Imperatives, interruption in conversation, and the rise of discourse markers: a study of Italian guarda *.
This paper investigates the history of discourse markers, with special reference to Italian guarda. This discourse marker has arisen from the imperative guarda! `look!'. The imperative `look!' may entitle the speaker to an interruption, because it conversationally implicates that she has to say something extremely important that requires the immediate attention of the conversation partners. It is argued that speakers will tend to use this imperative even in situations where they do not have something important for the others to look at. They will do so simply because this imperative is so useful for floor-begging. At this point, guarda! is no longer an imperative, but a discourse marker. The variety of DM functions can be shown to be already present as side-effects of the imperative use. Furthermore it will be argued, contrary to most current work, that the genesis of discourse markers is not an instance of grammaticalization.
1. The history of discourse markers
Recent years have seen a steep increase in research on discourse markers (DMs). (1) Most of the work is synchronic in orientation; yet, diachronic aspects have not been neglected (Abraham 1991; Onodera 1995; Brinton 1996; Jucker 1997; Traugott 1999; Diewald 1999). Given that forms that can serve as DMs are (or were, at some earlier point in time) very often also employed in a variety of other functions, the main problem of diachronic DM research is the explanation of how forms of language (the etyma, as it were) acquire the function of a DM. In the functional-typological tradition, the diachronic rise of DM is typically seen as an instance of subjectification, whereby forms denoting "objective," ideational meanings acquire more speaker-based, subjective, attitudinal meanings in the course of time (Traugott 1995, 1999).
Onodera (1995), Brinton (1996), and Traugott (1999) investigate the history of DMs in Japanese, Old and Middle English, and Modern English, respectively. Onodera studies the Japanese DMs demo and dakedo, which originated from intrasentential clause-combining devices. Today, demo and dakedo are DMs, both of which can be used for pointmaking, floor-claiming, opening of the conversation, and change of topic. Brinton (1996) offers a fine-grained in-depth study of a variety of DMs. For example, the Middle English DM gan stemmed from the Old English aspectual (ingressive) verb onginnan `to begin'. Brinton (1996: 80-81) reports that in Middle English this verb turned into a DM with textual meaning marking "new or important stages in the plot development," which then at a later stage underwent an additional change to an "interpersonal (emphatic, intensive)" function. Traugott (1999) and Schwenter and Traugott (2000) study the history of in fact, which developed from a prepositional phrase via a sentence adverbial into a DM. All of these findings confirm the overall tendency of subjectification in the sense of a development from propositional via textual to attitudinal, subjective meanings.
But subjectification in itself, in the sense I have outlined, does not explain the rise of discourse markers. In which kinds of context do the etyma first come to be used as discourse markers, and why do speakers begin to use them in that way? Subjectification is the description of a recurrent tendency of semantic change, but in itself it does not provide an answer to these questions. In more recent work, Traugott (1997, 1999) has integrated subjectification into the "invited inferencing theory of semantic change" (IITSC). Following this account, subjectification is related to the conventionalizaton of conversational implicatures, where implicatures are actively made by speakers with a rhetorical purpose. Detges (2000a, 2000b, 2001) has complemented this theory by identifying various types of functional change as respectively speaker- or addressee-based and by showing in detailed historical analyses how different types of pragmatic motivation lead to different types of functional change.
In this paper, I will take a look at the Italian DM guarda/guardi, arisen from the imperative guarda `look!' DMs like this one pose two problems of a different sort:
i. How did the etymon turn into a discourse marker?
ii. How did the functional range of the discourse marker evolve?
As for the first problem, I will argue that the imperative came to be used more and more frequently because it provided speakers with an excellent justification for self-selection at turn-taking, even at nontransition-relevance places. The result of this strategical overuse was the reanalysis of the imperative as a DM with the function `Listen to me, I have something important to say'. Detges (2001) shows that floor-seeking is a context type that is the origin of various types of functional change. I attempt to show that the DM guarda, too, and possibly also other DMs, originate in contexts of floor-seeking.
As for the second problem, the full functional range of the DM guarda was not yet present at that stage. I will argue that more steps of strategical overuse and subsequent reanalysis gradually enriched the functional range of the DM. Therefore I will provide a synchronic description of the DM functions of guarda. However, I do not aim at a most complete account of its functional range. My main goal is to provide a plausible diachronic pathway of the development of the imperative guarda! to the DM uses. I should add, however, that my analysis is provisional to the extent that no data from ealier stages of Italian (where guarda was not available as a DM) are taken into consideration.
Guarda and its counterparts in other languages have received some attention in the literature (Mara 1986; Manili 1986; Pons Borderia 1998b; Dostie 1998, and references cited therein). In their studies on the Spanish and the French counterparts of guarda (mira and regarde, respectively), Pons Borderia (1998b) and Dostie (1998) also relate the function of these DMs to the capturing of the addressee's attention, inherently implicated in the meaning of the underlying imperative mira!/regarde! `look!' But I am not aware of a previous study that accounts for WHY THE IMPERATIVE TURNED INTO A DM.
In section 2, I give an overview of the functions of the DM guarda in contemporary spoken Italian. In section 3, I discuss occurrences of the imperative guarda introducing an interruption. Section 4 explains why this kind of use should be held responsible for the development of the imperative into a DM. In section 5, further effects of the interruption are discussed. In section 6, consequences of this analysis for the theory of DMs are pointed out, especially with respect to the question of whether their historical rise can be considered an instance of grammaticalization.
2. Italian guardi/ guarda
2.1. The DM guarda
The DM guarda appears in a variety of contexts and functions. The following list is based on 336 occurrences of the string guarda in the LIP corpus (i.e. including homophonous forms such as the imperative and the 3SG present verb form).
(1) Turn-initial DM after transition-relevance place (LIP, MB9) B: ah hai visto ali poveretto e morto cosi l'avevano ammazzato DM have you seen the poor guy? he's dead they killed him A: tu dici? Do you think so? B: ma secondo me si' DM yeah, I think so A: madonna oh god B: <?> l'hanno ammazzato era ricchissimo qualcuno l'avra fatto fuori they killed him. he was very rich somebody murdered him A: guarda che soffriva di cuore _ eh? guarda he had a bad heart DM?
A and B talk about the death of some other person. B claims that this person has been killed. A is skeptical and introduces her argument that this person had a bad heart with guarda. The utterance with guarda occurs at a transition-relevance place. In this utterance, A continues the topic under discussion but introduces a new aspect (a "subtopic") to which she calls B's attention. Furthermore, guarda has an adversative nuance here: A does not seem to believe the murder version and insists on a natural death.
(2) Turn-final (LIP, FE15) B: che che butta? what what's she throwing away? H: butta carciofi artichokes B: non lo so se butta I don't know if she's throwing them away H: butta carciofi she is B: signora nessuno lei ci ha un'antagonista di la dell'altro sesso madam nobody you have an antagonist of the opposite sex H: eh no ma ho sbagliato ho sbagliato radio guarda DM no but I've made a mistake I've chosen the wrong radio station guarda B: ha sbagliato radio che voleva chiamare la Raffai she has chosen the wrong radio station [she] wanted to call up Raffai
In a radio program where people can call up the moderator directly, the caller H says that she has inadvertently chosen the wrong radio station and finishes her turn with guarda. H's contribution is thereby somewhat paradoxically highlighted, in the sense that she highlights her apparent wish to quit the situation.
(3) Turn-initial at non-transition-relevance place (LIP, MA2) B: prendi una stecchetta di legno e la fai con la stecchetta di legno e con gli adesivi get a wooden stick and you make it using the wooden stick and the glue A: si'rio con la stecchetta di legno cerco yes I'll try with the wooden stick ... C: guarda e piu semplice a colori quattro quattro <?> due chiodini e <?> basta # velocissimo rapido guarda it's much easier with colors four four <?> some tacks and <?> it's done, really fast A: <???> B: oppure la gente li mette sopra una poltrona sopra un tavolo sopra otherwise people put them on an armchair on a table on _
A, B, and C are discussing how to fix a wall hanging A owns. The wall hanging is, however, not present in the situation; as A says earlier in the discussion, it is still in her cupboard and she is unsure as to what to do with it. C has an idea and interrupts A, introducing the turn with guarda. C calls the others' attention to her idea.
(4) Turn-medial, topic shift (LIP, FE6) I: ciao tesoro hi sweetheart C: chi e? who is it? I: chi sara' mai? who could it be? C: Oraziella Graziella I: tesoro mio my sweetheart C: allora carache numero ti do? DM dear which number shall I give you? I: ottantaquattro eighty-four C: ottantaquattro guarda cara anch'io guarda ho tanta ogliohoi ogliohoi voglio andare a letto eighty-four guarda dear and me guarda I really want [yawns] I want to go to bed
Here, guarda occurs turn-medially. This chunk is taken from a radio program where listeners (here: the speaker I) can call in directly and choose a number (in this case, "84") for a kind of lottery. I (Graziella) starts the conversation on a confidential tone, thereby somewhat imitating the moderator himself, who talked to the previous participants in much the same way. The moderator C then introduces a new topic (his alleged tiredness) with guarda.
(5) Phatic marker (LIP, FE9) A: proporrei l'antipastino di mare bellino I would recommend the seafood starter sweetheart B: ah _ ah ah ah A: con cozze with mussels B: ah _ ah ah ah A: eh eccetera eccetera DM etcetera etcetera B: guarda guarda A: un bell'antipastino di mare the nice seafood starter
In this example guarda constitutes a turn by itself. B expresses admiration for the dishes A is offering. Bazzanella (1990: 640) classifies this use of guarda as an interactional cue conveying surprise on the part of the addressee. As Bazzanella (1990: 639) notes, this function goes beyond mere back-channeling.
The functional contexts (1)-(5) of guarda overlap with the functions of its Spanish counterpart mira as studied by Pons Borderia (1998b).
(6) Introducing quotes (LIP, FB19) A: e quindi dice la moglie guadagna tre milioni il marito un milione dice guarda facciamo una cosa--and then she says--the wife earns 3 million, the husband 1 million--DM guarda let's do like this sta- stai a casa te che almeno entrano tre milioni perche' se sto a casa io entra un milione solo you stay at home, so at least we get 3 millions because if I stay at home we will only get one million
In this example, guarda occurs turn-medially and introduces reported speech (it occurs after the verb form dice `s/he says'). Since guarda seems to occur relatively frequently in such a context, one might be tempted to consider it a quotative marker. But it could as well be that guarda is simply a part of the quote. The latter solution is especially plausible when considering examples like (7):
(7) (LIP, MB49) A: chiamarti al cellulare I'll call you on the mobile B: ma no lo sai che non ci son problemi no lo so non capisco e a XYZ gli ho detto comunque XYZ but you know there aren't any problems, no, I know, I don't understand and to XYZ I have told anyway XYZ guarda c'ho questo problema qua gli ho pure detto guarda probabilmente ci conviene partire domani mattina guarda I have this problem I told him as well guarda probably we have to leave tomorrow morning A: mh
The fact that guarda shows up at the beginning of the quote but after the name of the addressee (anonymized as XYZ) suggests that guarda is not a quotative marker but that it is a part of the quote. I think, however, that it is not by accident that guarda appears frequently at the beginning of quotations. Quotations are, by their very nature, introductions of a different viewpoint. They are therefore very often topic shifts. (2) In order to underline this, the speaker may introduce her quotation with guarda even if the one whose utterance is reported did not say this.
(8) Hesitation phenomenon (LIP, MB11) A: ma io no _ non mi piace pero' no _ no no q fare una guarda pubblicazione pura delle lettere DM no, I don't like it but no no I'd prefer to guarda publish the letters as they are B: pura delle lettere cosi? the letters exactly as they are? A: si _ che pero _ cioe con un cappello in cui spieghiamo ecco nonostante _ si e si e parlato di pacifismo yes but with an introduction where we explain that nevertheless pacifism has been discussed
Here guarda is a mere hesitation phenomenon: it occurs turn-medially, inserted between an article and the corresponding noun (una pubblicazione).
2.2. The imperative guarda!
The DM guarda stems from the 2SG imperative of the verb guardare `to look'. Of course, guarda is still in use also as a full-fledged imperative verb form:
(9) Imperative (LIP, RE1) A: # # questa e bella # guarda pure questa quant' e questa e rasata fuori e flanellata dentro this one is beautiful look at this one as well ... outside is satin and it has a flannel lining
In a clothing store, A shows a garment and invites the listener to look (guarda!) in order to see its particular properties. (9) seems thus to be an unambiguous example for the use of guarda as an imperative. As with other Italian verbs, the imperative also exists in the polite 2SG form (guardi) and the 2PL form (guardate). These forms have counterparts as DMs, too (Bazzanella 1994: 146, 154; 1995). (3) The fact that the DM guarda is subject to inflection might be taken as an index of the relative recency of the development of the DM uses, and consequently as an encouragement to base a study of this development on a contemporary spoken-language corpus. In this paper, I will focus on the form guarda.
Having two different functions (imperative and DM), guarda is heterosemous (Lichtenberk 1991). Heterosemy means that a certain linguistic form has two different but related meanings; at the same time, the two uses can also be distinguished on purely syntactic grounds. Of course, a DM occurs in other syntactic distributions than a verb form. From a diachronic viewpoint, the interesting question is why the imperative did develop into a DM and how exactly the DM functions are related to the imperative uses. A solution will be proposed in the next two sections.
3. The imperative `look?--a "priority pass" for self-selection at turn-taking
3.1. `Look!' introducing an interruption
The key to the development of the DM lies, I believe, in the imperative use where it introduces an utterance that INTERRUPTS the preceding one. The LIP corpus has a nice example for this kind of use:
(10) Imperative B: come trovare il subagente? how can I find the subagent? A: che domande che fai? what questions are you asking? D: ah pure il subagente vuole pure insomma gente troppo bella questa questa la devi segnare troppo ah s/he even wants a subagent! hey guys that's a great one to remember A: guarda guarda che aspetto che c'ha look! look! What s/he looks like D: e scusate il disturbo [ridono] troppo bello e tutto troppo bello I am sorry to disturb you [laughing] it's really great it's really great
In this vivid, partly polemic multiparty conversation, A, B, C, and D are talking about doing business in the importation of Latin American wood paintings. B seems to be a newcomer to this kind of business and asks a question (come trovare il subagente?) that A and D find silly. A criticizes B for asking this question (che domande che fai?), while D, who seems to be more rude, laughs at B and treats him openly as foolish. A then interrupts D with the exclamation guarda guarda che aspetto che c'ha, thereby distracting the others' attention from the allegedly silly question and pointing to something A finds funny, probably the way D looks while making the sarcastic comments. (In the ensuing discourse, A criticizes D for treating B that way. A's interruption is therefore probably an attempt to protect B.) D then returns to her/his point, repeating the sardonic judgment troppo bello. Interestingly, D begins that utterance with an excuse after which all laugh, as if D accepted A's interruption.
3.2. Interruption as a conversational practice
Of course, an interruption such as A's is a blatant violation of turn-taking rules such as formulated in the seminal paper by Sacks, Schegloff, and Jefferson (1974). The turn-taking system in conversation provides for speakers beginning to talk only if they have been explicitly selected as next speaker by the previous speaker or if the previous speaker has finished her turn without selecting the next speaker, thereby enabling self-selection of any of the participants of the conversation (Sacks et al. 1974). Self-selection is possible only at transition-relevance places. The previous speaker, D, had not finished her turn: the last word in that turn, the adverb troppo `too' does not have a head that it could modify. It is not a possible completion point of the turn. Hence, A's turn begins at a non-transition-relevance place, with the word guarda. But, as can be seen from D's reaction after that interruption, the violation of the turn-taking rules seems to be compatible with conversational practice.
In conversation analysis, it eventually turned out that the rules of the turn-taking system as formulated by Sacks et al. were too strict. Turn-taking in conversation is not exclusively governed by the "formal" rules of self-selection at transition-relevance places and selection by previous speaker. Interruption, or "unsmooth speaker change," is indeed part of conversational practice and to a certain extent tolerated. More precisely, when a participant has to say something extremely urgent or considers the current turn as irrelevant, self-selection may be possible at non-transition-relevance places (Jefferson 1978; Duranti and Ochs 1979; Dausendschon-Gay and Krafft 1991). What is "urgent" and what is "irrelevant" are of course subject to debate and to point of view. (4) Furthermore, interruption may even be a form of positive politeness. It can be a sign of cooperative joint formulation (Tannen 1984: 30; Bazzanella 1991). (5) Orestrom (1983) has established a typology of reasons for interruption that can be observed in conversational practice, such as anticommunication ("stop that!"), protest, check-up question, and so on. Bazzanella (1991) has added "psychological need" and "force majeure" to these categories.
If a participant of the conversation finds that what she has to say justifies an unsmooth speaker change and wishes to begin to speak without waiting for the next transition-relevance place, she is well advised to begin her turn in a way that makes the violation of the turn-taking system rules understandable for the other participants, that is, one that makes them actually listen to what she has to say rather than continuing the current turn or fighting down her attempt to talk. A candidate interrupter must make the high relevance of what she has to say immediately obvious to the conversation partners. Even though there is a considerable amount of research in conversation analysis on interruption, especially on the reasons speakers seem to have for it and gender-related differences in their behavior, the linguistic means that speakers select for this purpose have, to my knowledge, not been studied systematically. What kind of means is effective for successful floor-seeking depends of course on the type of speech act the candidate speaker purports to perform. An illuminating study in this respect is the work done by Duranti and Ochs on Italian left-dislocation (LD). Duranti and Ochs (1979: 404) report that 40% of the Italian turn-initial left-dislocations they studied show up at an unsmooth speaker change. This finding suggests that left-dislocation is an appropriate tool for interruption attempting a topic shift: "LDs are effective means of seeking and occupying the floor because they nearly always relate to some general concern under consideration. The left-dislocated referent itself may have appeared in the prior talk and, hence, constitute an explicit legitimizer of subsequent talk [...]. Or, the left-dislocated referent is semantically linked to general concerns at hand [...] LDs may be successful topic-shifters in part because, while shifting focus of attention, they nonetheless are semantically relevant to the prior focus of attention" (Duranti and Ochs 1979: 406-407).
A speaker may find that an urgency justifying interruption is provided by objects or events that require the immediate attention of the participants of the conversation, maybe because the object or event is present only for a very short time, or because the event calls for some immediate reaction (these are the "force majeure" cases reported by Bazzanella 1991). In these cases, an ideal means for making the rule violation acceptable seems indeed to be the imperative `look!' This imperative verb form strongly conveys that the speaker knows of some visual object present in the situation that has hitherto escaped the other participants' attention, while at the same time being of high relevance for them. Furthermore it conveys that the speaker has reasons to believe that the other participants should immediately look at that object, thereby putting behind them the current activity, that is talking. In this sense, the imperative `look!' is a "priority pass" for self-selection that confers on the speaker the enormous privilege of starting a turn even before the next transition-relevance place. It is therefore not really surprising that in many languages the imperative `look!' has a DM offspring, such as Spanish mira (cf. Pons Borderia 1998a, 1998b), English look, French regarde (cf. Dostie 1998), German sieh mal/sehen Sie/schauen Sie, Portuguese olha. (6)
One can observe such an interruption in example (10). Guarda in A's turn is clearly an imperative meaning `look!', not a DM, as can be seen from the substantive aspetto `look' in the same turn, which again makes reference to visual perception. As has been shown, A is successful in interrupting D, that is, D effectively stops talking and restarts only after A has finished. The "visual object" justifying the interruption seems to be the way D looks while talking in an agitated way. Obviously the way someone looks while talking furiously is a view available only for a very short time, that is, during the time that person actually talks. In combination with the overall polemic character of the conversation, A's interruption is therefore justified by the situation.
4. Reanalysis of the imperative as a discourse marker
The reason why I believe that the interruption contexts such as (10) are so important for the recruitment of guarda as a DM is that this imperative is a very powerful tool in conversation. As I have described in the previous section, it entitles the speaker to interrupt the current speaker rather than to wait for the next transition-relevance place. An attractive tool like this one, which is, after all, in everybody's reach, lends itself to "improper use." That is, it is likely that speakers will utilize it in contexts where they do not have anything urgent to have the other participants look at. They will simply say "look!" in order to be allowed to begin a turn before the next transition-relevance place. Still, they may consider what they have to say as particularly important, even if they are not in a position to point to some "important thing" visible in the situation. (7) The result of this "abusive" look! is the reanalysis of the imperative as a DM. (8) An example of this use of guarda is (3), of which I repeat the most important parts for convenience:
(3) A: si io con la stecchetta di legno cerco yes I with the wooden stick I search ... C: guarda e piu semplice a colori quattro quattro <?> due chiodini e <?> basta # velocissimo rapido guarda it's easier with colors four four <?> two tacks and <?> it's done really fast
C interrupts A. C has an idea as to how to fix a wall hanging. C seems to assume that this idea is of particular importance for the discussion the participants are engaged in, which, for her, apparently justifies a violation of turn-taking rules. To this extent, (3) resembles (10). But it is highly probable that in (3) we are no longer dealing with an imperative, but with a discourse marker. For it is unlikely that C wants to SHOW the others something. C has an idea that she explains. But she does not have a particular object to show to the others. It is therefore unlikely that C wants the others to LOOK at something particular, given that the wall hanging they are talking about is not visible in the situation. Guarda is hence no longer an imperative. It means, `I have something important to say that justifies an interruption'. A semantic change has occurred: what used to be a conversational implicature of the imperative use is now a conventionalized meaning (cf., for the conventionalizaton of conversational implicatures as a pathway of semantic change, Traugott and Konig 1991). A corollary of the meaning change is that guarda in this particular interpretation cannot be analyzed as a verb anymore. A discourse marker is born. This is schematized in (11).
(11) Imperative guarda DM guarda as in (3) Meaning Look!' `I have something important to say' Implicature `I have something important to say'
The semantic change that led from the imperative to the DM is the result of joint work of speakers and addressees: speakers employ the imperative "illegitimately" in situations where they claim to have something very important to say (without, however, having an object to have the others look at). They do so because the imperative `look!' lends itself to the solution of a frequently occurring problem in communication, namely floor-seeking. This in itself does not make for a semantic change. The change is completed by hearers who correctly understand that the speaker is "exaggerating" and, by the same token, replace the old imperative meaning, which they feel is no longer intended by the speaker, by the new DM meaning.
The view on semantic change expressed here follows work by Ulrich Detges (2000a, 2000b, 2001) and myself (Detges and Waltereit i.p.). There it is assumed that it is not inherent properties of the forms of language that provoke language change. Rather, it is assumed that functional change may be provoked by speakers who use a form in a new way that serves a frequently occurring communicative purpose. For example, resultative constructions may diachronically turn into perfects because speakers can use resultative constructions in order to invite the inference that some past event really did happen, given that the result they are talking about provides strong evidence for the event that led to that result (cf. Detges 2000a). It is indeed a frequently occurring communicative problem to have to provide evidence for some past event. The recognition or the "uncovering" of this technique on the part of listeners amounts to a reanalysis of the resultative construction as a perfect. Likewise, Detges (2001) convincingly suggests that the grammaticalization of subject pronouns (for example in the history of French) can ultimately be traced back to floor-seeking strategies at turn-taking (which is manifestly a frequently occurring communicative purpose). In order to highlight their own contribution to conversation as relevant and noteworthy (thereby justifying self-selection), speakers introduce their turn with a nonobligatory contrastive 1SG pronoun, even if the propositional content of their turn does not actually contrast with the preceding turn, as in Me, I like this color. This "improper use" of a contrastive device will lead to a reanalysis of the contrastive pronoun as a noncontrastive pronoun, which, in turn, will eventually end up in obligatorification of subject pronouns. In a similar way, the Spanish prepositional accusative can also be shown to go back to floor-seeking strategies (Detges 2001: 281-321).
Correspondingly, the imperative `look!' lends itself to rhetorical use because, as argued for in this paper, it provides to speakers an effective means for self-selection. Obviously, self-selection is a highly frequent problem in a locally managed and party-administered social practice such as conversation and it calls for effective solutions. One could go even further and argue that the imperative `look!' is a more powerful means of floor-seeking than simple turn-initial contrastive devices such as left-dislocation. Contrastive devices are effective to the extent that they allow speakers to make obvious to their conversation partners that they have a relevant contribution to make, relevant because it differs in some way from the previous contributions. To this extent, turn-initial contrastive devices claim effectively that the incipient turn will be worth listening to. The imperative `look!', however, is even better in this respect because not only does it effectively promise a contribution worth listening to, but it also justifies an interruption. That a speaker has something to say that is different from another speaker's talk is not really sufficient to justify an interruption, because everyone has, after all, equal rights to take and to finish her turn. But the imperative `look!' invokes a higher-order "force majeure" right that occasionally may justify an interruption.
5. Side-effects of interruption
As described in section 2.1, the DM guarda has a variety of distinct functions, not only the function of `indicating that the speaker believes she has something important to say that justifies an interruption'. At present I am not in a position to establish a diachronic order of appearance of these functions. It can be shown, however, that all of these functions have a conventionalized meaning that, in one way or another, separately encodes a conversational implicature of the imperative use of guarda!
The effects for the structure of discourse provoked by an unsmooth speaker change, introduced by the imperative guarda! as in (10), seem to include at least the following:
i. A speaker change occurs. By the very nature of the interruption, the imperative `look!' marks the border between two speakers in the overall sequence of discourse.
ii. The speaker has the floor. By virtue of the described semantic and pragmatic properties of the imperative `look!', the other conversation partners will believe that the speaker has something highly relevant to say and will, in cooperative conversation, give up the current turn and then listen to her.
iii. The speaker has some time to formulate her utterance. Given that the imperative `look!' does not directly encode what the speaker has to say but only announces something important, she still has some time (at least the time needed for uttering the imperative) to think about the formulation of what she wants to say.
iv. A topic shift occurs. The imperative `look!' implicates that the topic of the ensuing discourse will be one that the conversation partners have not touched upon in the preceding discourse.
v. The topic of the ensuing discourse is highlighted as particularly important and interesting (otherwise the interruption would not be justified).
Comparing now these side-effects with the DM functions of guarda, one easily sees that each function of guarda as listed in section 2.1 corresponds to one side-effect of the imperative use:
i. The quotative use (cf. ) exploits the change-of-speaker side-effect of the imperative. In the turn-medial occurrence in (6), the DM guarda is like a colon and announces direct speech, the quotation beginning with guarda. The change-of-speaker side-effect (or maybe only the topic-shift side-effect) of the imperative is preserved in this DM use.
ii. Turn-initial guarda after a smooth speaker change, (1), preserves the floor-seeking function of the imperative, without, however, claiming the right to interrupt the previous speaker.
iii. The hesitation-phenomenon (or speech-management) function of the DM guarda as in (8) preserves the floor-maintenance function of the imperative. As a hesitation phenomenon, guarda conveys that the speaker is momentarily unsure as to how to continue her current turn, without, however, abandoning her right to speak.
iv. The turn-medial topic-shift DM guarda as in (4) draws upon the topic-shift side-effect of the imperative, without direct consequences for turn-taking.
v. As a phatic marker (5), guarda seems to draw upon the highlighting side-effect of the imperative. The DM in this example constitutes a whole turn. The speaker expresses her surprise about what the other conversation partner says. The DM is like an echo to the preceding turn.
As I have argued in section 4, it is most plausible to assume that the turn-initial guarda at unsmooth speaker change is the first DM function to have emerged from the imperative, because this function is the direct result of a reanalysis of the floor-seeking overuse of the imperative. Once this reanalysis has taken place, guarda is recruited to the class of DMs. It has acquired the DM function `Listen to me, I have something important to say'. At this early stage, it is, however, unlikely to already have the full functional spectrum that it has today. More steps of language change must have happened in order for it to attain its actual functional range. The more recent DM functions are the respective results of further semantic changes of the DM, it currently not being possible to establish a chronological order of appearance of these further functions. (9) But each of these encodes a side-effect already present in the interrupting imperative use. I assume that the further steps of semantic change follow the model of the first step of semantic change. Note that the DM function of guarda, as described in section 4, is still a powerful tool in communication (even though maybe not as powerful as the imperative) and therefore has an inherent rhetorical potential for overuse. It can still be very effective at floor-seeking. That is, speakers can take advantage of this function in contexts that do not legitimately justify its use; and listeners recognize what the speaker really wants to say, thereby reanalyzing this overuse as a new function of the DM, which in turn may then be overused again in other contexts, and so on. A concatenation of changes of this type dissects, as it were, the side-effects of the imperative use, resulting in a variety of separate DM functions.
6. Implications for the theory of discourse markers
The observations made in this paper on the rise of the Italian DM guarda may shed some new light on issues of general interest in DM research.
6.1. Speaker motivation
I have argued in this paper that the DM guarda is the historical result of a strategically motivated overuse of the imperative `look!' insofar as this imperative entitles the speaker to begin a turn before the next transition-relevance place. That is, the recruitment of guarda to the class of DMs is the result of a speaker motivation that is itself related to the structure of the discourse and the interaction. A glance at other Italian (and other) discourse markers suggests that this type of motivation might be relevant also for other discourse markers. For example, the DM ecco, which has several textual and interpersonal functions (cf. Bazzanella 1995: 251-253), stems from the uninflected one-word presentative ecco `here you are' (cf. French voila). Given my analysis of guarda, it seems plausible to suspect that the presentative ecco can also justify an interruption, for example in order to direct the attention of the conversation partners to something they have been awaiting. Other verbs of perception, such as sentire `to hear', lend themselves to the recruitment to DMs in a somewhat similar fashion (cf. Pons Borderia 1998b and Dostie 1998 for studies on DMs based on verbs meaning `to hear' or `to listen'). The DM diciamo, which again has metatextual and interpersonal functions (Bazzanella 1995: 250-251), derives from the 1PL subjunctive verb form `let us say'. The sentence `let us say' has a strong social aspect because it makes the following utterance appear as joint work of speaker and addressee. At the same time, it has a mitigating function to the extent that by introducing an utterance with this verb form, the speaker invites the addressee to express agreement or disagreement. By marking an utterance as joint work of speaker and addressee, one indirectly requests a reaction of the addressee as to whether he agrees or not with this monopolization. Diciamo is hence an interactionally powerful word (and a sentence) that lends itself to eliciting phatic reactions and at the same time highlighting the adjacent utterance.
Of course, further work is needed to test these intuitions. But it seems that the type of motivation behind the DM guarda is not a completely isolated case.
In DM research, there is a long-standing debate as to whether the functional spectrum of DMs can be shown to be a set of context-dependent instantiations of one single meaning (monosemy) or whether it represents several meanings of the lexical item (polysemy) (see Mosegaard Hansen 1998:85-90 for a recent overview). The historical model presented here presupposes the polysemy stance, because the separate functions discussed are assumed to have developed one by one in discrete steps. This assumption is corroborated by language comparison (cf. note 6). A monosemy account would be unable to explain the fact that the DM form can have a varying number of functions at different points in time, given that such an account would require all functions to be equally derivable from the same underlying representation.
6.3. Is the rise of discourse markers an instance of grammaticalization?
In most current work it is assumed that the historical genesis of DMs is an instance of grammaticalization (Abraham 1991; Brinton 1996; Traugott 1997, 1999; Diewald 1999). Indeed, one can easily observe typical features of grammaticalization, such as a semantic change toward pragmatic strengthening in the rise of DMs (Brinton 1996: 60-64). Furthermore, by the very nature of their rise, DMs are recruited through a "decategorialization" of their source constructions (Brinton 1996; Pons Borderia 1998a: 185, 1998b: 218), which is again a typical feature of grammaticalization. In the case of guarda, this means that the imperative is no longer a verb form, but a DM. However, authors concede that the rise of DMs is not completely analogous to the "classical" examples of grammaticalization, like the grammaticalization of tense or of prepositions. In particular, DMs do not become obligatory and one does not observe the scope shrinkage typical for grammaticalization (Brinton 1996; Traugott 1997). Furthermore, it is not even immediately obvious that DMs are part of the GRAMMAR of a language in the first place. (10) Traugott (1997) redefines grammaticalization and even grammar so as to include DMs into the grammar and to subsume their historical genesis under grammaticalization. I would, however, like to adopt a more traditional view. It seems that there are certain properties of "classical" grammaticalization that are shared by the rise of DMs, and that there are other properties of "classical" grammaticalization that are not shared by the rise of DMs. Instead of minimizing these differences, I would like to suggest that linguists should be grateful for them and analyze them in more detail, because they might tell us important things about the nature of both discourse markers and grammaticalization. As a starting point for such an analysis (and as a suggestion for further research), I would like to take Lehmann's (1985) grammaticalization theory.
Strictly speaking, the question of whether DMs are part of the grammar is orthogonal to the question of whether their historical genesis is an instance of grammaticalization. Lehmann (1985) takes grammaticalization not as "recruitment to the grammar," but as the gradual loss of autonomy of a linguistic sign along his scales. It is therefore insightful to check the development of DMs against these scales, whether or not one takes DMs as part of the grammar. Lehmann's (1985) parameters of grammaticalization are listed in (12).
(12) Parameters and processes of grammaticalization (Lehmann 1985: 306)
Paradigmatic Syntagmatic Weight integrity (attrition) scope (condensation) Cohesion paradigmaticity bondedness (paradigmaticization) (coalescence) Variability paradigmatic variability syntagmatic variability (obligatorification) (fixation)
Lehmann defines grammaticalization as the gradual and simultaneous loss in paradigmatic and syntagmatic weight and variability, and gain in paradigmatic and syntagmatic cohesion. But only the parameters loss of integrity (attrition) and gain in paradigmaticity (paradigmaticization) prove correct for the rise of DMs. It is true that DMs may undergo phonetic shrinkage (cf. Brinton 1996), even though guarda is not (yet?) a case in point, and it is also true that they undergo semantic changes of the type Lehmann had in mind when referring to grammaticalization processes. Hence, the rise of DMs may imply attrition. Furthermore, it is true that DMs belong to a paradigm, the paradigm of discourse markers, whereas their etyma did not belong to this paradigm.
Discourse markers represent paradigmatic choices. But, as one anonymous reviewer suggested, they form only a very "loose" paradigm. So maybe even this parameter of grammaticalization is doubtful in the case of discourse markers. At any rate, none of the other parameters can convincingly be said to hold for the rise of DMs. DMs never become obligatory. Their scope usually does not reduce. On the contrary, it is typical for them to have larger scope than their etyma. They are always free morphemes with respect to their adjacent elements, hence they do not undergo coalescence. Furthermore, their position does not become fixed; on the contrary, they have the well-known property of being able to occupy various places in the utterance.
Hence, from an orthodox Lehmannian perspective, the rise of DMs cannot count as grammaticalization, because only two of its criterial properties (both of which belong to the paradigmatic axis) are fulfilled. None of the three syntagmatic parameters gives evidence for grammaticalization. In the following, I would like to make some speculations on the reasons for this.
A very important similarity of the rise of DMs and grammaticalization seems to be the fact that both grammaticalization and the recruitment of DMs are the historical relics of rhetorical overuse of a construction, as alluded to in the comparison with the rise of perfect markers in section 4. The initial rhetorical overuse gives rise to the development of a discourse technique, which speakers will employ routinely to achieve a certain frequently occurring communicative goal (Detges 2000a: 366-368). As a result of routinization, the construction is shifted out of the focus of attention and may then undergo attrition. Paradigmaticization is a natural side-effect of the recruitment to a functional class, be this a category of grammar or the category of discourse markers. These two parameters are therefore readily accounted for. But there seems to be a very important difference between the rise of DMs and grammaticalization: the source constructions of DMs are very often syntactically independent by themselves. The imperative guarda! (and its counterpart in other languages) is a short but syntactically complete sentence. The same holds for the subjunctive form diciamo and the presentative ecco, both of which I briefly alluded to in section 6.1. But the source constructions of grammaticalization are most often not syntactically complete. For example, nouns that turn into prepositions (Latin casa `house' > French chez `at') require, of course, at least a complement at the outset of the grammaticalization process. Likewise, tense markers often derive from inflected verb forms, which of course typically need complements. This difference between the source constructions of DMs and those of grammaticalization makes for a great difference in the subsequent language change. In both types of process, that is, the rise of DMs and grammaticalization, the forms undergoing a change are shifted into the background of attention. In the case of grammaticalization, this is a grammatically structured background, because the constructions are already integrated into a larger syntactic structure. With the rise of DMs, however, the communicative background is simply the sequential structure of the discourse (which is itself by definition not part of the grammar of a language), because the source constructions are grammatically independent.
Therefore, grammatical obligatorification of DMs will not occur for the simple reason that their source constructions are not integrated into a structure of sentence grammar. (11) The same fact seems to be responsible for the preservation (or even augmentation) of autonomy with regard to all parameters of the syntagmatic axis. DMs apply by their very nature to discourse units of variable size. As is well known in DM research (cf. Mosegaard Hansen 1998: 113-128 and references cited therein), DMs do not apply to form-based units, but to function-based or even action-based discourse segments. They cannot therefore be subject to syntagmatic restrictions that could be measured by solely GRAMMATICAL (i.e. form-based) terms. The same holds for their source constructions.
Given that they are themselves syntactically independent and complete, they cannot contract syntagmatic relations that are dictated by categories of sentence grammar.
Whereas grammaticalization is the result of a rhetorical overuse that concerns other items of the same proposition, the rise of DMs is related to rhetorical overuses that have to do with the discourse they are part of (Detges 2001: 431). When saying the imperative `look!' the speaker does not want to highlight some element of the same sentence (this would be literally impossible given that the sentence is made up of the imperative alone), but she wants to highlight some element in the turn she is about to begin. The DM guarda as a historical result of this highlighting strategy is therefore not an obligatory part of sentence grammar but a (by definition) optional element of discourse structure.
Received 26 June 2001 Revised version received 19 November 2001
* I would like to thank Sarah Dessi Schmid, Ulrich Detges, Paul Gevaudan, Peter Koch, Maj-Britt Mosegaard Hansen, Salvador Pons Borderia, Alessandra Rivetto, Nicoletta Rivetto, Scott A. Schwenter, Paul Shannon, the audience of a presentation at the University of Tubingen, and two anonymous reviewers for comments and suggestions. Correspondence address: Romanisches Seminar, Universitat Tubingen, Wilhelmstr. 50, D-72074 Tubingen, Germany. E-mail: email@example.com.
(1.) The following abbreviations will be used: DM = discourse marker, 1 = first person, 2 = second person, SG = singular, PL = plural.
(2.) Cf. for a similar analysis of French alors Mosegaard Hansen (1998: 335-348). The discussion of the "quotative" use of guarda has benefited a lot from remarks by M. B. Mosegaard Hansen on an earlier version of this paper and by an anonymous reviewer.
(3.) One does find, however, occurrences where inflection of the DM and address style of the communication do not match; cf. example (2). The DM is therefore not as sensitive to address style as verb forms are. This confirms an analogous finding by Pons Borderia (1998: 216-217) for guarda's Spanish counterpart mira.
(4.) The work on interruption in conversation has in fact largely been concerned with gender differences in interruption and hence with the question of to what extent interruption can be proved to be a sign and a practice of male dominance over women. See Zimmerman and West (1975), West and Zimmerman (1983 ), Beattie (1983), Murray (1985).
(5.) As observed by an anonymous reviewer, interruptions introduced by guarda are, however, unlikely to be taken as sign of positive politeness. As shown in section 2, guarda does not typically announce an attempt of cooperative joint formulation.
(6.) Even though in many languages there are DMs derived from the imperative `look', these DMs do not necessarily all have the same functions (Pons Borderia 1998a: 182). For example, the French DM regarde seems to have a much more restricted range of functions than its Italian counterpart (cf. Dostie's 1998 analysis of regarde).
(7.) One might say that the "deictic" use of `look!' is hereby transformed into an "anaphoric" (more precisely, cataphoric) use (Paul Gevaudan, personal communication).
(8.) Also Pans Borderia (1998b: 216-219), in his analysis of the Spanish DMs mira and oye, assumes that the function of capturing the addressee's attention of mira is the DM function closest to the imperative function.
(9.) A comparison of the DMs resulting from the imperative `look!' in several languages might, however, provide interesting insights into the typical sequence of changes, especially because not all of the functions of the Italian DM are analogously available in the other languages (cf. note 6).
(10.) In a relatively early paper, Fraser (1988: 32) claims that DMs are "part of the grammar of a language." In another place in the same paper, however, he characterizes them as "lexical expressions which are syntacticly [sic] independent of the basic sentence structure" (1988: 27). One might ask what kind of grammar allows grammatical items to be syntactically independent of sentence structure. In Fraser (1999) he seems to stick to the view that DMs are part of the grammar of a language.
(11.) It should be emphasized, however, that it is not properties of the source constructions as such that drive them to develop into DMs. Rather, they are chosen by speakers in order to attain a certain effect related to the structure of the discourse or the interaction because they have properties (for example, their syntactic independence) that are useful for such an effect.
Abraham, Werner (1991). The grammaticization of the German modal particles. In Approaches to Grammaticalization, Elizabeth Closs Traugott et al. (eds.), vol. 2, 331-380. Amsterdam: Benjamins.
Bazzanella, Carla (1990). Phatic connectives as interactional cues in contemporary spoken Italian. Journal of Pragmatics 14, 629-647.
--(1991). Le interruzioni "competitive" e "supportive" `verso una configurazione complessiva. In Dialoganalyse, III: Referate der 3. Arbeitstagung Bologna 1990, Sorin Stati et al. (eds.), vol. 1, 283-292. Tubingen: Niemeyer.
--(1994). Le facce del parlare. Un'approccio pragmatico all'italiano parlato. Florence: La Nuova Italia.
--(1995). I segnali discorsivi. In Grande grammatica italiana di consultazione, Lorenzo Renzi et al. (eds.), vol. 3, 225-257. Bologna: Il Mulino.
Beattie, Geoffrey (1983). Talk. Milton Keynes: Open University Press.
Brinton, Laurel J. (1996). Pragmatic Markers in English. Grammaticalization and Discourse Functions. Berlin and New York: Mouton de Gruyter.
Dausendschon-Gay; and Krafft, Ulrich (1991). Tache conversationelle et organisation du discours. In Linguistische Interaktionsanalysen. Beitrage zum 20. Romanistentag 1987, Ulrich Dausendschon-Gay et al. (eds.), 131-154. Tubingen: Niemeyer.
Detges, Ulrich (2000a). Time and truth: the grammaticalization of resultatives and perfects within a theory of subjectification. Studies in Language 24, 345-377.
--(2000b). Two types of restructuring in French creoles: a cognitive approach to the genesis of tense markers. In Degrees of Restructuring in Creole Languages, Ingrid Neumann-Holzschuh et al. (eds.), 135-162. Amsterdam: Benjamins.
--(2001). Grammatikalisierung. Eine kognitiv-pragmatische Theorie, dargestellt am Beispiel romanischer und anderer Sprachen. Unpublished Habilitation thesis, Universitat Tubingen.
--; and Waltereit, Richard (i.p.). Reanalysis vs. grammaticalization: a simplest systematics of functional change in grammar. Zeitschrift fur Sprachwissenschaft.
Diewald, Gabriele (1999). Die Entwicklung der Modalpartikel abet--ein typischer Grammatikalisierungsweg der Modalpartikeln. In Internationale Tendenzen der Syntaktik, Semantik und Pragmatik, Hans-Otto Spillmann et al. (eds.), 83-91. Frankfurt/Main: Lang.
Dostie, Gaetane (1998). Deux marqueurs discursifs issus de verbes de perception: de ecouter/ regarder a ecoute/regarde. Cahiers de lexicologie 73, 85-106.
Duranti, Alessando; and Ochs, Elinor (1979). Left-dislocation in Italian conversation. In Discourse and Syntax, Talmy Givon (ed.), 377-416. Syntax and Semantics 12. New York: Academic Press.
Fraser, Bruce (1988). Types of English discourse markers. Acta Linguistica Hungarica 38, 19-33.
--(1999). What are discourse markers? Journal of Pragmatics 31, 931-952. Jefferson, Gail (1978). Sequential aspects of storytelling in conversation. In Studies in the Organization of Conversational Interaction, Jim Schenkein et al. (eds.), 219-248. New York: Academic Press.
Jucker, Andreas (1997). The discourse marker well in the history of English. English Language and Linguistics 1, 91-110.
Lehmann, Christian (1985). Grammaticalization: synchronic variation and diachronic change. Lingua e stile 20, 303-318.
Lichtenberk, Frantisek (1991). Semantic change and heterosemy in grammaticalization. Language 67, 475-546.
LIP = De Mauro, Tullio; et al. (1993). Lessico di frequenza dell'italiano parlato. Ricerca a cura dell'Osservatorio linguistico e culturale italiano OLCI dell' Universita di Roma "La Sapienza". Milan: Etaslibri.
Manili, Patrizia (1986). Sintassi di connettivi di origine verbale. In Parallela 2. Aspetti della sintassi dell'italiano contemporaneo, Klaus Lichem et al. (eds.), 165-176. Tubingen: Narr.
Mara, Edith (1986). Per un'analisi dei signali discorsivi nell'italiano parlato. In Parallela 2. Aspetti della sintassi dell'italiano contemporaneo, Klaus Lichem et al. (eds.), 177-189. Tubingen: Narr.
Mosegaard Hansen, Maj-Britt (1998). The Function of Discourse Particles. A Study with Special Reference to Spoken Standard French. Amsterdam: Benjamins.
Murray, Stephen O. (1985). Toward a model of members' methods for recognizing interruption. Language and Society 13, 31-41.
Onodera, Noniko Okada (1995). Diachronic analysis of Japanese discourse markers. In Historical Pragmatics, Andreas Jucker (ed.), 393-437. Amsterdam: Benjamins.
Orestrom, Bengt (1983). Turn-Taking in English Conversation. Lurid: Gleerup.
Pons Borderia, Salvador (1998a). Conexion y conectores. Estudio de su relacion en el registro informal de la lengua. Valencia: Universitat de Valencia.
--(1998b). Oye y mira o los limites de la conexion. In Los marcadores del discurso. Teoria y analisis, Maria Antonia Martin Zorraquino et al. (eds.), 213-228. Madrid: Arco Libros.
Sacks, Harvey; Schegloff, Emanuel A.; and Jefferson, Gail (1974). A simplest systematics for turn-taking in conversation. Language 50, 696-735.
Schwenter, Scott; and Traugott, Elizabeth Closs (2000). Invoking scalarity: the development of in fact. Historical Pragmatics 1, 7-25.
Tannen, Deborah (1984). Conversational Style: Analyzing Talk Among Friends. Norwood, NJ: Ablex.
Traugott, Elizabeth Closs (1995). Subjectification in grammaticalisation. In Subjectivity and Subjectivisation, Dieter Stein et al. (eds.), 31-54. Cambridge: Cambridge University Press.
--(1997). The role of the development of discourse markers in a theory of grammaticalization. Paper given at the 12th International Conference on Historical Linguistics, Manchester 1995 (version 11/97).
--(1999). The rhetoric of counter-expectation in semantic change: a study in subjectification. In Historical Semantics and Cognition, Andreas Blank et al. (eds.), 177-196. Berlin: Mouton de Gruyter.
--; and Konig, Ekkehard (1991). The semantics/pragmatics of grammaticalization revisited. In Approaches to Grammaticalization, Elizabeth Closs Traugott et al. (eds.), vol. 1, 189-218. Amsterdam: Benjamins.
West, Candace; and Zimmermann, Don (1983 ). Small insults: a study of interruptions in cross-sex conversations between unacquainted persons. In Language, Gender and Society, Barrie Thorne et al. (eds.), 103-117. Rowley, MA: Newbury House.
Zimmerman, Don; and West, Candace (1975). Sex roles, interruptions and silences in conversation. In Language and Sex. Difference and Dominance, Barrie Thorne et al. (eds.), 105-129. Rowley, MA: Newbury House.
|Printer friendly Cite/link Email Feedback|
|Publication:||Linguistics: an interdisciplinary journal of the language sciences|
|Date:||Sep 1, 2002|
|Previous Article:||Reflexives and resultatives: some differences between English and German *.|
|Next Article:||Partial agreement in second-language acquisition *.|