Under sound change of stops in Kyungsang Korean, this study examined how the voice onset time (VOT) and F0 cues to the aspirated-lenis stop contrast were used in the productions of children, teenagers, young adults, and elderly speakers. Results showed that the three younger groups were more innovative in using VOT than the elderly speakers, but the use of F0 was not as innovative as the use of VOT. This study suggests that although there is some difference between Seoul and Kyungsang Korean, the sound change in Kyungsang Korean stops is incrementally similar to the change in Seoul Korean stops.

Korean has a typologically unusual three-way laryngeal contrast among voiceless stops, namely, fortis (/p′, t′, k′/), lenis (/p, t, k/), and aspirated (/ph, th, kh/) stops. In standard Seoul Korean, voice onset time (VOT) and F0 are the most important acoustic cues to differentiate the stop categories, indicating the VOT pattern of “fortis < lenis < aspirated” and the F0 pattern of “lenis < fortis ≤ aspirated” [e.g., Cho et al. (2002) and Lee et al. (2020)]. Importantly, the relative importance of the two acoustic cues has changed over the past 70 years. Previous apparent time studies examining the acoustic properties of the stops between younger and older Seoul adults showed that the VOT difference in distinguishing the aspirated-lenis contrast has become smaller, whereas the F0 difference for the contrast has become greater [e.g., Silva (2006) and Kang (2014)]. Unlike Seoul Korean, Kyungsang Korean (spoken by approximately thirteen million people in southeast Korea, which is the second largest population after Seoul speakers) is a pitch accent language in which the pitch accent pattern within a word can differentiate segmental homonyms [e.g., kácì (HL) “type” vs kàcí (LH) “eggplant”]. This means that Kyungsang speakers use F0 for two different purposes, namely, to cue the laryngeal category of any word-initial plosive, and to cue the word's lexical accent category. The presence of the lexical pitch accent in Kyungsang Korean makes the F0 cue as a function of laryngeal contrast less reliable compared to Seoul Korean, and the well-separated VOT distinction across stops compensates for the lesser role of F0 in the laryngeal contrast [e.g., Lee and Jongman (2012) and Lee et al. (2013)].

A recent acoustic study (Lee and Jongman, 2019) showed age-related differences in the use of VOT and F0 for the Kyungsang Korean stops, indicating a diachronic change in the same direction as the Seoul Korean sound change. When compared to Kyungsang elderly speakers in their 60 s and 70 s, Kyungsang speakers in their 20 s showed larger F0 differences and smaller VOT differences for the aspirated-lenis contrast. The F0 and VOT patterns of Kyungsang adults were similar to those of elderly Seoul speakers, based on which Lee and Jongman (2019) suggested that Kyungsang Korean is also undergoing the sound change that has affected Seoul Korean, although the Kyungsang sound change lags behind by about a generation. Considering the acoustic comparison between the adult and elderly speakers, the sound change in the stops seems to occur uniformly across Seoul and Kyungsang Korean merely on a different timeframe, which might indicate incremental sound change in both dialects of Korean [e.g., Beckman et al. (2014)]. However, it may be premature to conclude that the changes are identical given potentially different motivations for the change. While there is no doubt that the incremental sound change or phonetic motivation underlies the change of the Seoul Korean stops (i.e., change from below), it is not clear whether the change in the Kyungsang Korean stops is phonetically motivated or borrowed from Seoul Korean (i.e., change from above). In addition, if the change of the Kyungsang stops is due to a phonetic motivation, it also remains unclear whether the change started with the VOT neutralization and F0 divergence, similar to the Seoul change, or if it began with the neutralization of the pitch accent distinction [e.g., Lee et al. (2016)].

The comparison between the adult and elderly speakers in Lee and Jongman (2019) tells us whether the acoustic properties of the Kyungsang Korean stops have changed or not. Nevertheless, extending the inter-generational comparison further to children and adolescents might help us better understand the nature of the sound change by showing the sound change process from beginning to advanced stages over successive generations. Under regular sound change that Seoul Korean has undergone, we would expect continuous sound change over generations (Beckman et al., 2014). However, in Kyungsang Korean, whose pronunciation norms are not clear under the sound change context, it is questioned how children and teenagers utilize the innovative F0 and conservative VOT cues to distinguish the stop contrast compared to younger and older adults. If the type of sound change that Kyungsang Korean undergoes is the same as the one for Seoul Korean, and the two regional varieties share every stage of the change, Kyungsang pre-adult speakers would show gradual and continuous age-related variations. Specifically, we might speculate that as speakers are younger, they would be more likely to have interaction within the more innovative language network [e.g., Eckert (2017)]. Therefore, hypothesizing such incremental sound change over successive generations, we might predict the more innovative acoustic patterns in children and teenagers than in young adults. In contrast, if the change of the Kyungsang stop results from an abrupt borrowing affected by the stop change of Seoul Korean, such incrementally changing pattern would not be observed.

The current study examines the VOT and F0 distinction between the lenis and aspirated stops for children and teenager Kyungsang Korean speakers and compares the acoustic distinction with the stops produced by young adult and elderly South Kyungsang speakers whose data were adopted from Lee and Jongman (2019). Through this cross-generational comparison across four age groups, we hope to broaden our understanding of the developmental trajectory of the use of multiple acoustic cues for the Korean stop contrast as it undergoes sound change.

The production data newly collected for the current study were from thirty South Kyungsang Korean-speaking children [15 children (11 female, 4 male), mean age of 8.4 years (SD = 0.99)] and teenagers [15 young adults (15 female), mean age of 14.6 years (SD = 0.48)] in 2017. All participants, including the children, were literate in Korean, and all of them were born and educated in the South Kyungsang region (near Changwon city), with parents who also spoke the same dialect. The present study compared the data from these pre-adult speakers with the data from ten young adults and nine elderly South Kyungsang speakers collected in 2011 by Lee and Jongman (2019). The years of birth ranged from 2007 to 2010 for the children, from 2005 to 2006 for the teenagers, from 1988 to 1992 for the young adults, and from 1936 to 1950 for the elderly speakers. None of the participants reported language or hearing problems, and all of them provided informed consent.

This study recorded the same speech materials used in Lee and Jongman (2019) [27 disyllabic words = 3 initial stops (fortis, lenis, aspirated) × 3 accent classes (HL, HH, LH) × 3 places of articulation (bilabial, alveolar, velar)] [see Table 1 in Lee and Jongman (2019)]. Both the children and teenager Kyungsang speakers were instructed to read the words written in Korean orthography on index cards; pictures were provided to help participants understand the meaning of the words, especially for segmental homonyms. While most of the participants produced two repetitions of a randomized list of the test words, two children only did once due to fatigue. A total of 1566 tokens [(27 words × 2 repetitions × 28 speakers) + (27 words × 1 repetition × 2 speakers)] were recorded using a Marantz Digital Recorder (PMD661) and a SHURE head-mounted microphone at a sampling rate of 44 100 Hz.

This study measured the VOT of the stop and the F0 of the following vowel, using praat (Boersma and Weenink, 2017). The VOT and following vowel portion were first manually identified from the point of stop burst release to the onset of voicing, and the onset of F1 and the offset of F2, respectively, as seen in both the waveform and spectrogram. The onset of voicing was determined by the onset of the first full period. The VOT duration and F0 values at the midpoint of the vowel were later extracted using praat scripts. F0 values in Hz were converted to the semitone scale to minimize speaker variation across generation and gender based on the median value for each group. The measured VOT values in milliseconds were converted to log scale.

These VOT and F0 measures were analyzed with mixed-effects regression models, using the lme4 package (Bates et al., 2013) in R (R Core Team, 2020). The present study estimated the effect of the fixed variables Age (child, teenager, young adultreference, elderly) and Gender (male, femalereference) on dependent variables. This study converted the dependent variables by subtracting the VOT and F0 values of the lenis stop from those of the aspirated stop for each of the HL, HH, and LH accent classes to obtain the VOT and F0 differences between the aspirated and lenis stops in each accent class. The converted dependent variables were estimated in regression models constructed for each of the three accent classes. In so doing, we could test the effect of Age more directly on the F0 divergence and VOT convergence for the lenis and aspirated contrast. The model without Gender was compared with that adding Gender using a log-likelihood test. The p-values of the model coefficients were obtained using the lmerTest package (Kuznetsova et al., 2015).

Figure 1 shows the F0 ranges in fortis, lenis, and aspirated stops for child, teenager, young adult, and elderly Kyungsang Korean speakers as a function of log transformed VOT in milliseconds. Figure 2 presents the F0 difference between the aspirated and lenis stops for the four age groups in each accent classes as a function of log transformed VOT. In Figs. 1 and 2, the left and right sides of the interquartile ranges indicate the lower and upper quartiles of the group's log VOT values, respectively. Table 1 summarizes the parameter estimates for the Age and Gender fixed effects.

Fig. 1.

(Color online) Boxplots overlaid with scatterplots of F0 (semitones) in fortis, lenis, and aspirated stops for child, teenager, young adult, and elderly Kyungsang Korean speakers as a function of log transformed VOT in milliseconds. The x- and y-position of each dot represent the mean VOT and F0 values, respectively, averaged across accent classes and repetitions for an individual talker.

Fig. 1.

(Color online) Boxplots overlaid with scatterplots of F0 (semitones) in fortis, lenis, and aspirated stops for child, teenager, young adult, and elderly Kyungsang Korean speakers as a function of log transformed VOT in milliseconds. The x- and y-position of each dot represent the mean VOT and F0 values, respectively, averaged across accent classes and repetitions for an individual talker.

Close modal
Fig. 2.

(Color online) Boxplots of the F0 difference between the aspirated and lenis stops by child, teenager, young adult, and elderly Kyungsang Korean speakers for HL, HH, and LH accent classes as a function of the VOT difference between the aspirated and lenis stops.

Fig. 2.

(Color online) Boxplots of the F0 difference between the aspirated and lenis stops by child, teenager, young adult, and elderly Kyungsang Korean speakers for HL, HH, and LH accent classes as a function of the VOT difference between the aspirated and lenis stops.

Close modal
Table 1.

Estimates for Age of the mixed-effects models in the analysis of the VOT and F0 differences between the aspirated and lenis stops for each of the HL, HH, and LH accent classes.

HLHHLH
βt-valueβt-valueβt-value
VOT diff. for aspirated-lenis Intercept 0.44 0.35 0.34 
Agechild −0.03 −0.42 −0.03 −0.35 −0.11 −1.15 
Ageteenager −0.09 −1.08 −0.14 1.77c −0.06 −0.83 
Ageelderly 0.89 9.41a 1.03 10.95a 1.02 11.69a 
Gendermale    0.41 3.97a 
F0 diff. for aspirated-lenis Intercept 1.03 1.31 1.33 
Agechild −0.32 −2.38b 0.31 1.70c −0.58 −2.81a 
Ageteenager −0.04 −0.30 −0.05 −0.35 −0.14 −0.68 
Ageelderly −0.49 −3.31a −0.01 −0.10 −0.18 −0.80 
Gendermale   −0.88 −4.38a   
HLHHLH
βt-valueβt-valueβt-value
VOT diff. for aspirated-lenis Intercept 0.44 0.35 0.34 
Agechild −0.03 −0.42 −0.03 −0.35 −0.11 −1.15 
Ageteenager −0.09 −1.08 −0.14 1.77c −0.06 −0.83 
Ageelderly 0.89 9.41a 1.03 10.95a 1.02 11.69a 
Gendermale    0.41 3.97a 
F0 diff. for aspirated-lenis Intercept 1.03 1.31 1.33 
Agechild −0.32 −2.38b 0.31 1.70c −0.58 −2.81a 
Ageteenager −0.04 −0.30 −0.05 −0.35 −0.14 −0.68 
Ageelderly −0.49 −3.31a −0.01 −0.10 −0.18 −0.80 
Gendermale   −0.88 −4.38a   
a

p <0.01.

b

p <0.05.

c

p <0.1.

The young adults' estimated mean VOT difference between the aspirated and lenis stops was significant for the comparison with the elderly in all accent classes, the teenagers in HH, and the male children in LH. As expected, the mixed-effects regression models predicted significantly greater VOT differences between the two stops for the elderly group compared to those of the young adults in the HL, HH, and LH accent classes. The models showed a smaller VOT difference between the stops for the teenagers than the young adults in the HH class. Regarding Gender, boys patterned differently from girls only in the LH class, indicating that boys had a significantly greater VOT difference for the aspirated-lenis contrast compared to girls in LH accent words. Figure 1 shows that the VOT interquartile ranges for the aspirated and lenis stops are well separated for the elderly group, whereas they are very close together for the three younger groups. This Age effect is more clearly illustrated in Fig. 2, indicating that the VOT interquartile ranges on the x axis for the elderly do not overlap with those of the three younger groups for each accent class.

The young adults' estimated mean F0 difference between the aspirated and lenis stops was significant for several comparisons. For the HL class, the mixed-effects regression model predicted significantly smaller F0 differences between the two stops for the comparison with the children and with the elderly speakers. The HH model yielded an effect of Gender, predicting a significantly greater F0 difference between the two stops for girls than for young adults, but a smaller F0 difference for boys than girls, indicating that boys had a significantly smaller F0 difference for the aspirated-lenis contrast compared to girls in HH, as seen in Fig. 2. Finally, in the LH model the F0 distinction for the stop contrast was significantly smaller in the children. In general, the F0 distinction in teenagers patterned similarly with that of the young adults in all accent classes without the significant age-related differences, and the children (especially boys) and the elderly speakers showed similar patterns in terms of the absence of the significant group difference compared to the young adults. Figure 1 shows that the F0 interquartile ranges are separated between the aspirated and lenis stops more for the teenagers and young adult speakers compared to the children and elderly. Figure 2 also illustrates the regression results, presenting on the y axis the F0 interquartile range of the young adults overlapping with teenagers for each accent class. In Table 2, the four age groups are ordered according to the coefficients estimated by the regression model of each accent class.

Table 2.

The four age groups ordered by the coefficients of VOT and F0 differences in the aspirated-lenis contrast in the HL, HH, and LH regression models. [“≥ (or ≤)” indicates “less (or greater) than or equal to.”]

VOT differences (greater → shorter)F0 differences (smaller → greater)
HL Elderly > Adult ≥ Child ≥ Teen Elderly ≤ Child ≤ Teen ≤ Adult 
HH Elderly > Adult ≥ Child ≥ Teen Childmale ≤ Teen ≤ Elderly ≤ Adult < Childfemale 
LH Elderly > Childmale > Adult ≥ Teen ≥ Childfemale Child ≤ Elderly ≤ Teen ≤ Adult 
VOT differences (greater → shorter)F0 differences (smaller → greater)
HL Elderly > Adult ≥ Child ≥ Teen Elderly ≤ Child ≤ Teen ≤ Adult 
HH Elderly > Adult ≥ Child ≥ Teen Childmale ≤ Teen ≤ Elderly ≤ Adult < Childfemale 
LH Elderly > Childmale > Adult ≥ Teen ≥ Childfemale Child ≤ Elderly ≤ Teen ≤ Adult 

Table 2 indicates that the three younger groups have a shorter VOT difference in the aspirated-lenis pair compared to the elderly group, suggesting that the younger Kyungsang generations have undergone VOT convergence similar to standard Seoul Korean. In addition, the pre-adult teenagers and children, and the girls in particular, always had less of a VOT distinction in the aspirated-lenis contrast compared to the young adults, although this trend lacks the statistical significance. Regarding F0, however, the younger generations did not always show a greater F0 difference or diverging F0 in the aspirated-lenis contrast relative to the elderly group. In addition, the degree of the F0 difference for the children, and the boys in particularly, was as small as that for the elderly speakers.

This section presents the relationship between the VOT and F0 differences of the aspirated-lenis contrast for the four age groups to examine the source of the age-related variation in Kyungsang Korean through correlation tests (Fig. 3). It also examines the role of F0 as a function of segment distinction and F0 as a function of accent distinction to see whether the age effect of the F0 distinction is related to the change in the lexical pitch accent in Kyungsang Korean (Fig. 4).

Fig. 3.

(Color online) Correlations between the VOT and F0 differences of the aspirated-lenis stop contrast for child, teenager, young adult, and elderly Kyungsang Korean speakers averaged across accent classes.

Fig. 3.

(Color online) Correlations between the VOT and F0 differences of the aspirated-lenis stop contrast for child, teenager, young adult, and elderly Kyungsang Korean speakers averaged across accent classes.

Close modal
Fig. 4.

(Color online) Correlations between the F0 difference of the aspirated-lenis contrast and that of the HL-LH (upper panel) and HH-LH (lower panel) accent contrasts for child, teenager, young adult, and elderly Kyungsang Korean speakers.

Fig. 4.

(Color online) Correlations between the F0 difference of the aspirated-lenis contrast and that of the HL-LH (upper panel) and HH-LH (lower panel) accent contrasts for child, teenager, young adult, and elderly Kyungsang Korean speakers.

Close modal

Figure 3 confirms the regression results in Sec. 3.1. The three younger groups pattern similarly compared to the elderly group in terms of VOT. The VOT difference of the elderly group is greater than 60 ms, whereas that of the three younger groups is mostly shorter than that. As noted in Sec. 3.1, however, Fig. 3 shows less consistent F0 patterns across the accent classes. Figure 3 also illustrates the Gender effect: the boys show a greater VOT difference in the aspirated-lenis pair, but a smaller F0 difference mostly below 1.0 semitone. In addition to this confirmation, the elderly speakers exhibited less individual variation in both VOT and F0 compared to the younger groups, and among the four groups the children showed the largest individual variation in F0 ranging from 0 to 2.0 semitones. Finally, none of the correlation tests were significant, and the individuals who had greater (or smaller) VOT differences did not exhibit smaller (or greater) F0 differences between the two stops.

This paper also examined whether the F0 cue as a function of the segment distinction was related to the role of F0 as a function of the accent distinction, and Fig. 4 illustrates some age differences for these two F0 functions. Consistent with Lee and Jongman (2019), the elderly speakers have a greater F0 difference for the accent contrasts [(elderly: above 0.5 semitones), (younger: close to 0.0 semitones)] but a smaller F0 difference for the laryngeal contrast [(elderly: from 0.5 to 1.0 semitone), (younger: from 0.0 to 2.0 semitones)] compared to the three younger groups. Large individual variation is also observed in the children and elderly speakers regarding the F0 accent distinction. Finally, none of the correlation tests were significant; that is, individuals who used F0 more (or less) in distinguishing the aspirated-lenis contrast were not less (or more) likely to use F0 for the HL-LH and HH-LH accent contrasts.

The current study examined the VOT and F0 of the three-way laryngeal contrast of the Korean voiceless stops produced by children, teenagers, young adults, and elderly speakers of the Kyungsang dialect, focusing on the contrast between aspirated and lenis stops currently undergoing a sound change. Observing the acoustic properties of the stops over successive generations, this study aimed to explore the nature of the sound change as well as the developmental trajectory in using the two acoustic cues for the stops in a regional dialect of Korean. This study not only confirmed previous findings, but also showed novel findings providing some implications for the sound change in Kyungsang Korean stops.

In the first phase of the current analysis in Sec. 3.1, the Age effect on the VOT distinction tended to be consistent across the HL, HH, and LH accent classes, showing positive coefficients for the elderly and negative coefficients for children and teenagers. This suggests a weakening of the VOT distinction between the lenis and aspirated stops among all younger speaker groups, and that at least in terms of VOT the Kyungsang dialect of Korean is undergoing sound change in a similar direction as Seoul Korean. Notably, the comparison between the teenagers and young adults in the HH accent class was significant, indicating that the VOT distinction between the lenis and aspirated stops was converging more for the teenagers than the young adults, and the other comparisons with the children and teenagers revealed negative coefficients. This indicates that the children and teenagers tended to converge VOT between the lenis and aspirated stops more than young adults do, and the more innovative VOT pattern for the pre-adult speakers supports the incremental sound change of the Kyungsang Korean stops. Overall, the regression results on the VOT distinction suggest that the neutralization of VOT in contrasting the lenis and aspirated stops is prevailing for the younger Kyungsang generations.

An Age effect was also found for the F0 distinction, but it was less consistent than the age variation of the VOT distinction. Specifically, while the young adults exhibited a smaller VOT distinction between the lenis and aspirated stops (i.e., innovative VOT pattern) constantly across all accent classes compared to the elderly, the young adults showed a greater F0 distinction (i.e., innovative F0 pattern) only for the HL accent class. For HH and LH the F0 distinction was not different between young adults and elderly speakers. That is, the F0 divergence for the lenis and aspirated contrast is not constant across accent classes. In addition, the coefficient comparisons in Table 2 showed a smaller F0 distinction for the children than the elderly speakers, indicating that the children, and the boys in particular, were as conservative as the elderly in using F0. This suggests that in using F0 as a function of the laryngeal distinction the younger Kyungsang generations were not in a stage as advanced as they were in using the VOT distinction.

Noteworthy aspects of the data were the VOT and F0 variation regulated by gender and the large individual variation observed in younger generations, implying the stage in which the Kyungsang Korean stops are in the ongoing sound change. First, the gender-related variation supports the earlier discussion about the VOT neutralization before the F0 divergence. Specifically, most of the children had a shorter VOT distinction than the elderly speakers did, which contributed to the group-averaged pattern, but the boys had a larger VOT distinction than the girls did (see Tables 1 and 2). Likewise, for the F0 difference the boys had a smaller F0 distinction than the girls did, and some individual girls had an F0 distinction as great as the teenagers and young adults (see Figs. 3 and 4). That is, the girls in the present observation had more innovative VOT and F0 patterns than the boys did. Given that females are known to use more innovative phonetic form compared to males in the context of sound change, the gender-related individual variation in the current study suggest that Kyungsang-speaking children around the age of eight seem to start the neutralization of the conservative VOT cue before mastering the innovative F0 cue. Notably, this nature of the sound change in Kyungsang Korean stops is different than the stop change of Seoul Korean. Kang (2014) noted that adult female Seoul speakers who were in the most advanced stage of the sound change distinguished the stop contrast solely by the F0 difference, and suggested the emergence of the redundant F0 distinction before neutralizing the VOT distinction. What we observed in the current data, however, was opposite; VOT values were converging in all cases, but F0 values were not diverging constantly in all cases for the younger generations relative to the elderly group. Therefore, we might conclude that unlike Seoul Korean speakers, the Kyungsang speakers did not fully rely on the F0 difference for the lenis-aspirated stop contrast yet, even after they started neutralizing the VOT distinction. Second, the large individual variation also implies how the regional dialect speakers have initiated the sound change and how they have developed new phonetic forma that rely less on the conservative VOT cue, and learn the innovative F0 cue instead. Under sound change, we expect wide individual variation [e.g., Harrington and Stevens (2014)], and the individual variation in the current data was also larger for the three younger groups than the elderly in both VOT and F0 (see Fig. 3). That is, although the group-averaged pattern indicated a less advanced distinction for F0 than for VOT in the younger speakers, some of them already made the aspirated-lenis contrast with a greater F0 difference, suggesting the initiation of an F0 divergence across younger individuals.

In the second phase of the analysis in Sec. 3.2, we observed a relationship between the VOT and F0 cues for the laryngeal distinction and between F0 as a function of segment distinction and F0 as a function of accent distinction, speculating significant negative correlations between the variables, but did not find any. The results suggest that the change of the Kyungsang Korean stops might not be related to dialect-internal sound change, which might be phonetically motivated either by the VOT and F0 change or the neutralization of the lexical pitch accent distinction.

Congruent with these observations, this study suggests that the phonetic norm that Kyungsang Korean speakers pursue for the stop contrast, and the lenis and aspirated stops in particular, is the same as that of standard Seoul Korean, and that Kyungsang speakers are changing the acoustic cues through converging VOT and diverging F0 for the contrast. Notably, the nature of the change is different than that of the Seoul Korean stops, and the completion order of the VOT and F0 change is different between the two dialects of Korean in that Kyungsang Korean speakers did not start making a redundant F0 distinction before neutralizing the VOT distinction. However, whether or not the stops of the regional variety are undergoing an incremental dialect-internal sound change related to the phonetic motivation seems puzzling due to the lack of the support from the cue correlations. The regression analyses reported significantly shorter VOT for the teenagers than for the young adults in the HH class, indicating the gradual age variation. Considering the incomplete VOT change and the reassigning F0 in progress, we can expect to see more gradual age-group differences as the sound change is more advanced. However, the lack of significant correlations in Sec. 3.2 suggests that the sound change of the Kyungsang Korean stops might not be solely due to the neutralization of the lexical pitch accent, and multiple factors may be involved in the sound change. Therefore, this study suggests that the stop sound change that Kyungsang speakers are undergoing tends to be incremental, and further investigation with ample data would be able to confirm the type of underlying phonetic motivation underlying.

The Kyungsang dialect of Korean and its linguistic setting are different than standard Seoul Korean. Kyungsang Korean is a pitch accent language, whereas Seoul Korean is not; while the sound change that the Seoul Korean stops underwent was phonetically motivated, it was not clear what is motivating the change in Kyungsang Korean stops. This study has some implications regarding the trajectory of stop sound change for Kyungsang speakers with respect to the multiple acoustic cues and confirmed the dialectal differences of the sound change between Seoul and Kyungsang Korean. Different linguistic environments between the Seoul and Kyungsang dialects of Korean might result in different paths to reach the ultimate phonetic forms under the sound change. In future research, examining Seoul Korean stops produced by pre-adults in middle school and elementary school and testing the gender effect with more participants will provide a clearer picture about how exactly the nature of the sound change is different between the two dialects of Korean, and which stage the two dialects of Korean are at with respect to the change.

This work was supported by the Ministry of Education of the Republic of Korea and the National Research Foundation of Korea (NRF-2017S1A5A2A03068448).

1.
Bates
,
D.
,
Maechler
,
M.
, and
Bolker
,
B.
(
2013
). “
lme4: Linear mixed-effects models using S4 classes
,” R package version 0.999999-0, http://CRAN.R-roject.org/package=lme4 (Last viewed September 1, 2017).
2.
Beckman
,
M. E.
,
Li
,
F.
,
Kong
,
E. J.
, and
Edwards
,
J.
(
2014
). “
Aligning the timelines of phonological acquisition and change
,”
Lab. Phonol.
5
(
1
),
151
194
.
3.
Boersma
,
P.
, and
Weenink
,
D.
(
2017
). “
Praat: Doing phonetics by computer
” [computer software], http://www.Praat.org/ (Last viewed October 1, 2019).
4.
Cho
,
T.
,
Jun
,
S.-A.
, and
Ladefoged
,
P.
(
2002
). “
Acoustic and aerodynamic correlates of Korean stops and fricatives
,”
J. Phon.
30
,
193
228
.
5.
Eckert
,
P.
(
2017
). “
Age as a sociolinguistic variable
,” in
The Handbook of Sociolinguistics
(
Wiley
,
Hoboken, NJ
), Chap. 9, pp.
151
167
.
6.
Harrington
,
J.
, and
Stevens
,
M.
(
2014
). “
Editors' introduction: Cognitive processing as a bridge between phonetic and social models of sound change
,”
Lab. Phonol.
5
(
1
),
1
8
.
7.
Kang
,
Y.
(
2014
). “
Voice onset time merger and development of tonal contrast in Seoul Korean stops: A corpus study
,”
J. Phon.
45
,
76
90
.
8.
Kuznetsova
,
A.
,
Brockhoff
,
P. B.
, and
Christensen
,
R. H. B.
(
2015
). “
Package ‘lmertest,’
” R package version 2(0), https://github.com/runehaubo/lmerTestR (Last viewed March 15, 2018).
9.
Lee
,
H.
,
Holliday
,
J. J.
, and
Kong
,
E. J.
(
2020
). “
Diachronic change and synchronic variation in the Korean stop laryngeal contrast
,”
Lang. Ling. Comp.
14
(
7
),
e12374
.
10.
Lee
,
H.
, and
Jongman
,
A.
(
2012
). “
Effects of tone on the three-way laryngeal distinction in Korean: An acoustic and aerodynamic comparison of the Seoul and South Kyungsang dialects
,”
J. Int. Phon. Assoc.
42
,
145
169
.
11.
Lee
,
H.
, and
Jongman
,
A.
(
2019
). “
Effects of sound change on the weighting of acoustic cues to the three-way laryngeal stop contrast in Korean: Diachronic and dialectal comparisons
,”
Lang. Speech
62
,
509
530
.
12.
Lee
,
H.
,
Jongman
,
A.
, and
Zhang
,
J.
(
2016
). “
Variation and change in the nominal pitch-accent system of South Kyungsang Korean
,”
Phonology
33
(
2
),
325
351
.
13.
Lee
,
H.
,
Politzer-Ahles
,
S.
, and
Jongman
,
A.
(
2013
). “
Speakers of tonal and non-tonal Korean dialects use different cue weightings in the perception of the three-way laryngeal stop contrast
,”
J. Phon.
41
,
117
132
.
14.
R Core Team (2020). “
R: A language and environment for statistical computing
,” https://www.R-project.org/ (Last viewed May 1, 2020).
15.
Silva
,
D. J.
(
2006
). “
Acoustic evidence for the emergence of tonal contrast in contemporary Korean
,”
Phonology
23
,
287
308
.