Home > Speech Act Distinction in Syntax

Speech Act Distinction in Syntax

Page 1
Timothy Shopen (ed.) Language Typology and Syntactic Description. Cambridge: Cambridge University Press.
Speech Act Distinctions in Grammar
Ekkehard König & Peter Siemund 1 Speech acts and sentence types1 In contrast to the traditional view that the function of language is essentially a descriptive one, it is now generally accepted that in speaking we perform actions of various kinds. This is clearest in the case of so-called ��performative utterances�� like the following: (1) a. I (hereby) order you to leave the room. b. I promise you never to be late again. c. I hereby declare this meeting closed. d. I hereby christen this ship ��Queen Elizabeth��. Sentences like these are special insofar as their utterance in appropriate circumstances amounts to performing the action identified by the finite verb. The typical formal properties of such sentences in English include first person subjects, second person indirect objects, a present tense non-progressive active form of a speech act verb and the deictic adverb hereby, but performative sentences may also be in the passive voice, contain modal hedges and a nominalization instead of a verb: (2) a. You are cordially invited to come to my birthday party. b. I must admit that you have won that argument. c. I��ll come to see you next week, and that��s a promise. European languages, in particular, have large inventories of such ��performative verbs��, but ��performative utterances�� of the type (1) or (2), i.e. sentences whose meaning is such that we can perform the action named by the verb just by saying literally that we are performing it, are rarely used. Most frequently, they seem to occur in institutional settings, where they are part of more elaborate rituals. In fact many, if not all, ��performative verbs�� (e.g. ��marry��, ��chris- ten��, ��appoint��, ��resign��, ��baptize��, ��veto��, ��guarantee�� and even ��bet�� or ��promise�� etc.) pre- suppose the existence of the relevant extra-linguistic social institutions. A second important context seems to be the one where the action performed by an utterance is unclear and needs to be made fully explicit. In most verbal interactions, however, the kind of speech act per- formed by an utterance is only very weakly determined by the meaning of the sentence ut- tered. A simple imperative like ��Sit here��, for instance could be used as a command, request, offer, advisory or exhortation, depending on the context, as is shown by the following poten- tial responses: ��Yes, sir�� (command), ��Okay�� (request), ��No thanks�� (offer), ��What a good idea�� (advisory), ��Thank you�� (exhortation) (cf. Clark, 1996: 213). Examples such as these show that it is only the communicative potential of a sentence, a default interpretation, that is determined by its formal and semantic properties. The precise speech act performed by an utterance is the result of an interaction between these properties and various contextual fac- tors, such as the social situation, the current state of an interaction and the background knowl- edge of speaker and hearer. Moreover, our examples suggest that the precise function (illocu- tionary force) of an utterance may partly be the result of cooperative negotiations between speaker and hearer. Three basic sentence types are traditionally distinguished for European languages and have also been found useful for many other languages: declarative, interrogative and impera- tive sentences.
Declarative sentences are primarily and most frequently used for speech acts

Page 2
such as asserting, claiming, stating, but also accusing, criticizing, promising and guaranteeing. Interestingly enough, all performative sentences are also of the declarative type. Interrogative sentences are typically used for eliciting information, asking questions, introducing delibera- tions, etc. and imperatives have their basic use in all attempts to get or advise the hearer to do something, i.e. speech acts such as orders, requests, suggestions, prescriptions, appeals, etc. The following three examples are instances of the three basic sentence types in English, all involving the same sentence radical and proposition (cf. (3)). It is only in combination with other factors, both linguistic and contextual, that such sentences are restricted to expressing specific speech acts. (3) a. John is taking out the garbage. b. Is John taking out the garbage? c. Take out the garbage, John. Any attempt to characterize the patterns and limits of variation across languages in the do- main of sentence types cannot simply be based on the assumption that the three types men- tioned above (cf. (3)) are distinguished in all languages or exhaust the list of basic sentence types that may be distinguished. What this list shows, however, is that we are looking for grammatical distinctions that can be correlated with a certain use potential or illocutionary functions. What we need to correlate are, on the one hand, general semantic and/or functional distinctions, as found in various typologies of speech acts (cf. Austin, 1962; Searle, 1976). Searle��s typology, for example, is based on a variety of criss-crossing dimensions (the point or purpose of an utterance, the way the content is related to the world, whether obligations are introduced for the speaker or for the hearer, etc.) and distinguishes five basic types of speech acts: representatives, commissives, declarations, directives and expressives. On the other hand, a cross-linguistically useful definition of sentence types must also be based on formal criteria: the sentence types or, more specifically, the formal properties characterizing those types should ideally form a system of alternative choices that are mutually exclusive, such that each sentence token can be assigned to one type and no sentence token can be a member of more than one type (cf. Sadock & Zwicky, 1985: 158):
sentence types declarative interrogative imperative other
Figure 1: Sentence types
This formal criterion is ideally fulfilled in those cases where the formal markers (inflectional affixes, word order patterns, particles, etc.) identifying the basic sentence types in a language form a system of alternative choices.
In Greenlandic Eskimo (cf. Sadock & Zwicky, 1985; Sadock, 1984), for instance, the three basic sentence types are identified by different verbal affixes (all abbreviations are explained in a list at the end of the paper):
(4) Greenlandic Eskimo a. Iga-voq cook-DEC.3.SG ��He cooks.�� 2

Page 3
b. Iga-va cook-INT.3.SG ��Does he cook?�� c. Iga-git / -guk cook-IMP.2.SG / -IMP.2/3.SG ��Cook (something) / it!�� A similar opposition between three basic sentence types (declarative, interrogative, impera- tive) is found in Nama Hottentot (cf. Hagman, 1977), and in Khoisan languages in general, but in these languages the markers of sentence type are particles rather than affixes (cf. (5)). A noteworthy property of this system for marking basic sentence types is that only the particle marking declarative sentences is obligatory. Interrogative and imperative sentences are identi- fied via specific intonation patterns, but can also occur with the relevant particles. The occur- rence of a declarative particle in interrogative or imperative sentences is strictly ruled out. (5) Nama Hottentot a. Declarative: NP + DEC + PredP (DEC = ke, km) b. Interrogative: NP�� + (INT) + PredP (INT = kxa) c. Imperative: NP�� + PredP + (IMP) (IMP = r��) In Korean, by contrast, the system of moods (declarative, interrogative, imperative, adhorta- tive) interacts with honorification, i.e. with five or six different verbal affixes being used for each sentence type depending on the speech style (formal, polite, blunt, familiar, intimate, plain) and thus ultimately on the interplay between speaker, hearer and the participants of the situation talked about (cf. Table 1, taken from Chang, 1996: 191). Evidently, the differentia- tion is not drawn on all speech levels. Speech Level Sentence Type Declarative Interrogative Imperative Adhortative High Formal Polite (su)pnita (e)yo (su)pnikka (e)yo (u)psio (e)yo (u)psita (e)yo Mid Blunt Familiar so ney so na so key so sey Low Intimate Plain e ta e (nu)nya e la e ca
Table 1: Markers of sentence type in relation to speech levels in Korean
These, however, are by no means wide-spread situations. In many languages interrogative sentences can simply be derived from their declarative counterparts through the addition of a particle or tag. In Shona (Bantu), for instance, it is the addition of the particle here (yes/no- questions) or of an interrogative pronoun (constituent questions) that turns a declarative into an interrogative sentence: (6) Shona a. nd��-n��-t����r-�� 1.SG-PRES-speak-DEC ��I speak.�� b. nd��-n��-t����r-�� h��r��? 1.SG-PRES-speak-DEC INT ��Do I speak?�� 3

Page 4
And in Japanese, declarative sentences are differentiated from imperatives by a clear para- digmatic contrast, but are unmarked with respect to interrogatives: (7) Japanese a. Sakana-o tabe-ro/-te (kudasai)! fish-ACC eat-IMP ��Eat the fish!�� b. Sakana-o tabe-ru. fish-ACC eat-PRES.DEC ��I eat fish.�� c. Sakana-o tabe-ru-ka? fish-ACC eat-PRES.DEC-INT ��Do you eat fish?�� Another common difficulty with the view of a clear paradigmatic opposition between the three basic sentence types under discussion is the fact that the imperative is often expressed by a specific inflectional form even in languages which do not distinguish the two other types by morphological means. In such languages the imperative is often analyzed as being one option in a system of ��mood��, which also includes the categories ��indicative��, ��subjunctive��, ��conditional��, ��optative�� and perhaps others (cf. Palmer, 1986: 23ff.):
mood indicative subjunctive imperative other
Figure 2: Mood distinctions
A clear paradigmatic opposition between basic sentence types is sometimes found with em- bedded sentences, particularly of European languages. In such structures the relevant opposi- tion is expressed by different complementizers or the lack thereof (cf. (8)). Note, however, that imperatives cannot be embedded and that there is thus no ��imperative complementizer��. Such complementizers have no bearing on the speech acts that may be performed by the sen- tence containing them, unless they occur in an independent, non-embedded sentence (cf. (9)).
(8) a. I knew that John did it. (declarative) b. Fred wonders whether/if John did it. (interrogative) c. Fred asked John to help him. (jussive) (9) a. That I should live to see this! b. A: Bist Du m��de? – B: Und ob (ich m��de bin)! [German] are you tired and whether I tired am ��A: Are you tired? – B: Am I ever!�� The set of expressive devices used in the languages of the world for the differentiation of ba- sic sentence types includes all those that are generally considered to be part of grammatical systems: intonation, inflection (more specifically the addition or omission of inflectional af- fixes), word order and the addition, omission or substitution of constituents.
Intonation clearly has a special role, since it is never in paradigmatic contrast with another strategy. 4

Page 5
Moreover, any intonation contour, or at least most of them, can apparently be superimposed on any segmental structure, an option that is frequently exploited in so-called echo-questions. It is for these reasons that we hesitate to regard intonation as the most important strategy for the identification of sentence types and in fact hesitate to identify a formal sentence type in a language purely in terms of intonation. In spite of such difficulties of basing the distinction between sentence types on formal paradigmatic oppositions, it is still possible to draw such distinctions for most languages. De- spite a certain heterogeneity in the formal inventory used to identify different sentence types, sentences can usually be assigned to one and only one basic sentence type within a language without fulfilling additional conditions. The labels ��declarative��, ��interrogative�� and ��impera- tive�� can then be assigned to these formal types on the basis of their typical use.
Moreover, these three basic sentence types – which have played a prominent role in the analysis of Euro- pean languages – also seem to be more clearly identifiable across languages than other ��basic�� types mentioned in the descriptions of various languages. More often than not, these other basic types are simply the result of an interaction of two formal properties, rather than being marked by a single grammatical device: ��prohibitives�� (��Don��t do that��) can often be analyzed as negative imperatives, ��optatives�� (��(May) God bless you��) may simply be based on the 3rd person subjunctive, ��hortatives�� (��Let��s go��) may be further analyzable as first person plural imperatives and exclamatives are typically based on interrogatives or declaratives (Isn��t she wonderful!; He is such a nuisance!). Of course, all of these categories may find a unique ex- pression in a language – and it is for those cases that labels like ��prohibitive��, ��optative��, ��ex- clamative��, etc. should be reserved, but they may also be the combinatorial result of several features. The three basic types declarative, interrogative and imperative, by contrast, seem to be clearly identifiable on the basis of only one formal property in the vast majority of lan- guages, even if the relevant properties do not always constitute a neat paradigm of formal oppositions.
What the languages and data examined in this paper do not support, however, is the view that three or more sentence types can be differentiated and identified on the basis of formal properties in all languages. Allowance must be made for cases of overlap between two sentence types in specific contexts, as well as for cases where there is no clear formal distinc- tion between two types at all, typically between declaratives and interrogatives. Moreover, a clear assignment of sentence tokens to a specific type is often only possible on the basis of more than one criterion. There is a well known problem for the view that the basic sentence types have a certain functional potential or default interpretation that needs to be addressed at this point. So far we have assumed that declarative sentences have an essentially descriptive function, that impera- tives are primarily used for directive speech acts and that interrogative sentences are primarily used as questions. Such functional criteria are in fact essential for an identification of sentence types as declarative rather than interrogative across languages. Unfortunately, such a view of the connection between form and illocutionary function is clearly an oversimplification (cf. (10)). (10) a. It is terribly cold in this room. b. Could you please close the window? c. Who likes being criticized? d. You must be feeling very tired. e. Why don��t you buy some stocks? f. Waiter, what��s that fly doing in my soup? The examples in (10) show that in English both a declarative sentence (10a) and an interroga- tive sentence (10b) can be used to perform directive speech acts, i.e. as requests for the ad- 5

Page 6
dressee to close the window. The example in (10c) could be used as a rhetorical question, i.e. as a statement expressed by an interrogative sentence, and (10d) could be intended as a re- quest for information (question) expressed in declarative form. Together with a falling intona- tion contour and an accent on the participle, the example in (10c) would typically be used as a rhetorical question, i.e. as a statement expressed by an interrogative sentence. A sentence like (10d) could be intended as a request for information (question) expressed in declarative form and sentences like (10e) are more likely to be used as suggestion than as requests to supply a justification for a certain course of action. The only use imaginable for (10f) is that of an act of criticizing. Several ways of dealing with such asymmetries have been discussed in the literature (cf. Levinson, 1983: 263ff.). The most extreme position is to simply abandon the assumption of clear form–function correlations. On this view, illocutionary force is purely a matter of con- text and has no direct and simple correlation with sentence form and sentence meaning. But such a theory clearly misses very important and obvious generalizations, as will be shown in the body of this paper. Another theory (��Inference Theory��), which avoids such undesirable consequences, is based on the assumption that sentences like (10a-d) are indeed used as statements and questions, respectively, but in addition to having this illocutionary force they also have the force of a request (10a-b), a statement (��Nobody likes being criticized��), a ques- tion (��Are you tired?��) and a suggestion (��You should buy some stocks.��). This secondary function is indirectly expressed – hence the term ��indirect speech acts�� – and based on conver- sational inferences and principles of conversational interaction which ultimately trigger such inferences. Such inferences rely heavily on contextual factors. A statement of discomfort such as (10a) or I need something to drink can only be used as an indirect request if there is a hearer present who feels under an obligation to serve the interests of the speaker. In other words, such hints could simply be overlooked. This inference theory does not only solve the problems raised by examples such as (10), but also accounts for a variety of interesting facts and generalizations, only two of which will be mentioned here: Typical responses to requests expressed by interrogative sentences like (10b) include both a verbal reaction (��yes��) and a non-verbal one (carrying out the action) and thus support the view that two actions are performed by such indirect speech acts. Further- more, such comparative studies as have been carried out in this domain show that the sen- tences that can be used to perform speech acts indirectly in English have clear parallels in other languages and that therefore the principles for using sentences in indirect speech acts must be very similar for different languages and cultures. What may differ from language to language is the addition of certain expressions to such sentences which would normally be used for the direct performance of the relevant speech act. In German, for instance, assertive particles like schon are not only used in declarative sentences, but also in rhetorical questions. The German counterpart of (10c) may thus take the following shape: (11) Wer wird schon gern kritisiert? [German] who is
voluntarily criticized ��Who likes being criticized?�� The addition of schon, however, clearly indicates that the relevant sentence is used as a state- ment, rather than as a question, just like the addition of please to an interrogative sentence indicates that the sentence is used as a request (Could you please get me some coffee?). As a result such patterns may become conventionalized for a certain use and a new sentence type can be created. 6

Page 7
2 Declarative sentences Declarative sentences are conventionally and typically used to perform representative (de- scriptive) speech acts such as assertions, reports, acts of complaining and bragging, but also acts of predicting and promising. All such acts convey the belief of the speaker that the proposition expressed is true or will turn out to be true. Of course, declarative sentences can also be used for directive speech acts, such as requests (12a) and commands (12b): (12) a. Du räcker mig salt-et! [Swedish] you pass me salt-DEF ��Can you pass me the salt?�� (lit.: You pass me the salt!) b. Du bezahlst jetzt sofort deine Schulden! [German] you pay.2.SG now right now your debts ��You are going to pay your debts right now.�� As is shown by the preceding examples, however, a number of additional conditions have to be met for such a use to be possible: Only in combination with second-person subjects, with a non-past tense and with a non-stative predicate can a declarative sentence be used as a request or as an order (cf. (12)). 2.1 Declaratives in relation to the other basic types Among the three major types distinguished so far, declarative sentences are the ��unmarked�� member, in several senses of this term. They can be considered as unmarked relative to inter- rogative and imperative sentences for at least the following reasons: (i) Declarative sentences are the most frequent sentence type. (ii) The word order exhibited by declarative sentences (SOV, SVO, VSO, etc.) is normally regarded as the basic word order of a language. (iii) Declarative sentences are less restricted in their distribution compared to the two other types. Embedded clauses often have the same formal properties as declara- tive sentences and, as pointed out above, ��performative utterances�� are also of the declarative type. (iv) Declarative sentences exhibit the full paradigm of tense-aspect combinations available in a language, in contrast to imperatives. (v) Declarative sentences may be used to express most of the speech acts distin- guished in the typology by Searle, i.e. they can be used as representative speech acts (It is raining), as commissives (I will never again forget your birthday), as di- rectives (You know what you have to do), as expressives (I am sorry) and as decla- rations (He is guilty). (vi) In many languages, interrogatives sentences and partly also imperatives can be analyzed as being the result of some operations (adjunction, omission, change of word order) performed on declaratives, rather than the other way round. In languages that have systems of inflectional mood distinctions, declaratives are char- acterized by the indicative mood, of course, but so are interrogatives and sometimes even conditionals. So, perhaps, a falling intonation contour is a more reliable formal indicator of declaratives, but such a claim can only be made for the tiny subset among the world��s lan- guages whose prosodic properties have been examined in sufficient detail. Overall, the view that declaratives do not have a specific formal marker is wrong: Declarative sentences are characterized by precisely those formal properties which distinguish them from other sentence 7

Page 8
types, e.g. the absence of an interrogative marker and the absence of an imperative inflec- tional form or the presence of a specific finite verbal form. As already mentioned, interrogatives and partly also imperatives can frequently be de- scribed as being the result of modifying declaratives in some way; by changing the word or- der (inversion), adding a particle, etc. English interrogatives, for example, exhibit inversion of subject and auxiliary verb relative to declarative word order, unless the interrogative pronoun is the subject: (13) a. You are taking the train. b. Are you taking the train. c. Who did you see? d. Who saw the thief? Another typical situation is the formation of interrogative sentences from declarative sen- tences by adding a special particle. Such interrogative particles are quite frequent among the languages of the world. They are used in Tzotzil, a Mayan language spoken in Mexico (cf. (14)), and also in French (cf. (15)): (14) Tzotzil a. ch-a-bat
��You are going.�� b. mi cha-a-bat
��Are you going?�� (15) French a. Jean est malade. ��Jean is sick.�� b. Est-ce que Jean est malade? ��Is John sick?�� In those languages in which the formal means for marking the basic sentence types form a system of paradigmatic oppositions, declarative sentences receive special marking.
However, our overall impression is that languages with such paradigmatic oppositions are relatively rare. We have found paradigmatic oppositions of three different kinds. Firstly, the three major sentence types can be distinguished inflectionally within an elaborate mood system. In addi- tion to Eskimo (cf. (4) above), there are some language families in Brazil (Tariana, Tucano, Jarawara, etc.) with such mood systems, which typically include several imperatives. Sec- ondly, there may be different particles for declarative, interrogative and imperative sentences, which are mutually exclusive. The only language family in our sample illustrating such a paradigm is Khoisan (cf. (5) above). Thirdly, the three basic types of sentences may be as- signed different word order patterns. In German, for example, the basic word order is gener- ally assumed to be SOV, the word order manifested by subordinate clauses (cf. (16)). From this basic word order the sequential organization of the three sentence types can be derived as follows: (i) interrogatives are the result of moving the finite verb to initial position (cf. (17a));
(ii) declarative main clauses involve the additional operation of moving a constituent (the topic) in front of the finite verb, thus resulting in the order TVX, i.e. topic-verb-the rest (cf. (17b)); (iii) imperatives exhibit the same order as declaratives, but typically lack a (sec- 8

Page 9
ond person) subject and are also characterized by a special (impoverished) inflection (cf. (17c)); (16) Der Kanzler behauptet [dass ein Freund ihm das Geld gab]. [German] that a friend him the money gave ��The Chancellor claims that a friend gave him the money.�� (17) German a. Gab ein Freund ihm das Geld? ��Did a friend give him the money?�� b. Ein Freund gab ihm das Geld. / Das Geld gab ihm ein Freund. / Ihm gab das Geld ein Freund. ��A friend gave him the money.�� c. Gib ihm das Geld! ��Give him the money!�� The differentiation of the basic sentence types through word order is a typical feature of the Germanic languages (German, English, Norwegian, Swedish, Danish) and rarely found out- side of Europe.
2.2 Interaction with evidentiality In languages with markers signaling the kind of evidence (hearsay, common knowledge, first- hand visual evidence, etc.) on which a claim is based or the degree of strength with which an assertion can be made, these so-called ��evidential markers�� can normally only be combined with declarative sentences. This is little surprising given that declarative sentences are typi- cally used to express claims, assertions, statements about the world (of discourse) and thus indicate an attitude of belief (in the truth of the proposition expressed). However, this co- occurrence restriction is not sufficient proof for saying that evidentials are markers of declara- tive sentences. A language that is frequently mentioned in this context is Hidatsa (a Northern Plains Siouan language), which has five different particles (ski, c, wareac, rahe and toak), express- ing different kinds of evidential meaning. These morphemes occur at the end of a sentence, but do not co-occur with one another nor with the particles marking interrogatives or impera- tives (cf. Matthews, 1965: 99f.; Palmer, 1986: 70; Zaefferer, 1990: 222). Matthews provides the following names and glosses for these evidential particles: Evidential Marker Label Description ski emphatic definitive knowledge c period believed, desired or felt by the speaker wareac quotative general knowledge rahe report learnt from hearsay toak indefinite something the speaker does not know and thinks the listener does not either
Table 2: Evidentials of Hidatsa
A similar system can be found in Tuyuca (Brazil and Colombia) where five different types of evidential meaning are formally distinguished: (a) visual evidence; (b) non-visual (perceptual) evidence; (c) apparent/inferential; (d) secondhand/quotative; (e) assumed (cf. Barnes, 1984: 257; Palmer, 1986: 67). The following examples provide illustration for these 9

Page 10
distinctions. Each sentence would be translated as ��he played soccer�� into English, but is based on different evidence, as is indicated in brackets (cf. Palmer, 1986: 67): (18) Tuyuca a. d��iga ap��-wi (I saw him play.) b. d��iga ap��-ti (I heard the game and him.) c. d��iga ap��-yi (I have seen circumstantial evidence that he played.) d. d��iga ap��-yig (I obtained the information that he played from s.o. else.) e. d��iga ap��-hyi (It is reasonable to assume that he played.) Finally, in Jaqaru (Jaqi, South America) and related languages the verbal suffixes encoding evidential meaning stand in paradigmatic opposition to interrogative suffixes, may occur only once per sentence and thus appear to form a paradigm of contrastive sentence types (Hard- man, 1986: 129f.): (19) Jaqaru a. Amrucha-txi. (polar interrogative) ��Is X well?�� b. Amrucha-wa. (declarative, personal knowledge) ��X is well.�� c. Amrucha-mna. (declarative, knowledge through language) ��X is well, they say.�� In some languages evidential markers combine with all tenses. Quite frequently, however, they are restricted to certain tenses, typically the past or perfect tense. In Turkish, for instance, there is an obligatory choice for all past tense expressions: either the suffix –dI (marker of direct experience) is chosen or the marker of indirect experience (inference, hearsay) -mIs (cf. (20)): (20) Ahmet gel-mis. [Turkish] Ahmet come-QUOT/INFER ��Ahmet came/must have come.�� And in Tsez (Daghestanian, Northeast Caucasian) there is a morphologically marked distinc- tion between witnessed (-s(i)) and unwitnessed events (-n(o)) in the past (cf. Comrie & Polin- sky, 2002): (21) Tsez a. Kid mek��u-n. girl.ABS be hungry-PAST.UNWITNESSED b. Kid mek��u-s girl.ABS be hungry-PAST.WITNESSED ��The girl was hungry�� Languages which do not have such evidential markers either do not indicate the type of evi- dence their claim is based on at all or they express the relevant meaning by a higher clause (cf. (22)), by a modal verb (cf. (23)) or by a modal particle (cf. (24)), as in the following Eng- lish and German examples: (22) I see/hear/understand you are leaving this country. 10

Page 11
(23) German a. Karl soll mit dem Papst gesprochen haben. ��Charles is said to have talked to the Pope.�� b. Karl will mit dem Papst gesprochen haben. ��Charles claims to have talked to the Pope.�� (24) Du blutest ja! (visual evidence) [German] ��Why, you are bleeding.�� The preceding examples of a close interaction between declaratives and evidentiality show that there is no generally accepted inventory of categories and terms that can be used unambi- guously for a cross-linguistic description in this domain. Certain descriptive labels, such as ��visual evidence��, ��auditory evidence��, ��sensory evidence�� / ��non-visual��, ��hearsay��, ��quota- tive��, ��inferential��, ��personal knowledge��, ��general knowledge��, ��reportive��, ��(un)witnessed��, ��dubitative��, etc. are found quite frequently in individual descriptions and descriptive surveys (cf. Palmer, 1986; Chafe & Nichols, 1986), but it is not clear which of these terms describe similar, different or exactly the same phenomena. Nor is it clear what the maximal number of possible distinctions is, which categories are superordinate to others and whether all of the distinctions drawn in a language can be ordered on a single dimension. Mood oppositions as the ones found in Hidatsa (cf. Table 2) suggest that two different sub-systems should be dis- tinguished (Palmer, 1986: 53): judgements (strength of the speaker��s commitment) and evi- dentials (type of evidence, channel/source of information). A further interesting problem that should be mentioned at this point is the relationship between an affix or particle expressing ignorance or doubt (��dubitative��) and interrogative markers. In a variety of languages the same suffix or particle is used to indicate that the speaker does not know whether a sentence is true and may thus characterize the sentence as either declarative and dubitative or as inter- rogative (cf. Sadock & Zwicky, 1985: 169f.). 3 Interrogative sentences The fact that most, if not all, languages have sentences of special structural types for asking questions clearly demonstrates how central this activity is to human communication. These ��interrogative sentences�� are conventionally associated with the speech act of requesting in- formation. Interrogative sentences fall into two major classes depending on their syntactic and semantic properties. It is obvious that the two interrogative sentences given in (25) below have different syntactic structures and are typically used for different types of requests or in- quiries. (25) a. Do you believe in miracles? b. Who discovered America? Sentence (25a) is an instance of a so-called ��polar interrogative sentence�� or ��yes/no- question��, where, quite in contrast to (25b), the expected answer simply consists in providing a truth value for the corresponding declarative sentence. Polar interrogatives are typically used to inquire about the truth or falsity of the proposition they express. It should be borne in mind, however, that answers to polar questions can plausibly assume any value on a scale between ��true�� and ��false��, as e.g. ��perhaps��, ��possibly��, ��quite likely��, etc. ��Constituent inter- rogatives�� of the type exemplified by (25b), also known by the name ��information questions��, receive answers that provide the kind of information specified by the interrogative word (wh- words like who, when, how in English) contained in it, i.e. some expression denoting a human 11

Page 12
being in the case of (25b). Semantically, constituent questions can be analyzed as open propo- sitions with interrogative words signaling the relevant variable positions. As example (26) demonstrates, constituent interrogatives may contain more than one interrogative word. (26) Who did what to whom? Strictly speaking, a clear distinction has to be made between interrogatives as a type of sen- tences or clauses and their semantic counterpart, i.e. questions (cf. Huddleston, 1994). Such a distinction is not consistently found in the literature, however. The formal differentiation between polar interrogatives and constituent interrogatives is a stable cross-linguistic parameter and for each type of interrogative sentence a limited num- ber of recurring, and sometimes overlapping, coding strategies can be identified across the world��s languages (cf. Siemund, 2001). With respect to its meaning, another type of question can be distinguished: (27) Do you prefer beer or wine? In the prototypical case, such ��alternative questions�� are used to ask the addressee to decide which of two or more alternatives holds, i.e. is true or not. Apparently, alternative questions have a lot in common with polar questions, but strictly speaking the answer set is different.
Be that as it may, for the purposes of the present study alternative questions can be neglected since, at least from our current perspective, they do not seem to show any striking typological variation. Nevertheless, alternative questions will become important in the subsequent discus- sion of coding strategies for polar questions. 3.1 Polar interrogatives Essentially six ways of expressing polar questions are encountered across the languages of the world, by and large independently of their genetic affiliation. In decreasing order of their rela- tive and possibly even absolute frequency these are: (i) special intonation patterns, (ii) inter- rogative particles, (iii) the addition of special tags, (iv) disjunctive-negative structures, (v) a change in the relative order of constituents and (vi) particular verbal inflection. By far the most prominent among these six strategies are the use of a special intonation pattern as well as the addition of a particular particle. Interrogative tags are also quite frequent, but typically add some bias with respect to the answer expected. Disjunctive-negative constructions of the type ��A or not A�� are found in some Asian languages. A constituent order different from the one used for declarative sentences is mainly restricted to Indo-European languages. On the whole, these strategies are much less frequent in comparison to the two mentioned at the be- ginning. The encoding of interrogativity by verbal inflection, i.e. an interrogative mood, is extremely rare and mainly restricted to polysynthetic languages. Many languages have more than one strategy – although one of these is usually primary – and typically use intonation in combination with one of the remaining five strategies. 3.1.1 Intonational marking Broadly speaking, the intonation contour used in interrogative sentences is the opposite of the one found in declaratives. While it is typical of declaratives to show falling intonation, the great majority of languages use rising intonation in conjunction with interrogatives. Excep- tions to this generalization are very rare indeed. However, cases of rising intonation in de- claratives and a falling contour in interrogatives do occur and are, for example, reported from Fanti (Niger-Congo, Kwa) and Grebo (Niger-Congo, Kru). The reason for the predominance of rising intonation in interrogatives is usually seen in the fact that high pitch signals uncer- tainty, indecision, hesitation and also insecurity. Low pitch, by contrast, is assumed to convey 12

Page 13
confidence, assurance and certainty (cf. Ohala, 1983; 1994). The intonation contours of de- claratives and interrogatives thus provide good illustration for the principle of iconic motiva- tion. The Italian example in (28) illustrates the standard pattern. (28) Italian a. Suo marito �� ancora \malato. (statement) ��Her husband is still ill.�� b. Suo marito �� ancora /malato? (question) ��Is her husband still ill?�� Of course, the actual shape of the rising contour is not the same in all languages. There is a clear tendency for languages to place the rise towards the end of the contour, but interroga- tives marked by an initial rise do occur also, albeit somewhat sporadically. Moreover, Ultan (1978) reports a number of recurrent shapes which final rises apparently can assume, the most important of these being higher ultima (Vietnamese), higher penult (Chontal, a Hokan lan- guage), higher pitch on last stressed vowel (Bashkir, a Uralian language of the Altaic group) and rising toward last stressed vowel (Hebrew, Semitic). Even though declarative sentences with rising intonation contours can often be analyzed as expressing questions, it is highly doubtful whether they should be regarded as instances of the form type ��interrogative�� (cf. Huddleston, 1994: 428). Several facts argue against such an analysis. First of all, the domain of the rising contour can be wider than that of at least some of the other signals used for the interrogative type and comprise a coordination of clauses (cf. (29a)). On the other hand, the superordinate clause of a complex sentence may be outside the scope of the question proper (cf. (29b)). (29) a. So, Kim went to the meeting but you stayed at home? b. I don��t suppose Jack will contribute to our cause? Secondly, declarative sentences with rising intonation do not license negative polarity items like ever, any, at all, etc., in contrast to clear cases of interrogative sentences: (30) a. Have you ever met him? b. *You have ever met him? Thirdly, declarative sentences with rising intonation, like (31) in English cannot be used as neutral questions in many languages: (31) a. You stole the money? b. He believes in God? Like other declarative sentences, such sentences generally express a commitment. In contrast to declaratives with falling intonation, however, it is a commitment of the addressee and thus an assumption about the answer that is expressed. Therefore such sentences cannot be used as an opening move in a zero context, i.e. in a context where no assumptions about the answer are justified. Finally, rising intonation can be combined with any form type and it has been shown that ris- ing intonation plays a fairly minor role in signaling questions with either inverted or unin- verted structures in English (cf. Geluykens, 1988). If criteria like these have cross-linguistic relevance, the conclusion seems inescapable that languages which use declaratives with rising intonation to express questions do not have the sentence type ��interrogative��. On the other hand, there are languages where ��declarative sentences�� with rising contours meet at least 13

Page 14
some of the criteria generally considered relevant for the identification of genuine interroga- tives (licensing of negative polarity items, use in neutral contexts, etc.). In Russian, for in- stance, sentences without the interrogative marker li and rising intonation meet the relevant criteria so that their categorization as ��interrogatives�� seems justified (cf. Meyer & Zybatow, 2003). 3.1.2 Interrogative particles The addition of interrogative particles to declarative sentences is another way of deriving in- terrogatives. Their precise position is subject to considerable typological variation, but to have interrogative particles in sentence-final position seems the most widely-used option. Inter- rogative particles also occur in constituent interrogatives, but mostly optionally so. The con- trast between declarative and interrogative exemplified by the pair of Japanese sentences in (32) is representative of the situation found in many languages (cf. Hinds, 1984: 158). (32) Japanese a. Yamada-san wa ginkoo de hataraite-imasu. Yamada-Mr. TOP bank at working ��Mr. Yamada works at the bank.�� b. Yamada-san wa ginkoo de hataraite-imasu ka? Yamada-Mr. TOP bank at working
��Does Mr. Yamada work at the bank?�� Further examples of interrogative particles include French est-ce que, Polish czy, Finnish ko/, Mandarin ma, Slavic li, Turkish mi, Indonesian ka, Bengali ki, Kannada e:nu, etc. In Bengali and Kannada the interrogative particle is homonymous with the interrogative word for ��what��. The case of a sentence-initial particle is illustrated by the Persian example in (33), cf. Mahootian (1997: 9). Another such language is Tzotzil (cf. (14b) above). (33) Aya in gorbe-ye šoma-st? [Persian]
this cat-EZ you-is ��Is this your cat?�� For Russian, Turkish and Ute (Uto-Aztecan), but also Latin and Finnish, the relevant inter- rogative particles (li, mi, aa, ne, ko/respectively) should probably be categorized as affixes or at least clitics, because in these languages they are not construed with the entire sentence, but are always attached to a particular constituent. Giv��n (1984: 219f.) shows for Ute -aa that it always occurs after the first constituent and that it is in effect enclitic to it (cf. (34)). (34) Ute a. mam��-ci-aa ��u w��uka-pug��? woman-SUB-INT that.SUB work-REM ��Did the woman work?�� b. k��aw-aa p��ga-kway-kya? yesterday-INT leave-go-ANT ��Did (she) leave yesterday?�� Turkish mi usually occurs in sentence-final position immediately after the predicate. It takes scope over the entire sentence and there are good reasons to assume that it cliticizes onto the predicate (it shows vowel harmony with the stem and is never stressed although word stress in Turkish is word final). Nevertheless, the Turkish interrogative particle may also attach to con- 14

Page 15
stituents within a sentence. This can be observed in focussing constructions, where the scope of the particle is restricted to the relevant constituent (cf. Kornfilt, 1997: 191): (35) kitab- Has��n m Ali-ye ver-di? [Turkish] book-ACC Hasan INT Ali-DAT give-PAST ��Did HASAN give the book to Ali?�� Given that interrogative particles preferably occur adjacent to the predicate, at least in the unmarked case, it should be possible to make certain predictions concerning the position of interrogative particles and the basic word order pattern of a language. A plausible prediction, and this is by and large borne out by the data, would be that verb-final languages mostly have sentence-final particles whereas verb-initial languages tend to have sentence-initial particles (cf. Greenberg, 1966: 81). Nevertheless, there are also some counterexamples to this generali- zation, as the examples from Persian (33) and Ute (34) show (both SOV). Moreover, apart from general concerns about drawing a distinction between basic word order types, another complicating factor for any such generalization is that languages may have more than one interrogative particle the distribution of which may be quite different. There seem to be no preferences for the position of interrogative particles in SVO languages. Thai and Yoruba, for instance, have final particles whereas the particle in Lithuanian is initial, but all belong to the basic type SVO. Korean is one of those languages which have more than one interrogative particle, al- though they all occur at the end of a sentence. The factor governing their distribution is the level of formality (honorification): high style (su)pnikka/(e)yo, mid style so/na, low style e/(nu)nya (cf. also Table 1). The position indicated by ��X�� in (36) below can be occupied by any of these expressions (cf. Chang, 1996: 84). (36) Kui-nun cal cwumwusi-X [Korean] he-TOP well sleep-INT ��Does he sleep well?�� Interrogative particles are often closely related to expressions introducing conditional sub- clauses. As is shown by Russian esli, conditional markers can develop out of interrogative particles (est�� + li) and, in fact, consistently do so in language after language (cf. Traugott, 1985: 291). In Hua (a Papuan language, cf. Haiman, 1978: 570-571), interrogative particle and conditional marker are formally identical (cf. (37)). (37) Hua a. E -si -ve baigu -e come 3.SG.FUT
will stay 1.SG ��If he comes, I will stay.�� b. Fri -si -ve die 3.SG.FUT
��Will she die?�� 3.1.3 Interrogative tags Closely related to interrogative particles are interrogative tags, the main difference between the two kinds of expressions being that tags, apart from characterizing sentences as questions, also contribute a certain bias by raising expectations either towards a positive or a negative answer. This is illustrated by the English examples in (38), where the a-sentence expects a positive answer, but the b-sentence one that is negative.

Page 16
(38) a. You like ice-cream, don��t you? b. You don��t like ice-cream, do you? Although it is a reasonable approximation to the facts to say that negative tags presuppose a positive answer and vice versa, Ultan (1978) points out that the answer induced by a tag ques- tion depends to a greater extent on the polarity of the declarative sentence used for forming the interrogative and less so on the polarity of the tag. Tag interrogatives of type (38a) are clearly the most frequent pattern, but the one that comes second is a combination of affirma- tive sentence, affirmative tag and positive answer. Cases like (38b) belong only to the third most frequent type. Another important difference between interrogative tags and interrogative particles is that tags almost exclusively occur at the end of a sentence, quite independently of the basic word order pattern. One exception to this generalization is the Persian tag mœge (cf. (39)). Moreover, tags are formally not particles, but occur as either (content) words (40), phrases (41) or clauses (42). (39) mœge un mašin-e to nist? [Persian]
that car-EZ you isn��t ��Isn��t that car yours?�� (40) Ty ego slyšal, pravda? [Russian] you him heard true ��You heard him, didn��t you?�� (41) Er ist sehr reich, nicht wahr? [German] he is very rich not true ��He is very rich, isn��t he?�� (42) Ahmet d��n sinema-ya gi-ti, degil mi? [Turkish] Ahmet yesterday cinema.DAT go-PAST, NEG.COP
��Ahmet went to the movies yesterday, didn��t he?�� (lit. ��, isn��t it so?) 3.1.4 Disjunctive-negative structures A completely different strategy of forming polar interrogatives from the ones discussed so far is found in Mandarin Chinese and some other Asian languages as well as in certain languages spoken in Papua New Guinea (Amele, Kobon). What is remarkable about the relevant con- structions is their affinity to alternative interrogatives. The example given in (43) looks like an alternative interrogative in which the conjunction is missing and where the second con- junct is the negation of the first. This type of polar interrogative is also known as the ��A-not-A construction�� (cf. Li & Thompson, 1984). (43) ta z��i jia bu z��i jia? [Mandarin Chinese] 3.SG at home NEG at home ��Is s/he at home?�� The complete structure as shown in (43), however, is not used very frequently and usually replaced by either of the reduced constructions given in (44). 16

Page 17
(44) Mandarin Chinese a. ta z��i bu z��i jia? b. ta z��i jia bu z��i? What is noteworthy about Mandarin Chinese is that the very same construction, when ap- pended to a declarative sentence, serves the function of an interrogative tag. In other words, the resulting questions are not neutral any longer with respect to their expected answer. Nev- ertheless, the set of A-not-A structures used as tags is extremely limited and highly lexical- ized. Apart from du�� bu du�� ��right not right��, as shown in (45), there is only hao bu hao ��good not good��, sh�� bu sh�� ��is not is��, key bu key ��may not may�� and x��ng bu x��ng ��okay not okay�� that occur with significant frequency. (45) zhang-san xhuan he jiu, du�� bu du��? [Mandarin Chinese] Zhang-san like drink wine right NEG right ��Zhang-san likes to drink wine, right?�� 3.1.5 Change in the order of constituents Mostly restricted to Indo-European languages and predominantly to the Germanic branch, by contrast, is the use of a special word order for polar interrogatives. Outside this genetic group, this strategy is very rare. Two non-Indo-European languages known to us that also make use of it are Finnish and Malay. The most common case is to put the finite verb into sentence- initial position while retaining the relative order of the other constituents (cf. (46) – (49)).
(46) Swedish a. Lars läser tidningen. Lars reads the newspaper ��Lars is reading the newspaper.�� b. Läser Lars tidningen? reads Lars the newspaper ��Is Lars reading the newspaper?�� (47) a. She is a translator. b. Is she a translator? (48) a. She can swim. b. Can she swim? (49) a. He knows a translator. b. Does he know a translator? From this follows that inversion of the verb-fronting type is ruled out for VSO-languages. It can only occur in languages whose basic word order type is either SVO or SOV. Greenberg��s (1966: 83) universal 11 states that inversion with polar interrogatives presupposes inversion with constituent interrogatives. 17

Page 18
3.1.6 Verbal inflection Totally different from the strategies discussed so far is the way some polysynthetic languages like West Greenlandic (Kalaallisut), and Eskimo languages (Inuit) in general, encode polar questions (cf. Sadock, 1984: 190). As the minimal pair from West Greenlandic in (50) shows, the interrogative is derived from the declarative by morphological alternation. Additional lan- guages possessing an interrogative mood marked by inflection are Tariano (Arawakan, Bra- zil), languages of the Tucano family spoken in Brazil/Colombia and Blackfoot (an Algon- quian language). (50) West Greenlandic a. neri-vutit eat-IND.2.SG.PAST ��you ate�� b. neri-vit eat-INT.2.SG.PAST ��Did you eat?�� 3.2 Constituent interrogatives With the exception of interrogative tags and disjunctive-negative structures, all the strategies used for deriving polar interrogatives as discussed in the previous section can also be found with constituent interrogatives. However, their occurrence is less wide-spread and their use in many cases optional. Constituent interrogatives, so it seems, can be more readily recognized as questions than polar interrogatives. One notable exception to this general impression is the morphological marking of questions as found in West Greenlandic (cf. Sadock, 1984: 199), where the same marker appears obligatorily in both polar interrogatives and constituent inter- rogatives: (51) West Greenlandic a. neri-va-∅? eat-INT-3.SG ��Did he eat?�� b. su-mik neri-va-∅ what-INST eat-INT-3.SG ��What did he eat?�� However, with respect to the remaining strategies, such a conclusion seems indeed justified. For example, of the 36 languages in Ultan��s (1978) sample for which intonation as a means for marking polar interrogatives is attested (mostly rising intonation or higher pitch), only 12 (or one third) use the same or a similar intonational pattern also for marking constituent inter- rogatives. The overall impression that the data give is that most languages either do not mark constituent interrogatives by intonation at all (33.3%) or do so only optionally (33.3%). No intonational marking of constituent interrogatives is reported from Fula, Japanese and Taga- log, whereas languages like Amharic, English and Turkish at least have optional marking. A similar problem is the occurrence of interrogative particles in constituent interrogatives. From a functional point of view, it appears superfluous to have such particles in constituent inter- rogatives. The interrogative word should unambiguously type the relevant sentences as inter- rogative. What we can observe empirically, however, is that in approximately fifty percent of the languages interrogative particles are optionally added to constituent interrogatives. In the 18

Page 19
Japanese example in (52) below, the particle ka is not obligatory and may be added or simply left out. (52) Dare-ga kimasu (ka)? [Japanese] who-NOM come
��Who is coming?�� Finnish, by contrast, does not allow the interrogative particle ko/in constituent interroga- tives: (53) Kuka(*ko) tulee huomenna? [Finnish] who come.3.SG tomorrow.ESS ��Who is coming tomorrow?�� Attempts have also been made, although with limited success, to correlate the position of in- terrogative words with the position of interrogative particles in polar interrogatives. Baker (1970: 207), based on Greenberg��s (1966) data, hypothesizes that only languages which locate interrogative particles, provided they have such particles, in clause-initial position permit in- terrogative words in positions other than those of the constituents they replace, i.e. the posi- tion of such particles predicts whether a language has wh-movement or not (initial interroga- tive particle ⇒ wh-movement). Contrasting with this view is Cheng (1997: 13ff.) who argues, mainly on the basis of Chinese, that all so-called ��in-situ languages�� possess special particles to mark constituent interrogatives, although this marking may be covert, and all languages with such particles are in situ. Put in a nutshell, the distribution of interrogative particles in constituent interrogatives is clearly governed by certain constraints, but so far it has not been possible to identify them precisely. What seems to be relatively uncontroversial, by compari- son, is the assumption that interrogative tags do not occur in constituent interrogatives. Whether the reordering of constituents should be considered important for the identifi- cation of constituent interrogatives across languages depends to a large extent on the perspec- tive taken. Restricting the scope to cases of subject-verb inversion leaves very few languages as plausible candidates manifesting this phenomenon (mainly the Germanic languages). If, by contrast, we examine the position of interrogative words relative to the position of the con- stituents they substitute for, a much greater range of languages has the property in question. For instance, it is very common for languages, quite independently of their genetic affiliation, to place interrogative words in sentence-initial position. For some languages it is obligatory to do this kind of reordering (English, German, Hebrew, Supyire, Yoruba, Zapotec), others just show a strong tendency to do so (Egyptian Arabic, Kannada, Korean, Palauan). An example of such a fronting language (Finnish) is shown in (54); the example from Swahili in (55) illus- trates constituent interrogatives of optional fronting languages (Haiman, 1985: 245). (54) Finnish a. Maija ottaa omenaa. Maija take.3.SG apple.PAR ��Maija is taking an apple.�� b. Mitä Maija ottaa? what.PAR Maija take.3.SG ��What is Maija taking?�� (55) Swahili a. A-li-fika lini? 19

Page 20
3.SG-PAST-arrive when ��When did s/he arrive?�� b. kwa nini chakula ki-me-chelewa? why food 3.SG-PERF-late ��Why is the food late?�� In Mandarin Chinese, and also in Indonesian, Japanese and Lezgian, interrogative words re- main exactly in the position of the constituent which they replace (so-called ��in-situ lan- guages��). (56) illustrates this point for Mandarin (cf. Cheng, 1997: 5). (56) Mandarin Chinese a. Hufei mai-le y-ben-shu Hufei buy-ASP one-CL-book ��Hufei bought a book.�� b. Hufei mai-le sh��nme? Hufei buy-ASP what ��What did Hufei buy?�� The position of interrogative words depends, to a certain extent at least, on the basic word order type of a language. Greenberg (1966: 82) found a systematic correlation between VSO order and fronted interrogative words as well as between SOV order and the in-situ parame- ter. However, the correlation is much weaker in the case of SOV languages. No such correla- tion can be established for SVO-languages. Another point that is interesting from a cross- linguistic perspective is the behavior of languages when it comes to the co-occurrence of mul- tiple interrogative words. Fronting languages behave surprisingly different with respect to this parameter: Some of them neatly stack interrogative words at the beginning of a sentence while others only front one interrogative word and leave the rest in the positions where they logically belong (i.e. in-situ). As (57) and (58) demonstrate, English fronts only one interroga- tive word leaving additional ones in-situ, whereas Russian, and Slavic languages in general, assemble them at the beginning of a sentence.
(57) a. John gave the book to Mary. b. What did who give to whom? (58) Kto kogo ljubit? [Russian] who whom loves ��Who loves whom?�� 4 Imperative sentences Even a superficial glance at the strategies of imperative formation encountered across the world��s languages makes it clear that variation in this area is as least as extensive as in the case of interrogatives. It is certainly no exaggeration to say that most, if not all, languages have a least one strategy for identifying imperatives, i.e. constructions dedicated to the ex- pression of directive speech acts, i.e. orders and requests, but also invitations, the giving of advice, warnings, wishes, instructions, etc. (cf. Davies, 1986: 30ff.): (59) a. Please clean the bath after use. 20

Page 21
b. Have some more cake. c. Take plenty of exercise if you want to slim. d. Watch out for the dog. e. Sleep well. f. Ring bell for service. In most studies, the label ��imperative�� is reserved for sentence types expressing such speech acts when they are directed to addressees in the narrow sense of the word (second person).
Some authors extend this narrow definition to include commands, request, etc. addressed to the first and sometimes even to the third person (cf. Xrakovskij, 2001), i.e. to cases for which traditionally labels like ��hortatives��, ��optatives��, ��jussives�� and the like would be used. In keeping with the traditional definition, we understand imperatives as sentences with an under- stood second person subject. The most widespread strategy for marking imperatives seems to be a special inflectional form of the verb so that the traditional Western approach of subsuming the imperative under the category ��mood�� appears justified even from a cross-linguistic perspective (cf. van der Auwera & Lejeune, 2003a, 2003b; van der Auwera, Dobrushina & Goussev, 2003). This in- cludes cases of genuine imperative affixes, but also the use of the bare verb stem, special verb stems as well as morphological marking taken over from different domains, as e.g. subjunc- tive, aorist (perfect) and passive forms. A fairly general characteristic of inflecting languages is that morphological marking of the imperative is less extensive in comparison to other moods (indicative, subjunctive, etc.). Agreement affixes for person, number and gender but also those indicating tense, aspect, and the like are frequently suppressed. However, the pic- ture is complicated by the fact that the kind of imperative strategy employed frequently de- pends on the polarity of the sentence. Many languages use one strategy in affirmative sen- tences, but resort to a totally different strategy in negative sentences (so-called ��prohibitives�� or ��vetatives��; cf. Auwera & Lejeune, 2003b). What is also extremely common, if not univer- sal, is the suppression of the subject (pronoun) in imperatives. Although many languages al- low the optional use of a subject pronoun, mainly for contrastive purposes or to soften the force of the imperative (politeness/degree of formality), its absence clearly represents the un- marked case. These general patterns notwithstanding, the range of variation found in the do- main of imperatives is quite extensive. We will give an overview in the following paragraphs, starting with what appear to be the most widespread strategies for positive and negative con- texts. This will be followed by some brief remarks about more indirect means of expressing directive force as well as related constructions. 4.1 Positive imperatives One of the languages in our sample that possesses a true imperative marker, i.e. an affix ex- clusively dedicated to the expression of directive force, is Limbu, a Tibeto-Burman language spoken in Nepal (cf. Driem, 1987: 188). In Limbu the imperative marker  is used in the singular and plural, and always occurs in word-final position (cf. (60)). (60) Limbu a. ips- sleep-IMP ��Sleep!�� b. ips-tch- sleep-2.D-IMP ��Sleep (you two)!�� 21

Page 22
c. ips-amm- sleep-2.PL-IMP ��Sleep (all of you)!�� In many other languages, however, we find imperative markers of a fusional type which pro- vide further information in addition to illocutionary type, usually person and number. The imperative suffixes –a (SG) and –wch (PL) of Welsh are used to issue directive speech acts to a single addressee or to a group, respectively. Another such language is Acholi, a Nilo-Saharan language (cf. Crazzolara, 1955: 110; (61)). (61) Acholi a. lok-i turn-2.SG.IMP ��Turn!�� b. lok-wu turn-2.PL.IMP ��Turn (pl.)!�� In Finnish, by contrast, the imperative marker kaa/kää only appears in the plural (cf. (62)). In the singular, the imperative in Finnish has the same form as the first person singular of the present indicative without the final –n (cf. (63)). (62) sano-kaa [Finnish] talk-2.PL.IMP ��Talk (PL)!�� (63) Finnish a. sano-a talk.INF b. sano-n talk-1.SG.PRES.ACT ��I talk.�� c. sano talk.2.SG.IMP ��Talk (SG)!�� The imperative markers of Evenki, a Tungus language spoken in Siberia, also encode person and number (Nedjalkov, 1997: 18ff.): (64) Purta-va-s min-du bu:-kel [Evenki] knife-ACC.DEF-2.SG.POSS I-DAT give-2.SG.IMP ��Give (you.SG) me your knife!�� The relevant Evenki forms for all combinations of person and number are shown in Table 3. What should be borne in mind, however, is that the forms of the first and third person perform a somewhat different function. Morphological markers expressing commands or requests to the first person are usually called ��hortatives��, those addressing third persons are sometimes referred to as ��optatives�� or ��jussives��. 1.SG baka-kta ��let me find�� 22

Page 23
2.SG baka-kal ��find�� 3.SG baka-gin ��let him/her find�� 1.PL.EXCL baka-kta-vun/baka vvun ��let us find�� 1.PL.INCL baka-gat ��let us find�� 2.PL baka-kallu ��(you PL) find�� 3.PL baka-ktyn ��let them find��
Table 3: The imperative paradigm of Evenki
In addition to the paradigm shown in Table 3, Evenki has another complete paradigm for the encoding of orders, requests, commands, etc. that need not be executed immediately. Thus, what we find grammaticalized in Evenki is the distinction between a near future and a remote future imperative, cf. (65) vs. (66). At the same time, and quite plausibly, the remote forms are used as the polite imperative. (65) D��u-la-vi himat eme-kel [Evenki] home-ALL-POSS quick(ly) come-2.SG.IMP ��Come quickly to my place!�� (66) D��u-la-vi (gochin) eme-de:-vi [Evenki] home-ALL-POSS (next year) come-IMP-POSS ��Come to my place (next year)!�� Quite unexpected, at least from a European perspective, is the case of Macushi, an Amazonian language of the Carib family (Abbott, 1991: 49ff.), where there are special imperative mark- ers indicating motion towards or away from the speaker, cf. the contrast between (67) and (68). A movational imperative is also reported from Chontal (Hokan). (67) apo�� era��ma-ta, ta-��pî-i-ya [Macushi] fire go-IMP.MOT say-PAST-3-ERG �� ��Go get firewood,�� he said.�� (68) tuna era��ma-tane��kî, ta-��pî i-san-ya [Macushi] water get-IMP.MOT say-PAST 3-mother-ERG �� ��Come get the water,�� her mother said.�� What appears also quite common across the world��s languages is to have no special impera- tive marker at all and simply to use the bare verb stem. In many of these cases, however, a plural marker appears with commands directed to a group of addressees. This asymmetry cor- responds to well-established markedness patterns according to which the singular represents the unmarked category (cf. Croft, 1991). Imperative formation of this kind is found in Turkish (Kornfilt, 1997: 41), German, Persian, Punjabi, Shona and many other languages. The rele- vant paradigms for Turkish and German are given in (69). (69) Turkish German a. git-mek komm-en go-INF come-INF b. git komm go.IMP.2.SG come.IMP.2.SG c. gid-in komm-t go.IMP-2.PL come.IMP-2.PL 23

Page 24
In Spanish the imperative verb form of the second person singular is identical to the verb form of the third person singular present indicative. In the plural the suffix –d is added, which yields an unambiguous imperative form (cf. (70)). (70) Spanish a. canta sing.2.SG.IMP (= 3.SG.PRES.IND) b. cantad sing-2.PL.IMP Gulf Arabic is one of the few languages in our sample that draws a gender distinction in im- peratives. This, however, is done in the singular only, again in complete harmony with mark- edness patterns (cf. (71)). Another such language is Abkhaz. (71) Gulf Arabic a. iktib-∅ write-M b. iktib-i write-F c. iktib-u write-PL Special imperative stem forms can be found, inter alia, in German, Lezgian and Welsh, but whereas the relevant German stems are clearly related forms (cf. (72)), we find true supple- tion in Lezgian (cf. (73)). A few imperatives of Lezgian are derived from the stem by conso- nant reduplication. (72) German a. nehmen.INF nimm.IMP ��take�� ��Take!�� b. geben.INF gib.IMP ��give�� ��Give!�� (73) Lezgian a. atu-n.INF ša.IMP ��come�� ��Come!�� b. fi-n.INF alad.IMP ��go�� ��Go!�� Finally, it should be mentioned that many languages do not possess morphological markers dedicated to expressing imperative force. An obvious example is English together with many languages of Southeast Asia (cf. van der Auwera & Lejeune, 2003a). 4.2 Negative imperatives (prohibitives) Negative imperatives deserve special mention since they may be similar to, but also very dif- ferent from their positive counterparts. There are basically four strategies according to which languages encode negative directive speech acts (cf. van der Auwera & Lejeune, 2003b): (i) the use of a positive imperative verb in combination with the negative strategy found in declaratives; 24

Page 25
(ii) the use of a positive imperative verb in combination with a negative strategy not found in declaratives; (iii) the use of a verb other than the positive imperative and the negative strategy found in declaratives; (iv) the use of a verb other than the positive imperative and a negative strategy not found in declaratives; In addition, there is some minor variation on top of these basic distinctions. Some lan- guages even have special prohibitive morphology, i.e. affixes expressing negative directive speech acts without the relevant sentences being overtly negative. The first major strategy introduced above can be illustrated with Turkish and German, where negative commands are expressed by using the verb stem in combination with the negative marker found in declarative sentences (cf. (74)). As pointed out above, only the bare verb stem is used for positive imperatives addressed to one person (cf. (69)). (74) Turkish German git-me komm nicht go-NEG come
��Don��t go (SG)!�� ��Don��t come (SG)!�� Somewhat different, but still comparable to cases like Turkish and German, are languages that require a special auxiliary in negative contexts. The do-periphrasis known from English would be a case in point. What is striking about English is that negative imperatives require do-support even with the verb be, which can be combined with a following negation without do-support in declaratives and interrogatives: (75) Don��t be a fool. Evenki is quite similar to English in this respect (also the Carib language Wai Wai), but adds the imperative suffix to the negative auxiliary e- ��not to�� (cf. (76)). It should be stressed that the strategies of negation illustrated in (75) and (76) represent the normal means of sentence negation and have nothing to do with imperatives proper. (76) Tala e-kel girku-ra [Evenki] there NEG.AUX-2.SG.IMP go-PART ��Don��t go there!�� As for the second major strategy, i.e. the use of a special negative marker together with the positive imperative verb, this can be illustrated with the Vietnamese data in (77), where in negative imperatives the negative element ch�� appears, in contrast to the normal sentence negation không (cf. Thompson 1965: 210, 221). Another such language is Punjabi (Bhatia, 1993: 40). (77) Vietnamese a. Ch�� uông ruou!
drink alcoholic beverages ��Do not drink alcoholic beverages!�� b. Tôi không hieu I
understand ��I do not understand.�� 25

Page 26
A related, but slightly different strategy can be found in Welsh, Finnish and Samoan, where imperatives are negated by means of special negative auxiliaries which only occur in impera- tives (in contrast to English and Evenki above). (78) is an example from Finnish, (79) from Samoan (cf. Mosel & Hovdhaugen, 1992: 482). Note that Finnish marks both the main verb and the (negative) auxiliary as imperative (but only in the plural). The special imperative marker -ko is formally indistinguishable from the interrogative particle. (78) Äl-kää tul-ko [Finnish]
come-IMP ��Don��t come!�� (79) ��Aua e te fa��asaunoa ��i mea-ola [Samoan] don��t 2.SG ASP torture to thing-life.PL ��Don��t torture animals!�� An example of the third major strategy introduced above (negation of declaratives plus verb form different from the positive imperative) is Spanish, where negative commands, requests, etc. are expressed with the relevant verb forms in the subjunctive in combination with the normal sentence negation (cf. (80)). (80) Spanish a. canta cantad sing.2.SG.IMP sing.2.PL.IMP ��Sing!�� ��Sing (all of you)!�� b. no cantes no cant��is
��Don��t sing!�� ��Don��t (you all) sing!�� In Italian we find the infinitive used for negative imperatives, albeit only in the singular: (81) Italian a. canta cantate sing.2.SG.IMP sing.2.PL.IMP ��Sing!�� ��Sing (all of you)!�� b. non cantare non cantate
NEG sing.INF NEG sing.2.PL.IMP
��Don��t sing!�� ��Don��t (you all) sing!�� In the fourth major strategy of forming negative imperatives, finally, a form of the verb other than the one found in positive imperatives is used together with a negative element which does not occur in declarative sentences. Examples of this strategy are Malagasy (cf. Dez 1980: 33, 167), where negative imperatives are formed with the negative element aza (the declarative negation is tsy-) and a verb in the indicative, as well as Kannada (cf. Sridhar 1990: 36), which in negative imperatives requires a verb in the infinitive and a negative auxiliary instead of the normal sentence negators alla or illa (cf. (82) and (83)). (82) Malagasy a. Tongav-a! come-IMP ��Come!�� 26

Page 27
b. Aza mitomany!
cry.IND ��Don��t cry!�� (83) A: ka:De ho:g-a-be:Da! [Kannada] that side go-INF-NEG ��Don��t go that way!�� One of the few languages in our sample that possesses an unequivocal prohibitive marker is Lezgian (Haspelmath, 1993: 149f.). In this language (as well as other Caucasian languages), the imperative marker –a stands in paradigmatic opposition to the prohibitive marker –mir, cf. the contrast between (84a) and (84b). Such prohibitive sentences do not contain sentential negation. (84) Lezgian a. Wuna bagišlamiš-a, buba you.ERG forgive-IMP father ��Forgive me, father!�� b. Wa-z kic��e ze-mir you.DAT afraid be-PROHIB ��Don��t be afraid!�� Although true prohibitive markers appear to be a comparably infrequent phenomenon – there are just four languages in our sample (of about 70 languages, cf. note 3) that have a special morphological marker for the expression of negative directive speech acts (Lezgian, Macushi, Malayalam, Warekena) – the previous discussion has shown that it is relatively common for languages to treat negative imperatives differently from positive imperatives in one way or another. Overall, imperatives tend to preserve archaic forms and exhibit less compositionality than the other sentence types. 4.3 Indirect strategies For a few languages in our sample it has not been possible to identify a construction uniquely dedicated to the expression of directive force. What we also found are languages that in prin- ciple have a true imperative construction, but only very rarely make use of it and are gradually replacing it with a construction taken from a different domain. Such is the case in Modern Hebrew, where verb forms marked for future tense are the normal or unmarked means of ex- pressing directive speech acts (cf. (85)), even though a true morphological strategy is avail- able. The use of the future tense is obligatory for negative commands, since the imperative is not possible in such contexts (Glinert, 1989: 284ff.). Another language well-known for avoid- ing its genuine imperative strategy is English (cf. Thank you for not smoking, Could you ��?). (85) Te- sader [Modern Hebrew]
FUT- tidy
��Tidy!�� That languages use markers from the domain of tense and aspect for the expression of direc- tive force is also attested in other languages. For example, the imperative of Georgian is for- mally indistinguishable from the aorist and in Rapanui, a Polynesian language (du Feu, 1996: 27

Page 28
37ff.), directive force is expressed by normal declarative sentences in the present tense com- bined with a temporal adverb meaning something like ��now�� or ��just��, i.e. a momentary tem- poral unit (cf. (86)). The literal translation into English makes clear that such sentences can indeed be understood as commands: Now you wipe your face. Obviously, the distinction be- tween direct and indirect speech acts is not applicable to languages like Rapanui. (86) Ka amo te ��arina [Rapanui] now clean DET face ��Wipe your face!�� Similar observations can also be made for German, where the perfect participle can be used to express commands of a rather impolite kind: Jetzt aber aufgestanden! ��Get up!�� (lit. now but gotten up). As a matter of fact, it is quite conceivable that temporal and aspectual markers are a source, maybe even a major one, for the grammaticalization of imperatives. Another indirect strategy of imperative formation are subjunctives. The imperative of Lango, a Nilo-Saharan language spoken in Uganda (Noonan, 1992), is formed by dropping the subject agreement affixes from the relevant subjunctive forms. And in Hungarian (cf. Kenesei et al., 1998) as well as in Persian, the imperative marker (-j and be-/bo- respectively) is formally equivalent to the subjunctive marker, cf. the Hungarian examples in (87a) and (87a). (87) Hungarian a. M��sol-j egy kulcs-ot copy-IMP a key-ACC ��Copy (2.sg) a key!�� b. Nem sz��ks��ges, hogy P��ter meg-tanul-j-a a vers-et
necessary that Peter CPL-learn-SUBJ-DEF.3.SG the poem-ACC ��It isn��t necessary for Peter to learn the poem.�� The exact status of the imperative marker in Hungarian is a matter of some debate, but given the core meaning of the subjunctive, it is certainly not unexpected to find its range of uses extended in this way. By comparison, it appears more challenging to explain why the passive can also be used as an indirect strategy for imperatives, as is the case in Maori, a Polynesian language (Bauer, 1993: 32). (88) Patu-a te kurii raa [Maori] beat-PASS the dog DIST ��Beat that dog!�� 4.4 Related constructions Closely related to imperatives, i.e. constructions expressing directive speech acts such as commands, requests, advice, suggestions, invitations, etc., are formal markers frequently re- ferred to as ��hortatives��, ��optatives��, ��debitives��, ��rogatives�� and ��monitories��, which are typi- cally associated with the illocutionary forces given in Table 4. Moreover, there is a difference in person associated with some of these labels: The label ��imperative�� is often restricted to second person directives, whereas ��hortatives�� is found for first and third person directives and ��optatives�� for directions addressed to third persons. Applied in the strict sense of the word, which is usually not done, a language is said to possess any of these categories if it has special verbal morphology exclusively dedicated to the expression of the relevant illocution- ary functions. Such inflectional markers do indeed exist in many languages other than Eng- lish. In principle, one could imagine a language with full morphological paradigms for each of 28

Page 29
these functions, maybe even inflecting for person, number, etc. In practice, this happens very rarely, if at all. At least, we are aware of no such language. What does happen is that lan- guages pick a seemingly idiosyncratic mix of forms with often overlapping functions. An ad- ditional complicating factor is that the usage of these terms is not completely fixed and varies from author to author.
Category Illocutionary Force hortatives exhortations optatives wishes debitives obligations rogatives petitions monitories warnings
Table 4: Subcategories of imperatives
Of course, apart from morphological marking, there are always alternative means of express- ing the relevant illocutions. The English construction for the expression of exhortations, i.e. commands to the first person or a group of people including the speaker, is let��s, as in Let��s go to the movies. Wishes, obligations and warnings are either expressed by explicit performatives (I wish ��; I warn ��) or by modal verbs (You must leave now). True morphological hortatives are, inter alia, found in Lezgian (89), Evenki (90) and French (91).
(89) Sifte wun wi buba.di-z qalur-in [Lezgian] first you.ABS you.GEN father-DAT show-HORT (=1.SG.IMP) ��First let me show you to your father!�� (90) Bi oro-r-vi baka-kta [Evenki] I reindeer-PL-POSS find-HORT (=1.SG.IMP) ��Let me find my reindeer.�� / ��I��ll go and find my reindeer.�� (91) French a. chant sing.2.SG.IMP b. chantez sing.2.PL.IMP c. chantons sing.1.PL.IMP (=HORT) ��Let��s sing.�� Examples of languages with optatives, i.e. morphological markers expressing wishes (or third person directives), are Malayalam, Lezgian, Evenki, Greek, Turkish. An example from Maya- layam is shown in (92), and from Lezgian in (93). (92) avar samsaarikk-ae [Mayalayam] they speak-OPT ��Let them speak.�� 29

Page 30
(93) Quj wun ci C��kwer.a-z wax xu-raj [Lezgian] let you.ABS we.GEN C��kwer-DAT sister be-OPT ��May you be a sister for our C��kwer.�� Among the languages in our sample possessing debitives, i.e. verbal morphology reserved for the expression of obligation, are Malayalam (cf. (94)) and Evenki (cf. (95)). The relevant markers –aam and –mechin inflect neither for person nor for number. (94) nia naae tanne var-aam [Mayalayam] you tomorrow EMPH come-DEBIT ��You must come precisely tomorrow�� (95) Minngi girki-v ilan-duli chas-tuli suru-mechin-in [Evenki] my friend-1.SG.POSS three.PROL hour.PROL go.away-DEBIT-3.SG ��My friend must leave in three hours.�� The decision whether a language has a rogative (for petitions) or not depends to a large extent on the definition of this category. If it is defined broadly so as to comprise polite imperatives, then a not insubstantial number of languages will qualify for inclusion. We have already men- tioned that Evenki has two complete imperative paradigms, one expressing categori- cal/immediate commands, the other one being reserved for those that are polite/remote. Simi- larly, the polite imperative –wat�� of Chontal (Hokan) could plausibly be called a rogative: (96) n��y-wat�� [Chontal] ��Please, let me know.�� The final subcategory of imperatives to be discussed in this section is the category of ��moni- tory��, i.e. verbal inflection used for the expression of warnings. One of the few languages pos- sessing a true monitory is again Evenki (cf. (97)). Of course, there are numerous indirect strategies for expressing the same illocutionary function. For instance, warnings in Malaya- lam can be expressed by combining the imperative marker with a negative tag. (97) Er-tyki, tar-tyki iche-t-ne [Evenki] this-ALL that-ALL see-CONT-MONIT ��(Be careful and) look in different directions.�� It is an intriguing question whether there are any systematic relations between the subcatego- ries of the imperative and also between these subcategories and the imperative itself. A plau- sible assumption is that the distribution of prohibitives, hortatives, optatives, etc. is not totally random but subject to implicational generalizations of the kind that if a language has category X, it will also have category Y. In the ideal case, the categories under consideration here would permit a ranking on an implicational hierarchy. Unfortunately, only a minority of lan- guages in our sample draw the relevant morphological distinctions so that the formulation of any such strong hypothesis cannot even be attempted. Nevertheless, what our data clearly show is that some of these categories are very likely to occur together – although it is not clear in which order – and that there are also some implicational connections. For instance, Macushi and Malayalam have four of these categories, Lezgian, Turkish, Evenki and Wai Wai have three and Georgian has special morphological markers for at least two of them. Moreover, the existence of any of these subcategories implies the existence of a true (morpho- logical) imperative in a language (cf. van der Auwera, Dobrushina & Goussev, to appear). 30

Page 31
5 Some minor sentence types 5.1 Exclamatives Among the minor sentence types that can be distinguished across languages in addition to the three major ones and their subtypes, exclamatives are the most prominent one. In terms of the typology of speech acts mentioned above, exclamations, the semantic counterpart of so-called ��exclamative sentences��, are used for the performance of representa- tive speech acts, i.e. for speech acts expressing a state of belief and making a claim about the world. But in contrast to assertions, the point of an exclamation is not really to inform the hearer(s) about some situation, but to express an affective response to what is taken to be a fact. More specifically, exclamations convey the speaker��s surprise that some present situa- tion is remarkable and thus seem to be used as expressive speech acts of a type not included into Searle��s typology. Finally, exclamations relate to a scale or dimension and identify an extreme value. Exclamations can be expressed by a wide variety of formal structures and constructions (cf. Rosengren, 1992b; Michaelis, 2001). In English, declarative sentences, interrogative sen- tences, free relatives, isolated NPs, inversion, subordination to factive epistemic verbs may be used, inter alia, for that purpose:
(98) a. He is so stupid/such an idiot! b. Isn��t this great! c. How foolish he is! d. The speed they drive on the freeway! e. Man, is this kid intelligent! f. I can��t believe how much he has grown! A similar variety of structures can be found in many other languages (cf. Michaelis, 2001). In addition to the structures mentioned for English, declarative sentences introduced by the complementizer dass ��that�� can be used in German, as well as all free relatives introduced by an interrogative pronoun: (99) German a. Dass der immer nur Tennissocken trägt! that he always only tennis socks wears ��It is incredible that he always wears tennis socks.�� b. Wen die alles eingeladen haben! whom they all invited have ��The people they invited!�� Given this variety of structures and constructions that can be used to express exclamations, none of which can easily be dismissed as being an example of an indirect speech act, it is, of course, highly problematic to list exclamatives as a fourth basic sentence type alongside de- claratives, interrogatives and imperatives. The only common denominator of all these struc- tures seems to be intonation: Exclamative constructions are generally characterized by a fal- ling intonation contour and a focus on either the basic argument or the (scalar) predicate or on both. In view of these facts it seems justified to exclude exclamatives from the list of basic sentence types (cf. Rosengren, 1992b: 265f.). So-called ��exclamative sentences�� could simply be regarded as being the result of combining declarative or interrogative sentences with spe- cific syntactic, semantic and pragmatic properties, all of which are highly suitable and thus motivated for the expression of an exclamation. Zanuttini & Portner (2003) thus restrict the term ��exclamative�� to those sentences that contain an interrogative word and are factive. A 31

Page 32
different view is provided in one of the very few typological studies of exclamative construc- tions (cf. Michaelis, 2001). Michaelis regards exclamatives as constructions, i.e. as a set of semantico-pragmatic features, all of which must receive formal expression. 5.2 Echo questions Somewhat different from their central use is the use of questions for the purpose of seeking clarification on a preceding utterance. Although both polar interrogatives and constituent in- terrogatives easily lend themselves to this purpose, echoing something previously uttered in the form of a question often precludes the use of the standard interrogative constructions and necessitates a more indirect way of asking questions. In English, for instance, polar echo questions show a strong tendency to be phrased in terms of a declarative sentence with a ris- ing intonation. Moreover, using constituent questions as echoes means that interrogative words can and frequently are left in-situ: (100) A: I tell you he is a braggart. B: He is a braggart? / He is what? As a matter of fact, more careful analyses of echo questions have revealed that their discus- sion in the context of interrogative constructions and their frequent incorporation into this domain is problematic. One of the major problems for such an apparently straightforward approach is that all the three basic sentence types can be used as echo questions. The exam- ples given in (101) show declarative, interrogative and imperative sentences being used as echo questions. The only property that these echo questions have in common with interroga- tive sentences is that they contain an interrogative word (constituent echo questions at least). These and similar facts have led some authors to conclude that echo questions cannot be re- garded as instantiating a special sentence type, but are a phenomenon that can be superim- posed on any of the basic sentence types (cf. the discussion in Reis, 1992; Huddleston, 1994). (101) a. A: John lives in Paris. – B: He lives where? b. A: John lives in Paris. – B: Where does he live? c. A: Go to Paris! – B: Go where? Another point of some debate is whether echo questions are used to perform independent speech acts or whether they merely quote the preceding utterance. What, among other things, argues in favor of a quotational analysis is that echo questions do not represent the point of view of the speaker. The evaluation of a certain person��s mental capacity expressed in (102B) is evidently not that of the speaker. (102) A: I have met this idiot again. B: You have met which idiot again? Moreover, such an analysis is corroborated by languages which regularly use a quotative con- struction to express echo questions, as e.g. Turkish (cf. (103)). The quotative verb is not op- tional in this example (cf. Kornfilt, 1997: 32). (103) Turkish A: Sinema-ya gid-iyor-um cinema-DAT go-PRES.PROG-1.SG ��I am going to the movies.�� 32

Page 33
B: Sinema-ya gid-iyor-um mu de-di-n where-DAT go-PRES.PROG-1.SG
say-PAST-2.SG ��Did you say ��I am going to the movies��?�� In all the languages we surveyed, echo questions may be reduced, and in fact are often re- duced, to the constituent or the constituents on which clarification is required. In informal speech it is probably even more common simply to use the interrogative word for inanimate referents (Engl. what). Somewhat different from the examples of echo questions just dis- cussed are those that are used as a reply to a preceding question. In German and English, at least, such echo questions have the structure of embedded clauses and cannot be analyzed as quotations. Consider the example of a polar echo question in (104B) and the one of a con- stituent echo question in (105B). (104) German A: Hast du schon eingekauft? ��Have you done the shopping yet?�� B: (Du fragst, ) ob ich schon eingekauft habe? ��(You are asking) if I have done the shopping?�� (105) German A: Wo bist du gewesen? ��Where have you been?�� B: (Du fragst, ) wo ich gewesen bin? ��(You are asking) where I have been?�� 5.3 Non-finite presentatives Another minor sentence type that can be identified across languages are non-finite construc- tions that are neither declarative, interrogative or imperative and that are used to present a proposition with the purpose of rejecting it as absurd: (106) a. Him play the piano. Ludicrous! (English) b. Der und Klavier spielen. Lächerlich! (German) Constructions like (106) have been discussed under such labels as ��MAD magazine sentences��, ��left dislocations of argument and predicate��, ��sentential topics��, ��citations��, etc. The most interesting point about such constructions is that they are not tokens of one of the three major sentence types and could be analyzed as instantiating a ��presentative mode�� (cf. Zaefferer, 1990: 223f.). It is probably due to the marginal role of such constructions in communication that they have not been accorded that status. 5.4 Answers to questions Among the numerous sentence fragments that are used across languages, answers are easy to identify and also lend themselves to some noteworthy cross-linguistic generalizations. An- swers are declarative sentences with a specific focus marking. More often than not they are relatively short, reduced to their focus, and normally do not give more than the requested in- formation, i.e. a truth value in the case of polar questions and the information specified by the interrogative word(s) in the case of constituent questions (cf. (107)). Of course, more elabo- rate responses can also be encountered. 33

Page 34
(107) a. A: Are you leaving. – B: Yes (I am). b. A: Where are you going? – B: (I am going) to Paris. / Well, what do you think? There seem to be three different answering systems for polar interrogatives: (i) yes/no sys- tems, (ii) agree/disagree systems and (iii) echo systems, the main properties of which can be described as follows. In yes/no systems of the type employed in English, German, Turkish, etc. confirmation of the proposition expressed by the relevant question is indicated by supply- ing an answer of the same polarity whereas the polarity of the answer is opposite to the one of the question in case of non-confirmation. This applies to both positive and negative questions. (108) a. A: Did he bring a present? – B: Yes. (confirmation) b. A: Did he bring a present? – B: No. (non-confirmation) (109) a. A: Did he not bring a present? – B: Yes, he did. (non-confirmation) b. A: Did he not bring a present? – B: No. (confirmation) There is no difference between yes/no systems and agree/disagree systems as far as questions of positive polarity are concerned. Where the two systems diverge is in negative contexts, the confirmation/non-confirmation pattern of the respective answers being exactly reversed. The example in (110) simulates an agree/disagree system on the basis of English. (110) a. A: He did not bring a present, right? – B: Right. (confirmation) b. A: He did not bring a present, right? – B: Wrong. (non-confirmation) Languages possessing agree/disagree systems in our sample include Gulf Arabic, Japanese, Malayalam and Punjabi among others. For instance, in Gulf Arabic naam is the agreement particle whereas bala is used to express disagreement (cf. (111)). (111) Gulf Arabic A: maa indik fluus, muu chidhi
with you money
like-that ��You haven��t any money, right?�� B: naam ��It is true that I have no money.�� bala ��It is not true that I have no money, i.e. I have money.�� No special answer words at all can be found in the third type of answering system, i.e. the echo system, which works by using part of the question – usually the verb – as the answer. Welsh and Finnish are among the language in our sample possessing such an echo system (cf. the Welsh example in (112)). A special negative element (na(c)) is used in case the question cannot be answered affirmatively. (112) Welsh A: A welwch chwi hwy?
see you them ��Do you see them?�� B: Gwelaf ��(Yes) I see (them).�� see Na welaf ��(No) I don��t see (them).��
see 34

Page 35
One well-known problem of yes/no systems is that positive answers to negative questions can be confusing whenever the answer expected is biased towards an affirmation. As the example in (113) illustrates, answering such questions simply by ��yes�� leaves open whether what was meant is Yes, he did or Yes, he didn��t. (113) A: He didn��t bring a present, did he? – B: Yes. Evidently, this ambiguity arises because yes is mistakenly interpreted as a marker indicating confirmation, i.e. as part of an agree/disagree system. In order to make it unambiguously clear that a positive answer to a negative question is intended and that the expectations raised by the question are wrong, many languages with yes/no systems offer a third answering strategy besides ��yes�� and ��no��. For German this is doch (as opposed to ja), for French si (instead of oui), for Tigrinya bba (rather than wwa) and in English one can use a tag answer (e.g. Yes, he did) to achieve the desired effect. 6 Summary and conclusion The preceding discussion has shown that the communicative potential of a sentence, i.e. the potential for performing actions (speech acts) of various kinds, is consistently and pervasively encoded in the grammar of languages, even if not typically in terms of paradigmatic opposi- tions. In particular the distinction between three basic sentence types, declarative, interroga- tive and imperative, is overtly drawn in most, even if not in all, languages. So-called ��excla- mative sentences��, by contrast, do not seem to constitute a separate basic sentence type, but can simply be analyzed as the result of combining declarative or interrogative sentences with specific syntactic, semantic and pragmatic properties. Cross-linguistically, the three major sentence types are characterized by a limited set of recurrent strategies. Among these the im- perative exhibits the highest degree of further differentiation, a fact, which could find an ex- planation in the interactional risks associated with directive speech acts. Formal differentia- tion between imperatives and prohibitives, adhortatives, optatives, rogatives, debitives, etc. is a fairly wide-spread phenomenon among the languages of the world. As far as interrogative sentences are concerned, most languages seem to distinguish polar interrogatives from con- stituent (wh-) interrogatives and also the unmarked use of both from the echoic use (��echo questions��). Declarative sentences are often identified by formal markers that conflate indica- tive or declarative mood with modal notions, such as evidentiality and strength of assertion. Note, however, that these basic sentence types are compatible with a wide variety of specific uses or speech acts. So, strictly speaking, what we find in the grammar of a language are gen- eral distinctions of sentence types, semantic mood or illocutionary potential, rather than ��speech act distinctions��. Our finding, as those of others before us, show that the three major sentence types tradi- tionally distinguished for European languages can also be clearly identified in a wide variety of other languages and that further differentiations are typically based on these primary form types. Our findings therefore suggest that the distinction between declarative, interrogative and imperative can and ought to play an important role in grammatical theory and in the analysis of the interface between grammar and pragmatics. What our survey also reveals is the fact that the semantic analysis of these categories must be a very abstract one. It is only as a result of the interaction of these basic sentence types with a variety of other formal, semantic and contextual properties that an utterance has a specific use or function in a context. Some of these interacting features, such as intonation for instance, seem to function very similarly across languages. Others seem to be language-specific (as for instance modal particles in German). And such combinations of basic sentence types with bundles of language-specific 35

Page 36
features may develop into constructions whose use potential can no longer be derived from the interaction of all features in a compositional fashion. Suggestions for further reading There are very few comprehensive cross-linguistic studies on the form and function of sen- tence types. Apart from the present article there is only its counterpart in the 1985 edition (Sadock & Zwicky, 1985) and Palmer��s (very readable) volume on mood and modality (1986). Some methodological problems relating to the identification of sentence types are discussed in Croft (1994). Moreover, there are a few studies on specific sentence types (as well as related phenom- ena) across different languages, although the number of studies available is surprisingly low. Ultan (1978) is the classic typological article on interrogatives. Chisholm et al. (1984) con- tains useful descriptions of interrogative constructions from various languages (Japanese, Russian, Ute, etc.). Cheng (1997) offers a cross-linguistic study of wh-question in the frame- work of government and binding. A recent summary of the cross-linguistic properties of inter- rogative constructions can be found in Siemund (2001). The main source for imperatives and related constructions (hortatives, optatives, etc.) are Xrakovskij (2001), van der Auwera & Lejeune (2003a, 2003b) as well as van der Auwera, Dobrushina & Goussev (2003). A new major cross-linguistic study of imperatives is Aikhenvald (2003). Cross-linguistic aspects of exclamative constructions are discussed in Michaelis (2001). Declaratives are usually dis- cussed in connection with markers of evidentiality: Chafe & Nichols (1986), Johanson & Utas (2000). As for descriptions of sentence types of individual languages, the single best investi- gated language is probably German, with the two extensive volumes edited by Rosengren (1992, 1993) still forming the major point of orientation. More recent discussions of German sentence types can be found in Reis (1999) and Lohnstein (2000). For English interrogatives and questions useful starting points are Pope (1976) and Huddleston (1994). Davies (1986) provides detailed information on English imperatives. Comprehensive descriptions, however, are only available for relatively few languages (even among the well studied languages) so that an approach via reference grammars is mostly inevitable. Theoretical problems pertaining to the form-function-relationship of sentence types are discussed inter alia in Sadock (1974), Searle et al. (1980), Levinson (1983), Zaefferer (1990), Tsohatzidis (1994). The classic (still widely discussed ) studies on speech acts and the func- tion of utterances are Austin (1962) and Searle (1969) as well as Searle��s subsequent publica- tions. References Abbott, Miriam. 1991. ��Macushi��. In: Derbyshire, Desmond C. & Pullum, Geoffrey K. (eds.). Handbook of Amazonian Languages. Berlin: Mouton de Gruyter, vol. III. Aikhenvald, Alexandra Y. 2003. Imperatives and other Commands. Unpl. Ms. La Trobe Uni- versity. Altmann, H. 1984. ��Zur Problematik der Konstitution von Satzmodi als Formtypen��. In: Mei- bauer, Jörg (ed.). Satzmodus zwischen Grammatik und Pragmatik. T��bingen: Niemeyer, 22-56. Altmann, H. 1993. ��Satzmodi��. In. Jacobs, Joachim et al. (eds.). Syntax. Ein internationales Handbuch zeitgenössischer Forschung. Berlin: de Gruyter, 1006-1029. Akmajian, A. 1984. ��Sentence types and the form-function fit��. Natural Language and Lin- guistic Theory 2, 1-24. 36

Page 37
Austin, John. 1962. How to do Things with Words. Oxford: Clarendon. van der Auwera, Johan & Lejeune, Ludo. 2003a. ��The morphological imperative��. In: Dryer, Matthew; Haspelmath, Martin; Gil, David & Comrie, Bernard (eds.). World Atlas of Language Structures. Oxford: Oxford University Press. van der Auwera, Johan & Lejeune, Ludo. 2003b. ��The prohibitive��. In: Dryer, Matthew; Haspelmath, Martin; Gil, David & Comrie, Bernard (eds.). World Atlas of Language Structures. Oxford: Oxford University Press. van der Auwera, Johan; Dobrushina, Nina & Goussev, Valentin. 2003. ��Imperative-hortative systems��. In: Dryer, Matthew; Haspelmath, Martin; Gil, David & Comrie, Bernard (eds.). World Atlas of Language Structures. Oxford: Oxford University Press. van der Auwera, Johan; Dobrushina, Nina & Goussev, Valentin. To appear. ��A semantic map for imperative-hortatives��. In: Willems, D.; Colleman, T. & Defrancq, B. (eds.) Points of Comparison in Linguistics: From Morphology to Discourse. Basingstoke: Palgrave Macmillan. Baker, Carl Lee. 1970. ��Notes on the description of English questions��. Foundations of Lan- guage 6, 197-219. Barnes, J. 1984. ��Evidentials in the Tuyuca verb��. International Journal of American Linguis- tics 50, 255-71. Bauer, Winifred. 1993. Maori (Descriptive Grammars). London: Routledge. Besnier, Nico. 2000. Tuvaluan. A Polynesian Language of the Central Pacific. London: Routledge. Bhatia, Tej K. 1993. Punjabi (Descriptive Grammars). London: Routledge. Chafe, Wallace & Nichols, Johanna (eds.). 1986. Evidentiality: The Linguistic Coding of Epistemology. Norwood, New Jersey: Ablex Publishing Corporation. Chang, Suk-Jin. 1996. Korean. Amsterdam: Benjamins. Cheng, Lisa Lai-Shen. 1997. On the Typology of wh-Questions. New York: Garland. Chisholm, William S. & Milic, Louis T. & Greppin, John A.C. (eds.). 1984. Interrogativity: A Colloquium on the Grammar, Typology and Pragmatics of Questions in Seven Diverse Languages. Amsterdam: Benjamins. Clark, Herbert H. 1996. Using Language. Cambridge: Cambridge University Press. Comrie, Bernard. 1981. Language Universals and Linguistic Typology. Oxford: Blackwell. Comrie, Bernard & Polinsky, Maria. 2002. ��On the grammar of evidentiality. Witnessed and unwitnessed verb forms in Tsez��. In: Guentcheva, Zlata (ed.) Studies in Evidential Con- structions. Amsterdam: Benjamins. Crazzolara, P. (1955) A Study of the Acooli Language. Oxford: Oxford University Press. Croft, William. 1991. Typology and Universals. Cambridge: Cambridge University Press. Croft, William. 1994. ��Speech act classification, language typology and cognition��. In: Tso- hatzidis, Svas L. (ed.), 460-477. Davies, Eirlys. 1986. The English Imperative. London: Croom Helm. DeLancey, Scott. 1997. ��Mirativity: The grammatical marking of unexpected information��. Linguistic Typology 1, 33-52. Dez, Jacques. 1980. La syntaxe du malgache. Lille, France: Atelier de reproduction des th��- ses. Driem, George van. 1987. A Grammar of Dumi (Mouton Grammar Library, No 10). Berlin: Mouton de Gruyter. du Feu, Veronica. 1996. Rapanui (Descriptive Grammars). London: Routledge. Fortescue, Michael. 1984. West Greenlandic. London: Croom Helm. Geluykens, R. 1988. ��On the myth of rising intonation in polar questions��. Journal of Prag- matics 12, 467-485. Giv��n, Talmy. 1984. ��Ute��. In: Chisholm, William S. et al. (eds.). 37

Page 38
Glinert, Lewis. 1989. The Grammar of Modern Hebrew. Cambridge: Cambridge University Press. Greenberg, Joseph H. 1966. ��Some universals of grammar with particular reference to the order of meaningful elements��. In: Greenberg, Joseph H. (ed.). Universals of Language. Cambridge, Mass: MIT Press, 73-113. Gunlogson, Christine. 2001. True to Form. Rising and Falling Declaratives as Questions in English. Ph.D. Dissertation, University of California at Santa Cruz. Hagman, Roy S. 1977. Nama Hottentot Grammar. Bloomington: Indiana University Press. Haiman, John. 1978. ��Conditionals are topics��. Language 54, 564-589. Hamblin, Charles L. 1987. Imperatives. New York: Blackwell. Hardman, Martha J. 1986. ��Data-source marking in the Jaqi languages��. In: Chafe, Wallace & Nichols, Johanna (eds.). Haspelmath, Martin. 1993. A Grammar of Lezgian (Mouton Grammar Library, No 9). Berlin: Mouton de Gruyter. Hausser, Roland. 1983. ��The syntax and semantics of English mood��. In: Kiefer, Ferenc (ed.). Questions and Answers. Dordrecht: Reidel. Hinds, John. 1984. ��Japanese��. In: Chisholm, William S. et al. (eds.). Huddleston, Rodney. 1994. ��The contrast between interrogatives and questions��. Journal of Linguistics 30, 411-439. Johanson, Lars & Utas, Bo (eds.) 2000. Evidentials. Berlin: Walter de Gruyter. Kenesei, Istv��n; Vago, Robert M. & Fenyvesi, Anna. 1998. Hungarian (Descriptive Gram- mars). London: Routledge. Kay, Paul & Fillmore, Charles J. 1999. ��Grammatical constructions and linguistic generaliza- tions: The What��s X doing Y? construction��. Language 75, 1-34. Kornfilt, Jaklin. 1997. Turkish (Descriptive Grammars). London: Routledge. Levinson, Stephen. 1983. Pragmatics. Cambridge: Cambridge University Press. Li, Charles N. & Thompson, Sandra A. 1984. ��Mandarin��. In: Chisholm, William S. et al. (eds.). Lohnstein, Horst. 2000. Satzmodus – kompositionell. Zur Parametrisierung der Modusphrase im Deutschen. Berlin: Akademie-Verlag. Mahootian, Sharhzad. 1997. Persian (Descriptive Grammars). London: Routledge. Matthews, George H. 1965. Hidatsa Syntax. The Hague: Mouton. Meyer, Roland & Zybatow, Gerhild. 2003. ��Zur Satztypologie russischer Entscheidungsfra- gen: Satztypfestigung durch Intonation?�� Paper given at the 25th meeting of the Deut- sche Gesellschaft f��r Sprachwissenschaft, Munich 2003. Michaelis, Laura. 2001. ��Exclamative constructions��. In: Haspelmath, Martin; König, Ekke- hard; Oesterreicher, Wulf & Raible, Wolfgang (eds.). Language Typology and Lan- guage Universals. Berlin: de Gruyter. Michailovsky, Boyd. 1988. La langue hayu. Paris: Editions CNRS. Mosel, Ulrike & Hovdhaugen, Even. 1992. Samoan Reference Grammar. Oslo: Scandinavian University Press. Moyse-Faurie, Claire. Grammaire du futunien. Collection Universit��. Nedjalkov, Igor. 1997. Evenki (Descriptive Grammars). London: Routledge. Noonan, Michael P. 1992. A Grammar of Lango (Mouton Grammar Library, No 7). Berlin: Mouton de Gruyter. Ohala, John J. 1983. ��Cross-language use of pitch: an ethological view��. Phonetica 40, 1-18. Ohala, John J. 1994. ��The frequency code underlies the sound-symbolic use of voice pitch��. In: Hinton, Leanne; Nichols, Johanna & Ohala, John J. (eds.). Sound Symbolism. Cam- bridge: Cambridge University Press, 325-347. 38

Page 39
Önnerfors, Olaf. 1997. Verb-erst-Deklarativsätze. Stockholm: Almquist & Wiksel Interna- tional. Palmer, Frank. 1986. Mood and Modality. Cambridge: Cambridge University Press. Pope, Emily N. 1976. Questions and Answers in English. The Hague: Mouton. Reis, Marga. 1992. ��Zur Grammatik und Pragmatik von Echo-w-Fragen��. In: Rosengren, In- ger (ed.). 1992a. Reis, Marga. 1999. ��On sentence types in German: An enquiry into the relationship between grammar and pragmatics��. Interdisciplinary Journal for Germanic Linguistics and Se- miotic Analysis 4:2, 195-236. Rosengren, Inger. (ed.). 1992a. Satz und Illokution. T��bingen: Niemeyer, LA 278, vol. I. Rosengren, Inger. 1992b. ��Zur Grammatik und Pragmatik der Exklamation��. In: Rosengren, Inger (ed.). 1992a. Rosengren, Inger. (ed.). 1993. Satz und Illokution. T��bingen: Niemeyer, vol. II. Sadock, Jerrold M. 1974. Toward a Linguistic Theory of Speech Acts. New York: Academic Press. Sadock, Jerrold M. 1984. ��West Greenlandic��. In: Chisholm, William S. et al. (eds.). Sadock, Jerrold M. 1994. ��Toward a grammatically realistic typology of speech acts��. In: Tsohatzidis, Svas L. (ed.), 393-406. Sadock, Jerrold M. & Zwicky, Arnold M. 1985. ��Speech act distinctions in syntax��. In: Shopen, Timothy (ed.). Language Typology and Syntactic Description. Cambridge: Cambridge University Press. Sbis��, Marina. 1995. ��Speech act theory��. In: Verschueren, Jef; Östman, Jan-Ola; Blommaert, Jan (eds.). Handbook of Pragmatics. Amsterdam: Benjamins. Searle, John. 1969. Speech Acts. Cambridge: Cambridge University Press. Searle, John. 1976. ��The classification of illocutionary acts��. Language in Society 5, 1-24. Searle, John; Kiefer, Ferenc & Bierwisch, Manfred. 1980. Speach Act Theory and Pragmat- ics. Dordrecht: Reidel. Siemund, Peter. 2001. ��Interrogative constructions��. In: Haspelmath, Martin; König, Ekke- hard; Oesterreicher, Wulf & Raible, Wolfgang (eds.). Language Typology and Lan- guage Universals. Berlin: de Gruyter. Sridhar, S.N. 1990. Kannada (Descriptive Grammars). London: Routledge. Thompson, Laurence C. 1965. A Vietnamese Grammar. Seattle: University of Washington Press. Traugott, Elizabeth C. 1985. ��Conditional markers��. In: Haiman, John (ed.). Iconicity in Syn- tax. Amsterdam: Benjamins. Tsohatzidis, Svas L. (ed.). 1994. Foundations of Speech Acts Theory. Philosophical and Lin- guistics Perspectives. London: Routledge. Ultan, Russell. 1978. ��Some general characteristics of interrogative systems��. In: Greenberg, Joseph H. (ed.). Universal of Human Language. Stanford, CA: Stanford University Press. Wierzbicka, Anna. 1987. English Speech Act Verbs. Orlando: Academic Press. Wunderlich, Dieter. 1976. Studien zur Sprechakttheorie. Frankfurt/M.: Suhrkamp. Wunderlich, Dieter. 1991. ��Bedeutung und Gebrauch��. In: von Stechow, Arnim & Wunder- lich, Dieter (eds.). Semantik. Ein internationales Handbuch zeitgenössischer Forschung. Berlin: de Gruyter, 32-52. Zaefferer, Dietmar. 1990. ��On the coding of sentential modality��. In: Bechert, Johannes et al. (eds.). Towards a Typology of European Languages. Berlin: Mouton de Gruyter. Zanuttini, Raffaella & Portner, Paul (2003) ��Exclamative clauses: At the syntax-semantics interface��. Language 79:1, 39-81. 39

Page 40
Xrakovskij, Viktor S. 2001. ��Hortative constructions��. In: Haspelmath, Martin; König, Ekke- hard; Oesterreicher, Wulf & Raible, Wolfgang (eds.). Language Typology and Lan- guage Universals. Berlin: de Gruyter. Abbreviations
direct object
indirect object
object marker
past tense
verb 1 first person 2 second person 3 third person Notes
1 The financial support of the Deutsche Forschungsgemeinschaft (SFB 447/B8), the Max-Planck-Gesellschaft
and the Alexander von Humboldt-Stiftung is gratefully acknowledged. We would also like to thank Manfred Krifka, Claire Moyse-Faurie, Timothy Shopen and Anne Zribi-Hertz for constructive criticism and valuable suggestions.
2 Note that these three sentences types are also often distinguished in the orthography by different punctuation
marks (��.��, ��?��, ��!��).

Page 41
33 Apart from the language material found in the relevant literature, we have based our study on the following
language sample (convenience sample): Araona, Chontal (of Oaxaca), Dumi, English, Evenki, Finnish, French, Futunian, Georgian, German, (Modern) Greek, Gulf Arabic, Hayu, Hebrew, Hidatsa, Hua, Hungarian, Indone- sian, Italian, Japanese, Korean, Lango, Lezgian, Limbo, Macushi, Malayalam, Mandarin Chinese, Maori, Nama Hottentot, Persian, Punjabi, Rapanui, Russian, Samoan, Shona, Somali, Spanish, Swedish, Tsez, Turkish, Tzotzil, Ute, Wai Wai, Wardaman, Warekena, Welsh, Wolof, West Greenlandic, Zoque.
4 Additional examples illustrating these oppositions can be found in Fortescue (1984):
(i) angirlasi- nngil- anga be_homesick not 1SG.DEC ��I am not homesick.�� (ii) niri- riir- pit eat already 2SG.INT ��Have you already eaten?�� (iii) niri- guk eat 2SG.3SG.IMP ��Eat it!��
5 Some languages (e.g. German) exploit this use of complementizers in non-embedded clauses quite systemati-
cally for the characterization of minor sentence types.
6 Croft (1994: 462ff.) argues that the formal properties distinguishing the basic sentence types are particularly
salient ones, even from a cross-linguistic perspective. Moreover, he proposes that the ��distance�� between sen- tence types can be iconically motivated by taking into account their structural differences: The more substantial the formal differences, the farther the sentence types are apart in terms of their use potential or illocutionary force.
7 There are cases, however, where such an identification does not seem to be possible unless and until a second
property (e.g. intonation) is considered. Polite imperatives in German are a case in point. In these imperatives the distal (polite) form of address (Sie) has to be used as subject and thus the resultant sentences have the formal properties of interrogative structures. A sentence like Kommen Sie mit (lit.: ��Come you along��) would still be classified as an interrogative if it has a rising rather than a falling intonation contour. Similarly, verb-first struc- tures such as Kommt ein Mann zur T��r herein (lit.: ��Comes a man through the door��) can be used both as inter- rogative sentences and as declaratives. Such declaratives, however, require specific contextual conditions for their use (cf. Önnerfors, 1997).
8 We do not wish to rule out languages in which more than one formal property is necessary for marking basic
sentence types. Although such languages seem to be rare, the truth is that little systematic work has been done on this matter.
9 For oppositions of this kind the term ��marked declaratives�� has been proposed (cf. Sadock & Zwicky, 1985).
We will avoid this terminology since, as stated above, also so-called ��unmarked declaratives�� have specific for- mal properties which distinguish them from other sentence types.
10 Under specific circumstances verb first structures can also be used as declarative sentences, as exclamative
sentences and as conditional antecedents.
11 This picture is not fully adequate for German and can only be justified in terms of frequency. In the over-
whelming majority of cases, main clause declaratives are indeed verb-second structures, but verb-first structures can also be used as declaratives under specific conditions (cf. Önnerfors, 1997). Lohnstein (2000) has shown that it is the interaction between word order and verbal mood that differentiates between sentences types in German clearly and unambiguously.
12 As a matter of fact, according to their answer set alternative questions are like constituent questions. 13 Note that such bias may also be produced by polarity items such as yet and already:
(i) Have you not eaten yet? (ii) Haven��t you eaten already?
14 As is shown by our examples in (47) – (49)), in English (with the exception of be as a main verb and for some
speakers have) only auxiliary verbs can be shifted to a position preceding the subject. If the corresponding de- clarative sentence does not contain an auxiliary, the all-purpose auxiliary do is introduced. This do is fronted and carries the tense.
15 Languages also show differences in their inventory of interrogative words. One usually finds interrogative
words which replace the core constituents or arguments of a sentence and typically inquire about persons and things (who versus what) as well as interrogative words in an adverbial function which are typically used to seek information about (i) the location of a situation (where), (ii) its temporal setting (when), (iii) the manner of carry-

Page 42
ing it through (how) and (iv) the reason for it (why). Besides this core inventory one can also find interrogative words for determiners and/or adjectives (English which, Finnish kumpi ��which one of the two��), quantifiers (French combien ��how many��) and ordinal numbers (Finnish monesko, German der wievielte ��the how many- th��). Specific interrogative words for verbs or verb phrases like Tahitian eaha, Tagalog ano, Palauan mekera or Futunian a are frequently found in Oceanic languages but rare otherwise. Interrogative words for prepositions have so far not been attested.
16 A prototypical imperative implies a second person addressee. If there are further person distinctions made in
the same paradigm they follow the following hierarchy (cf. Aikhenvald, 2003): 2 > 1PL (inclusive) > 3SG/PL > 1SG and/or 1PL (exclusive).
17 Especially the frequent confusion of form and function makes information provided in grammatical descrip-
tions difficult to interpret.
18 Strictly speaking, the form chantons in (91c) is first person plural present indicative (nous chantons). What
makes it imperative or hortative is the omission of the subject.
19 Note that there are two properties that distinguish (98b) from (98a): (98b) invariably contains a negation and
can have rising intonation; structures of type (98a) have neither of those properties.
Search more related documents:Speech Act Distinction in Syntax

Set Home | Add to Favorites

All Rights Reserved Powered by Free Document Search and Download

Copyright © 2011
This site does not host pdf,doc,ppt,xls,rtf,txt files all document are the property of their respective owners. complaint#nuokui.com