Ordered alphabetically according to first author's surname.
For workshop and plenary paper abstracts, see separate pages.
According to Comrie (1985: 9), tense is 'grammaticalised expression of location in time'. Grammaticalisation involves the incorporation of certain (lexical) forms into the grammatical framework of a language. To be considered as grammatical forms, they must display two features in particular: morphological boundness and obligatory expression. By admission of the author himself, though, this definition "is a prototype definition, rather than a definition in terms of necessary-and-sufficient conditions" (1985: 10, n. 7), since languages not only show forms that are clearly either grammaticalised or lexicalised, but also forms that are not identifiable as belonging clearly to either of the two groups.
Another important issue is that tense is a deictic category, a system '... that relates entities to a reference point' (1985: 14). The reference point is typically the moment of speech, ie. '… the present moment (for time), the present spot (for space), and the speaker and hearer (for person)' (1985: 14). In his chapter referring to absolute tense, identified as '… a tense which includes as part of its meaning the present moment as deictic centre' (1985: 36), Comrie devotes a paragraph to those languages defined as tenseless, that is those which do not display grammatical/grammaticalised forms of temporal localisation. Among these languages he includes Burmese, by taking into account two sets of particles, the realis marker [tE] with its variants [thA*/hta], which refer to actual situations, and the irrealis marker [mE] with its variants [mA/hma], which refer to non-actual situations (Chung & Timberlake 1985: 241). These forms are sentence operators (Van Valin & La Polla 1997: 40-52), ie. they typically occur at the end of the sentence, and may be followed by markers of evidentiality, polarity, etc.
The situation in Burmese, though, is not as simple as it may seem at first glance. There are in Burmese locative/directional particles that are used as markers of localisation in time, as is clear from the analysis of forms by John Okell (1969). These particles are not necessarily linked to the verbal complex alone, but can also appear attached to lexical forms that normally occur at the beginning of the sentence. These lexical forms diffuse temporal information to the rest of the sentence, information that is further specified by the use of the above-mentioned realis/irrealis markers.
To illustrate this point, let us consider two particles in particular, [ka2] and [hma]. According to Okell (1969: 315-319), [ka2] is a directional marker corresponding to the English preposition from. What is interesting is that it is also used to mark what Okell glosses as past time (1969: 316-317):
| 1) | sAnene2-ga2 | twe2-dE |
| Saturday-PAST TIME | meet-REALIS | |
| '[We] met on Saturday' | ||
| 2) | hpehpe | shi2-doun-ga2 | lwE-dE |
| father | exist-when-PAST TIME | be easy-REALIS | |
| '[It] was easy when father was alive' | |||
| 3) | bE-doun-ga2 | jauQ-thA-lE1 |
| which-when-PAST TIME | arrive-REALIS-INTERR | |
| 'When [in the past] did [you] arrive?' | ||
| 4) | eindaun | mA-cha2-the1-doun1-ga2 | pyO-dE |
| household | NEG-be in-yet-when-PAST TIME | be happy-REALIS | |
| '[I] was happy when [I] was not married' | |||
[hma] is a locational marker that translates the English prepositions at, in, on. It is used in expressions with future time reference, associated with the IRREALIS sentential marker [mE]:
| 5) | sAnene2-hma | twe2-mE |
| Saturday-at | meet-IRREALIS | |
| '[I] shall see [him] on Saturday' | ||
The question then is: does Burmese have tense? Do we have to rule out the possibility for Burmese to be considered as a tensed language only because it does not necessarily display temporal anchoring onto the verbal complex alone? We will see how deixis and the relation between space and time play a major role in the description of the Burmese temporal system. I will try to address these and other issues while trying to devise an ontological mapping of temporal information in Burmese.
This paper introduces Andajin, a highly endangered language of the Kimberley, about which very little is known, the literature containing only sparse information.
The paper discusses initial fieldwork carried out on the language with possibly the last full speaker. Traditional Andajin country was interestingly surrounded by three languages from three different families: Wurla, a dialect of Ungarinyin (Worrorran family) to the west, Bunuba (Bunuban family) to the south-west, and Gija (Jarrakan family) to the east (McGregor 1988). Andajin seems to have linguistic connections to each of these languages.
Research on Andajin is in its early stages. This paper will discuss comparatively Andajin's relationship to the other languages. Phonologically Andajin is almost identical to Wurla, but has a sixth vowel a high central unrounded vowel. From the little grammatical information collected, Andajin seems to share a translative case function and proximad/distad verbal prefixes with Ungarinyin (Rumsey 1982:64-67, 110-112). This suggests it belongs to the Worrorran family. However Andajin has a lot of shared vocabulary with Gija, Bunuba, and with Gooniyandi the other Bunuban language as well as Wurla/Ungarinyin.
The emphasis will be be on vocabulary comparison, discussing lexical borrowing from the other languages. It is suggested that Andajin is a member of the Worroran family but has been heavily influenced by its neighbours through lexical borrowing. It has been minimally influenced phonologically. Some morphological borrowing has occurred resulting in some hybrid constructions. Reasons for borrowing and problems of classification will also be discussed.
It has been observed that in a number of Australian languages, a clitic marks the temporal relationship between a secondary predicate and the main predicate (Dench and Evans 1988: 14). This paper presents a detailed discussion of the functions of the clitic =oong in Jaminjung, and compares these functions to the functions of similar clitics in other languages.
Jaminjung =oong is found most frequently on both depictive and resultative secondary predicates, as in (1) and (2) below.
| (1) | nginy=oong | na-rooma-ny | woorrgoorroo=marraj |
| bare.teeth=CLITIC | 2sg-come-PAST | devil=LIKE | |
| ‘showing your teeth you came like a devil’ | |||
| (2) | jalag=oong | ga-w-irna, | klinbala=oong |
| good=CLITIC | 3sg-FUT-burn/cook | clean=CLITIC | |
| ‘it should cook (until it’s) good, (and) clean’ | |||
The same clitic, however, occurs in a wide range of other morphosyntactic contexts. For example, it can follow the main predicate, translating as ‘still’). It also occurs on local and temporal adverbials, and in some cases has to be regarded as lexicalised.
The contexts identified for Jaminjung =oong will serve as a starting point in comparing the range of uses of clitics occurring on secondary predicates in a number of other Australian languages. Although the functions of the clitics in these languages overlap to only some extent, they always include functions that have been subsumed under ‘expectation modifiers’ by McConvell (1983), i.e. the clitics translate in at least some contexts as ‘still’ or ‘only’. The findings suggest that clitics on secondary predicates, rather than marking a purely temporal relationship to the main predicate, originate as expectation modifiers.
There has recently been a concern over the scientific status of the linguistic theory (Botha, 1989; Yngve, 1996; Moshfeghi & Sharifian, in press). This paper argues that to achieve the status of an empirical science, linguistics should employ the hypothetico-deductive method of research introduced by Popper (1968). In this method, the hypothesis proposed by the scientist is considered as the major premise, to use the syllogistic terms; the minor premise would be a statement made about a real observation, and finally a conclusion statement is deduced from these premises. The conclusion statement is then tested for its validity. If the conclusion statement is true, then the hypothesis is said to have survived and if the conclusion statement turns out to be wrong, then the hypothesis would be falsified. An example may clarify this procedure. Consider the following sentence and a possible argument, formulated as a syllogism:
(1) We know [the president will approve the project], and [that congress will ratify his decision].
Major Premise: Only constituents belonging to the same category can be conjoined.
Minor Premise: The second italicized clause in sentence (1) is an S-bar (as it contains an overt complementiser, that).
Conclusion statement: The first italicized clause is an S-bar
It is now possible to follow the above-mentioned argument in two different ways:
Many people still think of pidgins, creoles and minority dialects (such as Aboriginal English) as corrupted or degenerate forms of the standardized language to which they are lexically related. Although these stigmatized varieties have been shown to be legitimate rule-governed forms of language, they are still banned from most classrooms because of persisting negative attitudes and fears that they will interfere with the students' acquisition of the standard. This paper critically examines this notion of interference. It begins by describing research on three types of educational programs in which stigmatized varieties have actually been used in the classroom. This research shows that contrary to the prevailing viewpoint, using the stigmatized variety in formal education seems to have a positive rather than a negative effect on the acquisition of the standard variety. One of the reasons proposed to explain this effect is that such educational programs enable students to "separate" the two varieties. The paper goes on to look at some research in psycholinguistics and second language acquisition regarding the notion of separation, and then at various teaching approaches which promote it. It argues that by looking at features of their own varieties, students are induced to notice features of the standard which are different. This awareness is necessary for developing a separate mental representation of the standard. Thus interference is reduced rather than exacerbated.
A striking feature of creoles and pidgins spoken by Aborigines in Australia is their relative similarity in structure and the amount of shared vocabulary. For pidgins to stabilise in similar ways over wide areas, showing similar features, it helps for there to be people carrying the conventions, acting as vectors for the diffusion of the features. The major vectors in Australia must have been English-speaking colonists and Aborigines, followed by Chinese-speaking colonists. However, for over seventy years, 'Afghans' travelled great distances over large parts of Australia, coming into contact with Aborigines at a time when many Aborigines spoke other languages, and English (or pidginized varieties of English) was used as the lingua franca. This, combined with the fact that English was a second language for the Afghans, and their close relations with Aborigines made the learner English of the Afghans a particularly suitable vector for the diffusion of features. Based on nineteenth and early twentieth century renditons by speakers of English and Aboriginal languages I examine features attributed to Afghans and compare them with markers of Aboriginal pidgins and creoles. This leads to a discussion of the representation of Afghans and the reliability of the data.
Andersen (1997) has claimed that “markedness relations can be observed in every variety of linguistic change, from its inception to its completion, both in the relations among variants and in the relations that that define the plethora of categories that typically condition the gradual process by which newer forms replace older correspondents”. In this paper, I shall discuss the extent to which the differential disappearance of agreement between the past participle and a direct object in the Romance compound past tenses formed with the auxiliary ‘have’ may be an instance of a linguistic change which is sensitive to markedness. Drawing on earlier work (Smith 1995; 1997) suggesting that this morphosyntactic development is due to perceptual factors and processing strategies, which sometimes yield patterns conforming to traditional notions of markedness and sometimes do not, I shall argue that the application of theories of markedness to morphosyntactic change is often Procrustean. For instance, Timberlake’s well-known claim (Timberlake 1977) that “the concept of markedness, or naturalness, must be understood with reference to the particular change involved” (a claim which finds an echo in the ‘system-dependent naturalness’ of natural morphology) suggests that explanations in terms of markedness may represent Pyrrhic victories: they salvage an account of the data, but only at the cost of demonstrating the Protean nature of markedness. The view that the actualization of morphosyntactic change is gradual, and proceeds according to a set of hierarchies, and that these hierarchies are determined by factors relevant to the change in question, is undoubtedly well founded; but the claim that they are markedness hierarchies is either uninteresting (if the markedness is defined as purely local) or unsubstantiated (if the markedness is to be defined in terms of universal or quasi-universal principles). An examination of the data leads me to conclude that the relation of hierarchies of actualization or extension to a general theory of markedness is perhaps best regarded as epiphenomenal rather than phenomenal.
This paper presents an analysis of the discourse strategies used by the members of a factory production team to communicate information or persuade others to do things as they talk with one another during the course of a shift. The strategies used range from very explicit ‘on-record’ utterances like the emphatic imperative Just do it…! in the title, to a variety of less direct, more consensual devices. The database consists of audio recordings and ethnographic information collected in a Wellington factory during the pilot phase of a collaborative action research project, as an extension of Victoria University’s Language in the Workplace Project (see Stubbe 1998).
The primary language of communication in this factory is English, but the workforce is multicultural and includes many people for whom English is a second language. Related literacy issues mean that spoken interaction is the primary channel of communication for most workers on the factory floor. Communication is often sporadic and predominantly involves the routine imparting of specific information or instructions. In this context, ‘getting the message across’ (i.e. making sure the referential meaning of a message is communicated fully and accurately), is perceived by factory personnel as being of prime importance, as any miscommunication at this level will often have negative (and highly visible) practical consequences (cf Coupland et al 1991).
However, analysis of the data shows that successful communication consists of much more than the simple transfer of information from one individual or group to another: rather, it involves a joint negotiation of meaning which may extend well beyond the boundaries of a single interaction. Both the sequential structures in an interaction and participants’ background knowledge (eg. about what has gone before, the role relationships involved, the kind of talk appropriate in each setting) provide the means by which participants jointly construct a particular social order and come to a shared interpretation of what is going on (Drew and Heritage 1992, Gumperz 1992). Moreover, because communication always involves both affective and referential meaning, the most direct and explicit strategies are not necessarily the most effective (Holmes et al in press, Stubbe and Vine in press). The strategies and linguistic forms chosen to express any given speech function play a vital part in building and maintaining good relationships, which in turn contribute to effective communication within the team.
The motivation of the paper lies in the observation that there is little interaction among four major areas of comparative linguistic research --- language typology, contrastive linguistics, translation studies and the computational modeling of multilingual processes as implemented in machine translation or multilingual generation. While language typology typically compares many languages and looks at only a few parameters of cross-linguistic variation, contrastive linguistics usually looks at only a couple of languages and covers many features of the languages under investigation. Where language typology and contrastive linguistics converge are in being system-oriented: the focus of study are differences and commonalities of language systems. Translation studies is at the other extreme: it is text-oriented, focusing on the relation between texts --- a source language text and its rendering in a target language --- rather than language systems. Machine translation and multilingual generation are in between the two: in both of them actual text has to be produced and thus they must have a model of both system differences between languages and their reflexes in texts. There is however one common concern in all four areas and this is to account for the differences and commonalities between languages and to make available a suitable model to do so.
In the paper I suggest that the common core of language typology, contrastive linguistics, translation studies and multilingual computational linguistics should be a general model of cross-linguistic comparison that accommodates both the system view and the text view on cross-linguistic variation. The model I present is based on Systemic Functional Linguistics (SFL; Halliday 1978, Halliday 1985). I show that the general representational categories of SFL --- metafunction, stratification, instantiation, rank, axis, delicacy --- can act as parameters along which cross-linguistic variation can be described. The fundamental assumption brought forward by SFL that acts as the unifier of concerns of the different areas involved in language comparison is that texts are ultimately instantiations of the language system under certain specifiable contexts of use. A model of cross-linguistic variation based on SFL thus bears the promise of opening up the text view for the system-oriented branch of cross-linguistic study, and the system view for the text-oriented branch.
I illustrate the model with data from several European languages (German, English, French, and some Slavic languages), concentrating on cross-linguistic variation between English and German as elaborated in Hawkins 1986.
The paper describes the main results of a comparison of the ToBI (Tones-and-Breaks-Indices) (Pierrehumbert 1980, Silverman et al. 1992} and the Hallidayan systems (Halliday 1967, Halliday 1970) for annotating speech data with information about intonation. The goal of this comparison is to define a mapping between the two systems, so that one can be translated into the other automatically.
This attempt has a two-fold motivation. First, it is motivated by computational application in concept-to-speech systems, in which text in spoken mode is automatically generated from an underlying abstract meaning representation. The concrete goal we are pursuing is to connect an off-the-shelf speech synthesizer for English (Festival; Black & Hunt 1996) with an automatic text generation system based on Systemic Functional Grammar (SFG; Halliday 1985) that generates English text. Since in the systemic approach, intonation is accounted for as part of grammar rather than as an independent component, it is straightforward to extend the grammar of a systemically based text generation system with an account of intonation. Connecting such a system to a speech synthesizer requires mapping the output of the generator to the input requirements of the speech synthesizer. In the Festival system, the intonation of the text to be synthesized can be manipulated by annotating it with ToBI labels. Therefore, a mapping between a systemic intonation annotation and ToBI is required. Second, there is a theoretical motivation. With a mapping between the ToBI and the SFG systems for intonation annotation, it will be possible to link the acoustic analysis of speech data to an interpretation of intonational meaning as it is offered by SFG. Existing speech corpora that are acoustically analyzed and annotated with ToBI can be used to test some of the assumptions brought forward by SFG about the nature of intonation.
In the paper we report on the analysis of a small speech corpus compiled from Halliday 1970 with ToBI and SFG. The intonation analysis is based on an acoustic analysis of the speech data in terms of fundamental frequency (F0). The data are represented in EMU (Cassidy & Harrington 1996), a data base system for storing speech data that provides for a multiple-tier analysis of acoustic (e.g., F0 contour and speech waveform) and phonological (segmental and suprasegmental) features. We present the major differences and commonalities between ToBI and SFG and show how the five tones assumed by Halliday can be described by unique sequences of ToBI labels, which thus makes a mapping between the two feasible in principle. We conclude with some remaining problems, mainly to do with differences in the kinds of units that are recognized in the two system, on the one hand, and the notion of rhythm, on the other hand.
This paper examines 'na' in South Efate, Vanuatu. 'na' is the verb 'to say', but has other meanings ('want', 'in order to') that appear to be an innovation, given the lack of mention of these meanings in detailed historical texts from late last century (e.g. Bible 1874). An example of its use as 'to say' follows:
| I | wel | tem | i | na | "Ore | i | wi" |
| 3sg | thus | father | 3sg | say | yes/right | 3sg | good |
| So the father says, "Yes, it is good." | |||||||
The 'want' meaning appears below:
| Nat | i | skei | i | na | ke | wat | kineu | napu |
| man | 3sg | one | 3sg | want | 3sgIRR | hit | 1sg | road |
| This man wants to hit me on the road | ||||||||
However, other forms with a similar set of meanings are found in languages to the south and north of Efate (see Clark (1985) Crowley (1992); Crowley (1998); Schütz (1969)) The paper points out the fallibility of historical texts in this instance in not representing a feature of the language that must have been present given its existence in other languages of the region.
The alternative two possibilities are that (1) the innovation of the meaning 'want' from 'say' has occurred in the last 100 years in Efate, independently of an identical innovation in other languages of the region, which is unlikely; and (2) the meaning diffused into Efate from these other languages over the same time period, perhaps more likely.
This paper presents an analysis of the Spanish Discourse Marker pues (generally translated as ‘well’ or ‘then’) as used in conversational Colombian Spanish. Although pues has received a great deal of attention in the literature on Spanish Discourse Markers, most studies are based on constructed and written examples (Beinhauer 1968, Fuentes 1985, Portolés 1989, among others). The use of pues in conversation differs markedly from its use in writing, and analyses of non-conversational uses of this marker therefore fail to account for many of the functions found in the database used for this study. The database comprises two hours of conversation, and presents a total of 114 tokens of pues. In this database, what is commonly identified as the most basic function of pues in the literature, that of marking causality, does not occur at all. There are also no clear-cut examples of another commonly identified ‘basic’ function, as a consecutive marker. In this paper, it will be argued that the most basic function of pues is to overtly mark a link between some aspect of the prior discourse and upcoming information, and in so doing, to highlight this information. This can be seen in the following example, taken from the data base, where the speaker is talking about playing the game ‘Who am I?’
‘Yo que dizque, Amparo Grisales. Pues, tenían que adivinar quién era yo.’ ‘I’m supposedly Amparo Grisales (Colombian actress and sex symbol). pues, they had to guess who I was.’
A definition will be proposed to account for this use, in accordance with the Natural Semantic Metalanguage approach (Wierzbicka 1996, and references therein), and it will be shown that this meaning of pues is semantically related to the functions of pues as a causal and consecutive marker discussed in the literature.
Japanese is a mora-based language, and as such presents English learners with timing problems in production and perception with regard to long vowels and geminate consonants. Native listeners perceive the distinction between short and long vowels and single and geminate consonants, by the relative duration of the target vowel or consonant. In a pilot study, we found that beginner level English learners of Japanese were able to perceive differences between long and short vowel and geminate and non-geminate consonants (though their production of these durational contrasts was poor) provided that there was no need for temporal (speech rate) normalization of the stimuli. In the present study, we report on the perception of consonant and vowel mora contrasts by learners of Japanese at two levels of L2 proficiency, under listening conditions that required speech rate normalization.
We hypothesized that native listeners would experience no difficulty in accommodating to rate variation in the stimuli, that beginners would experience great difficulty, and that advanced learners would show some ability to adapt, particularly where the L2 contrast was similar to one in L1 (e.g.: the Japanese moraic vowel contrast and the tense/lax contrast in English vowels). Ingram and Park (1997) found that normalization strategies developed in L1 acquisition may transfer to perception of a similar phonological contrast in L2. Results support this prediction. Learners were more successful at adapting to speech rate variation for the perception of moraic vowel than consonant contrasts. However, there was a complex interaction between speech rate and proficiency level with respect to the perception of the consonant length contrast, which indicated that even experienced non-native listeners respond in a qualitatively different manner from native listeners, possibly employing different cues or perceptual strategies.
In the paper we explore the basis for these differences in the perception of moraic timing between beginning and experienced L2 learners, and between learners and native listeners.
One feature of song texts in many Central Australian ceremonial genres noted by Strehlow for Arrernte (1971), Hale for Warlpiri (1984), Tunstill for Pitjantjatjara (1995), and Hercus and Koch for Wangkamadla (1995), is the phonetic variation of words at the start of text lines. Hale has identified a process of 'consonantal transfer' whereby the consonant of a line final suffix is transferred to the beginning of the next line.
This paper explores the consonantal and vowel variation in the text lines of one Kaytetye awelye series called Akwelye, and the extent to which consonantal transfer occurs. Other phonetic features of akwelye text are identified and compared with everyday speech. The variation is examined in relation to Kaytetye syllabic structure and the rhythmic requirements of the specific song items.
Awelye is a type of women's ceremony belonging to the Arandic speaking languages of Central Australia (known as yawulyu in Warlpiri). Kaytetye is an Arandic language of Central Australia and the Akwelye series discussed here resembles other Central Australian song styles as identified by Moyle (1986), Ellis(1964), Keogh (1990) and Strehlow (1971).
The prevailing view of the status of the French future tense in Romance Linguistic literature is that it has been all but eclipsed in the spoken language by the composite future (aller + infinitive ) for the function of temporal reference (eg. Posner 1996). An influential hypothesis (Fleischman 1982 ) is that future tenses begin as modal exponents, develop temporal meanings and then lose their temporal force to become modal exponents again. The French future tense is described by Fleischman as being in the last part of this putative cycle, having lost its temporal meanings to aller + infinitive, except in formal registers, and become an exponent of modal meaning in the spoken language. Research in France, on the other hand, indicates that the future tense is still being used for particular types of future temporal reference in spoken French (Jeanjean 1988).
In this present study, evidence from a popular daily newspaper, and from native-speaker informants supports Jeanjean's claim over the view held in Romance Linguistics circles. Data will be presented to support the view that the French future tense is a temporal rather than a modal exponent, and an alternative theory, that future tenses tend to loes their stability because of loss of modal rather than temporal force will be put forward.