An increasing number of publications have emerged from inter-disciplinary research into a
contentious area known to some as literary semantics (Eaton, 1979; Margolin, 1984), to others as stracturalist narratology (Barthes, 1966; Bremond, 1966; Chatman, 1978; Genette, 1966; Todorov, 1966), to the Russians as literary semiotics (Bakhtin, 1980; Lotman, 1975; Voloshinov, 1983), to some American academics as studies in narrative structure (Johnstone, 1987) but to other American literary critics (Booth, 1961 and the Chicago school) as part of the exercise in practical criticism. Other contributions came from hermeneutics (Ricoeur, 1984), stylistics (Cassirer, 1975; Enkvist, 1973; Fish, 1970) and different schools of literary criticism. Clearly, it is difficult to situate the analysis of narrative voice exclusively in any of these disciplines. However, following the lead of many linguists who have perceived scope for linguistics to have its say in the elaboration of narrative structures and strategies (Adamson, in progress; Greimas and Courtes, 1976, 1982; Fowler, 1986; Fowler et al, 1979; Fraser and Joly, 1979; Leech and Short, 1981; Wright, 1987), the main task of this chapter is to explore the tools that linguistics can offer in the identification of narrative voice. Among the fundamental premises underlying the methodology adopted herein it is believed that propositions are uttered by speakers who appropriate a given linguistic system for their own subjective purposes (to be defined below); that speakers necessarily leave traces in the utterance of their attitudes both to their message and to their addressee(s). As Lyons has put it (1977:724): If propositions are treated as psychological entities, rather than as purely abstract third-order entities, then it is natural to treat as their location the persons (or the minds, or brains, of the persons) who have what philosophers might describe as a propositional attitude (knowledge, belief, etc.). It is necessary therefore to adopt a linguistic framework which argues the case for the need to systematically study which linguistic forms produce which subjective effects. Then it is up to the narratologist to measure the frequency and degrees of intensity of these forms and come up with conclusions that will hopefully pave the way for a typology of narratives (Ducrot and Todorov, 1981).
I.1 Jakobson's shifters: the need for an extension:
Modern theories of enunciation (reviewed by Fuchs, 1979), the growing interest in egocentric particles (Russel, 1940), and the influential findings in the indexical aspect of language (Todorov 1970) especially on deixis (Benv‚niste 1966, 1970; Fillmore 1973; Lyons 1977, 1979) deal with various aspects of the same class of linguistic categories, namely Jakobson's class of shifters. The notion of duplex structures, Jakobson argues (1956), plays a vital role in the study of language. Speaking from a dualist Saussurian perspective (namely the dichotomy of langue vs parole), he distinguishes four duplex structures, two of which are circular (message referring to message, and code referring to code), and the other two overlapping (message referring to code and code referring to message) (1956:133): In order to classify the verbal categories two basic distinctions are to be observed:#1) speech itself (s), and its topic, the narrated matter (n);#2) the event itself (E), and any of its participants (P), whether "performer" or "undergoer".#Consequently four items are to be distinguished: a narrated event (En), a speech event (Es), a participant of the narrated event (Pn), and a participant of the speech event (Ps), whether addresser or addressee. When the code obligatorily refers to the message the category involved is what, following Jesperson, he calls shifters. Shifters are verbal categories akin to Pierce's indexical symbols, since, having a general conventional definition at the code level, they are partly symbols, and yet, being in existential relationship with the object they designate, they come under the category of indices. Thus shifters are categories implying compulsory reference to the speech event (../Es) (procŠs de l'‚nonciation) and/or its participants (../Ps) (namely the enunciator and enunciatee(s)). The discrimination of shifters from among the whole set of grammatical categories is a necessary exercise, Jakobson insists, in any systematic classification of such categories. According to whether or not grammatical categories meet the requirement of referring to (../Es) and/or (../Ps), they will qualify either as shifters or nonshifters. Starting from this distinction, Jakobson classifies eight categories in the following table: P involved P not involved --------------------------------------------------------------- non-shifters gender/number voice status/aspect taxis (Pn) (PnEn) (En) (EnEn) --------------------------------------------------------------- shifters person mood tense evidential (Pn/Ps) (PnEnPs) (En/Es) (EnEnsEs) --------------------------------------------------------------- fig .1. condensed from Jakobson's two diagrams (1956:136) The categories classified as shifters are what Jakobson calls "person" (Pn/Ps) characterising the participants of the narrated event with reference to the participants of the speech event; "tense" (En/Es) involving the narrated event with reference to the speech event; "mood" (PnEn/Ps) having to do with the relation between the narrated event and its participants with reference to the participants of the speech event reflecting thus the speaker's view of the character of the connection between the action and the actor or the goal; and finally what he calls, for want of a better name, "evidential" (EnEn/Es) involving a narrated event, a speech event and a narrated speech event (alleged source of authority about the narrated event). These categories will be examined in detail in the first part of the thesis mainly in the chapters on deictic subjectivity (chapters two, three, four and five) and on modal and affective subjectivity (chapters six and seven). One common denominator between them is their reference, whether explicit or implicit to the speech event (act of enunciation) and its participants (both enunciator and enunciatee(s)). This characteristic is of vital importance in the definition of shifters, and will be shown to be instrumental in the detection of subjectivity indices or cues within the utterance "énoncé".
I.2 Problems with Jakobson's "non-shifters":
The controversial categories are those classified as "non-shifters". The following problems could be detected in them. I.2.1 The categories of "gender" and "number": Jakobson claims that both "gender" and "number", abbreviated as "Pn" characterise the participants themselves "without reference to the speech event". Against this unqualified statement, it will be argued that both gender and number do involve an ideological investment on the part of the speaker besides the cultural mediation infiltrating through the language used. Thus, the choice of these categories within the utterance could be taken as a cue revealing the latent attitude of the speaker to the participant(s) of the narrated event (the délocutés). Pushed a little further, this selection could be part of the speaker's discursive strategy to influence the addressee and to produce the desired effect on him. A serious criticism of the presumed transparency of the category of gender comes from Violi (1987) who, being a semiotician, is mainly concerned with the uncovering of the cultural codes and symbolising processes underlying the grammatical forms in a language: ce n'est pas la donnée naturelle de la différence des sexes, sa "matérialité grossière" comme dirait Sapir, qui structure les catégories linguistiques, mais c'est la symbolisation que cette différence a déjà subie, sous des modalités largement antérieures et indépendantes des formes grammaticales spécifiques avec lesquelles les diverses langues les reflèteront. Postuler un investissement de sens antérieur … la forme linguistique, signifie lire la différence sexuelle comme une structure déjà signifiante, déjà symbolisée, et capable … son tour de produire un sens et des symbolisations. Grammatical gender does not transparently reflect natural distinctions. Behind the assignment of gender lies both the speaker's culture, as embodied in and imposed by the language used, and in varying degrees, the speaker's own strategies. The semantic act of gender assignment is not one of immediate objective observation but constructed and negotiated in the process of its genesis. That is why grammatical gender offers a wide range of possibilities for manipulation. For instance, Kress and Hodge (1979) argue that when the gender system is overt, it fixes a set of sexual associations for a culture. Such associations are often wider and more subtle than the overt system indicates. Their argument becomes stronger when they analyse the ideological manipulations of implicit or covert gender in advertisements. After comparing the gender status of "car" in German, French, and English, they conclude that the basis for the gender assignment is not biological but ideological. Similarly, the choice of number (singular, dual, plural or any other possible arrangement) is not simply dictated by objective reality. In fact, there are many cases where speakers enjoy considerable "freedom" to define things either as singular or plural [although the word "freedom" has to be qualified in view of the fact that a great deal of these ideological stances are already prestructured and imposed as mediating grids by the language used] (Lodge, 1981). As will be explored in the next chapter (section on person displacement), English, like French and, more markedly Arabic, offers its speakers possibilities to use the plural for other purposes than number; the pronouns "we" and "vous" are not simply the plurals of "I" and "tu" respectively. Besides, the speaker's point of view or vis‚e is a decisive factor in determining whether words like "the government", "the United States" or indeed "Liverpool" are singular or plural; or whether many substances are apprehended as countable or uncountable (Quirk et al 1972). In Arabic, virtually all inanimate objects or natural phenomena are conjugated in the third person female singular. But most importantly, the choice of number in newspaper editorials, political speeches, and any such well processed discourse, is widely used as an efficient means of persuasion to manipulate the readers or audience by foregrounding or exaggerating certain items and minimising or obscuring others (see Fowler et al 1979 for interesting exemplification ). In English, the plural has different effects ranging between imprecision, uncertainty and dullness of focus on the one hand and exaggeration or foregrounding on the other. The former effect is held by Kress et al (1979:89) to be a convenient means for obscuring facts and mystifying the addressees : The general effect of a plural is to indicate a compound field. A sentence in which everything is in the plural will feel more diffused and complex than a sentence which uses only singulars, even if the reality described is the same. The latter effect is commonly obtained in newsbulletins and correspondent reports from countries where a number of "victims" of oppression or of "champions" of a cause in a demonstration is deliberately amplified. The most blatant manipulations of number are found in military communiqués in a war. Thus, contrary to Jakobson's claim, the categories of gender and number do refer to the speech event, or at least to the ideological investment of the participants of the speech event. Their exclusion from the class of shifters is not justified. I.2.2 The categories of "status" and "aspect": The second misleading category in Jakobson's "non-shifters" is what he calls "status and aspect" characterising the narrated event itself "without involving its participants and without reference to the speech event". Presumably, by this he is referring to what other linguists (Benv‚niste, 1966; Joly & Roulland, 1981; Todorov, 1970) call sentential modalities [modalit‚s phrastiques] such as the affirmative, the negative, the interrogative, the subjunctive and the conditional "moods" (as they are traditionally known). Most of these categories will be discussed under the section on modality (chapter seven), but it may be useful here to point out where Jakobson is wrong. The affirmative has a very strong modal force, and by implication, a greater speaker investment in the utterance. As Lyons (1977:808-809) has put it: It would be generally agreed that the speaker is more strongly committed to the factuality of "It be raining" by saying It is raining than he is by saying It must be raining. It is a general principle, to which we are expected to conform, that we should always make the strongest commitment for which we have epistemic warrant. If there is no explicit mention of the source of our information and no explicit qualification of our commitment to its factuality, it will be assumed that we have full epistemic warrant for what we say....There is no epistemically stronger statement than a categorical assertion. Similarly, the imperative and interrogative are "you-centred forms", to use Adamson's terminology, because they are oriented toward the addressee, and by implying "you" necessarily presuppose "I". Such linguistic forms indicate the presence of the speaker indirectly by necessitating the presence of an addressee. The dialogical inter-dependence of the first and second persons entails that when one of these interlocutors is implied in an utterance, the presence of his dialogical counterpart is automatically established. As Adamson has perceptively stated (in progress1:22): In texts, therefore, YOU and YOU-centered forms are particularly important in establishing "I" as a speaking presence. By the YOU-centered forms, I mean imperative and interrogative forms, both of which prompt us to infer an I-YOU dialogue, even when the presence of an "I" or a YOU has not been explicitly stated. Indeed, in the major syntactic functions of enunciation in an utterance enumerated by Benvéniste (1970), categories such as questions and imperatives have the function of enlisting and calling for a response from the addressee(s) to the speaker's performative act. Thus they inevitably presuppose reference both to the speech event and especially to the participants of the speech event: L'INTERROGATION, qui est une énonciation construite pour susciter une "réponse", par un procès linguistique qui est en même temps un procès de comportement … double entrée. Toutes les formes lexicales et syntaxiques de l'interrogation, particules, pronoms, séquence, intonation, etc, relèvent de cet aspect de l'énonciation. (Benvéniste, 1970:15) Consequently, it could be argued that the occurrence of questions, imperatives, and other forms of intimation should be taken to provide significant clues indicating the speaker's presence, and his intention to influence his addressee(s) one way or another (Benvéniste, 1966). The study of performatives and the different forms of doing has been the major contribution of speech act theories (Austin, 1962; Grice, 1975; Searle, 1969) and pragmatics (Leech, 1983; Levinson, 1983). What is common to these findings is the construal of enunciating as a form of doing and the exploration of the forces contained in an act of enunciation (locutionary, illocutionary and perlocutionary forces). What is vital for the present argument is that, contrary to Jakobson's claim, these forms make a necessary reference to the present of enunciating (what Jakobson calls the speech event and its participants) and have to be included as shifters. I.2.3 The categories of "voice" and "taxis": The third category classified as a "non-shifter" is what Jakobson terms "voice" (PnEn) which is supposed to characterise the relation between the narrated event and its participants "without reference to the speech event or to the speaker", whereas the fourth category failing to satisfy Jakobson's criteria for shifters is "taxis" (EnEn) "charaterising the narrated event in relation to another narrated event and without reference to the speech event". The difficulty with these categories is that, although they do not explicitly involve reference to the speech event or its participants, they do make this reference implicitly through other channels. For instance, in narratives, (especially those in the first person), one of the main constraints on storytellers is to capture in their talk various levels of what Johnstone (1987:44) calls "footing": Footing, as Goffman defines it (1981:128) is "the alignment we take up to ourselves and the others present as expressed in the way we manage the production or reception of an utterance". It is the "projected self" of a speaker as this self emerges in interaction. A person involved in an interaction and telling a story about a previous interaction must manipulate footing on at least two levels: the level of the storytelling interaction and the level of the interaction in the story. One's alignment with respect to others changes in the course of interaction, and these changes, on both levels, must be encoded too. The two levels of story and discourse in structuralist narratology (Chatman, 1978; Genette, 1966) cannot be separated. In other words, the telling/narrating/discourse level is always underlying to and presupposed by the histoire level. In linguistic terminology, enunciation [énonciation] is necessarily and always presupposed by the utterance [énoncé]. There has to be an axis of reference with respect to which relations are perceived, this being either a deictic centre (Fillmore 1973), i.e.the I/here/now ordering personal and spatio-temporal relations, or a perceptual centre (Adamson, in progress) experiencing and perceiving things, or a cognitive centre (Gréimas and Courtès, 1976) known also as a sentient centre (Lotman, 1975) affectively and modally reacting to things and evaluating them.
I.3 Suggested adjustments to Jakobson's model:
What is important as a corrective to Jakobson is to argue that not to refer to the speech event or to its participants is a structural impossibility . In fact, as Gustave Guillaume has put it (quoted in Joly 1981:545): Tous les actes d'expression - sans exception aucune - sont affectifs vu que tous ont pour objet d'agir sur l'interlocuteur, de l'affecter. Il n'est pas de phrase qui ne soit affective. Herman Parret (1983) claims that speakers could be said to have an affective competence "compétence passionnelle" that inevitably marks their discourse. Therefore, from the preceding discussion, it should have become apparent that it is very difficult, if not virtually impossible, to stop the erosive effect of the categories of shifters. It will be argued, together with Guespin (1976) that there is no clearly defined borderline between shifters and non-shifters. A large and ever growing shady area lies in between. However, what could be deduced from Jakobson's argument is that grammatical (and particularly verbal) categories shift differently, or, better still, linguistic forms belong to different kinds of shifters and lend themselves to different degrees of transparency. In brief, there are tendencies either towards shifting in [embrayage] where the subjective presence of the speaker is more readily apparent, if not foregrounded (as is the case with deictic categories), or towards shifting out [débrayage] where the presence of the speaker is less readily available and a disjunction of the utterance from I/here/now onto what is not-I/not-here/not-now takes place (tendency towards objectivity). To use the terminology of Greimas and Courtès (1982), the subjectivising tendency will be called engagement whereas the objectivising tendency will be called disengagement. Instead of Jakobson's sharp distinction between shifters and non-shifters, a scale or cline of gradations could be envisaged with "total" engagement or disengagement at either extremity. The linguistic forms could then be assigned degree-of-shifting values ranging from the most minimal to the most maximal. Consequently,the dichotomy between "subjective" and "objective" utterances does not operate at the deep level since, by its very definition, an utterance is the result of a subjective act of appropriating a linguistic system by a speaker (see below). In so far as there is a speaker latent to any conceivable speech event, then every utterance could be said to be subjective. What warrants the dichotomy on the other hand is that there is in an utterance a number of cues giving the impressions either of subjectivity or of objectivity (Morot-Sir, 1982:128). In the discursivisation process (Gréimas and Courtès, 1982) the speaker inscribes in the utterance either an engaging egocentric force or else a disengaging objectivising force. Thus, the import of the categories enumerated by Jakobson is that, when used in discourse, they either tend to give the impressions of subjectivity or at least to prompt the reader to take into account their shifting reference (Jakobson's shifters called here overt shifters) or on the contrary give impressions of objectivity by suppressing overt reference to the speech event and its participants (Jakobson's "non-shifters" called here covert shifters). It is important to bear in mind the fact that the notions of subjectivity and objectivity are relative and set into relief the very problematic of indeterminacy in borderline cases.
I.4 Jakobson's linguistic functions:
Two more fruitful concepts could be borrowed from Jakobson's legacy. In his widely quoted closing speech (1973), Jakobson listed several functions of language, two of which are extremely important for the present argument. The first of these is the emotive function (involving the attitude of the speaker/locutionary agent toward his own discourse). Since the main concern of this thesis is the systematic search for clues indicating narrative voice, it stands to reason that an analysis of utterances in terms of this function is particularly rewarding. The second function to borrow from Jakobson is the conative function (involving the speaker's consciousness or vis‚e of his addressee) as could be reconstructed from elements in the utterance. This function is particularly helpful in understanding the working of person deixis and, in particular, what, following Voloshinov (1983), will be termed the social orientation of the utterance (see chapter two). It could be concluded thus that Jakobson's legacy has paved the way towards a linguistics of enunciation where the speaking subject has a central place.
I.5 Emile Benvéniste's legacy: the nature of linguistic subjectivity:
The problematic of shifters found a considerable boost in the work of Benv‚niste on linguistic subjectivity and the formal apparatus for enunciation (1965, 1966, 1970 in particular). Operating within the structuralist duality of langue vs parole, he defines linguistic subjectivity with respect to them (1966:263): Le langage est donc la possibilité de la subjectivité, du fait qu'il contient toujours les formes linguistiques appropriées … son expression, et le discours provoque l'émergence de la subjectivité, du fait qu'il consiste en instances discrètes. Le langage propose en quelque sorte des formes "vides" que chaque locuteur en exercice de discours s'approprie et qu'il rapporte … sa personne. The emptiness, but not meaninglessness, of these forms provides the key feature of what is generally understood by shifters as construed in an extended framework. Shifters are thus inexorably linked to the emergence of the speaking subject within the utterance. It is because these empty forms lend themselves for appropriation by an individual speaker who by so doing transforms what is conventional and codified into something personal and idiosyncratic that they are called shifters. Hence, the study of shifters finds a natural place as a vital integral part in the study of linguistic subjectivity. This is what has been admirably perceived by Benvéniste (1966:259-60): C'est dans et par le langage que l'homme se constitue comme sujet; parce que le langage seul fonde en réalité, dans sa réalité qui est celle de l'être, le concept d'"‚go". La "subjectivité" dont nous traitons ici est la capacité du locuteur … se poser comme "sujet"...n'est que l'émergence dans l'être d'une propriété fondamentale du langage. Est "égo" qui dit "égo". It is this process of appropriation of the linguistic system by an individual speaker which defines the concept of enunciation. The speaker by annexing for his own use the formal apparatus of a language necessarily leaves traces of his presence as a speaker on the surface of his utterance. The task of the linguist is thus to focus on the imprint (marks, traces, cues, indices, clues...) of the process of enunciation in the utterance (Benvéniste, 1970:14; Kerbrat, 1980). These cues are pervasive. Speech is so impregnated with subjective markers that it is inconceivable to study its function without recourse to them (Bevéniste, 1966:261). The occurrence of such cues emanates from the relationship between the speaker, the utterance and especially the other participants both in the speech event and the narrated event. As Ducrot and Todorov (1981) have pointed out, the presence of these cues in different degrees of intensity in every utterance is a presence-indicator, revealing information on the speaking subject. It is from the centrality of the speaker in his utterance that the linguistic indices of enunciation stem (Benvéniste, 1970:14): L'acte individuel d'appropriation de la langue introduit celui qui parle dans sa parole. C'est l… une donnée constitutive de l'énonciation. La présence du locuteur … son énonciation fait que chaque instance de discours constitue un centre de référence interne. Cette situation va se manifester par un jeu de formes spécifiques dont la fonction est de mettre le locuteur en relation constante et nécessaire avec son énonciation. Most modern theories of enunciation are indebted to Benvéniste for this systematic search for clues and their assigning to a centre or axis of reference. The gains to narratology from this linguistic approach are innumerable. Person, spatial and temporal deictic forms will only take their meaning when assigned to an internal centre (presented below as the deictic centre). Modal auxiliaries, adverbs and all other realisations of modality (chapter seven) presuppose the subjective attitude of a locus of consciousness (Greimas's cognitive centre). The selection of lexical items, and a miscellany of other forms betokens the presence of an affective/ideological or sentient centre. The linguistic items presupposing an interaction of the individual with the environment stem from a perceptual centre. The attribution of all these forms to their respective centres draws on what Kerbrat (1980:31-2) calls faits énonciatifs: Dans cette perspective restrainte, nous considérons comme faits énonciatifs les traces linguistiques de la présence du locuteur au sein de son énoncé, les lieux d'inscription et les modalités d'existence de ce qu'avec Benvéniste nous appellerons "la subjectivité du langage". Nous nous intéresserons donc aux seuls unit‚s "subjectives"...porteuses d'un "subjectivème". (italicised in the text) Expression is inseparable from expressivity (Joly, 1973, 1979, 1980). That is, encoding a message is a complex process where subjective elements infiltrate into the speaker's already culturally prestructured initial project at different levels of the encoding process (Greimas's discursivisation process) so much so that the final product (the utterance) is a necessarily modified version of an already subjective input (LeGoffic, 1980). Consequently, the speaker is present everywhere in his utterance whether we as addresses or overhearers perceive his presence as explicit or implicit. I.6 Narratology's need for enunciatory linguistics. One "literary-oriented" frame searching for subjectivity cues which explicitly attempts to bridge the gap between the contemporaneous disciplines of linguistics and literature on the one hand, and the historically related disciplines of rhetoric and stylistics on the other is Cassirer's argumentative stylistics defined (1975:39) thus: Its objective is to make semantical and logical analyses of argumentative and persuasive (influence exerting) texts to investigate which techniques the writer uses in order to achieve which effect, to reveal and expose illusory objectivity in allegedly factual statements, etc. Argumentative stylistics is consequently directed towards the question of the means used to achieve a particular effect and - vice versa - what these means as used in a text can indicate about a non-explicitly expressed intention or attitude of the writer. As adapted for the present argument, argumentative or interpretative stylistics requires a systematic search for indicators of voice (piecing together linguistic indices of subjectivity, or noticing their scarcity, attributing them to their respective centres, having insights into the particular world views and mindstyles (Fowler, 1986) underlying them, exposing the speaker's strategies and latent intentions, etc. Summary: We need a linguistic frame which undertakes to retrieve the most salient subjectivity markers in language, to inventory through illustrative examples from literary extracts some of their most striking effects, and to indicate how such cues could serve as presence indicators and primary clues in the identification of narrative voice. This is what the thesis precisely attempts to explore.