=============================================================

1. Title: More on the Implications of Framing Typology: A View from East Asia
Author: Toshio Ohori (University of Tokyo)

------------------------------------------------------------------------------

The verb-framed (=VF) vs. satellite-framed (=SF) typology of Talmy (1985, 1991, 2000) has attracted a lot of attention in cognitive linguistics and language typology. Among its possible correlates, the present study focuses on two issues, drawing data from narrative discourse.

First, as noticed by Slobin (e.g. 1996), VF languages tend to express fewer motion events with explicit mention of ground elements such as goal compared with SF languages. Below are examples from Japanese (a VF language) and Chinese (a SF language, though problematic in some respects).

(1)	Mata shibaraku aruite iku-to	panyasan mitaina tokoro-ga atte
	then awhile walk.PRC go-CONJ	bakery like place-NOM be.PRC
	'then, (he) walked a while and there was a place like bakery'

(2)	Binke jiu yizhi zou-zou	dao yi-ge mianbao-dian
	Pingu then straight walk-walk	reach:GOAL one-CLF bread-shop
	'Pingu then walked his way to the bakery'

Notice that the goal of motion ('bakery') is not construed as such in (1), and it is rather expressed as an autonomous entity, while MIANBAO-DIAN is clearly goal in (2). Interestingly, typologically similar languages such as Chinese, Thai, and Vietnamese show different degrees of preference for the expression of goal, despite their comparable encoding strategies. This fact opens a possibility of micro-typology by arranging these languages on the scale of how far the grammaticalization of satellites is advanced.

Second, when we turn to the encoding of goal/result in the semantics of verbs, there are differences which cross-cut the VF-SF typology. For instance, the translation equivalent of the Chinese example (3) is unacceptable in English, which is a typical SF language.

(3)	Wo shao le fangzi,	dansi ta mei-you shao zhao.
	I burn(Vt) ASP house	but it NEG burn(Vi) ASP
	'I burned the house, but it didn't burn.'

Further, a translation equivalent of (3) is acceptable in Japanese but it is highly marginal in Thai. This puzzle can be solved if we adopt the view that languages differ with respect to the lexical implementation of goal/result in verb semantics. That is, the Vendler system of Aktionsart may itself be universal, but verbs with seemingly equal semantic substance can have different profiles across languages. Verbs which denote intentional goal-directed activities but do not have goal/result as part of their lexical meanings are here categorized as "pre-accomplishment" class.

The VF-SF typology combines with the above fact in the following way. Strategies for encoding paths of motion can be extended to the marking of aspect. In languages like English with more accomplishment verbs, satellites are used to re-frame activity verbs as accomplishments (e.g. BLOW > BLOW OUT). In Chinese, where pre-accomplishment verbs dominate, satellites (i.e. semi-grammaticalized resultative complements) are used to overtly encode the goal/result. In Japanese, pre-accomplishment verbs also dominate, but reflecting its verb-framed character, the encoding of goal/result requires complex predication rather than the addition of less autonomous morphemes. But in real discourse, this niche of grammar is not widely exploited for this purpose and in many cases the accomplishment of goal/result is relegated to pragmatic inference, i.e. conventional implicature.

Finally, the question arises whether there is any significant connection between preferences about expressing goals in discourse and the lexicalization of goal/result in the semantics of verbs. A tentative conclusion is that there is no inherent connection, but the two characteristics sometimes conspire to make some languages look strikingly opposed to each other.

[ASP=aspect, CLF=classifier, CONJ=conjunction, NEG=negative, NOM=nominative, PRC=participle]

=============================================================

2. Title: Resultative constructions with verbs of "cleaning" in Thai and Japanese: a comparative study
Authors: Kingkarn Thepkanjana (Chulalongkorn University) and Satoshi Uehara (Tohoku University)

------------------------------------------------------------------------------

The term "resultative construction" is generally used to refer to a syntactic construction consisting of two predicates representing a causing event and a resulting state. There are many types of construction which may fit this definition, such as compound verbs, and two concatenated predicates with or without any intervening linker. The type of resultative construction which is the object of study in this paper is manifested in the form of two concatenated predicates, one of which is a transitive verb. The particular transitive-based resultative construction to be examined in this study is made up of a transitive verb with the meanings "wash", "wipe", and "sweep", all of which containing the semantic element "cleaning something". Such transitive verbs represent the causing predicate in the resultative construction. Such "cleaning" verbs are characterized by the fact that an agent must have in mind a certain state of the patient, which represents his intended goal in performing an action. In addition to the causing predicate, the resultative construction contains a direct object argument indicating an entity affected by the causing action, and another predicate, which expresses the resulting state of the causing action and which is realized as a stative verb in Thai and as an adverb in Japanese. Some examples of the transitive-based resultative construction with verbs of cleaning in Thai and Japanese are given below.

Thai :	khaw2	laaN4	kE'Ew3	sa?2aat2
	he	wash	glass	clean
	'He washed the glasses clean."
Japanese :	kare wa	gurasu o	kirei-ni	arat-ta
	he TOP	glass ACC	clean-ADV	wash-PAST
	"He washed the glasses clean."

This study aims to make a comparative investigation of the transitive-based resultative construction with verbs of cleaning in Thai and Japanese by focusing on two syntactic-semantic aspects, namely, (a) collocational constraints between the causing predicate and the resulting predicate in the two languages, and (b) constraints in negating different portions of the resultative construction in the two languages. The similarities and differences discovered in the two languages are accounted for within the cognitive-functional linguistic framework. It is found that the "cleaning" causing predicate in Thai can be collocated with a resulting predicate indicating an anticipated state such as the one meaning "clean", a resulting predicate indicating an unanticipated state such as the one meaning "dirty", or even a resulting predicate indicating a state located outside the conceptual axis leading to the intended goal such as the one meaning "broken" in the case of the causing predicate meaning "wash glasses." In contrast, the "cleaning" causing predicate in Japanese must be collocated with a resulting predicate indicating an anticipated goal state only. That means the resulting predicate in the "cleaning" resultative construction in Thai can serve to confirm the fulfillment/anti-fulfillment of the causing action, or even serve to indicate an unexpected "other-event result" of the causing action (Talmy 1991, 2000), whereas the resulting predicate in Japanese must confirm the fulfillment of the "cleaning" causing action only. It is argued in this paper that the different word orders in the two languages, which in turn reflect different conceptualization systems, bear on such collocational constraints.

The second syntactic-semantic aspect of the resultative construction to be investigated are negation constraints. It is found that the Thai resultative construction allows for more possibilities of negation than its Japanese counterpart. In other words, more portions in the former can be negated than in the latter. The various negation possibilities in Thai are accounted for by appealing to the notion of "discourse presupposition" postulated by Givon (1984).

=============================================================

3. Title: "Functional transfer": a case study of the "potential" constructions in Japanese and Thai
Author: Shoichi Iwasaki (UCLA)

------------------------------------------------------------------------------

"Functional transfer" refers to the process by which a certain function expressed in one grammatical structure transfers to another in order either to preserve the old function when the earlier structure is losing its place in grammar, or to expand or highlight the existing function to increase its efficiency and expressiveness. I will examine the second type of functional transfer in this paper, by focusing on the "potential" constructions in Japanese and Thai.

Thai has a general potential construction with the modal auxiliaries, "day" and "pen," as shown in (1) below. It should be noted, however, that their placement is peculiar in light of the fact that Thai is a relatively straightforward head initial SVO language, and always places modal auxiliaries before the main verb as in (2), with the exception of the potential modal auxiliaries.

(1)	kin may day/pen
	eat NEG can
	"(He) can't eat it."

(2)	khuan ca kin
	should IRR eat
	"(He) should eat it."

The peculiar placement of the potential auxiliary can be explained easily if we consider that the source of this construction is the resultative construction, such as (3) below.

(3)	kin may long
	eat NEG go.down
	"(I) can't eat (it)"

In (3) the action of eating is broken down into two phases, the initial eating (e.g. chewing) phase represented by "kin" (eat) and the later swallowing phase represented by "long" (go.down). Thus, (3) means "I (try to) eat but (the food) won't go down," and is readily interpretable as "I can't eat." The resultative verb "long" is specific, and the range of the possible initial verbs is restricted. On the other hand, the verbs "day/pen" in (1) are general and the range of verbs appearing with them is much larger.

I argue that the construction represented in (1) is a result of functional transfer motivated by the need for a more general means of expressing potential meaning. With a more general potential construction, not only verbs, but also complements can appear, as shown in square brackets in (4) below, and consequently expressiveness is increased.

(4)	[ kin tua duay takiap ]	may day/pen
	eat beans with chopsticks	NEG can.do
	"(I) can't eat beans with chopsticks."

Japanese has the "potential" suffix, -(ra)re- or -e- (see (5).) Despite the existence of the morphological potential, the language has created a newer periphrastic potential construction, consisting of the potential agent NP marked by "ga/wa", a complement with the nominalizer, "koto," marked by "ga," and the general potential verb "dekiru" (see (6).)

(5)	boku-wa	namazakana-ga/o	tabe-rare-ru
	I-TOP	raw.fish-NOM/ACC	eat-POT-NONPAST
	"I can eat raw fish."

(6)	boku-wa	[ namazakana-o taberu koto]-ga	dekiru
	I-TOP	raw.fish-ACC eat NOMLZER - NOM	can/do:NONPAST
	"I can eat raw fish."

This move can be also understood as a functional transfer, through which a potential meaning is highlighted, and the expressiveness increased.

The functional transfer in Japanese is motivated by the morphological and semantic instability of a sentence like (5). Morphologically, the case marking for the patient noun phrase fluctuates between "ga" and "o," and some potential forms are semantically ambiguous, with both potential and passive interpretations. The new peripheral construction completely eradicates such confusion. In addition, the new structure can easily accommodate any proposition within the complement.

=============================================================

4. Title: Verbal nouns in Japanese and Korean: cognitive typological implications
Author: Kaoru Horie (Tohoku University)

------------------------------------------------------------------------------

This paper probes into differing cognitive orientations underlying the manner of assimilating loanwords into the lexico-grammatical systems of Japanese and Korean.

Japanese and Korean have borrowed numerous lexical items from Chinese since the ancient times and assimilated them into their native lexicons. Of particular note is the fact that both languages have consequently obtained numerous Sino-origin nouns such as ‘syuppatu’ “departure” in Japanese and its Korean counterpart ‘chulpal’ in Japanese and Korean (1a, 2a) that can be converted into verbs by attaching verbalizing suffixes, i.e. ‘suru’ in Japanese and ‘hata’ in Korean, as shown in (1b) and (2b):

(1) (Japanese)	(a) Ressya-no syuppatu	(b) Ressya-ga syuppatu-sita.
(2) (Korean)	(a) Yelcha-uy chulpal	(b) Yelcha-ka chulpal-hayssta.
	train-GEN departure	train-NOM departure-do:PAST
	“the departure of a train”	“A train departed (lit. A train did departure).”

These nouns are called “verbal nouns” (Martin 1975) since they have both nominal and verbal properties. Though predominantly of Chinese origin, verbal nouns in Japanese and Korean include non-Sino loanwords, particularly those from English, as shown in (3):

(3) (English) copy => (Japanese) kopii-suru; (Korean) khapi-hata

The morpho-syntactic realization of verbal nouns in Japanese and Korean is apparently very similar, as shown in (1)-(3). However, verbal nouns in the two languages contrast sharply when they co-occur with argument-marking case particles such as accusative case particles, as shown in (4) and (5):

(4) (Japanese)
Sono ressya-wa,	tookyoo-eki-o syuppatu-go,	sugu bakuhatusita.
That train-TOP	Tokyo-station-ACC departure-after	soon exploded

(5)(Korean)
Ku yelcha-nun,	tohkyo-yek-ul chulpal-*(han)-hu,	kot phokpalhayssta.
That train-TOP	Tokyo-station-ACC departure-(do:PAST)-after	soon exploded
“That train exploded soon after it left Tokyo station.”

The phrase ‘syuppatu-go’ in Japanese (4), consisting of a Sino-Japanese verbal noun ‘syuppatu’ and a Sino-Japanese nominal suffix ‘go’ “post-”appears to be a full-fledged noun phrase like ‘sen-go’ “post-war”. Nevertheless, it actually behaves like a past tensed verb taking a direct object NP ‘tookyoo-eki’ (“after (it) left Tokyo station”). In contrast, the Korean counterpart phrase ‘chulpal-hu’ (5), consisting of a Sino-Korean verbal noun chulpal’ and a Sino-Korean nominal suffix ‘hu’, cannot co-occur with a direct object NP ‘tookyoo-yek-ul’ without the overt indication of verbhood ‘han’ , i.e. the adnominal past tense form of the verb ‘hata’.

This paper argues that the existence in Japanese of the surface structure ‘tookyoo-eki-o syuppatu-go’, wherein the verbal property is concealed under the appearance of a noun phrase, and the absence of its counterpart structure in Korean, are not accidental. Rather, this contrastive linguistic difference is motivated by a differing cognitive predisposition exhibited by each language toward borrowing from languages of different typological profiles such as Chinese (an isolating language). Japanese, an agglutinating language, has innovated a hybrid structure (4) in which a Sino-Japanese verbal noun phrase is allowed to take a clausal argument without any agglutinative morphological adjustment. In contrast, Korean, another agglutinating language, has avoided such a hybrid structure by enforcing agglutinative morphological adjustment on a Sino-Korean verbal noun when it co-occurs with a clausal argument (5). The differing predispositions toward borrowing are consistent with the contrastive tendencies manifested between the two languages toward surface structural ambiguities (Horie 1998), i.e. Japanese is more tolerant of one-to-many (or many-to-one) form-meaning correspondences and concomitant surface structural ambiguities than Korean, which tends more strictly to preserve one-to-one form-meaning mapping.

=============================================================

5. Title: Language typology and grammaticalization: Tense/Aspect marking in the languages of East and mainland Southeast Asia between the lexicon, pragmatics and grammar
Author: Walter Bisang (University of Mainz, Germany)

------------------------------------------------------------------------------

The languages of East and mainland Southeast Asia share the following three typological properties which seem to determine the way in which grammaticalization takes place in this area:

1. The smallest meaningful element is the syllable (Ke 1998, Bisang forth.).

2. Lack of obligatory grammatical categories (indeterminateness, Bisang 1998, also cf. "transparadigmatic variability" in Lehmann 1995)

3. Relatively weak determination of word class in the lexicon

On 1: Due to the first property, indicators of a high degree of grammaticalization such as erosion, fusion, loss, or affixation (cf. Heine & Reh 1984, Lehmann 1995) are relatively marginal in East and mainland Southeast Asian languages.

On 2: Obligatority defined in terms of Bybee (1997: 34) means that the absence of a category X entails the inference of non-X. There is almost no marker of any grammatical category triggering such an inference. For that reason, most tests for aspect (e.g. perfective vs. imperfective) systematically fail in the languages to be discussed, even if their semantics are quite comparable.

In paths of grammaticalization such as the one of future (Bybee et al. 1994: 256) or the one of perfective/simple past (Bybee et al. 1994: 105) we often find the same lexeme from the very beginning to the very end of the continuum. As far as mainland Southeast Asian languages are concerned, we even find a new lexeme for marking perfective/simple past, the verb "get, attain", which is not discussed by Bybee et al. (1994). From the multifunctionality in the above sense, which is called "split" by Heine & Reh (1984) or "divergence" by Hopper & Traugott (1993), I conclude that semantic bleaching or processes such as metaphor in terms of Heine et al. (1991), understood as a development from concrete to abstract concepts, are only of secondary relevance for grammaticalization in East and mainland Southeast Asian languages. Thus, I shall try to develop a pragmatic scenario based on Hopper & Traugott (1993) which incorporates constructions in terms of Construction Grammar. Syntagmatically, reanalysis is applied to constructions and triggers metonymic inferences by matching a certain position (within that construction) with a certain function. Paradigmatically, metaphoric inferences enlarge the set of lexemes which can cooccur with a grammaticalised element (cf. Bisang 1998, 2000).

On 3: This last point belongs to my current research on transcategoriality in Old Chinese. Moreover, it is based on the observation that even in a language like Khmer with a rather developed morphology, morphology is not (or only marginally) sensitive to syntax in the sense that it would determine the compatibility of a lexical item with a certain position within a construction or with a certain terminal node within a tree structure. If it turns out that lexical items are not that rigidly marked for parts of speech in the lexicon as it is assumed for languages such as English, French, or German, this may be another factor which facilitates the reanalysis of lexical items in positions associated with grammatical functions.