Case in a Topic-Prominent Language: Pragmatic and Syntactic Functions of Cases in Japanese
Abstract and Keywords
Discussions about topic-prominent language verus subject-prominent language are old but still of current interest. According to the now classic work of Li and Thompson (1976), topic-prominent languages possess the following characteristics: the topic is coded on the surface, that is morphologically and/or syntactically; passive constructions either do not or only marginally exist or carry a special meaning; there are no dummy or empty subjects; double subject constructions are available; it is not the subject but the topic that controls coreferential constituent deletion; verb-final languages tend to be topic-prominent; there are no constraints on what kind of constituent may be the topic; and topic-comment sentences are basic. The traditional grammar of Japanese distinguishes between two types of postpositional particles: kaku-joshi (case particles) on the one hand and kakari-joshi (relating/charging particles) on the other hand. Japanese reveals a relatively free word order, while maintaining a rigid verb-final position. This article sketches synchronic and diachronic case-drop phenomena, and exemplifies some other typologically related as well as unrelated languages for their functional parallelism.
54.1 Characteristics of topic-prominent languages
Discussions about topic-prominent vs. subject-prominent languages are old but still of current interest. According to the now classic work of Li and Thompson 1976, topic-prominent languages possess the following characteristics: (a) the topic is coded on the surface, that is morphologically and/or syntactically; (b) passive constructions either do not or only marginally exist or carry a special meaning; (c) there are no dummy or empty subjects; (d) double subject constructions are available; (e) it is not the subject but the topic that controls coreferential constituent deletion; (f) verb-final languages tend to be topic-prominent; (g) there are no (p. 780) constraints on what kind of constituent may be the topic; and (h) topic-comment sentences are basic.
Japanese actually fulfils these criteria more or less: (a) it has the topic marker ~wa; (b) there are adversative passives that are fundamentally not based on the active-passive diathesis and carry a special meaning (cf. Palmer 1994; Shibatani 1994); (c) meteorological or chronological expressions, for which subject-prominent languages typically employ a dummy subject, exhibit a referential subject or no subject at all (cf. Ogawa 2006); the characteristics (d), (e), and (f) fully apply, whereas (g) and (h) apply either to a certain extent or need more evidence and argument.
First, the topic-prominent character of Japanese will be discussed with special reference to its case system. I will focus on Li and Thompson's criteria (a), (d), and (g). After sketching synchronic and diachronic case-drop phenomena, some other typologically related as well as unrelated languages will be briefly exemplified for their functional parallelism.
54.2 Case system of Japanese
The traditional grammar of Japanese distinguishes between two types of postpositional particles: kaku-joshi (case particles) on the one hand and kakari-joshi (relating/charging particles) on the other hand. The latter historically needed a correlated verbal form (a sort of agreement) in the sentence final position (e.g. ~wa … + assertive verbal form), which is no longer true in Modern Japanese.
Japanese makes use of a number of postpositional particles in the case system such as: sensei~ga (teacher-nominative), ~wo (accusative), ~ni (dative), ~no (genitive), ~de (locative, instrumental), ~e (directional), ~kara (ablative), ~to (comitative). These ‘kaku-joshi’ occur in almost perfect complementarity with each other within a sentence and do not violate the ‘single case condition’, albeit in some cases (e.g. ~ni vs. ~de, ~ni vs. ~e) there are syntactic or semantic overlaps (cf. Yamanashi 1995).
While all of the above-mentioned postpositions mark cases, i.e. grammatical relations, the topic marker ~wa which belongs to ‘kakari-joshi’ assigns no case as such and roughly means ‘as for X’. Apart from marking a grammatical relation, the particle ~mo (in the sense of ‘also’, ‘even’) also plays an important role. However, it will not be discussed here, since the topic-prominent character of Japanese is sufficiently obvious from the behaviour of ~wa, especially compared with the ~ga case marking.
54.3 Pragmatic and syntactic functions of ~wa and related constructions
54.3.1 ~wa and ~ga
There are to some extent controversial opinions about the function(s) of ~wa. Although Iwasaki (1987) sees the core function of ~wa as ‘scope setter’ from which its topic function is merely derived, and although Kuroda 2005 even argues against ~wa as a topic marker (in a strict sense), it undoubtedly determines the topic-prominent character of Japanese.
The topicalizing function of ~wa can most clearly be shown in comparison with ~ga, because ~wa and ~ga can be considered as counterparts in many respects:
a. Kare-wa Yamada-san desu.
he-top Yamada-Mr./Ms. be
‘He is Mr. Yamada.’
b. Kare-ga Yamada-san desu.
he-nom Yamada-Mr./Ms. be
‘Mr. Yamada is he.’
The nominal marked with ~s the topic, the predicate nominal (plus the copula) the comment, as in (1a). In (1b), on the contrary, the predicate nominal is the topic, and the nominal followed by ~s the comment. Here terms such as topic and comment are used in reference to the information structure status of a constituent.
There have been ample discussions about the distinction between ~wa and ~ga for the information structure (cf. Chafe's 1971 differentiation between ‘oldinformation’ and ‘new information’, Kuno's 1973 survey on this theme, and a lot of works published in Japanese). In this sense ~ga functions not only as a nominative marker but also as an ‘anti-topic’ one.
It is predictable that ~wa can rarely be employed in subordinated sentences, for they essentially contradict the topic-comment structure. The subject marker ~ga, on the other hand, occurs also in embedded sentences.
(2) Kare-ga/*-wa kuru-node, …
‘Because he will come, …’
(p. 782) ~wa in its topicalizing function can lead to strongly ‘contrastive’ interpretations asin:
(3) Kare-wa eigo-wa hanas-eru.
he-top English-top speak-can
‘He speaks English.’
The sentence-initial referent ‘kare’ is already topicalized; the following one ‘eigo’, also marked with ~wa, is furthermore topicalized, being so to speak ‘picked up’ from other related or evocable ones. In (3) there hence occurs a strongly contrastive effect: ‘He can speak no other foreign language but English.’
~ga can function as an ‘exclusion’ such as:
(4) Kono shigoto-wa watashi-ga yarimasu.
this task-top I-nom do-dis
‘I will do this task.’
This sentence can be paraphrased like this: ‘Nobody else but me will do this task’. So persons other than ‘me’ are excluded. ‘Watashi’, the speaker, being attributed high topicality per se, essentially becomes a ‘comment’ through ~ga, resulting in a kind of mirror effect.
It seems that ‘contrast’ by ~wa on the one hand and ‘exclusion’ by ~ga on the other hand lead to similar – albeit contradictory – interpretations. (Such difficulties in differentiation between ~wa and ~ga motivate Kuroda 2005 not to regard ~wa as a topic marker.) The former, however, is based on the process in which one stresses the topic, i.e. old (already-provided) information or information that the speaker supposes the hearer to know (cf. Onoe 1995). The latter results from the process in which one emphasizes the comment to a still higher degree, that is, the speaker presents to the hearer some information as entirely new, unknown at all, even if the item marked by ~ga apparently is not, such as ‘watashi’ in (4). Both processes are commonly ‘picking up’, but from different perspectives, namely foregrounding of the topic vs. of the comment.
54.3.2 Word order
Japanese reveals a relatively free word order, while maintaining a rigid verb-final position. There are, however, some canonical word order patterns, to which the ‘nominative-dative-accusative’ belongs:
(5) Sensei-ga seito-ni hon-wo ageta.
teacher-nom pupil-dat book-acc give-past
‘The teacher gave the pupil a book.’
As for topicalization, two strategies are available: i) the use of the topic marker ~wa, as we have already seen, and ii) the left-dislocation of the nominal with its (p. 783) case marking unchanged. The former (frequently together with the latter, unless the nominal in question is the subject) leads to topicalization of the nominal:
(6) Sono-hon-wa sensei-ga seito-ni ageta.
this-book-top teacher-nom pupil-dat give-past
‘As for this book, the teacher gave it to the pupil.’
The latter lets the nominal in question become a comment, hence the rest is a topic in the information structure:
(7) Sono-hon-wosensei-ga/-wa seito-ni ageta.
this-book-acc teacher-nom/-top pupil-dat give-past
‘It is this book that the teacher gave to the pupil.’
This kind of left-dislocation, that is, not the topicalization through ~wa, often results in an ‘exclusive’ interpretation, as we have seen in the case of ~ga in (4). While ~wa marks no grammatical relation as such, ~ga is a nominative/subject marker, so it cannot be used for the accusative nominal in (7).
As in (6), the topicalization through ~wa can be applied not only to genuine subject nominals, but also to object nominals:
a. Taro-wa Hanako-wo aishiteiru. (originally nominative -ga)
Taro-top Hanako-acc love
‘Taro loves Hanako.’
b. Hanako-wa Taro-ga aishiteiru. (originally accusative -wo)
Hanako-top Taro-nom love
‘It is Taro who loves Hanako.’
c. Hanako-wa Taro-ga denwashita. (originally dative -ni)
Hanako-top Taro-nom phone-past
‘It is Taro who phoned Hanako.’
On the contrary, oblique nominals can hardly ever be topicalized by means of ~wa, such as:
d. *Hanako-wa Taro-ga Tokyo-e itta. (originally comitative -to)
Hanako-top Taro-nom Tokyo-dir go-past
‘It is Taro who went to Tokyo with Hanako.’
e. *Naifu-wa Taro-ga pan-wo kitta.
knife-top Taro-nom bread-acc cut-past
‘With the knife Taro cut the bread.’
(originally locative-instrumental -de)
But as long as the original case marker remains, a more argument-like (less oblique) nominal can be topicalized by attaching ~wa (cf. [8d]: ‘Hanako-to-wa’ is well-formed), a less argument-like (more oblique) one even then can not (cf. [8e]: ‘Naifu-de-wa’ is less possible).
(p. 784) With regard to the following statement: ‘in a topic-prominent language […] there are no constraints on what may be the topic’ (Li and Thompson 1976: 471), Japanese actually might be characterized as a strongly, but not fully topic-prominent language. In this respect the data under (8) suggest a parallelism to the Noun Phrase Accessibility Hierarchy (Keenan and Comrie 1977), according to which relativization follows the hierarchy within the case system; relativization is one of the strategies for topicalization.
54.3.3 Predication and double subject constructions
In the case of ~wa an extremely broad range of predicative relations can be attested. Some of them are:
a. Kare-wa Yamada-san desu.
he-top Yamada-Mr./Ms. be
‘He is Mr. Yamada.’
b. Kare-wa byouki desu.
he-top ill be
‘He is ill.’
c. Kare-wa koohii desu.
he-top coffee be
‘He ordered coffee.’
d. Kare-wa onnanoko desu.
he-top girl be
‘He has a daughter.’
Subject-prominent languages, on the contrary, allow only a few of these uses – (9a) (‘identification’) and (9b) (‘property’) – provided in copular constructions. Though in English, for example, one can say ‘You are a coffee, right?’ as spoken by a waitress in a restaurant. But the range of possibilities of such expressions is much smaller than in Japanese.
Ikegami 1981 characterizes the function of this copular (or copular-like) construction in Japanese as ‘A WITH B’. This can be interpreted as: A is in some relation with B. If, for example, ‘he’ orders ‘coffee’ in a restaurant, ‘he’ is in some relation with ‘coffee’. The sentence (9c) hence allows not only the interpretation glossed above, but also interpretations like: ‘He dislikes coffee’, ‘He plants coffee’, ‘He sells coffees’, etc. Which one is suitable can only be verified pragmatically.
To add: The copular-like construction can employ ~ga too, instead of ~wa. This results in a mirror relation in terms of the topic-comment structure, as seen in (7).
The function of ~wa can also be compared to ‘opening the file’ concerning the category about which an assertion should be made. The operation of ‘opening the file’ can typically be applied to so-called double subject constructions, such as:
(p. 785) (10) Zou-wa hana-ga nagai.
elephant-top trunk-nom long
‘An elephant has a long trunk.’
(11) Nihon-wa rokugatsu-ga ichban ame-ga ooi.
Japan-top June-nom most rain-nom much
‘In Japan it rains most in June.’
In (10) the file of the category ‘elephant’ is opened first. It evokes a number of ‘facets’ concerning the ‘elephant’; one of them is ‘the trunk is long’ (cf. Sawada 2002). (11) is a ‘triple’ subject construction, which is coherent with the idea of opening the file for multiple applications. The characterization of ‘staging’ through ~wa (Maynard 1980) also supports this idea. And for the process ‘opening the file’ or ‘staging’, the ‘aboutness condition’ (Shibatani 1994) is relevant.
Some double subject constructions semantically overlap with genitive constructions:
a. Zou-wa hana-ga nagai.
elephant-top trunk-nom long
‘An elephant has a long trunk.’
b. Zou-no hana-ga/-wa nagai.
elephant-gen trunk-nom/-top long
‘The trunk of an elephant is long.’
These two construction types, however, fundamentally differ. (12a) is an assertion about an elephant, whereas (12b) is an assertion about an elephant's trunk. This does not seem to be a case of possessor raising, but a kind of expansion of valency of a predicate (cf. Shibatani 1994, Ogawa 1997). Other double subject constructions, quite typical ones, cannot be paraphrased with genitive constructions:
a. Watashi-wa atama-ga itai.
I-top head-nom hurt
‘I have a headache.’
b. *Watashi-no atama-ga itai.
I-gen head-nom hurt
‘My head hurts.’
While genitive constructions must be adjacent, double subject constructions need not (cf. Heycock 1993):
a. Zou-wa umaretsuki hana-ga nagai.
elephant-top from.the.birth trunk-nom long
‘An elephant has a long trunk from birth.’
b. *Zou-no umaretsuki hana-ga/-wa nagai.
elephant-gen from.the.birth trunk-nom/-top long
‘The trunk of an elephant is long from birth.’
(p. 786) The possibility of discontinuity indicates the double subject as being a sentential constituent.
54.3.4 ‘Case drop’ synchronic and diachronic
The case marker can drop especially in colloquial speech, as long as its grammatical relation is reconstructable:
(15) Aitsu, mou ronbun kaita?
he-∅ already article-∅ write.past
‘Has he already written his article?’
Here the nominative argument aitsu as well as the accusative one ronbun respectively lack their case markers.
In general, case markers that are highly argument-like, i.e. nominative and accusative, can easily drop. Oblique case markers, however, can drop too, provided the argument in question is – mostly idiomatically – incorporated into the verb:
a. Sonoatode gakko itta-yo.
after.that school-∅ go.past-dis
‘After that I went to school.’
b. Mou ie kaeroo-ka?
yet house-∅ go.back-dis
‘Shall we go home?’
a. Watashi-wa samishii-no.
‘I am sad.’
b. Watashi-ga samishii-no.
‘It is I who is sad.’
c. Watashi samishii-no.
‘I am sad.’
Watashi-wa in sentence (17a) is strongly topicalized. The dichotomy of the topic-comment structure is in full use. The counterpart lies in (17b), where watashi-ga is the comment. The sentence (17c) without overt case marking can be located between them. The topic-comment structure is neutralized, watashi can be seen as ‘quasi-topic’. A ‘deictic/stage-dependent’ (Onoe 1995) argument, watashi representatively, possesses such high topicality that it tends to occur without ~wa. The (p. 787) sentence (17c) hence is an unmarked utterance, while (17a) and (17b) are topic-comment structurally marked. We rather should speak not of the dropping of ~wa but of its non-existence.
Historically ~wa triggered a sentence-final particle, which is known as kakari-musubi (‘charging-closing’). Recall:~wa belongs to ‘kakari-joshi’ (‘charging/relating particles’) in the traditional grammar of Japanese. The ‘kakari-musubi’ is again confirmation of the theoretical linguistic claim that the main function of ~wa lies in scope setting (Iwasaki 1987): the domain must have been set up between both the co-occurring particles. In the era of ‘kakari-musubi’ the kakari (‘charging’) particle ~wa assigned the beginning of the scope, and a corresponding musubi (‘closing’) particle overtly marked its end. Besides ‘kakari-musubi’ there existed (and – as we have seen in this section – partly still exists) zero case marking, especially in subject as well as object position, whereas the topic function attributed to ~wa in Modern Japanese was broadly borne by zero marking (Kinsui 1995). The ‘kakari-musubi’ itself progressively vanished in the Kamakura era (twelfth-fourteenth century).
This correlated with ~ga obtaining the status of subject in the Kamakura era and the following Muromachi era (fourteenth-sixteenth century). ~ga had persisted till the Heian era (eleventh century) as the genitive marker. A remnant of this can still be observed in its overlapping use with ~no (Taro-ga/-no katta hon: Taro-nom/-gen buy-past book).
54.4 Parallel data in some other languages
There are a number of grammatical strategies that distinguish topic and comment. Some languages employ their topic marker quite often, such as Japanese and slightly differently Korean or Burmese:
Ku saram-un/-i Park imnida.
this person-top/-nom Park be
‘He is Mr. Park.’
e:dilou pyo:da ha mәkaun:bu:no.
such a thing to.say-top bad-dis
‘It is bad to say such a thing.’ (Kobayashi 1984: 93)
(p. 788) For functional equivalence other languages use their article system and/or word order variations and/or different sentence types. For example:
a. Der Mann heiratet die Frau.
the man-nom marry the woman-acc
‘The man marries the woman.’
b. Die Frau heiratet der Mann.
the woman-acc marry the man-nom
‘It is the man who marries the woman.’
La tortilla la comió Pedro.
The omelette-acc/-nom it-acc eat-past Pedro
‘As for the omelette, Pedro ate it.’ (Noda 1996: 294)
a. Lao-Wáng zuótiān xiū le zhe liang zìxíngchē.
Mr. Wang yesterday fix prf this bicycle
‘Mr. Wang fixed this bicycle yesterday.’
b. Zhe liang zìxíngchē shi Lao-Wáng zuótiān xiū de.
This bicycle it Mr. Wang yesterday fix dis
‘As for this bicycle Mr. Wang fixed it yesterday.’
Example (20) uses the word order strategy with sentence-initial position; (21) shows left-dislocation, followed by an anaphoric pronoun. (22) employs the dislocation in a different sentence type; in (22) not a particle such as the Japanese ~wa but an anaphoric element (shi) is inserted, and corresponds to the sentence-final marker de. This recalls that in Old Japanese ~wa also required such a correlated verbal agreement in the sentence-final position (see section 54.3.4). An anaphoric element occurs in the Spanish example (21) too. In addition to these syntactic strategies for topicalization, prosodic ones play a role in all these examples.
To what extent a topic construction is more or less marked/unmarked than its counterpart or related construction can be clarified only relatively. As for Japanese, however, the topic marker alternates with the argument-like case markers (nominative, accusative, dative) and can also co-occur with the rest (oblique case markers). The bare left-dislocation of grammatically case-marked nominals also results in a topic-comment structure. Both undoubtedly indicate an unmarked base for a topic-prominent language.