ELSEVIER
Journal of Pragmatics 28 (1997) 781-794
On the co-construction of counterfactual reasoning Noriko Akatsuko Department of East Asian Languages and Cultures, UCLA, Los Angeles, CA 90095, USA
Abstract Using Japanese data collected following the 1994 Northridge earthquake, I would like to suggest that counterfactual thinking is an essential necessity for human existence. Specifically, a novel account of co-construction presented here provides an important insight into the nature of counterfactual reasoning by ordinary people in everyday life. It will be demonstrated that Japanese speakers spontaneously co-construct the line of reasoning, 'P [DESIRABLE], because if not P, then not Q [UNDESIRABLE]', where P and Q stand for FACT (e.g., P = 'the earthquake occurred at 4:30 a.m.' and Q = 'the casualties were very small'). These speakers were total strangers until they met at the data collection site to talk about their own earthquake experiences. By jointly and collaboratively producing elements of counterfactual reasoning, the pairs of speakers are congratulating themselves by conveying the message 'We are lucky; it could've been so much worse'. There is a striking similarity between Fauconnier's (1985) made up examples in English and the Japanese authentic discourse data, which leads us now toward a universal hypothesis that this line of reasoning is not only a natural one for every human being, it also serves to demonstrate and express sympathy and mutual sharing.
I. Introduction
This is an inquiry into the ordinary usage of counterfactual reasoning in everyday life. To the best of my knowledge, with the exception of Akatsuka (forthcoming a), this topic has remained virtually unexplored in the domain of theoretical linguistics. Our major research question will be: When do we typically invoke counterfactual reasoning, and why? Using the data collected in Japanese, Korean, and English by I would like to dedicate this study to Herbert Morris. Without his continuous support and encouragement during his tenure as Dean of Humanities at UCLA, it would have been a sheer impossibility to establish graduate studies in Japanese linguistics at UCLA. I am grateful to Susan Strauss for originally stimulating my interest in the dynamic, interactive nature of discourse, and particularly for helping me see how participants in conversation do indeed jointly participate in expressing and constructing ideas, all while reflecting the nature of the grammar of that particular language. I am also gratefld for her critical reading of this manuscript and of all earlier versions of it. Her input with respect to the content and her editorial comments have greatly enhanced the quality of this paper. 0378-2166/97/$17.00 © 1997 Elsevier Science B.V. All rights reserved PH S0378-2166(97)00074-X
782
N. Akatsuka / Journal ~f Pragmatics 28 (1997) 781-794
Iwasaki, Kawanishi, and Strauss, respectively, just following the 1994 Los Angeles earthquake (hereafter, the earthquake data), I hope to show how and to what degree a qualitative cross-linguistic analysis of counterfactual reasoning can shed light on the critical role of the speaker's evaluative stance of desirability, DESIRABLE/ UNDESIRABLE in understanding the inherent relationship between counterfactuality and everyday human existence. I will demonstrate that Japanese is particularly useful in making this relationship salient. Until quite recently, the study of counterfactual reasoning looked as if it belonged basically to the domain of logicians and philosophers. Traditionally, the 'if-then' construction has been regarded as the epitome of Man's rational capacity. As a linguistic device for valid reasoning and logical argumentation, the natural language conditional has been typically compared to the mathematical conditional, 'p D q', where the relevant notions are truth-values, i.e., TRUE/FALSE. The major concern of previous researchers of philosophy and logic has been the logical problem of truth conditions, i.e., to specify under what strict conditions a particular instance of counterfactual reasoning is 'true' or 'valid' (cf. Jackson, 1991; Fauconnier, 1985). The counterfactual conditional, in particular, has enjoyed a tremendously prestigious position in that rationalistic tradition. However, through years of research, I have become more and more convinced that the speaker's evaluative stance plays a crucial role in understanding many instances of conditional reasoning in everyday life. The ultimate goal of the present study, therefore, is to solidify my earlier claim that natural language conditionals are an important device for encoding the speaker's evaluative stance of desirability (e.g., Akatsuka, forthcoming a, b; 1991). The outline of the paper will be as follows. In Section 2, I will introduce Fauconnier's (1985) theory of mental spaces as one current view on counterfactuals presented by a linguist as an alternative to the traditional truth-conditional approach. I will systematically demonstrate why and how Fauconnier's theory fails to account for usages of counterfactual conditionals in everyday situations. In Section 3, I will examine instances of counterfactual reasoning by Japanese speakers excerpted from the earthquake data and will contrast these with Fauconnier's made-up examples. What is remarkable here is that these Japanese speakers are jointly and collaboratively producing elements of counterfactual reasoning (i.e., co-construction of counterfactual reasoning) and I will demonstrate that they are 'collaboratively' constructing the line of reasoning, 'P [DESIRABLE], because if not P, then not Q [UNDESIRABLE]'. Section 4 is the conclusion.
2. Fauconnier (1985) I would like to begin this inquiry into counterfactual reasoning by critically examining Fauconnier's (1985) view of counterfactuals. As part of the recent cognitive semantics movement, Fauconnier attempted to provide an alternative to the truth-conditional account of counterfactual conditionals within his theory of mental spaces. It is clear that Fauconnier wished to focus more on the 'ordinary' or everyday usages of counterfactuals than his predecessors from the tradition of formal
N. Akatsuka / Journal of Pragrnatics 28 (1997) 781-794
783
logic. Unfortunately, however, despite his intentions, Fauconnier (1985) is still very much under the strong influence of rationalistic tradition. Consequently, his theory of mental spaces shows no concern whatsoever for the speaker's mental attitude. Moreover, his theory does not seem to be at all interested in the question, 'when a speaker uses a counterfactual conditional, just what is s/he really doing?'. 2.1. 'Space-builders' The best way to understand Fauconnier's view of counterfactuality is to examine his theoretical construct, 'space-builders'. 'Space builders' include various linguistic elements such as conditionals ( i f _ _ ) , verbs like wish, and negatives such as not and prevent, and as such, they are supposed to be capable of setting up a counterfactual mental space which functions in such a way that 'some relation (sic)', which does not hold in the parent space (i.e., the speaker's belief world), may be satisfied in the counterfactual space. In this light, consider the utterances listed in (1), which appear in the original as (F3-F6) (emphasis original).
(1) (F3) If Lucky had won, I would be rich. I would have moved to Tahiti. (F4) I wish Lucky had won. I would be rich. (F5) Fortunately, the fire did not cross the highway. My house would have been destroyed. (F6) Luckily, the fire was prevented from crossing the highway. My house would have been destroyed. Note that none of the above examples are impersonal statements. The speakers of (F3) and (F4) sound unhappy or disappointed because what they wanted to happen did not happen. In contrast, the speakers of (F5) and (F6) sound extremely happy. They are saying in essence, 'Boy! Am I fortunate/lucky! My house wasn't destroyed!'. However, in Fauconnier's theory of mental spaces, the speaker's stance of desirability plays absolutely no role. Fauconnier states simply that in all four of the above utterances, "the first sentence can be understood to set up a counterfactual space (incompatible with the origin), and the second one expresses some relation satisfied in that counterfactual space" (1985:109-110). I turn first to Fauconnier's analysis of (F5) and (F6). Observe that Fauconnier totally ignores the significance of the attitudinal adverbs, 'fortunately' and 'luckily' in his account of counterfactuality in the second sentence. He states that "Clearly, counterfactuality may be lexically imposed, as it is by strong 'negatives' such as not and prevent in (5) and (6)" (1985:111). Interestingly, Fauconnier believes that 'not' and 'prevent' are the strongest lexical space-builders, much stronger than 'if' and 'wish', since the latter do not always build counterfactual space. However, as will be demonstrated shortly, it is not at all the existence of the elements 'not' and 'prevent' that has 'imposed' the counterfactuality. To illustrate this point in preview, notice that the omission of the attitudinal adverbs 'fortunately' and 'luckily' from Fauconnier's examples result in an unnatural occurrence of counterfactuals in the second sentence, as illustrated in (2) below:
784
N. Akatsuka / Journal of Pragmatics 28 (1997) 781-794
(2) (a) [from (F5)] ??The fire did not cross the highway. My house would have been destroyed. (b) [from (F6)] ? ? The fire was prevented from crossing the highway. My house would have been destroyed. Hence, the source of this counterfactuality must be elsewhere, as we shall soon see. 2.2. The Japanese pattern, P-te + yokatta
The critical role of the speaker's desirability stance becomes much clearer when we compare Fauconnier's examples with their Japanese counterparts. The English expression, 'Luckily/Fortunately + P (past tense)' corresponds most closely to the sentence pattern, P-te + yokatta (past tense) in everyday Japanese. What this grammar pattern expresses is nothing but the speaker's attitude, DESIRABLE, towards the past event/state of affairs described by the proposition, P. (3) P-te, yokatta. [DESIRABLE] was lucky '(I, it, etc.) was lucky that P' In P-te, the verb of the subordinate clause P takes the tenseless te-form or gerund, and the predicate of the entire sentence is yokatta, the past tense form of the adjective, ii 'good, lucky, fortunate'. There is no syntactic subject for yokatta. Now, compare (F5) with (JF5) below. An English translation which most closely resembles the syntax of (JF5) has been added for clarification. (4) (JF5) Kaji ga dooro no mukoo ni moeutsura-naku fire SUBJ highway over to cross-buming-not TE, YOKATTA. (Moshi) moeutsutte-itara, uchi ga was fortunate rp-teyokam] IRREALIS cross burning-if my house SUBJ yaketa daroo burnt down would have been 'I was lucky that the fire did not cross the highway. If it had, my house would have been destroyed.' As we note in Fauconnier's examples (F3)-(F6) using English, the second sentence is a mono-clausal counterfactual conditional. However, as pointed out by Fujii (1995), Japanese syntax normally does not tolerate mono-clausal conditionals and requires or prefers the explicit existence of the conditional antecedent, as in the case of (JF5). In fact, the P-te + yokatta pattern is typically followed by the counterfactual conditional, 'if not P, not Q', where 'not Q' is an UNDESIRABLE consequence. It is precisely because of the explicit existence of the counterfactual antecedent 'if not P', that we can begin to understand what the speaker is doing by invoking counterfactual reasoning: S/he is presenting the reason why s/he considers it DESIRABLE that the fire did not cross the highway (= FACT). The COUNTER-TO-FACT situation
N. Akatsuka / Journal of Pragmatics 28 (1997) 781-794
785
(= the fire's crossing the highway) would have led to the tragedy of the speaker's losing his/her home. We can informally represent the speaker's line of reasoning, as in (5): (5) 'P' was DESIRABLE. (Wily? Because) 'NOT P' would have led to UNDESIRABLE consequences. Contrary to Fauconnier's claim, it is not the lexical items 'not' and 'prevent' that build a counterfactual space - these become relevant only in the discussion of counterfactuality when the proposition P in which they are uttered is evaluated by the speaker in terms of desirability. Rather, it is the speaker who has invoked counterfactual reasoning as an integral step in the line of reasoning, as delineated in (5). The superficial nature of Fauconnier's concept of 'space-builders' becomes even more apparent when we examine examples (F3) and (F4) in the light of the above discussions. (F3) and (F4) are repeated below for convenience. (6) (F3) If Lucky had won, I would be rich. I would have moved to Tahiti. (F4) I wish Lucky had won. I would be rich. According to Fauconnier, the space-builders 'if' and 'wish' are responsible for the appearance of counterfactuality in the second sentence, but he does not mention what might be responsible for the appearance of counterfactuality in the first sentence. Also, notice that in his purely syntactic, surface-oriented approach to counterfactuality, he establishes no inherent relationship between (F3) and (F4), and between (F5) and (F6) from the point of view of counterfactual reasoning. Based on the foregoing examples, I would claim that Fauconnier (1985) is more concerned with the surface distribution of counterfactuality and that there is no actual analysis of 'counterfactual reasoning'. Now, compare Fauconnier's (F3) and (F4) with (F3)' and (F4)' below: (7) (F3)' Unfortunately, Lucky did not win. If he had won, I would be rich. If I were rich, I would have moved to Tahiti. (F4)' Unfortunately, Lucky did not win. I wish he had won. If he had won, I would be rich. I suggest that an adequate theory of counterfactuality in cognitive semantics must explain that the workings of counterfactual reasoning in Fauconnier's (F3) and (F4) are exactly the same as those in (F3)' and (F4)'. According to our desirability-based analysis, the counterfactual reasoning involved here can be illustrated informally as in (8): (8) 'P' was UNDESIRABLE..(Why? Because) 'Not P' would have led to DESIRABLE consequences. We have seen that the speaker's mental attitude, in particular the speaker's evaluative stance of desirability, is absent from Fauconnier's theory of mental spaces. Further, because Fauconnier's theory involves a fundamental lack of concern about
786
N. Akatsuka / Journal of Pragmatics 28 (1997) 781-794
the speaker's intentions in the use of counterfactuals, it fails to provide a successful framework for analyzing the relationship between counterfactuals and human emotions. Unwittingly though, Fauconnier's study serves as testimony to the crucial role of these two factors in our inquiry into ordinary usages of everyday counterfactual reasoning. Let us now turn our discussion away from constructed examples and examine authentic Japanese discourse excerpted from the earthquake data.
3. Co-construction in Japanese discourse 3.1. The 1994 LA earthquake data
On January 17, 1994 at approximately 4:30 a.m. an unusually strong earthquake hit the Los Angeles area. That day happened to be a national holiday and not a regular working day. Most people were at home, still fast asleep. Thanks to these fortunate factors, casualties were miraculously small. About five weeks following the earthquake, audio- and video-taped dialogues about individual personal experiences were collected from UCLA students who are native speakers of Japanese, Korean, and American English by Iwasaki (Japanese), Kawanishi (Korean) and Strauss (English). The students were total strangers until they met for the first time at the data collection site. They were paired up, left alone in the room with audio and video recording equipment and asked to talk for 20-25 minutes about their personal experiences during the earthquake. For the purpose of this study, I will restrict the analysis to particular excerpts from the co-construction of counterfactual reasoning by two-pairs of Japanese speakers. 3.2. Co-construction as more than 'collaborative finish'
The term, 'co-construction' refers to any type of collaborative participation among conversational interactants and is often understood to be synonymous with the notion of 'collaborative finish' (Lemer, 1991; Ono and Yoshida, 1996). Simply stated, the first speaker utters an almost complete sentence which the second speaker completes without waiting for the arrival of the normal transition relevance place (Schegloff et al., 1974), in other words, without waiting for his clear turn to speak. As suggested by the title of Lemer's (ibid.) pioneering work, 'On the syntax of sentences-in-progress', 'co-construction' is taken to be a fundamentally local 'syntactic' phenomenon, albeit an interactive, social activity. Lemer discusses many syntactic environments in English where the first speaker can provide a cue for the second speaker to come in at some appropriate place to actually complete the first speaker's sentence. The 'IF p, (THEN) q' construction is one typical example. According to Lemer, the occurrence of the conditional marker 'if' at the beginning of a speaker's utterance can foreshadow the upcoming second component, '(then) q'. Consider his example.
N. Akatsuka / J6,urnal of Pragmatics 28 (1997) 781-794
787
(9) (L6) [US] Rich: if you bring it intuh them Carol: ih don't cost yuh nothing Note, however, that Japanese conditional markers always occur at the end of the conditional antecedent, as is often the case with typical verb-final languages. That is, the Japanese conditional sentence takes the form of 'p + if, q'. In spite of this syntactic constraint of Japanese (and other SOV languages) which does not fit into Lerner's theory of the relationship between morpheme order and co-construction, I will show that the 'co-construction' of counterfactual conditional reasoning frequently does occur in the Japanese earthquake data and that this co-construction is not merely an instance of a collaborative finish. What is going on? Recently, Strauss (1995) and Strauss and Kawanishi (1996) have used the earthquake data for their seminal studies on assessment strategies in Japanese, Korean and American English. They argue that Japanese speakers often participate in co-construction activities to establish common ground. In this section, I would like to demonstrate that the co-construction of counterfactual reasoning must be indeed accounted for in the light of their findings, and as such, is far more than just 'collaborative finish'. Specifically, I hope to show that when speakers are co-constructing counterfactual reasoning, they are in fact collaboratively constructing the entire line of reasoning, 'P [DESIRABLE], because if not P, not Q [UNDESIRABLE]'. 3.2.1. Co-construction of counterfactual reasoning: Case (1) Consider excerpt (10) below. This stretch of talk followed a 15-minute exchange of the two speakers' respective experiences about things such as where they were at the time of earthquake, how frightening it was, and what kind of damages they suffered (relatively small). The structure of (10) is clearly divided into 3 parts: (a) the opening, (b) co-construction of counterfactuals, 'if not p, not q' by the two speakers, where p and q stand for FACT (e.g., p = 'the earthquake occurred at 4:30 a.m.' and q = 'the casualties were very small'), and (c) the closing. Note that the speakers are saying essentially the same thing in the opening and the closing. In the opening, speaker H says 'Asa de yokatta (Fortunately, it was in the morning)' and in the closing, speaker A repeats the same utterance verbatim within a larger stretch of talk, Asa de yokatta. (10) (a) Opening Hi: demo nee asa de yokatta desu yo ne? gozen yoji han de? Al atashi mo omotta-n desu yo. (b) 'if not p, not q' H2: moshi nee kore ga nee imagoro A2: = kore ga nee, yuugata no: H3: ima toka ne A3: rasshuji no go-ji toka ne ha ha ha ha
788
N. Akatsuka / Journal of Pragmatics 28 (1997) 781-794
H4:
(3)
soo soo soo yuugata no rasshuji no goji toka: A4: ha haha haha ha Hs: ano toshokan nanka de mada gakusei nanka ir < iru jikan dattara: As: iru san yoji toka: H6: moo atama kara saa hon ga ochitekite: A6: aa zaa - sugokatta desu nee. HT: haiuei furiiuei nanka mo zenbu nee A7: zenbu Hs: sundanshichattara moo A8: kanbotsushichatte: H9: shisha nanka ... juubai gurai n natta deshoo nee? Closing A9: nn atashi mo omotta-n su asa de y o k a t t n -n daroo na tte Hlo: ann A~o: minna neteru toki datta kara HII: honto hukoo chuu no saiwai desu yo: All: nn H12: =nn nn [English version of (10)] (a) Opening HI: but, see, we were lucky it happened in the morning, at four thirty, right? A~: I thought, about that, too.. (b) 'if not p, not q' H2: If it had happened around now {in the late afternoon}, A2: =(if it) had happened in the evening H3: like now or something A3: at rush hour like five o'clock ha ha H4: right, fight, right at rush hour at like 5:00 in the evening A4: ha haha haha ha Hs: well like at the library, if it happened at a time when students were still there As: like three or four o'clock or H6: well, the books would have fallen on their heads and A6: Oh, like 'crash' - it would've been terrible. H7: the highway, freeway, all of them. AT: All of them Hs: If they had been cut As: they would have collapsed and H9: The death toll would've been 10 times higher, fight?
N. Akatsuka / Journal of Pragmatics 28 (1997) 781-794
789
(c) Closing A9: So, I thought too that we were lucky it h a p p e n e d in the morning. HI0: Yeah. A10: Everybody was sleeping at that time, so Eli: Right, it was the only blessing within this disaster. A11: Yeah. H12: =Yeah, yeah. What is so remarkable is that the abstract structure of these Japanese data is quite similar to that of Fauconnier's examples in (1), (particularly (F5) and (F6)), even though the latter are monologic utterances and (10) is a dialogue. One important difference is that in the Japanese utterances, there is no occurrence whatsoever of the corresponding lexical items 'nol' or 'prevent'. Contrary to Fauconnier's claim, what 'imposes' a long stretch of co-construction of counterfactuals by the two speakers is the initial exchange of Asa d e y o k a t t a desu yo he? (Weren't we lucky it happened in the morning!) and atashi mo omotta n desu yo (I thought so, too!). Now, if we consider co-construction to be an essentially local syntactic phenomenon where the second speaker's contribution is simply to syntactically finish up the almost completed sentence, this excerpt is not very interesting theoretically, because there is not a single occurrence of the clear-cut syntactic 'division of labor' such that the first speaker would say, 'if not P' and then the second speaker would follow with '(then) not Q'. However, if we hypothesize that the two speakers here are actually 'collaboratively' constructing the abstract line of reasoning, 'P [DESIRABLE], because if not P, not Q [UNDESIRABLE]', then this stretch of conversation becomes suddenly an exciting resource for the study of counterfactual reasoning as well as the study of co-construction. The key to shedding light on what is going on here is the insights provided by the previous articles, Strauss (1995) and Strauss and Kawanishi (1996). First, in both studies, it is suggested that Japanese speakers often express their agreement by 'repeating' or 'echoing' what the previous speaker has just said, as illustrated in the following excerpt (Strauss, 199.5: 187-188). (11) Hide: sonde heta suruto ame-ame nanka hunnai to moo ne m i z b u s o k u ni narushi nee? Ai: m o o mizubusoku n narushi koko datte mo
790
N. Akatsuka / Journal of Pragmatics 28 (1997) 781-794
Hide: originally it's a desert, right? Ai: It's a desert and it's like people force the desert into becoming a livable place Hide: right, right, right. Moreover, it is pointed out that Japanese speakers often participate in co-construction activities to establish c o m m o n ground, contrary to the claim made by others, notably Ono and Yoshida (1996) that speakers do not 'finish each other's sentences'. In fact, example (10) used here is excerpted from one of Strauss' examples illustrating collaborative finishes in Japanese. It is clear that in her case the term 'collaborative finish' is not used in the usual way of 'syntactically completing a sentence started by the other speaker' but rather it is used to mean "to co-construct each other's story to the degree that it is no longer possible to determine who is the primary speaker and who is the interlocutor" (Strauss, 1995: 187). When we re-examine (10), we now realize that the two speakers are co-constructing 'If not P' by the extensive use of an 'echoing' strategy. Also, notice that they are clearly enjoying their 'collaborative work' as illustrated by laughter tokens ha ha ha ha ha and an enthusiastic endorsement of the other party's comments soo soo soo. Observe: (12) (b) if not p, not q H2: moshi nee kore ga nee imagoro A2: =kore ga nee, yuugata no: H3: ima toka ne A3: rasshuji no goji toka ne ha ha ha ha H4: soo soo soo yuugata no rasshuji no goji toka: A4: ha haha haha ha [English version] H2: If it had happened around now (in the late afternoon) A2: (If it) had happened in the evening H3: like now or something A3: at rush hour, like at five o'clock ha ha ha ha H4: right, right, right, at rush hour, at five o'clock in the evening
A4:
ha haha haha ha
The two speakers are not only 'jointly' constructing the 'if not p' component - they are also collaboratively construction the '(THEN) not Q' component as well. Both H6-A6 and A8-H9 are cases in point. Consider the latter, represented below as (13): (13) H 8: sundanshichattara moo ---) As: kanbotsushichatte: --+ H9: shisha nanka ... juubai gurai n nattara deshoo nee?
N. Akatsuka / Journal of Pragmatics 28 (1997) 781-794
791
[English version] Hs: if they had been cut oh A8: they would have collapsed and H9: The death toll would've been 10 times more, right? It is now time to re-pose the question: What are the two speakers doing here by invoking counterfactual reasoning? It is clear that they are not co-reasoning to create new ideas; there is absolutely no exchange of new information going on here. In these extended stretches of talk the speakers are 'collaboratively' repeating what they have been repeating to themselves and to others and what they have heard repeated on so many occasions since the earthquake first hit five weeks earlier. And by repeating what they know and what basically everybody else already knows, they are, in a sense, congratulating themselves by conveying the message 'We are really lucky. It could've been so much worse'. These two individuals were total strangers until they met that day at the small office of a UCLA instructor to talk about their own earthquake experiences. They were strangers, but by exchanging personal anecdotes of their common earthquake experiences for 15 minutes, they have reaffirmed that they had recently shared an unusually undesirable experience, and were able to come out of it relatively unharmed. Temporarily, therefore, they have established a kind of 'in-group' relationship. The instances of co-construction occurring here are not simple acts of providing 'collaborative' finishes, nor are they just part of any ordinary and interactive social activity; this activity of co-construction is an act of establishing and building up an emotional solidarity based on shared experiences. 3.2.2. Co-construction of counterfactual reasoning: Case 2 At first glance, the following excerpt between K and S, a second pair of subjects from the Japanese earthquake data, would appear to be a 'garden variety' collaborative finish where the second speaker completes the first speaker's unfinished sentence.
(14) K: ... maa demo hontoni ... hiruma de nakute (YOKATTA) ... a hiruma dattara sootoo na higai nat ... tteta-njanai kanaa to omo S:
nnn. d e s u y o nee.
[English version] K: ... Well, but really . . . . it wasn't in the daytime and so (WE WERE LUCKY) ... if it had been during the day, there would have been considerable damage, I thiA: (I) t h i n k so too.
In (14), speaker K has just started to utter the main verb omou 'think', when speaker S comes in and completes the sentence. Note that the further addition of the sentence-final particles, yo + ne makes it clear that speaker S is not just helping
792
N. Akatsuka / Journal of Pragmatics 28 (1997) 781-794
speaker K finish the sentence, but that he is eager to show that he is strongly agreeing with K's point of view. S seems to be totally agreeing with K's entire line of reasoning, 'P [DESIRABLE], because if not P, not Q [UNDESIRABLE]'. This point becomes even clearer when we examine how their conversation proceeds. In (15) below, from the same interaction as above, S, after digressing onto another topic for about one minute, initiates (15), the structure and content of which is strikingly similar to that of (10). Note, for example, that exactly the same expression, hukoochuu no saiwai desu yo (it was a blessing within this disaster), appears to be a concluding remark in both cases (10) and (15). The absence of an explicit opening in (15) is quite understandable since the two speakers have just confirmed each other's stances hiruma de nakute yokatta (we were lucky it didn't happen in the daytime) in (14). There are important differences, however, the most striking of which is the following: There is no syntactic occurrence of 'not Q', and yet, notice that the second speaker K enthusiastically agrees with S. Observe: (15) (b) 'if not p, (not q)' S: demo are na no ga nee/hiruma dattara honto shikamo are ga shukujitsu janakute K: ee S: nee sono hito ga gakko ittari K: heejitsu de nee. ee. S: kaisha ittari suru hi de K: (.) soo desu yoo ne S: rasshuawaa dattara chotto nanjikan ka zuretetara nee -+ K: hontoni (c) Closing are wa yoji han goro deshita kke? honto ni hito ga ichiban inai toki ni (.) de: honto ni maa yokatta to iu. S: hukoochuu no saiwai desu yo nee.
--.)
---)
[English version] (b) 'if not p (not q)' S: but see that earthquake, if it had happened in the daytime, really And if it had not been on a holiday K: Right S: and see, if people had gone to school K: just a regular weekday, yeah S: just a regular working day K (.) That's really true. S: If it had been during rush hour, or if it happened a few hours later, K: Exactly. (c) Closing was it around 4:30? it was the time very few people were up and around. We were really lucky S: it was the only blessing within this disaster.
N. Akatsuka / Jc,urnal of Pragmatics 28 (1997) 781-794
793
While there is no syntactic 'coUaboration' of any sort between the two speakers, I believe that (15) is also an instance of co-construction of counterfactual reasoning. The two speakers are 'collaboratively' (re)constructing the line of reasoning that 'P [DESIRABLE], because if not P, then not Q [UNDESIRABLE]'. One might wonder how I can support my claim when there is no 'not Q' element in (15). I would argue that ordinary people such as K and S in this interaction do not even realize that 'if not P' is an incomplete statement. Put differently, they intuitively know that 'IF not P' means underlyingly 'IF not P, TERRIBLE THINGS WOULD HAVE HAPPENED', when P represents past events/states of affairs which they consider DESIRABLE. All utterances of 'if not p' in (15) were said with full emotion, which makes them almost like exclamations. In the rationalistic tradition of the study of counterfactuals, instances of 'If not P' without '(then) not Q' have been completely ignored because there is no possible truth-value for such an utterance;, thus rendering them as non-conditionals. However, authentic discourse data such as those presented here, and particulary in (15) clearly demonstrate that for ordinary people these utterances are indeed full-fledged conditionals on a par with 'If not P, not Q'. In many or perhaps even most languages of the world, the 'if not P?' forms have been grammaticalized as exclamations expressing the speaker's deep-felt emotions such as REGRET, LAMENT, WISH, etc., e.g., 'If only I hadn't said that to him?' Ultimately, this research, which has at its foundation the notion of human desirability in natural logic, will be able to provide a principled explanation for this highly deserving candidate for absolute linguistic universals, which according to Comrie (1981) are extremely rare.
4. Conclusion While this is primarily a study of counterfactual reasoning, the study of co-construction has provided essential insight into the nature of counterfactual reasoning in ordinary life. For example, since the speakers in the Japanese earthquake data spontaneously engage in co-construction in the line of reasoning, 'P [DESIRABLE], because if not P, then not Q [UNDESIRABLE]', we can hypothesize that this line of reasoning must be in some intuitive sense quite natural to ordinary Japanese people, and probably to every human being. The striking similarity of the abstract structure of Fauconnier's made-up examples in English and the Japanese authentic discourse data provides strong support for such a universal hypothesis. I have noticed that speakers of Korean and American English also invoke counterfactual reasoning when narrating personal anecdotes. Interstingly, however, I have not found such an extensive co-construction as our Case (1) and Case (2) in neither the Korean nor the American English data. The study of different styles of co-construction activities across languages will be a fruitful area of research for the future. References Akatsuka, Noriko, forthcoming a. Towards a theory of desirability in conditional reasoning. In: Akio Kamio and John Whitman, eds., Function and structure. Philadelphia, PA: Benjamins.
794
N. Akatsuka / Journal of Pragmatics 28 (1997) 781-794
Akatsuka, Noriko, forthcoming b. Negative conditionality, subjectification and conditional reasoning. In: Ren6 Dirven and Angeliki Anthansiadou, eds., On conditionals again. Amsterdam: Benjamins. Akatsuka, Noriko, ed., 1994. Japanese/Korean Linguistics, Vol. 4. Stanford, CA: CSLI. Akatsuka, Noriko, Shoichi Iwasaki and Susan Strauss, eds., 1996. Japanese/Korean Linguistics, Vol. 5. Stanford, CA: CSLI. Akatsuka, Noriko, Hajime Hoji, Shoichi Iwasaki, eds., to appear. Japanese/Korean Linguistics, Vol. 7. Stanford, CA: CSLI. Comrie, Bernard, 1991. Language universals and linguistic typology. Philadelphia, PA: Benjamins. Dirven, Ren6 and Angeliki Anthansiadou, eds., forthcoming. On conditionals again. Amsterdam: Benjamins. Fauconnier, Gilles, 1985. Mental spaces. Cambridge, MA: MIT Press. Fujii, Seiko Y., 1995. Mental space builders: Observations from Japanese and English conditionals. In: Masayoshi Shibatani and Sandra Thompson, eds., Essays in semantics and pragmatics, 73-90. Philadelphia, PA: Benjamins. Jackson, Frank, ed., 1991. Conditionals. Oxford readings in philosophy. Oxford: Oxford University Press. Lerner, Gene H., 1991. On the syntax of sentences-in-progress. Language in Society 20: 441-458. Cambridge: Cambridge University Press. Ono, Tsuyoshi and Eri Yoshida, 1996. A study of co-construction: We don't finish each other's sentences. In: Noriko Akatsuka, Shoichi Iwasaki and Susan Strauss, eds., Japanese/Korean Linguistics, Vol. 5:115-130. Stanford, CA:CSLI. Sacks, Harvey, Emanuel Schegloff and Gail Jefferson, 1974. A simplest systematics for the organization of turn-taking for conversation. Language 50: 696--735. Strauss, Susan, 1995. Assessment as a window to sociolinguistic research: The case of Japanese, Korean, and (American) English. In: Misato Tokunaga, ed., Gengo henyoo ni kansuru taikeiteki kenkyuu oyobi sono nihongokyooiku e no ooyoo, 177-191. Mombushoo Research Report 0044510992. Strauss, Susan and Yumiko Kawanishi, 1996. Assessment strategies in Japanese, Korean, and American English. In: Noriko Akatsuka, Shoichi Iwasaki and Susan Strauss, eds., Japanese/Korean Linguistics, Vol. 5: 149-166. Stanford, CA: CSLI. Traugott, Elizabeth C., Alice ter Meulen, Judy S. Reiley and Charles A. Ferguson, eds., 1985. On conditionals. Cambridge: Cambridge University Press. Texts: 1994 Los Angeles Earthquake data: English data: Susan Strauss, 1994 Japanese data: Shoichi lwasaki, 1994 Korean data: Yumiko Kawanishi, 1994