©Hyuna B. Kim Coyote Papers 21 (2013) UA Linguistics Tucson, AZ, U.S.A. Pragmatic repair driven by indexicality Hyuna B. Kim UNM, CE [email protected] Abstract This paper aims to account for Double Accessibility effects found in Korean. Clearing out confusions in the previous discussion on the phenomenon, it claims that Korean does not have a Double Access reading in a semantic sense, unlike English, but Double Accessibility effects arise as a result of pragmatic repair which is employed in order to interpret a focused indexical element causing a conflict in the interpretation process. The advantages of the pragmatic analysis defended in this paper over the movement analyses proposed in the literature will be shown in details. 1 Introduction English present under past sentences have a peculiar reading of ‘Double Accessibility,’ which is illustrated in the following Abusch’s examples (1991): (1) a. John believed that Mary is pregnant. b. John said that Mary lives in Chelsea. Kim, p.2 1 INTRODUCTION The basic intuition is that the embedded present tense in (1a) or (1b) does not overlap with the utterance time only, but also with the time of the believing or saying event of the matrix clause, as Comrie (1985) and Smith (1978) pointed out. More specifically, in (1a), the event of Mary’s being pregnant holds at two time intervals, the time of John’s believing and the utterance Time, yielding the peculiar reading where Mary was pregnant at a certain past time when John had such a belief and her pregnancy continues up to the utterance time. The continuous event from one past time interval to the utterance time characterizes double access sentences1. Thus the phenomenon might not be best couched in the term ‘double accessibility’ because with that term one might reasonably think of the case of temporal access to two separate or discounted temporal intervals, which is not available for double access sentences as in (1a) and (1b). Anyways, the double access sentences in (1a) and (1b) present a very interesting case such that the embedded present tense does not behave as a simple deictic tense, but somehow it receives a so-called ‘double’ temporal interpretation. A question to arise from a typological perspective is whether the same phenomenon can be found in other languages such as Japanese and Korean. Ogihara (1996) affirmed it by providing the Japanese data below: (2) Taroo-wa Hanako-ga ima Taroo-TOP Hanako-NOM now Tookyoo-ni i-ru to it-ta-yo. Tokyo-at be-PRS that say-PST ending Reading #1: ‘Taro said that Hanako was in Tokyo at that time’ Reading #2: ‘Taro said that Hanako was in Tokyo at that time and (the speaker assumes that) she is still in Tokyo at the utterance time’ 1 The double access reading of (1a) and (1b) would become clearer when compared to the following sentences: (i) John believed that Mary has been pregnant (ii) John said that Mary has lived in Chelsea The sentences in (i) and (ii) are also considered as double access sentences, and they are considered to have the very similar meaning with (1a) and (1b), except that the continuity of the embedded event in question becomes obvious by the presence of the aspectual auxiliary ‘has’ in (i) and (ii). Kim, p.3 1 INTRODUCTION According to Ogihara (1996), the sentence (2) has two readings, one of which, Reading #2, is a double access reading. In Reading #2, the subordinate event of ‘Hanako’s being in Tokyo’, the speaker assumes, continued from the past time of Taro’s saying it up to the utterance time2. But, there is an obvious difference between English in (1) and Japanese in (2), because the double access reading in Japanese is just one of possible readings with present under past, unlike the English case in (1) where the double access reading is obligatory. Ogihara (1996) attributes the difference to the presence / absence of obligatory tense movement of the embedded present tense. Ogihara (1996) claims that English that has 2 Ogihara (1996) shows that the double access reading for (2) can survive even when a past time denoting adverb that prevents the upstairs and downstairs event from overlapping due to time conflict is added to the matrix clause, as below: (i) Taroo-wa kinoo Hanako-ga Taroo-TOP yesterday Tookyoo-ni i-ru to it-ta-yo. Tokyo-at be-PRS that say-PST ending Hanako-NOM ima now Reading #1: ‘Yesterday, Taro said that Hanako was in Tokyo at that time’ Reading #2: ‘Yesterday, Taro said that Hanako was in Tokyo at that time and (the speaker assumes that) she is still in Tokyo at the utterance time’ JOgihara (1996) presents the sentence in (i) as the evidence for a double access reading in Japanese. However, there are several things that have to be checked out before we consider the readings of (i) as supporting evidence. First, we need to have a grammatical device to preclude from possible readings under discussion a direct quotation interpretation which is most of time available with the complementizer ‘-to’ in Japanese and ‘ko’ in Korean. It could be the main source of confusion. Given that, Reading #1 has to be excluded from the discussion because it is a direct quotation interpretation. The past time denoting adverb ‘kinoo’ (yestereday) and the utterance time denoting one ‘ima’ (now) definitely cannot go with each other in the simultaneous reading between the matrix and embedded clauses. What is means that the simultaneous reading, Reading #1, is made unavailable when the time conflicting adverb ‘kinoo’ (yesterday) is added to the sentence (unless you assume that Japanese can have a non-deictic time adverb ‘ima’). But, in a direct quotation construction, such a reading is possible. By still including Reading #1 for (2), Ogihara (1996) unintentionally brings confusion to discussion. Kim, p.4 1 INTRODUCTION an obligatory movement triggered by an inherent deictic present tense gets an obligatory double access reading with present under past, while Japanese with only optional tense movement for a ‘sometimes’ deictic present tense gets an optional double access reading. Yet, note that Ogihara (1999), turning from Ogihara (1996), tried to provide a generalized account for double accessibility across languages by positing that that both English and Japanese alike have deictic and anaphoric tenses and hence they have a double access reading in the same fashion. The only difference between the two languages lies in the fact that the double access reading in Japanese is somehow obscured by the presence of a strong simultaneous reading of present under past3. Now, would the story work for other Japanese type languages such as Korean and so could we claim that Korean also has double access sentences? It looks like we can find in Korean an example corresponding to the Japanese double access sentence in (2), as below: (3) Taro-nun Hanako-ka cikum Tokyo-ey Taro-TOP Hanako-NOM now Tokyo-in iss- ø-ta-ko be-PRS-DCL-COMP malhay-ss-ta say-PST-DCL Reading #1: ‘Taro said that Hanako was in Tokyo at that time’ Reading #2: ‘Taro said that Hanako was in Tokyo at that time and (the speaker assumes that) she is still in Tokyo at the utterance time’ Reading #2 shows that the sentence in (3) is to be associated with two time intervals, some past time and the utterance time, yielding a double access reading. Then, can we conclude that Korean has an optional double access reading) as in Japanese? Given the parallel pattern, 3 The Ogihara’s (1999) improving effort does not seem to be successful. He is trying to provide a unified account for English and Japanese double access sentences, but it costs him explanatory adequacy because he inevitably ignores the apparent differences between the English and Japanese double access sentences. For instance, while a double access reading is available only with certain predicates like ‘live’ or ‘be pregnant’ in English, but such a restriction does not apply to Japanese or Korean double access sentences. The Ogihara’s (1999) account fails to provide an explanation for the difference. We will come back to this issue in the following section. Kim, p.5 1 INTRODUCTION it looks just simple but the actual picture of the double access reading in Japanese or Korean is complicate. Before we jump into the simple conclusion, we need to eliminate all the interfering factors and make clear whether or not the alleged double access reading is real in a semantic sense in Korean. In a nutshell, I reject the parallel analysis for English and Korean and claim that Korean does not have a double access reading in a semantic sense, defending a pragmatic analysis for the double access effects in Korean. In section 2, I will explain why I am being skeptical about the claim that Japanese or Korean has a double access sentence and identify the true status of a so-called double access reading in Korean, thereby clearing out the confusions in Ogihara (1996) and (1999). In section3, I will turn to the double access effects arising with an indexical adverb ‘cikum’ (now) in Korean and sketch the movement based syntactic solutions suggested in the literature. In section 4, I will propose a pragmatic analysis for the double access effects in Korean. It would explain better the pragmatic aspects of the double access sentences that are strongly associated with the speaker’s assumptions. Section 5 is conclusion and remaining issues 2 Clearing out the confusion 2.1 Securing the target interpretation One interfering factor always coming in with a complementizer ‘-ko’ in Korean is a direct quotation interpretation4. As the example (5) below shows, both direct and indirect speech interpretations are available with the embedded clause headed by ‘-ko’: (4) Taro-nun Hanako-ka Tokyo-ey iss- ø-ta-ko Taro-TOP Hanako-NOM Tokyo-in be-PRS-DCL-COMP malhay-ss-ta say-PST-DCL Reading #1: ‘Taro said, “Hanako is in Tokyo.”’ 4 Japanese patterns with Korean with respect to the direct quotation interpretation of a clause headed by ‘-to.’ Kim, p.6 2 CLEARING OUT THE CONFUSION Reading #2: ‘Taro said that Hanako was in Tokyo (at that time).’ The English glosses given in (4) can differentiate Reading #1, a direct speech interpretation from Reading #2, an indirect speech one, using different grammatical devices such as quotation marks for the former and a complementizer ‘that’ for the latter. But, Korean does not have such a tools to differentiate them except for the marked form of ‘-lako’ for a direct speech interpretation, which yet can be often shortened to ‘-ko.5’ Note that in (4) they end up with having the same interpretation, which serves as a source of confusion6. Even when the indirect interpretation is made unavailable, Korean native speakers would get the reading through the still available direct quotation interpretation which happens to be identical. Let us see how confusion arises when we add an indexical adverb ‘cikum’ (now) to (4): 5 There are other formal tools to differentiate them, for example, a pause placed right after the direct quotation marker ‘-ko’ along with placing a high peach on it. Using the phonetic information, speakers / hearers can disambiguate the interpretations, but when presented in a written form lacking such information, the two types of interpretation are always available, bringing confusion to the current discussion. 6 Things go different with the past under past case and an ambiguity issue does not arise: (i) Taro-nun Hanako-ka ecey Taro-TOP Hanako-NOM yesterday Tokyo-ey Tokyo-in iss-ess-ta-ko be-PST-DCL-COMP malhay-ss-ta say-PST-DCL Reading #1: ‘Taro said, “Hanako was in Tokyo yesterday.”’=‘Taro said that Hanako was in Tokyo the day before Taro said it.’ Reading #2: ‘Taro said that Hanako was in Tokyo yesterday.’ As the Reading #1, the direct quotation reading shows, ‘yesterday’ inside a direct quotation ends up with referring to ‘yesterday’ of ‘Taro’ which turns out to be the day before Taro spoke out the sentence. In contrast to this, in an indirect quotation reading of Reading #2, ‘yesterday’ remains as ‘yesterday’ of the speaker. What it means is that the issue of direct or indirect quotation does not bring a confusion to the interpretation of a past tensed embedded clause headed by ‘-ko,’ but it does in the present tense case as in (4). Thus, one has to be careful in dealing with a present tensed ‘-ko’ clause. Kim, p.7 (5) 2 CLEARING OUT THE CONFUSION Taro-nun Hanako-ka cikum Tokyo-ey iss- ø-ta-ko Taro-TOP Hanako-NOM now be-PRS-DCL-COMP Tokyo-in malhay-ss-ta say-PST-DCL Reading #1: ‘Taro said, “Hanako is now in Tokyo.”’ ð ‘Taro said that Hanako was in Tokyo at that time.’ Reading #2: ‘Taro said that Hanako was in Tokyo *now.’ ð ‘Taro said that Hanako was in Tokyo at that time and (the speaker assumes that) she is still in Tokyo at the utterance time’ The direct quotation reading in Reading #1 is straightforward. In the reading, ‘cikum’ (now) is isolated in the direct quotation, not being able to referring to the utterance time of the speaker. It is Taro’s ‘now’ when he made the utterance about Hanako’s being in Tokyo. In contrast to it, a conflict arises when ‘cikum’ (now) is added to the indirect speech. The embedded tense is being anaphoric to the matrix past tense, resulting in the past tense meaning of the embedded event, which the utterance time referring ‘cikum’ (now) cannot modify7. If we assume that ‘cikum’ (now) modifies the embedded event properly referring to the utterance time, the embedded tense should be interpreted as a deictic present tense only, which is the same configuration as in the English double access sentence in (1). Consequently, we would have to say that the only available reading for (5) is Reading #2, the double access reading, but not (1). It is so because Reading #1 is a direct speech reading and is not available for the indirect speech reading. Nevertheless, Reading #1 is often considered as a possible reading for the indirect speech case of the sentence in (5), with no attention to the interfering fact of the direct speech reading, as Ogihara (1996) did in the corresponding Japanese case in (2), where confusion kicks in. The direct speech reading in Reading #1 keeps interfering, making things look like Reading #1 is a primary reading for (5) and Reading #2 is a secondary one, which is incorrect. Moreover, the existence of the wrong interpretation, Reading #1, for the indirect 7 The situation is very similar to the English case in which ‘now’ cannot modify the embedded event ‘was in LA’ that is interpreted anaphoric to the matrix past event in the English gloss in (5) because of the time mismatch. Yet obviously there is a difference in morphological features that the embedded tenses in English and Korean carry. They would make the English case work differently with respect to the adverb and tense interaction. Kim, p.8 2 CLEARING OUT THE CONFUSION speech case of (5), obscures the semantic conflict between the adverb ‘cikum’ (now) and the embedded anaphoric tense because in Reading #1 the same indexical ‘cikum’ (now) does not raise any problem. In order to remove such an interfering factor that brings confusion to the discussion, we need to use an element requiring a dependency between the matrix and embedded clause, which would be not allowed to appear in direct quotation in isolation. A long distance anaphor can make a good candidate for the job. When a long distance anaphor ‘caki’ (self) is inserted inside the embedded clause lacking a suitable antecedent, it needs to build a dependency across a clausal boundary to find its antecedent (Yoon 1989, Gill 1999, Kim & Yoon, 2006), which rules out the direct quotation reading, as follows: (6) Taro-nun caki chinku-ka Tokyo-ey Taro-TOP self Tokyo-in friend-NOM iss- ø-ta-ko be-PRS-DCL-COMP malhay-ss-ta say-PST-DCL Reading #1: ‘Taro said, “a friend of *himself is now in Tokyo.” ’ Reading #2: ‘Taro1 said that his1 friend was in Tokyo at that time. In (6), Taro, the matrix subject serves as an antecedent for a long distance anaphor ‘caki’ (self) because there is no available antecedent in the clause headed by ‘-ko.’ If it appears inside direct quotation, the whole sentence will become ungrammatical or unacceptable because a dependent relation needed for ‘caki’ cannot be built anywhere in the sentence. Hence, the direct quotation reading, Reading #1, is removed when ‘caki’ (self) is added, thereby letting us avoid the confusion caused by the interfering factor8. The sentence form in (6) will be the basic one to be used in the rest of our discussion in this paper. 8 A long distance anaphor does its job for differentiating indirect speech from direct one, better than other tools such as anaphoric pronouns, honorifics, or polarity items which also need a syntactic / semantic dependency. However, it is not a perfect solution because there are some deviant usages of ‘caki’ (self) where ‘caki’ (self) does not require a syntactic antecedent in a sentence and can get its antecedent from a context, which is shown below: (i) (Talking about Taro) Kim, p.9 2.2 2 CLEARING OUT THE CONFUSION Reasons for being skeptical Let’s come back to the initial question and see if the Korean sentence in (6) can have a double access reading. Given the parallelism between Japanese and Korean, Ogihara’s (1996, 1999) would propose that it is the case. But, a closer look at the case led me to find several reasons to cast serious doubt on such a proposal. So, in this section, I will explain why I am being skeptical about it even if in fact it would be much simpler and neat to say that (6) has a double access reading as in English and Japanese. Our discussion starts with the sentence (6) with no temporal adverb. The temporal indexical ‘cikum’ (now) is omitted intentionally for us to make a move step by step in the discussion. We will get to the case with ‘cikum’ (now) in the next section. Caki-nun calmos-ha-n kes-i eps-e? Self-TOP mistake-make-PAST.REL thing-NOM not.be-PRS-INT ‘Can he say that he never made a mistake?’ There is no available syntactic antecedent for ‘caki’ in (i), but because a context provides ‘Taro’ for it, the sentence in (i) appears to be grammatical. We do not have a space in this paper to see what would happen when it is embedded inside a clause headed by ‘-ko,’ but it should be interesting enough to look into. Keeping it in mind, consider the following sentences with a direct quotation construction: When a typical direct quotation marker ‘-lako’ replaces ‘-ko,’ the degree of ungrammaticality gets stronger, as shown above. (ii) */?? Taro-nun Taro-TOP caki self chinku-ka friend-NOM iss- ø-ta-lako Tokyo-ey Tokyo-in malhay-ss-ta be-PRS-DCL-COMP say-PST-DCL Reading #1: ‘Taro said, “a friend of *himself is now in Tokyo.” ’ And, when the embedded clause takes an informal speech clausal ending which can appear only in direct speech, but not in indirect one, ungrammaticality gets enhanced: (iii) *Taro-nun Taro-TOP caki self chinku-ka friend-NOM iss- ø-e-lako be-PRS-DCL-COMP Tokyo-ey Tokyo-in malhay-ss-ta say-PST-DCL Reading #1: ‘Taro said, “a friend of *himself is now in Tokyo.” ’ Kim, p.10 2 CLEARING OUT THE CONFUSION The first reason is that there is no real syntactic or semantic force for double accessibility in Korean, one of non-Sequence of Tense languages. In Sequence of Tense language like English, an embedded past tense can be interpreted simultaneous to the matrix past tense. It goes hand in hand with the fact a present tense is rarely used for a simultaneous reading, whether embedded or not. Hence, the main source for double accessibility in English, one of Sequence of Tense language, is the English present tense is deictic, which cannot be interpreted with no reference to the utterance time. However, Korean, one of non-Sequence of Tense language, does not have such a feature. The Korean present tense can be purely anaphoric when embedded so that it does not have a strong deictic force in it, as Ogihara (1996) indicated. Going back to the example in (6), the embedded present tense is interpreted simultaneous to the matrix past tense because it is the way the Korean present tense is normally interpreted. There is no syntactic or semantic reason to induce double accessibility like deicticity or something as in English. Then, why should we go a step further and add another reading of double accessibility to (6) with no reason? Secondly, a double access reading is a highly marked temporal interpretation so that it is available only with certain types of predicates such as ‘be pregnant’ and ‘live’ in English. In contrast, Korean does not have such a restriction on predicates which allow a double access reading. Rather, the reading can readily arise with any kind of predicates, as bellow: (7) a. Taro-nun Hanako-ka Tokyo-ey iss-∅-ta-ko Taro-TOP Hanako-NOM Tokyo-in be-PRS-DCL-that ‘Taro said that Hanako was in Tokyo.’ b. Taro-nun Hanako-ka Taro-TOP Hanako-NOM Tokyo-ey sa(l)-n-ta-ko malhay-ss-ta say-PST-DCL malhay-ss-ta Tokyo-in live-PRS-DCL-that say-PST-DCL ‘Taro said that Hanako was living in Tokyo.’ c. Taro-nun Hanako-ka Tokyo-ey ka-n-ta-ko Taro-TOP Hanako-NOM Tokyo-in malhay-ss-ta go-PRS-DCL-that say-PST-DCL ‘Taro said that Hanako was going to Tokyo.’ d. Taro-nun Hanako-ka Tokyo-ey tochakha-n-ta-ko malhay-ss-ta Taro-TOPHanako-NOM Tokyo-in arrive-PRS-DCL-that say-PST-DCL ‘Taro said that Hanako was arriving in Tokyo.’ Kim, p.11 2 e. Taro-nun Hanako-ka Taro-TOP Tokyo-ey cip-ul Hanako-NOM Tokyo-in CLEARING OUT THE CONFUSION cis-nun-ta-ko house-OCC build-PRS-DCL-that malhay-ss-ta say-PST-DCL ‘Taro said that Hanako was building a house in Tokyo.’ All of the sentences in (7) have the primary simultaneous reading as given in the English glosses, and a double access reading would be available with each of them when a proper context is given, putting aside the issue whether it is a separate reading or not. The fact indicates that Korean double access sentences do not share the main characteristics with English ones. Last but not least, the alleged double access reading for the Korean sentence in (6) is better explained in terms of pragmatic inferences based on the speaker’s assumption. (6) is repeated below: (8) Taro-nun caki chinku-ka Tokyo-ey Taro-TOP self Tokyo-in friend-NOM iss- ø-ta-ko be-PRS-DCL-COMP malhay-ss-ta say-PST-DCL ‘Taro1 said that his1 friend was in Tokyo at that time. With (8), one can naturally think of two logical possibilities with respect to the current state of the embedded event at the utterance time, as in (9): (9) Logical possibilities: (i) The event of Taro’s friend’s being in LA might have been going on up to the utterance time, (ii) Or, it might have ceased before the utterance time. When a context provides a reason for the speaker or hearer to make an inference based on the possibility (i), we can get a so-called double access reading such as ‘Taro said that Hanako was in Tokyo at the time of Taro’s speech and the speaker assumes that she is still Kim, p.12 2 CLEARING OUT THE CONFUSION there at the utterance time.’ Note the logical possibilities in (9) are not a part of an asserted meaning of (8) in Grice’s (1989) sense. Given that, the double access reading is not a separate reading in a semantic sense, but the alleged reading is what we get when we add our pragmatic inference to the asserted meaning of (8). In other words, double accessibility is not grammatically encoded in Korean unlike English, but out of contextual information enhancing certain pragmatic inferences of speakers and hearers available in conversation. Yet, it is just one of natural consequences resulting from information processing and updating which are also found in other languages. Consider the following English example9: (10) Taro said that his friend was in Tokyo. ‘Taro said that his friend was in Tokyo at the time of Taro’s speech.’ Possibilities: (i) The event of Taro’s friend’s being in Tokyo might Have been going on up to the utterance time, (ii) Or, it might have terminated before the utterance time. Just like in the Korean example in (8), two logical possibilities arise with respect to the current state of the embedded event for (10). And, when the speaker or hearer adds his / her pragmatic inference based on (i) which is most compatible with context, he / she would be able to get the so-called double access reading. However, it is apparently differentiated from the real or grammatical double access reading whole part of which composes of an asserted meaning as in (11) below: (11) a. John said that Mary lives in LA. b. John said that Mary has lived in LA. The sentences in (11) are considered to have a double access reading while one in (10) is not, even though similar temporal information with respect to double accessibility might involve with both cases. Likewise, the alleged double access reading for (8) is the same kind of 9 Let us ignore another back-shifted reading available for (9) and focus on the Sequence of Tense reading, where the embedded past tense is interpreted as simultaneous to the matrix past tense. Kim, p.13 2 CLEARING OUT THE CONFUSION reading or information that we get from (10) in English, which means that it is not another reading to treat semantically. 2.3 The double access reading effects The discussion in the previous subsection led me to the conclusion that the Korean sentence in (8) does not have a double access reading in a semantic sense. Yet, one can still get similar information through working pragmatic inferences, which we call ‘Double access effects.’ Let us see how things go when an indexical ‘cikum’ (now) is added to (8). (12) Taro-nun caki chinku-ka cikum Taro-TOP self now iss- ø-ta-ko friend-NOM be-PRS-DCL-COMP Tokyo-ey Tokyo-in malhay-ss-ta say-PST-DCL ‘Taro1 said that his1 friend was in Tokyo at that time (and the speaker / hearer assumes that his friend is still there at the utterance time). The semantic conflict brought by the insertion of the indexical ‘cikum’ (now) was explained in the previous subsection. The utterance time denoting adverb ‘cikum’ (now) cannot modify the anaphoric tense in the embedded clause and fails to be interpreted properly. But, the sentence in (12) does not turn out to be ungrammatical but receives an interpretation with double access effects, yet somewhat stronger than the one associated with pragmatic inferences as in (8). Hence, an account needs to be provided for (12). In the following section, we are going to briefly go over the movement analyses put forth in the literature, and the proposed pragmatic analysis will be given in detail in Section 4. Kim, p.14 3 Movement analyses 3.1 Tense movement 3 MOVEMENT ANALYSIS The syntactic movement solution have been suggested for a double access reading, as in Abush (1991, 1997), Stowell (1993) and Ogihara (1997). Even though it is couched in different terms and structures, the important assumption that the proposals share in common is that the English present tense cannot stay under the scope of the matrix past tense and should scope out to get access to the utterance time. Let us briefly review the tense movement analyses and discuss their limitation in dealing with double access effects found in Korean. Abush’s (1991, 1997) temporal de re analysis accounts for the English double access reading in (1) in terms of a temporal de re and the Upper Limit Constraint. In her analysis, the access to the utterance time is achieved by a ‘scoping out’ mechanism, as shown in (13): (13) John PAST1 say [s PRES3 λ3λ2 [s Mary t3 be pregnant ]] Abusch (1991, 1997) moves the embedded present tense, a temporal de re, out of the intensional domain of λ2 to the position where the present tense can be evaluated with respect to the Utterance time, t0. As for the other access to the past time of John’s saying event, Abusch (1997)’s ‘Upper Limit Constraints10’ do its work: the time denotation of the trace of Pres3, t3 inside the intensional domain cannot exceed the time of John’s saying, the upper limit. Hence, the denotation of present tense also has to include the time interval including ‘the past time of John’s saying’ through the interpretation of the trace remaining in the intensional domain, resulting in a double access reading. Ogihara (1996) proposes another movement analysis for a double access reading. 10 The upper limit constraint is a constraint suggested to account for the general pattern of the interpretation of the embedded tense in English. Abusch’s (1997) wanted to explain why past under past cannot receive the reading where the embedded past tense follows the matrix past tense in a complement clause. It is reinterpreted by Heim (1982) in terms of temporal presupposition. Kim, p.15 3 MOVEMENT ANALYSIS Given his assumption that utterance time oriented indexicals cannot appear in the translation of the intensional argument of an attitude verb (Ogihara 1996, p 211), he claims (i) in English, an absolute present tense, an indexical element in nature, obligatorily moves out of an intensional domain at LF. Whereas, (ii) in Japanese, a relative present tense, not necessarily tied to the speech time, optionally moves out of an intensioal domain at LF. In his analysis, the double accessibility is accounted for as follows: the access to the utterance time is obtained by a tense movement out of the intensional domain as below: (14) 9 [CPPRES2 [SJohn PAST1 say t2 [CPthat [S Mary t1 be pregnant ]]]] A cyclic movement to the empty Comp position, leaving a trace, t1 and t2 at LF makes PRES2 / t2 accessible to the utterance time, as in (14). As for the access to the time of John’s saying, by stipulation, t2, a tense that moves out of an intensional domain, obtains a new index distinct from the initial one t1; t1 is accessible to the time of John’s saying. 3.2 Limitations The important implications that these tense movement analyses of a double access reading have about Korean cases is that, if Korean patterns with Japanese, it is expected that Korean has an optional tense movement which results in a double access reading in the same fashion as in English. However, the discussion in the previous section shows that there are significant differences between English and Korean with respect to the characteristics of a double access reading. Ogihara (1996) made the story simple by ignoring the important differences and wanted to provide a generalized account for English and Japanese with a minor language variation depending on the optionality / obligatoriness of the tense movement. Let us point out that the movement analyses run into several problems. First of all, it fails to account for the case where a tense movement should be involved but no double access reading is found. The example in (15) below shows that tense movement does not necessarily involve with a double access reading: Kim, p.16 (15) 3 MOVEMENT ANALYSIS John said that this guy who is now sitting in front of you killed the woman at the bar. The deeply embedded present tense inside the relative clause has to move out up to the matrix tense so that it could get an access to the Utterance time, with which it is to be evaluated. However, the tense movement does not result in a double access reading for (15). It seems that the present tense just freely moves out to the most outer clause; the present tense does not hold at two time points, but it is associated with a single time point, that is, the utterance time. Thus, the movement analyses turn out to run into a problem in dealing with this case. The next issue is on the interpretation of indexical items. If an indexical present tense should move up due to its deicticity, do all the indexical items such as indexical time and space adverbs have to move out to get interpreted with respect to the utterance time? But, in general it is assumed that indexical adverbs are to be interpreted in situ. If they remain in situ and can be interpreted deictically, why should the deictic tense undergo movement? And, even if we assume that adverbs are somehow special, having an ability to get access to the utterance in situ, how can they properly modify the trace or copy of the present tense left after movement which is to be interpreted as anaphoric to the matrix past tense. There are quite a few unanswered questions about the interpretation of indexical items in a double access reading. The last one is the matter of filling up between the two time intervals in a double access reading. The movement analyses provide an account for how to get access to each time intervals, that is, the utterance time and some past time interval. But, the important thing of a double access reading is that the embedded event continues from some past time provided by a sentence up to the utterance time. Yet, the movement analyses do not say much about the aspectual property of a double access reading. Kim, p.17 4 DEFENDING A PRAGMATIC ANALYSIS 4 Defending a pragmatic analysis 4.1 A semantic conflict Let us go back to the Korean sentence with an indexical ‘cikum’ (now) in (11), which is repeated in (16) below: (16) Taro-nun caki chinku-ka cikum Taro-TOP self now iss- ø-ta-ko friend-NOM malhay-ss-ta be-PRS-DCL-COMP Tokyo-ey Tokyo-in say-PST-DCL ‘Taro1 said that his1 friend was in Tokyo at that time (and the speaker / hearer assumes that his friend is still there at the utterance time). As discussed before, the indexical item ‘cikum’ (now) referring to the utterance time cannot be interpreted properly with the anaphoric present tense in the embedded clause in (16), resulting in a crash in interpretation, which turns out to be a source for double access effects. Yet, the semantic conflict is obscured and not easy to see so that we need to find a way to show the conflict clearly. The comparison to another structure, that is, a present tensed relative clause with ‘cikum’ (now) would reveal it. Consider the following example: (17) Taro-nun [ e1 Taro-TOP Tokyo-ey Tokyo-in cenwha-lul hay-ss-ta call-OCC iss- ø –nun ] be-PRS-REL caki self chinku1-eykey friend-to do-PST-DCL Reading #1: ‘Taroi made a phone call to a friend of hisi who is in Tokyo now. Reading #2: ‘Taroi made a phone call to a friend of hisi who was in Tokyo at that time. Kim, p.18 4 DEFENDING A PRAGMATIC ANALYSIS The present tensed relative clause typically receives two readings, a deictic reading and an anaphoric one. But, when the utterance time denoting adverb ‘cikum’ (now) is added to (17), the anaphoric reading, Reading #2, is removed as in (18): (18) Taro-nun [ e1 cikum Tokyo-ey iss- ø –nun ] Taro-TOP now Tokyo-in caki chinku1-eykey be-PRS-REL self friend-to cenwha-lul hay-ss-ta call-OCC do-PST-DCL Reading #1: ‘Taroi made a phone call to a friend of hisi who is in Tokyo now. Reading #2: ‘Taroi made a phone call to a friend of hisi who was in Tokyo *now. The disambiguating effect by ‘cikum’ (now) in (18) above shows that the indexical ‘cikum’ (now) cannot go with the anaphoric present tense in the relative clause because it causes a semantic conflict. Otherwise, there is no reason Reading #2 becomes unavailable by the presence of ‘cikum’ (now). Given this, it is expected that there should be a conflict of the same kind in (16). The interesting thing is that the conflict would normally lead (16) to be ungrammatical because there is no any other reading available for it, in contrast to the situation in (18), but in fact the sentence in (16) is saved by the double access effect made available with it, maintaining the indexical ‘cikum’ (now) within the anaphoric reading of the embedded clause. Then, how can we explain it? First of all, de we have to say that the Korean sentence in (16) gets a double access reading just as in English because it turns out to be the only possible reading? Then, would a double access reading which was absent in (8) with no ‘cikum’ (now) be able to come to life when the adverb is added in (16)? The short answer would be that (16) does not have a double access reading but rather stronger double access effects, along the same line that we have put forth so far11. The double 11 Our discussion can go on when the grammaticality of (16) is assumed. If it is ungrammatical, it is due to the semantic conflict, and our discussion stops there. But, many of Korean native speakers confirmed the grammaticality of (16). Kim, p.19 4 DEFENDING A PRAGMATIC ANALYSIS access effect is indicated in the parenthesis in the English gloss for (16). The reason the double access effect is easier to see in (16) is because a pragmatic repair in relation to the effect is done on (16) due to the semantic conflict discussed. The pragmatic solution will be addressed in detail in the following subsection. Yet, before we move onto it, let us check whether the indexical ‘cikum’ (now) is really referring to the utterance time, as follows: (19) Taro-nun caki chinku-ka cikum Tokyo-ey Taro-TOP self now friend-NOM iss- ø-ta-ko be-PRS-DCL-COMP sasil now malhay-ss-ciman say-PST-CONJ ku-nun cikum in fact he-TOP Tokyo-in Seoul-ey iss-ø-ta. Seoul-in be-PRS-DCL. ‘Taro1 said that his1 friend was in Tokyo at that time (and the speaker / hearer assumes Taro1 believe that [his1 friend]2 is still there at the utterance time) but he2 is in Seoul at the utterance time. In (19), ‘cikum’ (now) in the matrix clause is a deictic adverb which is referring to the utterance time with no exception. The fact the embedded event and the conjoined event show a contrast about the current status of the location of Taro’s friend indicates that both of the two instances of ‘cikum’ (now) in (19) refer to the utterance time in the same manner. It means that ‘cikum’ (now) in the embedded clause in (19) is for sure a deictic adverb which would cause a semantic conflict in an anaphoric interpretation. 4.2 Strengthening and Weakening Now we turn to the reason why (16) is still grammatical in spite of the semantic conflict between the embedded tense and the temporal indexical ‘cikum’ (now) in (16)12. I propose that it is so because a pragmatic repair on the meaning of ‘cikum’ (now) is made, which make it possible for ‘cikum’ (now) to modify the embedded anaphoric tense in (16). But, 12 When the sentence in (16) is presented to Korean native speakers, a delay in processing time takes place in general. Kim, p.20 4 DEFENDING A PRAGMATIC ANALYSIS such a pragmatic repair does not take place for no reason. It comes in the way only when the conflicting temporal indexical ‘cikum’(now) strengthens the possible inferences associated with (16). See the following examples: (20) Taro-nun caki chinku-ka Taro-TOP self iss- ø-ta-ko Tokyo-ey friend-NOM Tokyo-in malhay-ss-ta be-PRS-DCL-COMP say-PST-DCL ‘Taro1 said that his1 friend was in Tokyo at that time. We saw before that a double access effect arises with (20) when a context supports the pragmatic inference such that the event of Taro’s friend’s being in Tokyo might have been going on up to the utterance time. When ‘cikum’ (now) is added, it strengthens the inference because it tries to tie the embedded event to the utterance time, which is a part of the inference, despite of the apparent semantic conflict. Thus, if the temporal indexical ‘cikum’ (now) were made to modify the embedded event, it would make the context supported pragmatic inference stronger. It is claimed that such strengthening triggers a pragmatic repair on the indexical ‘cikum’ (now). If it is on a right tract, it is predicted that when such a pragmatic inference is weakened by a context or other adverbs, a pragmatic repair does not take place. See below: (21) I Two nyen year iss- ø-ta-ko cen-ey before be-PRS-DCL-COMP Taro-nun caki chinku-ka Tokyo-ey Taro-TOP self Tokyo-in malhay-ss-ta friend-NOM say-PST-DCL ‘Two years ago, Taro1 said that his1 friend was in Tokyo at that time. The past time denoting adverb ‘two years ago’ is added to (20), the inference associated with double access effects is weakened. In that case, the further addition of ‘cikum’ (now) is not able to strengthen the inference so that no pragmatic repair takes place, leading to the unacceptability / ungrammaticality of (22): Kim, p.21 (22) 4 ??/# I Two nyen cen-ey Taro-nun year Taro-TOP Tokyo-ey Tokyo-in before caki chinku-ka self iss- ø-ta-ko be-PRS-DCL-COMP DEFENDING A PRAGMATIC ANALYSIS friend-NOM cikum now malhay-ss-ta say-PST-DCL ‘??Two years ago, Taro1 said that his1 friend was in Tokyo (and the speaker assumes that his friend is still in Tokyo at the utterance time.)’ The ungrammaticality does not simply come from the mismatch between the temporal adverbs in matrix and embedded clauses in (22). Compare it to (23) below: (23) [Context: Taro knows that his friend came to Tokyo for business and it would take several ?Ecey weeks to get every things done.] Taro-nun caki chinku-ka Yesterday Taro-TOP self friend-NOM Tokyo-ey iss- ø-ta-ko Tokyo-in be-PRS-DCL-COMP cikum now malhay-ss-ta say-PST-DCL ‘Yesterday, Taro1 said that his1 friend was in Tokyo (and the speaker assumes that his friend is still in Tokyo at the utterance time.)’ Normally, the past time denoting indexical ‘ecey’ (yesterday) would not go well with the utterance time denoting one ‘cikum’ (now). But when a proper context is provided, it would not weaken the inference for a double access effect, and in turn allow ‘cikum’(now) that is strengthening the inference. 4.3 A pragmatic repair: Extended ‘now’ Then, what kind of pragmatic repair would be made on ‘cikum’ (now) to make it work? The idea that I implement for it is that we can extend the meaning of ‘cikum’ (now) to the extent it can overlap with the past time event of the matrix clause in (16). Stretching the meaning Kim, p.22 4 DEFENDING A PRAGMATIC ANALYSIS of ‘now’ is not in fact new. ‘Cikum’ (now) can be stretched showing a flexibility of appearing with past or future tenses, as follows: (24) a. Taro-nun Taro-TOP cikum chulpalhay-ss-ta. now leave-PST-DCL. ‘Taro left just now.’ b. Taro-nun Taro-TOP cikum chulpalha-n-ta. now leave-PST-DCL. ‘Taro is leaving now.’ c. Taro-nun Taro-TOP cikum chulpalha-l-keya. now leave-FTR-DCL. ‘Taro will leave now/soon.’ The sentences in (24) show that ‘cikum’ (now) can be stretched to modify even the past or future event. Yet, the stretch is limited and it can only modify the events that are very close enough to overlap with the utterance time. And, such a stretch has been identified with ‘cikum’ (now) with respect to the temporal interaction between the matrix and embedded clauses. For a formal side, we adopt McCord’s (1978)’s pragmatic extended now theory, which was originally proposed to account for the meaning of the English present perfect, as below: (25) McCord (1978)’s pragmatic Extended Now theory a. English Preterite: a marker of prior events which is concluded and separate from the overall period of the present, ‘Extended now.’ [+then] b. English Perfect: a marker of prior events which is included within ‘Extended now.’ [-then] According to McCord (1978), the difference between English Preterite and English Perfect with respect to their relation to the ‘Extended Now’ comes from different pragmatic inferences Kim, p.23 4 DEFENDING A PRAGMATIC ANALYSIS associated with them, which unfortunately is not explicitly presented in McCord (1978)13. In this paper, it is proposed that in order to interpret the temporal indexical ‘cikum’ (now) properly, it is introduced in pragmatics an Extended Now interval XN(r) whose left boundary is contextually provided and whose right boundary is the utterance time indicated by the indexical. The configuration can be shown as in the picture below: (26) Extended now and DAR effects Extended now the utterance time Being in Tokyo 13 Adopting McCord’s (1978) idea, von Stechow (1999) and many others developed a semantic Extended Now theory, as follows: (i) a. von Stechow(1999)’s semantic Extended Now theory A perfect marker denotes an Extended Now interval XNP(r) whose right boundary is the reference time r and whose left edge is a contextually salient time. XNP(r) is a restrictor of an adverb of quantification b. Musan(2002)’s representation: VP at t. EXt [IN (EN (t))] [VP-(t)]= There is a time t properly included in the ExtendedNow such that c. Pancheva(2004)’s representation: [[PERFECT]] =λp<i,t>. λti. ∃ti [XN(tʹ′,t) & p(tʹ′)] where XN(tʹ′,t) iff t is a final subinterval of tʹ′ Kim, p.24 4 DEFENDING A PRAGMATIC ANALYSIS The inference strengthened Because the stretch of ‘cikum’ (now) is temporary and limited as the context allows, it cannot go far freely. So, when there is no temporal adverb that designates the exact temporal position of the matrix clause, the matrix event serves as the left boundary of Extended now, which requires the matrix event to be interpreted as close as the utterance time. When ‘cikum’ (now) is extended from the utterance to the extent that it can modify a recent past event, the semantic conflict is resolved so that the sentence in (16) turns out to be acceptable. A double access effect is just there because the inference is strengthened by ‘cikum’ (now). Kim, p.25 5 5 CONCLUSIONS AND REMAINING ISSUES Conclusions and remaining issues It is always worth clearing out interfering factors and resulting confusions that would keep us from making a progress in discussion. I tried to contribute to the discussion of double access sentences across languages on that aspect in the first part of this paper. I showed why it would be so important to eliminate a direct quotation interpretation from the sentences embedding a ‘-ko’ clause. On the basis, I claimed that Korean does not have a double access reading in a semantic sense, unlike English, but double access effects can arise from pragmatic inferences associated with temporality. Especially, the Korean sentences with double access effects can be accounted for pragmatically in terms of the strengthening effect and Extended ‘now’ for ‘cikum’ (now). One of the remaining issues is what makes a difference between complement clauses and relative ones with respect to the availability of a double access reading. General pragmatic inferences are available in a present tensed relative clause, but strong double access effects do not arise in a relative clause with ‘cikum’ (now) in (27) below: (27) Taro-nun [ e1 cikum Tokyo-ey Taro-TOP now Tokyo-in iss- ø –nun ] caki chinku1-eykey be-PRS-REL self friend-to cenwha-lul hay-ss-ta call-OCC do-PST-DCL Reading #1: ‘Taroi made a phone call to a friend of hisi who is in Tokyo now. Reading #2: ‘Taroi made a phone call to a friend of hisi who was in Tokyo *now. It is not clear why Reading #2 becomes unavailable in (27) while double access effects arise in a complement clause, as in (16). It calls for investigation. Kim, p.26 REFERENCES References Abush, D. 1997. Sequence of tense and temporal de re. Linguistics and Philosophy 20. Comrie, B. 1985. Tense. Cambridge, England: Cambridge University Press. Dowty, D. 1986. The eﬀects of aspectual class on the temporal structure of discourse: Semantics or pragmatics? Linguistics and Philosophy 9. Higginbotham, J. 2002a. Why is sequence of tense obligatory? Logical Form and Language, Oxford University. Gennari, S. 1999. Embedded present tense and attitude reports. In Proceedings of Semantics and Linguistic Theory (SALT) IX , pages 91–108. Ithaca, NY: CLCPublications. Giorgi, A. 2006. From temporal anchoring to long distance anaphors, Nat Lang Linguist Theory 24. Giorgi. A., & Pianise, F. 1997. Tense and Aspect: From semantics to morphosyntax, New York: Oxford University Press. Kratzer, A. 1998. More structural analogies between pronouns and tense. Proceedings of SALT VIII. CLC publication, Ithaca, NY. Kubota, Y., Lee, J. Smirnova, A and Tonhause, J. 2009. On the cross linguistic interpretation of embedded tenses, In Arndt Riester and Torgrim Solstad eds., Proceedings of Sinn und Bedeutung 13. Proceedings of SuB13, Stuttgart. Ogihara, T. 1996. Kluwer. Tense, Attitudes, and Scope, Studies in Linguistics and Philosophy 58. Ogihara, T. 1999b. Double-Access Sentences Generalized, Proceedings from Semantics and Linguistic Theory IX (SALT9), CLC Publications, Cornell University, pp. 224-236. Ogihara. T. 2003. The scope theory of tense and adnominal modifiers. Empirical and theoretical investigation into language, ed. by S. Chiba et al. Tokyo. Sharvit, Y. 2003. Embedded tense and universal grammar. Linguistic Inquiry 34(4):669–681. Schlenker, P. 2000. Propositional attitude and indexicality Ph.D. dissertation. MIT. Stowell, T. 2007. The syntactic expression of tense. Lingua 117. von. Stechow, A. 1995. On the proper treatment of tense, Proceedings of Semantics and linguistic theory 5. CLC publication. Cornell university, Ithaka, NY.
* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project