http://dx.doi.org/10.17959/sppm.2017.23.3.325

325

The phonology and semantics of expressive lengthening in Korean* <

Eon-Suk Ko (Chosun University) Ko, Eon-Suk. 2017. The phonology and semantics of expressive lengthening in Korean. Studies in Phonetics, Phonology and Morphology 23.3. 325-347. Expressive lengthening is a universal process with language-specific variation in its semantic and phonological implementation. This paper suggests that there are two types of expressive lengthening in Korean, i.e. intensification and diffusive lengthening. Intensification augments the core semantic attribute of a word, and its locus of lengthening is the last segment of the initial syllable. Diffusive lengthening, often applied to adjectives and adverbs derived from –ha verbs, yields the effect of spatial and/or temporal sustainment of the semantic property denoted by a word. The target locus for diffusive lengthening is often the stem-final syllable, with some exceptions. I suggest that the locus of expressive lengthening is determined by the semantic intent of the speaker rather than a morphological class. The data on intensification sheds fresh insight on the phonological representation of Korean laryngeal consonants. (Chosun University) Keywords: expressive lengthening, intensification, laryngeal, focus < <

1. Introduction <> 1.1 Preliminary remarks The meaning of a word with gradable semantic properties can be augmented or abated by various devices. Many words in Korean undergo dramatic lengthening of a phonologically conditioned segment in a syllable to intensify the degree or extent of the meaning conveyed by the word. In fact, this phenomenon is pervasive throughout world’s languages and has variously been called expressive lengthening (Bolinger *

This paper is a development of the ideas preliminarily laid out in Ko (2002). I thank the consultants who made judgments for a list of words to help me prepare the data, and three anonymous reviewers for their helpful comments. This study was supported by a research fund from Chosun University in 2016.

326

Eon-Suk Ko

1961) or emphatic lengthening (Kawahara and Braver 2014). Despite the many linguistic issues that the phenomenon can relate to, it did not draw much attention of generative linguistics. The purpose of this paper, therefore, is to document the phenomenon in Korean in the context of cross-linguistic universality and variation, and discuss its implication on phonological issues in Korean. Expressive lengthening is a phenomenon whereby a phonological modification of a segmental duration affects the meaning of a word. It is, therefore, a process taking place at the interface of phonology and lexical semantics (Potts 2005). Lengthening does not change the attributes, events, or other ontological properties denoted by the word per se, but rather has an effect of shifting the reference point in the scale of the property denoted by the word. Thus a word with a reference point close to the upper end of the scale (e.g. huge, tall) will receive an augmentation effect whereas a word denoting a reference point in the lower end of the scale will receive a diminution effect (e.g. tiny, small). This phenomenon is somewhat deviant from the design features of human language (Hockett 1973) in that it represents a non-arbitrary relation between sound and meaning. That is, by elongating a certain part of a word, the gradient semantic attribute of the word is augmented as a linear function of the lengthening (Kawahara and Braver 2014). Such iconic (de Saussure 1916) processing of semantic gradience seems to be a universally adopted strategy for expressing emphasis. Some languages might use the elongation process pervasively whereas others use periphrastic ways of intensifying the property of words more broadly by using modifiers such as very, really, awfully and extremely. Speakers might also emphasize the meaning of a word by repeating them1. 1

The repetition of a noun functions to boost the core meaning of the lexical item. See the English examples below (Julia Hirschberg, p.c.): (i) a. Did you date-date or did you just go out together a couple times? b. Is she giving a speech-speech or just saying a few words? A similar trend is recently being found in Korean: (ii) a. !"!"#$ eki eki hata ‘baby baby do’ (meaning ‘very cue’) b. %&%&#$ jəca jəca hata ‘woman woman do’ (meaning ‘very feminine’)

The phonology and semantics of expressive lengthening in Korean

327

1.2 Semantic and phonological constraints on expressive lengthening Not all words, and not even all adjectives, undergo expressive lengthening. In this section, I discuss some of the semantic and phonological constraints on this phenomenon, drawing data from English. Languages vary in the specific constraints on this phenomenon, but an in-depth look at the English data will help us have a sense of the range of such constraints. English speakers stretch the vowel of a stressed syllable to enrich the semantic attributes of a lexical item. Let us take some examples in (1), where words that undergo expressive lengthening are bold typed and underlined. (1) a. I saw a huge walrus at the zoo. b. He speaks so fast! c. Did you see that teeny little mouse in the lab? Speakers can lengthen the duration of the vowel in words such as huge, fast, tiny (or teeny), and big to boost the meaning of a property upwards from an assumed norm (Quirk et al. 1985)2. The right scalar property of an adjective or an adverb is certainly an important prerequisite for a word to be eligible for lengthening. In the literature of lexical semantics, the group of words whose core meaning involves a relation to a scalar concept are called gradable adjectives (Kennedy and McNally 2005). Adjectives such as dead and pregnant are not gradable by any means, and they cannot undergo expressive lengthening.

2

The correlation between the vowel duration and the magnitude of intensification (Kawahara and Braver 2014) seems to suggest that this could be some sort of an iconic semiotic system often called sound symbolism (Perniss et al. 2010). Interestingly, however, the examples (1b) and (1c) suggest that the process of lengthening is not entirely iconic because elongating the vowel in these cases shifts the reference point of the semantic scale in the opposite direction from what might be implicated by lengthening. That is, the elongation of the vowel in “fast” (1b) or “teeny” (1c) has an effect of slowing down the speaking rate or magnifying the overall duration of the word, respectively. This suggests that lengthening is not simply an onomatopoeic effect, but is a systematic linguistic process which functions to drive the reference point of a scalar word towards the extreme end of the denoted scale.

328

Eon-Suk Ko

(2) a. #He’s long been dead. b. #Isn’t it amazing that she finally got pregnant? Interestingly, however, not all gradable adjectives are eligible candidates for lengthening. As we see in (3), words such as possible and eligible are gradable but we cannot elongate the stressed vowel in these words to boost the extent of the meaning conveyed by these words. (3) a. #A possible solution to the problem is raising the tax rate. b. #He sounds like an eligible person. The ineligibility of the adjectives in (3) for expressive lengthening might be due to the phonological restriction on this phenomenon. That is, most words that undergo expressive lengthening are monosyllabic, although there are occasional exceptions as in (4). (4) a. Today is a fantastic day. b. That is amazing. In (4), it is possible to elongate the stressed syllable for intensification although they are multisyllabic. Note that multisyllabic gradable adjective such as fantastic or expensive can undergo intensification periphrastically with a modifier such as so, very, and such if they cannot undergo expressive lengthening. Gradable adjectives in (3), however, are not eligible for intensification even periphrastically. Note also that, although lengthening would not normally apply on the underlined words in (3), when it does apply, it induces an interpretation that evokes a nuanced interpretation entailing scalar implicature. (5) A: Are you coming tonight? B: #It is possible. In (5), for example, the speaker B’s answer with an elongated stressed syllable does not really enhance the possibility of B’s coming, but instead yields an implicature towards the opposite direction, i.e. B is likely not coming (p. c.

The phonology and semantics of expressive lengthening in Korean

329

Cassandre Creswell). Clearly, more research is required to view the full picture of the semantic and phonological restriction on this phenomenon. 1.3 Previous research In the literature of generative linguistics, little attention has been paid to the effects expressive lengthening yields on the meaning of a word. A notable exception is Woodbury (1987), who analyzed expressive lengthening in Central Alaskan Yupik Eskimo as “a meaningful phonological process”. Some of the data are presented in (6), where a plain form is followed by an intensified one. (6) Expressive lengthening in Central Alaskan Yupik Eskimo (Woodbury 1987) a. [quyá:kacá:γáγ̊luní] ‘he was really (-kacaγaγ̊-) grateful (quya-)’ a’ [quyák:acá:γáγ̊luní] ‘he was REALLY grateful’ b. [puqí:kapíxtúq] ‘he is really (-qapixtə-) smart (puqiγ-)’ b’ [puqík:apíxtúq] ‘he is REALLY smart’ c. [tamá:qáp:iáγ̊luní] ‘absolutely (-qapiaγ̊aγ̊-) all (tamaγ̊-) of us (-mta)’ c’ [tamáq:áp:iáγ̊luní] ‘ABSOLUTELY all of us’ In this language, lengthening applies to both consonants and vowels. For consonants, the lengthening takes place as a form of gemination when the consonant in the initial position of the postbase is preceded by a stressed open syllable, i.e. CV. Vowels undergo lengthening if it is part of a short stressed open syllable. These two rules are in counterfeeding relationship, thus the lengthening in vowel should be applied after the application of consonantal lengthening. According to Woodbury, the application of lengthening in Central Alaskan Yupik Eskimo heightens the emotional impact of the augmentative postbase, and conveys a particular attitude of the speaker. Expressive lengthening, often considered a paralingusitic phenomenon, sometimes sheds light on core phonological issues. The phenomenon in Finnish, first noted in Carlson (1978, cited in Hanson and Kiparsky 1996), provides an interesting piece of supportive evidence for a foot structure independently motivated in poetry. In an emphatic pronunciation, expletives are lengthened according to the pattern in (7). (7) Finnish (Hanson and Kiparsky 1996) a. perkele [per:kele]

330

Eon-Suk Ko

b. saatana [saa:tana] c. kamala [kama:la] d. vituttaa [vitu:t:taa] If the initial syllable is heavy, as in (7a,b), segments in its second mora are lengthened; if it is light, as in (7c,d), all moraic segments in the second syllable are lengthened. Hanson and Kiparsky (1996) take such a pattern as evidence for both initial H and initial LH forming a minimal foot, where expressive lengthening applies to the weak branch of the initial (head). Thus, the phenomenon of expressive lengthening serves as an important piece of evidence for understanding the metrical structure of Finnish. The phenomenon is further reported in several other languages. In Tamil (Christdas 1997), there is a hierarchy among the syllables of a word that can undergo the lengthening. That is, lengthening first applies to a syllable containing a long vowel, if available, and, otherwise, in the order of a syllable with a coda consonant, and an open syllable. Interestingly, this locus of lengthening is independent from the metrical head. In addition, Kâmv'iri, a language of Afghanistan, and also known as Bashgali, is reported to make use of expressive lengthening for various semantic coloring. Interestingly, it seems that the locus of lengthening varies depending on whether the lengthening indicates augmentation (a pre-accentual consonant) or diminution (accented vowel) (Strand 2017). Also, Reddy (1999) reports that length in Telugu works as a marker of emphatic stress in word initial syllable. As such, the phenomenon of intensification seems to be quite a general crosslinguistic phenomenon. In the rest of the paper, I will discuss the case of Korean with the following organization. In section 2, I will provide a detailed sematic and phonological description of expressive lengthening in Korean. In section 3, I will compare expressive lengthening with the phrase-level intonational focus. In section 4, I will conclude the discussion.

2. Expressive lengthening in Korean Many Korean words, in particular adjectives and adverbs, undergo lengthening to deliver a nuanced enhancement of the meaning of a word. There are several studies on the phenomenon (e.g. C-S Kim 1991, S-C Kim 2011) but very few from a generative approach.

The phonology and semantics of expressive lengthening in Korean

331

I argue that there are two types of expressive lengthening in Korean. The first involves enhancing the gradable core property of a word, similar to the case of English we discussed in 1.2. I will refer to this process as intensification. The second type involves sustainment of the word’s semantic property in the temporal or spatial dimension. I will refer to this type of lengthening as diffusive lengthening. (See C-S Kim 1991 for a similar account). In this section, I will describe these two types of expressive lengthening in more detail. Before presenting the data, I should note that expressive lengthening is an optional and probabilistic process. The data I describe in this paper are based on the judgments of eight consultants, including two Ph. D.’s, who have completed graduate-level courses in phonology. Four of them spoke the Seoul dialect, three Gyeongsang and one Jeonnam 3 . The survey was conducted over internet using Google Forms. A superset of examples in section 2 were presented in multiple sections. Consultants were asked whether or not and where expressive lengthening may apply in the word. Multiple answers were allowed for the locus of lengthening though the majority were answered with only one location. Data were included in the paper if at least half of them agreed with the author’s intuition for at least two thirds of the words in the data set. Specific rates of agreement are reported in places where necessary. 2.1 Intensification Expressive lengthening that functions to intensify the core semantic property of the word takes place on the initial syllable. More specifically, lengthening occurs on the last segment of the first syllable. Thus if the syllable of the target is open, it is the vowel that is lengthened (8).

3

I admit the limitations of the survey coming from the small number of respondents, lack of controlling dialectal differences as well as the need for instrumental inspections. However, the goal of the survey was more of a sanity check for the author’s intuition-level judgment than an in-depth behavior-level investigation, which is certainly worth taking up as a follow-up study. Let me also note that some of the respondents changed their judgment and became more lenient when I demonstrated a live example of lengthening in person. I did not change the reported rate based on these anecdotal interactions, but this suggests that the level of agreement could have been increased if stimuli were presented with audio in an appropriate semantic context.

332

Eon-Suk Ko

(8) Lengthening on the vowel of the initial open syllable a. !" [me:u] ‘very’ #" [kjə:u] ‘barely’ $% [kje:sok] ‘continuously’ &'( [a:muri] ‘regardless’ )* [nu:rən] ‘yellow’ b. +,[no:rakhe] ‘yellow’ ./[sə:rəpke] ‘sorrowfully’ 0/[tə:rəpke] ‘dirty’ 12[kwi:hake] ‘preciously’ 34,- [khə:tarakhe] ‘large’ The consultants had a robust intuition on the locus of lengthening for these sets of data, yielding almost a unanimous agreement. Note, however, that, unlike the monomorphemic adverbs in (8a), adverbs derived from adjectives with the –- –ke suffixation in (8b) may also elongate the second syllable as in +,- [nora:khe] ‘yellow’. My proposal in this paper is that the locus of lengthening is not arbitrary but is systematically governed by the semantic effects yielded by the elongation. Thus, the same word can undergo lengthening in two different locations depending on the speaker’s semantic intent. Such a semantic bifurcation of expressive lengthening will be discussed again in 2.3. If the initial syllable is closed, the coda consonant tends to undergo lengthening as in (9). (9) Lengthening on the coda of the initial syllable 567 [chuŋ:punhi] ‘enough’ 897 [saŋ:daŋhi] ‘ample’ :;7 [wan:cənhi] ‘completely’ 4 <=> [cuk:torok] ‘desperately’ ?@[mot:tøke] ‘maliciously’ ABC [pan:tɨsi] ‘necessarily’ 4

The locus of lengthening here was determined on an articulatory basis since the closure phase of /k/ and /t/ are indistinguishable acoustically. That is, I assume that speakers maintain the tongue at the velar position and do not move the tongue tip towards the alveolar ridge until the release phrase of the consonant cluster.

The phonology and semantics of expressive lengthening in Korean

DE/F(2GHI-

[ciŋ:kɨləpke] [jəŋ:lihake] [əm:chəŋnake]

333

‘extremely’ ‘smartly’ ‘humongously’

At least five respondents shared the intuition that the locus of lengthening is the coda of the first syllable in the words above5. The rate of agreement on this set of data was lower compared to the data in (8). However, the tendency to prefer the coda to the initial vowel as the locus of lengthening was clear6. An instrument-based behavioral study would certainly be beneficial to ascertain the probabilistically preferred locus of lengthening in such cases. Interestingly, if an open syllable is followed by an aspirate (10) or a tense stop or affricate (11), lengthening tends to take place on the closure phase of the laryngeal consonant rather than on the vowel of the initial syllable7. (10) Lengthening on the intervocalic aspirate stops and affricates C3J/sikhəməkhe/ K [sik:khəməkhe] ‘black’ LMN/echəropke/ K [et:tʃəropke]8 ‘pitiful’ OP/kəchilke/ K [kət:tʃilke] ‘tough’ 1Q& /kwichana/ K [kwit:tʃana] ‘bothersome’ RS/sechake/ K [set:tʃake] ‘strongly’

5

6

7

8

!"# [ma:lkakhe] ‘clearly’, which also ends in a closed initial syllable, was included in the list of survey but the respondents almost unanimously judged the initial vowel rather than the coda to be the locus of lengthening. The semantics behind such a deviant pattern is to be answered by semanticists. Previous research considered the locus of lengthening in words such as $%& [cəl:tero] ‘absolutely’ to be the stop closure of [t] (W. Huh 1985, cited in S-C Kim 2011). However, the responses of the consultants were split even between the vowel and the coda for this particular item, thus not included in the data. At least half of the respondents agreed with the author’s intuition on the data presented in (10) and (11). I should, however, acknowledge that there were items that were eliminated because the majority preferred the initial vowel to the coda as in '(# [i:pɨke] ‘pretty’. I transcribe the Korean affricates as alveolars, i.e. [c, ch, c’] (Ko 2013). For examples involving the lengthening of the aspirate affricate, however, I use the symbol for the postalveolar [tʃ] instead illustrate the elongation of the closure phase more clearly.

334

Eon-Suk Ko

(11) Lengthening on intervocalic tense stops and affricates CT/- /sik’ɨrəpke/ K [sik:k’ɨrəpke] ‘noisily’ UVW /sep’alkan/ K [sep:p’alkan] ‘red’ XV( /cep’ali/ K [cep:p’ali] ‘swiftly’ YZ- /ət’əke/ K [ət:t’əke] ‘so much’ Y[I /əc’ina/ K [ət:ʃ’ina] ‘extremely’ The status of laryngeal consonants as having one or two timing slots or roots has long been debated. The leading view is to consider the tense as a geminate but the aspirate as a singleton (e.g. Avery and Idsardi 2001). If a singleton, we would expect the elongation of the vowel to take place on the initial syllable as in (8). However, the data in (10) and (11) suggest that lengthening tends to occur on the laryngeal onset of the second syllable instead of the open initial vowel for both the aspirate and the tense consonants. What seems to be a deviance here actually turns out to be an important piece of evidence for the phonological representation of the laryngeal consonants. That is, the lengthening of the second syllable onset in (10) and (11) can be explained if the laryngeal consonants are analyzed as ambisyllabic, i.e. the consonant is initiated in the first syllable and released in the second syllable (Borowsky et al. 1984). In other words, the aspirate and tense consonants are linked to the first and the second syllable simultaneously, in which case the initial syllable is no longer open. Thus, the elongation takes place on the closure phase of the consonant linked to the initial syllable, similar to the lengthening of coda consonants in (9). The lengthening of intervocalic laryngeal consonants thus demonstrates a case where the aspirate and the tense stops pattern together as geminates. The data on the intensification of words containing the lenis and fortis fricatives /s/ and /s’/ sheds some further insight on issues in Korean laryngeal phonology. Korean /s/ is interesting in that it is phonetically aspirate, with spread glottis, but is phonologically lenis. It thus patterns with the aspirate consonants in phonetically motivated phenomena such as intervocalic voicing and boosting of the F0 on the following vowel, but with lenis consonants in phonologically conditioned phenomena such as germination in co-compounding (Ko 2003). In expressive lengthening, the locus of lengthening for words with an intervocalic lenis /s/ is the open initial vowel:

The phonology and semantics of expressive lengthening in Korean

335

(12) lengthening on the open initial vowel for /s/ Y\] [ə:sekhan] ‘awkward’ ^R] [mi:sehan] ‘minute’ __] [so:sohan] ‘trivial’ Consultants’ responses on these data were almost unanimous. The patterning of /s/ with other lenis consonants in (8) rather than aspirate consonants in (10) suggests that lengthening is a phonologically rather than phonetically conditioned phenomenon. We, therefore, conclude that lenis /s/ is not ambisyllabic but is solely affiliated to the second syllable. Turning to the data on fortis /s’/, we expect this tense consonant to be ambisyllabic like other tense consonants in (11) above. Unlike the intervocalic tense stop consonants, however, lengthening simply occurs throughout the frication phase for /s’/ (13). (13) Lengthening on the intervocalic tense fricative consonants `a] [os:’akhan] (*[ots:’akhan]) ‘scary’ bc- [pis:’ake] (*[pits:’ake]) ‘expensively’ Consultants’ response on these data were unanimous. Fricatives in Korean are normally neutralized to /t/ in the coda position (e.g. d /os/ ! [ot] ‘clothes’, ef /kas’-ko/ ! [kat-k’o] ‘went-CONJ’). The absence of neutralization of the fricative in (13), i.e. *[ots:’akhan] or *[pits:’ake], is expected as geminates are often immune to phonological processes applying to a singleton. Formally, this is an example of the Linking Constraint (Hayes 1986) in action, whereby the association lines of a rule is interpreted exhaustively. 2.2 Diffusive lengthening If the intensification discussed in §2.1 serves to boost the core semantic property of a word, there is another kind of semantic effect achieved by lengthening a different part of a word. For example, by lengthening the second syllable in gh2- [phjəna:n(h)ake] 9 ‘comfortable’, we achieve the effect that the state of being comfortable is temporally 9

/h/ in parenthesis, i.e. (h), indicates the application of the optional /h/-deletion (KimRenaud 1975).

336

Eon-Suk Ko

sustained rather than being momentary. To take another example, by lengthening the second syllable of the word such as ii [t’okt’o:k] ‘knock-knock’, we achieve the effect that the sound of knocking propagates through the space. In these cases, lengthening yields an effect of temporal or spatial sustainment of the semantic property, and I will call this type of lengthening diffusive lengthening. The majority of the words that undergo diffusive lengthening are associated with the suffix –hake ‘do-ADV’ as in jR2- [case-hake] ‘in detail’ or its variant –hi as in jR7 [case-hi] ‘in detail’, which originates from the light verb ha- ‘do’. In words involving ha- ‘do’ derivation, diffusive lengthening targets the stem-final vowel as in (14) and (15). (14) stem-final vowels undergoing diffusive lengthening in monosyllabic stems k2[mjo:hake] ‘curiously’ O2[kə:hake] ‘grandiosely’ l2[mə:ŋhake] ‘absent-mindedly’ m2[ci:n(h)ake]10 ‘dark, thickly’ (15) disyllabic stems with an open second syllable jR7 [case:hi] ‘in detail’ "&2[ua:hake] ‘graciously’ nn2[susu:hake] ‘modestly’ h op2[c ora:hake] ‘in a shabby form’ qr2[ɨsɨ:khake] ‘dim and remote’ h st2[kəlc’u:k ake] ‘thickly’ uv2[t’uljə:thake] ‘clearly’ Consultants’ agreement on the stem-final syllable as the locus of elongation in these data was almost unanimous. Thus, in an open second syllable as in jR7 [case:hi] ‘in detail’, the stem-final vowel /e/ undergoes lengthening. In a closed second syllable as in qr2- /ɨsɨ:k-hake/ ! [ɨsɨ:khake] ‘dim and remote’, resyllabification applies and the stem-final vowel undergoes lengthening. Again, however, this generalization is probabilistic, and I can imagine instances where lengthening applies on the closure phase of the intervocalic aspirate consonant /kh/ instead. 10

In /cinhake/, the coda /n/ may serve as the locus of lengthening if /h/ deletion does not apply.

The phonology and semantics of expressive lengthening in Korean

337

In words without the –ha ‘do’ suffixation as in Yw- [ətup-ke] ‘dark’, either the stem-final vowel /u/ or the stem-final coda consonant /p/ undergoes lengthening as in (16). (16) disyllabic stems with a closed second syllable Yw[ətu:pke] ~ [ətup:ke] xy[kanɨ:lke] ~ [kanɨl:ke] xz[kapjə:pke] ~ [kapjəp:ke] OP[kəchi:lke] ~ [kəchil:ke]

‘dark’ ‘thin’ ‘light’ ‘rough’

In words with the suffix –hake and stems containing three or more syllables, the locus of lengthening is reported to be the 3rd syllable (S-C Kim 2011). In fact, 4 to 5 of the eight respondents chose the third syllable as the locus of lengthening in (17). But it seems that the second syllable is also a possible locus of elongation as was indicated by 3 respondents. There was also 1 respondent who answered the lengthening occurs in the initial syllable. Thus there was a greater variation in speaker’s intuition with longer stems. (17) trisyllabic stems a. C{|2- sikhɨntuŋhake ‘uninterested’ h h [sik ɨntu:ŋhake~sik ɨ:ntuŋhake] b. }~2- k’amatɨkhake ‘far away’ [k’amatɨ:khake~ k’ama:tɨkhake] c. €n2- təpsulukhake ‘shaggy’ [təpsulu:khake ~ təpsu:lukhake] When the locus of lengthening is the third syllable, the target segment in the syllable also tends to vary between the nucleus vowel and the coda consonant. In contrast, when the same stem combines with the suffix –hi instead, the locus of lengthening is the second syllable as in }~7 [k’ama:tɨkhi] ‘far away’. One might postulate the possibility that the suffix –hake has some sort of morpho-phonological specification in the lexicon to draw the locus of the lengthening to the stem-final syllable. This seems to be supported by the alternation in [k’amatɨ:khake] ~ [k’ama:tɨkhi] ‘far away’, where the same stem lengthens different part of the word depending on the derivational suffix attached.

338

Eon-Suk Ko

However, not all adjective or adverbs with the derivational suffix –ha have the stem-final syllable as the locus of the lengthening. See the examples in (18) where the words undergo lengthening on the initial syllable despite the presence of the –ha suffix in the word. (18) words with –ha suffix undergoing lengthening on the initial syllable ii2[t’ok:t’okhake] #[t’okt’o:khake ‘clearly’ ‚ƒ2[ju:mjəŋhake] #[jumjə:ŋhake] ‘famously’ „…2[k’ɨm:c’ikhake] #[k’ɨmc’i:khake] ‘terribly’ †‡2[chəl:cəhake] #[chəlcə:hake] ‘thoroughly’ At least 6 out of the eight respondents agreed that lengthening applies on the initial syllable in the words above. Note also that not all words with the –hake or –hi suffix undergo expressive lengthening. The following are examples of words for whom speaker’s intuition about their eligibility for lengthening was not robust. (19) Words with –ha suffix that do not generally undergo lengthening ˆx‰2- [pulkapihake] ‘inevitably’ 4Š7 [taheŋ(h)i] ‘luckily’ Four out of the eight consultants judged the words in (19) to be ineligible to undergo lengthening. In previous research, it has been considered that adjectives derived with the suffixation of –ha undergoes expressive lengthening on the stem-final syllable (C-S Kim 1991). However, here we also see examples of initial lengthening (18) and the ones hard to lengthening (19). I, therefore, suggest that the explanation about whether or not, and where the lengthening takes place, should incorporate a more elaborate understanding of the semantic properties of the words rather than being attributed to a morphological class. The locus of the lengthening being limited to the initial syllable in words in (18) seems to suggest that the semantic properties of these words are more compatible with intensification rather than diffusive lengthening. In turn, the semantic properties of the majority of the words with –ha or –hi suffixes might be more compatible with diffusive lengthening than intensification. There is another morphological process, i.e. reduplication, which seems to be

The phonology and semantics of expressive lengthening in Korean

339

compatible with the idea that there might be a semantic bias for the class of words undergoing a certain morphological process. Most of the reduplicated words in Korean are onomatopoeic. Reduplicated onomatopoeic words are often already intensified through the morphological repetition of a single morpheme (cf. fn. 1). Yet, they often undergo expressive lengthening for further embellishment of semantic nuance. The locus of lengthening for reduplicated words is often the second syllable. The examples are as follows: (20) Reduplicated words undergoing lengthening on the second syllable x‹x‹ [kama:n kaman] ‘silently’ ŒŒ [əŋkɨ:m əŋkɨm] ‘crawling’ ŽŽ [toŋkɨ:l toŋkɨl] ‘round’ ‘‘ [cojo:ŋ cojoŋ] ‘quietly’ C’C’ [siwə:n siwən] ‘pleasantly’ One might hasten to conclude that the locus of lengthening in reduplication is fixed on the second syllable, but this is not true. Some reduplicated words such as “”“” [k’aŋchoŋ k’aŋchoŋ] ‘hopping’ can undergo lengthening on the initial as well as the second syllable (21). (21) Reduplicated words allowing lengthening on the initial syllable “”“” [k’aŋ:choŋ k’aŋchoŋ ~ k’aŋcho:ŋ k’aŋchoŋ] ‘hopping’ ŒŒ [əŋ:kɨm əŋkɨm] ‘crawling’ ŽŽ [toŋ:kɨl toŋkɨl] ‘round’ •• [cu:lcul] ‘dripping’ –– [sa:lsal] ‘softly’ —— [si:lsil] ‘unwittingly’ When the initial syllable is lengthened in “”“” [k’aŋ:choŋ k’aŋchoŋ] ‘hopping’, it has the effect of emphasizing the core meaning of the word “” [k’aŋchoŋ], which depicts the bouncy hopping motion of an animal such as a rabbit. In comparison, lengthening on the second syllable has the effect of depicting the bouncy hopping motion as being sustained for a certain temporal duration rather than being momentary. Thus, the different locus between intensification and diffusive lengthening might not be simply an effect of morphological process per se such as –ha suffixation or

340

Eon-Suk Ko

reduplication, but is governed by the semantic nature of the stems that combine with the suffix –hake or –hi, and the morphemes that can undergo reduplication. Admittedly, however, further research is required for a more complete explanation of the data presented in this section. 2.3 Semantic bifurcation between intensification and diffusive lengthening In previous research, it has been noted that certain lexical items such as ˜˜24 [s’ols’olhata] ‘pleased’ can lengthen either the first or the second syllable (B-G Lee 1986) though no particular explanation was provided for this observation. More examples of the cases with varying location of lengthening are in (22). (22) Words with –ha suffix undergoing both types of lengthening ™š2- /k’ek’ ɨthake/ K [k’ek:k’ ɨthake~ k’ek’ ɨ:thake] ‘clean’ ›œ2- /t’at’ɨthake/ K [t’at:t’ɨthake~ t’at’ɨ:thake] ‘warmly’ —7 /hwaksilhi/ K [hwak:sil(h)i~ hwaksi:l(h)i] ‘clearly’ 567 /chuŋpunhi/ K [chuŋ:pun(h)i~chuŋpu:n(h)i] ‘sufficiently’ Three to five consultants agreed that lengthening in either syllable is acceptable in these words11. When intensification is applied to ™š2- [k’ek’ ɨthake] ‘clean’ with lengthening on the initial syllable, the degree of cleanness is boosted, i.e. spotless. In comparison, when diffusive lengthening is applied on the second syllable of the same word, the effect achieved is somewhat different in that the cleanliness seems to diffuse spatially, reaching every corner of the space. Similarly, an application of intensification on the first syllable of ›œ2- [t’at’ɨthake] ‘warm’ achieves the effect of emphasizing the degree of warmness whereas an application of diffusive lengthening on the second syllable of the same word achieves the effect of the warmness temporally sustained rather than being momentary. My argument here, therefore, is that the variation of the locus in expressive lengthening is not arbitrary but governed by the speaker’s intent on the semantic effect yielded by lengthening.

11

For some of these words, more effort is required to think of a situation involving temporal or spatial sustainment. With appropriate semantic contexts, therefore, the acceptability rate could have gotten higher.

The phonology and semantics of expressive lengthening in Korean

341

3. Expressive lengthening and semantic focus As discussed in §1, expressive lengthening seems to be a cross-linguistic phenomenon. Nevertheless, this phenomenon has not attracted much attention of linguists. Part of the reason may have to do with the locus of expressive lengthening overlapping with that of informational focus in English. That is, both phenomena target the stressed syllable as the locus of acoustic manifestation, and this might have led many to confound the two phenomena. The claim I put forth in this paper is that the elongation of the vowel in expressive lengthening is to be differentiated from the long vowel duration in pitch accent as an acoustic manifestation of focus. The examples below illustrate this point: (23) a. A: So did you see that huge walrus? B: Yeah, it was a hu:ge one. b. A: What size ice cream did you buy? B: I bought a huge one. c. A: Did you buy a small ice cream cone? B: No, I bought a huge one.

(intensification) (informational focus) (contrastive focus)

The underlined huge in example (23a) cannot be a focused item according to the definition of focus in the sense of formal semantics (e.g. Rooth 1985, 1992) as it is neither an answer to a wh-question (23b) or a contrastive item (24c). Let us now turn to the case of Korean. Earlier, I mentioned that expressive lengthening generally occurs only in adjectives or adverbs but not in nouns or verbs. (24) Expressive lengthening not allowed on nouns or verbs *!"# $%&' ()*+ !:",- ./. kjəŋce-ka chimchetweə siləpljul-i cɨ:ŋkahako is’ta economy-NOM depressed unemployment-NOM increase-PROG ‘Due to an economic recess, unemployment rate is on an increase.’ Note, however, that nouns and verbs, as well as adjective and adverbs, can always be the focus of a sentence.

342

Eon-Suk Ko

(25) Nouns and verbs can be focused. !"

#$

%&'() *+,- ./'0 !"12- 342.

nophɨn kjəŋce səŋcaŋljuleto pulkuhako siləpljuli cɨŋkahes’tako hapnita high economic growth

despite

unemployment-NOM increased said

‘They say that despite high economic growth, unemployment rate is on an increase.’

When in focus, a noun or a verb will be in the initial position of a prosodic domain called Accentual Phrase (Jun 1993). The inapplicability of expressive lengthening on nouns or verbs, on the one hand, and their eligibility for focus, on the other, thus suggests that expressive lengthening is a phenomenon set apart from informational focus. When verbs or nouns are focused, their acoustic implementation does not solely rely on duration. Some of the most often discussed prosodic characteristics of focus in Korean are phrasing and higher pitch (Jun 1993). Phrasing refers to the phenomenon where a focused word initiates an Accentual Phrase, removing any other phrasal boundaries that might exist within the same Intonational Phrase. Thus, a focused word is pronounced with the following words in the same AP, which also serves as a domain for other phonological processes. (26) Phrasing in neutral and focused sentence (adapted from Jun 1993) a. {'0} {1 23}. əsə pap məkca quickly rice eat ‘let’s eat rice quickly’ b. {4} {+ 1} {5678} na i pap pərilejo I this rice throw ‘I am going to throw away this rice’ In (26a), the combination of 1 [pap] ‘rice’ and 23 [məkca] ‘eat’ is a highly predictable sequence whereas the combination of rice with the verb 5678 [pərilejo] ‘throw away’ in (26b) is unexpected. Objects and verbs are normally phrased into a single AP, but in the case of (26b), the verb is phrased into its own AP and realized with a focus prosody. Expressive lengthening and focus differ in the distribution of lengthening. Unlike the case of expressive lengthening, informational focus does not accompany a

The phonology and semantics of expressive lengthening in Korean

343

dramatic lengthening beyond the relatively small lengthening typically applied to an accent. (27) a. {9:# ;<} {#:$% =>?}. (exp. lengthening) t’alki-ka acu p’alkahke seŋkjəs’ne strawberry-NOM very red look ‘The strawberry looks really red.’ b. {'"} {@AB0} {&' @CD E'}. (focus) əce suphə-esə noran supakɨl pwas’ ə yesterday market-LOC yellow watermelon-ACC saw ‘I saw a yellow watermelon in the market.’ c. #{'"} {@AB0} {&:' @CD E'} (exp. lengthening) əce suphə-esə noran supakɨl pwas’ ə yesterday market-LOC yellow watermelon-ACC saw ‘I saw a yellow watermelon in the market.’ In (27a) the redness of the strawberry is intensified by elongating the coda [l] of the first syllable. In (27b), the yellowness of the watermelon is new information thus focused. In (27c), however, the application of expressive lengthening on [noran] ‘yellow’ is judged to be semantically awkward. Typically, if a watermelon is yellow, it is more likely to be new information worthy of focusing as in (27b) rather than intensifying the hue as in (27c). Thus the dramatic lengthening in (27c) beyond the level of lengthening found in pitch accent is deemed to be awkward. Going back to the issue of phrasing, expressive lengthening also tends to initiate an AP. Unlike the case of focus, however, the placement on an intensified word at the left edge of an AP does not seem to be a requirement. (28) {F44# &:(%} {G HI?} banana-ka norakhe cal ikəs’ne banana-NOM yellow well ripened ‘The yellow banana is ripened nicely’

(expressive lengthening)

Whether or not a word undergoing expressive lengthening necessarily triggers dephrasing is also not clear. Focused word is also realized with higher overall pitch. Words with an expressive

344

Eon-Suk Ko

lengthening might also accompany higher pitch, but little is known about the issue. It could be that the probability or magnitude of pitch raising might be different in the two phenomena. Clearly, these are issues which require a systematic investigation based on acoustic data.

4. Closing remarks This paper described expressive lengthening in Korean and provided a rough phonological and semantic analysis for the phenomenon. Although I attempted to generalize the systematic relationship between the phonological patterns of lengthening and their semantic effects, there are many remaining issues that need to be further addressed. For example, the intuition on the locus of the lengthening is quite probabilistic. Although the data is based on a consultation of speakers, it is likely that not every reader agrees with the data in this paper. The situation can be compared to the judgment of vowel length in traditional Korean. That is, even though most people agree that there is a long phonemic vowel in the standard variety of Korean, the intuition on vowel length substantially varies from individual to individual. However, failing to achieve a unanimous agreement on data should not discourages us from investigating the phenomenon. To the contrary, it might provide an opportunity to deepen our understanding of the variability, and I hope this paper could serve to invite a more elaborate study on this issue from a theoretical as well as a behavioral perspective. Another issue is the question of to what extent the locus of lengthening can be explained based on the semantic effects rendered by lengthening. For example, in the Jeonnam dialect, žŸ- [sanapke] ‘fierce’ and ' - [musəpke] ‘scary’ are similar in their meaning and their phonological composition is also very similar. Yet, žŸ- [sanapke] tends to undergo lengthening on the initial syllable whereas the locus of lengthening in ' - [musəpke] is the second syllable. Also, lengthening applied to the second syllable in [sanapke] would be more acceptable than lengthening applied on the first syllable of [musəpke]. This is likely to be a question of how expressive lengthening interacts with the metrical head (Ko 2013), thus a careful look at this variation will enrich our understanding of Korean word-level prosody. The phenomenon of expressive lengthening provides critical insight to some

The phonology and semantics of expressive lengthening in Korean

345

longstanding issues in Korean phonology. I touched on how the data on expressive lengthening sheds light on the phonological representation of Korean tense and aspirate consonants. In addition, the phenomenon could help us better understand the Korean word- and phrase-level prosody by investigating its interaction with metrical head and intonation though I could not go into these issues deeper in this paper. However, I hope the questions raised and the analyses provided in this paper have reinforced the value of the so-called paralinguistic data to reap insight for core theoretical issues. This paper thus seems to have put forth more questions than have answered. Future research to address the questions raised is expected to advance our understanding of the organization of our lexicon, and the architecture of the grammar allowing for the interface between phonology and semantics.

REFERENCES <¡•> AVERY, PETER and WILLIAM IDSARDI. 2001. Laryngeal dimensions, completion and enhancement. In Tracy A. Hall and Ursula Kleinhanz (eds.). Studies in Distinctive Feature Theory, 41-70. Berlin: Mouton de Gruyter. BOLINGER, DWIGHT. 1961. Generality, Gradience and the All-Or-None. The Hague: Mouton. BOROWSKY, TONI, JUNKO ITO and ARMIN MESTER. On the formal representation of ambisyllabicity: evidence from Danish. In Proceedings of NELS 14, 34-48. University of Massachusetts, Amherst: GLSA. CHRISTDAS, PRATHIMA. 1997. Syllable stress and prominence in Tamil. Ms, University of Pennsylvania. DE SAUSSURE, F. 1916. Cours de Linguistique Generale. Paris: Payot. HANSON, KRISTIN and PAUL KIPARSKY. 1996. A theory of metrical choice. Language 72, 287-335. HAYES, BRUCE. 1986. Inalterability in CV phonology. Language 62, 3221-251. HOCKETT, CHARLES. 1973. Man’s Place in Nature. New York: McGraw Hill. JUN, SUN-AH. 1993. The Phonetics and Phonology of Korean Prosody. PhD Dissertation, Ohio State University.

346

Eon-Suk Ko

KAWAHARA, SHIGETO and AARON BRAVER. 2014. Durational properties of emphatically lengthened consonants in Japanese. Journal of the International Phonetic Association, 44.3, 237-260. KENNEDY, CHRIS and LOUISE MCNALLY. 2005. Scale structure, degree modification, and the semantics of gradable predicates. Language 81, 345-381. KIM, CHANG-SEOB. 1991. ‘hata’ hyengyongsa eyseeuy phyohyencek cangeum (Expressive lengthening in ‘ha-ta’ adjectives). Kwukehakeuy Saylowun Insikkwa Cenkay (New Recognition and Development in Korean Linguistics), 744-762. Seoul: Mineumsa. KIM, SEON-CHEOL. 2011. Kwuke hyengyongsawa pusaeuy phyohencek cangeumhwa (Expressive lengthening of Korean adjectives and adverbs). Eoneohag 59, 5169. The Linguistic Society of Korea. KIM-RENAUD, YOUNG-KEY. 1973. Korean Consonantal Phonology, PhD Dissertation. University of Hawaii. ________________________. 1975. On h-Deletion in Korean. Journal of Korean Linguistics 3, 45-64. The Society of Korean Linguistics. KO, EON-SUK. 2002. The phonology and phonetics of word-level prosody and its interaction with phrase-level prosody: A study of Korean in comparison to English. PhD Dissertation. University of Pennsylvania. ___________. 2003. The laryngeal effect in Korean: Phonetics or phonology? In Jeroen van de Weijer, V. J. van Heuven and H. van der Hulst (eds.). The Phonological Spectrum, volume 1, Current Issues in Linguistic Theory series 233, 171-191. John Benjamins, Amsterdam and Philadelphia. __________. 2013. A metrical theory of Korean word prosody. The Linguistic Review 30.1, 79-115. KO, INSUNG. 2013. The articulation of Korean coronal obstruents: Data from heritage speakers and second language learners. PhD Dissertation. University of Hawaii. PERNISS, P., THOMPSON, R. and VIGLIOCCO, G. 2010. Iconicity as a general property of language: Evidence from spoken and signed languages. Frontiers in Psychology 1(DEC), 1–15. POTTS, CHRISTOPHER. 2005. Lexicalized intonational meaning. In Shigeto Kawahara (ed.). University of Massachusetts Occasional Papers 30 (UMOP 30), 129146. Amherst, MA: GLSA

The phonology and semantics of expressive lengthening in Korean

347

QUIRK, RANDOLPH, S. GREENBAUM, G. LEECH and J. SVARTVIK. 1985. A Comprehensive Grammar of the English Language. London: Longman. REDDY, K. NAGAMMA. 1999. Vowel and consonant length contrasts in Telugu. Proceedings of the ICPhS 1999, 2231-2234. ROOTH, MATS. 1985. Association with Focus. PhD Dissertation. University of Masschusetts. ___________. 1992. A theory of focus interpretation. Natural Language Semantics 1, 75-116. STRAND, RICHARD. 2017. The Sound System of Kâmviri. http://nuristan.info/Nuristani /Kamkata/Kom/KomLanguage/Lexicon/phon.html WOODBURY, ANTHONY. 1987. Meaningful phonological processes: A consideration of Central Alaskan Yupik Eskimo prosody. Language 63, 685-740.

Eon-Suk Ko Department of English Language and Literature Chosun University 390 Philmundaero, Dong-gu, Gwangju Korea 61452 e-mail: [email protected]

received: November 16, 2017 revised: December 15, 2017 accepted: December 22, 2017