A Syntactic Approach to the 1st Person Restriction of Causal Clauses in Korean

The main purpose of this paper is to provide a syntax-based analysis of the differences between the two Korean causal clauses, i.e. ese-clauses and nikka-clauses. Focusing on the various aspects of Mood distinction, we claim that nikka and eseclauses can be analyzed as indicatives and subjunctives, respectively. Such an analysis enables us to provide syntactic explanations for issues—what we call the 1st person restriction of ese-clauses and its obviation—which might be considered merely semantic/pragmatic issues. 1 Main puzzle: The 1st person restriction of ese-clauses Korean has two causal connectives, -nikka and -ese ‘because’. In most cases, they can be used interchangeably without noticeable differences in their meanings: (1) Pi-ka o-ase/nikka ttang-i cecnunta. rain-Nom come-because ground-Nom wet ‘Because it rains, the ground is getting wet.’ The two connectives, however, exhibit different distributions with respect to the main clause subject when they contain a Contrastive Topic (CT) marking: while ese-clauses display person restrictions on the main clause subject, nikka-clauses do not. In (2), for example, both –ese and –nikka allow CT-marking when the main clause subject is 1st person. In this case, the CT conveys the implied message that the speaker believes that Hoya cannot fulfill other conditions required for marriage, such as a full-time job position. (2) a. pro2 cip-un sa-se, na1-nun pro house-CT buy-ese I-Top Hoya2-wa kyeolhonha-yss-ta. Hoya-with marry-Past-Decl b. pro2 cip-un sa-ss-unikka, na1-nun pro house-CT buy-Past-nikka I-Top Hoya2-wa kyeolhonha-yss-ta. Hoya-with marry-Past-Decl ‘I married Hoya because he (at least) bought a house.’ However, the two connectives behave differently when the main clause subject is not 1st person. As in (3a), an ese-clause does not allow CT-marking, when the main clause subject is 3rd person. We name this constraint the 1st Person Restriction (PR). Unlike ese-clauses, nikka-clauses allow CTmarking regardless of whether the main clause subject is 1st person or not. In (3b), Yuna is the one who believes Hoya at least meets the minimum condition for marriage, but it is possible that he cannot fulfill any other conditions.1 (3) a. #pro2 cip-un sa-se, Yuna1-nun pro house-CT buy-because Yuna-Top Hoya2-wa kyeolhonha-yss-ta. Hoya-with marry-Past-Decl ‘Yuna married Hoya because he (at least) bought a house.’ 1 As far as we know, this kind of difference doesn’t seem to be found in any other languages—even though some languages use two words for ‘because’ (e.g., German denn and weil). PACLIC 30 Proceedings 287 30th Pacific Asia Conference on Language, Information and Computation (PACLIC 30) Seoul, Republic of Korea, October 28-30, 2016 b. pro2 cip-un sa-ss-unikka, Yuna1-nun pro house-CT buy-Past-nikka Yuna-Top Hoya2-wa kyeolhonha-yss-ta. Hoya-with marry-Past-Decl ‘Yuna married Hoya because he (at least) bought a house.’ Note that an ese-clause displays this restriction only in cases where it contains a CT marker: (4) pro2 cip-ul sa-se, na1/Yuna1-nun pro house-Acc buy-ese I/Yuna-Top Hoya2-wa kyeolhonha-yss-ta. Hoya-with marry-Past-Decl ‘I/Yuna married Hoya because he bought a house’ More surprisingly, PR can be circumvented when even one of the arguments in the ese-clause is coindexed with the subject of the main clause. It is commonly claimed that as an applicative construction -e cwuadds a goal argument of the beneficiary relation (Jung 2014, etc.). PR can be obviated due to a co-indexation of the goal argument with the main clause subject, as in (5). (5) pro2 cip-un pro1 sa-cwu-ese, pro house-CT pro buy-give-because na1/Yuna1-nun Hoya2-wa kyeolhonha-yss-ta. I/Yuna-Top Hoya-with marry-Past-Decl ‘Yuna married Hoya because he (at least) bought a house for her.’ Just like (3b), the implied massage in (5) is that Yuna believes it is possible that Hoya meets the minimum condition for marriage, but he cannot fulfill other conditions. Thus, (3b) and (5) show that the unacceptability of (3a) cannot be attributed to a semantic/pragmatic anomaly. 2 Causal clauses and Contrastive marking To explain PR, it is necessary to understand the CT-marking in causal clauses. As widely pointed out in the previous literature, CT-marking is restricted in embedded contexts: while it is allowed in a causal clause but not in a temporal/conditional clause: (e.g., Hara 2008, Tomioka 2015 for Japanese, Park & Hoe 2015, etc.) (6) * Hangsang aitul-un cip-ey always children-CT house-to o-l ttay, kay-ka cic-nun-ta. come when dog-Nom bark-Pres-Decl ‘When (at least) children come to our house, dogs always bark.’ (based on Hara 2008) (7) Sacangnim-un John-i ilpone-nun president-Top John-Nom Japanese-CT hal cul-a-nikka/al-ase chayonghay-ss-ta. do can-because hire-Past-Decl ‘Because John can speak (at least) Japanese, the president hired him.’ (based on Hara 2008) According to Hara (2008), a CT-marking conveys an implied message that some epistemic bearer (mostly the speaker of the ‘utterance context’ (CU, hereafter)) entertains the possibility that the stronger scalar alternatives to the asserted proposition are false (e.g., it is possible that John cannot speak both Japanese and Korean in (7)). For such a scalar comparison, the use of CT requires some kind of epistemic bearer who has limited knowledge. Given this, Hara (2008) claims that CT-marking is licensed in causal clauses because causal clauses can fulfill the requirement about the epistemic bearer by introducing their own contexts. The idea of introducing an additional context can also be supported by Davidson’s (1963) primary reason. According to Davidson, doing something for a reason means doing something intentionally. Thus, to accept a rationalization of an action, we should be able to say what caused the agent to do the action (e.g., attractions, obligations, etc.). Davidson claims that when somebody does something for a reason, he must have a primary reason that consists of (i) a pro-attitude toward an action of a certain kind (e.g. desires come from moral believes, social conventions, etc.), and (ii) a belief that doing his action is of that kind, and (iii) this belief and desire cause him in the right way to do the action. In this way, the primary reason can be understood as a cause of the action. In order to accept only intentionally qualified causal relations, a kind of screening of the causal relation is added (expressed by “in the right way”). The relation between the reason/cause and its unintended outcomes cannot be qualified as a “right” causal relation. As exemplified in Davidson (1963), if someone turned on the light and by doing so he happened to alert the burglar in the room, the relation between the two events cannot be qualified as a right causal relation if the pro-attitude is supposed to be an intention. Given this, we can say that the use of causal clauses adds some additional context corresponding to the qualifying process. We call