1038 8. X. Syntactic Typology References Birjulin, Leonid. 1994. Semantika i pragmatika russkogo imperativa [Semantics and pragmatics of the Russian imperative] (Slavica Helsingiensia, 13.) Helsinki: University of Helsinki. Bolinger, Dwight. 1974. “Do-Imperatives”, Journal of English Linguistics 8: 1⫺5. Davies, Eirlys. 1986. The English imperative. London: Croom Helm. Hamblin, Charles L. 1987. Imperatives. New York: Blackwell. Šteling, Donat A. 1982. “O grammatičeskom statuse povelitel’nogo naklonenija” [On the grammatical status of the imperative mood]. Izv. AN SSSR. Serija literatury i jazyka, 41.3. Wierzbicka, Anna. 1972. Semantic primitives. Frankfurt am Main: Athenäum. Xrakovskij, Viktor S. (ed.) 1992. Tipologija imperativnyx konstrukcij [Typology of imperative constructions] St. Petersburg: Nauka. Xrakovskij, Viktor S., Volodin, Aleksandr P. 1986. Semantika i tipologija imperativa. Russkij imperativ. [Semantics and typology of the imperative. The Russian imperative] Leningrad: Nauka. Viktor S. Xrakovskij, Sankt Petersburg (Russia) Hare, Richard M. 1949. “Imperative sentence”. Mind, 58: 21⫺39. 79. Exclamative constructions 1. 2. 5. 6. Introduction The conceptual basis of the exclamative category The exclamative sentence type The cross-linguistic expression of exclamative meaning Conclusion References 1. Introduction 3. 4. Exclamative constructions form a variegated class, both within and across languages. This article presents a typological survey of the coding of exclamations. This survey will proceed in the following fashion. First, we will explore the conceptual basis of the category exclamation, and in particular the category of degree exclamations (section 2). Second, we will examine the formal, semantic and pragmatic constraints which jointly define the exclamative sentence type ⫺ a type whose characterization is based, as in Grimshaw (1979), on the degree class (section 3). Third, we will look at cross-linguistic manifestations of this type, with particular attention to those recurrent formal properties which reflect components of the exclamative sentence type and which suggest general tendencies in the grammaticalization of exclamative constructions (section 4). In a concluding section (section 5), we will consider the question of whether, on the basis of the data considered, one can establish the existence of a formfunction fit in the area of exclamations. 2. The conceptual basis of the exclamative category Any attempt to identify a sentence type in a given language, or to compare instances of a given sentence type across languages, relies on an understanding of the function pole of the form-function pairing. Such an understanding has been particularly elusive in the case of exclamations, since the terms exclamation and exclamative have often been taken to refer to emphatic or expressive utterances in general, as in the following definition from a study on exclamative intonation: “L’exclamation est généralement définie comme la manifestation linguistique d’un état émotionnel de l’énonciateur […]” (Morel 1995: 63). Further, the label exclamation has often been applied to related expressive phenomena, like interjections and news-reporting declaratives. For example, Makkai (1985) purports to examine the diachronic sources of exclamations, but focuses largely on interjections. Speech-act theory (Austin 1962, Searle 1979, Geis 1995) has not helped to refine our understanding of the exclamative type, since expressive speech acts, with the exception of formalized locutions like apologies, are not readily analyzed with regard to preparatory, essential and sincerity conditions. It is perhaps because researchers lack a solid conceptual foundation for the exclamative type that this type has not figured prominently in typological syntactic research (exceptions being Elliott 1974 and Sadock & 1039 79. Exclamative constructions Zwicky 1985). Therefore, as a prelude to our typological exploration of exclamative constructions, we will now examine semantic and pragmatic criteria which define exclamations and distinguish them from interjections and declaratives. 2.1. The coding of surprise Exclamations, like The nerve of some people! or the French Comme il fait beau! (‘How lovely it is!’), are grammatical forms which express the speaker’s affective response to a situation: exclamations convey surprise. Surprise may be accompanied by positive or negative affect. In the model assumed here, surprise is not merely a response (startled or otherwise) to a situation which the speaker had failed to predict. For example, I might not have predicted a hallway encounter with a colleague, but I would not necessarily find that encounter surprising (even if my colleague had startled me). Instead, surprise entails a judgement by the speaker that a given situation is noncanonical. A noncanonical situation is one whose absence a speaker would have predicted, based on a prior assumption or set of assumptions, e. g., a stereotype, a set of behavioral norms, or a model of the physical world (Michaelis 1994b). In conveying surprise, exclamations resemble a subset of interjections (Ehlich 1986, Makkai 1985). Interjections resemble exclamations in that they express the speaker’s appraisal of a situation. While some interjections (like Yay! or Damn!) express the speaker’s evaluation (positive or negative) of the situation, some, like Hey! or Oh my God!, have a function like that of exclamatives ⫺ expressing what Fillmore & Kay & O’Connor (1988) call a noncanonicity judgement. 2.2. The expression of speaker viewpoint The individual whose surprise is expressed by an interjection or exclamation is ⫺ by default ⫺ the speaker. A judgement expressible as an exclamation, like that expressible as an interjection, can, however, be attributed to someone other than the speaker, as in (1) and the presumably veridical quote in (2): (1) She couldn’t believe how few people came to help her. (2) He’s like, ‘Hey! You’re not supposed to be here!’ Sentence (1) is vague as to whether the speaker shares the judgement attributed to the subject-referent (that the paucity of helpers was surprising). Sentences like (3⫺4), however, appear to invite the hearer to share the speaker’s judgement: (3) You won’t believe who spoke up. (4) Du würdest nicht glauben, wer sich zu Wort gemeldet hat. ‘You wouldn’t believe who spoke up.’ Exclamatives like (3⫺4), in which the hearer is the judge, have a marked status: such exclamations generally require a futurate or subjunctive main verb. This modal marking imparts a hypothetical flavor; sentences like (3⫺4) presuppose that the hearer is not currently in a position to make the relevant judgement. The modal marking appears to be crucial: utterances like (5) are anomalous, whether or not they are construed as exclamations: (5) ??You don’t believe who spoke up. The irrealis flavor of (3⫺4) suggests that the speaker is by default the source of the noncanonicity judgement, as does the interpretation of (6), in which the source of the relevant viewpoint is not overtly expressed: (6) It’s incredible how little you can spend there. In (6) the relevant judgement (concerning the unusually low prices) is attributed to the speaker (at least), and perhaps also to people in general. The expression of speaker viewpoint appears intrinsic to the exclamative speech act, and utterances like (1), despite having formal hallmarks of exclamatives (e. g., a wh-complement following an epistemic predicator), are not clear examples of exclamations, just as (2), despite containing a form otherwise identifiable as an interjection, is not a prototypical example of an interjection. One can presume instead that examples like (1⫺2) involve perspectival shift (Fillmore 1982). Just as the demonstrative adjective this is not clearly proximal when the viewpoint of someone other than the speaker is invoked (as in, say, a narrative), so an exclamative form like (1) does not clearly perform an exclamative function when the viewpoint expressed is not the speaker’s. 2.3. Propositional content We have observed that exclamations and interjections share two semantico-pragmatic properties: the function of expressing a non- 1040 X. Syntactic Typology canonicity judgement and the indexical function of expressing speaker perspective. The major semantic feature which distinguishes exclamations from interjections is also the major semantic property that exclamations share with declaratives: recoverable propositional content. Both exclamations and declaratives linguistically encode a proposition which the speaker assumes to be true. Interjections lack this property. For example, the interjection Hey! does not express a proposition. While one might analyze this interjection as conventionally expressing a meaning that can be represented by a proposition of the form ‘I am surprised at some aspect of the present situation’, such a proposition is not encoded, elliptically or otherwise, by the linguistic form in question. By contrast, the exclamation in (7): (7) It’s so hot! denotes the proposition ‘It’s hot to a particular degree’. The claim that exclamations lexically encode a proposition requires some justification in light of examples like (8): (8) In The Mask, Carrey plays Stanley Ipkiss, a bank clerk whose timidity is quickly demonstrated in a series of opening sketches. Poor Ipkiss! The indignities that the world heaps on him! ⫺ Michael Covino, The East Bay Express 8/15/94 (⫽ Michaelis & Lambrecht 1996a (32 d)) Although the exclamation in (8) consists of a NP alone, the interpretation of this NP is identical to that of a clausal exclamation like The world heaps so many indignities on him. Grimshaw (1979) refers to exclamations like that in (8) as hidden exclamatives. The proposition conveyed by the hidden exclamative in (8) (‘There is some number of inidignities’) is distinct from propositions we might use to represent the speech-act force of the utterance, which in (8) is presumably the expression of surprise at the high number, etc. of indignities suffered by Ipkiss. Thus, exclamatives and declaratives, unlike, say, questions and imperatives, express propositions. What properties distinguish exclamations from declaratives? Sadock & Zwicky (1985: 162) describe the difference between the two sentence types as follows: Exclamations are intended to be expressive whereas declaratives are intended to be informative. Both represent a proposition as being true, but in an exclamation, the speaker emphasizes his strong emotional reaction to what he takes to be a fact, whereas in a declarative, the speaker emphasizes his intellectual appraisal that the proposition is true. Sadock & Zwicky frame the distinction between the two sentence types as one of emphasis: the declarative emphasizes the truth of the proposition, while the exclamation emphasizes the speaker’s emotional reaction to the proposition (qua situation). The different emphases of the two speech-act types can be described in terms of the well known semantico-pragmatic property of presupposition. Exclamations, unlike declaratives, presuppose that the proposition expressed is mutually known by speaker and hearer. The presupposed proposition is one which involves a scalar degree. The degree itself is not mutually presupposed; the speaker purports to know it, but assumes that the hearer does not, since the speaker’s purpose in exclaiming is to inform the hearer that the degree in question is extreme. Thus, the propositions which are presupposed in exclamative utterances can be represented as open propositions like ‘It is hot to x degree’. The presupposed status of this open proposition is reflected in use conditions. A speaker could use (7) when the general ambient temperature is mutually known to be warm. A speaker would not be inclined to use (7) to report on the weather if neither the speaker nor the hearer know whether it is cold, hot or temperate outside. On the view that exclamations presuppose the propositions which they express, newsreporting utterances (like They dismissed the Paula Jones case!) do not qualify as exclamations. Although this sentence does convey the affective stance associated with exclamations (surprise), it does not instantiate any exclamative construction. It also fails to qualify as an exclamation with respect to the scalarity criterion. However, scalarity, while a necessary condition for exclamative status, is not a sufficient one. Scalarity must be coupled with presupposition. Thus, (9) is not an exclamation on the view taken here: (9) It’s very hot! Although (9) contains a degree adverb, very, this degree adverb differs from anaphoric degree adverbs like so. As Zwicky (1995) observes, the syntactic behavior of degree adverbs of the very-class is systematically distinct from that of degree adverbs in the so-class. As shown in (10), degree adverbs of the so- 1041 79. Exclamative constructions class have cataphoric reference in, e. g., the inverted resultative construction (10a), while degree adverbs of the very-class do not (10b): (10) (a) I almost fainted, the sun was so hot. (b) *I almost fainted, the sun was very hot. With respect to exclamations in particular, the degree adverb very does not collocate with interjections (like God!) and matrix predicators (like I can’t believe) which express the speaker’s affective stance. This is shown in (11), where anomalous exclamations containing very are contrasted with well formed exclamations containing so. (11) (a) (a1) (b) (b1) ??God, it’s very hot! God, it’s so hot! ??I can’t believe it’s very hot! I can’t believe it’s so hot! The distinct behavior of two classes of degree adverbs finds a parallel in Italian. The anaphoric cosı́ is appropriate in the exclamative context (12a); the nonanaphoric molto is not (12b): (12) (a) Non ci posso not sia credere che it can.1sg believe.inf that cosı́ imbecille. is.sbj.3sg so stupid ‘I can’t believe he’s so stupid!’ che (b) *Non ci posso credere not sia it can.1sg believe.inf that molto imbecille. is-sbj.3sg very stupid ??‘I can’t believe he’s very stupid!’ The fact that exclamatives in English and Italian (as well as other languages to be discussed in section 4) use anaphoric degree adverbs like so and cosı́ makes sense on the assumption that the scalar proposition expressed in the exclamation is presupposed. The use of an anaphoric adverb like so relies upon the hearer’s ability to recover the relevant scale from the context. 2.4. Speech act function An exclamation counts as an assertion that the degree in question is higher than the speaker would generally expect. The speaker’s affective stance toward the propositional content can be overtly expressed by a negated epistemic predicator like I can’t believe or by an interjection denoting the speaker’s surprised affect. However, the speaker’s affective stance is not necessarily encoded at the lexico-grammatical level. For example, in (8), an isolated-NP exclamative, the speaker’s affec- tive stance is not lexically expressed. In such cases we can say that the speaker’s affective stance can be inferred by the hearer from the semiotic value conventionally attached to the form employed. 3. The exclamative sentence type In this section, we will pull together the semantic and pragmatic properties which were attributed to exclamations in section 2, while introducing an additional property, which is closely related to the presuppositional property discussed in section 2.3: referent identifiability. We will view these properties as defining a sentence type ⫺ a conventional pairing of form and function. The formal expression of this sentence type is not specified, since, as we have seen, there is a many-to-one mapping of form to function. Instead, we will posit only a constraint on realization: all semantico-pragmatic components of the exclamative sentence type receive formal expression. Certain of these components can be realized through metonymic construal (a construal associated with hidden exclamatives) or through a type of pragmatic construal similar to that found in instances of null complementation (Fillmore 1986, Kay & Fillmore 1998). The semantico-pragmatic features shared by exclamatives are summarized in (13): (13) (a) Presupposed open proposition (with a degree as the variable); (b) Expression of commitment to a particular scalar extent; (c) Expression of affective stance toward the scalar extent; (d) Person deixis (judge is the speaker by default); (e) Identifiability of the referent of whom the scalar property is predicated. As stated in (13e), the entity of whom the scalar property is predicated must be identifiable. An identifiable referent is one for which a shared representation exists in the minds of speaker and hearer at speech time (Lambrecht 1994). Identifiable referents surface as definite NPs or, if activated in discourse, pronominal NPs. Notice, for example, the anomaly of the sentences in (14): (14) (a) *What a nice cake no one ate! (b) ?I can’t believe how much a guy spent! (c) ??Someone is so messy. 1042 X. Syntactic Typology The identifiability constraint exemplified in (14) can be motivated by reference to the requirement of pragmatic presupposition. If a proposition is presupposed, then its arguments are necessarily identifiable to both speaker and hearer. (21) Qu’est-ce qui est devenu de notre ville! ‘What’s become for our city!’ (22) C’est incroyable qu’est-ce qui est devenu de notre ville! ‘It’s incredible what has become of our city.’ 4. As Grimshaw points out, the matrix verbs and adjectives which appear in sentences like (22), which Milner refers to as indirect exclamatives, are necessarily factive, i. e., they presuppose the truth of their complements. These verbs and adjectives also presuppose a norm with which the outcome or situation at issue is implicitly compared. The adjectives in this class may appear in right-dislocation structures like (22), in which there is a referential pronominal subject with which the postverbal clause corefers (J Art. 80), or in extraposition structures, which lack a referential subject. Exclamative constructions involving extraposition are exemplified in (23⫺26) for English, Italian, Croatian and Palestinian Arabic, respectively: The cross-linguistic expression of exclamative meaning Exclamative constructions are characterized by the following formal features: co-occurrence with interjections, complementation structures involving factive epistemic matrix verbs, topic constructions, anaphoric degree adverbs, question words, NP complements, ellipsis with NPs, and inversion. 4.1. Co-occurrence with interjections The close relationship between interjections and exclamation was brought out in section 2. Those interjections which convey surprise typically co-occur with exclamative constructions. Some of these interjections are invocations; others, like wow in English and aman in Turkish, have no recognizable source. Examples are given for English, German, French, Italian, Turkish and Mandarin: (15) Jesus, what a mess! (16) Mein Gott, ist es heiss! ‘My God, is it hot!’ (17) (18) (19) (23) It’s amazing how much noise they make. (24) È pazzesco quanto rumore fanno. ‘It’s amazing how much noise they make.’ (25) Croatian Za ne-vjerovati to Qu’est-ce qu’il est con, sainte vierge! ⫺ Reiser, Les Oreilles Rouges ‘Holy virgin, what a fool he is!’ Mamma, quante ore ho speso in vano! ‘Mamma, how many hours I have spent in vain!’ Turkish Aman, bu ne je potrošila. ‘It’s unbelievable how much she spent.’ (26) Palestinian Arabic Mish maø? uul addaysh not reasonable how.much daføat. paid.3sg.f sicak! ‘Wow, it’s so hot!’ Mandarin Nàme guı̀ koliko is.3sg spend.PaP.sg.f ‘It’s amazing how much she paid.’ interj this how heat! (20) je neg-believe.inf is.3sg how.much ya! that.much expensive interj ‘Wow, so expensive!’ 4.2. Subordination to factive epistemic verbs Both Grimshaw (1979) and Milner (1978) differentiate between main-clause exclamatives, like (21), and constructions containing exclamative complements, like (22): In such constructions, the matrix adjective denotes the property of causing disbelief, for the speaker and for people in general. The generic interpretation is possible because the identity of the judge is not overtly specified. Adjectives may also appear in constructions containing a referential subject denoting the source of the noncanonicity judgement. An example is given for English in (27) and for Setswana in (28): (27) I’m amazed at how much time it took. 1043 79. Exclamative constructions (28) Setswana Ke makatswa ke gore o I amazed by that she dirisitse bokae used how.much ‘I’m amazed at how much she spent.’ Exclamatives with cognizer subjects also commonly contain matrix predicators headed by verbs. The lexical verb is typically a negated form of the verb which means believe. This verb may also be accompanied by a modal element denoting ability, in which case it is the expression of ability which is negated. Examples are given in (29⫺34) for Italian, Turkish, Malay, German, Setswana and Mandarin: (29) Non ci posso credere Not it can.1sg believe.inf che hai speso cosı́ tanto. that has.3sg spent that that.much ‘I can’t believe that she spent that much.’ (30) Turkish Nereye kadar yüzmüşşün ki where extent swam.2sg excl inanmtyorum believe neg.pres.1sg ‘I don’t believe how far you swam!’ (31) Malay Saya tak percaya siapa yang I not believe who rm bercakap. spoke ‘I don’t believe who spoke up!’ (32) Ich kann nicht glauben, wer sich zu Wort gemeldet hat. ‘I can’t believe who spoke up.’ (33) Setswana Ga ke dumele se re se boneng neg I believe rp we om found. ‘I don’t believe what we found!’ (34) Mandarin (Wŏ) jiănzhı́ bù găn xiāng xı̀n (I) simply not dare believe tā doū nàme dà le! 3sg even that.much big perf ‘I simply can’t believe that he’s so big now!’ The indirect exclamative has a strong semantico-pragmatic motivation. Since the assertion of surprise is an essential condition upon the exclamative speech act, and since no lan- guage lacks for psychological predicates denoting disbelief or the property of inducing disbelief, it is natural that languages should use such predicates in complementation structures denoting the speaker’s affective stance toward a scalar proposition. The subordinating predicator merely expresses the otherwise implicit affective stance of the speaker. 4.3. Topic constructions Lambrecht (1994) and Lambrecht & Michaelis (1998) distinguish between two kinds of pragmatic presupposition, which correspond to different kinds of assumptions a speaker may have concerning the addressee’s state of mind at the time of an utterance: knowledge and topicality presuppositions. Knowledge presuppositions concern the assumed knowledge state of a hearer at the time of utterance; they are what linguists typically have in mind when they use the term (pragmatic) presupposition. They are manifested in the complements of factive verbs, in sentential subjects, in various constructions involving open propositions, in definite descriptions, etc. According to our analysis of the exclamative sentence type, the propositional content of exclamations is knowledge presupposed. Topicality presuppositions concern the assumed statuses of referents as topics of current interest in the discourse. In accordance with Lambrecht (1994), we can define a topic as a referent (an entity or proposition) which the speaker assumes to be a relatively predictable argument of predications in the conversation. Topic constructions, like left dislocation and right dislocation, differ according to whether the referent in question is an established topic. Lambrecht observes that the referents of right-dislocated topic expressions, which he refers to as antitopics, tend to be more established as topics than those of topicalized and left-dislocated topic expressions, which, as observed by Prince (1981), are often contrastive (J Art. 80). An important prosodic characteristic of antitopics is that they are pronounced with a low pitch accent characteristic of established topics (Pierrehumbert & Hirschberg 1990). Examples of right dislocation and left dislocation are given in (35⫺36). Sentence accents (high pitch accents) are marked by small caps: (35) (a) She’s pretty sharp, my mom. (b) That’s certainly a shame, that he’s not willing to discuss it. 1044 X. Syntactic Typology (36) (a) The first one, I’m not so sure about it. (b) That there’s liability, this seems obvious. The (a) examples involve topical entities, while the (b) examples involve topical propositions. Topical propositions, and their syntactic encoding, will be of interest to us here. We have said that the propositional content of exclamations is knowledge presupposed. That is, e. g., an open proposition of the form ‘It’s hot to some degree’ is taken for granted by a speaker who employs the exclamative form It’s so hot! A knowledge presupposed proposition may be either topical or nontopical. Some exclamative constructions treat the scalar propositions which they presuppose as topical. Such exclamations often take the form of right-dislocation structures. Examples are given for French and Italian in (37⫺38); the resumptive pronouns are in boldface: (37) (38) C’est incroyable comment elle nous traite. ‘It’s incredible how she treats us.’ Non ci posso credere che Not it can.1sg believe.inf that hai speso cosı́ tanto. has.3sg spent that that.much ‘I can’t believe that she spent that much.’ Related to (37⫺38) is an English exclamative construction which, although lacking the resumptive pronoun characteristic of right dislocation, features a rightward declarative clause pronounced with the low pitch accent characteristic of antitopics. This construction, exemplified in (39), is referred to by Michaelis & Lambrecht (1996a) as the antitopic exclamative: (39) (a) God it’s hot out there. (b) My goodness you’re late. In this construction, a clause-initial interjection which denotes the epistemic stance receives the sole sentence accent. The clause following the interjection does not contain a degree adverb, but this clause must express a scalar proposition, as shown by the ill formedness of (40): (40) *God that’s an even number. Topic expressions used to express exclamative meaning may also take the form of unlinked topics. Unlinked topics, described by Aissen (1992) for Tzotzil and Lambrecht (1994) for English and French, are leftward topic expressions which lack a syntactic relationship to a clause that predicates something relative to that topic. An example of an unlinked topic is given in (41): (41) Most cities, you can’t walk alone at night. There is an exclamative construction of Turkish which appears to instantiate an unlinked topic construction. An example is given in (42): (42) Turkish Nereye kadar-yüzmüşşün ki where extent swam.2sg excl gözlerime inanmtyorum eyes.my believe. neg.pres.1sg ‘How far you swam! I don’t believe my eyes.’ While the translation of (42) uses two separate clauses, one to express the topical proposition and the other to express the speaker’s epistemic stance, the Turkish construction does not reflect this division. Instead, one can analyze the scalar proposition as an unlinked topic, and the following clause as providing additional information about this topic (i. e., expressing the speaker’s attitude toward this proposition). Exclamative constructions which invoke right-dislocated and unlinked topics are strongly motivated in terms of the exclamative sentence type, which involves knowledge presupposition of a scalar proposition. Since knowledge presupposed propositions, like identifiable entities, are often topical, it makes sense that some exclamative constructions in a language should additionally express the topic status of the proposition which they presuppose. 4.4. Anaphoric degree adverbs We will use the general label anaphoric as a cover term for both cataphoric and anaphoric uses of words like so, on the assumption that both anaphoric and cataphoric uses involve a word whose interpretation requires the hearer to find an appropriate reference point in the conversational context. In section 2.3., we also distinguished between anaphoric degree adverbs, like so and cosı́, which are generally found in exclamations, and nonanaphoric degree adverbs like very and molto, which are not involved in the expression of exclamative meaning. 1045 79. Exclamative constructions Using as a diagnostic of anaphoricity the ability of a degree word to appear in anaphoric contexts like the correlative resultant-state construction exemplified in (15), we find that when languages use degree words other than question words in exclamative constructions, these are anaphoric degree words analogous to so (in English and German), Italian cosı́, and French tellement. Examples of anaphoric degree words in exclamative contexts, and in the diagnostic resultant-state context, are given in (43⫺44) for Malay, in (45⫺46) for Croatian, and in (47⫺48) for Turkish: (43) ‘I can’t believe she spent so much.’ (1985) observe this tendency, and ascribe it to the fact that both exclamatives and interrogatives are nonassertoric. Given this commonality, they say, it stands to reason that the two sentence types should share formal features. However, it is not the class of interrogative constructions per se whose properties (whatever they might be) are found in exclamatives. The interrogative type which exclamations most closely resemble typologically is the information (or ‘wh’) question. In exclamatives like the Vietnamese example in (49), we see a structural relationship with information questions, in that an argument, determiner, or adjunct role is filled by an (in situ) question word (which is sometimes, as in Vietnamese, identical to the set of indefinites): Cuaca panas sangat sampai (49) Saya tidak percaya banyak I believe neg sangat duit so (44) much dia dah guna. money s/he past use weather hot so until saya hampir pengsan. I ngu’o{’i! almost faint Tako je So vruc« e. people Bilo je tako vruc« e da be.PaP.sg.n is.3sg so sam se hot that skoro onesvjestila. is.1sg refl almost faint.PaP.sg.f ‘It was so hot I almost fainted.’ (47) Öyle zenga so ki! rich.3sg.pres excl ‘He is so rich!’ (48) Öyle zengin so ‘There are so many people there!’ (50) is.3sg hot ‘It’s so hot!’ (46) ki, bao nhiêu là be.at there have how-many ints ‘It was so hot I almost fainted.’ (45) Vietnamese { O’ Ió có yat bile rich.3sg.pst result yacht even aldi. buy.pst.3sg ‘He was so rich that he even bought a yacht.’ As we noted in section 2.3, the prevalence of anaphoric degree words in exclamatives can be explained on the assumption that the scalar proposition expressed in the exclamation is presupposed. If a speaker in using an exclamative construct like He is so rich is invoking a knowledge presupposition of the form ‘He is rich to some extent’, then an anaphoric degree adverb like so can then be used to refer to that point on a scale of wealth. 4.5. Information-question form Perhaps the most prevalent source, cross linguistically, of degree words in exclamations is the set of question words. Sadock & Zwicky Vietnamese { O’ Ió có bao nhiêu ngu’o{’i? be.at there have how-many people ‘How many people are there?’ While (49⫺50) show that exclamatives and information questions may look alike, constructions which instantiate the two sentence types have distinct formal markings. For example, nonsubject information questions feature subject-auxiliary inversion in English, while wh-exclamatives lack inversion. Intonational distinctions between exclamations and information questions are also obvious in English, where question-word exclamations feature a tune distinct from the H* L L% pattern which Pierrehumbert & Hirschberg (1990) identify with declaratives and whquestions. Exclamations are also characterized by exclamative markers, like the intensifying postmodifier là in Vietnamese (exemplified in (49)), and the postclausal exclamative particle ki in Turkish, exemplified in (47). Further, the syntactic behavior of question words may differ in the two classes of constructions. In English, e. g., the modifier what may precede an indefinite NP in an exclamative (What a sad story!) but not in an information question (*What a sad story did you read?). Hence, while exclamations may owe aspects of their form to information questions, their formal properties are determined by exclamative constructions. 1046 X. Syntactic Typology The idiomatic nature of question-form exclamations is also evident when we look at their external syntax and semantics (i. e., their behavior as complements). Milner’s label indirect exclamative suggests an analogy between exclamative complement clauses and indirect questions ⫺ an analogy which appears to have been overstated. An indirect question, as in the boldfaced portion of (51), does not have the illocutionary force of the corresponding direct question. Sentence (51) is a declarative, and not a question. By contrast, an indirect exclamative, as in the boldfaced portion of (52), has the same force that its complement clause would if it were used as a matrix exclamative. Both (52) and its complement clause count as exclamations: (51) I know who left. (52) I can’t believe how few people really care. Why does the complement of (52) represent the same kind of speech act as (52)? The answer is straightforward when we recall that aspects of the exclamative speech-act scenario (like the source of the noncanonicity judgement) may be recovered pragmatically rather than directly encoded. Matrix exclamatives, such as question-form exclamatives and anaphoric degree exclamations, require the interpreter to recover the affective stance appropriate to the semantico-pragmatic model which these constructions instantiate: the exclamative sentence type. One puzzle that arises here is the following. As Elliott (1974: 236⫺237) and McCawley (1988: 717) have observed, not all question-form exclamatives which serve as indirect exclamatives are equally able to serve as matrix exclamatives. In English, the only indirect exclamatives which correspond to wellformed matrix exclamatives are wh-clauses introduced by the degree word how. Other wh-clauses cannot stand alone as matrix exclamatives. Well-formed indirect exclamatives are given in the (a)-sentences of (53⫺55), with corresponding anomalous matrix exclamatives in the (b)-sentences: (53) (a) You won’t believe who they hired. (b) ?Who they hired! (54) (a) I can’t believe where they go! (b) ?Where they go! (55) (a) I’m amazed at what we found. (b) ?What we found! One likely source of the grammaticality facts evidenced in (53⫺55) is this: the wh-complements in these examples do not intrinsically invoke a property scale. The person uttering (53a) invokes a presupposed proposition ‘They hired someone’. This proposition does not obviously evoke a scale. Nevertheless, the sentence is not easily taken to mean: ‘They hired a certain person, and this surprises me’. Instead, the sentence does seem to evoke a scale. In accordance with Fillmore & Kay & O’Connor 1988, we assume that individuals are assigned positions on scales. Sentence (53a) presupposes or rather creates the presupposition that the person hired deserves to be ranked on the scale of incompetence. The sentence asserts that this ranking is remarkably high. The fact that (53a) can be construed as presupposing a propositional function of the form ‘The person they hired is at some point on the incompetence scale’ is a fact about constructional meaning, and the way in which constructions can impose meaning on their lexico-grammatical ‘fillers’: the wh-clause receives the appropriate scalar interpretation only in the context of the indirect-exclamative construction. When used as a matrix exclamation, as in (53b), the whclause lacks the syntactic context needed to force the scalar interpretation. How-clauses in English are unique not only in their ability to serve as matrix exclamatives, but also in their ability to yield a committed reading (Cruse 1986) in nonexclamative factive contexts like (56): (56) I realize how hard you tried. (J you tried hard) Other wh-clauses, like Where you are, since they do not evoke a scale, cannot be said to be commited with regard to a scalar degree in factive contexts (like I realize where you are). Hence, in English, the ability to yield a commited reading in factive contexts may be the property which enables how clauses to serve as matrix exclamatives. The constraint exemplified in (53⫺55) is not universal. There are languages which allow matrix exclamatives like those in the (b)-sentences of (53⫺55). As shown in (57), Italian is among these languages, as is Turkish, as shown in (58⫺59): (57) Dove si arrampicano, questi Where they climb.3pl these ragazzi! boys ‘The places they climb, these boys!’ 1047 79. Exclamative constructions (58) Turkish Kimleri gördük, (61) Everyone’s afraid that the next cutback will involve them. You wouldn’t believe the bickering that goes on! ⫺ ‘For Better or for Worse’ 8/5/94 (62) C’est incroyable le bruit qu’ils font. ‘It’s incredible the noise they make.’ (63) È pazzesco il rumore che fanno. ‘It’s incredible the noise they make.’ (64) Unfassbar, der Krach, den sie machen. ‘Unimaginable, the noise that they make.’ (65) Swetswana Ga ke dumele ka moo a (kim)! who.pl.obj saw.pst.1pl who ‘The people we saw!’ (59) Turkish Neler bulduk, (neler)! what.pl find.pst.1pl. what.pl ‘The things we found!’ The English translations of (57⫺59) employ definite NPs, and thus represent hidden exclamatives of the type to be discussed in sections 4.6.⫺4.7. The widespread use of question forms in exclamations, both direct and indirect, has a straightforward semantico-pragmatic basis. It has long been maintained by a variety of scholars that an information question presupposes a propositional function in which the argument, adjunct or modifier encoded by the question words is represented as a variable (Jackendoff 1972, Prince 1986, Rooth 1992, Raymond & Homer 1996, Lambrecht 1994, Lambrecht & Michaelis 1998). Thus, for example, the open proposition presupposed by (60a) is (60b): (60) (a) How much did he spend? (b) He spent x amount As we have seen in sections 2 and 3, it is reasonable to propose that an exclamation of the form I can’t believe how much he spent or How much he spent! also presupposes (60b). Exclamations and questions differ with regard to what is asserted. In using (60a), the speaker asserts the desire to know where the spending ranks on a numerical scale (Lambrecht & Michaelis 1998). In using an exclamative like How much he spent!, the speaker asserts that the spending ranks high on that numerical scale. However, both speech acts have the same pragmatic starting point: the speaker takes for granted, and presumes that the hearer is willing to take for granted, the proposition in (60b). Since exclamations and information questions have identical presuppositional structure, it makes sense that this shared pragmatic feature should be reflected in a formal overlap between these two sentence types. 4.6. NP Complements The tendency to use hidden exclamatives as exclamative complements is widespread. Examples are given for English, French, Italian, German, Setswana and Turkish in (99⫺104), respectively: neg I believe clf way she dirisang madi ka teng uses.prog money prt prt ‘I don’t believe the way she spends money.’ (66) Turkish Yaptıklari gürültü make.pst.rp.pl.obj noise.obj inanmtyorum! believe.neg.prs.1sg ‘I can’t believe the noise they make.’ Each of these NP complements is readily translatable by a question-form complement introduced by how much. It is not, e. g., the noise itself that engenders disbelief, but the duration or amplitude of the noise. Hence, Michaelis & Lambrecht (1996a, 1996b) claim that these NPs refer metonymically to a point on a scale. The metonymic target is often indeterminate, since a sentence like I can’t believe the people you know may be used to invoke the number, the variety, or the peculiarity of the people in question. In typically requiring context for recovery of the appropriate scale, hidden-exclamative complements resemble question-form complements like that in (53a), which require the hearer to invoke an appropriate scale on which to rank the person in question. An interpretively vague nominal head which frequently appears in hidden exclamatives is one denoting manner, as in the Setswana sentence (65), or the English sentence in (67): (67) I can’t believe the way they treat us. The hidden exclamative in (67) has the same indeterminacy as the question-form complement in (68): (68) I can’t believe how she treats us. 1048 X. Syntactic Typology Both (67) and (68) are indeterminate as to whether the relevant scale for treatment is cruelty, condescension, etc. The use of a NP to denote a scalar degree is motivated in terms of semantico-pragmatic properties of the exclamative sentence type. The proposition presupposed by an exclamation refers to a scalar extent. A scalar extent is something which can be indexed, as we noted in the discussion of anaphoric degree expressions in section 4.4. Something which can be indexed counts as referential, i. e., as an entity. Since nouns prototypically refer to entities (Croft 1990: 64⫺154), it stands to reason that a noun should be used to refer to a scalar extent in a construction which serves to comment on that extent. The particular use of definite NPs in exclamative contexts (in those languages which express definiteness) can be motivated by reference to the presupposed status of the open proposition denoted by the NP. If a sentence like (61) presupposes a proposition like ‘There is some degree of bickering’, then this degree is also mutually identifiable to speaker and hearer. The claim that factivity motivates the definiteness of hidden exclamatives is substantiated by the use of definite NPs in nonexclamative factive contexts, as in (69): (69) (70) Le bruit qu’ils font! ‘The noise they make!’ (71) Der Krach, den manche Leute machen! ‘The noise that some people make!’ (72) Turkish Gittikleri ‘The places they go!’ (73) 4.7. Free NPs The exclamative use of a free NP is exemplified in (8) for English. The free-NP type of exclamation is exemplified for French, German, Turkish, Setswana and Korean in (70⫺ 74): Setswana Mo.dumo o clf.noise ba o dirang! rp they om make.prog ‘The noise they make!’ (74) Korean Ah, cheo sori! interj the sound ‘The noise!’ While English and German generally require that isolated-NP exclamatives contain a relative clause, French does not, as seen in the following attested example: (75) I regret the trouble we caused. Since it contains a factive matrix verb, sentence (69) can be seen as presupposing a proposition of the form ‘We caused some degree of trouble’. The presupposed status of this proposition can then also be seen as rendering this degree identifiable. Referent identifiability has already been mentioned as a semantico-pragmatic constraint on exclamatory statements (13e). The statement in (13e) pertained to the entities of which scalar properties, like that of spending a large amount of money, are predicated. In this section, however, we see that identifiability is a property that we can use to characterize two referents in an exclamation: the described entity and the degree. Insofar as this is the case, exclamations are double predications: they not only predicate a scalar property of a given referent, but also predicate a property (that of violating expectation) of a degree. yerler! go.pst.rp.pl.obj place.pl [Child looking at a man’s large stomach.] Le bide! […] Le gros bide comme ça! [gestures]. […] Le plus gros bide de l’année. ⫺ Reiser, Les Oreilles Rouges ‘The stomach [on this guy]! A stomach like this! [gestures] The biggest stomach of the year!’ The motivation for the exclamative use of free NPs is the same as that brought out in the discussion of matrix exclamations which contain question words and anaphoric degree words. Exclamative constructions, like other expressive forms, need not overtly specify the speaker’s affective stance toward the content encoded. This stance can be inferred from the speaker’s choice of an exclamative form. 4.8. Inversion A minor pattern instantiated by matrix exclamative constructions is the inversion of subject and finite verb, discussed by McCawley (1973) for English and exemplified for English and German in (76⫺77), respectively: (76) The narrative is pretty jerky, but, man, can this kid direct second unit! ⫺ Time 5/19/97 (77) Hast du Glück gehabt! ‘Did you luck out!’ Both McCawley (1988) and Sadock & Zwicky (1985) have related the use of inversion here 79. Exclamative constructions 1049 Ehlich, Konrad. 1986. Interjektionen. Tübingen: Niemeyer. to the use of inversion in interrogative contexts. The use of the inversion pattern in both interrogative and exclamative contexts is motivated for these theorists insofar as both of these sentence types express nondeclarative speech acts. If we focus only on the use of inversion in yes-no questions, the adduced motivation seems valid: only yes-no questions share with inversion exclamatives the property of using inversion as the sole syntactic feature which marks a deviation from declarative syntax. Yes-no questions deviate from the declarative prototype in that their content is not asserted. Similarly, exclamatives, as discussed in section 2, do not assert their propositional content, but rather presuppose it. Geis, Michael L. 1995. Speech acts and conversational interaction. Cambridge: Cambridge University Press. 5. Goldberg, Adele. 1995. Constructions: A construction grammar approach to argument structure. Chicago: University of Chicago Press. Conclusion As Bybee & Perkins & Pagliuca (1995: 3) argue, explanation in linguistics requires one not merely to describe the functions associated with a given construction, but also to address the question of why that form has the functions it does. This question can be answered in both a synchronic model, which concerns the way in which grammatical structures are motivated via relations of association (formal and semantic overlap), and a diachronic model, which concerns patterns in the semantic extension of forms. This typological survey has shown that exclamations are characterized across languages by several recurrent formal features ⫺ most saliently, the presence of information-question forms and anaphoric degree adverbs. Appeal to semantico-pragmatic features of the exclamative sentence type has enabled us to show why these grammatical forms are used to express noncanonicity judgements which involve semantic scales. 6. References Elliott, Dale E. 1974. “Toward a grammar of exclamations”. Foundations of Language 11: 231⫺46. Fillmore, Charles. 1982. “Toward a descriptive framework for spatial deixis”. In: Jarvella, Robert J. & Klein, Wolfgang (eds.). Speech, place and action. London: Wiley, 31⫺59. ⫺. 1986. “Pragmatically controlled zero anaphora”. Berkeley Linguistics Society 12: 95⫺107. Fillmore, Charles & Kay, Paul & O’Connor, Mary C. 1988. “Regularity and idiomaticity in grammatical constructions”. Language 64: 501⫺538. Grimshaw, Jane. 1979. “Complement selection and the lexicon”. Linguistic Inquiry 10: 279⫺326. Hoeksema, Jacob & Napoli, Donna Jo. 1993. “Paratactic and subordinative so”. Journal of Linguistics 29: 391⫺314. Jackendoff, Ray. 1972. Semantic interpretation in generative grammar. Cambridge, MA: MIT Press. Kay, Paul & Fillmore, Charles. 1998. Construction grammar. Stanford: Stanford University Press. Lambrecht, Knud. 1994. Information structure and sentence form. Cambridge: Cambridge University Press. ⫺. 1996. “On the formal and functional relationship between topics and vocatives: Evidence from French”. In: Goldberg, Adele (ed.). Conceptual structure, discourse and language. Stanford: CSLI, 267⫺288. ⫺ & Michaelis, Laura. 1998. “On sentence accent in information questions”. Linguistics and Philosophy 21: 477⫺544. Makkai, Adam. 1985. “Where do exclamations come from?” In: Makkai, Adam & Melby, Alan K. (eds.). Linguistics and philosophy: essays in honor of Rulon S. Wells. Amsterdam: Benjamins, 445⫺472. Aissen, Judith. 1992. “Topic and focus in Mayan”. Languages 60: 48⫺80. McCawley, James D. 1988. The syntactic phenomena of English, 2. Chicago: University of Chicago Press. Austin, John L. 1962. How to do things with words. Oxford: Clarendon Press. McCawley, Noriko, 1973. “Boy, is syntax easy!” Chicago Linguistics Society 9: 369⫺377. Bybee, Joan & Perkins, Revere & Pagliuca, William. 1994. The evolution of grammar: Tense, aspect and modality in the languages of the world. Chicago: Chicago University Press. Michaelis, Laura. 1994a. “A case of constructional polysemy in Latin”. Studies in Language 18: 45⫺ 70. Croft, William. 1990. Typology and universals. Cambridge: Cambridge University Press. Michaelis, Laura. 1994b. “Expectation contravention and use ambiguity: The Vietnamese connective cũng”. Journal of Pragmatics 21: 1⫺36. Cruse, D. A. 1986. Lexical semantics. Cambridge: Cambridge University Press. Michaelis, Laura & Lambrecht, Knud. 1996a. “Toward a construction-based model of language func- 1050 X. Syntactic Typology tion: The case of nominal extraposition”. Language 72: 215⫺247. Prince, Ellen. 1981. “Topicalization, focus movement and Yiddish movement: A pragmatic differentiation“. Berkeley Linguistics Society 7: 249⫺264. Raymond, William R. & Homer, Kristin. 1996. “The interaction of pragmatic roles and thematic structure in the selection of question form”. Berkeley Linguistics Society 22: 316⫺327. Rooth, Mats. 1992. “A theory of focus interpretation”. Natural Language Semantics 1: 75⫺116. Sadock, Jerrold, & Zwicky, Arnold. 1985. “Speech act distinctions in syntax”. In: Shopen, Timothy (ed.). Language typology and syntactic description, 1. Cambridge: Cambridge University Press, 155⫺ 196. Searle, John R. 1979. Expression and meaning. Cambridge: Cambridge University Press. Slobin, Dan & Aksu, Ayhan. 1982. “Tense, aspect and modality in the use of the Turkish evidential”. In: Hopper, Paul J. (ed.). Tense aspect: Between semantics and pragmatics. Amsterdam: Benjamins, 185⫺200. Zwicky, Arnold. 1995. “Exceptional degree markers: A puzzle in internal and external syntax”. Ohio State University working papers in linguistics 47: 111⫺23. Prince, Ellen. 1986. “On the syntactic marking of presupposed open propositions”. Chicago Linguistics Society 22: 208⫺222. Laura A. Michaelis, University of Colorado, Boulder (USA) ⫺. 1996b. “The exclamative sentence type in English”. In: Goldberg, Adele (ed.). Conceptual structure, discourse and language. Stanford: CSLI, 375⫺ 390. Milner, Jean-Claude. 1978. De la syntaxe à l’interpretation. Paris: Seuil. Morel, Mary-Annick. 1995. “L’Intonation exclamative dans l’oral spontane”. Faits de Langues 6: 63⫺70. Pierrehumbert, Janet B. & Hirschberg, Julia. 1990. “The meaning of intonational contours in the interpretation of discourse”. In: Cohen, Philip & Morgan, Jerrold & Pollack, Martha (eds.). Intentions in communication. Cambridge, MA: MIT Press, 271⫺311. 80. Dislocation 1. 2. 6. 7. Definition and terminology Dislocation and superficially similar constructions The structure of dislocation sentences The grammatical status of dislocated constituents Discourse functions of Left-Dislocation and Right-Dislocation Special abbreviations References 1. Definition and terminology 3. 4. 5. A dislocation construction (also called detachment construction) is a sentence structure in which a referential constituent which could function as an argument or adjunct within a predicate-argument structure occurs instead outside the boundaries of the clause containing the predicate, either to its left (left-dislocation, henceforth LD) or to its right (rightdislocation, henceforth RD). The role of the denotatum of the dislocated constituent as an argument or adjunct of the predicate is represented within the clause by a pronominal element which is construed as coreferential with the dislocated phrase. Typically, the dislocated phrase is marked with special prosodic features. The above definition involves four criteria: (i) extra-clausal position of a constituent, (ii) possible alternative intra-clausal position, (iii) pronominal coindexation, (iv) special prosody. These four criteria apply in prototypical instances. However, there are many instances in which one or more of them fail to apply. Only criterion (i) is a necessary (though not sufficient) condition for a sentence construction to qualify as an instance of dislocation. Our definition will be modified as we go along. Examples (1a) and (2a) are attested instances of LD and RD in English, followed by their canonical (i. e. non-dislocated) counterparts in (1b) and (2b). The dislocated constituents are enclosed in square brackets, for easy recognition. The coreference relation between the dislocated constituent and the intraclausal pronominal element is indicated by subscripts. Following a common orthographic practice, the clause boundary is signalled by a comma; this comma does not indicate a pause. The small capitals indicate the