benveniste (1971: 203) suggests that in languages lacking a formal inclusive/exclusive contrast, for which referentiality of the pronouns is not grammatically assigned (e.g. Indo-European languages), “we” is something other than a junction of definable elements, and the predominance of ‘I’ is very strong in it, to the point that under certain conditions, this plural can take the place of the singular. In languages such as English that do not have a morphological or lexical inclusive/exclusive contrast, referentially ambiguous uses of ‘we’ in discourse have important functions. Expression of group membership or 382/ JOANNE SCHEIBMAN maintenance of positive and negative face, for example, may motivate language users to align themselves with a diffusely conceived class of referents. For example, the speaker of the utterance in (3) is a district attorney who is meeting with a witness in a case she is working on. The witness is also a victim of the crime being discussed. Because of the speaker’s multiple alliances, we in this utterance is at once singular (the lawyer herself who will prosecute the case), inclusive plural (the lawyer and the witness and the witness’s husband who is also present), and exclusive plural (the lawyer and other members of the district attorney’s office). (3) if we get a conviction just on one case here, J1074.821 The inclusive sense allows the lawyer to show support for the witness and also to endow her (the witness) with agency in the legal process, both expressions of positive politeness. The exclusive interpretation permits the lawyer to identify with the district attorney’s office, thus adding authority to her position (which may reassure the witness/crime victim as well). The attorney’s aligning herself with the district attorney’s office also provides an implicit cushion for her own accountability in the event she is not successful in her prosecution. This example illustrates that ambiguity, or fluidity, of expression of number and referentiality in first person plural pronouns can be socially and discursively functional in English conversation. The rest of this paper offers descriptions and analyses of usages of we from a corpus of American English conversational utterances. The interactional pivot of the expression of inclusion in these analyses is the question of ‘whether or not the speaker(s) are empowered to speak on behalf of the reference class as a whole, thus conveying a joint/common communicative intent’ (Margolin 1996: 118). Such interpersonal negotiations have structural consequences for linguistic patterning in English. 2 Description of the Data Utterances with we subjects were taken from sixteen audiotaped conversations and entered into a spreadsheet application for coding. Fifteen of the conversations are part of the UCSB Corpus of Spoken American English (CSAE), and one conversation was taped and transcribed by me. The fifteen conversations total five hours and sixteen minutes of audiotaped talk and yielded 378 clauses with first person plural subjects.2 The conversations 1 Numbers locate conversational utterances in the database. 2 Let’s imperatives are not included in the corpus. INCLUSIVE AND EXCLUSIVE PATTERNING / 383 themselves represent the language of fifty-three speakers in total, though only thirty-nine speakers produce first person plural utterances. Brief sociolinguistic characteristics of participants in the conversations (sex, age, years of education, primary dialect of American English) are available as part of the CSAE but do not figure directly in the current study. The utterances were coded for a variety of structural and functional features. Most relevant to this analysis is the coding of subject referentiality and number (exclusive/inclusive/generic, dual/plural), referential specificity (individuated, class), modality, and tense (present, past, modal).3 Coding categories are briefly discussed below. Referentiality: The values used to code referentiality of we subjects are: inclusive dual, inclusive plural, exclusive dual, exclusive plural, and generic. (4) INCLUSIVE DUAL (speaker addressing a friend in the conversation) 4 we better go up to Dillards. (5) INCLUSIVE PLURAL (6) EXCLUSIVE DUAL (7) EXCLUSIVE PLURAL (8) GENERIC T609.7 (five family members celebrating one of their birthdays together) we gotta get a picture. A679.36 (one member of a couple referring to herself and her partner, speaking to a dinner guest) we read this great book. P557.34 (speaker referring to herself and her classmates at the school where she is studying to be a farrier) we have to put ointment on em and stuff. K711.78 5 we take our air for granted. K1360.62 The process of coding linguistic material is immensely useful when investigating linguistic structure in interactive discourse; however, in practice, 3See Scheibman (2002) for general discussion of grammatical coding of conversational utterances. 4The following transcription symbols are used in this paper (Du Bois et al. 1993). . final transitional continuity .. pause , continuing transitional continuity @ laughter ? appeal transitional continuity (H) inhalation [] speech overlap Q quotation -truncated utterance YWN yawn = lengthening X indecipherable 5 These very global generic usages are not analyzed in this study. 384/ JOANNE SCHEIBMAN linguistic elements do not always fall into discrete classes. For example, in (3) above, repeated below as (9), we may be interpreted as inclusive or exclusive. The subject pronoun in this utterance was coded inclusive plural. The rationale for privileging the inclusive sense in such cases is because one of the main goals of this study is to look at structural properties of utterances whose first person plural subjects include addressees versus those that do not. This example illustrates the difficulties in collapsing complex discourse usages into discrete categories. (9) if we get a conviction just on one case here, J1074.82 Referential specificity: Most grammatical presentations of pronominal paradigms assume that singular and plural morphemes reliably designate singular and plural referents. However, in English usage inclusive plural and exclusive plural subjects may refer to a class of referents as well as to individual referents. For this reason, we subjects in the database were assigned an additional code to indicate the referential specificity of the pronoun. The two coding values are individuated (specific individuals) and class (nonindividuated group). (10) INDIVIDUATED (Three people cooking dinner together) ROY: Æ (11) @So, we can dispense with the garlic and the butter. PETE: <X Right X>. P234.78 CLASS (The speaker is a bank supervisor in a meeting discussing a par- ticular money instrument. We in this utterance refers simultaneously t o the bank employees present in the meeting and to the institution itself.) we've kind of hesitated in .. in offering those, O931.48 Modality: English central modals appearing in the conversational utterances are entered in this field (can, can’t, could, couldn’t, might, might not, must, shall, should, shouldn’t, will, ’ll, won’t, would, ’d, wouldn’t). Other modal elements, or intermediate function verbs (e.g. gotta, ought to, be able to), are listed in a separate category. Tense: The coding values for this category are present, past, and modal. Modal is used to code predicates containing modal auxiliaries, for which time reference is usually irregular. INCLUSIVE AND EXCLUSIVE PATTERNING / 385 3 Distributional Patterns of Referentiality, Tense, and Modality Table 1 displays the distribution of tense and modality of utterances in the corpus by referentiality of we. Of the 378 first person plural utterances in the database, 136, or 36 percent are inclusive; 212, or 56 percent, are exclusive; and 30, or 8 percent, of the tokens are coded generic.6 Most striking about these distributions is that 88 percent of all past tense predicates have exclusive subjects, and 68 percent of utterances containing modal auxiliaries occur with inclusive we. Moreover, 47 percent of clauses with exclusive subjects are past tense (99/212), compared to only 9 percent of inclusive tokens (12/136). Additionally, the percentage of utterances containing central modals is far greater for inclusive tokens (35 percent) than for exclusive uses (9 percent). Referentiality inclusive exclusive generic Total Percent Present 76 39.18% 94 48.45% 24 12.37% 194 100.00% Past 12 10.62% 99 87.61% 2 1.77% 113 100.00% Modal 48 67.61% 19 26.76% 4 5.63% 71 100.00% Total 136 35.98% 212 56.08% 30 7.94% 378 100.00% Table 1. Utterances by referentiality and tense/modality (n=378) 4 4.1 Inclusive We in Conversation Introduction Table 2 summarizes the distributional properties of utterances with inclusive subjects in the database. Only 9 percent of these predicates are marked past tense and over a third of inclusive utterances (35 percent) contain a modal auxiliary. Inclusive dual subjects are strikingly less frequent in this corpus than are inclusive plurals (17 and 119, respectively). This difference in frequency between the dual and the plural may be due to the fact that twelve of the sixteen conversations used as data for this study had more than two participants; this might reduce the number of opportunities for two people to 6Percentages in tables are rounded to whole numbers in discussion (e.g., 36 percent is used for 35.98 percent). 386/ JOANNE SCHEIBMAN have separate conversations with one another. Or it may be the case that in these multiparty exchanges the recording event itself could motivate participants to treat the conversation as a special occasion and to maintain group interaction. Because the inclusive dual and plural tokens exhibit similar structural features (e.g. high frequency of modal elements, low percentage of past tense), they will be analyzed together as one group. Referentiality inclusive dual inclusive plural Total Percent Present 7 41.18% 69 57.50% 76 55.88% Past 5 29.41% 7 5.83% 12 8.82% Modal 5 29.41% 43 36.67% 48 35.29% Total 17 100.00% 119 100.00% 136 100.00% Table 2. Inclusive utterances by referential specificity and tense/modality (n=136) In an earlier study examining relationships between subjects and predicates in American English conversation, I found that clauses coocurring with second person singular subjects (you) are informationally mediated in a variety of ways (Scheibman 2002). Second person singular utterances exhibit a very low frequency of past tense predicates (demonstrating the infrequency of direct assertion with addressee subjects), a high percentage of modal elements (e.g. (12)), and second person singular expressions often appear as questions or subordinate clauses following epistemic or evaluative clauses (e.g. that’s good in (13)). (12) you can use this for your muffins. (13) that’s good you’re getting r- good rest. (Scheibman 2002: 77-8) These properties of second person singular utterances demonstrate that English speakers hedge assertions directed toward addressees to maintain addressees’ negative face. With respect to the present study, first person plural inclusive utterances in the database exhibit these same features, suggesting that when English speakers use inclusive we they are sensitive to the interpersonal stakes involved in negotiating consensus for their assertions in the same ways that motivate them to mediate assertions toward addressees using you. INCLUSIVE AND EXCLUSIVE PATTERNING / 387 4.2 Modal Elements in Inclusive Utterances Table 2 shows that 48 of the 136 inclusive predicates in the database contain a central modal (35 percent). The most frequently appearing modals in the database are can, could, should, and ’ll. Predicates with modal auxiliaries in these inclusive utterances highlight the speaker’s stance or commitment toward a proposition, and they also soften violations of negative face that are potentially part of the act of referentially subsuming other discourse participants under we. This situation is in contrast to predicates without modal elements which more directly assert a subject referent’s participation in, or compliance with, the state/activity designated by the main verb (e.g. going vs. can go or want to go). This polite intersubjective use of inclusive we utterances containing modal auxiliaries is illustrated in (14). In this conversational episode, Marilyn and Pete are discussing what they will make for dinner for themselves and their guest, Roy. In Marilyn’s utterance, We can make um, … garlic bread or something, the modal can mediates the force of the assertion, providing an opportunity for the other participants to negotiate the activity proposed by the speaker. That is, can gives Pete and Roy the choice whether or not to be referentially included in Marilyn’s we. Note, too, the presence of other markers that mediate assertion in this utterance, such as um, the pause between intonation units (represented by three dots), and the speaker’s opening the other participants’ options with the phrase or something. These features, in addition to can, mark this inclusive we utterance as a polite suggestion. (14) Æ MARILYN: ... We can make um, ... garlic bread or something. ROY: .. <FOOD Oh, that [sounds] fun FOOD>. PETE: [Yeah]. MARILYN: ... Okay. PETE: Yeah, we could have a little .. garlic bread. R214.79 Many inclusive predicates contain other types of modal elements as well. Like modal auxiliaries, these intermediate function verbs (e.g. be able to, be gonna, have to, or (had) better) also open the proposition expressed in the predicate to the addressee’s approval or acquiescence. Twenty-two inclusive predicates in the database contain a modal element such as those described here. Seventeen appear in present tense predicates, two with past 388/ JOANNE SCHEIBMAN tense predicates, and three clauses contain both a central modal and an intermediate function verb (e.g. we would then have to if they had a CD pull that in). For example, in (15), Sam suggests going to a department store (Dillards) with Doris and uses the expression (had) better. This marker opens the assertion to Doris’s approval. And indeed, Doris does negotiate her involvement in the activity by questioning whether or not they would have to drive up to Phoenix. (15) Æ SAM: (H) We better go up to Dillards, and see if they still have those muumuus. DORIS: ... You mean up to Phoenix? SAM: ... No=, well- they might have them here, in town. T609.73 4.3 Inclusive Utterances as Polite Questions In a study of questions in English conversation, Weber (1993) finds that 41 percent of the questions in her corpus do not occur as interrogative forms; instead questions are expressed in a variety of ways (e.g. as clauses, phrases, or lexical items). It is also the case that in English and in other languages, interrogatives can serve as polite requests (e.g. Why don’t you sit down?). In the database, inclusive we often appears in utterances functioning as polite questions or requests. Eighteen inclusive tokens are questions in the conversations. For example, in (16) Marilyn uses an inclusive plural we to inquire whether or not the other discourse participants want potatoes for dinner. This we is functioning as a de facto you; however, by including herself in the food decision, the speaker makes the inquiry less direct. (16) MARILYN: Æ ROY: ... okay, so did we decide we do or do not want potatoe=s. ... I think potatoes are excessive. I think we have enough food here. P202.00 The majority of inclusive first person plural pronouns that occur in polite questions—thirteen of sixteen tokens—refer to specific individuals, not to a class of people. This is because the interpersonal consequences are greater when addressees are individual people as opposed to an undifferentiated group. Speakers do not have to mediate referential inclusion of other people when these others—the nonspeaker referents of we—are part of a generally construed class of referents. INCLUSIVE AND EXCLUSIVE PATTERNING / 389 4.4 Inclusive Utterances Following Epistemic/ Evidential Clauses The last group of inclusive utterances discussed in this section are those that follow clauses containing epistemic or evidential verbs (Thompson and Hopper 2001), such as I think, I don’t know, I mean, and I guess (see also Schiffrin 1987, Scheibman 2000, Scheibman 2002, Thompson 2002). In these uses of we, the epistemic/evidential expressions temper speakers’ assertions about addressees (as coreferents of the subject). In (17), the speaker’s I wonder (along with her use of you know and the rise in pitch at the end of the intonation unit) marks the contribution as an utterance that is open to another participant’s agreement (e.g. as a question or perhaps an indirect request)—one that is structurally receptive to the addressee’s point of view. (17) 4.5 ... this darn dog keeps ... breathing, <X and like X> ... dreaming, Æ you know I wonder if we should wake her up? E542.77 Summary of Inclusive Utterances The frequency of the structural properties characteristic of inclusive we utterances surveyed in the section—the presence of modal elements, the appearance of we in questions or polite requests or following epistemic/evidential clauses, and the paucity of past tense reports—mediate direct assertion in conversation. These same linguistic features have been found to characterize second person expressions in conversation as well (Scheibman 2002). These findings are compatible with typological studies that document associations between second person elements and inclusive ‘we’ forms in languages that mark an inclusive/exclusive contrast morphologically or lexically. What is apparent in the structure of inclusive clauses in English conversation is a formal mediation of assertion contingent on the addressee’s approval or consensus. This situation reveals the delicate interpersonal negotiations that are prominent in face-to-face communication when speakers use we to speak for another participant. In contrast, these mediating properties occur much less frequently in exclusive clauses. 5 5.1 Exclusive We in Conversation Introduction Fifty-six percent of first person plural subjects in the database are exclusive (n=212). Table 3 indicates that 47 percent of exclusive predicates are past tense (compared to 9 percent inclusive past tense tokens), and only 9 percent 390/ JOANNE SCHEIBMAN of exclusive clauses contain a modal auxiliary (compared to 35 percent of inclusive utterances). This section considers the discourse functions of exclusive utterances in English conversation with respect to this frequency of past tense predicates and infrequency of modal elements, relative to inclusive clauses. Exclusive we indexes the speaker and other nonaddressees. These third person referents may be specific individuals (coded individuated) who may or may not be present in the discourse context (see Section 5.2), or they may be a generally conceived group (coded class) often affiliated with an institution (e.g. bank, university course or program). Table 3 presents the distribution of exclusive utterances by tense/modality and referential specificity. All exclusive duals refer to specific people (individuated) as do 25 percent of exclusive plurals (17/67). On the other hand, the majority of exclusive plurals (50/67, or 75 percent) refer to the speaker and to a class of referents. Referentiality exclusive dual exclusive plural individuated class Total Percent Present 72 49.66% 22 32.84% 3 17.65% 19 38.00% 94 44.34% Past 61 42.07% 38 56.72% 13 76.47% 25 50.00% 99 46.70% Modal 12 8.28% 7 10.45% 1 5.88% 6 12.00% 19 8.96% Total 145 100.00% 67 100.00% 17 100.00% 50 100.00% 212 100.00% Table 3. Exclusive utterances by referential specificity and tense/modality (n=212) The most frequent referential category in the corpus is exclusive dual (145, or 68 percent of exclusives, and 38 percent of all we tokens). Exclusive plurals account for 32 percent of the exclusive group (67/212). That exclusive predicates occur in the past tense much more frequently than do inclusive predicates, and are less likely to include modal elements than their inclusive counterparts, suggests that these structures participate less in interpersonal negotiations and more in direct assertion than do inclusive clauses. The convergence of these frequent properties of exclusive utterances in the database—past tense, absence of modals, and the presence of a specific human nonspeech act participant subject referent—have also been found to characterize utterances with third person singular human subjects in English conversation (Scheibman 2002). The crosslinguistic association between INCLUSIVE AND EXCLUSIVE PATTERNING / 391 exclusive first person plural markers and third person elements is demonstrated in the function and distribution of structural properties of exclusive expressions in English conversation as well. 5.2 Exclusive Dual Utterances Forty-two percent of exclusive dual predicates are past tense. All but one of these 145 utterances refer to the speaker and to a specific person with whom the speaker has an intimate or very familiar relationship, such as a partner/spouse, mother, brother, child, boss, or good friend. For example, in (18) the speaker uses we to refer to herself and her partner/husband (who is not a participant in the conversation). We used in this way to report past experiences is a common function of exclusive dual utterances in the database. (18) you know, Æ we showed up there and they were all gone. .. That really irritated me. Z839.14 Table 4 shows that 66 percent of the exclusive dual pronouns in the database (95/145) have nonspeaker referents who are unaddressed participants in the conversation, and 50 exclusive dual tokens refer to people who are not present in the discourse. Goffman (1981: 133) makes a distinction between addressed and unaddressed recipients in multiparty encounters; he explains that ‘the speaker will, at least during periods of his talk, address his remarks to one listener, so that among official hearers one must distinguish the addressed recipient from “unaddressed” ones’. In a study of the effects of audience on speaker style, Bell (1984) makes a similar distinction using the terms addressee and auditor. Auditors (unaddressed recipients) are third persons who are present and ratified hearers, but are not directly addressed by the speaker. Bell (1984: 160) suggests that with respect to the effect of audience role on stylistic variation, the auditor accounts for variation, though it is less influential than effects of the speaker and addressee. Table 4 indicates that auditors, or unaddressed recipients, do appear to influence speakers’ use of exclusive we in conversation. That is, these utterances more frequently show signs of mediation than those that refer to a third person who is not present. 392/ JOANNE SCHEIBMAN Unaddressed Recipient Present Past Modal Total Present in conversation 54 31 10 95 56.84% 32.63% 10.53% 100.00% Not present in conversation 18 30 2 50 36.00% 60.00% 4.00% 100.00% Total 72 61 12 145 Percent 49.66% 42.07% 8.28% 100.00% Table 4. Exclusive dual utterances by presence of unaddressed recipient and tense/modality (n=145) Though this data set is small, Table 4 shows an interesting distribution of exclusive dual usages based on whether or not the nonspeaker referent is also a discourse participant. Specifically, there are a greater proportion of utterances with modal auxiliaries (11 percent) and a lower proportion of past tense predicates (33 percent) when we refers to two discourse participants (the speaker and an unaddressed recipient) than when the nonspeaker referent is not present in the conversation (modal: 4 percent; past tense: 60 percent). In (19), the speaker, Wendy, is using we to refer to herself and her husband, Kevin. Both her use of the modal ’ll and the epistemic clause I don’t think mediate her assertion in this utterance. (19) Æ WENDY: .. Well I don't – .. I don't think we'll be taking the end tables to Bulgaria. KEVIN: ... I'm sure we will not. A1010.19 5.3 Exclusive Plural Utterances Exclusive plural usages account for slightly less than a third of all exclusive tokens (67/212). Unlike the exclusive dual forms discussed above, none of the exclusive plurals in the database refers to unaddressed recipients. Moreover, in contrast to the exclusive dual subjects whose nonspeaker referents frequently refer to specific people, only 17, or 25 percent, of the exclusive plural subjects are coded individuated (Table 3). In contrast to exclusive dual utterances, then, the majority of exclusive plural subjects in the corpus (50/67, or 75 percent) refer to undifferentiated groups (Table 3). Using these class tokens, speakers typically make reference to an institutional group such as a university class or a place of employment that the speaker is affiliated with. For example, in (20), we refers to the speaker and the people she works with at a medical laboratory and also to the institution as a whole. INCLUSIVE AND EXCLUSIVE PATTERNING / 393 (20) Æ we do like ... o=h, about, four hundred samples a day, G19.9 Though the numbers are small, the distribution of tense in individuated vs. class uses of exclusive we in Table 3 suggests that there is proportionally more past tense for exclusive first person plural subjects that point to individuals (coded individuated) than for exclusive utterances that index a group or an institutional function (coded class) which points to their different functions in discourse. While utterances with exclusive we referring to individuals tend to be used by speakers to narrate specific past tense events, the most common use of exclusive plural expressions is to make general predications (often with habitual meaning) about the speaker’s activities as a member of a group. 5.4 Summary of Exclusive Utterances Utterances with exclusive first person plural subjects are more frequently marked past tense and less frequently contain modal elements than utterances with inclusive subjects. Predicates occurring with exclusive we subjects, then, are less mediated than those with inclusive subjects. Exclusive duals in the database are used by speakers to report on activities accomplished by speakers and people close to them (partners, spouses, family members, and friends). Distribution of tense for these exclusive dual forms also indicates that when exclusive we refers to another person in the conversation—an unaddressed recipient—there is proportionally less past tense and a greater percentage of modal elements than when the exclusive dual subject refers to an individual who is not a participant in the conversation. This suggests that when there is a third person referent of we present in the conversation, English speakers tend to mediate their assertions as they do when using inclusive we subjects. On the other hand, the majority of exclusive plural subjects index groups or institutional functions instead of specific individuals, and these utterances show proportionally less past tense usage than their referentially specific counterparts. Speakers tend to use exclusive plural we subjects with more generalizing predicates to discuss activities and beliefs relevant to their membership in a group. Analyses of these data, then, suggest that exclusive individuated referents and exclusive class referents have different functions in English conversation. 394/ JOANNE SCHEIBMAN 6 Conclusions Investigations of inclusive and exclusive first person plural markers tend to cluster in two broad areas of language study: (1) typological examinations of personal pronoun systems, and (2) analyses of the social and rhetorical uses of ‘we’ in various spoken and written discourse genres. The present study combines the spirit of these two approaches by attending to the formal and functional contexts in which American English speakers use we inclusively and exclusively in conversation. By using a method that focuses on usage—in particular, conversational interaction—an inclusive/exclusive contrast becomes apparent in English. This analysis shows that in English interactive discourse pragmatic and structural distinctions are made by speakers even when there are no morphological contrasts of the type found in other languages. The distribution of tense and modality of we utterances in the conversational database are in line with typological studies that document formal associations between second person elements and inclusive markers on the one hand, and third person morphemes and exclusive markers on the other. In English conversation, mediated assertions overwhelmingly occur with second person singular subjects and also with inclusive we subjects (frequent modal elements, few past tense reports). In the same way that the presence of a second person singular subject softens the force of an utterance in English conversation, the implicit ‘you’ indexed by inclusive we also motivates speakers to mediate their predications. In contrast, clauses with third person singular human subjects and those with exclusive we subjects less frequently contain modal elements and are often past tense. Assertions with these subjects are more direct. Exclusive we utterances in English conversation also demonstrate social and interpersonal functions. While exclusive dual forms are overwhelmingly used by speakers to narrate specific activities that they and a close or intimate person participated in, the majority of exclusive plural pronouns in the database do not function as true plurals. Instead, these uses of we tend to express generic reference, indexing the speaker as a member of a group or aligned with an institutional point of view. These expressions can be as mundane as discussing rote procedures in one’s place of employment or as ideologically consequential as showing solidarity with a political position. Both of these situations, however, are referentially expansive activities. In general, then, analyses of the formal properties of inclusive and exclusive usages in English conversation demonstrate the interpersonal functions of these utterances and also offer a view of the interactional underpinnings of grammatical categories as they appear in conversational contexts. INCLUSIVE AND EXCLUSIVE PATTERNING / 395 References Bell, A. 1984. Language Style as Audience Design. Language in Society 13: 145-204. Benveniste, E. 1971. Problems in General Linguistics. Translated by M. E. Meek. Coral Gables, FL: University of Miami Press. Bybee, J. 2001. Phonology and Language Use. Cambridge: Cambridge University Press. Du Bois, J., S. Schuetze-Coburn, S. Cumming, and D. Paolino. 1993. Outline of Discourse Transcription. Talking Data: Transcription and Coding in Discourse Research, eds. J. A. Edwards and M. D. Lampert, 45-89. Hillsdale, NJ: Lawrence Erlbaum. Duszak, A. 2002a. Us and Others: An Introduction. Us and Others: Social Identities across Languages, Discourses and Cultures, ed. A. Duszak, 1-28. Amsterdam and Philadelphia: John Benjamins. Duszak, A. (ed.). 2002b. Us and Others: Social Identities across Languages, Discourses and Cultures. Amsterdam and Philadelphia: John Benjamins. Goffman, E. 1959. The Presentation of Self in Everyday Life. New York: Doubleday. Goffman, E. 1981. Footing. Forms of Talk, 124-59. Philadelphia: University of Pennsylvania Press. Haas, M. R. 1969. ‘Exclusive’ and ‘Inclusive’: A Look at Early Usage. International Journal of American Linguistics 12: 1-6. Helmbrecht, J. 2002. Grammar and Function of We. Us and Others: Social Identities Across Languages, Discourses and Cultures, ed. A. Duszak, 31-49. Amsterdam and Philadelphia: John Benjamins. Hopper, P. J. 1987. Emergent Grammar. Berkeley Linguistics Society 13: 13957. Ingram, D. 1978. Typology and Universals of Personal Pronouns. Universals o f Human Language, Vol. 3: Word Structure, ed. J. H. Greenberg, 213-47. Stanford, CA: Stanford University Press. Ivins, Molly. 2002. ‘We’ is Not Me or a Lot of Us. Fort Worth Star-Telegram July 21: 4E. Jacobsen, W. H. 1980. Inclusive/Exclusive: A Diffused Pronominal Category i n Native Western North America. Papers on the Parasession on Pronouns and Anaphora, eds. J. Kreiman and A. E. Ojeda, 204-27. Chicago: CLS, University of Chicago. Kuo, C-H. 1999. The Use of Personal Pronouns: Role Relationships in Scientific Journal Articles. English for Specific Purposes 18: 121-38. Mao, L. R. 1996. Chinese First Person Pronoun and Social Implicature. Journal of Asian Pacific Communication 7: 106-28. Margolin, U. 1996. Telling Our Story: On ‘We’ Literary Narratives. Language and Literature 5: 115-33. 396/ JOANNE SCHEIBMAN Mühlhäusler, P., and R. Harré. 1990. Pronouns and People: The Linguistic Construction of Social and Personal Identity. Oxford: Basil Blackwell. Nichols, J. 1992. Linguistic Diversity in Space and Time. Chicago: The University of Chicago Press. Pyykkö, R. 2002. Who is ‘Us’ in Russian Political Discourse. Us and Others: Social Identities Across Languages, Discourses and Cultures, ed. A. Duszak, 234-48. Amsterdam and Philadelphia: John Benjamins. Romaine, S. 1992. The Inclusive/Exclusive Distinction in Tok Pisin. Language and Linguistics in Melanesia 23: 1-11. Scheibman, J. 2000. I dunno ... A Usage-Based Account of the Phonological Reduction of don’t in American English Conversation. Journal of Pragmatics 32: 105-24. Scheibman, J. 2002. Point of View and Grammar: Structural Patterns of Subjectivity in American English Conversation. Amsterdam and Philadelphia: John Benjamins. Schiffrin, D. 1987. Discourse Markers. Cambridge: Cambridge University Press. Schmidt, W. 1919. Die Personalpronomina in den Australischen Sprachen. Vienna: Akademie der Wissenschafen. Skarżyńska, K. 2002. We and They in Polish Political Discourse. Us and Others: Social Identities Across Languages, Discourses and Cultures, ed. A. Duszak, 249-64. Amsterdam and Philadelphia: John Benjamins. Thompson, S. A. 2002. ‘Object Complements’ and Conversation: Towards a Realistic Account. Studies in Language 26: 125-64. Thompson, S. A., and P. J. Hopper. 2001. Transitivity, Clause Structure, and Argument Structure an Conversation. Frequency and The Emergence of Linguistic Structure, ed. J. Bybee and P. J. Hopper, 27-60. Amsterdam and Philadelphia: John Benjamins. Valle, E. 1996. Representation of the Discourse Community in Scientific and Popular Writing. Essays & Explorations: A ‘Freundschrift’ for Liisa Dahl, ed. M. Gustaffsson, 157-70. Turku, Finland: University of Turku. Weber, E. 1993. Varieties of Questions in English Conversation. Amsterdam and Philadelphia: John Benjamins.