How Metalinguistic Negation Differs from Descriptive Negation: ERP Evidence Chungmin Lee (clee@snu.ac.kr) Department of Linguistics, Gwanak-ro 1, Gwanak-gu Seoul, 151-742, Korea Abstract cannot intervene. See *I’m not HAPPY at all; I’m ECSTATIC. A metalinguistic use of negation cannot be This talk explores degree adverbial modifiers licensed exclusively by metalinguistic negation (MN), and compares replaced by a prefixal negation, either, as in *I’m unhappy; them with those licensed by descriptive negation (DN) such I’m ECSTATIC. Therefore, we cannot include Geurts’ as NPIs. It shows how MN-licensing is more marked than (1998) ‘propositional’ denial as one of the MN-like denials. DN-licensing in prosody and then attempts to show how Irony also has some sense of refutation, based on the anomalies arising from misplacing MN-licensed adverbs in general or mutual assumption, expectation or hope for ‘a DN-requiring short form negation sentences elicit the picnic day’ as a mental representation or thought, as in (4) approximate N400 but not the P600 in ERPs. This strongly suggests that such anomalies are meaning-related and tends to (‘echoic use’ (Sperber and Wilson 1986; Carston 1996). It is support the pragmatic ambiguity position by Horn than the negative, although expressed affirmatively. contextualist or relevance-theoretic approach. (4) It’s a lovely/fine/great day for a picnic! MN is an echoic rebuttal of whatever aspect of an Keywords: metalinguistic negation; descriptive negation; expression in a previous utterance to assert a rectifying markedness; prosody; ERPs; N400; pragmatic ambiguity; contextualist expression. Therefore, the speaker’s implicit inner alternative Q in Contrastive Focus can be assumed to 1. Markedness of MN Adverbials precede it, as in (3’) and its initial reply equivalent to MN can be assumed to be (5a), with the pair of expressions So far researchers have worked more on negative polarity connected by SN but (sino Spanish and sondern German), arguments and modifiers, which are licensed by descriptive and its bi-clausal manifestation with no but is (5b), whose negation (DN). The NPIs here simply reinforce the intonation is the L*(+H) L- H% of incredulity, distinct from falsification of the propositional contents. They are the Contrastive Topic intonation L+H* L- H% (Lee 2006, therefore emphatic in general (Potts 2010, Israel 2004). Constant 2012). Crosslinguistically and diachronically, NPIs have typically (3’) Are you HAPPY or ECSTATIC? developed from minimizers with ‘even’ (Lee 1993, Y. Lee (5) a. I’m not HAPPY but ECSTATIC. and Horn 1994, Lee 1999, Lee 2010 a.o.). b. I’m not HAPPY; I’m ECSTATIC. (1) a. amwu-to o-ci anh –ass-ta (Korean = K) anyone-even This paper explores degree modifiers licensed by MN, come-not-PAST-DEC and compares them with those licensed by DN and shows ‘Not anyone came.’ =b. ∼∃x (x: person’ (x)) [came (x)] how MN-licensing is more marked than DN-licensing in b. dare-mo ko-nakat-ta (Japanese = J) prosody first. The MN-licensed degree modifier A LITTLE c. shwei-ye mei-you lai (Chinese = C) in (6) forms a rising high peak of 254Hz after another peak (2) a. theibul-i tomwuci wumciki-ci anh-nun-ta (K)table – of not (MN) in Fig. 1. This is in sharp contrast with those NOM at all move-CI not-PRES-DEC ‘The table does not NPI-like minimizers licensed by DN in (7), one of which move at all.’ =b. ∼∃x (x: way’ (x)) [move’ (t)(in x)] forms the a bit/a little !H downstep with 211.7Hz, preceded b. teeburu wa mattaku ugoka-nai (J) by a high H* not. Because of the distinct and marked MN c. zhuo-zi gen-be budong (C) intonation for (6) and other cases, the rectification or MN, on the other hand, is used to reject, object to or rectify clarification clause may not follow; the conveyed meanings a previous utterance ‘on any grounds whatever’ ((Horn which may be called conventional implicatures, not 1985), (Ducrot 1972)). In (3), what is negated is not the cancellable, seem to be more assertive than ‘implicatures.’ proposition ‘I am happy’ in its reference or truth but the As a result, the purport of (6) is affirmative whereas that of degree of happiness expressed by the adjective ‘HAPPY’ in (7) is negative, although their written form is one and the the scale of happiness. The speaker objects to the way how same, creating ambiguity in English. it is put by the interpocutor. Typically, the expression (6) She is not A LITTLE upset. (She is VERY upset.) ‘HAPPY’ occurs or is assumed to occur in a previous (7) She is NOT a little upset. [even a little] (She is not upset utterance. Because the first clause in (3) does not falsify its at all, is quite composed.) Sentences for our phonetic positive proposition but object to the degree of happiness, experiments are modified from Bolinger (1972). the following clarification clause can assert a higher degree of happiness – ‘ECSTATIC’ without creating a contradiction, even though ecstatic entails happy in the Horn or entailment scale. (3) I’m not HAPPY; I’m ECSTATIC. (No contradiction arises) In this metalinguistic use of negation, a negative polarity item such as at all, which co-occurs with DN, as in (2), Fig 1 a-little-MN: a double of rising accent peaks 698 In Korean, the marked intonation of the MN-licensed negation of (11) can be assumed to be external (or cleft) S adverbial POTHONG ‘commonly,’ with a high pitch of negation in the Conrastive Focus construction. The MN 375Hz on the adverb, is sharply contrasted with the construction is crucially connected to the SN ‘but’ intonation of the adverb of the same form with the scalar coordination in C as in (12), anira in Korean, naku in J, ma marker –to ‘even’ [pothonguro–to] attached to function as in Vietnamese, etc. (Lee 2010). an NPI for DN (as in ‘--- not do well even commonly’), (10) a. Ta bu gao. #Ta feichang gao. (C) (cf. Wible et al 2000) which generates a comparatively low pitch of 295Hz on the 3sg NEG be tall 3sg be extremely tall adverb. The MN adverbial is prosodically marked. b. Ta bu rang wo qu. #Ta bi wo qu. 3sg NEG let 1sg go 3sg force 1sg go Now turn to the syntactic aspects of Korean negation to (11) a. Ta bu shi gao. Ta shi feichang gao. see how MN is syntactically marked as well. The MN- 3sg NEG FOC tall 3sg FOC extremely tall licensed stressed degree modifiers POTHONG and YEKAN, b. Ta bu shi rang wo qu. Ta shi bi wo qu. both ‘commonly,’ require external negation, as in (11a), 3sg NEG FOC let 1sg go 3sg FOC force 1sg go c. Ta bu hui rang wo qu. Ta hui bi wo qu. long form negation, as in (11b), or copula negation, as in 3sg NEG able let 1sg go 3sg able force1sg go (9c), but they cannot occur in a positive declarative S, as in (12) a. Wo bu shi xihuan ta, er-shi ai ta. (9d). In contrast, short form negation is typically for DN in I not like her but love her Korean. Therefore, if the MN-licensed stressed degree ‘I don’t LIKE CF her but LOVECF her.’ modifiers POTHONG or YEKAN occurs in short form b. Ta bu shi gao, ershi pang. [content also matters] negation sentence, the result is anomalous, as in (8). 3sg NEG FOC tall SN fat (8) a. Mia-ka POTHONG yeppu-n kes-i ani-i -ya. ‘(S)he is not tall but fat.’ M -Nom commonly pretty-PreN COMP-Nom not-Cop-Dec [extern-neg] Likewise in Chinese, YIBANde ‘commonly’ is an MN- ‘Mia is not COMMONLY pretty.’ ~> Mia is exceedingly licensed degree adverb and freely occurs in an MN pretty.’ sentence, as in (13a), conveying a higher degree expression. But it cannot occur in a positive sentence, as in (13b), nor in b. Mia-ka POTHONG/Yekan yeppu-ci anh-a a DN sentence, as in (14). Similarly in Japanese, the degree M -NOM commonly pretty-CI not-DEC (= a) [long-f modifier fuTSUU is typically licensed by MN to convey a neg]1 higher degree, as in (15). c. Mia-ka POTHONG(-i) ani-ya. [cop-neg]2 (13) a. Ta bu shi yibande piaoliang. (C) M -NOM common(NOM) not-DEC she MN commonly beautiful ‘Mia is not COMMON/ORDINARY.’ ~> Mia is extraordinary. ‘She is not COMMONLY beautiful .’ ~> (S)he is very beautiful. d. *Mia-ka POTHONG/Yekan yeppu-e. (with no negation) b. *Ta yibande piyaoliang.3 M-NOM commonly/relatively pretty-DEC (14) *Ta bu yibande piyaoliang . (C) (9) * Mia-ka POTHONG an yeppu -e. [short form neg] (s)he NEG commonly beautiful (K) (15) a. fuTSUU-no kawaisa ja-nai [--- ja naku honto-no kawaisa- M -NOM commonly not pretty –DEC da] (J) Cf. Mia-ka cenhye an yeppu -e [NPI] common –of prettiness not MN much-of prettiness at all ‘(She) is not COMMONLY pretty.’ ~> She is very pretty. ‘Mia is not pretty at all.’ b. fuTSUU janai [fuTSUU ja naku sugoi] In C, if bu ‘not’ co-occurs with an immediately following common (Adj) not MN extraordinary main predicate to negate, it is interpreted as DN, not ‘Not COMMON.’ (EXTRAORDINARY) allowing a rectifying clause, as in (10). If it is, however, followed by the Focus marker shi (from ‘be’) first and then Crosslinguistically in general, if ds is the echoic standard the main predicate, it forms a bi-clausal MN construction degree of the predicate, its metalinguistically negated with shi in the rectifying clause, as in (11). An overt (or utterance generates its positive proposition with a higher covert) modal may replace shi for MN-licensing. The degree d > ds of the same predicate. The epistemic agent is the speaker in a simple sentence, but it can be the subject in an embedded reported speech or complex attitude sentence. 1 YEKAN in Korean and YIBANde in Chinese are fixed as The syntactic form of external negation may favor MN both in Korean and English but external negation is not a sufficient condition for MN. MN-licensed modifiers whereas POTHONG(uro) in Korean An NPI in the complement clause is not happily licensed. and fuTSUU in Japanese may have their unstressed uses in positive utterances; pothong as an adverb is used in a (a) ??It is not the case that anyone came. (ExtN) 3 (b) ?? amu-to o-n key ani-ya (ExtN) (K) Sojung Im (pc) brought this to my attention. The string bu yibande in 2 This may be regarded as a variant of external negation, as property (14) was not found in the Peking University corpus and the anomaly of (14) negation. was confirmed by several native speakers of Chinese. 699 different quantificational meaning ‘usually’ and as a c. To measure individual subjects’ brainwave responses to each predicative noun pothong in K and fuTSUU in J they have stimulus, the waves by each stimulus were divided by the time their positive degree meaning of ‘common standard.’ 4 units at which each stimulus was presented. In Experiment 1 English has no counterpart of the MN-licensed echoic with Set A, the averages of the divided waveforms from all the electrodes were measured to get respective significant P-values. standard degree modifier ‘common,’ except the stressed By targeting the average of all subjects’ ERP responses, we MN-licensed below the middle degree modifier ‘A produced the final, grand average curve of ERP responses with LTTLE’/’A BIT,’ previously discussed. the N400, as shown in Fig 12. With those marked prosodic features and/or syntactic environments, MN-licensed degree modifiers can take place cross-linguistically, as opposed to DN-licensed ones. We will turn now to the next step: ERP studies. 2. ERPs for MN Adverbials We conducted ERP experiments with MN adverbials data twice. In the two experiments, we tried to see what happens when MN-requiring adverbials are placed in a short form negation (typically exclusively used for DN) in Korean, not Discussion of Experiment 1 on Written Visual Data properly in an external negation or a long form negation. What do the results of Experiment 1 say? The N400 ERP Naturally we presented well-formed MN sentences with MN results on Cz in Fig 12, the grand average of four subjects’ adverbials and ill-formed short form negation sentences brain-wave curves, reveal that some meaning-related with MN adverbials in contrast. In Experiment 1, written anomaly occurred from data Set A of the contrast between sentences were presented visually, whereas in Experiment 2, the well-formed external MN sentences with the MN- spoken sentences were presented auditorily. licensed degree adverbials and the ill-formed short form negation sentences with the same MN-licensed degree ERP Experiment 1 Data Set A: Well-formed External adverbials. In the Set A experiment, when a subject’s eyes Negation with STRESSED MN adverbial in red color vs. in the external negation condition reach the MN-licensed ill-formed Short Form Negation with STRESSED MN degree adverb marked in red, (s)he must expect an adjective adverbial all in red. 10 well-formed (with 5 POTHONG or adverb to be modified by the MN adverb and the sentences and 5 YEKAN sentences), 10 ill-formed complement clause ending, followed by external negation. sentences (with 5 POTHONG sentences and 5 YEKAN But in the short form negation condition, when the subject’s sentences), with 80 fillers, counterbalanced and presented to eyes reach the same MN-licensed degree adverb marked in each. red, (s)he must expect exactly the same external negation 요즘 │ 아이들은 │ 보통 │큰 게 │ 아니야 (or a long form negation) that can license the MN degree these days children commonly tall-Comp not-Cop-Dec adverb but in fact (s)he encounters the short form negation ‘It is not that these days children are COMMONLY tall.’ in the fourth column, followed by an adjective or adverb to Fig 2 well-formed: MN-licensed 보통 is in external be modified. (S)he would then be in a conflict between the MN adverb and the DN. An MN adverb cannot be licensed negation or interpreted by DN, which implies that MN and DN are 저 영화 │ 어제 │ 보통 │안 │ 졸렸어 distinctly used at least in pragmatic meaning. The adverb in red must have been charitably interpreted that movie yesterday commonly not boring as a stressed MN adverb. Similarly, even without red for the ‘It is not that that movie yesterday was commonly boring.’ adverb in the case of the intended ill-formed unstressed Fig 3 ill-formed: MN-licensed 보통 is in short form adverb condition in the external negation sentence in Set 1, negation because of the forceful MN bias of the external negation, participants seem to have interpreted the adverb in black Procedure, EEG Measurement and Analysis charitably as (stressed) MN-licensed degree adverb and that a. Subjects were presented with written sentences visually by E- Prime 2.0 our stimulus presentation software. seems to be why no results appeared. b. Ag/AgC1 electrodes and Brainamp were used;. VEOG and HEOG were employed with online filtering at 0.1Hz-70Hz, Experiment 2: ERP Analysis of MN Adverbials in sampling rate at 500Hz, and the impedance of electrodes under Spoken Sentences 10 kΩ. 4 Method See the degree expressions with a copula in a positive utterance, all unstressed: Subjects a. Pothong-i-ya (K) b. FuTSUU –desu (J) Comm 15 undergraduate subjects (4 females and 11 males) with on-COPULA-DEC Common-COPULA-DEC ‘That’s commo a mean age of 23.53 years (range: from 20 to 34, n (ordinary) (in degree/standard).’ undergraduate Seoul National University students) 700 participated for a cash payment of W25, 000 (about the mismatch (ill-formed) (S10 in the E-Prime) $25/hour). All were standard (Seoul-Gyeonggi) Korean condition and the match (well-formed) (S20 in the speakers, right-handed, not weak-sighted, with no history of E-Prime) condition were obtained. 5,000 times neurological disorders. These conditions were announced repeated; α=0.05, [IMG1]. beforehand in the internet recruitment and were met in the b. ANOVA: The following were examined: subjects’ written experiment protocol in the lab. . (i) subjects (random) x experiment manipulation Stimuli (repeated measures) (ii) electrodes (random) x In Experiment 2, recorded auditory sentences, unlike the experiment manipulation (repeated measures) written sentences in Experiment 1, were presented. The match (well-formed) condition with the stressed MN- An F1 repeated measures ANOVA with hemispheres (2) licensed degree adverb in external negation sentence vs. the x ROIs (electrodes) x manipulation is desirable but will be mismatch (ill-formed) condition here with the same stressed addressed in a later refinement with the total raw data. MN-licensed degree adverb in short form negation sentence is the same as in Experiment 1 (Set A). The only difference Discussion of Experiment 2 lies in that the MN adverb was in red in written sentences of external negation and short form negation in Experiment 1 As indicated, the N400 effect was elicited from the five but the same MN adverb was heard or auditory in recorded electrode sites near the center on both hemispheres sentences of external negation and short form negation in including C4 in Experiment 2 with the spoken sentences in Experiment 2. which MN-licensed degree adverbs placed in the matching In the match (well-formed) condition, 30 external external (MN) sentences vs. those placed in the negation sentences (15 with pothong ‘commonly’ and 15 mismatching short-form negation (DN) sentences. A certain with yekan ‘ordinarily’) were prepared, and in the mismatch difference with the results of Experiment 1 with the written (ill-formed) condition, 30 short from negation sentences (15 sentences lies in that the N400 effect was elicited from with pothong ‘commonly’ and 15 with yekan ‘ordinarily’), channel Cz (center) in Experiment 1. The difference may be 60 experimental sentences in total, were prepared, as well due to visual vs. auditory data. The same perspicuous as 80 filler sentences, totaling 140 sentences. The MN- negativity with the N400 effect in Experiment 2, however, licensed degree adverbs were all stressed in the spoken should be caused by the same meaning-related anomalies. sentences. Each subject heard all these types, but with each The N400 is ‘qualitatively distinct’ from the P600, which is sentence randomly assigned to one type. a reflection of syntactic anomalies such as number and The Well-formed Condition sentences and the Ill- gender agreement, phrase structure, verb subcategorization, formed Condition sentences were constructed in the same verb tense, constituent movement, case, and subject-verb fashion as done for Experiment 1. honorification agreement to be added in this work (see Osterthout et al (1999) for the distinction, stating that the Procedure, EEG Measurement and Analysis ERP brain responses to semantic/pragmatic anomalies In order to keep the participants attentive during the whole (selection restriction violation etc.) is dominated by a large session, they were told to press M if the sentence just heard increase in the N400 component and the response to a is natural and to press Z if not natural, at the end of each disparate set of syntactic anomalies is dominated by a large- sentence heard. From this test, we could distinguish a group amplitude positive shift. See Kutas et al (2011) for a survey of seven participants who made the wrong opposite of ERP N400 and meaning. responses 11 to 30 times from the rest who made less than six wrong responses. We eliminated the seven ill-behaved subjects from the analysis. Because a last minite E-Prime programming error (of placing a pair of anomalous sentences in a row) was found, one relevant subject was also eliminated and the total left for analysis was seven (7) subjects. Significant differences were detected at the five electrode sites near the center (particularly C4) with the N400 effect in Experiment 2. This is slightly different from Experiment 1, where the locus was exactly Cz (center) of the scalp. In order to decrease the noise effect, the ERP Fig 19: The N400 elicited at C4. signals were down sampled to 30Hz (and the +-200uv ones (30-40 out of 115~117) were eliminated). By employing the t-value of the T-Test as the Test 3. General Discussion of ERPs for MN Statistics in Permutation Test, we obtained the following: Adverbials (16) a. From the five electrode sites (C4, The markedness hierarchy of the three different types of S must be: CP2, CP5, P4, P7) significant differences between 701 (17) MN S> DN S> Affirmative S5 (DN = descriptive negation) Unlike the contradictory pairs with explicit or implicit negation involved in the past experiments, which often didn’t elicit any MN reveals phonetic and/or syntactic prominence in Contrastive immediate N400 effect and needed previous proper linguistic Focus (CF) in contrast to DN in English/Korean. Because the contexts for due expectations (Staab et al 2008), the distinction stressed POTHONG/YEKAN in Korean cannot appear in a positive between MN and DN is not necessarily context-dependent because sentence, as in (11d), researchers so far could not distinguish this of MN’s marked prosodic and/or lexical features that require MN from NPIs in Korean linguistics (Cho et al 2002; Whitman et al and the necessary conveyed implicature or following clarification 2004). But crucially they cannot co-occur in a negative sentence. A clause. long form negation in Korean can license either an NPI or an MN I give an independent support to my claim that pragmatic adverb but only separately. See (1a) with an NPI and (11b) with an meaning anomalies elicit the N400. Sakai’s (2013) ERP studies on MN adverb, both licensed by long form negation. Not the same Japanese honorific processing show: If you address a boy by negation can, however, license both NPI and MN-adverb at the “Kato-sama” honorifically, it is mismatched with the context and same time.6 Observe (18). elicits the N400 when in contrast with calling him “Kato.” (18) *amwu yeca–to POTHONG/YEKAN yeppu-ci anha Noh et al (2013) report in a rare valuable psycholinguistic eye- tracking experiment on MN that the subjects’ processing times at any woman-even commonly pretty-conn not(LF) the clarification clauses were not different between MN and DN in ‘Not any woman is commonly pretty.’ (Intended) their eye-tracking experiments, claiming that their results support Regarding the distinct functions between MN and DN, unlike the contextualist or relevance theory. As indicated, this theory has scholars such as Russell (1905) and Karttunen & Peters (1979), no separate use or pragmatic ‘meaning’ and therefore no who advocate the semantic ambiguity position, Horn (1985, 1989) ambiguity; MN is also truth-functional for them. But the Korean takes the pragmatic ambiguity position. Horn’s position is based on examples this study employed are dubious; the first “MN” example the unavailability of the implicated upper bound of weak scalar the authors provided is the following short form negation an ‘not’: predicates (e.g. ---we don’t like coffee, we love it), which he argues (18) (7) a. Yuna-nun ton-ul an pel-ess-e; ssule moa-ss-e. is pragmatic. It is a denying of the assertability or felicity of an Yuna-TC money-AC not make-PST-DC; rake in-PST-DC utterance or statement rather than negating the truth of a “Yuna didn’t make money; she raked in money.” proposition. His pragmatic ambiguity must be between two uses As we already explained, the short form negation an ‘not’ is MN and DN in his still one semantic negation monoguist position. typically used as DN in Korean. Then, what can we expect from Levinson’s (2000) criticism that even a semantically negated the bi-clausal construction in (18)? Sheer contradiction and it is. statement doesn’t have any implicatures is not tenable. Some more Native Korean speakers who are not biased will all agree. The echoic, nonveridical contexts may license MN uses, often English bi-causal MN construction is prosodically marked and rhetorically. I argue that the prosodically frozen MN uses of A cannot allow for the concessive But/but before the clarification LITTLE, POTHONG (K), and fuTSUU (J) and lexicalized MN clause. Therefore, if the combined use condition is met, MN can uses of YEKAN (K) and YIBANde (C) have their pragmatic involve even truth-conditional entailment cases and that’s why meaning associated with MN. On the other hand, the context- Horn’s definition has the expression ‘on any grounds whatever.’ driven or relevance-theoretic approach by Sperber & Wilson The following utterance: (1986), Carston (1988, 1998), Noveck et al (2007), Breheny et al. (19) I’m not HAPPY; (*but) I’m MISERABLE (2006) and Noh et al (2013) also as monoguists argue that there is is an MN case for Horn even though miserable entails ∼happy, no pragmatic ‘ambiguity’ or separate MN use/meaning and that not creating any contradiction. The first clause of (19) objects to scalar implicature is by the pragmatic enrichment of the scalar term the expression HAPPY and asserts the salient alternative involved. So, the literal form a or b as excluding a and b is due to clarification clause. 7 Compare it with (3), where not leads to a the contextual enrichment from inclusive (‘literal’) to exclusive, contradiction if read descriptively. This is not an MN for not by default for them. But consider ‘not a or b’ by DN becoming contextualists. Of course, there are quite a few researchers who do ‘not a and b’=’neither a nor b.’ We need MN to get a and b from a not adopt this claim and narrow down the range of MN cases. or b.’ To settle the debate, we need empirical, experimental Although this is still debatable, taking such “DN” examples evidence. occurring in external negation that typically licenses MN is not In the case of English and other intonation-based MN languages, convincing; for Horn, they are simply other cases of MN. This is prosody distinction elicits the MN vs. DN ambiguity (with the particularly true of pairs of expressives or emotion-charged frozen MN ∼ MN adverb intonation), as in (6) vs. (7). Here expressions such as wangtaypak ‘hit the jackpot’ vs. semantically weak degree adverbs like ‘a little’ were involved. In phwungpipaksan ‘break into fragments,’ occurring in MN- Korean and Japanese, stress (prosody) distinction (less in J) elicits licensing constructions in Korean. Either one of the two the same ambiguity but on the standard degree adverb such as expressives may be metalinguistically negated. The participants ‘commonly.’ Furthermore, some lexicalized MN-licensed degree might have skipped ‘non-sensible’ MNs quickly ‘with a fast effect’ adverbs developed in K and C, as in yekan ‘ordinarily’ and ibande (in their sensicality test, the mean sensicality of MNs was ‘commonly.’ The MN-licensed adverbs placed in short form significantly lower than that of DNs) and might have read sensible negation (DN) sentence in contrast to those in external negation MNs slower than DN ones with a slow effect, resulting in ‘no (MN) sentence elicited the N400. difference’ between conditions. As the reviewer supposed, this is rather in support of the ‘meaning’ approach than their contextualist 5 position. MN-licensing is most optimal in external negation and far Giora (2006) takes the symmetry position between (descriptive) less optimal in long form negation. The long form negation tends negation and affirmation. 6 A similar phenomenon in English has been indicated: an NPI cannot to lead to DN by default, although it can license MN. The intended appear in MN, as in (a). (Karttunen et al (1979:46 47). (a) *Chris didn’t manage to solve any of the problems---he managed 7 In German, the SN ‘but’ is employed for this situation: Ich bin nicht to solve all of them. (Horn 1989, 374). glueclich, sondern ungluecklich. 702 MN alternatives in contrast may become more easily non-sensible Horn, L. (1985). Metalinguistic Negation and Pragmatic Ambiguity, in long form negation than in external negation and they are Language 61, 121-74. doomed to be non-sensible in short form negation. Israel, Michael (1996). Polarity Sensitivity as Lexical Semantics. Linguistics & Philosophy 19, 619-666. 4. Concluding Remarks Kuno, S. and J. Whitman (2004). Licensing of Multiple Negative Polarity Items, in Studies in Korean Syntax and Semantics. Seoul: Pagijong. Kutas, Marta Kutas and Kara D. Federmeier (2011). Thirty Years and We made the distinction between two types of modifiers: those Counting: Finding Meaning in the N400 Component of the Event- licensed exclusively by MN and those by DN. The former are Related Brain Potential (ERP), Annual Review of Psychology 62:14.1– some MN-licensed degree adverbs, which are prosodically, 27. lexically and syntactically conditioned, and the latter are NPIs, Lee, Chungmin (1993). Frozen expressions and semantic representation, which reinforce negation unlike the former. The distinction Language Research 29: 301-326. suggests that MN and DN have distinct functions and uses, even if Lee, Chungmin 2006. Contrastive Topic/Focus and Polarity in Discourse, we assume that there is one single logical negation, departing from Where Semantics Meets Pragmatics (K. von Heusinger and K. Turner Russell (1905) and Karttunen et al (1979). Horn’s (1985, 1989) (eds)), CRiSPI 15, 381-420, Elsevier. pragmatic ambiguity position is in contrast to the context-driven or relevance-theoretic approach by Sperber et al (1986), Carston Lee, Young-Suk and Laurence Horn 1994. Any as indefinite plus even. MS. (1988, 1998), who deny that there is pragmatic ‘ambiguity’ and Yale University. claim that scalar implicature is by the pragmatic enrichment of the Levinson, S. (2000). Presumptive Meaning: The Theory of Generalized scalar term involved. How can we settle the debate? Conversational Implicature. MIT Press, Cambridge, MA. We are curious about possible empirical, experimental evidence Noh, Eun-Ju, Hyeree Choo, Sungryong Koh (2013). Processing that may shed light on the debate. A hypothesis can be: if the metalinguistic negation: Evidence from eye-tracking experiments, stressed MN-licensed degree adverb POTHONG/YEKAN co- Journal of Pragmatics 57: 1-18. occurs with short form negation (DN) in a sentence, the adverb Noveck, Ira and Dan Sperber (2007). The why and how of experimental will not be licensed by MN, which is absent, and as a result the pragmatics: The case of ‘scalar inferences,’ in Noel Roberts (ed.) sentence will be anomalous. But would it be meaning-based or Advances in Pragmatics, Palgrave. Osterhout, Lee and Janet Nicol (1999). On the distinctiveness, structure-based? With this in mind, we conducted two types of independence, and time course of the brain responses to syntactic and ERP experiments on MN for the fisrt time as far as we know: in semantic anomalies. Langauge and Cognitive Processes 14:3, 283-317. Experiment 1 (pilot), the pair of written sentences (with the Potts, Chris (2010). On the negativity of negation. In David Ludz and Nan stressed adverb in red) was presented and by targeting the average Li (eds.) Proceedings of SALT 30. of all the four subjects’ ERP responses, we produced the final, Recanati, Francois (1993). Direct Reference: From Language to Thought, grand average curve of ERP responses with the N400 over Cz, the Blackwell. central site. In Experiment 2, fifteen subjects participated. In the Russell, B. (1905). On denoting. Mind, 14. Blackwell. well-formed condition, 30 external negation sentences, with Sakai, H. (2013). Computation for Syntactic Dependency at Language pothong ‘commonly’ and yekan ‘ordinarily,’ and in the ill-formed Culture Interface: A View from ERP Studies on Japanese Honorific condition, 30 short form negation sentences, with stressed pothong Processing. Hiroshima U. Konkuk U Talk. and yekan, as well as 80 fillers, were presented all in recorded Sperber, D. D. Wilson. (2004). Relevance Theory, in G. Ward and L. Horn (eds) Handbook of Pragmatics. Oxford: Blackwell, 607-632. sound. The N400 effect ranging near 400ms from onset was Staab, Jenny, Thomas P. Urbach, and Marta Kutas (2008). Negation elicited from the five electrode sites near the center including C4 in Processing in Context Is Not (Always) Delayed, in Jamie Alexandre this experiment with the spoken sentences. Also, a significant (Ed.) Center for Research in Language, UCSD. negativity signal around 700ms was detected. This is an interesting Whitman, John and Susumo Kuno (2004). Licensing of Multiple Negative difference with the results of Experiment 1, where a rather typical Polarity Items. Studies in Korean Syntax and Semantics ed. by Susumu N400 effect was observed. However, nothing like the P600 was Kuno, 207-228. Seoul: Pagijong. detected. Wible, David and Eva Chen (2000). Linguistic Limits on Metalinguistic We need more data and analyses but we tentatively claim that Negation: Evidence from Mandarin and English, Language and Linguistics. the N400 effect was elicited from the two conditions and that if this turns out to be valid it shows that the anomaly is meaning- Acknowledgments related, though pragmatic. This tends to be in support of the I thank Sung-Eun Lee, and Sungryong Koh for their technical pragmatic ambiguity position than the contextualist non-ambiguity contributions to the ERP experiments and to Yoonjung Kang and Jeff approach. This is just the first step in the direction of researching Holliday for their contributions to the phonetic experiments. I am also brain responses to anomalies involving MN-licensed degree grateful to Larry Horn and Michael Israel for their comments on one of the modifiers. earliest versions and the CIL19 presentation. This work was supported by the National Research Foundation under (Excellent Scholar) Grant No. 100- 20090049 through Korean Government. References (Selected) Breheny, R., Katsos, N., Williams, J. (2006). Are scalar implicatures generated by default? Cognition 100 (3), 434-463. Burton-Roberts, Noël, (1989). On Horn’s dilemma: presupposition and negation. Journal of Linguistics 25, 95--125. Carston, Robyn. 1996. Metalinguistic Negation and Echoic Use, Journal of Pragmatics 25, 309-330. Cho, S. and Lee, H. (2002). Syntactic and Pragmatic Properties of NPI Yekan in Korean. In N. Akatsuka et al (eds.) Japanese/Korean Linguistics 10. CSLI. Ducrot, O. (1972). Dire et ne pas dire. Hermann. 703