Berman et al. / Discourse stance / 01-4-3 / p. 1
Final version
Discourse stance
Ruth Berman,a Hrafnhildur Ragnarsdóttir,b and Sven Strömqvistc
aTel Aviv University / bIceland University of Education / cLund University
In press, Written Language and Literacy 5,2 2002.
(R. A. Berman, H. Ragnarsdóttir, & S. Strömqvist. 2002. Discourse stance. Written Languages and Literacy, Volume 5, 2 )
The aim of this article is to integrate findings reported in the preceding articles in this collection, employing a global discourse perspective labeled discourse stance. The paper attempts to clarify what is meant by this notion, and how it can contribute to the evaluation of text construction along the major variables of our project: target Language (Dutch, English, French etc.), Age (developmental level and schooling), Modality (writing vs. speech), and Genre (personal experience narratives vs. expository discussion). We propose a general conceptual framework for characterizing discourse stance as a basis for an empirically testable potential model of this key aspect of text construction and discourse analysis. Unlike the cross-linguistically data-based studies reported in the rest of this collection, which involve quantitative as well as well as qualitative analyses, this concluding article presents selected pieces of text from our sample to serve as case studies that illustrate our general line of reasoning, rather than to test specific hypotheses.
1. Introduction
The term stance has been used in the discourse literature in different ways. For example, Biber & Finegan (1989) define stance as “the lexical and grammatical expression of attitudes, feelings, judgements, or commitment concerning the propositional content of a message”(1989:124) – to include adverbs, verbs, and adjectives which mark affect, certainty, doubt, hedges, emphasis, possibility, necessity, and prediction. Ochs 1990, 1996 specifies “stance” as one of four dimensions that she discusses in considering the relation between language and culture. She defines stance as “a socially recognized disposition,” distinguishing epistemic stance, “a socially recognized way of knowing a proposition, such as direct (experiential) and indirect (e.g., secondhand) knowledge, degrees of certainty and specificity,” vs. affective stance, a “socially recognized feeling, attitude, mood, or degree of emotional intensity” (1990:2).
These studies derive from quite different perspectives: Biber and his associates (cf. Biber 1995, Biber et al. 1998) analyze recorded written and spoken texts, in terms of the statistical distribution of different clusters of linguistic markers, as expressing a particular “stance style.” These researchers deliberately proceed from analysis of linguistic forms,1 with no a-priorirelationship to a particular discourse context or communicative setting. In contrast, Ochs and her associates (e.g. Ochs & Schieffelin 1983) proceed from the communicative context of situation to analysis of linguistic forms occurring in different socio-cultural settings. They focus on conversational interaction, and advocate an ethnographic methodology to assess how children acquire the ability to “use language constitutively,” on the assumption that “epistemic and affective stance has … an especially privileged role in the constitution of social life” (Ochs 1996: 420).
The framework for analysing discourse stance which we propose below is parasitic on the above research, and on a large body of other literature that ranges across literary studies (e.g. Bakhtin 1986, Leech & Short 1981); sociolinguistic analyses of narrative and conversational interactions (Labov 1972, Tannen 1989); psycholinguistic research on conversational usage (Clark 1986, Clark & Gerrig 1990); studies focused on the comparison of written vs. spoken discourse (Tannen 1982, Chafe 1994); and research on children’s developing discourse abilities (Shatz 1985, Reilly 1992). Relevant notions that have been alluded to in the literature include the following (in roughly chronological order).
(a) Evaluation. This critical notion in studies of narrative discourse, since the pioneering work of Labov & Waletzky 1967, refers to those elements of a narration which flesh out the sequentially ordered events that it describes (in a presumably objective, descriptive fashion) by providing the narrator’s personal commentary on those events, and subjective interpretation of them, and so renders a story more expressive and interesting to the listener.2
(b) Involvement. Chafe 1982 and Tannen 1985 use this term to characterize the interactive features of texts. Tannen points out, importantly from our viewpoint, that involvement need not be confined to prototypically interactive situations of face-to-face conversation, or even to narrative type texts. She refers instead to the “relative focus of involvement,” noting that literary language, like ordinary conversation, is dependent for its effect on interpersonal involvement. It fosters and builds on involvement between speaker and hearer rather than focusing on information or message. It also depends for its impact on the emotional involvement of the hearer. In contrast, expository prose, associated with literate tradition … depends for its impact on impressing the audience with the strength and completeness of its argument, that is, with aspects of the lexicalized message (Tannen 1985:139–40).
(c) Perspective. Our view of discourse stance also interacts with the notion of perspective —although, again, the two are not the same. The term perspective is used in linguistic analysis primarily in discussion of grammatical aspect, as in the distinction made by Smith 1991 of “situation-type aspect” (or [space] Aktionsarten)vs. “viewpoint aspect” (cf. Goldsmith & Woisetschlaeger 1982). In developmental studies of discourse analysis, perspective has been considered largely in terms of “agentivity,”, with grammatical voice a major distinguishing feature of different perspectives on a situation (Budwig 1990, Berman 1993, Berman & Slobin 1994:515–38). This is shown in our analysis contrasting the use of passive voice in several languages (Jisa et al. 2002). Others talk about point of view,which Brown & Yule (1983:146–48) relate to topic ordering in narratives. Relatedly, Li & Zubin 1995 assume that “choice of an anaphoric referring expression — full NP, pronoun, or zero — might be a function of context-dependent cognitive factors”, in distinguishing linear from rhetorical continuity, as well as in the two perspectives of expressive vs. reportive framing in narrative discourse. These and related ideas are perhaps most broadly articulated by Chafe (1994:132), in terms of “point of view” and of “immediacy” vs. “displacement,” taking as a starting point the “fact that consciousness is oriented from the point of view of an experiencing self.”
(d) Distancing devices. In work which formed the background to the present analysis, we considered the nature of such dev ices, in the sense of linguistic means used to express a particular discourse stance along a range of distinctions — including personal(ized) vs. general(ized), immediate vs. detached, involved vs. distanced, specific vs. general, and subjective vs. objective (Berman 1999, Jisa & Vigué 1999, Ravid & Cahana-Amitay 1999, Tolchinsky 1999). The present study attempts to refine and clarify these distinctions in a more principled frame of reference (§2), as background to specification of the linguistic devices which serve the over-all purpose of expressing discourse stance (§3).
Our proposal aims at a “top-down” approach to the analysis of discourse stance, at the same time specifying the forms of linguistic expression which speaker/writers use in realizing this aspect of text construction. We start by trying to define the functional parameters involved in this notion (§2) and then examine the linguistic forms which speaker/writers deploy in expressing stance (§§3–4). In this bidimensional approach to form/function relations, we are aided by the methodology evolved for data collection and analysis, as laid out in Berman & Verhoeven 2002 (§1). Our sample provided us (uniquely in the research literature, to the best of our knowledge) with directly comparable texts dealing with shared thematic content in narrative vs. expository discourse, in both speech and writing, across four different age groups. The fact that exactly parallel procedures were adopted across different languages means that we can directly address the impact of available structural devices and of rhetorical preferences in a given target language — a recurrent theme in the preceding articles of this collection.
Our characterization of discourse stance thus takes into account two genres of monologic texts (for comparisons of narrative with other modes of discourse, see Bruner 1986, Giora 1990, Stutterheim & Klein 1999), both written and spoken, in developmental perspective. Our major motivation is to examine the complex interaction between linguistic forms and discourse functions by considering a broad array of linguistic devices as giving expression to several different dimensions of discourse stance. In the present context, we aim to provide a functionally based overview that integrates topics discussed elsewhere in this collection, including the lexicon, noun-slots, verb-slots, voice, and propositional attitudes.
2. Conceptual framework
We consider the notion “discourse stance” as referring to three interrelated dimensions of text-construction: Orientation (Sender, Text, Recipient); Attitude(Epistemic, Deontic, Affective); and Generality(of reference and quantification —specific vs. general). These are functional dimensions which apply across texts, and so differ from what we term “Propositional Attitudes”, whose scope is the (semantic) proposition or something like a (syntactic) sentence (see Reilly et al. 2000).3 Central to our present proposal is the idea that all or any of these three dimensions of stance -- orientation, attitude, and referential specificity or generality -- can be alternated within a piece of discourse. A given text, may start out with a “sender” orientation as a deictic center, and then switch to taking the text or even the recipient-addressee as a point of reference and then either return to the speaker/writer perspective or not. Similarly, a single text may contain any one or more of the three types of attitudes we are identifying — epistemic, deontic, and affective — and it may be both specific and general in reference to persons, places, and times.
2.1 Orientation
This dimension concerns the relation between the three participating elements in text production and interpretation: sender (speaker or writer), text (narration or exposition), and recipient (hearer or reader). A sender orientationis subjective, and is deictically centered on the speaker/writer. It tends to be deontically judgemental or affective in attitude, and specific in reference; it reflects personal involvement in the content of the text, relating to events and ideas that the speaker/writer has experienced or thought about. These distinctions are always relative; e.g., expressions like I think, je trouve, creo contain an epistemic predicate, yet they proceed from a deictic, sender-oriented viewpoint. A recipient orientationis communicatively motivated; it takes into account, or at least appears to be addressing, the hearer/reader quite directly. This is found in expressions like you know, or use of 2nd person pronouns in a non-personal sense, with generic reference (Spanish “tu” arbitrario). Thus, when an English-speaking woman in the course of an oral narrative makes a generalization to the effect that people who you know they take advantage of your trust”, this might be construed as more recipient-oriented than the semantically corresponding people that are known to / who clearly take advantage of one´s / a person´s trust).4 A text orientationtakes the object that is being produced orally or in writing as a conceptual or cognitive point of reference. It relates the representation of the content of the piece of discourse itself (cf. I’m not quite sure how to formulate the problem, or What I´m going to talk (or write) about is …) to a totally distanced, impersonal metatextual level of orientation, e.g. When discussing issues such as this …, or In considering the topic of … In our database, expressions like the latter are confined to the older subjects, mainly among university graduate adults, occasionally in the texts of high-school adolescents.
2.2. Attitude
Distinctions of attitude also apply at the more local level of propositional attitudes (Reilly et al. 2000). But as has been noted, e.g. by Ochs 1996, such distinctions express a quite general discourse stance as well. An epistemic attitudeconcerns a relation between a cognising speaker/writer and a proposition, in terms of possibility, certainty, or the evidence for the individual’s belief that a given state of affairs is true (or false). A deontic attitudeadopts a judgemental, prescriptive, or evaluative viewpoint in relation to the topic. An affective attitude, in contrast to the epistemic, concerns a relation between cognising speaker/writer and their emotions (desire, anger, grief etc.) with respect to a given state of affairs.5 These distinctions can thus be ranged on a cline — from the more objective, abstract, and universalistic epistemic attitudes; through socially conditioned deontic attitudes, shared within a group familiar to the speaker/writer; and on to the most subjective reactions and personal feelings that an individual holds in relation to a given topic. Psychological studies on socio-cognitive and moral development (e.g. Hersh et al. 1979), as well as the findings from discourse analysis in our own sample (Berman & Verhoeven 2002, §4.1), indicate that the ability to combine and interrelate these different attitudes flexibly and appropriately, in a single discourse context, is the hallmark of a socially developed adolescent/adult.
2.3 Generality
This dimensionconcerns the degree of generalization vs. specificity of reference to people, places, and times referred to in the text. To a large extent, this is a function of, or parasitic on, the two previous dimensions, since speaker orientation is necessarily highly specific, while cognitive attitudes may be quite general and universalistic in scope. We distinguish three levels of expression in this respect: Personal or Specificin Reference (e.g. I / my parents think, my / this boy’s father made me / him apologize); Generic(e.g. People / We tend to think, It depends on your / one’s attitude);and Impersonal(It´s well known, the fact that, Spanish se sabe, French il faut). As the examples indicate, the linguistic means for expressing these different levels of generality depend on the available devices and typological properties of the different languages. But distinctions in generality of reference, as in orientation and attitude, are assumed to be relevant regardless of the particular target language.
This three-pronged approach to discourse stance represents a deliberate attempt to view language use and discourse rhetoric along clines of interacting factors, rather than in dichotomous terms of written vs. spoken, personalized vs. objective, or involved vs. distanced. This over-all conception is supported by our analysis of developing text construction abilities, which aims at integrating top-down and bottom-up approaches to discourse analysis, from the perspective of a multiplicity of linguistic forms that can be recruited to express a range of discourse functions. It is facilitated by the database at our disposal, which allows us to do a careful examination of comparable, specially elicited texts produced by non-expert subjects in different languages, at four levels of Age (and schooling), in two Genres (narrative and expository), and in two Modalities (speech and writing).
3. Linguistic forms of expression
Under this heading, we move from “function” (the notion of discourse stance) to “form,” in the sense of overt linguistic markings of stance — morphological, syntactic, and lexical. The breakdown in Table 1 was devised in the framework of a panel presentation concerned with “Talking and writing about conflict situations at different ages and in different languages” (Berman 1999).
Table 1 near here
Papers presented at the panel (Ravid & Cahana-Amitay 1999 for Hebrew, Jisa & Viguié 1999 for French, and Tolchinsky 1999 for Spanish) focused on the linguistic forms used to distinguish the direct, immediate, and highly personalized perspective of personal-experience narratives from the more distanced, abstract, and impersonal rhetoric of expository discourse. This analysis was based on a contrast we drew between these two types of texts, working with a database collected at a prior stage of the study reported on in here. In our current thinking about discourse stance, as formulated in the preceding section, we have abandoned this rather dichtomous view to take account of the complex nature of the topic as well as the form and content of the texts that we have analyzed.
Nonetheless, there are good grounds, psychological and linguistic as well as developmental, for setting narrative apart from other discourse genres (as cogently argued by Bruner 1986, von Stutterheim & Klein 1989). This is particularly true in the case of personal experience narratives, as compared with discussions of a topic such as we elicited (Berman 2001b). For example, across age groups and languages, the dominant tense in the expository texts is the (timeless) present, as compared with a preference for past tense forms in the narratives (Ragnarsdóttir et al. 2002). Across Age and Language, nominals functioning as surface subjects tend to be more generic, impersonal, and/or lexical in expository texts, with a higher proportion of personal pronoun subjects in narratives (Ravid et al. 2002). Again across Age and Language, expository texts contain more modal-type predicate modifiers (like should, can)than the narratives, which have relatively more aspectual verbs (like start, keep on).Where modal expressions do occur in narratives, they are typically “agent-directed” (Reilly et al. 2002.)
The ideas presented in §2.2, concerning “a clineor continuum of rhetorical means for moving from the personal to the general, from concrete to abstract, from specific to general, from immediate to distanced, from involved to detached,” thus seem to provide a useful starting point for examining our proposed characterization of discourse stance. Relevant linguistic distinctions are as shown in Table 1, along the dimensions of word-internal morphology, lexicon, syntax, and semantic content:
The way and the extent in which the devices in Table 1 are deployed will vary across a number of dimensions: text type or genre — e.g. a personal narrative vs. fictive short story and these compared with an academic text, a newspaper report, or a procedural text; modality – speech or writing; target-language typology; available structural and rhetorical options; as well as the rhetorical preferences and style of individual speaker-writers. For example, as shown in earlier articles in this volume, use of passives and reliance on pronominal subjects and impersonal constructions interact markedly with whether a language requires a surface subject in simple clauses (a typological property which sets Spanish and Hebrew apart from the other languages in our sample). And while all the languages in our sample have passive constructions which are structurally quite productive, they show different distributional patterns across the texts in the five languages examined for this topic.
The listing above demonstrates the multiple levels and types of linguistic devices involved in expressing discourse stance. It also has an advantage over certain prior analyses since it departs from a strictly dichotomous division in favor of a continuum or cline. The way it is presented implies a directionality that is helpful for purposes of analysis, rather than being “correct” in principle. It should not be taken to mean that the features to the left of the chart are in some sense inferior to, or more juvenile, less developed, or less expressive than those to its right. The claim is, rather, that a maturely expressive and rhetorically proficient text will be weighted to one end of the scale or another in keeping with (a) the particular context of discourse and (b) the communicative goals of speaker/writers on a given occasion. In fact, a hallmark of skilfully proficient speaker/writers is that they use a wide range of these different devices in conjunction, and that they do so flexibly, appropriately, and without putting the consistency of the text at risk.