• No results found

CHAPTER 4. TOPICALIZATION

4.2. Previous Analyses in GB Framework

4.2.2. Topicalization in Korean

4.2.2.2. Base-Generation Analysis

Y.-S. Kang (1986) argues that Korean topics must be base-generated in the left­ most topic node, since they allow for resumptive pronouns and violate CNPC and

Subjacency. If the topic phrase involves movement, as argued in (17)-(19), how can we explain the presence of casin ‘himself in (20b) ? Under the movement theory, the empty position in (20a) should be filled by a trace left by ku-nun ‘he-Top’ which has moved to the sentence-initial position. However, the empty position can be filled by a lexical item casin ‘himself,’ as shown in (20b) (from D.-W. Yang (1973)).

(20) a. [cp ku-nun [ip [ e] ku il-ul hay-ss-ta.]] he-Top the job-Ace do-Pst-Dec ‘He did the job.’

b. [c p ku-nun [ip casin-i ku il-ul hay-ss-ta.]]

he-Top himself-Nom the job-Ace do-Pst-Dec ‘He did the job himself.’

The above examples in (20) support the claim that the topic construction should be base-generated. If ku-nun ‘he-Top’ in (20) is moved from the subject position in IP, casin-i ‘himself-Nom’ cannot be inserted since the subject position is filled by the trace left by the moved ku-nun ‘he-Top’. The same analysis can be applied to (21).

(21) Ku il,-un [ Mary-ka taumtaley (ku kes,-ul) halye] ko sayngkakhakoiss-ta. thejob-Top Nom next month thejob-Acc do Comp think-Dec

‘As for the job, Mary is thinking of doing it next month.’

The topic phrase in (21) allows the occurrence of a resumptive pronoun. This imphes that the topic phrase is not moved from the gap position but is base-generated in the current position. The example in (21) implies that ku il ‘the job’ does not move from the position filled by ku kes ‘if to the surface position. Now let us consider the

example (22) where the resumptive pronoun is not allowed, unlike (21). I reproduce the following examples in (22) from D.-W. Yang (1973).

(22) a. Seoul/-ey [ Mary-ka taumtaley t, kalye] ko sayngkakhakoiss-ta.

Log Nom next month go Comp think-Dec ‘It is Seoul that Mary is thinking of going to (there) next month.’

b. * Seouf-ey [ Mary-ka taumtaley keki,-ey kalye] ko sayngkakhakoiss-ta.

Log Nom next month there-Loc go Comp think-Dec

‘It is Seoul that Mary is thinking of going to there next month.’

Crucially, the sentence-initial phrase in (22) is not marked with nun. That is, the example in (22b) does not involve topicalization but just scrambling of Seoul-ey ‘to Seoul’ (or movement). A resumptive pronoun is not allowed in the gap of the scrambled element, since scrambling involves movement. Therefore the resumptive pronoun in the gap position is not permitted since the gap position is already filled by a trace. Now we can account for the contrast between (21) and (22) by topicalization involving non-movement and scrambling involving movement, respectively.

The topic construction in (23b) is not subject to island constraints but the scrambling construction in (23a) is. I reproduce the following examples from G.-S. Moon (1987). The grammaticality judgement also is G.-S. Moon’s.

(23) a. * ce moca-lul John-i [rc t, ssuko-i-ss-ten] salam-ul cal alkoiss-ta. that hat-Acc Nom wear-be-Pst-AM person-Acc well know-Dec ‘That hat, John knows the person who was wearing it well.’

b. ce moca-nun John-i [rc e, ssuko-i-ss-ten] salam-ul cal alkoiss-ta.

that hat-Top Nom wear-be-Pst-AM person-Acc well know-Dec ‘As for that hat, John knows the person who was wearing it well.’

The scrambled phrase ce moca-lul ‘that hat’ in (23a) is moved out of the relative clause, resulting in a CNPC violation. In contrast, the topicalized ce moca-nun ‘that hat’ in (23b) is base-generated in the surface position. If the topic were moved out of the relative clause, it would violate the CNPC. Unlike (23a), (23b) is grammatical. This is another piece of evidence for the base-generation of the topic construction in Korean.

Kuno (1972) also points out that there is a problem with the movement approach, by presenting the topical sentences like (24a,b). It is not possible to argue that kkot

‘flower’ or mullihak ‘physics’ involves movement.

(24) a. Kkot-un [ cangmi-ka ceyil-i-ta.] Flower-Top rose-Nom first-be-Dec ‘Speaking of flowers, roses are the best.’ b. Mullihak-un [chwicik-i elyep-ta.]

physics-Top employment-Nom difficult-Dec ‘Speaking of physics, finding a job is difficult.’

Given this problem with the movement approach to topicalization, a natural alternative approach is to argue for the base-generation of the topic element in the surface (sentence-initial) position. The topic construction in Korean is different from the scrambling construction which involves movement in that the former may appear with resumptive pronouns, whereas the latter may not.

Based on Kuno (1972), D.-W. Yang (1973) postulates that the deep structure of a topical sentence embeds its corresponding non-topical sentence, as in (25) (from D - W. Yang (1973)).

(25) i salam-\m [s John-i i salam-\x\ manna-ss-ta.] This man-Top Nom this man-Acc meet-Pst-Dec ‘As for this man, John met this man.’

The topicalization rule then deletes the object NP i salam ‘this man’ under coreference with the topic NP. This account does not involve movement. The relationship between the topic and coreferential NP as in (25) can be explained in terms of the coreferential NP deletion.

However, the non-movement analysis of D.-W. Yang (1973) still cannot explain the contrast between (26a) and (26b). (25) and (23b) are repeated as (26a) and (27) below, respectively.

(26) a. i salam-un [ John-i i salam-ul manna-ss-ta.] This man-Topic Nom this man-Acc meet-Pst-Dec ‘As for this man, John met (him).’

b.* i salamrun [John-i ku-lul, manna-ss-ta.] this man-Top Nom he-Acc meet-Pst-Dec

‘As for this man, John met him/himself.’

(27) ce moca-nun John-i [rc (ku-kes,-ul) ssuko-i-ss-ten] salam-ul cal alkoiss-ta. that hat-Acc Nom the-thing-Acc wear be-Pst-AM person-Acc well know-Dec

‘As for that hat, John knows the person who was wearing it well.’

D.-W. Yang (1973) suggests that the pronominal-deletion is obligatory in (26b) but optional in (27). But this explanation is ad-hoc. He does not explain why that is so.

Now we are faced with a dilemma: examples like (17a) and (19) should involve movement but examples like (20b), (21), (24) and (27) involve base-generation in the surface position.