A Pitch Analysis of Two Types of wh-Clauses in Korean · A Pitch Analysis of Two Types of wh-Clauses in Korean ... The issue of the disambiguation of wh-words or wh-phrases/clauses

154

A Pitch Analysis of Two Types of wh-Clauses in Korean

DUK-SOO PARK The University of Sydney

ABSTRACT In Korean, wh-words have dual functions: one as a wh-pronoun in a wh-question, and the other as an indefinite pronoun in a yes/no-question. In speech, however, such ambiguity does not occur as native speakers distinguish the two interpretations according to prosodic differences between the two types of questions. This phonetic experiment tries to uncover the salient prosodic features by which Koreans disambiguate the two types of questions. The main finding of the experiment is that there are two loci of contrast for the two types of questions, i.e. the boundary tones (or sentence final intonation) and the pitch contour of the wh-words themselves. Predominantly, the yes/no-question ends in a high rising intonation, while the wh-question may end in either a medium rising or a falling intonation. Compared with indefinite pronouns used in yes/no-questions, wh-pronouns in wh-questions generally have a higher pitch contour.

INTRODUCTION Wh-words in Korean have two functions: one as a wh-pronoun, as in a wh-question, and the other as an indefinite pronoun, as in a yes/no-question. For example, the wh-word [nugu] can function either as the wh-pronoun ‘who’ or as the indefinite pronoun ‘someone /anyone’, so that the sentence [nuguwanni] can be ambiguous, at least at the segmental level or in writing, between the interpretations “Did anyone come?” or “Who came?”. In actual speech, however, such ambiguity does not occur because native speakers are able to distinguish the two interpretations according to prosodic differences between the two types of questions. In this paper, I will investigate the salient prosodic features by which native speakers of Korean disambiguate the two types of questions. The wh-clauses, which include the following five wh-words, were analysed using an open-source speech analysis software, Praat.

(1) wh-word wh-pronoun indefinite pronoun [әnʤe] when sometime

[әdi] where somewhere/anywhere [nugu] who someone/anyone

[mwә] what something/anything [әt’әkhe] how somehow

According to Oh and Lee (1993) and Jun and Oh (1996), wh-phrases (wh-clauses in my terms) in Korean can be ambiguous at the segmental level in three ways due to their dual lexical functions or to the syntactic properties of a wh-phrase. They are (1) a wh-question, (2) a yes/no-question, and (3) a so-called incredulity (or echo) question. The incredulity question is an echo question expressing surprise concerning the referent of the wh-word, used by the listener when s/he cannot believe what the speaker says.

I exclude incredulity questions from my analysis because of their predictability. While prosodic features may disambiguate different types of questions with wh-words,

155

there are other things which also help listeners distinguish between them. Potential disambiguating factors include an extra morpheme, word, or phrase, or even the conversational situation. Incredulity questions, for example, are the only type of question which can be produced in the context of an unbelievable or surprising story. In other words, incredulity questions are easily predictable because of the situation in which they occur.

Of course, in reality, the other two types of question may also be accompanied by clues other than prosodic features. Often, an additional word or phrase in a wh-clause can be a clue. For example, having an extra word like /ʧom/ ‘a little/a bit’ (which may surface as [ʤom] as a result of intersonorant voicing) makes the clause [mwәlʤommәgilk’a] more likely to be a yes/no-question, “Shall we eat a little (of) something?”, than a wh-question, “What shall we eat a little (of)?”, as the latter is clearly an unusual expression.

PREVIOUS STUDIES The issue of the disambiguation of wh-words or wh-phrases/clauses has been discussed in terms of different prosodic notions in Korean phonology, such as ‘pitch accent (Chang 1973, Choe 1985), ‘boundary tones’ (Martin 1951, Choe 1985, Lee 1990), or ‘high versus low pitch’ (Cho 1990).

Choe (1985, 114), for example, maintains that if a wh-word is phonologically prominent compared with other elements of the sentence, that word is interpreted as a wh-pronoun. He does not, however, spell out what this ‘phonological prominence’ means in terms of its phonetic realization. Cho (1990, 56) also claims that a wh-word is intonationally focused when it functions as a wh-pronoun, thus forming one phonological phrase together with the following word, given that the following word does not belong to a separate phonological phrase. In the phonetic experiment reported in this paper, I look into the source of the notions phonological prominence or the intonational focus from point of view of the phonetic manifestation.

A similar issue has been noted in Tokyo Japanese (Maekawa 1991) and Kumamoto Japanese (Maekawa 1994). Using a perception test involving synthetic intonation, Maekawa (1991, 1994) finds that a phrase boundary marked by a falling tone between a wh-word and the following word is perceived as a yes/no-question, while the lack of a falling tone in the same position is perceived as a wh-question (Jun and Oh 1996, 38). In this paper, using natural rather than synthetic data, I endeavour to find if there is any similar tone pattern in Korean.

While there have been a number of phonological descriptions of the issue, Jun and Oh (1996) (henceforth J&O) is one of the first phonetic analyses on this topic in Korean. The findings of J&O are that the [three] types of questions are distinguished by the combination of boundary tones, accentual phrasing, pitch ranges, duration, and peak amplitude. The following are a summary of J&O’s findings under each notion:

Boundary tones. “. . . There was no single boundary tone type specific to one type of question: Yes/no-questions [and incredulity questions] were most often realized with High boundary tones (H%), although they sometimes appear with LH boundary tones (LH%) for all [four] speakers. The most common boundary tone for wh-questions was LH%, although H% and HL% were also observed, and one speaker used all four boundary tones (including HLH%) with roughly comparable frequencies. In addition to the different types of boundary tones, there seems to be a difference in the phonetic realisation of the high boundary tone. Within a given category of boundary tone, a high tone was in general significantly higher that the high tone in yes/no-questions, which was itself higher than that of wh-questions. . . .” (1996, 44) (See also Figures 3 and 4, and Table 2 on pp. 45-47).

156

Accentual phrasing. “Accentual phrasing was categorically different between the yes/no-questions and the other two types of questions. All speakers produced the wh-phrases as two separate Accentual Phrases—a wh-word and the following verb each forms one Accentual Phrase—in yes/no-questions, but as one Accentual Phrase in wh-questions [and incredulity questions].” (1996, 46-47) On the basis of Jun (1993)’s theory on six boundary tones in Seoul Korean (i.e. L%, H% LH%, HL%, LHL%, and HLH%) and another principle that an Accentual Phrase in Seoul Korean begins with a low tone (L) and ends in a high tone (H), accentual phrasing is achieved.1

Pitch range. J&O compares the pitch range and duration between incredulity- and wh-questions only, examining the pitch range of three points—i.e. the final H of the pre-wh-phrase, the initial L of the wh-phrase, and the peak of the wh-phrase. J&O states that “For all speakers, incredulity questions generally showed a larger pitch range than that of the other two types of questions, but this pitch range was more useful to distinguish incredulity questions from wh-questions rather than yes/no-questions from other two, because accentual phrasing is the same for both incredulity- and wh-questions, but not for yes/no-questions.” (1996, 50) So, the finding is not relevant to my experiment, which compares wh-questions with yes/no-questions.

Duration. “All speakers except Speaker O [i.e. three out of four] produced the peak F0 of the wh-phrase significantly later for incredulity questions than they did for wh-questions. This is due to the lengthening of the first syllable of the wh-word in incredulity questions relative to the same syllable in wh-questions. But incredulity questions are not longer overall than wh-questions, as Speakers Y an L [i.e. two out of four speakers] reduced the rest of the wh-phrase in incredulity questions.” (1996, 51)

Peak amplitude. J&O measures the peak amplitude (RMS, arbitrary scale) of the final syllable of interrogatives (“a question particle” in J&O) and wh-words for all three types of questions and compared them. Both the peak amplitude of wh-words and question particles were greatest in incredulity questions. “[For three out of four speakers], the peak amplitudes of wh-words were greater in wh-questions than those of yes/no-questions, while the amplitude of question particles [i.e. the last syllable of questions which relates to boundary tones] was variable across speakers.” (1996, 52)

For this fact, J&O assumes that the impression of greater phonological prominence for a wh-pronoun over an indefinite pronoun (Chang 1973, Choe 1985) seems to be based on the greater amplitude of wh-words in wh-questions than in yes/no-questions. (1996, 52)

In the experiment reported in this paper, I will examine whether some of these findings of J&O hold true. Note that this investigation is limited to the pitch, amplitude and duration of wh-words and the boundary tone in the two (rather than three) types of questions.

EXPERIMENT Method Subjects. Four native speakers of Seoul Korean in their twenties participated in the experiment: two female (Speakers A and B) and two male (Speakers C and D). All subjects are undergraduate students of the University of Sydney, Australia. Speaker A was born and raised in Seoul until she came to Australia at the age of 12. Since then she has been educated in Australia. Speaker B was born and raised in Seoul. She came to Australia at the age of 20. Speaker C was born and raised in Seoul, and left Korea at the age of 14, and since then he has been educated in Australia. Speaker D was also born and raised in Seoul until he left the country at the age of 16. All subjects reported that they speak

1 A schematic representation of tonal realizations of the basic tone patterns of an Accentual Phrase in Seoul Korean are L(HL)H for up to three-syllables, LHLH for four syllables and LH…LH for five or more syllables. See Jun (1993 & 1996) for details.

157

Korean to other Korean friends and acquaintances. Only Speaker C reported that he speaks Korean at home as his family is in Sydney. The other three subjects are international students without family in Australia.

Material. Ten brief dialogues were designed with five wh-words (see (1)) in two types of questions—wh-questions and yes/no questions. Each wh-word is followed by a verb, making up what I call a wh-clause. Some wh-clauses are preceded by a phrase to make the dialogues more natural (See Dialogue Pairs 1, 2 and 4): Sentence Pairs 1 and 2 begin with the vocative use of ‘Mr Park’ [paksәnsæŋnim], and Dialogue Pair 4 begins with an adverbial phrase ‘while I was out’ [næganagainnɨndoŋane]. Every question is followed by a relevant response to make it more natural, and each question ends in either [yo] or [k’a], the final syllable of the so-called Polite or Deferential style interrogative, respectively: Sentence Pairs 3 and 5 end in [k’a] and all others in [yo]. Notice in (2) that every second line is a phonemic transcription of each dialogue. For Dialogues 3a and 4a, there was an optional word or phrase in parentheses, which implies that the speaker may optionally choose to say it.

Dialogue Pairs

1a. /pak sәnsæŋnim әti kaseyo/ /ɨŋ әti ʧom kanta/ ‘Mr Park, are you going somewhere?’ ‘Yes, I am going somewhere.’

1b. /pak sәnsæŋnim әti kaseyo/ /uʧekuke ka/ ‘Mr Park, where are you going?’ ‘I am going to a post office.’

2a. /pak sәnsæŋnim әnʧe hankuk kaseyo/ /ɨŋ kɨlәnte wæ/ ‘Mr Park, are you going to Korea sometime?’ ‘Yes, I am, but why?’

2b. /pak sәnsæŋnim әnʧe hankuk kaseyo/ /taɨmʧu suyoillal kanta/ ‘Mr Park, when are you going to Korea?’ ‘Next Wednesday, I’m going.’

3a. /mwәl (ʧom) mәkɨlk’a/ /kɨlæ pækopʱɨta mwәl ʧom mәkʧa/ ‘Shall we eat (a bit of) something?’ Ye, I am hungry. Let’s eat a bit of something.

3b. /mwәl mәkɨlk’a/ /nan ʧ’aʧaŋmәn mәkɨllæ nәn/ ‘What shall we eat?’ ‘I will eat Tchajangmyŏn, how about you?’

4a. /næka naka is’nɨtoŋane nuka was’әs’ә/ /ɨŋ nuka was’әs’ә/ ‘While I was out, did anyone come?’ ‘Yes, somebody came.’

4b. /nuka was’әs’ә/ /kimsutoŋilan salami was’әs’ә/ ‘Who came?’ ‘A person called Kim Sudong came.’

5a. /әt’әhke ʧal twæs’ɨlk’a/ /ɨŋ әt’әhke ʧal twæs’ɨlkәya/ ‘Do you think it went well somehow?’ ‘Yep, I think it went well somehow.”

5b. /әt’әhke twæs’ɨlk’a/ /kim sәnsæŋnimi towa ʧusyәsә ʧal twæs’ɨlkәya/ ‘How do you think it went?’ ‘I believe it went well since Mr Kim helped.”

Procedure. The dialogues in (2), written on a piece of A4-size paper, were presented to each subject with a brief explanation of two functions of wh-words in Korean. After one practice session, the dialogues were recorded in a digital recorder, MicroTrack 24/96 Mobile Digital Recorder, with a Stereo Electret Microphone, although the recording was done in mono. Each subject read the question part, and the author read the responding part of each dialogue. The digitalised audio file was analysed by using Praat, a speech analysis software developed by Paul Boersma and David Weenink, Professors of Phonetic Sciences at the Institute of Phonetic Studies, University of Amsterdam, The Netherlands. The

158

version used was Praat 5.0.07 for Windows. From each speaker’s long audio (MP3) file, ten wh-clauses were extracted and analysed by Praat. The length of long audio files for each speaker was between 45 to 101 seconds, and the length of wh-clauses was around 1 second.

The pitch contour of each wh-clause was drawn by the software, and then measurements of the mean pitch of the wh-word and the sentence-final syllable, as well as the pitch range were taken. The mean intensity in dB (µ) and the duration (in seconds) of wh-words were also measured to see if they could be disambiguating factors for the two types of questions.

RESULTS AND DISCUSSION From each pitch contour, the mean pitch (in Hz) and the pitch range (of the minimum and maximum pitch points) were measured for wh-words and the sentence-final syllable. The former is to see if there is prominent pitch for one type of wh-words, and the latter for the boundary tone distinction. For both wh-words and boundary tones, then, the increase (or decrease) of pitch was measured to see how much pitch rise occurs in terms of intonation.

Table 1 is the mean pitch data for wh-words, and Table 2 is that of the sentence-final syllable, and a description and interpretation of these data follow each table. The full data from which Tables 1 and 2 are drawn is attached at the end of this paper as Appendix 1.

Table 1: The pitch pattern of wh-words (in Hz)

Pitch in Wh-words As shown in Table 1, the mean pitch of wh-pronouns is significantly higher than that of indefinite pronouns: 43.4 Hz higher by Speaker A, 18.7 Hz higher by Speaker B, 13.2 Hz by Speaker C, and 13.4 Hz by Speaker D. While the two male speakers show a similar gap (of around 13 Hz) between the mean pitch of the two types of wh-words, the gaps shown by the two female speakers were quite different. This is related to the pitch ranges that the two female speakers have in their speech. The pitch range of Speaker A was around 450 Hz (from the high 100s Hz to the high 500s Hz), while that of Speaker B was slightly more than 200 Hz (from low 100s Hz to mid 300s Hz).

Yes/No-Question Wh-Question

Indefinite pronoun Wh-pronoun

gend

er

spea

ker

mean

rise

min max mean rise min max

A 222.4 24.3 228.4 252.0 265.8 71.8 241.4 313.2 f

B 235.9 15.2 237.4 252.6 254.6 44.0 247.8 291.8

C 129.9 15.5 123 138.5 143.1 60.5 120.8 181.3 m

D 114.0 1.7 122 123.7 127.4 41.2 106.8 148.0

159

As we compare the pitch rise from the minimum to the maximum points in wh-words, again, we consistently see greater pitch values in wh-pronouns both in the peak points as well as the rise range. The difference between the peak points of the two types of wh-words was 61.2 Hz by Speaker A, 39.2 Hz by Speaker B, 42.8 by Speaker C, and 24.3 Hz by Speaker D. Here, the minimum pitch value can be lower in wh-pronouns as this increases the rise range, as shown in the cases of Speakers G and I.

In addition to the difference in the mean pitch (in Hz) of wh-words, distinctively different pitch contour (or intonation) was found between the two types of questions. Figure 1 shows the pitch contour diagrams of the two tokens of [әdigaseyo] by Speaker A: ‘Are you going somewhere?’ (above) and ‘Where are you going?’ (below). Figure 2 shows the extracted diagrams of the wh-word portions, which are marked by two vertical dotted lines in Figure 1.

Figure 1: Pitch contour of the two types of questions [әdigaseyo] by Speaker A:

160

Figure 2: Extracted pitch contours of ‘somewhere’ (left) and ‘where’ (right)

by Speaker A

Figure 1 shows the most typical pitch patterns of the two types of questions. A similar pattern was shown in the speech of all other speakers and in many other pairs of wh-words, except for those with a falling boundary tone for wh-questions. A falling intonation was observed in the wh-questions of Dialogue Pairs 3 and 5, ie. ‘What shall we eat?’ and ‘How do you think it went?’ For Speakers B, C and D, the rising intonation is not as high as Speaker A, but the highest point of the boundary tone in yes/no-questions was always higher than that of wh-questions. Boundary Tones This experiment confirms J&O’s claim that the boundary tone of yes/no-questions is significantly higher than that of wh-questions. In English, a rising boundary tone (or intonation) is a distinctive feature for yes/no-questions, while a falling intonation is for wh-questions. One may assume, therefore, that the Korean language may have the same or similar intonation patterns for each of the two types of questions. This experiment shows that while the yes/no-questions occur with a high rising tone all the time, a falling tone does not always occur for the wh-questions. In fact, among other patterns, many wh-questions had a rising intonation, while the pitch range was much lower than that of yes/no-questions. I would call this a medium rising tone in wh-questions in comparison with a high rising tone in yes/no-questions.

At the end of yes/no-questions, all speakers used predominantly high rising tones. Out of twenty yes/no-questions, only two were with other than a high rising tone: one with a small dip before the rising tone (henceforth indicated by √-tone in Appendix 1) (i.e. 142-120-210 Hz; See C5 in Appendix 1) and the other with a more or less flat tone (indicated by – in Appendix 1) at around mid/high 90s Hz (i.e. 95-98 Hz; See D2 in Appendix 1). Nevertheless, even for these speakers the boundary tones’ pitch values in yes/no-questions are much higher than those wh-questions.

On the other hand, wh-questions ended with various boundary tones, 14 medium rising tones, 4 falling tones, 1 √-tone, and 1 up-and-down tone (indicated by Λ-tone). The rising tones of wh-questions were much lower than those of yes/no-questions in terms of their pitch range and the peak point. The mean peak of boundary tones in wh-questions was very similar to the mean in the wh-pronouns: i.e. Mid 200s Hz for both female speakers and low and mid 100s for male speakers.

161

Let us now consider the interpretation of individual speaker’s data. Female Speaker A’s speech exhibited the most distinctive boundary tones for the two question types: The average boundary tones for yes/no-questions ranged from 261 to 517 Hz (with a mean of 393.6 Hz and a mean rise of 256 Hz), while those of wh-questions ranged from 227.4 to 292.8 Hz (with a mean of 257.38 Hz and with a mean rise of 65.4 Hz). Speaker A, like the other three speakers, always used a high rising intonation for yes/no-questions, reaching up to the mid 500s Hz, but predominantly a medium rising intonation up to mid/high 300s Hz or a falling intonation for two wh-questions (i.e. from 287 to 186 Hz, and from 254 to196 Hz in /mwәl mәkɨlk’a/ ‘What shall we eat?’ and /әt’әhke twæs’ɨlk’a/ ‘How do you think it went?’). (See the wh-question column of A3 and A5 in Appendix 1.)

Female Speaker B’s data also confirms a higher boundary tone in yes/no-questions than in wh-questions, with a much narrower range of somewhere in between that of female Speaker A and those of the male speakers: The mean pitch range of her yes/no-questions was from 260.6 to 341.4 Hz (with a rise of 80.8 Hz), and the mean pitch range of her wh-questions was from 223.8 to 250 Hz (with a a rise of 26.2 Hz). Notice that, for the wh-questions, the boundary tone’s pitch increase is 54 Hz for Speaker B, while that of the other female, Speaker A, is 190 Hz. This is, as a matter of course, due to the different pitch range of each individual. Notice that the valley points of boundary tones were similar to each other, around the low and middle 200s Hz, but their peak points were quite different: Speaker A’s peak points were much higher than those of Speaker B, i.e. up to the mid/high 500s Hz vs the mid 300s Hz.

While the speech of the male subjects also shows greater mean pitch values (such as the mean pitch and the pitch range) in the boundary tones of yes/no-questions than those of wh-questions in general, there were three individual cases where the pitch values are the other way around—i.e. the boundary tone values of yes/no-questions are lower than those of wh-questions (See C2, D2, and D5 in Appendix 1). My interpretation for this relates to the fact that there are two loci to achieve the contrast—i.e. the pitch of the wh-word and the pitch of the boundary tone—but that, in order to disambiguate the two types of questions, speakers can use only one locus of contrast. In these three cases, the distinction is in the pitch value of wh-words rather than that of boundary tone.

In fact, quite the opposite phenomenon was also found in the speech of Speaker D: In the Dialogue pair 1a and 1b by Speaker D, the wh-pronoun ‘where’ has a lower pitch value than that of the indefinite pronoun ‘somewhere’, except for the peak point (146 Hz vs 139 Hz). In this case, rather than the pitch of the wh-words, the boundary tones function as the distinguishing factor. These two cases reveal that native speakers do not always make use of the two loci of contrast. In spite of having two possible loci of contrast, the use of only one seems to be sufficient. In the cases of C2, D2 and D5, while the boundary tones are not sufficiently contrastive, the wh-words are. On the other hand, in the case of D1, when the pitch contours of wh-words are lower than those of indefinite pronoun counterparts (opposite to the general pattern found in the experiment), the boundary tones play a more decisive role.2

2 In fact, I found the same phenomenon in my own speech (which is not included in this experiment), where the pitch of a wh-pronoun is often lower than that of its indefinite pronoun counterpart, meaning that the distinction depends on the boundary tones. For the cases like this, in a phonological description, we may need to say that the tonal distinction in boundary tone is a more decisive factor than the pitch contour of wh-words.

162

Table 2: Boundary tones in the two types of questions (in Hz)

Intensity and Duration As mentioned previously, J&O (1996, 52) assumes that the phonological notion described as phonological prominence by Chang (1973) and Choe (1985) might be due to the greater amplitude of wh-pronouns from the measurement of RMS (arbitrary scale). My experiment, however, shows that the mean amplitudes of the two types of wh-words are not different from each other, and the data was inconsistent among subjects too. The inconsistency is marked by > or < (greater or less than) in Table 3 below. Contrary to the claim by J&O, for Speaker B, both the mean amplitude as well as the mean duration of wh-pronouns was shorter than that of yes/no-questions. Even for the overall average figure with 0.37 dB (µ) difference, I strongly doubt that this much difference in amplitude could be a distinctive feature for the purpose of disambiguation of the two types of questions. I found in some cases both pitch and intensity were greater in wh-pronouns than in indefinite pronouns, but the difference in intensity was minor, while that in pitch was clearly noticeable. I also found that the duration of wh-words may not be a significant feature either, as the overall average difference was merely 0.0257 second. See the data in Table 3 and Appendix 2 for details.

Table 3: Intensity and Duration of Wh-words

Yes/No-Question Wh-Question

S-final syllable S-final syllable

spea

ker

mean rise min max mean rise min max

A 393.6 259.0 261.0 517.0 257.4 65.4 227.4 292.8

B 308.3 80.8 260.6 341.4 257.7 26.2 223.8 250

C 169.5 93.8 125 218.8 153.0 103.6 104.8 208.4

D 135.9 20.0 124.8 144.8 121.9 6.3 116 122.3

Intensity in dB (µ) Duration in second

spea

ker

Indefinite pronoun Wh-pronoun Indefinite

pronoun Wh-pronoun

A 66.27 < 68.16 0.1973 < 0.2793

B 64.23 > 59.84 0.3301 > 0.2838

C 66.68 < 67.52 0.3074 > 0.3043

D 62.67 > 65.79 0.2653 < 0.3352

Overall avrage 64.96 65.33 0.2750 0.3007

163

CONCLUSION The predominant pattern that emerges from this experiment is that Korean speakers distinguish the two types of questions—the wh-question and the yes/no-question—by the tonal contrast in two loci: the boundary tone (or the sentence-final intonation) as well as the pitch contour of the wh-words. With regards to the boundary tone, the yes/no-question ends in a high rising intonation, while the wh-question may end in either a medium rising or a falling intonation. With regards to the wh-words, these generally have greater pitch contours in wh-questions than in yes-no questions, in terms of their mean pitch, peak pitch point and rise range. It is, however, worthy of note that there are some speakers who sometimes rely on a contrast in only one of these two loci: either the boundary tone or the pitch contour of the wh-words. This experiment also confirms that phonological notions like phonological prominence or pitch accent correlate with the higher pitch in wh-pronouns than in indefinite pronouns. The experiment also shows that the notion of phonological prominence does not correlate with greater amplitude or duration in wh-pronouns than in indefinite pronouns.

REFERENCES Cho, Y. Y. 1990. Syntax and phrasing in Korean. In The phonology-syntax connection, eds. S. Inkelas & D.

Zec, 47-62. Chicago: University of Chicago Press.

Choe, J.-W. 1985. Pitch-accent and q/wh words in Korean. In Harvard studies in Korean linguistics, eds. S. Kuno et al., 113-123, Cambridge: Harvard University.

Gussenhoven, C. 2004, The phonology of tone and intonation. Cambridge: Cambridge University Press.

Jun, S.-A & Oh, M. 1996. A prosodic analysis of three types of wh-phrases in Korean. Language and Speech 39 (1): 37-61.

Jun, S.-A. 1993. The phonetics and phonology of Korean prosody. Unpublished PhD dissertation, The Ohio State University. [published in 1996 by New York: Garland].

Kang, H.-S. 1996. Acoustic and intonational correlates of the informational status of referring expressions in Seoul Korean. Language and Speech 39 (4): 307-430.

Ladd, D. R. 1996. Intonational phonology. Cambridge: Cambridge University Press.

Ladefoged, P. and I. Maddieson. 1996. The sounds of the world’s languages. Oxford: Blackwell Publishers Ltd.

Ladefoged, P. 1962. Elements of acoustic phonetics. Chicago & London: University of Chicago Press.

Maekawa, K. 1991. Perception of intonational characteristics of wh and non-wh questions in Tokyo Japanese, Proceedings of the XIIth International Congress of Phonetics Science, 202-205. Aix-en-Provence: France.

Maekawa, K. 1994. Intonational structure of Kumamoto Japanese: A perceptual validation, Proceedings of 1994 International Conference on Spoken Language Processing: Vol.1, 119-122.

Oh, M. & Lee, H.-Y. 1993. Teaching intonation of Korean as a foreign language. Paper presented at IAKL, Seoul, Korea.

Yang, B.-G. 2003. Praat-rŭl iyonghan ŭmsŏngbunsŏg-ŭi iron-gwa silje [The theory and practice of speech analysis by the use of Praat], Seoul: Mansuchulpansa.

164

165

Appendix 2: Intensity and Duration of wh-words

Contact Address Dr Duk-Soo Park, Senior Lecturer, Department of Korean Studies, School of Languages and Cultures (A18), The University of Sydney, NSW2006 Australia. [email protected]

Intensity in dB (µ) Duration in second ind-pron wh-pron ind-pron wh-pron A1 where

68.82 67.78 0.2369 0.2924

A2 when 64.97 70.25 0.2307 0.2778 A3 what 67.31 71.41 0.2109 0.1772 A4 who 67.48 67.76 0.1208 0.2610 A5 how 62.79 63.61 0.1870 0.3883 mean of A

66.27 68.16 0.1973 0.2793

B1 65.35 63.40 0.2675 0.1886 B2 62.96 69.30 0.2660 0.3588 B3 68.53 68.30 0.2586 0.3845 B4 64.72 67.11 0.3210 0.3829 B5 59.57 56.28 0.3927 0.3789 mean of B

64.23 59.84 0.3301 0.2838

C1 60.79 67.37 0.2523 0.2713 C2 70.23 69.80 0.5769 0.3359 C3 66.70 68.26 0.1368 0.2618 C4 71.71 69.17 0.1368 0.2667 C5 63.98 63.02 0.4341 0.3859 mean of C

66.68 67.52 0.3074 0.3043

D1 64.90 65.99 0.2424 0.2825 D2 63.27 68.69 0.2243 0.4279 D3 63.99 66.70 0.2025 0.2810 D4 63.47 66.66 0.2117 0.2682 D5 57.70 60.89 0.4456 0.4164 mean of D

62.67 65.79 0.2653 0.3352

mean of all

64.96 65.33 0.28 0.30