Post-adolescent changes in the perception of regional sub-phonemic variation

Highlights

• We examined stop perception in 164 Korean speakers at different education levels.

• Participants represented both a standard and non-standard dialect of Korean.

• The primary perceptual cue, either VOT or f0, differed across dialects.

• University students accommodated supralocal variants more than high school students.

• Sociolinguistic experience affects perceptual tolerance of phonetic variants.

Abstract

Previous studies have shown that the experience of higher education can influence speakers’ use of local and supralocal variants, but there has been less work examining its effect on perception. In the current study, we investigated the effect of higher education on perceptual cue-weighting by comparing high school and university students speaking two different dialects of Korean: Standard Seoul Korean (SSK) and Kyungsang Korean (KK). SSK speakers are known to perceptually weigh f0 over VOT in the stop laryngeal contrast, whereas KK speakers weigh VOT over f0. 117 high school and university students completed a stop identification task by responding to auditory stimuli built from VOT and f0 continua. Results revealed that while dialect-specific cue-weighting patterns existed among both SSK and KK listeners, the cue-weighting of university students in both regions was less dialect-specific than their respective high school counterparts. Comparing these patterns with those of 47 elementary school students confirmed that the trend is not directly correlated with the listeners’ ages. These findings suggest that the sociolinguistic experience accompanying the transition into higher education motivates listeners to flexibly accommodate supralocal phonetic variation regardless of dialectal prestige.

Introduction

Diachronic sound change is naturally accompanied by synchronic phonological variation. Sound changes spread across speakers in a community over time, and therefore at any given time during a sound change it is inherently true that some speakers will be more advanced in the change than others (De Decker, 2006, p. 65). Because innovative variants are more likely to be observed in speakers from younger generations as they spread across geographic space (e.g. Takada, 2008), the chronological age of speakers is useful information in understanding the structure of linguistic variation and the progress of a sound change. However, it is also true that individual speakers find themselves in different social networks and involved in different institutions at different stages in their lives. For example, much linguistic research is carried out in societies in which it is common for post-adolescents to pursue tertiary education and attend university—and in fact, much published linguistic data is collected from university students themselves (e.g. Andringa & Godfroid, 2020). But the transition from secondary school to university is often accompanied by the establishment of new social networks and exposure to new linguistic variants (Bigham, 2012), which could serve to promote or hinder, as the case may be, the adoption of a sound change. In other words, the difference between secondary school and university students is partly a difference in chronological age, but there also exists an important difference in the range and nature of linguistic variation they are exposed to in their peer groups.

The linguistic experience of a listener is also known to affect their perception of variation (Brunelle, 2009, Dalcher et al., 2008, Kirby, 2010, Lee et al., 2013), suggesting that the social changes that speakers experience as they enter adulthood may also affect how they perceive linguistic variables. This possibility leads us to the overarching research question addressed by the current study: How does the age of a speaker, beyond merely “young” or “old”, influence their perception of a phonological contrast undergoing sound change? We address this question by investigating the perception of the Korean three-way stop laryngeal contrast by three different groups of “young” listeners in different institutional settings: elementary school, high school, and university students from two dialect regions, Seoul and South Kyungsang, that differ in their stage of a change in progress (Lee & Jongman, 2019). Our focus, however, is on the two older groups, with an aim to understand how the perception of a linguistic variable may change as a result of the sociolinguistic experience that often accompanies post-adolescence in South Korea.

The investigation of sound change is naturally focused on differences between speakers of different ages. For example, earlier studies that offered evidence of sound change in the Korean stop laryngeal contrast, such as Silva, 2006, Kang and Guion, 2008, did so by showing that younger speakers were producing the contrast with phonetic cue weightings different from older speakers. But the spread and adoption of sound change is conditioned on a range of factors, one of which is the specific community that the speaker is a part of: who are the speaker’s peers, and what social or institutional pressures might influence the speaker’s adoption of an innovative variant? While this topic has been examined in many different domains, the subset of previous research most relevant to the current study consists of studies that investigated post-adolescent linguistic variation in a higher education setting.

Entering university often means finding oneself in a new peer group, with potentially different values or social practices. Some evidence suggests that a speaker’s orientation to their new peer group may influence their adoption of new linguistic variants (Bigham, 2010, De Decker, 2006). Bigham (2010) further showed that this accommodation does not necessarily happen by adopting new variants wholesale, but often by shifting the category space in a way such that less extreme variants from the local repertoire can still be included. Other studies have shown that the social salience of the variable in question may also play a role. Prichard and Tamminga (2012) examined four vowel features in students from the same Philadelphia neighborhood. In comparing the students who attended a national university versus those who attended a community college or completed no higher education at all, differences were observed only in the two features that were subject to overt negative evaluation. In the case of the vowel features that were undergoing change but below the level of awareness, they observed no effect of higher education. A similar pattern was observed in the speech of the Ohio undergraduate students reported on in Campbell-Kibler, Walker, Elward, and Carmichael (2014). These results suggested the accommodation to new linguistic variants in a higher education setting was at least partly socially driven, rather than a purely mechanistic accommodation driven by mere exposure.

Of course, higher education is not unique in this regard: there are many other institutional or social settings that also offer speakers exposure to a wider range of linguistic variation. What makes higher education worth investigating here, however, is that it is so common. In South Korea in 2019, 70% of people aged 25 to 34 had attained a tertiary degree (OECD, 2021), suggesting that most Koreans spend at least part of their 20s attending university. Thus, in light of previous production studies (e.g. De Decker, 2006, Bigham, 2010), this widespread attainment of higher education calls attention to the question of how this experience may impact listeners’ perception of phonological contrasts.

Outside of a university setting, studies on the effect of linguistic experience on the perception of dialectal variation have shown that the perception of a phonological contrast may be affected by the variation the listener has been exposed to (Brunelle, 2009, Dalcher et al., 2008, Kirby, 2010). Relevant to the current study, it has also been shown that native Korean listeners from Seoul and South Kyungsang use different phonetic cue weightings to identify members of the Korean three-way stop laryngeal contrast (Lee, Politzer-Ahles & Jongman, 2013), to be discussed in more detail in the next section.

One of the few studies that has investigated changes in both production and perception in the course of higher education, Evans and Iverson (2007), reported that students from northern England were perceived to sound more southern after having attended university for 2 years, and acoustic analysis of their vowel productions showed evidence of shifting to a more southern pronunciation for the vowel in bud and cud. There was no change, however, in their perception of the best exemplars for this vowel category. In other words, while students did seem to accommodate their production to the wider linguistic variation encountered among their university peers, this linguistic experience did not seem to change their perceptual prototypes of the vowels whose pronunciation shifted. Evans and Iverson (2007) speculated that such changes in perception could take longer than the timescale covered in their study.

Several questions remain, however. Can the perception of a phonological contrast change in a higher education setting if given enough time? Does it matter whether the category is currently undergoing sound change, and whether the variable is socially salient or negatively evaluated? Lastly, to our knowledge, all previous studies on differences in the production or perception of variation between high school and university students have focused on English vowels, and in the context of either England or the United States. Thus, another goal of the current study is to expand the empirical record by investigating the perception of the laryngeal contrast in Korean stops by high school and university students in both Seoul and South Kyungsang, a region whose local variety is one of the most recognizably different from that of Seoul (Shin et al., 2020). In the next section we provide an overview of the phonetic properties of this contrast, and how the sound change has unfolded across the two dialect regions in question.

Korean has a three-way laryngeal contrast among voiceless stops, namely, fortis (/p’, t’, k’/), lenis (/p, t, k/), and aspirated stops (/p^h, t^h, k^h/), with the phonetic and phonological properties of this contrast having been thoroughly documented over the past six decades (e.g., Kim, 1965, Han and Weitzman, 1970, Cho et al., 2002, Kong et al., 2011, Lee, 2016, Lee et al., 2020). Research in the past 15 years, however, has focused on a sound change in the stop contrast of standard Seoul Korean (e.g., Kim and Duanmu, 2004, Silva, 2006, Kang and Guion, 2006, Kang and Guion, 2008, Kang, 2014). Production data has shown that while the three stop types used to be differentiated by VOT alone, in the order fortis < lenis < aspirated, the VOT of lenis and aspirated stops has become overlapped in younger Seoul speakers born after roughly 1965. Rather than neutralizing these two stops, however, these young speakers rely on the f0 of the following vowel as a primary cue to the contrast, as it is lower after the lenis stop than after fortis or aspirated stops, and VOT as a secondary cue. According to the meta-analysis presented in Beckman, Li, Kong, and Edwards (2014), this sound change is mostly complete for speakers born after 1985.

Studies have also demonstrated an analogous shift in the weighting of the VOT and f0 cues in perception (Kang, 2010, Kim et al., 2002, Kong et al., 2011, Schertz et al., 2015, Kong and Lee, 2018). The general finding reported by these studies is that while Seoul listeners primarily rely on VOT in identifying fortis vs. non-fortis stops, they use both VOT and f0 in distinguishing lenis from aspirated stops. And while studies vary somewhat in the specific reported weighting of VOT vs. f0, it is clear that VOT is no longer the sole perceptual cue in perceiving the lenis-aspirated contrast, and there is a growing consensus on the enhanced role of f0 in the perception of younger listeners.

This sound change has spread beyond Seoul, however, and is in progress in the southeastern region of Kyungsang¹ as well, albeit about one generation behind the change in Seoul. The Kyungsang region comprises two provinces, North Kyungsang and South Kyungsang, which have eponymous and similar dialects. Among Korean dialects, North and South Kyungsang Korean combined have the second largest number of speakers after Seoul Korean. In perceptual dialectology studies, however, Kyungsang dialects are consistently rated as non-standard (Long and Yim, 2002, Kang, 2010, Jeon and Cukor-Avila, 2015), and Kyungsang speakers themselves self-report lower proficiency in standard Korean and using it far less than speakers from other regions (National Institute of Korean Language, 2015).

Unlike Seoul Korean, the North and South Kyungsang dialects have lexical pitch accents (Chang, 2007, Lee, 2008, Lee, 2013, Ramsey, 1975). Previous studies have examined how the use of f0 for pitch accent affected the role of f0 as a cue to the laryngeal contrast in both production (Holliday and Kong, 2011, Jo, 2004, Kenstowicz and Park, 2006, Lee and Jongman, 2012) and perception (Lee, Politzer-Ahles & Jongman, 2013). In these production studies, Kyungsang speakers, compared to Seoul speakers, relied on f0 less and VOT more, resulting in less overlap in VOT across stop categories. In perception, Lee, Politzer-Ahles, and Jongman (2013) showed that while both Seoul and Kyungsang listeners relied on short-lag VOT to identify fortis stops, in distinguishing lenis from aspirated stops Kyungsang listeners mainly relied on VOT, whereas Seoul listeners used both VOT and f0. These findings suggested that although the presence of lexical pitch accent in Kyungsang Korean made f0 a less reliable acoustic cue than in Seoul Korean, it was compensated for by an increased role of VOT.

There are reasons to question, however, whether this dialect-specific cue-weighting pattern is consistent across Kyungsang speakers of different generations. Korean language ideology has emphasized linguistic homogeneity in favor of standard Seoul Korean (Silva, 2011), and under increasing influence from prestigious Seoul Korean, the phonetic properties of regional varieties have changed (Jang, 2013, Lee and Jongman, 2016, Lee, 2013). For example, nominal pitch accent contrasts have become less acoustically distinct for younger speakers in their 20s, neutralizing both spectral and temporal f0 distinctions (Lee and Jongman, 2015, Lee et al., 2016). Recent production studies (Hwang et al., 2019, Lee and Jongman, 2019) have reported age-related differences in the use of VOT and f0 to cue the stop contrast, with the direction of the generational change similar to the sound change in Seoul. In Lee and Jongman (2019), the VOT and f0 patterns observed for younger Kyungsang speakers in their 20s were not exactly the same as those of same-age speakers in Seoul, but rather similar to older Seoul speakers, whereas older Kyungsang speakers in their 60s and 70s relied on f0 even less. Thus, although Kyungsang speakers lag behind Seoul speakers in this sound change, the change in progress in Kyungsang Korean stops is in the same direction as the change that took place in Seoul.

To date, there are far fewer studies on age- and dialect-related differences in the perception of this contrast, and studies that have examined age effects at all have not typically considered distinctions finer grained than “younger” vs. “older” whole generation-level differences. And while acoustic cue-weighting in production and perception are probably linked, there must be more flexibility in perception given the range of production variation encountered across talkers (Beddor, 2015), raising the possibility that acoustic cue-weighting in perception may not mirror exactly what has been observed in production.

The aim of the current study was to investigate the use of VOT and f0 in identifying the stops of the Korean three-way laryngeal contrast by six groups of relatively young listeners. These groups differed in age (operationalized as educational setting: elementary school vs. high school vs. university) and dialect (Seoul vs. Kyungsang). In terms of dialect, we predicted that Seoul listeners would overall rely on f0 more than Kyungsang listeners do, based on previous acoustic and perceptual findings (e.g., Lee and Jongman, 2012, Lee et al., 2013). The primary comparisons of interest, however, were between the high school and university students within each dialect.

Our first research question is whether the differences between high school and university students can be best understood as primarily driven by the sound change in progress. While the mean difference in chronological age between the two groups is not large, and most of the listeners in Seoul would probably have already completed the sound change (Beckman et al., 2014), it is still possible that high school students could show more innovative cue-weighting patterns than university students simply because they have a younger peer group who has progressed further along in the sound change. In particular, it has frequently been observed that in a sound change in progress, adolescents adopt innovative forms at higher rates than pre- or post-adolescent speakers (e.g. Cedergren, 1988), a phenomenon known as an “adolescent peak” (Tagliamonte & D’Arcy, 2009). And while the sound change is more or less complete in Seoul, it is ongoing in Kyungsang, raising the possibility that adolescent Kyungsang speakers may perceive the stop contrast in a way that reflects the direction of change (i.e. more reliance on f0 and/or less reliance on VOT) more than Kyungsang university students do.

Note that this hypothesis assumes that listeners’ cue-weighting in perception should mirror their cue use in production. While some studies have suggested that this may be the case (e.g. Newman, 2003, Coetzee et al., 2018), other evidence suggests that, in the context of a sound change, the degree to which cue use in production and perception align could vary depending on how far along in the sound change the speaker is. Pinget, Kager and Van de Velde (2020) found that in the diachronic devoicing of Dutch /v/, speakers for whom the change is complete still perceived voicing to some degree, whereas speakers for whom the change was still ongoing were less sensitive to voicing in perception than their production would suggest. In other words, when a sound change is ongoing, perception and production may be misaligned such that innovative speakers are more conservative in their perception, whereas more conservative speakers are more innovative in their perception. Pinget et al. (2020, p. 678) suggest that this result may be due to listeners receiving input from a wide range of speakers who may be at different stages of the sound change.

To further explore the possibility that differences in cue-weighting between the university and high school students are simply due to the progression of sound change alone, we also checked whether the cue-weightings of the elementary school students in each dialect followed the same trend. While there are important sociolinguistic differences among these three age groups, the purpose of this first research question is simply to check whether differences in chronological age are the best explanation for the observed differences in perception.

If the differences between the high school and university students cannot be adequately explained by chronological age alone, we will then turn to our second research question: How could the experience of higher education affect perceptual accommodation in the context of a sound change? In light of previously reported sociophonetic differences between high school and university students, we hypothesize two possible effects in the current study. First, the experience of entering higher education could have a straightforward effect of making listeners more sensitive to the higher prestige standard variety of Korean, an effect that would play out differently in Seoul and Kyungsang. In Seoul, the local variety already carries overt prestige, and while Seoul speakers studying at a university in Seoul would come into contact with speakers from other regions whose stop contrast may not be as further along in the sound change, there would be little motivation to perceptually accommodate to such speakers. Thus, we may not observe large differences between Seoul high school and university students. In Kyungsang, on the other hand, the experience of higher education would manifest in university students shifting their perceptual cue-weighting to better accommodate the standard Seoul stop contrast, by increasing their reliance on f0 over VOT.

The second hypothesized effect is that the increased exposure to non-local speakers often found in a university setting would motivate perceptual accommodation to supralocal variation, regardless of prestige, an outcome in line with the findings of Pinget et al. (2020). For the Kyungsang listeners, this outcome would not look different from that of the first hypothesized effect, as the primary supralocal variety is the higher prestige standard variety. But for the Seoul listeners, this effect would manifest in university students adopting a less innovative perceptual cue-weighting, reducing reliance on f0, presumably as a result of having more contact with peers from outside Seoul.