17th-19th August 2000
Organized by The Linguistic Association of Finland
at the University of Helsinki

Abstracts of Section Papers

What comes after getting up?
Textual Motivations for the 'V(get up) and V' Patterns in Finnish

Anu Airola & Kari K. Pitkänen
University of Helsinki

Our purpose in this paper is to discuss the grammatical status and the text semantic motivation of Finnish coordinated patterns _V(nous-)+ ja +V_ 'to get up and V'. The discussion in e.g. Kalliokoski (1989) illustrates well the problematic nature of coordinated _V and V_ phrases. He analyses such strings in Finnish either as elliptical clauses or as coordinated verbs according to how tightly the two lexical items seem to belong together. We argue that textual conventions a structure has need to be taken into account, when the status of parts of speech categories is considered. In this sense, the question arises in the case of these coordinated verbs, whether they could be defined as a single lexicalized unit; i.e. as serial verb constructions.

Modern corpus linguistic approaches offer new ways to systematize the grammatical description and to consider the concept of grammatical structure itself. E.g. Biber & al. (1999) introduce a corpus-linguistic innovation, namely, a 'lexical bundle', by which they mean purely statistically defined, recurring lexical strings. Among these, they mention some frequently occurring _V and V_ binomial phrases in English. In the "Pattern Grammar" framework (Hunston & Francis 2000), too, _V and V_ is mentioned among the 60 different patterns of verbs, which are defined on the basis of systematic investigations of very large corpora.

In our corpus, the pattern _V(nous-)+ ja +V_ 'to get up and V' seems to be fairly conventionalized in certain types of texts: there are about 45 occurrences of the pattern per million words in fiction, while the corresponding figure e.g. in newspapers is one. It seems that the instances of the pattern form a continuum from highly lexicalized types, e.g., _nousta ja laskea_ 'to go up and down' to more loose schematic types, e.g. _nousta_ followed by an activity verb _(V(nous-)+ ja + V(activity))_, such as _nousta ja lähteä/mennä_ 'to get up and leave/go'. The subtypes of the _V(nous-)+ ja +V_ pattern will be discussed more fully in the presentation, as well as the functional interpretations which explain why the patterns exists.

Why, then, is _V(nous-)+ ja +V_ so frequent in narratives? Actually, in many instances the verb _nousta_ 'to get up' seems redundant, since the lexical information that _nousta_ itself bears, is in fact included in the second verb of the pattern. That is, the second item often presupposes the standing posture of the moving entity, or, depending on the preceding context, the change of its posture, too. _Nousta_ does not seem to be any kind of syntactic marker, either, since in most cases e.g. the burden of expressing the aspectual nature of the whole situation falls on the second verb.

However, the grammatical status of the pattern is easier to see if one considers it in relation to a larger textual context. Namely, this structure is highly iconic. In our paper, we will argue that the iconicity is used for imitating an observer's perspective. This is important for creating the illusion of the reader's presentness within the textual world. We also argue that textually, this structure is used as a genre specific narrative strategy tied to culturally defined communicative needs. It is tied to various shared, socio-cultural situations. In this respect, it also triggers cognitive changes in the textual world, which are interpreted in relation to this type of knowledge. Additionally, it is used for manipulating textual meaning. It encodes e.g. changes from descriptive sequences to action or motion.

Thus, the form, as can be seen by the coordinated _V(nous-)+ ja +V_ structure, should be investigated in relation to the communicative needs and the contextual interaction it has in specific discourse types. This motivates the meaning evoked by the structure as a part of the text.


BIBER, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad & Edward Finegan 1999. Longman Grammar of Spoken and Written English. Longman: London.

HUNSTON, Susan & Gill Francis 2000. Pattern Grammar. A corpus-driven approach to the lexical grammar of English. John Benjamins Publishing Company: Amsterdam / Philadelphia.

KALLIOKOSKI, Jyrki 1989. Ja. Rinnastus ja rinnastuskonjunktion käyttö. SKS: Helsinki.

Grammatical descriptions vs. corpora:
The use of the dual category in four northern Uralic languages
(Ume Saamic, Khanty, Nenets and Selkup)

Antti Arppe
University of Helsinki

Grammatical descriptions of a language often simply lay out the full range of permissible combinations of morphological categories, which constitute the set of various inflected forms that each member of a particular word class may have in that language. However, one could very well expect that not all inflected forms are used equally often - or at all - when observed in corpora, and the same applies individual morphological features.

The purpose of this paper is to study the use and appearance of the dual feature in corpora (Suihkonen) of four northern Uralic languages (Ume Saamic, Khanty, Nenets and Selkup). These results will be compared to general grammatical descriptions of these languages (Abondolo). Furthermore, the association patterns of the dual feature with both word classes in general in these languages and other morphological features will be observed.


Abondolo, Daniel (editor) (1998). The Uralic Languages. Routledge, London. UK.
- Khanty: Abandolo, Daniel. pp. 359-386.
- Nenets: Salminen, Tapani. pp. 516-547.
- Saamic: Sammallahti, Pekka. pp. 43-95.
- Selkup: Helimski, Eugene. pp. 548-579.

Suihkonen, Pirkko (1998). Documentation of the Computer Corpora of the Uralic Languages at the University of Helsinki. Technical Reports TR-2. University of Helsinki.

Auxiliaries as light verbs: evidence from south Asian languages

Tanmoy Bhattacharya, Universitaet Leipzig
Hanybabu M.T., Friedrich-Schiller-Universitaet, Jena
Kalyanmalini SAHOO, University of Trondheim

The empirical fact that both Aux-V and V-Aux orders are available in languages in general suggests a possible symmetry in word order. However, it has been noticed in typological studies that although adverbs can intervene between the Aux and the V in Aux-V languages, they cannot in V-Aux languages (shown for Bangla in (1)). Our study is an important footnote to this latter demonstration of antisymmetry as we believe that it has major consequences for the theory.

(1) *likh-aste-ch-i(Bangla)
Intended meaning: 'I am writing slowly'

In investigating the distribution and nature of auxiliaries in four South Asian languages, three of which are Indo-Aryan (Bangla, Hindi and Oriya) and one Dravidian (Malayalam), we come to the conclusion that auxiliaries in these languages are really heavily grammaticalized light verbs. That is, V-Aux is really V-v in SA languages. However, in the spirit of antisymmetry and the Linear Correspondence Axiom, this order is derived from an underlying v-V order, i.e., the light verb forms an outer shell of the verbal extended projection.

Previous work on light verbs in SA languages (Hook (1991) and Butt (1997)) although agreed that they constitute a V-V complex predicate, differ as to the semantic content of the light verb. Our study differs crucially from all previous work on this topic firstly by treating auxiliaries also as light verbs (though heavily grammaticalized) and by comparing Dravidian with the general IA pattern. On the basis of the fact that the Perfective and Progressive in Malayalam are fully grammaticalized, we claim, that the Bangla and Oriya equivalents, derived from the copular, are actually heavily grammaticalized light verbs.

(2) likhe-ch-i
'(I) have written'

The implication here is that originally this verbal complex denoted a sequence of an event and a state like {writing} and {being}. This is borne out by crucial evidence more or less unnoticed in the literature that the verb stems in both language types are actually made up of the root and a particle, which, unlike the light verb, is not derived from any verbal root (for (2) it is likh+e where -e is teh particle). We present evidence that this particle is actually a reflex of a union of events. Since these particle are affixa land denote completion of an event, we take these to contribute to the aspectual information of the verbal complex. Such a structure of the verbal complex reflects the fact that these complexes are a result of a union of two separate events. However, the clearest evidence in favor of a break in the extended verbal projection comes from a whole range of possible adjunctions inside the complex.Thus Malayalam shows clefting, coordination and restrictive particle attachment and Bangla/ Oriya show attachment of topic marker, emphatic marker, non-grammaticalized light verbs and modals at this site. The structure in addition predicts the typological finding that since it is no longer a V-V sequence adverbial adjunction is not possible in V-Aux languages (see (1)).

On defining auxiliary verb

Edward L. Blansitt, Jr.
The University of Texas at El Paso

The auxiliary verb (henceforth AUX) is recognized in grammatical descriptions of many languages; the AUX concept is, however, not necessarily identical in the grammars of different languages nor in different grammatical frameworks. Many English grammars have used a language-particular definition of AUX, often using the placement of negative NOT or N'T in postverbal position as the crucial diagnostic test; using this criterion, MUST is an AUX but HAVE TO is not. Postverbal negation identifies a significant language-particular subclass of verbs in English, just as preverbal placement of clitic pronominals does for Spanish, but grammatical typology, contrastive grammar, and diachronic grammar require a cross-linguistic definition of AUX.

In adapting a term that has been used for overlapping language-particular categories to a cross-linguistic category, it an indisputable desideratum is that the cross-linguistic category will include as many as possible of the language-particular categories for which the term has been used; a cross-linguistic AUX category should not arbitrarily exclude language-particular categories for which the same term has been in wide use. The following features are proposed for the AUX category.

1. AUX IS RECOGNIZABLE AS A VERB ON MORPHOLOGICAL AND/OR SYNTACTIC GROUNDS: If a language has verb inflection, AUX may, as in some English cases, lose some or all inflection. There must still be verbal syntactic criteria that apply to AUX, e.g. restricted co-occurrence with temporals.

2. AUX DOES NOT HAVE ITS OWN VALENCE DISTINCT FROM THAT OF THE LEXICAL VERB: The valence of a predicate is normally equal to the valence of the lexical verb (henceforth LEX); an exception is the case of an AUX that makes an adjunct a complement. An example is Spanish AUX 'llevar'(+ gerund), with which a durational becomes obligatory. The complements of LEX, however, will be complements of the complex predicate; AUX and LEX, for example, cannot have competing subjects or objects.

3. AN AUX MAY BE GOVERNED BY AN AUX: This may violate a principle of some grammars that an operator is a dependent that cannot in turn be a governor: IS HAVING TO STUDY is distinct from HAS TO BE STUDYING. In some cases the AUX order may be irrelevant, as in the case of AUX HABER and AUX PODER in the conditional in Spanish.

4. AUX MAY INCLUDE ADPOSITIONS OR ADPOSITION-LIKE ELEMENTS: Spanish TENER QUE includes QUE which is generally a subordinating conjunction. French ETRE EN TRAIN DE includes a phrasal adposition.


The discussion of the definition of AUX is cross-linguistic but with special attention to Spanish; verbal periphrases in this language have been studied in detail in recent years, especially by Olbertz and by Fernandez de Castro.

The categorial exceptionality of English 'ago' in historical and typological perspective

Philippe Bourdin
York University, Toronto

English ago, as head of a phrase functioning as "scalar localizer" of a past event, has been variously categorized as an adjective, an adverb or a preposition (!). Though falling short of capturing the "full picture", it is shown that each of these categorizations has some empirical legitimacy. However, the mechanisms underlying such categorial idiosyncrasy offer a much tougher explanatory challenge.

If the coding of "scalar localization in past time" is viewed, along the lines of H. Seiler and Ch. Lehmann, as a problem for languages to solve, the emergence of ago(n) in Middle English and its subsequent grammaticalization are inseparable from the demise, in late Old English, of one of the two temporal uses of the adposition for, as exemplified by such ambiguous constructions as (nu) for nigon gearum, 'nine years ago' / 'for the past nine years'. Fine-grained textual evidence offers some clues as to why the latter meaning won out over the former, an issue which is crucial to understanding the genesis of ago(n). As well, typological evidence, involving such languages as Finnish, Estonian, Latin, Danish, German and English itself (three years back), points to the problematically variable relevance of the front/back and anteriority/posteriority axes to the "problem-solving" task of coding scalar localization.

Syntactically, there is a "small clause" feel to the internal structure of NP ago(n) which has characterized the grammaticalization of ago from the very beginning. This is also a defining feature of functionally cognate constructions in several Romance languages as well as in Dutch. Whether -- and to what extent -- we are dealing with a formal reflex of the particular semantics at play here is an open question.

Again, a two-track approach would appear to be promising.

From a language-internal viewpoint, ago(n)/agoo(n) was originally the past participle of the verb agon (OE agan), a variant of gon (OE gan), which itself later evolved into go. To that extent, ago instantiates one of the grammaticalization pathways travelled by go in English; as such, it is to be contrasted with the endproduct of other pathways (It's gonna rain, This statement will not go unchallenged) as well as set against the temporal uses of the deictic converse of go (Come September, the rainy season will start).

While deicticity is inherent in the very semantics of 'ago'-type constructions, its coding is crosslinguistically variable, ranging from the formally covert (German, Somali) to the overt. In the latter case, a number of different strategies are attested, which point to different conceptualizations of the flow of time: Polish, Irish, Rumanian and English are representative of a few strategies of deictic anchoring which appear to be, at least superficially, starkly divergent.

The story of ago exemplifies how the categorial exceptionality of a grammaticalized item can withstand the regularizing pressure of morphosyntactic change over time. As far as ago goes, there are demonstrably sound typological grounds for such remarkable resistance to have prevailed.

Numerals as a lexical category: the case of zero

Bert Bultinck

Disregarding the observation that higher numerals are more noun-like than the lower ones (Jespersen 1969, Corbett 1978, Hurford 1987), the class of numerals is usually regarded as a rather straightforward subclass of the quantifiers, functioning either as pronoun or as determiner (Poutsma 1928, Schachter 1985, Quirk et al. 1991). Nevertheless, English pronominal and adnominal usage is very different and is tightly connected to the different functions of numerals. The polyfunctionality of numerals is even more striking when we look at zero. Traditionally considered to be a numeral, zero differs in its contemporary usage from other, more prototypical numerals like two. A corpus analysis, making use of the BNC, shows not only that the distribution of zero diverges from other numerals, but also that its semantics is rather special. The analysis of various uses of zero will be the basis for the thesis that zero, rather than being a straightforward numeral, constitutes a borderline case between cardinal numerals (zero occurrences), (pro)nouns (Turnover falls to zero, From zero to hero), and non-cardinal, name-like nominal adjuncts (zero quotas, zero outlay option). The correlation between the part-of-speech status and the semantic function of the different usages will be analyzed.


Corbett, G. (1978), "Universals in the syntax of cardinal numerals", Lingua 46: 355-368.

Hurford, J. (1987), Language and Number. The Emergence of a Cognitive System. Oxford, New York: Basil Blackwell.

Jespersen, O. (1969), Analytic Syntax, New York: Holt, Rinehart and Winston.

Poutsma, H. (1928), A Grammar of Late Modern English. For the Use of Continental, especially Dutch, Students, Groningen: Noordhoff.

Quirk, R., S. Greenbaum, G. Leech and J. Svartvik (1991), A Comprehensive Grammar of the English Language, London, New York: Longman.

Schachter, P. (1985), "Parts-of-speech systems", in: Timothy Shopen (ed), Language Typology and syntactic description, 3- 61.

Shopen, T. (1985), Language Typology and Syntactic Description, Cambridge, London, New York: Cambridge University Press.

The Hawaiian pepeke system

Kenneth William Cook
Hawaii Pacific University

What does one do if one needs to teach the grammar of Hawaiian to speakers of English who have very little knowledge of grammatical categories in any language? There are at least three possible approaches: (1) use English grammatical terms even while speaking Hawaiian, (2) use the Hawaiian translations of the English grammatical terms given in the Hawaiian-English dictionary (Pukui and Elbert 1986), or (3) create new terms that fit the phonology and syntax of Hawaiian and sound enough like existing Hawaiian words to remind students of what the terms designate.

The pepeke system of Hawaiian grammatical terminology employed in Kamanâ and Wilson (1990 & 1991) takes the third approach. It employs as its model for a sentence (pepeke) the body of an octopus (he'e in Hawaiian). The three essential parts of the octopus are the head (po'o), tentacles ('awe) and the place where the head and tentacles meet (piko). These parts correspond to the grammatical categories of predicate (po'o), subject (piko), and object/adjunct ('awe).

Students in Hawaii are familiar with the anatomy of an octopus; hence the pepeke analogy gives them a reference model by which they can understand the structure of Hawaiian grammar. Hawaiian word order is predicate-subject-object/adjunct, which is the order of the body parts (po'o-piko-'awe) of an octopus swimming from right to left.

Native speakers, however, are not familiar with the pepeke terminology, which impedes communication about grammar between students who only know the pepeke system and native speakers who, if anything, know the grammatical terms that appear as entries in Pukui and Elbert's dictionary.

Although the pepeke system is used widely in Hawaii to teach the Hawaiian language and some of its terms are used to classify nouns and verbs in Mâmaka Kaiao (the 1996 dictionary of new Hawaiian words), the majority of the terms of the system have not been defined in print other than in Kamana and Wilson (1990 & 1991). In other words, definitions of the pepeke terms (in English or Hawaiian) are not readily available to students using other texts nor to the general public.

This paper defines (in English) and illustrates the pepeke terms, including those that the system uses to classify parts of speech, and it shows how the noun categories that the system proposes for Hawaiian are justifiable given the case markings these categories require when they occur in certain grammatical positions.


Kamanâ, Kauanoe and William H. Wilson. 1990. Nâ Kai 'Ewalu [The Eight Seas]: Beginning Hawaiian Lessons (revised ed.). Hilo: Hale Kuamo'o.

---- 1991. Nâ Kai 'Ewalu [The Eight Seas]: Papa Makahiki 'Elua [Second Year Class] (revised ed.). Hilo: Hale Kuamo'o.

[no author] 1996. Mâmaka Kaiao [Carry Forward into the Dawn]: He Puke Hua'olelo Hawai'i Hou [A Book of New Hawaiian Words]. Hilo: Hale Kuamo'o.

Pukui, Mary Kawena and Samuel H. Elbert. 1986. Hawaiian Dictionary. Honolulu: University of Hawaii Press.

The status of the term particle in the Finnish grammar-writing tradition

Ilona Herlin, University of Helsinki
Eeva-Leena Seppänen, Research Institute for the Languages of Finland

Up to the end of the 20th century Finnish grammars use the term particle to refer to uninflected words. This traditional use of the term has its origins in Greek and Roman grammar tradition (cf. Kärnä 2000). It took several centuries, however, before this state was reached in the Finnish grammar-writing tradition.

This paper will analyse the use of the category named particle in the tradition of Finnish grammars, from the first grammar (Petraeus 1649) until the 20th century: how this category came to the Finnish grammar-writing tradition, which elements of language have been included in its scope and for what reasons, and how its use was gradually stabilized - to become questioned again at the end of 20th century. The main focus of the paper will be on the 19th century.

The first monography dealing with Finnish particles was published in 1890 ("Suomen partikkelimuodot" `the particle forms in Finnish' by Arvid Genetz). Although Genetz was the first scholar to concentrate on particles, he referred to a stabilized tradition in the definitions and categorizations he gave. The growth of the tradition can be seen in grammars written in the 19th century. The grammars written in the 17th and 18th centuries used a category called liitepartikkeli, "clitic-particle", `clitic'. The first one to use particle as a morphologically defined category was Renvall (1840). Defined this way, the category particle included the traditional categories adverb, pre- and postposition, conjunction and interjection. Renvall's use was immediately followed. Later the category also got a semantic definition: particles "do not mean either doing or name, --- they support words of other categories" (Euren 1852: 38).

The history of the term particle in Finnish grammars allows one to discuss the factors behind linguistic description. Grammar-writing is guided by international tradition: the parts of speech which are relevant in other European languages had to be found in Finnish, too. On the other hand, the rich inflection morphology of Finnish has made morphological criteria most salient in the grammatical descriptions of it. How does tradition - both international and language-specific - guide linguistic observation? How are traditional categories matched with empirical observations? And how do new ideas get into the main-stream and affect tradition?


EUREN, G. E. 1852: Suomalainen kielioppi suomalaisille. Turku.

GENETZ, ARVID 1890: Suomen partikkelimuodot. Helsinki.

KÄRNÄ, AINO 2000: Die Kategorie `Partikel' gestern und heute. Ein über blick über griechische, lateinische und deutsche Grammatiken. Prepublications and Internal Communications 4. Department of General Linguistics, University of Helsinki.

PETRAEUS, ESKIL 1649: Linguae Finnicae, Brevis Institutio. Aboae.

RENVALL, GUSTAF 1840: Finsk språklära. Åbo.

Noun vs. Verb?

Esa Itkonen
University of Turku

In this paper I intend to discuss three languages where the noun-verb distinction may be called into question, namely Tongan (Broschart 1997), Ancient Tamil (Lehmann 1994, Asko Parpola: personal communication), and Cayuga (Sasse 1988, 1993).

In one type of Tongan sentence-structure there is first a predication slot and then a reference slot such that the former is preceded by a temporal particle and the latter by an article. Both thing-words and action-words (all of which are uninflected) may occur in these two positions. Therefore there is no formal justification for calling the former nouns and the latter verbs . - Broschart (1997) adduces several reasons for doubting the validity of the noun-verb distinction in Tongan, but to me this seems the most cogent one. Of course, much depends on how pervasive, exactly, the interchangeability of thing-words and action-words is.

Both the indigenous tradition (expounded in the old grammar Tolkaappiyam) and the mainstream Western scholarship agrees that the noun-verb distinction exists in Ancient Tamil. The situation is not quite so simple, however. On the face of it, to be sure, Ancient Tamil (just like Modern Tamil) seems to have the SOV structure. However, the finite verb is identical with the nominative of (the principal allomorphic variant of) the active participle. (There are many functionally more or less similar non-finite verb-forms.) Thus, when this participle is the subject of the sentence, the subject and the verb are formally indistinguishable (although the word order maintains the reference-predication distinction). And if the action- word which performs the predication is noun-like (in being a participle), there seems no principled reason to consider it as formally different from a bona fide noun (= thing-word) functioning as the subject and thus performing the referring function. - Notice that in this case the action-word is a subtype (= nominative) of the thing-word .

In Cayuga, both thing-words and action-words have the structure of two-place predicates, with prefixed portmanteau-morphs for pronominal agent-patient marking functioning as arguments. Semantically one-place predicates have one empty argument exactly corresponding to the English it or the German es. Thus, a meaning like The man sees the girl is expressed, in principle, by the following type of structure: (he-her)-see & (he-it)- person & (she-it)-child. The predicates with action-like meanings are inflected in all combinations of persons and in all tenses, aspects, and moods, whereas the inflection of predicates with thing-like meanings is much more restricted. Yet, it would be wrong to call the former verbs and the latter nouns because noun and verb are opposite notions whereas here the thing-words are a subtype of, or included in, the action-words qua two-place predicates. Hence, the situation is opposite to the one prevailing in Ancient Tamil.

The implications are far-reaching. It is well known that among the word-classes the status of A and P is weaker than that of N and V, because the former are lacking in many languages. To the extent that the N vs. V distinction is less than universal, the kernel of the generativist UG, formulated as X-bar theory, disappears. With hindsight, one wonders how it was possible that the word-classes of English were accepted as universal for so long, given that the universality of the crucial N vs. V distinction has been contested on cross-linguistic evidence at least since Winkler (1887).


Broschart, Jürgen. 1997. Why Tongan does it differently. Linguistic Typology.

Lehmann, Thomas. 1994. Grammatik des Alttamil. Stuttgart: Franz Steiner Verlag.

Sasse, Hans-Jürgen. 1988. Das irokesische Sprachtyp . Zeitschrift für Sprachwissenschaft.

_____. 1993. Das Nomen - eine universale Kategorie? Sprachtypologie undUniversalienforschung.

Winkler, Heinrich. 1887. Zur Sprachgeschichte: Nomen. Verb und Satz. Berlin: Dümmler.

The Noun/Verb Distinction and Deverbal Nouns in Japanese

Yasufumi Iwasaki
Mie University

In Japanese, verbs and adjectives, which inflect, constitute closed classes, while nouns, which do not inflect, are an open class (Uehara 1995, cf. Sells 1996).

To compensate for the limited vocabulary of verbs and adjectives, so-called verbal nouns (VN) and adjectival nouns (AN) combine with light verbs, functioning as verbal and adjectival predicates (Iwasaki 1997, 1999).

This paper first proposes that deverbal nouns (DN) behave like VN in combining with a light verb (i.e. "suru" meaning 'do') in nonsubject honorifics, and then explores theoretical consequences of this proposal.

(1) a. John-gaSmith-kyoozyu-omukae-ru.
' John welcomes Prof. Smith.' (Plain style)
b. John-gaSmith-kyoozyu-oo-mukaesu-ru.
'John welcomes Prof. Smith.' (Honorific style)

In (1b) the speaker shows deference to Prof. Smith by means of "o (honorific prefix (HP))-mukae suru." Harada (1976) analyzes "mukae" in (1b) as the infinitive form of the verb "mukaeru," but three pieces of evidence suggest that it is a noun zero-derived from the infinitive (cf. Sells and Iida 1991). First, the infinitive "mukae" and the honorific "o-mukae" differ in that the former, but not the latter, can take the suffix "nagara" meaning 'while.'

(2) a. John-ga Smith-kyoozyu-o mukae-nagara, ...
'While John welcomes Prof. Smith, ...'
b. *John-ga Smith-kyoozyu-o o-mukae-nagara, ...

Second, they differ in their syntactic function: unlike "mukae," "o-mukae" cannot conjoin sentences.

(3) a. John-ga Smith-kyoozyu-o mukae, Bill-ga Joan-o mukaeru.
'John welcomes Prof. Smith, and Bill welcomes Joan.'
b. *John-ga Smith-kyoozyu-o o-mukae, Bill-ga Joan-o mukaeru.

Finally, "o-mukae" can head its own NP and (4) is a paraphrase of (1b).

(4) John-ga[Smith-kyoozyu-noo-mukae]-osu-ru.

These facts strongly suggest that "mukae" in (1b) is not an infinitive form but a DN.

Drawing on Saito and Hoshi's (1998) analysis of light verb constructions involving VN and "suru," I take the "o-DN" and "suru" to form a complex predicate in the syntax (cf. Mohanan 1997). Since the DN shares an argument structure with the verb "mukaeru," (1b) parallels (1a) with respect to the number of arguments and to their theta-roles and case-markings. "Suru" functions similarly to the inflectional/tense ending "-ru." One argument for my analysis comes from the fact that the focus particle "sae" meaning 'even' can be inserted between the "o-DN" and "suru," but not between "mukae-" and "-ru."

(5) a. o-mukae sae suru (syntactically formed predicate)
b. *mukae-sae-ru (lexical verb)

Another argument concerns the fact that the VN "demo" meaning 'demonstration' can combine with the ending "-ru" in the lexicon and with "suru" in the syntax.

(6) a. demo-ru
b. demo suru

On the analogy with "demo," the DN "mukae" in (1b) can be taken as corresponding to the stem of the verb "mukaeru" in (1a). This, in turn, suggests that the difference between V and N (i.e. VN and DN) in Japanese can be attributed to the verbal inflectional properties that the ending "-ru" has. (492 words)


Harada, S. I. 1976. Honorifics. In Syntax and semantics 5: Japanese generative grammar, ed. Masayoshi Shibatani, 499- 561. New York: Academic Press.

Iwasaki, Yasufumi. 1997. Clausal Case, aspectual nouns and the extended DP analysis. Proceedings of the Fourth Seoul International Conference on Linguistics, 728-737.

Iwasaki, Yasufumi. 1999. Three subcategories of nouns in Japanese. Doctoral dissertation, University of Illinois at Urbana-Champaign.

Mohanan, Tara. 1997. Multidimensionality of representation: NV complex predicates in Hindi. In Complex predicates, ed. Alex Alsina, Joan Bresnan, and Peter Sells, 431-471. Stanford, Calif.: CSLI Publications.

Namai, Kenichi. Subject honorification in Japanese. Linguistic Inquiry 31: 170-176.

Saito, Mamoru and Hiroto Hoshi. 1998. Control in complex predicates. Paper presented at the Tsukuba Workshop on Complex Predicates.

Sells, Peter. Case, categories and projection in Korean and Japanese. Proceedings of the 1996 Seoul International Conference on Generative Grammar, 47-62.

Sells, Peter and Masayo Iida. Subject and object honorification in Japanese. Proceedings of the Seventeenth Annual Meeting of the Berkeley Linguistics Society, 312-323.

Uehara, Satoshi. 1995. Syntactic categories in Japanese: A typological and cognitive introduction. Doctoral dissertation, University of Michigan.

The category 'particle' - past and present

Aino Kärnä
University of Helsinki

Indeclinable words, especially 'particles', have always presented a problem for grammarians. In the grammars of different languages there has been no lack of suggestions about how to deal with indeclinable words. Partly this is due to the multifunctionality of particles, since they can appear as conjunctions or as adverbs, too. This paper attempts to clarify the situation by presenting an account of the history of the grammatical category 'particle' and related terms like 'conjunction' and 'adverb'.

Among the subclasses of particles there has always been one subclass, which is especially mysterious: in Greek these little words were called 'parapleromatikoi', in Latin they appeared as 'completivae' or 'expletivae', and presently they are called 'pragmatic particles' or 'discourse particles'. A typical feature of these elements is that in many languages they have a homonymous form conveying other functions. The function and meaning of a particle can only be specified in the context of its appearance.

I intend to sketch how particles are viewed in rhetorics, logics, text analysis and grammatical studies. Aristotelian rhetorics, Alexandrinian linguistics and Stoic language philosophy are dealt with as the origins of the representation of the particles. Subsequently, I will follow up their description in early grammars and briefly touch the particles in grammars of the 17th and 18th century. I also intend to show that some questions about particles have been discussed repeatedly.

Parts of speech in Sango of Central Africa

Valeri Khabirov
Ural Pedagogical University, Ekaterinburg

To classify words into parts of speech in Sango - the creolized language of Central Africa has always been a problem as the traditional definitions of parts of speech are largely unworkable because of overlappings and confusions in sango words through the lack of morphological criteria lost during the period of creolization. The controlling criterion for classifying words into parts of speech in creolized sango is syntactical. During creolization were lost the affixes serving to form in the source-language (sango-yakoma) verbs with the iterative meaning (-kV), verbs with the meaning of joint action (-ngbi), with the meaning of intensified action (-kV) and concretizing action (-ngo), with the meaning of resultative action (nda). In the creolized sango verbs with the above mentioned affixes and verbs without such affixes function as free variants with the same meaning, for example bo and bongbi have the same meaning "collect" whereas in the ethnic language sango-yakoma the word bongbi means "collect together". There are two types of words in sango. Some of them are marked semantically and belong to a definite class of words like da "house" which is a noun or biaku "immediately" which is an adverb, and others are marked semantically only partially or not marked at all, and occurring in different environments actualize each time part of their dictionary form thus belonging to different parts of speech. For example the word pendere is a noun in pendere ti wali so "the beauty of this woman", an adjective in pendere wali "a beautiful woman", an adverb in a kanga so pendere mingi "he closed it very beautifully", a verb in kodoro so a pendere "this village is beautiful", part of a compound verbal predicate in wali so a yeke pendere mingi "this woman is very beautiful", with the only productive suffixe -ngo forms a verbal noun in penderengo ti wali so "the (fact) of beauty of this woman". Thus taking into consideration semantic markedness and potential syntactic function it is possible to classify sango words into parts of speech.

[We apologise for the missing diacritics in the examples of this abstract]

The Function of (English) Adjectives

Helena Kullenberg
Lund University

Although the main function of adjectives is often claimed to be that of description, a number of scholars have argued for other functions as well. One of the first to do so was Teyssier (1968), followed by, among others, Bache (1978), Warren (1984) and Halliday (1994). These writers all suggest that there are two more functions that (premodifying) adjectives may perform besides that of pure description, namely what most call the classifying and the identifying function respectively.

Classifiers and identifiers are claimed to differ from descriptors in that they somehow restrict the range of the head noun; the former point to a subcategory, and the latter indicate a certain referent or group of referents within the class denoted by the noun. Descriptors, on the other hand, are seen as pure 'qualifiers', "indicating some specific but non-limiting, 'non-restrictive' quality" (Teyssier: 229). An example of a typical identifier would be 'best' in 'This is his best book', where 'best' 'picks out' the intended referent from the class of books; an example of a typical classifier would be 'red' in 'I bought some red peppers', where 'red' indicates a subcategory within the class of peppers, and an example of a typical descriptor, finally, would be 'lovely' in 'They have such a lovely house', where 'lovely' simply adds descriptive information about the house in question.

In the present paper I will suggest a modified, although still tentative version of the three-fold division into identifying, descriptive and classifying uses, a version that takes into consideration the referential use of the noun phrase as a whole, and the impact this has on the function of the component adjectives. In doing this, I will also argue for the existence of a fourth function, namely that of specification, in which the adjective is used to delimit what I will call a 'conceptual referent'. An example of a specifier would be 'green' in 'I want a green car', where - in the non-specific reading - the adjective helps delimiting what it is the speaker wants.


Bache, C. (1978). The Order of Premodifying Adjectives in Present-Day English, Odense:Odense University Press.

Halliday, M.A.K. (1994). An Introduction to Functional Grammar, London: Edward Arnold.

Teyssier, J. (1968). "Notes on the Syntax of the Adjective in Modern English", Lingua 20, 225-249.

Warren, B. (1984). Classifying Adjectives, Gothenburg: Acta Universitatis Gothoburgensis

"You" and "I" in Japanese - Or what do "personal pronouns" do in Japanese discourse? -

Riikka Länsisalmi
National Museum of Ethnology, Osaka

It is common knowledge that in Japanese subjects are frequently ellipted. This is the reason there is often no need for an overt term referring, for example, to the first or second person. Ellipsis usually occurs in what is referred to as unmarked situations, i.e., cases where the referent can be identified through a grammatical construction or inferred from honorific mode or deictic expressions, or it represents the (paragraph) topic or can be understood through contextual cues. Studies of ellipsis in Japanese abound, but the natural extension of such inquiries, namely, those aimed at answering the question "What functions do terms that are generally ellipted have when they are present?" are fewer in number.

This paper is intended to fill in some of the gaps left by earlier studies dealing with (so-called) Japanese personal pronouns. The question of whether Japanese has an independent category of personal pronouns has been a controversial topic in Japanese linguistics, and researchers appear to be divided on the issue. Some make no distinction between nouns and personal pronouns (and other pronouns), while others prefer to employ nonstandard terms, such as "person terms" or "ninshoo meishi" 'personal nouns', instead of "personal pronouns" or talk about a noun-pronoun continuum, and still others simply speak of a class of Japanese personal pronouns with no reference to any possible unsettled questions concerning this topic. The latter generally consider Japanese to have an extremely high number of personal pronouns, from which a speaker of the language must select the most appropriate one in any given situation, taking into consideration the relations existing between the speaker, addressee and other persons present, the formality of the situation, etc. Although I have adopted the standard term "personal pronoun" in the present study, I intend to demonstrate that, in many cases, linguistic entities covered by this term in Japanese do not correspond to personal pronouns in languages such as English or Finnish, for example.

I carry out my examination by analyzing first and second person pronoun use in conversational interaction depicted in five Japanese films and discuss the following points: (1) various pronominal forms and restrictions related to their use in conversation, (2) structural environments requiring the use of pronominal forms, (3) personal pronouns as vocative terms, (4) personal pronouns as affect keys, (5) personal pronouns as "strategic tools".

Towards a Typological Characterization of Aspectual Light Verbs:
Evidence from Mandarin Chinese

Dianyu Li
Gothenburg University

It has long been observed that most prepositions in Chinese have verbal origins. Standard criteria for testing the verbhood of these words seem unable to decide their categorial status, especially for a subset which shows syntactic properties that make it heterogeneous from other prepositions. The present paper presents an approach to analysing this subset from the perspective of their cognitive develop- ment 'path' by which their polysemic structures are formalized. The semantic content and the variant context-dependent interpretation of these words are defined in the specific constructions they appear in where notions such as aspect and bounded- ness are applied to investigating their grammatical functions and their categorial status. As a result of the study, it is proposed that these words be categorised as aspectual light verbs in contrast to their full verb status when the two cases show distinct but related properties in terms of aspectuality.

Key Words: aspect, continuity, instantaneity, boundedness, polysemy, cogntive semantics

The lone adjective of Toqabaqita

Frantisek Lichtenberk
University of Auckland

Toqabaqita is an Austronesian language spoken in the Solomon Islands.

Toqabaqita has a well marked distinction between nouns and verbs, and most lexical items only belong in the former or in the latter category. At the same time, there are some areas of the grammar where the two categories exhibit similar properties: both nouns and verbs can form the head of a predicate, and both can modify nouns. The large majority of verbs used attributively are intransitive, but there are also a few transitive verbs so used.

Besides the nouns and the verbs, there is one lexeme that is significantly different from those two categories. This unique lexeme is an adjective; its meaning is 'small, little (in size or quantity)'. First, unlike the nouns and the verbs, it can only be used attributively in noun phrases to modify the head noun; it cannot be used predicatively. Second, while nouns and verbs used attributively in noun phrases follow the head noun, the adjective precedes it. And third, the lexeme is manifested by three partially similar word-forms, whose use is sensitive to the following distinctions associated with head nouns: singular-plural, animate-inanimate, and count-mass. None of these distinctions apply to attributively used nouns and verbs.

The paper investigates the morphosyntactic properties of the adjective and its differences from the nouns and the verbs. The history of the adjective also is discussed. There is evidence that the adjective derives historically from a noun meaning 'child'. While the meaning 'small' is expressed by an adjective, its opposite, 'big', is expressed by the nouns 'mother' and (archaic) 'father'.

Dixon (1994:34) suggests that a closed class of adjectives will have "anything between about five and around one hundred [members]". As the Toqabaqita case demonstrates, the number of adjectives in a language may be as low as one.


Dixon, R.M.W. 1994. Adjectives. In R.E. Asher (editor-in-chief), The encyclopedia of language and linguistics, vol. 1, pp. 29-35. Oxford: Pergamon Press.

Adverbs everywhere or something else?
Estonian syntax and some problems related to parts of speech.

Peep Nemvalts
Uppsala University

What is the role of parts of speech, or word classes, in describing syntax? How should words be classified? What kind of classification is purposeful? It is generally accepted that word-class definition is problematic, often because it attempts to embrace all the information lying in the language structure as well as the conditions of use. Although it is not possible to avoid cross-classification, it would perhaps be useful to adopt the viewpoint of either the sender or the receiver of messages. In the first instance we would get word classes described more semantically: substantives, determiners, indefinators, quantifiers etc. In the other we would get word classes defined more on morphosyntactic grounds: nouns, verbs, adverbs or particles, pre- and postpositions.

I shall present my view on which word classes should be distinguished for a rational description of Estonian syntax, mainly on semantic grounds; the prototypical determiners are qualifiers (incl. adjectives), identifiers and quantifiers. Unlike adjectives, identifiers do not determine the referent of a substantive qualitatively, but draw attention to the referent, helping to identify this referent either (i) as an imprecise member or an inexactly defined subset of a set or class, or (ii) as a specific member or subset of a set or class. Various terms have been used for these kinds of determiners in linguistics. The first type are often called indefinite determiners, the others specifying determiners. Neither of these terms, nor the term identifier, have been employed in Estonian linguistics before Nemvalts 1996. I do think however that they are also quite useful from a semantic point of view and relevant in describing Estonian. I call the first type indefinators, the others specifiers.

Crucial questions are: what are adverbs? What is the relationship between adverbs and particles? And how are adverbs and adverbials related to each other? There has often been a confusion in using the terms 'adverb' and 'adverbial'. Is 'adverb' a universal notion or can it be used language-specifically? In Estonian grammars, terms like afiksaaladverb 'affixal adverb', kvantoradverb 'quantifier adverb', kvantiteediadverb 'quantity adverb', modaaladverb 'modal adverb', määraadverb 'amount adverb', iseseisvad adverbid 'independent adverbs', adverb adjektiivatribuudina 'adverb as adjectival attribute', fraasiadverbiaal 'phrase adverb' are used. I am unsure as to whether all these words should be defined as adverbs.


Nemvalts, Peep 1996: Case Marking of Subject Phrases in Modern Standard Estonian. Acta Universitatis Upsaliensis. Studia Uralica Upsaliensia 25. Uppsala. 151 pp.

Proprietive as a mixed category

Irina Nikolaeva
University of Konstanz

A number of studies (Corbett 1987, 1995) have shown that the so-called possessive adjectives in Slavic demonstrate an unusual syntactic behaviour: the noun underlying them retains certain controlling properties. With the assumption that only inflectional morphology is transparent for syntax, this indicates that possessive adjectives are formed by an inflectional process. In contrast, other pieces of evidence rather point towards a derivational formation. This suggests that, first, there is no clear-cut distinction between derivational and inflectional morphology, and second, that word-class membership for such forms cannot be unambiguously established.

This conclusion is supported by the data from Tungussic. In Tungussic, nouns and adjectives have similar inflectional properties, but undoubtedly differ in their distributional characteristics, such as position with respect to head, the possibility of raising and the co-predicative use. A proprietive/comitative meaning ('with X') can be rendered by various adnominal constructions, some of which are clearly "nouny" and others clearly adjectival. However, the proprietive forms in -xi/-chi/-ku/-ko demonstrate a mixed behaviour. Distributionally they behave like typical adjectives, and so, according to the standard tests, should be classified as adjectives. However, the noun underlying these forms is accessible for syntactic processes: it can be pluralized and take a possessive inflection; it can head its own syntactic phrase; in some languages it triggers agreement on the modifier, and it is available for cross-reference by an anaphoric pronoun. So the external syntax of the proprietive forms is adjectival, while with respect to their internal syntax they behave as inflectional forms of nouns, just as is the case with Slavic possessive adjectives.

Such a distribution of the properties of mixed categories lying "in between" nouns and adjectives is not accidental. Adjectives are prototypical adnominal modifiers, so if a noun is used as an adnominal it may acquire certain adjectival properties (cf. Plank 1995). These will be exactly those syntactic properties that characterize an adjective as a modifier (such as, for example, position with respect to head). On the other hand, even when used adnominally, the underlying noun may retain its referentiality. In this case it may function as a regular discourse referent, that is, trigger anaphoric processes. The degree of "nounyness" for such mixed categories seems to depend on the extent to which the underlying noun is referential (definite, specific). Evidence from Tungussic and some other languages suggests that proprietives derived from definite human nouns are typically the most "nouny", while proprietives derived from abstract or non-referential concrete nouns are close to regular adjectives. In other words, the categorial status of morphologically complex adjectival and nominal adnominals may be represented in a form of a continuum. The position of each particular form on the scale is to a large extent determined by the referential characteristics of the base word, but other factors (such as the origin of the form) should perhaps be taken into consideration as well.

Verbal noun constructions in Japanese

Masako Ohara
University of Essex

This paper examines characteristics of verbal nouns (VNs) in Japanese. VNs have mixed properties of verbs and nouns, and the examination of VNs not only reveals their properties, but also casts light on the characteristics of nouns and verbs in general.

Although a VN typically appears with a light verb _suru_ 'do', and functions like a verb, it is not the only environment where it behaves like a verb. For example, when an aspectual affix _-tyuu_ '-while' is attached to a VN, the VN behaves like a verb in the sense that it takes (an) argument(s). As illustrated in (1a), arguments of a VN _HAITATSU_ 'delivery' are marked by nominative and accusative cases respectively, which is commonly observed when a verb takes (an) argument(s). Interestingly, it is also possible for a VN to behave like a noun. In (1b), the arguments are marked by genitive case. If genitive case marking on (an) argument(s) is considered to be a characteristic of a nominal head, _HAITATSU_ 'delivery' in (1b) may be regarded as a noun. Yet another intriguing property of the VN construction is that a mixture of case marking patterns is also possible.

Thus, in (1c), one argument is marked by nominative case, while the other is marked by genitive case. However, if we mark the subject argument with a genitive, and the object argument with an accusative marker, the sentence is ungrammatical as in (1d).

(1) a. [Taroo-ga piza-o HAITATSU-tyuu]-ni, jisin-ga okotta.
Taro-Nom pizza-Acc delivery-while-at earthquake-Nom occurred
'An earthquake occurred while Taro was delivering pizzas.'
b. [Taroo-no piza-no HAITATSU-tyuu]-ni, jisin-ga okotta.
Taro-Gen pizza-Gen delivery-while-at earthquake-Nom occurred
c. [Taroo-ga piza-no HAITATSU-tyuu]-ni, jisin-ga okotta.
Taro-Nom pizza-Gen delivery-while-at earthquake-Nom occurred
d. *[Taroo-no piza-o HAITATSU-tyuu]-ni, jisin-ga okotta.
Taro-Gen pizza-Acc delivery-while-at earthquake-Nom occurred

Thus, a VN shows various case marking patterns, typical of a verb, a noun, and a mixture of both. In examining the patterns exemplified in (1), we can ask the following questions: a) what sort of role(s) do case particles play in the syntax, which appear in various patterns with VNs; and b) what kind of properties do VNs have.

In this paper, it will be argued that case particles can select the type of category appearing in the governing phrase. Verbal case marking (i.e. nominative, accusative, etc.) appears inside a VP, whereas nominal case marking (i.e. genitive) appears inside an NP. Concerning the property of VNs, it will be proposed that they have morphological properties typical of nouns and argument structure of verbs. While an ordinary V or an ordinary N can appear as a head of VP and NP respectively, a VN can appear either as a head of VP or NP because of its mixed characteristics. The dual property of VNs, in accordance with the characteristics of case particles, determines the overall appearance of the sentences with VNs.

The examination of VNs offers the key to understanding interaction of morphology and syntax. It is a morphological item like a particle which can help to define overall appearance of phrase structures.

State denoting local case expressions in Finnish - nouns, adverbs, or periphrastic stative constructions?

Tiina Onikki
University of Helsinki

"A good part of language behavior takes place between, rather than within, linguistic categories such as word classes." (Heine, Claudi & Hünnemeyer 1991: 251.) In this paper I will offer an overview of such a case in Finnish. It concerns the polysemy of Finnish local cases and a continuum which leads to a particular stative predication type, state denoting local case expressions. On one hand they are frozen lexemes, on the other hand components in a grammaticalized construction type.

Some of them are traditionally categorized as adverbs separated from the productive declination of their stem (e.g. hän on sukkasillaan sukka+s(e)+i+lla+aan 'sock+derivative suffix+plural+external case+possessive suffix' 'he/she is having (only) socks on'; hän on nukuksissa nuk+uks(e)+i+ssa 'sleep+derivative suffix+plural+internal case' 'he/she is feeling sleepy' or 'sleeping'), others could be seen as part of the declination of nouns (hän on kovassa humalassa; 'hard+internal case drunk+internal case' 'he is heavyly drunken' notice the agreement of the attribute and its head in inessive case). This means that the demarcation line between nouns and adverbs goes inside the category of state denoting local case expressions. I will survey the status of this kind of expressions in the comparison with the criteria for parts of speech as prototype categories. I will also relate this expression type to grammaticalization paths which lead to periphrastic stative constructions (cf. Stassen 1997).

Syntactic and Semantic Categorization of the Finnish Terminative = Particles asti and saakka

Pia Päiviö
University of Turku

The term "Terminative" is generally used to designate a morphological case that indicates a border or a limit of motion or other kind of relation. In Finnish, terminativity is generally expressed with the local cases (cf. Alhoniemi 1978), and Finnish terminative expressions can be divided into four main groups: spatial, temporal, qualitative and quantitative terminatives. In addition, there are a few terminative particles used to strenghten the terminative meaning of the utterance, and to disambiguate the interpretation of the locative-case expression as primarily terminative not locative (e.g. Turku+un [Turku+ILL] 'to Turku' vs. Turku+un asti [Turku+ILL asti] 'all the way to Turku; no further than Turku, etc.').

The Finnish terminative particles asti and saakka, which are synonymous to a large extent, have generally been categorized as independent clause-level adverbials. However, some scholars (e.g. Penttilä 1957) have suggested that they should be classified as postpositions with the locative-case element as their modifier. However, the fact that this assumed modifier is declined in the local cases instead of the grammatical cases partitive and genetive has been problematic for the classification (e.g. Ikola 1989), since it has been assumed since Setälä (1952) that Finnish postpositions only take modifiers in the genetive and the partitive, not in the local cases.

In my presentation I will discuss the different viewpoints presented by Finnish linguists on the categorization of asti and saakka, and will shed some new light on the categorization problem of these two particles through a semantic study of their uses. My main argument will be that these elements are to be classified as focus particles (in the sense of König 1991), which focus on terminative expression, suppressing the locative interpretation.

Poor linguists - on the outskirts of the category of adjectives

Carita Paradis
Lund University

Traditionally, adjectives are defined in terms of notional, morphological and syntactic criteria. Adjectives are property concepts. They may undergo comparison. They are modifers of nouns and they occur both attributively and predicatively. As is well known, these criteria do not hold good for all the words that are traditionally considered to be adjectives. Rather, they are characteristic of what are considered to be typical adjectives. Within the framework of cognitive linguistics, grammatical categoreis are notionally characterized. Langacker (1987a, b) makes a broad distinction between two kinds of predications - those that designate entities (nouns) and those that designate relations (verbs, adjectives, adverbs and prepositions). He also says that there are two types of relational categories. There are verbs which are connected to time. They designate temporal relations. Adjectives, on the other hand, are atemporal and designate states.

Typical adjectives are content words with a characterizing function (a tall building, a good book, a nasty boy). From the assertion It is a tall building, we can infer It is a building and It is tall. Characterizing adjectives exhibit all the morphological and syntactic characteristics that are traditionally used as criteria of adjectivehood. They may occur both attributively and predicatively (a tall building, the building is tall), and they may be used in the comparative and the superlative (taller, tallest).

There is another group of adjectives whose function is to classify the nouns they modify (a financial problem, a pictorial atlas). These adjectives have a categorizing function. The adjective and the noun (a financial problem) designate, in this case, a particular kind of problem namely a financial problem. 'Financial problems' represent a subset of 'problems'. Classifying adjectives can be said to have nominal functions. From a morphosyntactic point of view, classifying adjectives differ from characterizing adjectives in that they occur primarily in attributive position, and they cannot undergo comparison.

The focus of the present paper is on items that are traditionally considered to be on the outskirts of the category of adjectives, namely adverbal adjectives, such as fast food, frequent visitor, real gentleman, certain winner, perfect idiot and utter nonsense. The issue under inquiry is: What do these adjectives have in common and on what grounds do they form a subclass of adjectives?

Adverbal adjectives are not descriptive of the noun per se but of a particular aspect associated with the noun (Warren 1984: 229). I am using poor linguists to illustrate several interpretations of an adjective in one and the same expression. Poor has three interpretations:

i) poor = not rich (characterizing)
ii) poor = bad (adverbal)
iii) poor = I feel sorry for them (adverbal)

The relatively complex semantics of linguist makes several adjectival readings possible. The reason why poor may take three different readings here is that it latches on in three different ways to linguist. In cognitive terms, three different valence relations are profiled. In the first reading, poor is descriptive of the noun linguist. Linguists are 'representatives of a group of employees' and they are badly paid. The entailment relation that 'poor linguists are linguists who are poor' holds.

The second reading of poor linguists is 'bad linguists' In such a context 'the activity of doing linguistics' is drawn out. As a linguist this person is not good. Poor maps on to the verbal aspect of doing linguistics and it modifies it with respect to manner. Poor can be said to be an adverbal adjective, since it evaluates the aspect of doing linguistics. It differs from characterizing poor in that the entailment relation that 'poor linguists are linguists who are poor' does not hold good. But, it is similar to characterizing adjectives in that it may under go comparison.

The third reading of poor linguists differs from the others in that it is not descriptive of linguists in any way. It reveals the speaker's attitude to linguists. It has a modal function. This reading of poor can only be used attributively. The entailment relation does not hold good ('Poor linguists' does not entail 'They are linguists and they are poor'). Poor cannot undergo comparison. Adverbal adjectives such as poor linguists! differ from the manner adjective in that they have very little propositional meaning. It is an exponent of a subjectivity on the part of the speaker (Paradis 2000). Obviously, poor meaning 'I feel sorry for' does not only have very little in common with characterizing and classifying adjectives, but it is also far removed from poor, meaning 'bad' as a modifier of 'somebody doing linguistics' in such a way. It is not a property concept, but an epistemic marker. If the members of the above three types of modifiers are adjectives, we may have to reduce the definition an adjective to an atemporal relational item which connects to a noun.


Langacker, R. 1987a. Nouns and verbs. Language 63.1.53-94.

Langacker, R.W. 1987b. Foundations of cognitive grammar. Stanford: Stanford University press.

Langacker, R.W. 1988. The nature of grammatical valence. In Topics in cognitive linguistics, ed. by B.Rudzka-Ostyn. 91 =96125. Amsterdam: Benjamins.

Paradis, C. 2000. Reinforcing adjectives. A cognitive semantic perspective on grammaticalization. In Generative theory and corpus studies, ed. by R. Bermudez-Otero, D. Denison, R.M. Hogg & C.B. Mc Cully. Berlin: Mouton de Gruyter.

Warren, B. 1984. Classifying adjectives. Gothenburg Studies in English 56.

Noun and verb fluency in Alzheimer's disease

Seija Pekkala
University of Helsinki

Background: Alzheimer’s disease (AD) is characterized by the impairment of multiple memory-related systems. Due to a disorder in the language specific semantic memory, which contains our knowledge of concepts, of words and their meaning, the semantic relations of nouns have been shown to be distorted in AD. The AD patients are impaired in retrieving and identifying the names of words. In category fluency task with nouns, their ability to produce category-related words, to organize semantic information into clusters of subcategory items, and to shift mental set in their search for information is deteriorated. The nature of the impairment of the semantic system is debated. Several different interpretations have been proposed, such as lack of access to relatively preserved representation of knowledge and an impairment of the processing of this information, or decay of information itself within semantic system.

Methods: In order to study the integrity of both the noun and verb categories and the search of subcategory words within the semantic fields, the semantic fluency task was performed by 20 mildly and 20 moderately demented Alzheimer’s (AD) patients and 30 normal controls (NC). The semantic categories of concrete nouns were clothes, vegetables, vehicles and animals and the categories of concrete verbs were to cook, to do sports, to build and to clean. The total number and types of words, the number and type of errors, as well as number and type of clusters of related subcategory names were recorded. The participants were given 60 s to generate words into each semantic category.

Results: NC participants were superior to both AD groups, the mild AD group being less impaired than the moderate AD group. An increase in errors, such as repetitions, category violations and intrusions, and a decrease in correct responses contributed to the performance of the AD group. The AD participants showed less clustering within categories and smaller cluster size. In all groups, more nouns were produced than verbs. In the moderate AD group, fewer pure verb forms were found. Instead, the number of less specific verbal expression (a more general verb form and its arguments) and deverbal nouns was bigger than in the other groups.

Conclusions: The AD participants do not perform the semantic fluency tasks in the same manner as normal participants. AD participants tend to violate the boundaries of the semantic categories and they generate fewer exemplars from subcategories, because the knowledge necessary to organize exemplars in semantically meaningful fashion is lost or degraded, or they are unable to access or to use their semantic knowledge appropriately. The differences between noun and verb production suggest category structure influences on retrieval processes. The category verbal fluency task seemed to be sensitive to semantic memory deficits in AD.

From preposition to verb: the short and the long route

Frans Plank
University of Konstanz

The prepositions (and adverbs) UP and DOWN in English, so far as we can reconstruct their past, derive from an ideophone (*(I)UP 'fast upward movement') and a prepositional phrase (OF DUNE '(down) from the hill'),respectively. The local (and perhaps comitative) prepositions GHAND 'at' in Maltese and INTA- in Coptic, so far as we can reconstruct their past, seem to derive from an accusatival verbal noun (*IMM-A 'embracing, grasping'), which, once a preposition, took an NP complement (*IMM-A JAD-I 'at the hand of'). These are typical instances of the grammaticalization of function words (prepositions) from lexemes (nouns, verbs, ideophones), or rather from lexemes in syntactic constructions.

But the story doesn't end here. There are way and means for these very function words just quoted to become proper lexemes again. The very popular short route back to verbhood or nounhood is illustrated by English: LIFE HAS ITS UPS AND DOWNS, TO UP ONE'S IMAGE, TO DOWN ANOTHER PINT, etc. The mechanism is conversion, and despite attendant historical changes in this respect (these and the immutable constraints will be briefly touched on in the paper), it is questionable whether using this mechanism as such means changing one's grammar. But the long route, the main interest of this paper, is indubitably one of change of both morphology and syntax, and if this is not degrammaticalization then I don't know what is. Both in Maltese and in Coptic (and elsewhere in Afroasiatic and perhaps beyond) local/comitative prepositions end up as transitive verbs of possession ('to have'), in both cases retaining a few odd inflectional traits of their prepositional past. The rise to verbhood in both cases involves reanalyses of grammatical relations, conditioned by marked word order (left dislocation of complement of the preposition in Maltese: 'the cow [is] at the uncle' > 'the uncle, [is] at-him the cow'; VSO-to-VOS in Coptic: 'is a wife at the father' > 'is at the father a wife'). A difference is that in Maltese the preposition on its own gets reanalysed as a verb, with no copula present in present tense clauses ('the uncle has-he the cow'), while in Copic, which has an overt present tense copula (existential verb), the preposition is reanalysed as belonging with the existential verb rather than with its complement and is actually univerbated with it, yielding 'have' ('is-at/has the father a wife').

Note: In the examples quoted there are all sorts of special symbols missing, owing to the medium.

A Re-examination of the Parts of Speech in the Southern African Languages

George Poulos
University of South Africa

In traditional analyses of the Southern African Languages, the Parts of Speech were identified and classified in terms of morphological, syntactic and semantic criteria. This approach culminated in the creation of so-called 'major' parts of speech, which were further sub-divided into sub-types.

A typical classification within this traditional framework of analysis was as follows:

Major Parts of Speech Sub-types/Sub-divisions
i)The Noun
ii)The PronounAbsolute
iii)The Demonstrative
iv)The QualificativeAdjective
v)The Verb
vi)The CopulativeIdentifying
vii)The AdverbInstrumental
viii)The Ideophone
ix)The Interjection
x)The Conjunction
xi)The Interrogative

In this paper, the parts of speech will be investigated from a different angle to that used by the traditional grammarians. This analysis will in essence encompass typological considerations. To this end, the core data will be representative of four major linguistic groups , namely Nguni (i.e Zulu, Xhosa, Swati, Ndebele), Sotho (i.e Southern Sotho, Northern Sotho, Tswana), Tsonga and Venda. These languages are typically agglutinating in their structure, and thus tend to reveal in many instances, the dynamics of language change from true lexical items to morphological elements in various developmental stages. With this in mind , it will be shown how certain linguistic processes such as grammaticalisation can throw more light on the nature of the 'basic' lexical categories, and in so doing, further our understanding of the true status of the parts of speech in these languages.

It will thus be argued that typological considerations as well as an insight into the dynamics of language change can lead to a different perspective on the classification of the parts of speech. A distinction will be made between the basic parts of speech, inter alia, the noun and the verb, and those which are merely derivative forms. It will also be shown that the basic parts of speech and their derivative forms are not discrete categories as 'traditionally' assumed. Instead, they operate on a type of linguistic continuum, which reflects to a large extent our underlying conceptualisation of the universe.

From nouns to prepositions: a grammaticalization process in Thai

Amara Prasithrathsint
Chulalongkorn University, Bangkok

Thai is an isolating language, in which inflection is completely absent. Therefore, it is more difficult to differentiate parts of speech in Thai than in inflectional languages. The lack of morphological marking makes intransitive verbs and adverbs look very much alike superficially. Also, it is almost impossible to differentiate a prepositional phrase from a verb phrase containing a transitive verb followed by an object noun or from a noun phrase containing two nouns.

Therefore, the preposition category in Thai is problematic. Unlike in some other languages, such as English, there is no consensus as to which words are prepositions. One linguist even goes to extreme by taking the position that there is no preposition in Thai (Warotamasikkhadit 1996: 152-158). However, most linguists, including myself, maintain that prepositions are a part of speech in Thai.

Some prepositions in Thai are intrinsic; i.e., as far as evidence shows, they have been used only as prepositions for more than 700 years; for example, /kap/ 'with', /kae/ 'to', /phuea/ 'for', /dooy/ 'by/ (Prasithrathsint 1994). Others are derived from verbs; e.g., /khaam/ 'across' (<'cross'), /caak/ 'from' (<'leave'), /thueng/'to' (<'arrive'), /suu/ 'towards' (<'head towards') (Intratat 1996). There is yet another group of prepositions in Thai which seem to derive from nouns; e.g., /nay/ 'in' (<'inside'), /bon/ 'on' (<'top'), /naa/ 'in front of' (<'face, front'), /lang/ 'behind' (<'back'), /khaang/ 'beside' (<'side'). Some syntacticians regard them as prepositions, but others label them as "relator nouns". This conflict has been observed for a long time but has not been deeply investigated. In this paper, I will maintain that they are prepositions. My purpose is to analyze them and provide evidence to support the grammaticalization theory that prepositions can derive from nouns.

The analysis is based on a two-million-word corpus of current Thai. It is found that the processes that support the recognition of such words as grammaticalized prepositions are: reanalysis, obligatorification, grneralization, persistence, semantic bleaching, and metaphorical extension. (concepts proposed by Hopper 1991, Heine 1991, Hopper and Traugott 1993, Bybee et al. 1994).


Bybee, Joan L., R.Perkins, and W. Pagliuca. 1994. The Evolution of Grammar. Chicago: The University of Chicago Press.

Heine, Bernd. 1991. Grammaticalization A Conceptual Framework. Chicago: The University of Chicago Press.

Hopper, Paul J. 1991. On some principles of grammaticalization. In Approaches to Grammaticalization, Vol. I: 17-35, edited by Elizabeth Traugott and Bernd Heine. Amsterdam: John Benjamins Publishing Co.

Hopper, Paul J. and Elizabeth Traugott. 1993. Grammaticalization. Cambridge: Cambridge University Press.

Intratat, Charatdao. 1996. Grammaticalization of verbs into prepositions in Thai. Ph.D. dissertation, Chulalongkorn University, Bangkok, Thailand.

Prasithrathsint, Amara. 1994. Three intrinsic prepositions in Thai. Ramkhamhaeng Journal, Vol. 17, No.2, pp. 89-104.

Warotamasikkhadit, Udom. 1996. Thai Reference Grammar (written in Thai). Bangkok: Rongphimchuanphim.

Areal Effects on the Preservation and Genesis of Slavic Postpositions

Donald F. Reindl
Indiana University, Bloomington

Despite the overwhelming syntactic tendency toward prepositions in the Slavic languages, several Slavic languages occasionally make use of postpositions as well, for example, in Upper Sorbian:

'because of the bad weather'

Although Slavic postpositions have been pessimistically characterized as marginal, they remain a integral part of the living languages. This paper attempts to account for the presence of these syntactically unexpected elements in Slavic languages.

Three main hypotheses may be proposed to account for postpositions in Slavic: preservation of an archaic syntactic system, evidence of an underlying left-branching phrase structure, or borrowing at both the lexical and syntactic levels.

Old Church Slavic does, in fact, exhibit a small number of postpositions; however, several modern Slavic languages admit more postpositions than are attested in Old Church Slavic--exactly the opposite of what would be expected on the basis of the archaism theory.

The phrase structure argument for Slavic postpositions is based on typological arguments; namely, that VO and VS orders correlate with prepositions, and that OV and SV orders correlate with postpositions. However, no matter how many postpositions are exhibited in Slavic languages, they remain basically prepositional. Also, lack of correlation with the placement of adjectives and adnominal genitives argues against an underlying phrase structure explanation.

Borrowing is advanced as the best solution to the origin of the great majority of Slavic postpositions. Sociolinguistic and areal factors point to German as the likely source language. The syntactic borrowing that this implies is cross-linguistically comparable to similar phenomena in Armenian and Creole French.

A number of German postpositional constructions have been identified as possible sources for semantically parallel constructions in the Slavic languages. For example, as a counterpart to (1), compare the (now dated):

(2) Meines Bruders wegen bin ich gekommen.
my brother.GEN because-of aux I come.PTCP
'I came because of my brother.'

The issue of the (now defunct) pleonastic prepositions in Slavic is touched upon and excluded from the analysis as a separate issue.

The hypothesis of German influence on Slavic postpositions is in line with other observations of syntactic and lexical effects of German on adjacent Slavic languages. Furthermore, the development of both the Slavic and German postpositions is in line with widely recognized patterns of grammaticalization.

Speech act(ion) expressions (SAX) as parts of speech - Overcoming the word boundary -

Rudolf Reinelt
Ehime University, Matsuyama, Japan

1. Introduction: Aiming beyond the word boundary While so far research in word classes has often been limited by the word boundary, this paper attempts at introducing a part of speech whose members each consist of a small number of separate lexical items. As an example, we consider speech act(ion) expressions(SAX) comprising traditional "speech act verbs" and other elements with related functions.

2. Speech act verbs in Indoeuropaean and other languages Considering only the Indeuropaean languages, we find speech act verbs (Verschueren 1998) the dominant means for indicating or performing a speech act(ion). Grammatically, they share all syntactic and most semantic features with other verbs. Usually, but not always, there is a quotation marker, and, in writing, an indicator of direct speech. The situation changes, however, completely, if we extend our view to other languages, where sometimes "there is, quite simply, no viable distinction between nouns and verbs" (Gil 2000: 174). This also affects the characteristics of SAX and their analysis.

3. SAX across languages From both a diachronic and a synchronic perspective, this paper will demonstrate how SAX are realized in a few classical languages and literatures, such as in Sumerian/ Akkadian (Gilgamesh: He talked to his mother and said ...) and Ancient Chinese (On day X ...Y prophesied and said....(shell bone inscriptions)) , and consider a contemporary Japanese short story example: Seiteki Ningen by Oe Kensaburo (new ed. 1996).

4. Result: The SAX quintuple In summary, we can characterize speech act(ion) expressions(SAX) as a quintuple of elements which
- span several traditonal "parts of speech",
- can be conflated, and
- consist of, at least, the following constitutive elements:

(The SAX quintuple)
Kind of speech act(ion) indicator + Ifid + Expression of speaking (verb, noun) + quotation marker + contents of utterance

Depending on the language and/or type of text, only some or all 5 of these may appear on the linguistic surface. Each can take over all the functions of the other elements on its own (+zero). While sharing a number of facultative categorial features, the main characteristic is , however, to establish a speech act of a specific category.

5. Towards a wider theory of parts of speech Finally, we want to show that our discussion of SAX fulfills Croft's (Croft 2000: 83) criteria for a proper theory of parts of speech by providing:
- distinguishing criteria vs other morphosyntactically defined subclasses,
- a uniform set of grammatical criteria, and
- the distinction between language universal and particular language facts.


Croft, William (2000). "Parts of speech as language universals and as language-particular categories". In: Vogel, Petra & Comrie. Bernard, (eds.) Approaches to the Typology of Word Classes. Berlin/New York, Mouton de Gruyter. p. 65 - 102.

Gil, David (2000). "Syntactic categories, cross-linguistic variation and universal grammar". In: Vogel, Petra & Comrie. Bernard, (eds.) Approaches to the Typology of Word Classes. Berlin/New York, Mouton de Gruyter. p. 173 - 214.

Verschueren, Jef. (1998). Speech Act Verbs. In: Mey, J. (ed.) Concise Encyclopaedia of Pragmatics. Amsterdam: Elsevier.

Clause-Like Structures Internal to Adpositional Phrases

Carol Rosen
Cornell University

While the verb/noun distinction will always be useful, syntactic theories have much to gain by taking as a primitive an over-arching notion of predicate which cuts across parts-of-speech categories.  It is a truism that common nouns are predicates, and the familiar idea that phrases headed by nouns are somewhat clause-like in their internal structure continues to resurface (as in the DP hypothesis) and to find increasing corroboration.  But another path of research, not so well worn, begins from the recognition that prepositions and postpositions too are predicates.

With the data displayed in this talk, I want to suggest that the study of cross-categorial parallels can achieve dramatic advances by working with the following hypotheses:

(1) Predicates which are morphologically non-verbs have argument structures which are best represented in terms of the same concepts RG's 1, 2, 3).

(2) The foregoing applies not just to the internal structure of noun phrases, but to adpositional phrases as well (I focus on the latter). The grammatical relation which an adposition assigns to its "object" is open to cross-linguistic variation, with interesting correlates.

Many linguists are aware of the conspicuous evidence for (1) and (2) that comes from paradigm identity phenomena -- where, for instance, an adposition registers features of its "object" by means of an affix taken from a paradigm identical to a verb agreement paradigm. While the approach I favor does give a clear view of how those parallels arise, there are much more novel results to be had when we apply two additional ideas, both independently motivated:

(3) It is possible for multiple predicates to combine and form a single clause. This notion, explored by Nichols and Woodbury 1985, Davies and Rosen 1988, and others, has extensive applications in the study of auxiliation, serialization, and "complex predicates".

(4) Clausemate serial predicates need not all be verbs. The present author has argued, for instance, that a noun predicate may serialize with a verb, thus possibly introducing arguments of its own as direct dependents of the clause (hence the phenomenon of noun incorporation and the associated "possessor ascension", so called).

Likewise, as will be shown, an adpositional phrase can either be subordinated, or it can occur in serial format, i.e. with the adposition actually situated as a predicate of the clause, in which case the arguments of the adposition are direct dependents of the clause. I illustrate with data from Italian, English, and several languages of North America. In the English part, in particular, we will see why English "pseudo-passives" are compatible only with an unergative main verb (as in: the table was danced on by revelers) and not an unaccusative (as in: *the table was remained on by debris).

Nouns and Verbs in Language Contacts

Alexander Rusakov
St. Petersburg

It is known that nouns and verbs may behave in the different ways in the language contact situations. Maybe the most striking examples are Michif which combines French noun system with Cree verb system (Bakker) and Mednyj Aleut which combines Aleut noun system with Russian finite verb morphology (Menovshchikov, Golovko). The reason of such differences are not very clear. There have been attempts to explain them by the great similarity of the Russian and Aleut noun systems (the verb systems are rather different) for Mednyj Aleut (Thomasson) and by the "unanalyzable" character of the Cree verb stem for Michif (Bakker & Pappen). It is important, however, that in both cases noun and verb systems behave as two separate grammatical subsystems.

An other interesting case is the North Russian Romani Dialect (NRRD), strongly interfered dialect, spoken in the Northern part of Russia. For this dialect the phenomenon of Code Mixing is strictly characteristic. However, an interesting peculiarity is observed in this dialect. Russian nouns and adjectives are used practically always in adapted form (with some special morphemes, intruded between the stem and the flexion of a borrowed lexeme). On the contrary, the Russian verbs are used mostly in an unadapted forms and are conjugated according to the Russian model. One should mention that in the most part of Romani dialects the borrowed verbs are used in the adapted form, too (Boretzky, Bakker). We could consider such unadapted verbs in NRRD as EL Islands (according Myers- Scotton), but there isn't any difference between Russian nouns (and adjectives) and verbs from the functional point of view. It is worth mentioning that not only unadapted but adapted Russian elements also are "introspected" by NRRD native speakers as Russian inclusions, which do not change, however, the Romani character of the text.

A possible explanation of this phenomenon is the simpler nature of noun adaptation mechanisms, as well as the similarity of the Russian and Romani sets of cases (cf. Thomasson on Mednyj Aleut). The adaptation mechanism for verbs is more complicated due to the complex character of the Russian verb morphonology. The Romani verb paradigm also contains rather difficult morphonological rules. A further point can be made. Because of the high degree of interference we can nearly always establish a simple correlation between the Romani and the Russian grammatical forms. The absence of the infinitive in the Romani grammatical system constitutes an exception to this generalization which might provoke the intrusion of some unadapted Russian infinitive forms at an early stage (cf. Leksa Manush on Hungarian Romani dialect). Then the use of unadapted Russian infinitives might trigger the unadapted use of finite verb forms. It is interesting that there is an other contact phenomenon in NRRD verb system: the borrowing of the full system of Russian prefixes. Such a borrowing changed all temporal-aspectual system of NRRD. Maybe both phenomena (the usage of unadapted Russian verbs and borrowing of Russian prefixes) are connected. Possibly we may pose a question about the different character of nominal (more formal?) and verb categories realized in the contact situations.

On noun/adjective distinction in Russian.

M.Rusakova & S.Sai
St. Petersburg

1. It is traditionally assumed that in Russian adjectives and nouns are distinct parts of speech. The purpose of the present paper is to observe possible manifestations of this distinction in various domains.

2. Morphology.
A). Form. The majority of the Russian adjectives have sets of inflections differing from those of the nouns. However, the boundary between nouns and adjectives is not strict, since there are numerous nouns with "adjectival" inflectional paradigms; besides, there are some borrowed indeclinable adjectives (along with older indeclinable nouns).
B). Function. Adjectival inflections traditionally are thought to be semantically redundant (their main function being signalling agreement). However, psycholinguistic evidence is obtained showing that usually they have their semantic value, comparable to that of nouns' inflections.

3. Syntax.
A). Elements of the NP. It is traditionally assumed that nouns are obligatory elements of the NP, while adjectives are optional. However, spontaneous speech recordings abound in "NPs" consisting of attributes without nouns. Such "NPs" are usually treated as elliptic, that is, the nouns in these phrases are thought to be omitted according to syntactic patterns. But sometimes the omission of the noun is not determined by pure syntax and may be only explained semantically or pragmatically. Besides, there are situations when it is impossible to speak about "omission" at all; rather, adjective may have independent referential function the noun thus being communicatively redundant (sometimes in these cases noun slots are filled by syntactic "dummies" - nouns without clear lexical meaning).
B). Word order. The so-called "basic" word order in the Russian NP is Adjective-Noun. However, spontaneous speech data show that word order in the NP depends on pragmatic factor, namely, communicatively more important element (would it be noun or adjective) tends to come first. Besides, it is observed that the situations when adjective is in distant position and/or the preposition is repeated are not infrequent. It is argued that in these cases the unity of the NP is destroyed and thus nominals are in apposition both adjective and noun being a constituent of the clause.
C). Agreement. It is widely assumed that nouns are the controllers in the NPs. There is, however, psycholinguistic evidence undermining this assumption. Adjectives' grammatical setting may depend on external syntactic, semantic and pragmatic context and thus be independent from noun. Moreover, adjectives uttered before the noun tend to be the controllers of agreement.

4. Semantics.
It is well known that semantic distinction between adjectives and nouns is rather quantitative than qualitative in the languages of the world. Russian is very rich in the number of adjectives; adjectives are used to express a wide range of various meanings, not only prototypical adjectival ones.

5. Conclusion.
A number of aspects in which adjectives and nouns are thought to be distinguished in Russian were observed. A conclusion is made that this distinction is not strict in any of them. The Russian adjectives share many common features with the nouns. The nature of the adjectives includes the ability to function like nouns (but not vice versa).

Where did the words learn how to behave? Lexical categories in some endangered Austronesian languages of Taiwan

Jozsef Szakos
Providence University, Shalu, Taiwan

The purpose of this paper is to provide an overview of the parts-of-speech systems of three Austronesian languages in Taiwan, with a special focus to the noun/verb distinctions and to the problems of particles and clitics. The three languages under investigation include Tsou, Kanakanavu and Saarua, all belonging to the oldest, Formosan branches of Austronesian languages. Tsou still has over two thousand speakers, but Kanakanavu and Saarua are highly endangered, with only about a hundred speakers each, in the middle part of the central mountain range.

Previous studies were concerned with the genetic classification (Li, Dyan, Blust, Tsuchida) and syntactic phenomena (Starosta), as well as with general documentation of these languages (Japanese scholars, Szakos). Over ten years of intensive study of these languages (preparing dictionaries, textbooks) have led me back to the problem of word-classes. For a practical grammar we may as well apply the categories of Indo-European languages, but when using it for instruction we are confronted with the fact that they do not really fit in the classical sense. Even if we accept some prototypical senses of these categories, they lack descriptional adequacy.

Therefore, at first I intend to explain the paradigms of "verbs" (inflectional changes, mostly the focus system and modality), "nouns" (plurality, incorporation), "pronouns" in Tsou, Kanakanavu and Saarua, then compare their derivational techniques and discuss whether categories of adverbs, adjectives, numerals can be posited or not. Upon that, I plan to introduce types of nominalizations (phrase to noun, past tense to noun and other forms with enclitics) and discuss the appearance of new classes (measure words) which may be attributed to language contact with Chinese. Finally I attempt to re-define the categories in terms of the semantic feature clusters that they may include, as compared to those of the Indo-European languages.

Syntactic studies were able to elegantly avoid the problem of categorizing the sentence particles by simply calling them "construction markers" (CM), however these exhibit features of articles and deictic particles. They were also shown as independently "floating", while I try to contend that these are proclitics, pronounced together with the preceding word and not with the "noun" they are referring to. (V-Proclitic ...N-Procl. N-Procl. in a VOS language of Tsou and in the other two VSO languages).

Finally I try to draw on typologically related languages (although materials are rare) for possible descriptive categories of parts-of-speech. I am confident that the original materials to be presented will enrich the further discussion on these categories.

Verbal particles and word classes

Ida Toivonen
Stanford University

It has previously been claimed in the literature that verbal particles in Germanic (UP and OUT, for example) make up a word class of their own (see Noren 1996 and others). According to this view, there is a syntactic category PARTICLE which contrasts with nouns, verbs, etc. The arguments for this position include the following:

1) Particles often differ in meaning and function from homophonous prepositions. For example, ON as a particle differs in meaning from ON as a preposition. Compare `he carried ON dancing' to `he put it on the table'.

2) Particles differ in distribution from other word classes.

The present paper argues against the position that particles form a word class of their own, However, it also argues agains authors that claim that all particles can be viewed as intransitive prepositions (Emonds 1972, Svenonius 1994, Den Dikken 1995).

The arguments of this paper are based on data from Swedish. I will argue that particles do not form a separate word class. Instead, words of any word class can function as a particle. The difference between particles and other words is that particles do not project phrases the way other words do: they never project beyond the X level in X-bar theory and they must head-adjoin (in Swedish, they must adjoin to V). This accounts for the word order differences found between particles and other words. Another advantage of the present approach is that the similarities between particles and words of other categories is explained, since a particle can be a preposition, a verb, etc.

The claim will be formalized using the theoretical framework of Lexical-Functional Grammar. The relevant phrase-structural facts can be captured with the following rules:

VP -> (V) (NP) (NP) (XP)

V -> (V) (X)

The X can be P, A, N, or V.

In the approach advocated here, the phrase structure is kept very simple, and most of the explanatory force is in the lexicon. I will especially rely on the correspondence mapping between the a(rgument)-structure and the lexical conceptual structure (LCS). To account for the idiomatic and semi-idiomatic expressions that verb-particle combinations are involved in, I develop a model of LCS that is similar in spirit to the formalism developed in the Construction Grammar framework (see Goldberg 1995, Fillmore and Kay 1999, and references).

An Attempt at Characterizing the Semantics and Functions of (English) Adjectives

Beatrice Warren
Lund university

Adjectives are relational: they are invariably connected to nouns as modifiers or complements, carrying information about the referent of the noun. There are, however, different types of adjective-noun combinations, each of which-it will be argued- requires its own mode of semantic analysis.

The first type of adjective-noun combination represents a basic ENTITY-to-ENTITY relation. Examples include dirty shoes (>shoes having dirt on them), noisy children (>children producing noise), healthy lifestyle (>lifestyle conducive to health). As can be seen from these examples, ENTITY here represents concrete as well as abstract, bounded as well as unbounded phenomena.

The second type of adjective-noun combination represents a basic ENTITY-to-STATE/EVENT relation. There are two subtypes: one in which the adjective represents basically an ENTITY and the noun a STATE/EVENT (parental permission >parents permit, bodily harm >(something) harms/ed body). Such combinations qualify as nominalization. In the other subtype, it is the adjective that represents the STATE/EVENT and the noun that represents the ENTITY. Consider running water, continuous gunfire. These types of adjectives can be characterized as verbal or deverbal.

The third type of adjective-noun combination represents a Qualifier-of-STATE (utter darkness) or a Qualifier-of-STATE-of-ENTITY relation, e.g. old friend >person (ENTITY) whose friendship (STATE) is longlasting (Qualifier). Qualifier here is a functional rather than a semantic term. It can represent semantic notions such as DEGREE and MANNER. Adjectives in these types of combinations can be characterized as adverbial.

It will also be argued that premodifying adjectives (at least of the ENTITY-to-ENTITY type) serve to characterize, identify or classify their heads and that these functions fundamentally influence the syntactic behaviour of adjectives.

A number of empirical studies (Ljung (1970), Levi (1978), Aarts and Calbert (1979), Warren (1984) and Leitzky (1989)) have shown that there is a limited number of regularly occurring connections between the "entities" in ENTITY-to-ENTITY combinations. These will be discussed as will the intriguing finding by Warren that only certain types of connections are compatible with adjectives that have purely characterizing function.


Aarts, Jan and Joseph Calbert. 1979. Metaphor and non-metaphor. The semantics of adjective-noun combinations. Tübingen: Max Niemeyer Verlag.

Leitzky, Eva. 1989. (De)nominale Adjektive im heutigen Englisch. Tübingen: Max Niemeyer Verlag.

Levi, Judith. 1978. The syntax and semantics of complex nominals. New York, San Francisco, London: Academic Press.

Ljung, Magnus. 1970. English denominal adjectives. Gothenburg Studies in English 21.

Warren, Beatrice. 1984. Classifying adjectives. Gothenburg Studies in English 56.

Resultative Verb Complements in Mandarin Chinese: An Aspectual Perspective

Yuanzhong Zhang
University of Arizona

This paper explores the aspectual functions of Resultative Verb Complements (RVCs) in Mandarin Chinese by examining their syntactic and semantic properties. RVCs are a special class of participles associated with the predicate. They have the potential to form syntactically acceptable constructions with the predicate when the morphemes de (able) or bu (not) are inserted in between. They signify, among other things, the aspectual meaning of the event(s) described in a sentence. In contrast to other aspectual markers that have been largely grammaticalized and thus lost their original meanings in Chinese, most of the RVCs retain their inherent lexical meanings, or are lexicalized as part of a larger lexicon item such as an idiom.

RVCs can be classified into two broad categories: directional and non-directional. Non-directional RVCs consist of two sub-groups: result and phase RVCs (Smith, 1991). Directional RVCs profile the spatial and/or temporal changes involved in the progression of an event. Result RVCs are concerned with the state that emanates from an action. Phrase RVCs highlight the ultimate outcome of an action. Different kinds of RVCs share the common denominator of procedurality, a more fundamental construct that underlies both tangible temporal and/or spatial changes and abstract variations stretched over a continuum.

Such a classification scheme suggests that RVCs are able to mark both the perfective and the imperfective aspect in Chinese. The notion of perfectivity is redefined by drawing on Langacker's (1986, 1987) criterion that emphasizes the perceptibility of changes, be they temporal, spatial or conceptual, in the duration of the occurrence of an event. Seen in this light, the perfective aspect deals with situations wherein changes prevail, while the imperfective aspect relates to those devoid of perceptible changes. The aspectual distinction of RVCs is thus made by the perception of alternations from the original state or quality along the enactment of the action. When RVCs are used in verb constellations that signify the maintenance of the state or quality, they often indicate the imperfective. The perfective uses of RVCs can be realized not only through completive events, but also through durative events. The durative event type entails internal metamorphoses of state that makes up a dynamic process. A verb-copying test "V-le-you (again)-V" is proposed to examine the compatibility of verb constellations with the concept of duration. The completive event type purports to the change of state as a result of the engagement of the action. The aspectual functions of RVCs are established through investigating the conditions including {[-internal change], [-state/quality], [+completion]}, {[+internal change], [-state/quality], [+completion]}, {[+internal change], [-state/quality], [-completion]}, {[-internal change], [+state/quality], [-completion]}.

The current data sources show that the aspectual functions of RVCs bear upon, and would be overridden by the aspectual meaning of the predicate and its argument. This corresponds to the hierarchical relationship between the predicate and RVCs. In addition, this underscores Smith's (1990, 1991) assumption that RVCs may not necessarily affect the event type of the sentence. The aspectual relevance of RVCs to the aspectual implications of the predicate and of the sentence is, therefore, contingent upon the context in which the event occurs.


Langacker, R. W. 1986. "Abstract Motion", In V. Nikiforidou, M. Vanclay, M. Niepokuj, & D. Feder (eds.). Proceedings of the Twelfth Annual Meeting of the Berkeley Linguistics Society, 455-471. Berkeley: Berkeley Linguistics Society, University of California at Berkeley.

Langacker, R. W. 1987. Foundations of cognitive grammar (vol.1) Stanford: Stanford University Press.

Smith, C. S. 1990. "Event types in Mandarin," Linguistics, 28: 309-336.

Smith, C. S. 1991. The parameters of aspect. Dordrecht: Kluwer Academic Publishers.

Back to Parts of Speech front page
Back to SKY homepage