8.

Typology and Discourse Analysis

JOHN MYHILL
Subjec Linguis ics ! Discourse Analysis DOI" #$.####%b.&'8$()#*$+&(8.*$$).$$$$&.,

$ In roduc ion
The relationship between typology and discourse analysis has been characterized by, on the one hand, a general ideological compatibility and, on the other, inherent practical difficulties in combining the interests of the two subdisciplines. The general ideological compatibility is the result of the subdisciplines sharing the view that the study of language should be based upon analysis of empirical data rather than thought experiments. In syntax, semantics, and even pragmatics, intuitions have played a central role in gathering data, and interaction with discourse analysis has tended to be controversial from the outset because of the emphasis in discourse analysis upon the empirical analysis of linguistic data. A discourse analyst interested in the subdiscipline's relationship with, for example, syntax must immediately discuss performance data and phenomena that mainstream syntacticians simply reject as irrelevant thus any syntax combined with discourse analysis can only be nonmainstream syntax. !o such ideological problem arises in the case of typology" intuitions play essentially no role in the data analyzed by typologists, and typologists are only too happy, in principle, to consider the possible relevance of discourse phenomena to the problems they investigate. In practice, however, it has been difficult to integrate wor# in these two disciplines, because of various empirical difficulties. There has, therefore, been relatively little research which can be said to have been the product of the interaction between these subdisciplines. $uch of what I write here will therefore be programatic, although I will also discuss findings in this area to exemplify what can be done. %efore proceeding, it will be necessary to describe what I am ta#ing to be &typology.' The prototypical typological study has data from a wide variety of genetically unrelated languages, analyzed within a common descriptive paradigm which ma#es it possible to directly, systematically, and (relatively) simply compare data from these various languages and propose hypotheses regarding human language in general. This type of study was pioneered by *oseph +reenberg (,-..a, ,-..b), who categorized a large number of languages according to, e.g., most common order of subject, verb, and direct object, order of adposition and noun, etc., and, on the basis of this categorization, determined correlations which could be hypothesized as characteristic of human language in general (e.g. verb/object languages are very li#ely to have prepositions rather than postpositions). It is the use of a systematic common descriptive paradigm, allowing for direct comparison between a wide variety of languages, which distinguishes the methodology of typology from that of other approaches. 0ithin the field of discourse analysis, there have been many studies which have compared different languages but which would not, on this understanding, be considered to be specifically typological, because they are not focused upon developing a system for direct,

systematic, and universal comparison of a wide variety of languages as +reenberg's studies were (e.g. Tannen ,-1, %rown and 2evinson ,-13 %lum/4ul#a ,--,). 5uch wor#s are discursive in nature, typically comparing 6nglish with one (or very rarely two) other language(s) and selecting examples which show how the languages differ in certain respects, or, alternatively, how they can fulfill similar discourse functions using constructions which may superficially appear to be different. Typically, there is no systematic, exhaustive, and 7uantitative analysis of a database, the examples are selected anecdotally depending upon which point the author wishes to ma#e in a particular article without systematic demonstration that they represent a general pattern, comparison between actual usage in the languages is unsystematic, and it is not clear how additional languages would fit into the comparative framewor# of the study. Thus, although such studies are comparative, they are not really directed toward establishing a systematic universal framewor# for categorizing discourse phenomena in the way that +reenberg's studies established a systematic universal framewor# for categorizing syntactic phenomena. %ecause I am discussing the relationship of typology and discourse analysis, then, I will in the present chapter discuss those approaches which have been more similar to +reenberg's in this respect. 5ection , of this chapter will describe general problems associated with methodology combining typology and discourse analysis. 5ections 8 and 9 then discuss two approaches to these problems, the use of universal conceptual systems of classification and the use of translation data.

# -roble.s o/ Typological Discourse Analysis

The study of discourse phenomena in a typological framewor# presents some inherent difficulties which are not found in other areas of typology. Traditional typological studies (+reenberg ,-..a %ybee ,-1: ;roft ,--<) use as their main source of data reference grammars from a wide variety of languages, and the linguistic phenomena they consider are those which are li#ely to be found in a reference grammar, e.g. typical word order (of subject=object=verb, adposition=noun, etc.), structural characteristics of voice alternations, phonological inventory, etc. >nfortunately, this is not possible with the sort of phenomena typically of interest to discourse analysts. 6xisting reference grammars of less/#nown languages generally have very little in the way of discourse analysis, and what limited analyses they do have are not written in a way to allow for cross/linguistic comparison by someone who does not #now the language very well. ?or example, if I were to attempt to do a typological study of the functions of contrastive connectives similar to 6nglish but, I could probably gather a list of words in a wide variety of languages with some type of generally similar function, but it would be impossible on the basis of the descriptions of these particles in reference grammars to understand and then compare the functions of these different words. Another problem for typological discourse analysis as compared with more traditional discourse analysis is the degree of familiarity of the researcher with the languages to be analyzed in a typological study the linguist is not going to #now all the languages under investigation very well, while in a traditional discourse study the investigator is li#ely to be a native spea#er of or very proficient in the language(s) under investigation. 0hile there are recorded cases of individuals #nowing a large number of languages, these are typically closely related or at least related languages in typological studies, on the other hand, it is typical to have data from languages from ,:@8< different language families. Although this problem can be alleviated to some extent through reliance on texts with interlinear glosses, this still does not entirely ma#e up for the researcher's lac# of in/depth #nowledge of the

language additionally, the languages in which there are a large enough number of such texts of reasonable length which can be the basis for a discourse study are concentrated in just a few language families (Austronesian, Australian, and 5emitic in particular), while the great majority of language families have no languages at all with a large number of texts with interlinear glosses. %ecause such studies are not really feasible, linguists interested in discourse analysis and typology have instead focused upon using a narrower range of languages with which they themselves have some expertise. 6ven in this case there has to be much more dependence upon observations of textual patterns (so that longer texts must be used) and much less upon introspective judgments than would be the case for linguists wor#ing in their native languages. In such a situation, we cannot expect the relatively 7uic# and impressive types of language/universal generalizations which individual typological studies of, e.g., word order patterns have been able to produce in fact, it is unli#ely that any single researcher will be able to conduct studies of a genetically diverse enough group of languages to allow for the degree of confidence in universality which typologists are accustomed to. Aather, in order to achieve an extensive genetic spread, it is necessary for a variety of discourse analysts, each wor#ing in a number of languages, to develop a uniform means of systematically comparing their results from these different languages. Aside from the 7uestion of which data to analyze, it is also necessary for typological discourse analysts to consider the nature of the discourse categories to be used. It is very common for linguists describing discourse categories in different languages to use the same words to describe something in the language they are investigating, e.g. &topic,' &focus,' &contrast,' etc., but this does not mean that they are referring to the same discourse phenomena. ?or example, although the term &topic' has been used to refer to a supposedly discourse/based category in a wide variety of languages, there is no cross/linguistic agreement about what a &topic' is. In each language, &topic' actually refers to whatever discourse properties result in a certain language/specific structure being used, so that the definition is a result of the language/specific pattern, and these structures in different languages actually serve clearly distinct functions. Thus in *apanese, anything mar#ed with the postposition wa is called a &topic' (4uno ,-39), while in 6nglish, the term &topic' might be used for a clause/ initial constituent whose syntactic role would call for some other position (e.g. That boo# I don't like), though usage differs (see ?irbas ,-.. and various articles in 2i ,-3. and +ivBn ,-19a)., Though linguists specializing in each of these languages may develop some sort of ostensibly discourse/based &definition' of a &topic' in this language (e.g. &0hat a sentence is about' or something which &sets a spatial, temporal, or individual frame/wor# within which the main predication holds' (;hafe ,-3." :<)), these definitions are invariably 7uite vague. Thus, in practice, the only objective way to determine whether a constituent is actually a &topic' has been to apply some language/specific structural test (e.g. to see if it is mar#ed with wa in *apanese). As a result, the &topics' in the different languages do not have the same discourse function at all, e.g. the translation of I like Mary into *apanese would have &I' as a &topic' (mar#ed with wa), the translation of I read the book into Tagalog would have &the boo#' as a &focus' (mar#ed with ang), etc. There is, then, no cross/linguistic idea of a &topic,' and so such a category cannot serve as a basis for cross/linguistic comparison. In order to deal with this problem of cross/linguistic comparison of function, linguists wor#ing in typological discourse analysis have focused upon developing a set of criteria

which ma#e it possible to give an objective, cross/linguistic definition of the discourse function of a particular form or construction. >sing these criteria, a linguist can go through a text in a given language, note all the occurrences of a given form or construction in that language, determine numerical scores for that form or construction according to various parameters (e.g. for an !C, how recently its referent has been mentioned, whether it refers to a human being, etc.), and then compare these scores with those of other constructions in other languages. The 7uestion, of course, is exactly which scores should be used in which cases, and this is a matter of ongoing research. A second approach to the problem of comparison is to use translation data we can get some idea of the functional similarity of and difference between constructions in different languages by seeing how often and in what circumstances they translate as each other. In section 8, I will discuss parameters used in classification of discourse function in section 9, I will discuss the use of translation data.

* 0ni1ersal Sys e.s o/ 2lassi/ica ion o/ Discourse 3unc ion

I will describe here various text/count methods which have been developed to give an objective, cross/linguistically applicable description of the discourse function of a given construction. The use of such text counts does not suggest that spea#ers themselves go through any calculations similar to those of the linguist, nor does it imply that a given text/ count score will predict with ,<< percent accuracy which construction will be used on each occasion. Aather, such counts are purely descriptive tools to allow for cross/linguistic comparison. The most widely used text/counts, associated particularly with Talmy +ivBn and students of his, are called Aeferential Distance (AD) and Topic Cersistence (TC). ?or each !C in a text, AD counts the last time its referent was referred to (including zero anaphora) in the preceding text (e.g. AD E 8 if it was referred to two clauses before), while TC counts how many times it is referred to in the following text (e.g. TC E , if it is referred to again in the following clause but not in the clause after that). 0e can say that an !C is generally more topical if its AD is low and its TC is high, but of course we are really measuring two types of topicality here, anaphoric (AD) and cataphoric (TC). AD and TC counts ma#e it possible to give a functional profile of a given construction or !C type. ?or example, suppose that we are trying to give a general characterization of the function of the active/passive alternation in 6nglish, e.g. Bill wrote that book vs. That book was written by Bill. 0e go through a text, collecting all active transitive and passive constructions, and then count the average ADs for the Agents of actives (Bill in Bill wrote that book), the Agents of passives (Bill in That book was written by Bill), the Catients of actives (that book in Bill wrote that book), and the Catients of passives (That book in That book was written by Bill). 0e then calculate the mean and median AD and TC scores for active Agents, passive Agents, active Catients, and passive Catients, or list the populations in a table. %y doing similar studies in a variety of languages, we can systematically compare the discourse functions of active and passive constructions in different languages. This approach has been useful in providing a typological perspective on functional alternations, clarifying the discourse motivations underlying these alternations, and also sharpening the descriptive tools for typological descriptions it does not suggest that spea#ers ma#e such calculations in deciding which construction to use (although AD can be interpreted as being generally

correlated with cognitive accessibility). In the remainder of this section, I will discuss a number of studies which have been done using these measures. Foice alternations in different languages have been characterized in various descriptive grammars in a variety of ways, in particular Active vs. Passive, Direct vs. Inverse, Ergative vs. Antipassive, and (for Chilippine languages) Agent ocus vs. !oal ocus. Gowever, the basis for such characterizations has often been unclear. ;onsider, for example, the following constructions in Tagalog"

(,) %umasa ang lala#i ng diyaryo. read man newspaper, &The man read a newspaper.' (8) %inasa ng lala#i ang diyaryo. read man newspaper, &The man read the newspaper.'

;ase functions in Tagalog are mar#ed by prepositions, here ang and ng. It is clear that ang mar#s intransitive subjects (e.g. "atalino ang lalaki &intelligent ang man' E &The man is intelligent'). The 7uestion here is what general function to ascribe to ang and ng. Hne possibility is to say that ang mar#s subjects (both intransitive and transitive) and ng mar#s direct objects and obli7ue !Cs. Then (,) would be an active construc tion, with lala#i as the subject and diyaryo as the direct object, while (8) would be a passive construction, with diyaryo as the subject and lala#i as the obli7ue Agent. Alternatively, we might say that ang is an absolutive case mar#er (mar#ing intransitive subjects and direct objects), while ng is an ergative (transitive subject) and obli7ue case mar#er. Then (,) would be an antipassive construction (grammatically intransitive), with lalaki as the intransitive subject mar#ed with the absolutive preposition ang, and diyaryo (which is in this case an obli7ue rather than direct object) mar#ed with the obli7ue preposition ng, and (8) would be an ergative construction, with lalaki as the transitive subject, mar#ed with the ergative preposition ng, and diyaryo as the direct object, mar#ed with the absolutive preposition ang. In fact, earlier studies of Chilippine languages (e.g. 5chachter and Htanes ,-38) used yet another type of terminology, referring to ang as mar#ing &focused' constituents (which causes confusion of another type in terms of cross/linguistic comparison, since the term &focus' is usually used with some sort of entirely different meaning) and ng as mar#ing certain nonfocused constituents, so that (,) is an &Actor ?ocus' construction while (8) is a &+oal ?ocus' construction. 5imilar labeling problems arise in many other languages (see +ivBn ,--I). The result of all of this has been that grammars of different languages have used a bewildering variety of labels for different constructions and it is unclear how to compare these. In response to this problem, linguists interested in functional factors such as discourse role began to develop discourse criteria for distinguishing these different types (see +ivon ,--I). The general criteria which have come out of these studies are"

, The functionally unmar#ed type, which I will refer to by the general name direct (including constructions which have been called &Active' and &6rgative'), typically has an Agent (A) which is somewhat more topical (e.g. lower AD, higher TC) than its Catient (C). 8 If a construction is particularly used when the Catient is very high in topicality, this construction is referred to as an inverse. 5uch constructions can be used even when the Agent is relatively topical as well, in situations where the relatively high topicality of the Agent would prevent the use of a Cassive. 9 If a construction is particularly used when the Agent is very low in topicality, this construction is referred to as a passive.

I If a construction is particularly used when the Catient is very low in topicality, this construction is referred to as an antipassive.

2et us now see more specifically how text counts can be used as diagnostics for categorization of particular constructions in particular languages (table 1.,). Table 1., Foice alternations in 4oyu#on and Dyirbal N ARD ATP P RD Data sources" T4o.pson 5#&&67 58oyu9on7: 2oore.an 5#&887 5Dyirbal7 4oyu#on (a) ,<< 8.88 :.I: 8.-, (b) ,,< I.-9.-< ,.:, (c) :< J J 1.I: Dyirbal (a) 88: 9.I8 8.<< :.,(b) II ,.I: 8.8< ,<.:3 FTP

9.3. ..19 ,.1. ,.,. <.1.

6ach of the labels (a), (b), and (c) in table 1., refers to a particular construction in these languages, and the data in these tables can be used in combination with the characterizations of the different voice types given above to label these constructions in a cross/linguistically comparable and consistent manner. ?or both of these languages, the (a) construction is direct=active, having an Agent which is somewhat higher in topicality than its Catient (but the difference is not as great as would be characteristic of an antipassive construction). The 4oyu#on (b) construction has a C which is very high in topicality (the lowest AD and highest TC of any of the constructions here), and its A is not particularly high or low in topicality this is therefore an inverse construction. The A in the 4oyu#on (c) construction, on the other hand, is very low in topicality (in fact obligatorily absent), and so this is a passive construction. The Dyirbal (b) construction is particularly characterized by having a very nontopical C (high AD, low TC), and so we can call this an antipassive construction (see +ivBn ,--I for similar discussion of a number of other languages). 2inguists have also applied AD to investigating word order variation. 5tudies from a variety of languages have found that preverbal arguments have on average a higher AD than postverbal arguments (there does not seem to be any corresponding clear pattern relating TC and word order). Table 1.8 shows data in this regard from a number of languages. Table 1.8 AD and word order in four languages Ute (s) Ute (o) Biblical Hebrew Spanish Chamorro Notes" Nu.bers are ;D 5N<si=e7. All da a are /or subjec s> e,cep 0 e 5o7> ?4ic4 is /or direc objec s. Data sources" @i1An 5#&8)b7 50 e7: 3o, 5#&8)7 5Biblical Hebre?7: Ben i1oglio 5#&8)7 5Spanis47: 2oore.an 5#&8)7 524a.orro7 Costverbal ,.1, (1.) I.8, (,I) ..:8 (9:3) 9.:I (I,) 3.I: (8<<) Creverbal :.I- (,,I) 3.31 (I.) ,<..I (,,8) 1.:: (,3<) ,<.-< (-.) These data have been ta#en from languages which are generally verb/initial (%iblical Gebrew and ;hamorro), 5FH (5panish), and where the verb most often follows both the subject and the object (>te) (see other studies in +ivBn ,-19a showing a similar pattern) thus there is

reason to suppose this may be a universal pattern. At first this appears surprising, because an often/repeated theme of functional linguistics is that &old information precedes new information' (e.g. ;ontreras ,-31), whereas the data in table 1.8 suggest the reverse, that arguments are more li#ely to precede the verb if their AD is higher, so that they represent newer information. Gowever, it is possible to suggest a resolution to this apparent contradiction (although this is speculative and should be chec#ed against more data). ;laims that old information generally precedes new information have been made on the basis of data from 6uropean languages which are generally 5FH, using an existential/presentative construction li#e #n the roo$ stood a chi"ney, where the preverbal roof is old information and the postverbal chimney is new information. It is possible that the distinctive use of F5 order in this existential/presentative construction is specific to 5FH languages, that such constructions constitute the only basis for the general claim that &old information precedes new information,' and that if these constructions in these 5FH languages are excluded, the reverse is generally true, and &new information precedes old information,' as suggested by the data in table 1.8. 5upporting this idea is the fact that in the data from the only 5FH language here, 5panish, the researcher specifically excluded existential/presentative constructions from the counts (see %entivoglio ,-19) if these constructions are included, the picture changes, as the postverbal subjects have a higher AD (,,.--, ! E ,I,) than the preverbal ones (1.88, ! E ,1<).8 Another criterion for categorizing discourse function in different languages is temporal se7uencing or foregrounding. Introduced in 2abov (,-38) (as the concept &narrative clause'), this was first extended to data in a variety of languages in Gopper (,-3-). According to this criterion, a clause is temporally se7uenced if it has past time reference and refers to the next event in a story line (e.g. the second clause, but not the first, in I was reading in the library and this guy ca"e up to "e K). The se7uencing function has been related to alternations in word order, voice, and verb form. ?or example, 5chiffrin (,-1,) shows that the 6nglish historical present is associated with temporally se7uenced clauses, while Gopper (,-3-) shows that temporal se7uencing is associated with the use of the verbal forms with a di%pre$i& in $alay. $yhill (,--8) argues that, in languages with a relatively high fre7uency of F5 order, se7uencing is particularly associated with F5 word order, while 5F order is associated with unse7uenced clauses. Hn the other hand, in languages with a lower fre7uency of F5 order, this correlation is not found. This is shown by the data in table 1.9 (see also data from Hld 6nglish in Gopper ,-3-). The %iblical Gebrew data here are particularly stri#ing, in that they show that when the language changed to a lower fre7uency of F5 order, the association between temporal se7uencing and F5 order disappeared. The concept of temporal se7uencing therefore ma#es it possible to ma#e a typological generalization regarding word order type. Table 1.9 0ord order and temporal se7uencing T ot il !BH Chorti Spanish "BH Romanian Notes" Nu.bers are CSD 5N<si=e7. EBH F Early Biblical Hebre?> LBH F La e Biblical Hebre?. Data sources" @i1An 5#&''7 5Hebre?7: My4ill 5#&867 5o 4ers7 All 1< (1--) .: (,<--) :, (,1I) II (8<<<) I< (I8<) 9, (::I) 5e7uenced -8 (8II) 1< (:I.) 38 (98<) :1 (9,.) 8< (1:) 88 (,,9) >nse7uenced 3. (.::) I- (::9) I3 (,:8) I, (,.1I) I. (99:) 99 (II,)

2inguists have proposed other types of text/counts which can be useful in giving a profile of the discourse function of a construction. $yhill and Ling (,--.) propose a definition of the term &contrast' which can be objectively applied to naturally occurring usages so as to categorize individual clauses as contrastive or not (and also to distinguish between different subtypes of contrast), so that one or another contrastive function can be shown to be statistically associated with the use of a certain word order, intonation pattern, or particle (e.g. *apanese wa, 4orean /(n)"). In ?orrest's (,--I) study of voice alternations in %ella ;oola, in addition to counts associated with !C information status such as AD and TC, she also uses a text/count distinguishing between !Cs which refer to major characters in a story and those which do not, and shows that variation on this parameter correlates with the use of one or other voice construction. A related and more objective and universally applicable (though also more time/consuming) type of measure is Topicality Muotient, described in Thompson (,-1-). To determine this, one counts the number of clauses in which a referent is referred to in an entire text, divides this by the number of total clauses in the text, and then assigns this score to every mention of this referent. Hther possible counts categorize referents according to their humanness, animacy, number, referentiality, function in previous clause, form (e.g. pronoun, unmodified noun, modified noun, common noun, proper noun, etc.), or, for that matter, anything else the linguist thin#s is important which can be coded objectively.

) Transla ion Da a
Translation provides another means of comparing discourse functions in different languages. It is useful in that it gives some idea of the functional similarity or difference between constructions in different languages. ?or example, in Dryer's (,--I) study of voice in 4utenai, he as#ed a bilingual 4utenai/6nglish spea#er to translate a 4utenai text into 6nglish. Ge found that, out of 3< clauses using a certain 4utenai construction clearly associated with highly topical Catients, only nine were translated into 6nglish as passives, the rest being translated as actives, suggesting that this 4utenai construction is functionally li#e an inverse rather than a passive. 5ometimes, translation data show that text/counts such as AD and TC do not give a true picture of the functional similarity or difference between different constructions in different languages. ?or example, 5un and +ivBn (,-1:) use data such as AD and TC to argue that object/fronting constructions in ;hinese and %iblical Gebrew serve basically the same function. Gowever, $yhill and Ling (,--9) show that, if we loo# at translation data, we see that the object/fronting constructions in these languages are fre7uently not translated as each other for example, of 18 HF constructions in a %iblical Gebrew database, I1 (:-N) do not use an HF construction in the ;hinese translation, while of ,-9 HF constructions in the ;hinese translation, ,:- (18N) do not use an HF construction in the Gebrew original. In other words, in the majority of cases, an HF construction in one language would not be used where an HF construc tion would be used in the other language. This shows that the Gebrew and ;hinese HF constructions clearly differ significantly in discourse function, in spite of their AD and TC scores. In such a situation, where established criteria for cross/linguistic comparison suggest functional similarity which is demonstrated to be incorrect by translation data, linguists interested in cross/linguistic comparison must develop other criteria which will capture these differences. In the case of the comparison of %iblical Gebrew and ;hinese object/verb constructions, $yhill and Ling (,--.) develop a text/count for contrast (see section 8.9 above), distinguishing between several subtypes of contrast, in order to describe exactly how

these constructions are similar in function and how they are different they find that certain types of contrastive functions result in HF order in both languages, but for other contrastive functions, only %iblical Gebrew fronts objects, while for still others, only ;hinese does. Another use of translation data can be to ma#e it possible to distinguish between different functions which a particular construction can serve in a manner which is objective and uses parameters which languages themselves treat as significant the studies described in section 8., ma#e no such functional distinctions but simply lump all structurally similar constructions together. $yhill and Ling (,--I) is a contrastive study of voice in ;hinese, 6nglish, and %iblical Gebrew of this type, using ;hinese and 6nglish translations of +enesis. $yhill and Ling divide up the database into one of a number of types of clauses, where all of the clauses in each type use a particular Gebrew construction translated as a particular ;hinese construction and a particular 6nglish construction, and give a functional characterization of the type in general. Thus, for example, the combination of an 6nglish passive, a ;hinese Catient/verb construction (suppressing the Agent), and a Gebrew niphal (an intransitive form often li#e a passive) occurs ,8 times in the translation database, characteristically having an obscure Agent and an inanimate Catient (e.g. The $ountains o$ the deep and the $loodgates o$ the sky were stopped up (+enesis 1"8)), while the combination of an 6nglish passive, a ;hinese active, and a Gebrew niphal occurs ,- times, characteristically with future time reference and a first or second person Agent (implied in the 6nglish and Gebrew, e.g. By this you will be put to the test, but ;hinese wo yao shiyishi ni"en, lit. &I will test you' (+enesis I8",:)), so that ;hinese is the only one of the three languages which does not use an agent/ suppressing construction to avoid mentioning first or second person Agents in such a potentially sensitive situation similar patterns were found with other combinations of translations. Translation data of this type ma#e it possible to divide up the functions of each of the constructions involved into subtypes and ma#e the functional differences between the constructions in the different languages explicit. Although translations are helpful in comparing functions across language, they have limitations. The most basic problem is the fact that, for many pairs of languages (e.g. 2uganda and Ouni), it is hard to get direct translation data. Hne possibility is to use material from a third language which has been translated into both (the %ible is the most li#ely source here) another is for the linguist to get native spea#ers to ma#e translations, although this will often be problematic as it will li#ely have to be done through a third language. Another problem with translations is that there is some tendency to translate according to certain conventions, with certain constructions translating as certain constructions and certain words translating as certain words even in cases where this might not result in the most idiomatic translations for this reason, translation data are most significant when the translator does not follow the usual translating conventions, as this identifies cases where the functional differences between the constructions or words in 7uestion are great enough to overcome fixed translation practices. A good source of cross/linguistic data is The Pear 'tories (;hafe ,-1<a). A silent film was shown to people with a variety of first languages (more than :< of 6nglish, at least 8< of ;hinese, *apanese, $alay, Thai, Cersian, +ree#, +erman, Gaitian ;reole, and 5acapultec ($ayan)), and they were as#ed to retell the story in their own language. In this case there is no actual language/to/language translation, although the texts in the different languages are to some extent parallel. The linguistic studies which have been done of these stories, however, have thus far not been typological in nature, either focusing upon the 7uestion of how people report the plot of the movie in a single language (e.g. ;hafe ,-1<b) or comparing 6nglish

with a single other language without any attempt to integrate this into a general cross/ linguistic framewor# for typological analysis (e.g. Tannen ,-1<).

6 2onclusion
Typology and discourse analysis are fields which have much to offer each other. Typological studies can provide a basis for discourse studies by offering a point of reference for discourse phenomena other than comparison with individual &other languages,' which almost always turn out to be 6nglish. Discourse analysis offers typology a way of comparing different constructions in different languages and sorting through the enormous terminological confusion and inconsistency found in reference grammars which have plagued typological studies. There has thus far been relatively little wor# integrating these approaches, because of the inherent problems I have discussed, but some progress has been made in this regard. The most li#ely source of a brea#through in this area is the development of a megacorpus of translation materials from a single text into a wide variety of genetically unrelated languages, with interlinear glosses, using perhaps the %ible, some other widely translated wor#, or The Pear 'tories a large enough corpus would to some extent mitigate the problems inherent in translation data. It will also be necessary to underta#e further studies along the lines of +ivBn (,-19a, ,--I), where a number of researchers apply similar text/count methodology to study similar structural phenomena in texts from a variety of genetically diverse languages, but the methodology should be expanded to include not just AD and TC (useful though these have shown themselves to be) but other text/count methods as well. Another welcome move would be the development and application of similar text/count methods to the cross/linguistic analysis of other phenomena of concern to discourse analysts, e.g. politeness, definiteness, discourse particles, etc. %ecause typological discourse analysis has developed out of traditional typology, it has focused more upon issues such as word order and voice, which can be more directly related to syntax, but there is no reason why this has to continue to be the case in the future. , A similar problem arises for &focus,' which means a completely different thing when referring to 5omali, Tagalog, Gungarian, or other languages. 8 Gerring (,--<) suggests a different type of universal account of word order patterns, one based upon categories which she refers to as &continuous topic,' &shifted topic,' &contrastive focus,' and &presentative focus,' and relating the position of these in a given language to the unmar#ed order of subject and verb in that language. Although Gerring's proposals are interesting, she does not give 7uantitative data or an objective definition of exactly what she means by e.g. &continuous topic,' nor does she attempt to integrate her findings with those of the papers in +ivBn (,-19a) this should be addressed in future research. Cayne (,--8) has a number of papers discussing factors affecting word order in a variety of languages from a functional perspective. I am not focusing on the papers in that volume because of space limitations and because they are not intended to provide a systematic comparative typological framewor# as are the papers in, e.g., +ivBn (,-19a, ,--I).

;E3E;EN2ES
%entivoglio, Caola. ,-19. Topic continuity and discontinuity in discourse( a study o$ spoken )atin%A"erican 'panish . In +ivBn ,-19a, pp. 8:: 9,8.

%lum/4ul#a, 5hoshana. ,--,. Interlanguage prag"atics( the case o$ re*uests . In A. Chillipson, eds, oreign+'econd )anguage Pedagogy ,esearch , pp. 8:: 38. ;levedon" $ultilingual $atters . %rown, Cenelope and 5tephen, ;. 2evinson. ,-13. Politeness( 'o"e -niversals in )anguage -sage ;ambridge" ;ambridge >niversity Cress . %ybee, *oan 2. ,-1:. Morphology Amsterdam" *ohn %enjamins . ;hafe, 0allace 2. ,-3.. !ivenness, contrastiveness, de$initeness, sub.ects, topics, and point o$ view . In 2i ,-3., pp. 8: :.. ;hafe, 0allace 2., eds ,-1<a. The Pear 'tories !orwood, !*" Ablex . ;hafe, 0allace 2. ,-1<b. The deploy"ent o$ consciousness in the production o$ a narrative . In ;hafe ,--<a, pp. - :<. ;ontreras, Geles. ,-31. El orden de palabras en espanol . $adrid" ;atedra . ;ooreman, Ann. ,-19. Topic continuity and the voicing syste" o$ an ergative language( /ha"orro . In +ivBn ,-19a, pp. I8: -<. ;ooreman, Ann. ,-11). Ergativity in Dyirbal discourse . )inguistics , (8.) "3,3 I.. ;roft, 0illiam. ,--<. Typology and -niversals ;ambridge" ;ambridge >niversity Cress . Dryer, $atthew. ,--I. The discourse $unction o$ the 0utenai inverse . In +ivBn ,--I, pp. .: ,<<. ?irbas, *. ,-... 1on%the"atic sub.ects in conte"porary English . Traveau& )inguisti*ues de Prague (8) . ?orrest, 2inda %. ,--I. he de%transitive clause in Bella /oola( passive vs2 inverse . In +ivBn ,--I, pp. ,I3 .1. ?ox, Andrew. ,-19. Topic continuity in Biblical 3ebrew narrative . In +ivBn ,-19a, pp. 8,: :I. +ivBn, Talmy. ,-33. The dri$t $ro" 4'# to '4# in Biblical 3ebrew( the prag"atics o$ tense% aspect . In ;. !. 2i, eds, Mechanis"s o$ 'yntactic /hange , pp. ,1, 8:I. Austin" >niversity of Texas Cress . +ivBn, Talmy, eds ,-19a. Topic continuity in discourse . Amsterdam" *ohn %enjamins . +ivBn, Talmy. ,-19b. Topic continuity and word%order prag"atics in -te . In +ivBn ,-19a, pp. ,I, 8,I. +ivBn, Talmy, eds ,--I. 4oice and Inversion Amsterdam" *ohn %enjamins . +reenberg, *oseph. ,-..a. 'o"e universals o$ gra""ar with particular re$erence to the order o$ "eaning$ul ele"ents . In *. G. +reenberg, eds, -niversals o$ !ra""ar , 8nd edn, pp. 39 ,,9. ;ambridge, $ass." $IT Cress . +reenberg, *oseph. ,-..b. )anguage universals, with special re$erence to $eature hierarchies . 5anua )inguaru", 'eries Minor, 67 . The Gague" $outon de +ryter . Gerring, 5usan ;. ,--<. In$or"ation structure as a conse*uence o$ word order type . Berkeley )inguistics 'ociety (,.) ",.9 3I. Gopper, Caul *. ,-3-. Aspect and $oregrounding in discourse . In T. +ivBn, eds, Discourse and 'ynta& , pp. 8,9 I,. !ew Por#" Academic Cress . 4uno, 5usumu. ,-39. The 'tructure o$ the 5apanese )anguage . ;ambridge, $ass." $IT Cress . 2abov, 0illiam. ,-38. The trans$or"ation o$ e&perience in narrative synta& . In 0. 2abov, )anguage in the Inner /ity , pp. 9:I -.. Chiladelphia" >niversity of Cennsylvania Cress . 2i, ;harles, eds ,-3.. 'ub.ect and Topic !ew Por#" Academic Cress . $yhill, *ohn. ,-1I. A study o$ aspect, word order, and voice . >niversity of Cennsylvania, ChD dissertation. $yhill, *ohn. ,--8. 8ord order and te"poral se*uencing . In Cayne ,--8, pp. 8.: 31. $yhill, *ohn and Ohi7un, Ling. ,--9. The discourse $unctions o$ patient $ronting( a co"parative study o$ Biblical 3ebrew and /hinese . )inguistics (9,.,) " 8: :3.

$yhill, *ohn and Ohi7un, Ling. ,--I. A co"parison o$ the $unction o$ voice in Biblical 3ebrew, /hinese, and English . )anguage 'ciences (,..8) "8:9 19. $yhill, *ohn and Ohi7un, Ling. ,--.. Towards an operational de$inition o$ discourse contrast . 'tudies in )anguage (8<.8) "9,9 3<. Cayne, Doris, eds ,--8. The Prag"atics o$ 8ord #rder le&ibility Amsterdam" *ohn %enjamins . 5chachter, Caul and ?e, T. Htanes. ,-38. Tagalog ,e$erence !ra""ar %er#eley" >niversity of ;alifornia Cress . 5chiffrin, Deborah. ,-1,). Tense variation in narrative . )anguage , (:3) "I: .8. 5un, ;haofen and Talmy, +ivBn. ,-1:. #n the '#4 word order in Mandarin /hinese . )anguage (.,) "8. Tannen, Deborah. ,-1<. A co"parative analysis o$ oral narrative strategies( Athenian !reek and A"erican English . In /ha$e 97:;a , pp. :, 11. Tannen, Deborah. ,-1,. Indirectness in discourse( ethnicity as conversational style . Discourse Processes (I) "88, 91. Thompson, ;had. ,-1-. 4oice and obviation in 1ava.o . In Proceedings o$ the ourth Annual Meeting o$ the Paci$ic )inguistics /on$erence , pp. I.. 11. 6ugene" >niversity of Hregon Department of 2inguistics . Thompson, ;had. ,--I. Passive and inverse constructions . In !iv<n 977= , pp. I3 .I.

2i e 4is ar icle
$PGI22, *HG!. QTypology and Discourse Analysis.Q The 3andbook o$ Discourse Analysis. 5chiffrin, Deborah, Deborah Tannen and Geidi 6. Gamilton (eds). %lac#well Cublishing, 8<<9. %lac#well Aeference Hnline. <9 $arch 8<<3 Rhttp"==www.blac#wellreference.com=subscriber=tocnodeS idEg-31<.9,8<:-.1Tchun#Tg-31<.9,8<:-.1-U