Ordered alphabetically according to first author's surname.
For workshop and plenary paper abstracts, see separate pages.
Many Australian languages often dispose of overt framers of quoted speech like '(s)he said' and rely purely on prosodic cues, particularly intonation, to signal the transition from narrated action to reported speech or thought. Although this pattern is widespread, to judge by the frequency with which unframed material in quotation marks appears in published texts, there has as yet been no instrumental phonetic study of the intonational cues employed. Yet the phenomenon is central to making sense of narrative texts, and, at a more specific level, can effect the interpretation of semantically ambiguous argument prefixes on the verb, counterfactual particles, and ignorative lexemes, among others.
The present paper reports on preliminary analyses of intonational framing of quoted speech in five non-Pama-Nyungan languages: Mayali and Dalabon (both Gunwinyguan), Iwaidja and Ilgar (both Iwaidjan) and Kayardild (Tangkic). In the main body of the paper we survey the main intonational cues associated with quotative framing, most notably, pitch reset and top line pitch movements. These cues are sufficiently distinctive that quotation can be clearly identified in the majority of cases in our initial corpus. We conclude by looking more briefly at four related questions: (a) does the intonational framing differ when an overt lexical marker of quotation is present? (b) in the case of represented dialogue, are polyphonic base lines employed systematically to represent different speakers? (c) does the intonational treatment of single quoted words differ from that of longer passages (d) are different intonational cues used in the framing of speech and of thought?
In languages like English, some words are made more prosodically prominent than others by carrying an intonational target or pitch accent. Furthermore, pitch accents are phonologically associated with the rhythmically most prominent syllable in the word. However, accentual prominence in English is not just cued by intonational targets or movements, rather, it can also be cued by greater loudness and length. In addition, accented vowels are often "hyperarticulated" relative to unaccented vowels. In other words, the paradigmatic contrast between vowels is enhanced. In recent studies (e.g. Harrington, Beckman, Fletcher, Palethorpe, JASA, 1997, 102, 3205A), it was shown that talkers sharpen the syntagmatic contrasts between a rhythmically strong vowel and both an initial consonant and a following rhythmically weak vowel in accented words relative to their deaccented equivalents. Thus accented vowels can be contrasted with unaccented vowels along a syntagmatic as well as paradigmatic dimension.
The focus of this study is to see whether similar kinds of paradigmatic and syntagmatic differences among accented and unaccented vowels is apparent in a typologically different language from English. We set out to investigate the tonal and non-tonal correlates of different levels of prosodic prominence in Dalabon, spoken in Western Arnhem Land. This study forms part of a much larger project that examines intonation and prosody in some Northern Australian languages. Dalabon is a non-tonal language which has lexical stress.
Two lengthy Dalabon narratives formed the basic data set in this study (roughly 8 minutes of connected speech). The original field recordings were digitised and acoustically labelled at the segment and syllable level. An acoustic intonational analysis was also performed to locate accented syllables and these were labelled using a version of the ToBI (Tones and Break Indices) intonational transcription system that has been developed in our previous work on Mayali which is a closely related language to Dalabon. F1 and F2 formant values, RMS amplitude peaks, as well as F0 targets were extracted for accented and accented vowels. Acoustic durations were also measured.
Our preliminary results suggest, that like English, accentual prominence is cued in these Dalabon data by, on average, higher F0 targets in accented vs unaccented syllables and greater mean RMS amplitude, duration, and marginally more dispersed formant frequencies in accented vs unaccented vowels. However, further analysis will reveal whether our Dalabon data show the same extent of syntagmatic differences between accented vowels and neighbouring unaccented or unstressed vowels.
Representation is a central concept in phonology, yet generally receives much less attention than its key role warrants. This lack of attention has been singled out from time to time as a contribution to various problems in theory and practice.
Thus, there have been moves to constrain the phonetic representation used in phonological theory - for example, by distinguishing more clearly between acoustic and articulatory features, by invoking a notion of 'controllability' (by speakers) of phonetic differences, or simply by demanding more detail and less presupposition of phonological analysis (consider the work of Browman and Goldstein, Pierrehumbert, Blumstein, Nolan, Diehl, Anderson, Keating, Fowler and others).
There has also, more famously, been much discussion of the need to constrain phonological representation, in terms of its abstractness, the units in which it is couched, its implications for the kinds of processes that must be invoked to relate it to the phonetic represenation, its 'psychological reality', and so on (consider the key role of representation in the development of models such as natural phonology, non-linear phonology, optimality theory; as well as theoretical discussion by Linell, Ohala, Goldsmith, Anderson, Local, Dressler, and many others).
In this paper I consider, rather metatheoretically, the nature of the constraints that have been brought to bear in these analyses of the problems of representation, and suggest that these constraints themselves might fruitfully be modified.
In doing this, I consider the general nature of representation (of anything) as involving 'something represented', 'someone representing' and a 'context'. On this basis, I recommend analysis of representations in terms of a series of simple questions: 'of what?', 'by whom?', 'in what context?'.
This type of analysis is proving helpful in some practical applications: disentangling representations like those of linguists and non-linguists (in dictionary pronunciation guides), those of native speakers and learners (in ESL pronunciation), those of writing and transcription (in orthography development).
In this paper I argue that some of the theoretical problems of phonology are also usefully seen as being caused by the lumping together in one representation of aspects of speech which require different answers to the questions above. I will have time for only one example of how this style of analysis can illuminate theoretical issues - the definition of 'lenition'.
Despite a large literature in linguistics, philosophy, psychology and literary studies (cf. Ortony 1983, Gibbs 1994, Goatly 1997), certain basic questions about the identification and interpretation of metaphors remain unresolved. This paper tackles only what are variously called active, live or fresh metaphors, e.g. 'The past is a foreign country', 'Unemployment is a contagious disease', 'Language is a mirror of the mind'. Needless to say, in English the use of active metaphors is an important feature of journalism, religious and political discourse, popular science writing, and many other contexts. My semantic framework is the natural semantic metalanguage (NSM) appproach originated by Anna Wierzbicka (1996). The NSM inventory of semantic primes provides a core of non-metaphorical meanings which can used to analyse active metaphors in a clear and non-circular fashion and in terms which can be readily transposed between languages.
The first part of the paper concerns identification. I argue that "metalexical awareness", i.e. a metalinguistic awareness of a potential contrast or dissonance between the speaker's intended meaning and "what the words say", is a distinguishing feature of active metaphor, more important than the parameter of originality vs. conventionality. I propose a metapragmatic "script" for active metaphorising as a valued speech practice of English, and Western cultures generally. Following Boguslawski (1997), I also suggest that linguistic tests, such as compatibility with "metalexical tags" such as 'so to speak', 'as it were', and 'if you like', can furnish a valuable tool for identifying active metaphors. A widely held belief that there is a continuous cline between fixed, conventional and active metaphors is therefore invalid.
The second part of the paper concerns interpretation. I show that (except for some metaphors in poetry, which demand separate treatment) active metaphors generally have specific and determinable meanings, which can be stated in explications framed in the NSM metalanguage. For the simple predicative metaphors dealt with in this paper, it turns out that the necessary explications have a three-part structure: they begin with a piece of background knowledge related to the topic of the metaphor, then follows what might be called the metaphorical dictum or assertion (the "point" of the metaphor), then finally comes the comparison or analogy between the metaphorical topic and vehicle. My argument opposes a commonly held view that metaphors have an open-ended set of interpretations which defy paraphrase.
This paper examines the nature of negation in Xhosa, a Southern Bantu language. The focus is specifically on the distinction between negation in verbal and copular predicates (particularly those based on adjectives). In the latter (but not the former) case there is significant variation in the way negatives are formed. In particular one finds two emergent strategies of negation, the one revealing the employment of auxiliary structures much as in English (but unusual in Bantu), the other showing the use of negative morphology similar to verbs, e.g.:
| Positive | ||
| Le | nto | intle |
| This | thing | it-beautiful |
| Negative: showing 'verbal' morphology | ||
| Le | nto | ayintlanga |
| This | thing | not-it-beautiful-not |
| Negative showing 'auxiliary' -kho | |||
| Le | nto | ayikho | intle |
| This | thing | not-present | it-beautiful. |
Reasons for this are offered. These include the possibility of English influence as well as the 'mid-way' properties of adjectives in terms of verbal and nominal morphology. Additional issues explored include:
The general conclusions of the paper are contextualised within the broader framework of negation in language, as well as the particular types of language change that languages in contact demonstrate.
Blake (1988: 7) reconstructs the following free pronoun paradigm for a proto-language ancestral to the Non-Pama-Nyungan (NPN) languages.
| (1) | 1 | 1+2 | 2 | 3masc | 3fem | |
| Minimal | *ngay | *nya | *nginy | *nu | *ngaya | |
| Augmented | *nyi-rrV | *nyi-rrV | *nu-rrV | *ku-rrV | *pu-rrV |
However, I argue that it is not a free pronoun paradigm which can be reconstructed for the NPN languages, but rather a prefix paradigm. Consider the following free pronoun and prefix forms from Kija, Ndjebbana and Ngalakgan: three widely separated languages, which do not otherwise appear to be related except as descendents of Proto-Australian.
| (2) | Kija | Ndjebbana | Ngalakgan | Reconstruction | |
| 2Aug Pronoun | nengkerre-pen | nu-rra-ppa | nu-rr-kkaq | ||
| 2Aug Prefix | na-rrV- | na-rru- | nu-rru- | *nV-rrV- | |
| 3Aug Pronoun | purru | pa-rra-ya-ppa | pu-rr-kkaq | ||
| 3Aug Prefix | pu-rrV- | pa-rru- | pu-rru- | *pV-rrV- |
There is no consistent relationship between the free pronouns as wholes. There are some consistent general relationships between portions of the free pronoun forms, but these portions are also the portions of those free pronoun forms which are either identical or similar to the corresponding verbal prefix. This follows from the fact, illustrated by the Ndjebbana and Ngalakgan forms, that free pronouns in the NPN languages often consist of the corresponding prefix + a base morpheme. The portions of pronominal paradigms which are cognate across the NPN languages, are the forms prefixed to verbs and pronominal bases. A comparison of prefix paradigms across the NPN languages does show the kinds of consistent general relationships which support the reconstruction of complete protoforms.
The pronominal base portions of the free pronouns do not correspond to one another. As such, it appears that paradigms of prefix+base free pronouns are commonly replaced by new prefix+base paradigms diachronically. This is a natural pattern of development in languages with obligatory cross-reference, such as the NPN languages, where free pronouns appear chiefly in topic functions. There are a number of different foci in the continuum of topicality (e.g. contrastive vs non-contrastive topicality). Languages commonly have more than one paradigm of free prefix+base pronouns, marking different domains of topicality. Paradigms marking one focus of topicality are likely to extend their range to displace paradigms marking other foci of topicality. The greater the time depth of prefixing, the more often prefix+base paradigms will be replaced.
This pattern of historical influence from prefixes to free pronouns has not otherwise been described, and is predicted to be rare (Heath 1978: 108, Nichols 1986: 87-88). As such, this reconstruction is of interest for historical linguistics generally. It is also of interest for Australian historical linguistics more specifically. It shows that head marking (Nichols 1986) is an old pattern within Australia.
Butcher, in unpublished research, has characterised Australian sound systems as "place-rich, manner-restricted". In the languages of Central Australia (Arandic, Western Desert and Warlpiri) the coronal consonants have up to five contrasts:
- apical: retroflex (sublaminal postalveolar) and alveolar;
- "prepalatal" (prepalatalised apical);
- laminal: "palatal" (lamino-postalveolar) and "interdental" (lamino-dental).
I present data that show that diachronically and synchronically, across and within languages and dialects of languages, the five coronal contrasts are reduced and confounded in a variety of ways. Synchronic facts include:
There is also a range of synchronic comparative data that points to systematic but complex diachronic relationships between sets of coronal contrasts:
These facts, though diverse in kind, show a common pattern of confounding or reduction of the coronal contrasts:
To understand these it is necessary to consider the phonetic facts of the dynamics of articulatory gestures (Butcher).
Although the phonetics of allophonic prepalatalisation has been investigated for Eastern Arrernte (Henderson 1998), there is a need to investigate it in the northern Arandic languages/dialects (Kaytetye, Eastern Anmatyerr, Alyawarr), where prepalatalisation leads to phonemic contrasts.
A better understanding of the phonetics and phonology of coronal contrasts in (Central) Australian languages is needed to aid the search for putative cognate sets, and for the formulation of historical sound changes (see for instance the questions raised by Koch 1997 p280, in the section on the putative sound change rule "(SC5) Prepalatalisation of apicals before i", and its ordering with respect to "(SC6) Vowel centralisation").
There is also a need to better understand the foundations of coronal phonemic contrasts and allophonic variation in the Arandic languages, particularly prepalatalisation, in order to be able to formulate a pan-Arandic practical orthography, perhaps by adhering less strictly to phonemic analyses, and taking allophonic and phonetic facts into account.
This paper focuses on how dictionary making in developing societies has become a most important ideological tool in the process of democratisation of one's language/s. While descriptive linguistic claims to have liberated languages from the fetters of prescription, in the present day, dictionaries entail hidden prescriptions, and even hidden agendas.
Today dictionary-makers should address the problems of :
The paper demonstrates these features by looking at two Tamil dictionaries brought out by Cre-A publishers for the language institute MOZHI, Cre-A's contemporary Tamil Dictionary and the Dictionary of Idioms. Both have aimed at unravelling the layers of the contemporary Tamil - how even the unlettered man is contributing to the written sphere and how there is an inexhorable process of democratizations leading to empowerment of the people.
An unusual feature of the Western dialect of Warlpiri which distinguishes speakers from those of the Eastern dialect is the presence of a stop - flap contrast in word initial position (tarruku 'sacred thing', rdarri 'raw') and its productive use intervocalically in post-tonic (post-stress) position (wapirti 'yam', wapirdi 'on arrival'). Initial position is highly disfavored for flaps in the languages of the world. Apical stops frequently weaken to flaps or taps intervocalically, but the maintenance of a phonemic contrast between apical stops and flaps in this position is unusual. Eastern Warlpiri has scattered examples of an apical stop-flap contrast, but the distinction appears not to be phonologized in this dialect (Laughren, 1999; i.e.: a few minimal pairs may be found, but the contrast appears not to be productive). Warlpiri word stress falls on the initial syllable. Post-stress (post-tonic) intervocalic position is the expected environment for lenition of stops to flaps in many languages, most notably, English. We have noted that Warlpiri stops in post-tonic position have unusually long closure periods, and abrupt release gestures, akin to geminate stops. This raises the possibility that Warlpiri stops in post-tonic position are `protected' from lenition as are geminates or fortis stops in other languages.
The main goal of our research is to understand the phonetic mechanism for maintaining the intervocalic stop-flap contrast in Western Warlpiri and how it subsequently spread to initial position. As an initial step towards this goal, we report an acoustic phonetic analysis of the Warlpiri stop - flap contrast in initial and post-tonic intervocalic position from extensive studio recordings of word forms and phrases elicited from two female informants of Western Warlpiri.
This is a report of work in progress. At the time of writing, some 60 tokens of the stop flap contrast in word initial position from two Western Warlpiri speakers have been analyzed. Individual differences are prominent in the phonetic realization of flaps in this position. In the paper we will report findings from post-tonic intervocalic and non-post-tonic intervocalic position and draw comparisons with phonetically related sounds in these environments.
This paper has two aims. One is to acoustically-phonetically describe the word-level pitch realisations of Kagoshima Japanese (KJ) by means of z-score normalisation, the other is, on the basis of this acoustic-phonetic description, to propose a tonal representation of the KJ words, using the target-tone model proposed by Pierrehumber and Beckman (1988). All raw data was collected from two native speakers of KJ (one male and one female), and analysed with the Computerised Speech Laboratory.
KJ is different from Standard Japanese (SJ), in that all words exhibit only two way accentual oppositions; (L)0HL and (L)0H. In the former pattern, only the penultimate syllable of a word has a high pitch, and every other syllable has a low pitch, for example [machiyakuba (LLLHL)] "town hall". In the latter pattern, only the last syllable of a word has a high pitch, and every other syllable has a low pitch, for example [miyagemono (LLLLH)] "present". Another significant difference of KJ from SJ is that, for KJ, the syllable is the tone bearing unit (or the unit of pitch contrast). That is to say, that whilst SJ pitch changes occur at mora boundaries, as you can see in four mora word [kantan (LHHH)] "simple", on the other hand, pitch changes occur at syllable boundaries in KJ, for example the three syllable word [bookuugoo (LLHHLL)] "air-raid shelter".
This study firstly presents how syllable structure influences pitch realisations of (L)0HL and (L)0H, by comparing the z-score normalised descriptions of words having different syllable structures. Possible syllable structures of the KJ are (C)V, (C1)VC2 (C2 is the initial part of a geminate), (C)VV, and (C)VN (N is a syllable-final nasal). Disyllable words having these syllable structures (e.g. [mune (HL)] "chest", [konban (LLHH)] "tonight", [taijyuu (HHLL)] "weight") were used for comparison. The results of the comparisons clearly show that the pitch realisations of (L)0HL and (L)0H are significantly influenced by some syllable structures, especially in terms of the pitch target value of the initial syllable.
Secondly, the pitch realisations of (L)0HL and (L)0H are examined by comparing the z-score normalised descriptions of poly-syllable words consisting of 2 to 5 CV syllables. The results of the comparisons clearly show that the pitch realisations of (L)0HL or (L)0H are significantly different from each other at almost all pitch target points.
Pierrehumbert and Beckman (1988) demonstrated that fully-specified tonal representation is not the best kind of analysis for surface tone pattern in SJ, and argued that accentual phrases need to be underspecified positing target tones. Their target model is adopted in this study, and a tonal representation of the KJ words is discussed on the basis of the above mentioned descriptive data and comparisons.
Australian Aboriginal Pidgin English was formed from the interaction between English and Aboriginal languages, beginning in Sydney in 1788. It subsequently spread with the moving frontier, being used as a lingua franca by Aboriginal people in the inland pastoral industries. From the 1860s a Melanesian offshoot of Australian Pidgin was developed in the Queensland sugar plantations; this variety was taken home by the workers to Vanuatu, the Solomon Islands and eventually to Papua New Guinea. Many of the features now found in Melanesian pidgin and creole languages are attested first in the Aboriginal Pidgin of New South Wales and Queensland. The question thus naturally arises: to what extent were features of this pidgin influenced by the Aboriginal languages of southeast Australia? (cf. Baker, Philip. 1993. Australian influence on Melanesian Pidgin English. Te Reo 36:3-67.)
I examine closely two of the most characteristic grammatical features of all the English-based pidgin/creole languages of the southwest Pacific:
| VERB-im NOUN | e.g. kill-im bullock |
| ADJ-fela NOUN | e.g. big-fela kangaroo |
proposing an account of their origin that is consistent with: the early documented facts of the pidgin, what is known about the grammar of the relevant Aboriginal languages, principles of linguistic change, and theories of pidgin/creole development. I argue that both constructions result naturally from a mechanism whereby Aboriginal people reanalysed English sentences according to the grammatical patterns of their indigenous languages. A key role is played by their interpretation of English forms in terms of constructions involving "nothing", i.e. syntactic zeroes - as object of the verb in the first case and as head of the noun phrase in the second.
A number of Australian languages have noun classes or gender agreement. Most include a masculine feminine distinction. Few make a distinction between singular and non-singular number apart from prefixes referring to groups of humans. Noun class or gender agreement may be by prefix eg. Wardaman (Merlan 1994), Yanyuwa (Bradley 1992) and others described in Harvey, M and Reid N. eds. 1997 Ngan’gityemerri (Reid), Marrithiyel (Green) and Murrinhpatha (Walsh), by noun markers, Dyirbal (Dixon 1972:44) or by pronominal agreement, Ungarinjin (Rumsey 1982: 30-41). In none of these is it shown by suffix.
The Jarragan language Gija has overt nominal gender shown by suffix to the noun and by agreement with adjectives (suffix agreement), demonstratives (form and suffix agreement), nominal interrogatives (form and suffix agreement) and all forms of third person pronouns including prefixes and suffixes to verbs and enclitic pronouns. The other Jarragan languages, Miriwoong and Gajirrabeng, have overt gender suffix marking only on some nouns that refer to human females. All other nouns have covert gender which can only be determined by observing concord with adjectives (shown by suffix), demonstratives (form and suffix), and all forms of third person pronouns. All three languages have masculine and feminine genders but Gija has an additional gender which has been glossed as ‘non-singular’. On preliminary analysis this appears to correspond closely with the neuter collective and human plural found in Ungarinjin its western neighbour (Rumsey 1982: 30-41).
It occurs with all plural animates e.g. jiyiliny ‘man’, jiyilem ‘men’, ngalil ‘woman’ ngalim ‘women’. (-ny - masculine, -l feminine and -m - non-singular). It occurs with mass nouns such as goorrngam ‘water’ and marnem ‘fire’, with all citation forms of body parts and with most generics miyalem ‘meat’, mayim ‘non-meat food’ and ngarem ‘bush honey’.
When referring to specific species of edible plants, animals or native bee hives, the singular forms of the generics which must agree in gender with the specific plant, animal or honey species are used.
Further study of the occurrence of the non-singular versus the singular shows that the choice of number/gender marker is more complex than a simple comparison of number. This can be seen in the two sentences quoted below in which plural ‘lice’ are contrasted with the idea of ‘louse’.
| "Yiwi-m-a | bamberr-any-nha=birri=yoo." |
| louse-n-EMPH | 2nsS.3nsO.IMP-CARRY.IMP-2S=3nsIO=d |
| Bring some lice for these two. | |
| Boorral=gala | woomberrayi=yoo | yiwi-ny." |
| feed=IMP | 3nsS.M13b.GIVE.IMP=d | louse-m |
| Let them have a meal of louse. (Gija TC HJ J tu J 88) | ||
The nature of the singular versus non-singular in Gija will be explored. Its occurrence in this language will be compared with that in the other Jarragan languages Miriwoong and Gajirrabeng.
The merging of EAR/AIR diphthongs in New Zealand English (NZE) is well-documented (research by Holmes & Bell 1992; Maclagan & Gordon 1996; Watson et al.1998), as is the chain shift raising of the short front vowels HEAD and HAD, and the centralizing of HID. With regard to the merging diphthongs, it has been hypothesized that NZE AIR has become involved in the short front vowel raising, and in particular with the raising of the HEAD monophthong. Evidence in support of this supposition is the observation that the target of NZE AIR is much more raised than in Australian English. The implied relationship between the raising of HEAD, and the merging of EAR and AIR, merits further investigation. A study was set up with the aim of finding such a correlation.
Speech data from NZE speakers of various ages were analyzed for their acoustic properties. It was hypothesized that the results would yield degrees of merging of the diphthongs, which is characteristic of near mergers, and, in addition, degrees of raising of the HEAD token. The interest of this research lies in the documentation of the relationship between these two factors. First and second formant readings were taken from the EAR/AIR, and HID/ HAD/ HEAD tokens obtained. These were analyzed statistically to ascertain whether there was any evidence of the hypothesized correlation. This involved normalizing procedures, comparison of HEAD formant readings with averaged formant readings, and calculation of the distance between the onsets of the diphthongs using a Euclidean distance metric.
Preliminary data analysis by auditory means revealed a preference for merging EAR/AIR towards the closer EAR diphthong. The loss of contrast is evident in the difficulty Australian English judges had in categorizing the tokens on a continuum. Acoustically, F1/F2-F1 graphs reveal that the onset targets for both diphthongs tend to congregate close to the HEAD target, an observation which supports the raising of the AIR onset in conjunction with the raising of the HEAD vowel. Statistical results from the scaled figures (in progress) are expected to support these observations and shed further light on any connection between the NZE falling diphthong merger and the chain shift raising of the NZE HEAD and HAD.
In Warlpiri finite clauses an auxiliary (AUX) complex of functional categories {COMP/NEG - (evidential) - TENSE/ASPECT- PERSON, NUMBER [of Subject and non-Subject]} occurs either in initial position or in Wackernagel's position, following the first phrasal constituent of the clause. The pre-AUX constituent is not contrained by its grammatical function (or deep structure syntactic position) (Hale 1982, Nash 1986, Simpson 1993), but this position is not totally free of syntactic constraints.
Austin and Bresnan (1996) propose a constituent structure where AUX is equated with I(nflection) and where a finite main clause is an IP whose complement is S. The pre-AUX constituent is thus the SPEC of IP. Where AUX is clause initial, IP is argued to lack a SPEC. To account for the fact that the verb, but not the VP, may occupy the pre-AUX position, despite the fact that SPEC of IP can only be occupied by a phrasal constituent (XP), not a head (X°), Austin and Bresnan invoke a prosodic rule of verb-AUX inversion to avert a phonological infelicity if AUX fails to meet the phonological requirements for occupation of the clause-initial position. This rule also applies where a preverb is separated from its verb by AUX.
In my paper I will examine these claims by taking a closer look at what can and cannot appear in pre-AUX position. In particular I will focus on the constraints imposed on what may occur in pre-AUX position if the AUX includes the negative element 'kula'. I will argue that no V-AUX inversion rule is required, since the pre-AUX position is, as Austin and Bresnan note, a focus position which I interpret as part of the projection of a higher functional category than IP. I will also present more data about what elements can constitute the left-most position in AUX in order to arrive at a better understanding of the syntactic structure underlying the surface string.
Gumperz and Levinson’s 1996 Rethinking Linguistic Relativity is a valuable collection of contemporary research on the linguistic relativity issue. Not one of the contributors, however, uses Whorf’s reasoning about the ‘linguistic relativity principle’ as a basis for argumentation. In particular, Whorf’s constructs, ‘isolate of experience’ and ‘isolate of meaning’, are not mentioned, although a) they are central to the logic of his claims and b) assessment of their viability is essential to any evaluation of the contemporary relevance of his definitions.
Whorf drew on gestalt psychology discoveries of figure/ground universals in human perception to argue that languages, by operationalizing specific configurations of experiential isolates as isolates of meaning, encourage language users to systematically attend to some ‘bits of experience’ in preference to others not so highlighted. In providing alternative conceptual frames for dealing with experiential data, different languages are thus compared with the different coordinate systems of Einstein’s observers in space, a linguistic relativity principle analogous to Einstein’s relativity principle being taken to operate at the core of that domain of human understanding which is linguistically mediated. (Lee 1996, 1999, In press).
Can the isolate notion be usefully applied to the diverse range of linguistic data presented in support or contestation of the notion of linguistic relativity? In this presentation, examples from the G & L collection will be used to demonstrate that the isolate construct has heuristic value at least. For instance, in mapping crosslinguistic variation in spatial categorization, Bowerman shows that linguistic resources provide conventionalized ways of conceptualizing aspects of scenes. From a Whorfian perspective, notions of containment and support associated with English in and on, respectively, may be regarded as abstractions from universally observable contiguous relations of objects. As such they constitute one category of experiential isolates amenable to logical extension from apparently canonical to less canonical scenes e.g. an apple in a cup to a video cassette in its case. Korean kkita, however, can be shown to cut across the in/on dichotomy by drawing obligatorily on an alternative relational isolate, that of three dimensional meshing or fit, which is not highlighted in equivalent English references to the same scenes. Contiguity isolates, although not perhaps as obviously ‘figural’ as those evoked by lexemes such as apple, box, run, strike, are nevertheless universally available in experiential terms.
In another article, contrasts between the way speakers of English, Spanish, German and Hebrew attend to durative and punctual aspects of events are highlighted by Slobin who argues that languages promote specific styles of thinking to support linguistic expressions of thoughts (‘thinking for speaking’. Again, the isolates he deals with are all available, regardless of enculturating languages, emphasising that aspect of Whorf’s reasoning which anchors linguistic relativity in universals of experience. In a straw man contestation of Whorf’s ideas, Kay shows how a single language (English) can provide alternative perspectivizations of events through differential assignment of thematic roles to framal participants. (e.g. buy/sell/pay/cost contrasts) and differential assignment of grammatical functions to thematic roles (as in active/passive sentence alternations). In each case, we may construe the gestaltic shifts of focus as constrasting isolations from a single event, confirming the relevance of the linguistic relativity principle to a single language and thus supporting rather than undermining Whorf’s reasoning as Kay intended. Other articles deal with discourse data, the isolate notion usefully clarifying, not only Whorf’s reasoning about linguistic relativity, but also the continuing value of the isolate construct as a means of generalizing across diverse linguistic data.