Verbal Predicates in Flemish Sign Language (VGT) and South African Sign Language (SASL)

Verbal Predicates in Flemish Sign Language (VGT) and South African Sign Language (SASL) Mieke Van Herreweghe Ghent University Myriam Vermeerbergen Lessius Antwerpen/KULeuven University of the Free State, Bloemfontein, South Africa 1. Introduction In 2006 and 2007 a project on differences and similarities between SASL and VGT was set up for which a number of grammatical mechanisms and structures were being compared across the two signed languages, using a corpus consisting of similar VGT and SASL-data of a various nature. The overall goal was to contribute to a further understanding of the degree of similarity between unrelated signed languages. Part of the grammatical comparison focused on word order issues (published in Vermeerbergen et al., 2007). However, while annotating and analysing the data a number of problems emerged. These concerned sentences with two or more elements which at first sight seemed to function like verbal predicates. We shall present a cross-linguistic analysis of these sentences and try to determine the grammatical/typological status of the predicate-like elements. It will be shown that determining that status is not straightforward. 2. Research Design The methodology for the cross-linguistic study was driven by the Volterra et al. (1984) picture elicitation task first used in a study of word order in Italian Sign Language (LIS). This research design was chosen because it has frequently been used in signed language research so that it would yield results that would easily be comparable to results found in other signed languages. In addition to the initial study on LIS, the elicitation task has been used in research in at least six other signed languages: French Swiss Sign Language (Boyes-Braem et al., 1990), Sign Language of the Netherlands (Coerts, 1994), Flemish Sign Language (Vermeerbergen, 1996; 2004), British Sign Language (Saeed et al., 2000), Irish Sign Language (Leeson, 2001) and Australian Sign Language (Johnston et al., 2007). In

Mieke Van Herreweghe & Myriam Vermeerbergen Sze s (2003) work on Hong Kong Sign Language a similar method was used. The main purpose of these studies has been to analyse whether the signed language(s) studied exhibit a systematic ordering of constituents in declarative utterances with two arguments, and if this is the case, to determine which orders occur. In the Volterra et al. task, eighteen pairs of drawings with only one contrastive element are used to elicit sentences describing three distinct types of states of affairs: I. Six pairs of locative states of affairs presenting the motion and/or location of two entities relative to each other: 1. The tree is behind / in front of the house 2. The flowers are next to / in the vase 3. The man is far in front / right in front of the car 4. The cat is under / on the chair 5. The car goes under / on the bridge 6. The ball is under / on the table II. Six pairs of non-reversible states of affairs where only one of the illustrated entities, the animate or human one, would be likely to be the agent of the action: 7. The boy closes / opens the door 8. The girl / boy eats cake 9. The man builds / paints the wall 10. The girl watches T.V. / looks at the painting 11. The woman / man cuts the string 12. The man washes the dog / the car III. Six pairs of reversible states of affairs in which either illustrated entity could possibly be the agent: 13. The car is towing the truck / The truck is towing the car 14. The little boy hugs granny / Granny hugs the little boy 15. The boy pushes the girl / The girl pushes the boy 16. Mother brushes the child s hair / The child brushes mother s hair

Verbal Predicates in Flemish and South African Sign Language 17. The Native American stabs the cowboy in the back / The cowboy stabs the Native American in the back 18. The girl strokes the boy s cheek / The boy strokes the girl s cheek In this task, informants work in pairs of signers and interlocutors. The signer is videotaped and s/he has the drawings before him/her. For each pair one of the drawings is marked with an arrow. The interlocutor, another signer who is not being videotaped, has the same drawings but in his/her drawings there are no arrows. The signer is asked to sign one sentence describing the drawing marked with the arrow; the interlocutor is asked to indicate which of the two drawings of each pair is being described. For each signed language, two male and two female (near-)native signers representing two different regions per country contributed to the data collection. Their interlocutors were always (near-)native signers from their own region. For VGT four adult signers were videotaped, i.e. two male signers from West-Flanders and two female signers from East-Flanders. They ranged in age between 30 and 45, were all native signers and acquired Flemish Sign Language at home from their signing parents. For SASL one female black native signer from Bloemfontein, one female Indian native signer from Johannesburg, one male white near-native signer from Bloemfontein, and one male black near-native signer from Johannesburg were videotaped. For the first signer the interlocutor was a hearing signing colleague from work engaged in SASL research and teaching; for the other signers the interlocutor was the first informant. After the data elicitation, the videotaped data were transcribed (with the help of two South-African colleagues 1, cf. Vermeerbergen et al., 2007), annotated and analysed. 3. Results and Discussion The analysis and comparison of the data yielded a number of sentences with two or more elements which needed further scrutiny. These elements seemed to function as verbal predicates but they may not all belong to the class of verbs or verb constructions. The sentences discerned were (1) sentences with SIT, STAND and KNEEL-DOWN, (2) sentences with GIVE, (3) split sentences, and (4) sentences with so-called verb sandwiches. Before discussing these problematic 1 We would like to thank Philemon Akach and Emily Matabane, whose help in the transcription of the SASL data has been invaluable.

Mieke Van Herreweghe & Myriam Vermeerbergen constructions, we will first provide a brief discussion of the difference between verb signs and verb constructions in signed languages. 3.1. Verb signs and verb constructions All signed languages studied so far exhibit both an established lexicon and a productive lexicon. The established lexicon consists of those items which occur so frequently in the language that we can list them in dictionaries (Brennan, 1990:181) i.e. the codifiable lexical items. The citation form of established signs always shows the same component parts and these signs have a fixed meaning or set of meanings. They are normally transcribed with upper-case letters, e.g. EAT: EAT This is not the case for the items or constructions representing the productive lexicon. Brennan (1990: 163) discussing the productive lexicon in British Sign Language, calls these mix n match signs : mix n match involves selecting the component parts and putting them together in appropriate ways to create particular kinds of effect. In the literature on Flemish Sign Language, productive lexical items are usually called gebarenconstructie in Dutch or sign construction in English; sign constructions functioning as a predicate are labelled werkwoordelijke constructie or verb construction (vc). They are normally transcribed with lower-case letters and quotation marks, e.g. vc: eating-a-pieceof-cake : vc: eating-a-piece-of-cake

Verbal Predicates in Flemish and South African Sign Language When describing a specific action, signers often have a choice between a description using one or more (established) verb signs or a (productive) verb construction. This can be illustrated by the difference between (1) MOTHER TAKE BOWL TAKE (cf. also http://gebaren.ugent.be/alfabet.php?id=18213) and (2) MOTHER BOWL vc: take-bowl-using-both-hands vc: take-bowl-using-both-hands Signers may also decide to combine both a lexical verb and one of its productive counterparts in one and the same sentence (see Section 3.5). An exhaustive discussion of the established versus productive lexicon in signed languages lies beyond the scope of this paper (see Vermeerbergen (2006) for some important observations on this subject). One important issue to mention here, because of its relevance to the discussion of verb sandwiches later in this chapter,

Mieke Van Herreweghe & Myriam Vermeerbergen is the relationship between word order and the use of a verb sign/verb construction. Vermeerbergen et al. (2007:46) notice the following: From previous studies on word order issues in VGT it seems that (probably more in non-reversible than in reversible sentences) there might be a difference between sentences containing (plain) lexical verbs and sentences with verb constructions. Briefly, verb constructions tend to appear sentence-finally, while (plain) lexical verb signs tend to be produced earlier in the sentence and follow the subject. Some processes adding meaning to a lexical verb (e.g. the simultaneous representation of the object on the nondominant hand) seem to result in the verb being placed at the end of the sentence as well (Vermeerbergen 1996; 2004). Observations concerning certain types of verbs behaving differently as far as word order is concerned have been made by other researchers for the sign languages they study (e.g. by Sze 2003 for Hong Kong Sign Language). The analysis of word order in non-locative sentences in VGT presented in Vermeerbergen et al. (2007) indeed shows that verb constructions as well as lexical verbs incorporating the object (either because the object is referred to on the non-dominant hand or because the location or orientation of the verb sign is modified so as to refer to the location of the object) are produced sentence-finally, whereas the verb sign in the SVO-constructions is without exception a plain lexical verb. This tendency does not hold for SASL where verb signs and verb constructions seem to always occur sentence-finally. 3.2. SIT, STAND and KNEEL-DOWN A first type of sentences that needs to be examined in more detail comprises sentences with SIT, STAND and KNEEL-DOWN on the one hand 2 and sentences with localising classifiers 3 on the other. Both may attribute a specific 2 Other positioning verbs (e.g. HANG) may be involved as well, but we have limited ourselves here to those verbs that were encountered in the data. 3 The term classifier is used in the signed language literature to refer to a handshape or a combination of a handshape with a specific orientation which represents a referent. Although the parallel between classifiers in signed languages and one of the four types of spoken language classifiers in Allan s (1977) typology often referred to in early publications has later been questioned, the use of the term for signed languages is still widespread. We refer the reader to Schembri (2003) for an in-depth discussion.

Verbal Predicates in Flemish and South African Sign Language locus to the referent involved in the signing space, but only SIT, STAND and KNEEL-DOWN are generally regarded as verbs. An examples in VGT is: (3) GIRL SIT CAKE vc: eat-cake One possible explanation for the occurrence of this type of sentence in VGT is that they are influenced by comparable constructions with positional verbs in Dutch ( Het meisje zit cake te eten. / The girl sits eating cake ). SIT etc. could then be regarded as semi-auxiliaries used to express a durational aspect as in Dutch, i.e. the girl is eating cake and sits down while doing so. Whereas borrowings from Dutch are indeed attested in VGT (Akach et al., 2009), we hesitate to analyse SIT and STAND as related to their Dutch counterparts, mainly because there are similar structures in other signed languages (see the SASLexamples in this study for instance) where the surrounding spoken languages (e.g. English, Afrikaans, etc.) do not feature the structures present in Dutch. Moreover in VGT KNEEL-DOWN can also be used in a similar manner: (4) MAN KNEEL-DOWN DOG vc: wash but kneel-down cannot be used as a semi-auxiliary in Dutch ( *De man knielt de hond te wassen / The man kneels-down washing the dog ). In sentence (3) above SIT could also be regarded as a postmodifying adjective ( the girl sitting ) or as an embedded clause ( the girl who is sitting ) -although for the latter we would expect a different non-manual marking- (cf. Johnston et al., 2007), but in our opinion it seems to be very similar to a classifier functioning as a localising device (with semantic content). In the SASL-data, the latter occurred even more frequently than the SIT and STAND constructions. In the SASL-data, the description of the non-reversible states of affairs fairly often resulted in rather complex sentences, including sentences where the articulation of the sign for the object or the subject is followed by the production of a classifier (construction), as can be seen in the following example: (5) Right hand: WOMAN------------------------------ STRING vc: cut-string Left hand: woman-located-at-loc a STRING-------------------- After the sign WOMAN (signed with the right hand), the left hand produces a classifier construction by means of which the referent woman is located in the signing space as is shown here:

Mieke Van Herreweghe & Myriam Vermeerbergen WOMAN------------------------------- woman-located-at-loca In such examples, the classifier functions as a kind of localising device i.e. it attributes a spatial locus/specific location within the signing space to the referent. All SASL-signers produced at least one example of this. In our VGT-data, the use of classifiers as localising devices is rare, instead localising verbs such as SIT are used. Such localising verbs do not only express the fact that the girl is sitting, but also attribute a locus to GIRL which may later be used to refer back to. The localising function is clearer in sentences with two animate arguments as in the next example: (6) BOY SIT a // MOTHER STAND b COMB a BOY SITa MOTHER STANDb

Verbal Predicates in Flemish and South African Sign Language COMBa Here the boy is localised by means of the verb SIT in front of the signer, and the mother by means of the verb STAND (and by the fact that the signer slightly leans back) relative to the locus attributed to BOY, i.e. behind the boy (in order to be able to comb his hair). As such, STAND as it is used here seems to have more in common with constructions involving the use of a classifier handshape following the production of a noun and used to attribute a locus to the referent, as in the following SASL example: (7) BOY boy-located-at-loca GIRL girl-behind-boy vc: comb-hair a COMB -self vc: comb-hair a girl-behind-boy In this sentence ( The girl combs the boy s hair ) the boy is localised in front of the signer, i.e. loc a, by means of a classifier handshape used to refer to persons in SASL (i.e. a kind of thumbs-up sign). The second thumb, referring to the girl, is localised behind the first so that we know that the girl is positioned behind the boy. Afterwards the signer signs that the girl is combing the boy s hair. Instead of classifier handshapes simple pointing signs would have been perfectly acceptable here as well. Therefore verbs like SIT, STAND and KNEEL-DOWN seem to have a similar function to classifier handshapes following the production of a noun in order to attribute a locus to the referent. The only difference is that more information is

Mieke Van Herreweghe & Myriam Vermeerbergen given about the posture of the referent by means of the verb (sitting, standing, kneeling down, etc.) whereas this is less important or sometimes even totally irrelevant when classifier handshapes or pointing signs are used. Hence elements like SIT, STAND and KNEEL-DOWN can be analysed in two different ways: either as localising verbs with positional content or simply as localising devices (not with any verbal function). In the first case utterances (3) (GIRL SIT CAKE vc: eat-cake ) and (4) (MAN KNEEL-DOWN DOG vc: wash ) have to be analysed as SVOvc, in the second case as SOvc, a remarkable difference. To make things even more complex the following example is added in which it is very difficult to decide whether the V-hand (as in the second photograph below) stands for the verb sign STAND or is simply a localising classifier handshape (in Flemish Sign Language the same handshape is used for both): (8) COWBOY V-hand loc a / NATIVE-AMERICAN STAB a COWBOY cl:v-hand loca In this sentence ( The cowboy is stabbed by the Native-American ) the cowboy is located in front of the signer, i.e. loc a, by means of the right hand which can be regarded as the verb sign STAND or as a classifier handshape used to refer to persons (the left hand constitutes a fragment buoy of the sign for cowboy). This locus is then used as end point of the sign STAB a indicating that the cowboy is being stabbed. Instead of a classifier handshape a simple pointing sign would have been perfectly acceptable here as well. Again this has implications when discussing word order in this utterance. When treating the V-hand as the verb sign STAND we obtain the order SVOV but when treating it as a localising device we obtain the order SOV. Therefore it is very important for researchers to clearly explain how they treat signs like SIT, STAND and KNEEL-DOWN in their analyses of word order issues. Unfortunately, this was not always done in signed language research using

Verbal Predicates in Flemish and South African Sign Language the Volterra materials so that comparative research between these studies is sometimes difficult to even impossible (cf. also Johnston et al., 2007). 3.3. GIVE A second problematic type comprises examples featuring GIVE (which only occurred in VGT and not in SASL). One example is: (9) BOY GIVE GIRL PUSH The question is which grammatical status can be attributed to GIVE: ditransitive lexical verb (with both GIRL and PUSH as objects), monotransitive lexical verb in a split sentence construction (see 3.4), light verb, auxiliary, or preposition (cf. Van Herreweghe & Vermeerbergen, 2004). In previous research on VGT word order, similar examples with GIVE, but also ON, were identified and discussed (Vermeerbergen 1996). What is striking is that ON and GIVE seem to have a similar semantic function, but seem to appear in different word orders: ON tends to introduce the object (i.e. the recipient) in a SVO-structure, whereas GIVE seems to introduce the object (i.e. the recipient) in a SOV-structure. Previous analysis has also demonstrated that GIVE can be spatially modified in order to show agreement with both arguments and can in this way formally illustrate the direction of the action. However, the movement of GIVE does not need to be modified i.e. GIVE can also be used in its citation form. Van Herreweghe & Vermeerbergen (2004) claim that even though GIVE in these sentences seems to be very similar to the directional verb GIVE, the sign as it is used here appears to have a different function. The semantic meaning of GIVE, although not altogether empty, seems to be less important than the grammatical function: GIVE relates the agent of the action (i.e. the referent named just before GIVE) to the recipient of the action (i.e. the referent named after GIVE) and as such GIVE explicitly indicates the direction of the action. The same function was attributed to ON, generally considered a preposition, in similar structures. Comparable examples have been described for a number of other signed languages e.g. Sign Language of the Netherlands, German Sign Language, Irish Sign Language, Catalan Sign Language, Greek Sign Language and Australian Sign Language (Vermeerbergen & Leeson, 2010). Johnston et al. (2007) discuss the use of ON preceding the undergoer in sentences with reversible arguments in Australian Sign Language and the production of TO after the verb and before the second argument in the same type of sentences in Irish Sign Language. Both AUSLAN ON and ISL TO are labelled prepositions. However, for Sign Language

Mieke Van Herreweghe & Myriam Vermeerbergen of the Netherlands (NGT) Bos (1994) discerns an element with a similar semantic function (i.e. recipient marker) but calls it an auxiliary verb, viz. ACT-ON. Steinbach & Pfau (2007) discuss similar elements in a number of signed languages in terms of different grammaticalisation processes, with some elements derived from verbs and others from prepositions. Clearly, these recipient markers are not typical of VGT only. Van Herreweghe (2010) claims that it might be possible to analyse different instances of GIVE as different stages of a grammaticalisation process. Following the principle of divergence (Hopper 1991:24) the different forms can exist next to each other at the same time and the variants can be put in a hierarchy from less to more grammaticalised (the following examples are taken from De Vriendt, 2009): (a) Modified verb sign GIVE with an incorporated classifier handshape referring to the direct object: SOMEONE BOOK GIVE -cl.book TO BOY ( Someone gives a book to a boy and the thickness of the book is indicated by means of the classifier handshape); (b) Conventionalised sign GIVE with no incorporated handshape, but with spatial agreement with subject and indirect object: SOMEONE BOOK 1GIVE 3l TO BOY ( Someone gives a book to a boy ); (c) GIVE in a verb sandwich construction (see section 3.5) in which the first GIVE is the conventionalised citation form without any spatial agreement (so more grammaticalised); the second one has an incorporated classifier handshape and spatial agreement as in (a): WOMAN GIVE PRESENT TO BOY 1 GIVE -cl.present3r ( A woman gives a present to a boy, and the size of the present and the way in which she handles it is indicated by means of the classifier handshapes); (d) Auxiliary GIVE with spatial agreement but reduced semantic content functioning as recipient marker: INDEX m RABBIT 1 GIVE 3l MAN SHOOT ml ( A rabbit shoots a man ); (e) Preposition GIVE without spatial agreement and a reduced phonological form (simply a fairly short horizontal forward movement starting from the signer) introducing the recipient of the action: RABBIT GIVE MAN SHOOT ml ( A rabbit shoots a man ). Following Van Herreweghe (2010) it seems that recipient marker GIVE functions in some of the examples encountered in the present data as an auxiliary (with

Verbal Predicates in Flemish and South African Sign Language spatial agreement) and in others as a preposition (with a reduced form) resulting in the following word orders respectively: SauxOV and SOV. Again comparison with other signed languages is only possible when it has been made clear whether they consider similar recipient markers as prepositions, auxiliary verbs or even other elements. 3.4. Split Sentences Split sentences make up a third problematic type which occurs in both signed languages. Both in the VGT and SASL-data elicited for the study of word order, there are some examples of sentences with two verbs/verb constructions each preceded by one argument e.g. (10) CAR TOW TRUCK FOLLOW (VGT) (11) BOY PUSH GIRL vc: stumble-forward (SASL and VGT) (12) MAN KNEEL-DOWN DOG vc: wash (VGT) Vermeerbergen (1996) analyses such examples as split sentences (cf. Volterra et al., 1984) containing two clauses, each consisting of a subject and a predicate. Such an interpretation seems acceptable in example (10), but especially for example (12) such an analysis does not seem to be acceptable for all (signed) linguists. The main argument against such an interpretation is that in this reading the relation between the second subject DOG and its predicate vc: wash is not one between the referent who carries out the action (the logical subject) and the action. To date, it is not yet fully clear whether and how the notion grammatical subject needs to be defined for Flemish Sign Language. In most studies analysing the syntax of Flemish Sign Language the definition of subject as logical subject is adopted: the subject is the constituent indicating the referent which performs the described action, or, if no action is described, the constituent indicating the referent to which the condition, the process, the quality, etc. named by the verb is attributed (Vermeerbergen, 1997a). If one applies this definition to the arguments in the sentences above, it is not always possible to interpret the second arguments as subjects. However, if one is willing to accept a psychological definition of subject here, i.e. the particular about whom/which knowledge is added (cf. Chafe, 1976: 44), it does seem possible to analyse the second argument in all of the above sentences as the (psychological, but not logical) subject of the second verb (Vermeerbergen,

Mieke Van Herreweghe & Myriam Vermeerbergen 1997a). The transliteration of sentence (12) would then be something like: The man kneels down and the dog is being washed, implying that the idea that it is the man who is washing the dog is not overtly expressed. Vermeerbergen (1997a) discusses similar subject + verb/predicate combinations. One example is: (13) LAST WEEK MY WIFE MY TOOTH vc: extract-tooth / AFTER POORLY Last week my wife had a tooth extracted, afterwards she felt poorly/unwell. When the definition of logical subject is applied, this sentence does not contain any subject at all. We know that sentences in which no subject is overtly expressed are not exceptional in Flemish Sign Language (Vermeerbergen, 1996). In example (13), informants do identify the constituent MY WIFE as the subject of the verb construction. However, MY WIFE does not refer to the referent performing the action referred to by the verb construction extract-tooth, i.e. the constituent MY WIFE is the psychological but not the logical subject. Engberg-Pedersen (1993:153) discusses similar ambiguous clauses with a single argument and a verb/predicate in Danish Sign Language and claims that such examples point in the direction of a pragmatic analysis of the verb s arguments in Danish Sign Language as can be found in topic-prominent languages. She refers to the work of Chao (1968) on spoken Chinese and the English translation of Chinese sentences. The English translation of the sentence Jeh-jiann shyt tzao fabeaule is This matter has long been published but that is not 100% accurate because fabeaule is translated by a passive form while in the Chinese sentence there is no (formal) indication that there is received action. A more accurate rendering would be: (As for) this matter, (one) has long published (it). According to Chao [a]ll such renderings in English, however, are limited by the exigencies of English grammar requiring a clear actor-action relation, (...), thus entailing a number of parenthetical devices which never were in Chinese, which simply said: This matter has long ago published (Chao, 1968:70). In example (10), the subject of the first clause also functions as the object (undergoer or patient) of the verb in the second clause. This is often the case for split sentences, however, not all argument verb/vc argument verb/vc combinations can be interpreted in this way. In some examples, it is the subject of the second clause which is also the object of the verb of the first clause. In these

Verbal Predicates in Flemish and South African Sign Language structures, especially if there is no clear indication of a clause break, the second argument can be seen as a pivot functioning at the same time as the O of the first verb and the S of the second, as in the following example: (14) BOY PUSH GIRL vc: stumble-forward This is something that is quite common in (spoken) languages across the world and it is often thought that such pivot constructions are the predecessors of relative clauses (Lehmann, 1992). Split sentences show even more clearly that a researcher s specific analysis of the constituents in the sentence has important consequences for the analysis of constituent order. 3.5. Verb sandwiches The notion verb sandwich was probably introduced by Fischer and Jannis (1990) discussing these constructions in American Sign Language. Fischer and Jannis (1990: 287) schematize a verb sandwich as: SUBJECT VERB (OBJECT) (ADJUNCTS) VERB+LOTS-OF-INFLECTION.... The prototypical verb sandwich includes one subject followed by two instances of the same verb, with the second instance of the verb carrying additional information. Fisher and Janis (1990) present examples from American Sign Language, showing that there are a number of different ways in which the second verb can carry additional information, e.g. expression of plurality, inclusion of a classifier and aspectual information. The latter case is illustrated with the following example: (15) SALLY THERE HMM TYPE -T-E-R-M-PAPER TYPE (asp: unrealized inceptive)... Interestingly, discussing verb sandwiches in Croation Sign Language, Milković, Bradarić-Jončić and Wilbur (2006: 184-185) too present an example where the second instance of the verb includes aspectual information: it is produced with a repeated movement, indicating iterative aspect. Indeed, the literature on signed languages shows that many different, related and unrelated, signed languages share this construction (Vermeerbergen & Leeson, 2011). According to Fischer & Jannis, the existence of this construction is motivated by the amount of information one single instance of the verb can carry. When the verb becomes too heavy, for example because there are too many morphological modifications, the signer introduces a second instance of the same

Mieke Van Herreweghe & Myriam Vermeerbergen verb to help carry some of the load. Another reason to split the verb may be the wish to use two different classifiers, which cannot be combined in one single form. However, both the analyses of similar constructions in the signed language literature, as well as our own data, show that there are also examples of verb sandwiches showing two identical instances of the same verb. Two examples from our own study are: (16) GIRL SIT LOOK-AT -T-V- LOOK-AT (for SASL) (17) CAR TOW TRUCK TOW (for VGT) This has also been called verb echo (Pinsonneault, 1994). Yet other examples combine a lexical verb and a related productive verb construction, e.g. (18) GIRL EAT CAKE vc: eat-cake (VGT) Here we see the lexical verb sign EAT in combination with the productive verb construction vc: eat-cake. It seems possible that the existence of the latter construction is connected to the verb construction = sentence-finally (plain) lexical verb sign = early/earlier in the sentence -tendency we discussed earlier in this chapter (cf. also Vermeerbergen, 1996) but we should not forget the examples with two identical instances of the same verb. Again, different interpretations of the verbs in this structure and of the constituent order are possible. Johnston et al. (2007) analyse these sentences as A1 V A2,V i.e. they counted the second verbal element as a separate clause. However, they also point out that it is possible that these strings represent a distinctive form of constituent ordering and structuring apparently found in many signed languages, as argued by Vermeerbergen (1996). Volterra et al. (1984) analyses constituent order in these sentences as SVO, treating the final verb as a repetition of the first one. Again, different analyses will lead to different descriptions of constituent order and researchers should be more explicit about their interpretations. 4. Conclusion Our study of constituent order in Flemish Sign Language and South African Sign Language showed that there are important similarities between the two signed languages studied, some of which were not really expected since they are

Verbal Predicates in Flemish and South African Sign Language unrelated, but at the same time it also shows that there are remarkable differences (Vermeerbergen et al., 2007). A larger-scale study, involving other types of language data and more signers, is needed in order to get a full(er) picture of constituent order in VGT and SASL and the cross-linguistic comparison. The study also showed, once more, that analysing word order is far from simple, even in a small study like this with a limited set of data consisting of elicited declarative sentences only. Identifying word order implies making decisions, decisions on clause boundaries, decisions on how to interpret the relationship between the verb and a specific argument (cf. the discussion of subject above), decisions on how to deal with simultaneity, etc. Johnston et al. (2007) discuss a broader range of practical problems and theoretical issues that arise from the study of word order. In this paper, we have concentrated on the problems that relate to the analysis of (would-be) verbal elements. We have shown that often different analyses are possible, i.e., although we ourselves may show a preference for a specific analysis, other interpretations are possible as well. And of course, different decisions on what constitutes a verb and what not, will lead to different analyses of constituent order. The lack of consensus when it comes to labelling constituents, structures, elements,... in signed language research in itself make it very hard to conduct cross-linguistic work on signed languages and to establish typological generalisations. Moreover, all too often authors do not include (enough) information on how exactly they use a specific notion and/or on the several decisions they make when analysing signed language data. Because of this what look like differences across different signed languages, may actually be differences in the interpretation of the raw data. Another important insight resulting from our discussion presented here relates to the description of verbs in signed languages. Traditionally, the discussion of verb classes includes a discussion of plain verbs, agreement verbs (or indicating verbs) and classifier predicates. Generally speaking, there is only limited attention for categories such as auxiliaries and localising verbs in signed languages. As is shown here, there clearly is a need for more work on verbal and related categories in signed languages. References AKACH, P., DEMEY, E., MATABANE, E., VAN HERREWEGHE, M. & VERMEERBERGEN, M. (2009). What is South African Sign Language? What is the South African Deaf community? In B. BROCK-UTNE & I. SKATTUM (eds), Languages

Mieke Van Herreweghe & Myriam Vermeerbergen and Education in Africa: a comparative and transdisciplinary analysis. Bristol: Symposium Books, 333-347. ALLAN, K. (1977). Classifiers. Language, 53: 285-311. BOS, H. 1994. An Auxiliary Verb in Sign Language of the Netherlands. In I. AHLGREN, B. BERGMAN & M. BRENNAN (eds), Perspectives on Sign Language Structure. Papers from the Fifth International Symposium on Sign Language Research. Volume 1. Durham: ISLA, 37-53. BOYES-BRAEM, P., FOURNIER, M.-L., RICKLI, F., CORAZZA, S., FRANCHI, M.-L. & VOLTERRA, V. (1990). A Comparison of Techniques for Expressing Semantic Roles and Locative Relations in Two Different Sign Languages. In W.H. EDMONDSON & F. KARLSSON (eds), SLR 87: Papers from the Fourth International Symposium on Sign Language Research. Hamburg: Signum, 114-120. BRENNAN, M. (1990). Word Formation in British Sign Language. Stockholm: University of Stockholm. CHAFE, W. (1976). Givenness, Contrastiveness, Definiteness, Subjects and Topics. In C. LI (ed.), Subject and Topic. New York: Academic Press, Inc., 25-55. CHAO, Y. (1968). A Grammar of Spoken Chinese. Berkeley: University of California Press. COERTS, Jane (1994). Constituent Order in Sign Language of The Netherlands. In M. BRENNAN & G. TURNER (eds), Word-order Issues in Sign Language. Durham: International Sign Linguistics Association, 47-72. DE VRIENDT, T. (2009). Vlaamse Gebarentaal (VGT): De grammaticale functie van het werkwoord GEVEN en de voorzetsels OP, AAN, NAAR en TEGEN. Unpublished master dissertation, Ghent University. ENGBERG-PEDERSEN, E. (1993). Space in Danish Sign Language. Hamburg: Signum- Verlag. FISCHER, S.D. & WYNNE, J. (1990). Verb Sandwiches in ASL. In S. PRILLWITZ & Th. VOLLHABER (eds), Current Trends in European Sign Language Research: Proceedings of the 3rd European Congress on Sign Language Research, Hamburg, 1989. Hamburg: Signum, 279-294. HOPPER, P.J. (1991). On some Principles of Grammaticization. In E.C. TRAUGOTT & B. HEINE (eds), Approaches to Grammaticalization, vol. I. Amsterdam: John Benjamins. 17-35. JOHNSTON, T., VERMEERBERGEN, M., SCHEMBRI, A & LEESON, L. (2007). Real data are messy : Considering Cross-Linguistic Analysis of Constituent Ordering in Australian Sign Language (Auslan), Vlaamse Gebarentaal (VGT), and Irish Sign

Verbal Predicates in Flemish and South African Sign Language Language (ISL). In P. PERNISS, R. PFAU & M. STEINBACH (eds), Visible Variation: Comparative Studies on Sign Language Structure. Berlin: Mouton de Gruyter, 163-205. LEESON, L. (2001). Aspects of Verbal Valency in ISL. Ph.D. dissertation, Centre for Language and Communication Studies, Trinity College, University of Dublin, Dublin. LEHMANN, W.P. (1992 3 ). Historical Linguistics. London: Routledge. MILKOVIĆ, M., BRADARIĆ-JONČIĆ, S. & WILBUR, R.B. (2006). Word Order in Croatian Sign Language. Sign Language & Linguistics, 9(1-2): 169-206. PINSONNEAULT, D. (1994). Verb Echoes in LSQ (Langue de Signes Québécoise). In M. BRENNAN & G.H. TURNER (eds), Word-order Issues in Sign Language - Working Papers. Durham: ISLA, 113-132. SAEED, J.I., SUTTON-SPENCE, R. & LEESON, L. (2000). Constituent Order in Irish Sign Language and British Sign Language: a Preliminary Examination. Poster presented at the Seventh International Conference on Theoretical Issues in Sign Language Research (TISLR 7), Amsterdam, July 2000. SCHEMBRI, A. (2003). Rethinking Classifiers in Signed Languages. In K. EMMOREY (ed.), Perspectives on Classifier Constructions in Sign Languages. Mahwah, N.J./London: Lawrence Erlbaum Associates, 3-34. STEINBACH, M. & PFAU, R. (2007). Grammaticalization of Auxiliaries in Sign Languages. In P.M. PERNISS, R. PFAU & M. STEINBACH (eds), Visible variation: Comparative Studies on sign language structure. Berlin and New York: Mouton de Gruyter, 303-339. SZE, F.Y.B. (2003). Word order of Hong Kong Sign Language. In A. BAKER, B. VAN DEN BOGAERDE & O. CRASBORN (eds), Cross-linguistic Perspectives in Sign Language Research. Selected Papers from TISLR 2000. Hamburg: Signum, 163-192. VAN HERREWEGHE, M. (2010). Grammaticalisation Processes in Flemish Sign Language. Poster presented at the 10th International Conference on Theoretical Issues in Sign Language Research (TISLR 10), West Lafayette, Indiana, USA, 30 September 2 October 2010. VAN HERREWEGHE, M. & VERMEERBERGEN, M. (2004). The Semantics and Grammatical Status of Three different Realizations of GIVE (GIVE): Directional Verb, Polymorphemic Construction and Auxiliary/Preposition/Light Verb. Poster presented at TISLR 8, Eight International Conference on Theoretical Issues in Sign Language Research, Barcelona, Spain, 30 September-2 October 2004. VERMEERBERGEN, M. (1996). ROOD KOOL TIEN PERSOON IN. Morfosyntactische aspecten van gebarentaal. Doctoral dissertation, Vrije Universiteit Brussel, Brussel.

Mieke Van Herreweghe & Myriam Vermeerbergen VERMEERBERGEN, M. (1997a). Some Remarks on the use of the Notions Subject, Object and Topic in Sign Language Research. Leuvense Bijdragen, 86(4): 505-512. VERMEERBERGEN, M. (1997b). Grammaticale Aspecten van de Vlaams-Belgische Gebarentaal. Gentbrugge: Cultuur voor Doven. VERMEERBERGEN, M. (2004). The Quest for Basic Word Order in Flemish Sign Language. In A.-M. BERTONNEAU & G. DAL (ed), La linguistique de la LSF: Recherches Actuelles. Silexicales, 4. Villeneuve d Ascq: Université de Lille 3, 257-267. VERMEERBERGEN, M. (2006). Past and Current Trends in Sign Language Research. Language & Communication, 26(2): 168-192. VERMEERBERGEN, M. & LEESON, L. (2010). European Signed Languages Towards a Typological Snapshot. In B. KORTMANN & J. VAN DER AUWERA (eds), Field of Linguistics: Europe. Berlin: Mouton de Gruyter, 269-287. VERMEERBERGEN, M., VAN HERREWEGHE, M., AKACH, P. & MATABANE, E. (2007). Constituent order in Flemish Sign Language and South African Sign Language. A cross-linguistic study. Sign Language & Linguistics, 10(1): 25-54. VOLTERRA, V., LAUDANNA, A., CORAZZA, S., RADUTSKY, E. & NATALE, F. (1984). Italian Sign Language: the Order of Elements in the Declarative Sentence. In F. LONCKE, P. BOYES-BRAEM & Y. LEBRUN (eds), Recent Research on European Sign Languages. Lisse: Swets & Zeitlinger B.V., 19-48.