Introduction
Zhongjiang Chinese (中江话) is a variety of Mandarin Chinese, one of the ten language families in China (Mandarin, Cantonese, Xiang, Min, Gan, etc.,). Sometimes the term “Mandarin” is also used to refer to the national lingua franca “Putonghua”. We refer to the national lingua franca as “Standard Mandarin” and reserve the term Mandarin (官话) to refer to the language family, as distinct from other families, e.g., Cantonese, Xiang, Min, Gan, etc. Within Mandarin, Zhongjiang is a member of the Southwestern dialects (西南官话) (Language Atlas of China 1988). It is spoken in the urban areas of Zhongjiang county (中江县) in Sichuan province (四川省 ) of China. Zhongjiang county belongs to the Deyang (德阳) region, which is located in the middle of Sichuan province, about 100 kilometers northeast from the provincial capital of Chengdu (成都). Zhongjiang is just beyond the Chengdu Plain (成都平原), so the terrain is mostly hilly, with little flat land. It has a population of 1.37 million residents and an area of 2,200km2 (data from Zhongjiang official website, updated July 7, 2021).
Existing studies on Zhongjiang Chinese reporting aspects of the phonetics include the Sichuan Dialects Investigation Report (Yang, Reference Yang1984), based on data recorded in 1946. The data for this report was collected in what was called Dasang town, currently Longtai town, which represents the rural village dialect and is phonetically different from the urban areas of Zhongjiang county. Longtai (formerly Dasang) is 20 kilometers southeast of Zhongjiang City. The Yang (Reference Yang1984) report is based on one speaker, who also lived in Chengdu for six years and in Emei (a county in Southwest Sichuan) for three years. Another phonetic description is the Phonology of Sichuan Dialects (1960), which recorded about 150 dialects in Sichuan province including Zhongjiang county. The aim of that work was to identify the variation across varieties in consonants, vowels and tones by comparing cognate words.
The aim of the present illustration is to zoom in on one of these varieties.
(A) Map of China with Sichuan province indicated in grey; (B) Inset of the DeYang region of Sichuan, which contains Zhongjiang, including Zhongjiang City, indicated by a double circle, and surrounding towns, indicated by single circles.
Although Zhongjiang is typically characterized as a Mandarin variety, it has some phonetic features that are shared with the Xiang language family. Xiang is a distinct family of languages, primarily spoken in Hunan province. Arguably, the shared phonetic features between Zhongjiang and Xiang are related to language contact. Mass immigration in the Qing dynasty, approximately 400 years ago, brought large numbers of Xiang speakers from Hunan province to Sichuan province, including to areas around Zhongjiang (Cao, Reference Cao1997; Zhou, Reference Zhou2013).Footnote 1 Today, there remain a few Xiang language islands in the rural areas of Zhongjiang county, such as Yongxing (Cui, Reference Cui1985). Due to the local prestige of Southwestern Mandarin, Xiang language islands are becoming endangered, and younger speakers may be less likely to acquire Xiang-influenced phonetic features, even in the Zhongjiang variety (see also Rao & Shaw, Reference Rao and Shaw2021).
This current article is based on recordings from five speakers (three male) of Zhongjiang Chinese, producing a range of different materials. Of the male speakers, Speaker 1 (Sp 1) was born in 1960, Speaker 2 (Sp 2) was born in 1959, and Speaker 3 (Sp 3) was born in 1997. Of the two females, one (Sp 4) was born in 1953 and the other was born in 1991 (Sp 5). All five speakers were born and raised in Zhongjiang City (double circle in Figure 1), where they also currently reside. All speakers use Zhongjiang Chinese in their daily lives. We selected older speakers, particularly those with less formal education, to minimize the influence of Standard Mandarin. Speakers were also screened by the first author to make sure that they were representative of the urban Zhongjiang variety. Additionally, at the end of the recording sessions, we administered part of a standardized test, used for assessing Standard Mandarin proficiency. This involves reading a story designed to elicit features of Standard Mandarin. There was variation across speakers in Standard Mandarin proficiency. Most of the examples submitted with this article are drawn from speaker 2, who showed the least Standard Mandarin proficiency, reading the Standard Mandarin passage with phonetic features characteristic of the Zhongjiang variety.
The tradition in Chinese dialectology, including past work on Southwest Mandarin has been to describe consonants, vowels, and tones in their surface phonetic form; thus, most past descriptions can be interpreted as surface allophones which may or may not support a phonemic contrast. In this description, we take the distribution of allophones into account in arriving at a phonemic inventory. As a result, our phonemic description contains considerably fewer consonants than past descriptions. We describe systematic allophonic variation after introducing the phonemic inventory.
In this article, we sometimes explicitly contrast the pronunciation of Zhongjiang Chinese with Standard Mandarin. To make it clear which variety an IPA transcription refers to, we adopted the convention of notating Zhongjiang Chinese transcriptions with “(ZC)” and Standard Mandarin with “(SM)”. Transcriptions that are not notated with this convention refer to Zhongjiang Chinese.
Consonants
In the consonant table, square brackets show allophones with restricted distributions, which can be derived from other consonants, as we describe in greater detail below. Briefly, [w] and [j] are in complementary distribution with the vowels [u] and [i]; [ɲɟ] occurs only in the environment preceding [i]/[j]; and, [l], which is variably nasalized, [l̄], surfaces only in onset position, in complementary distribution with [n]. Although these consonants have restricted distributions, we listed them in the table in square brackets as they are often included in descriptions of related Mandarin varieties.
The consonant table below depicts the inventory of Zhongjiang Mandarin. Square brackets indicate consonants with limited (non-contrastive) distributions.
Plosives
Plosives are produced at three places of articulation in Zhongjiang Chinese: bilabial /p pʰ/, as in [pɑ31] 八 ‘eight’, [pʰɑ31] 爬 ‘crawl’; alveolar /t tʰ/, as in [te31] 得 ‘to get’, [tʰe31] 特 ‘special’; and, velar /k kʰ/, as in [ko31] 各 ‘each’, [kʰo31] 壳 ‘shell’.
At each place of articulation, there is a contrast in aspiration. Table 1 reports the mean and standard deviation of Voice Onset Time (VOT) for aspirated and unaspirated plosives at each place of articulation. The mean and standard deviation were calculated across monosyllabic words (between 27 and 54 words per phoneme) for one male speaker. The list of words entering the analysis is provided in the appendix.
VOT of aspirated and unaspirated plosives in different places of articulation. The items that contributed VOT values, 258 monosyllabic words, are listed in appendix 1. All VOT values are in milliseconds (ms).
Our impression is that aspiration in Zhongjiang is generally stronger than in Mandarin, but this would require a more comprehensive study to test. The VOT measures for this speaker, averaged across a single production of 27–54 words per category, are particularly long for bilabial and denti-alveolar plosives, as compared with published data on other Chinese varieties. For Tianjin Mandarin, Li (2019) reports the following mean VOT values: /pʰ/ 91 ms, /tʰ/ 100 ms, /kʰ/ 116 ms. For Standard Mandarin, Rochet & Fei (Reference Rochet and Fei1991) report /pʰ/ 99.6 ms, /tʰ/ 98.7 ms, and /kʰ/ 110.3 ms. The observation that bilabial and denti-alveolar plosives have particularly heavy aspiration in Zhongjiang is consistent with past reports (Yang, Reference Yang1984). Additionally, some cognate words are produced as aspirated in Zhongjiang but unaspirated in Standard Mandarin: e.g., [pʰɑ31] 拔 ‘to pull’ (ZC) vs. [pa35] (SM), [tʰɐɯ451] 抖 ‘to tremble’ (ZC) vs. [təɯ214] (SM); [tɕʰe31] 截 ‘section’ (ZC) vs. [tɕje35] (SM); [tsʰɑɯ324] 造 ‘to create’ (ZC) vs. [tsau51] (SM).
Affricates
Affricates in Zhongjiang Chinese have two places of articulation: denti-alveolar /ts tsʰ/,Footnote 3 as in [ʦɑ31] 杂 ‘mixed’ and [ʦʰɑ31] 茶 ‘tea’, which involve the tongue tip generating frication at the teeth, and alveolo-palatal /ʨ ʨʰ/, as in [ʨi31] 集 ‘collection’, [ʨe31] 节 ‘holiday’, [ʨʰi31] 七 ‘seven’, [ʨʰe31] 切 ‘to cut’. Place of articulation for affricates is contrastive before all vowels except for /u/. Before /u/, only the denti-alveolars are found: [tsu31] 竹 ‘bamboo’, [tsʰu31] 出 ‘come out’. Denti-alveolars (but not alveolo-palatals) can also be followed by a labiovelar glide, as in [tswɑŋ34] 装 ‘pretend’ and [tsʰwɑŋ324] 窗 ‘window’. In addition to these two affricate series, Standard Mandarin Chinese also has retroflex (post-alveolar) affricates, which contrast with the denti-alveolar, as in [ʦɑ35] 杂 ‘mixed’ and [tʂɑ35] (SM) 炸 ‘fry’; whereas, in Zhongjiang, [ʦɑ31] represents both of these meanings.
We also note that the denti-alveolar affricate as well as the denti-alveolar fricatives (described below) condition allophonic variation in the following vowel /i/. In denti-alveolar contexts, /i/ surfaces as an ‘apical vowel’, which we represent as [ɹ̺]: (affricate) [tsɹ̺31] 汁 ‘juice’, [tsʰɹ̺31] 吃 ‘eat’; (fricative) [sɹ̺34] 丝 ‘silk’. Although we do not have articulatory data for ZC, other Chinese varieties have shown that the tongue position and shape for the apical vowel is nearly the same as for preceding fricatives/affricatives, i.e., denti-alveolar in the denti-alveolar fricative/affricate context and retroflex in the retroflex fricative/affricate context (see Faytak & Lin 2015 for Standard Mandarin and Faytak 2018 for Suzhou Chinese).
Finally, we note that in longer words the affricates are sometimes produced without full closure, surfacing instead as fricatives: for example /i53tɕʰã31/ 以前 ‘before’ surfaces as [ji45ɕʰã31].
Nasals
Nasals occur at four places of articulation, bilabial [m], alveolar [n], alveopalatal [ɲɟ], and velar [ŋ], and they alternate in some cases with laterals [l]. All of these consonants are often produced with an oral release. The oral release is particularly prevalent in the environment of /i/ and /j/. Since the alveopalatal nasal only occurs in this environment, we transcribed it as [ɲɟ] throughout, including in the consonant chart.
The bilabial /m/ can only occur in syllable onset position, where it is often pronounced with an oral release, i.e., as [mb]. This is also noted in The Report of Sichuan Dialect (Yang, Reference Yang1984). Figure 2 shows the waveform (top) and spectrogram (bottom) for a representative token, the /m/ from [mbe31] 麦 ‘wheat’. A decrease in energy in the upper formants can be observed in the spectrogram just before the release burst (the release burst is visible in the waveform). In particular, when we play back this token from the temporal mid-point of the /m/, we have the percept of /b/, which is not the case for fully nasalized tokens in other varieties. Although the nasals can be perceived as pre-nasalized stops, due to the oral release, they function phonologically in Zhongjiang as nasals.
Waveform (top) and spectrogram (bottom) of [mbe31] ‘wheat’.
In Zhongjiang, the velar nasal /ŋ/ can occur in both onset position, as in [ŋɡɑɯ31] 熬 ‘to stew’, and coda position, as in [fɑŋ31] 房 ‘house’, (also [ŋɡɑŋ31] 昂 ‘raise’). As with the other nasals, the velar nasal in onset position often has an oral release, [ŋɡ]. Figure 3 compares /ŋ/ across positions in the same word: [ŋɡɑŋ31] 昂 ‘raise’. In onset position, the bandwidth of the first formant is wide, characteristic of the nasal formant, but there is little energy at higher frequencies, particularly near (immediately preceding) the release burst. The syllables that begin with velar nasals in Zhongjiang are typically onsetless in Mandarin Chinese, for example [ŋɡai324] 爱 ‘love’ (ZC) corresponds to [ai451] 爱 ‘love’ (SM). The presence of the velar nasal in onset position is another way in which Zhongjiang differs from Mandarin Chinese and from North Mandarin dialects in general.
Waveform (top) and spectrogram (bottom) for [ŋɡɑŋ31] 昂 ‘raise’ illustrating velar nasals in onset and coda position, within the same word. The onset nasal has a wider first formant bandwidth and less energy at higher frequencies than the coda nasal. See sound file ‘52-raise’.
The Zhongjiang variety also has an alveolo-palatal nasal, [ɲɟ] with a limited distribution. [ɲɟ] only occurs before the front high vowel, [i], or palatal glide, [j]. In these environments, it contrasts with the lateral, [l]: [ɲɟi31] 泥 ‘mud’ (ZC) forms a minimal pair with [ldi31] 梨 ‘pear’ (ZC), which, like the nasals, often has an oral release. Standard Mandarin Chinese, in contrast, does not have an alveolo-palatal nasal. Cognate words in Standard Mandarin are pronounced either with a denti-alveolar nasal, /n/, e.g., [ni35] 泥 ‘mud’ (SM) or without an initial consonant, e.g,. [i51] 艺 ‘art’ (SM), although variable pronunciations of /n/ as /ɲ/ have also been described for Standard Mandarin. For example, Duanmu (Reference Duanmu2007) describes a pronunciation of [njan35] 年 ‘year’ as [ɲan35] (SM).
Comparison of initial [l] in different vowel environments: /i/ as in [lᵈi 31] 立 ‘stand’ (left) and /ɑ/ as in [lɑ31] 辣 ‘spicy’ (right).
The alveolar nasal [n] is generally restricted to coda position. However, the lateral approximant, which does occur in initial position, can be nasalized, [l̄], and even produced with an oral release, [ld], just like the initial nasals at other places of articulation. The degree of nasalization may vary according to context. Various patterns of nasal-lateral merger and blending have been documented across Sichuan varieties of Southwest Mandarin. Wang (2004:151) noted that merger of /n/ with /l/ in Sichuan varieties is common. The merged phoneme is usually recorded as /n/, and it is reported that merger of nasal /n/ with lateral /l/ could happen in any context. However, in some varieties, there are also intermediate sounds, i.e., nasalized laterals or lateralized nasals. The phonology of Sichuan dialect (1960:5) recorded that the /l/ in Sichuan dialect has three varieties: [n], [l] and [l̄], the usage of which varies across speakers as well as within speakers. Often, [n] tends to occur in front of /i/ or /y/, while [l] and [l̄] tend to occur before other, more open, vowels. This generalization is largely consistent with our observations of ZC.
We observe systematically different spectral patterns for [l]∼[l̄] across vowel contexts: /i/∼/j/ vs. others, e.g., /ɑ/. An example is shown in Figure 4. The left panel shows [ldi31] 立 ‘stand’; the right panel shows [lɑ31] 辣 ‘spicy’. Differences include that the [l] in [li31] shows the presence of high frequency energy, in the range of F2/F3 for [i], which are absent for [ɑ] in [lɑ31]. Additionally, the F1 formant bandwidth is wider for [l] in [ldi31] than in [lɑ31]. This may indicate greater nasal airflow in [ldi31] than in [lɑ31] (for a review of acoustic correlates of nasality, see Styler (Reference Styler2017) and Carignan (Reference Carignan2018)); however, F1 bandwidth can have many articulatory sources, including degree of lateralization. Despite possible differences in degree of nasalization across vowel contexts, we denote the Zhongjiang lateral as [l]∼[ld].
Summary of the distribution of nasal consonant allophones.
Examples marked with an asterisk, ‘*’, indicate a lexical gap; examples without * indicate an attested word.
Finally, we add that, in coda position, the place of oral articulation, alveolar vs. velar, of the nasal is dictated by vowel contrast, with the alveolar nasal following non-back vowels and the velar nasal following back vowels.
To summarize the nasal consonants, we present a table of allophones (Table 2). The first column summarizes the set of environments that condition variation. The top row lists phoneme to allophone mappings: (1) /n/ realized as [n]; (2) /n/ realized as [l]∼[ld]; (3) /ŋ/ realized as [ɲɟ]; (4) /ŋ/ realized as [ŋ]; and, for completeness, (5) /m/ realized as [mb]. The last three rows show the distributional facts in coda position, where /n/ is realized as [n] only following non-back vowels. Coda nasals in the back vowel environments of /ɑ/ and /o/ are realized as [ŋ], and no coda nasals occur after /u/. Since [n] and [ŋ] are in complementary distribution in coda position, it may be possible to analyze them as either /n/ or /ŋ/. In onset position, there are minimal pairs differentiating [ld] from [ɲɟ] and [l] from [ŋ]. Both of these contrasts, we have analyzed as /n/ vs. /ŋ/, with [l] and [ld] derived from /n/ and [ɲɟ] derived from /ŋ/.
We end the section on nasals with a brief comparison to other Mandarin varieties. In Standard Mandarin, there is a clear contrast between the alveolar lateral /l/ and the alveolar nasal /n/ in onset position: e.g., [lu51] 鹿 ‘deer’ vs. [nu35] 奴 ‘slave’; [loŋ35] 龙 ‘dragon’ vs. [noŋ35] 浓 ‘heavy’. Word pairs such as these are homophonous in the Zhongjiang variety. To our ears, the initial /n/ retains a lateral quality [l]∼[l̄]. Figure 5 displays spectrograms for the relevant comparison. The distribution of spectral energy for the initial consonant is similar, even in the presence of a nasal coda (c.f., Xiangxiang Chinese, [Zeng, Reference Zeng2019]).
Comparison of initial [l] in words without (left) and with (center, right) a nasal coda.
Finally, in coda position, Standard Mandarin shows a contrast in nasal place of articulation, e.g., /in/-/iŋ/, /ɐn/-/ɐŋ/ are contrastive, as in [ɕin55] (SM) 新 ‘new’ vs. [ɕiŋ55] (SM) 星 ‘star’ and [kɐn55] (SM) 根 ‘root’ vs. [kɐŋ55] (SM) 耕 ‘plow’. The cognate words are homophones in Zhongjiang Chinese, where /in/-/iŋ/ are both pronounced as [in] and /ɐn/-/ɐŋ/ are both pronounced as either [oŋ] following labials onsets, [p], [pʰ], [m], [f], and as [ɐn] following all other onset consonants.Footnote 4
Fricatives
There are five contrastive fricatives in Zhongjiang Chinese, /f/, /s/, /x/, /ɕ/, /z/. None of the fricatives can occur in coda position. Before /e/, there is a minimal quintuplet: [ɕe31] 雪 ‘snow’, [se31] 色 ‘color’, [ze31] 热 ‘hot’, [xe31] 黑 ‘black’ and [fe31] 或 ‘or’. In the /i/ environment, there is a three-way contrast between /s/, /z/, and /ɕ/, with allophonic variation in the following vowel: /si/ surfaces as [sɹ̺], as in [sɹ̺34] 丝 ‘silk’; /zi/ surfaces as [zɹ̺], as in [zɹ̺31] 日 ‘sun’; and /ɕi/ surfaces as [ɕi], as in [ɕi34] 西 ‘west’.
Figure 6 shows spectra of the five contrastive fricatives in the /e/ environment. /s/ has a peak in the spectrum around 3,600 Hz, which begins a plateau of relatively high amplitude in the frequency range from 3,600 Hz to 9,500 Hz. This contrasts with a lower frequency peak in /ɕ/. /z/ has an even lower frequency peak, presumably due to voicing, and a plateau of high amplitude energy from 2,700 Hz to 7,200 Hz. The velar fricative, /x/, is not voiced, but has low frequency energy peaks; amplitude falls sharply at higher frequencies, in contrast to the high amplitude energy plateaus for the coronal fricatives. The velar fricative /x/ may be realized as a uvular fricative [χ] when followed by a low vowel and falling tone, as in /xau3/ [χau451] 好 ‘good’. Even outside of this environment, the place of articulation of /x/ is produced with a relatively posterior constriction, evidenced by the particularly low spectral center of gravity compared to other languages (Rao & Shaw, Reference Rao and Shaw2021). Finally, the spectrum for the labiodental, /f/, is diffuse, with relatively stable energy levels across frequencies. Although the common place of articulation of this fricative is labial-dental, bilabial variants have also been observed (Rao & Shaw, Reference Rao and Shaw2021).
Spectra of five fricatives of Zhongjiang Chinese: /s/, /z/, /x/ and /f/.
The contrast between /f/ and /x/ is neutralized in some environments. The distribution of /f/ and /x/ is summarized in Table 3. The dominant pattern across speakers is that before /u/, only /f/ (and not /x/) can occur. We see the opposite pattern preceding the rime /oŋ/. In the environment of /oŋ/, only /x/ (and not /f/) can occur. Neither fricative can occur before /w/.
Distribution of /f/ and /x/
The distribution in Table 3 is the result of conditional mergers between /x/ and /f/. Velar fricatives in the environment of [w] were reinterpreted as /f/, e.g., [xwɑ31] 滑 ‘slippery’- -> [fɑ31] 法 ‘law’; for a more detailed account of the development of the velar and labiodental fricative distribution, including patterns of individual differences and discussion of historical development, see Rao and Shaw (Reference Rao and Shaw2021). Acoustic measurements of the non-coronal fricatives in Zhongjiang Chinese are available in Rao et al. (Reference Rao, Shaw and Dockum2022).
We note that the voicing distinction between /s/ and /z/, the denti-alveolar fricatives, is the only pure voicing contrast in the language. Other Chinese varieties have more extensive voicing contrasts, including within the fricatives. For example, Shanghai Chinese, has a distinction in voicing across the stop and fricative series, including a voicing contrast for labiodental fricatives: e.g., in [vu3] 父 ‘father’ vs. [fu1] 夫 ‘husband’ (Chen & Gussenhoven, 2015). In Zhongjiang, some aperiodic energy can be observed in the high vowels, including /u/ but not so much that we would describe it as a [v] allophone of /u/.
Approximants
Zhongjiang Chinese has two approximants: the glides [w j], as in [wu451] 五 ‘five’, [je214] 夜 ‘night’. These are in complementary distribution with the high vowels: [w] is in complementary distribution with [u]; [j] is in complementary distribution with [i]. The glides, [w] and [j], occur before vowels while the vowels, [u] and [i], occur elsewhere. Since it is not generally possible for consonants to surface as syllable nuclei (although see discussion of the “apical vowel”), we analyze the approximants as positional variants of the high vowels: /u/ and /i/.
In pre-vocalic position, there are some restrictions on the environments in which the glides can surface. Some restrictions apply only to [w], some apply only to [j], and some apply to both. Table 4 summarizes the attested and unattested vocalic environments of each glide.
Vowel environment of glides (see text for examples)
[w] can precede rimes beginning with [e] (as in [we451] 喂 ‘hello’ and [kwe31] 国 ‘country’), and [ɑ] (as in [wa34] 蛙 ‘frog’ and [kwɑ34] 瓜 ‘melon’), except for [ɑɯ]; but does not occur before rimes beginning with [i], [u], [o]. Additionally, [w] can occur before [ɐn] (as in [kwɐn451] 滚 ‘roll’) but not before [ɐɯ].
[j] can precede all rimes beginning with [ɑ] (as in [l̄ jɑŋ451] 凉 ‘chilly’) and [o] (as in [jo31] 药 ‘medicine’) but it cannot occur before [i] or [u]. Additionally, [j] can occur before [e] (as in mje31] 灭 ‘exterminate’) and [ɐɯ] (as in [l̄jɐɯ451] 料 ‘material’) but not before [ei] and [ɐn]. There are also phonotactic restrictions on the onset consonants that precede [j]. Of the consonants represented in Table 1, only [p, pʰ, t, tʰ, tɕ, tɕʰ, m, n, [ɲɟ] ɕ] can precede [j]. This is the same set of consonants that can precede the vowel [i].
Some tokens of [j] in our recordings have a high level of frication. This is most common in words with homorganic vowels, e.g., /ji31/ — ‘one’. Although the voiced fricative variant of the palatal approximant is not consistently produced across Zhongjiang speakers, it’s possible that it is more common in Zhongjiang than in other Mandarin varieties.
The above restrictions on glide distribution generalize to cases in which the glide is preceded by an onset consonant. For example, [w] can occur before /ai/, as in [wai324] 外 ‘outside’ but not before /ɑɯ/, *[waɯ], and this same restriction applies to [kw]; [kwai324] 怪 ‘strange’ is attested but *[kwɑɯ] is not. The restrictions therefore apply to the rime environment following the glide, even when the glide is incorporated into a complex onset.
The phonotactic restrictions on glide-vowel combinations cannot be stated with reference to the immediately following vowel only; it is necessary to state the entire rime. For example, [ɐɯ] cannot follow [w], but this is not due to a restriction on [w] and [ɐ] generally, because [ɐn] can, in fact, follow [w]. Similarly, after [j], [e] but not [ei] is permitted. Thus, the restrictions appear, at least in some cases, to refer to the entire rime as opposed to local combinations of a glide and a following vowel.
In many cognate words, Zhongjiang differs from Standard Mandarin in the presence/absence of a labiovelar glide. Differences include cases in which a [w] in Standard Mandarin is absent in Zhongjiang and cases in which a [w] in Zhongjiang is absent in Mandarin. There are two main environments for this variation.
One is the rime environment [ɐn]. In this environment, several words that contain a glide in Standard Mandarin, lack a corresponding glide in Zhongjiang. For example, the following Zhongjiang words are produced without a glide: [tsʰɐn34] 村 ‘village’,Footnote 5 [tsʰɐn324] 寸 ‘inch’, [tɐn34] 蹲 ‘squat’, [lɐn31] 轮 ‘wheel’, [sɐn451] 笋 ‘bamboo shoot’, [sɐn34] 孙 ‘grandson’, [tʰɐn34] 吞 ‘swallow’, [tsʰɐn34] 春 ‘spring’. The Standard Mandarin cognates of these words are produced with a glide: [tsʰwɐn55] (SM) 村, [tsʰwɐn451] (SM) 寸, [twɐn55] (SM) 蹲, [lwɐn35] (SM) 轮, [swɐn214] (SM) 笋, [swɐn55] (SM) 孙, [tʰwɐn55] (SM) 吞, [tsʰwɐn34] (SM) 春. This variation is not due to a categorical prohibition of the labiovelar glide before [ɐn]. As noted in Table 4, the labiovelar is allowed in this environment. Although it is absent in some words, it is retained in others. The following words contain a labiovelar glide in Zhongjiang, making them homophonous with their Standard Mandarin pronunciation: [kwɐn451] 滚 ‘roll’/[wɐn34] 温 ‘warm’/[kwɐn324] 困 ‘sleepy’.
A second environment for labiovelar glide variation is between a lateral consonant and the diphthong [ei]. In this environment we observe a labiovelar glide in Zhongjiang that is absent from cognate words of Standard Mandarin. Examples include: [lwei31] 雷 ‘thunder’, [lwei324] 累 ‘tired’, [lwei324] 泪 ‘tears’, [lwei451] 垒 ‘build’, [lwei324] 内 ‘inner’, which are pronounced without the glide in Standard Mandarin, i.e., as [lei35] (SM), [lei451] (SM), [lei451] (SM), [lei214] (SM), [nei451] (SM), respectively. This pattern of labiovelar glide retention also occurs in other Southwestern Mandarin varieties, including the Chengdu variety.Footnote 6
Like other Mandarin varieties, Zhongjiang has, in a restricted set of environments, an apical vowel, which we represent as [ɹ̺], e.g., [sɹ̺324] 四 ‘four’. Since this sound can only occur in syllable nucleus position (preceded by a homorganic consonant), we will discuss it in the vowel section (see also affricate section). In Zhongjiang, we analyze the apical vowel as an allophone of /i/.
Vowels
Monophthongs
There are seven contrastive monophthongs: five oral, non-rhotic vowels, one rhotic vowel, and one nasal vowel. In addition, there is an apical vowel, which we analyze as an allophone of /i/. These eight monophthongs are shown on the chart in Figure 7. The apical vowel, since it is not phonemic in this variety, is shown in square brackets.
Monophthong vowels.
A formant plot of the eight monophthongs, including the non-contrastive apical vowel, is provided in Figure 9. Formants were tracked using the Burg method in Praat. The tokens of the oral, non-rhotic vowels draw from six tokens each, produced in different onset consonant environments but with the same tone [𠀳𠀱] ‘falling tone’. This was not possible for the rhotic vowel, which has a more limited distribution (see discussion at the end of pg. 16). The tokens of the nasal vowel draw from the same onset consonant environments as the non-rhotic oral vowels but with more variation in tone. The tokens of the apical vowel as well necessarily draw from its limited distribution (see discussion on pg. 17, first paragraph). A list of the items that went into the formant plot is provided in Appendix B. The mean formant values of each monophthong are labeled with the IPA symbol for the vowel in large font. The individual tokens are shown as the IPA category label in a smaller font that matches the color of the IPA category label for the mean. We used the following eight labels for the monophthongs: /i/, /e/, /u/, /o/, /ɑ/, /ã/,Footnote 7 /ɚ/, [ɹ̺]. However, the specific qualities of the vowels each require some additional commentary.
The position of /i/ in the vowel space is high and front. In contrast to Standard Mandarin, we’ve noticed that the high front vowel in Zhongjiang sometimes triggers aperiodic energy. This is particularly common in the falling tone context.Footnote 8
The position of /e/ in the vowel space is also relatively high, i.e., close to /i/. Another relevant difference between Zhongjiang Mandarin and Standard Mandarin is that Zhongjiang does not have a high front rounded vowel /y/. Additionally, Standard Mandarin has restrictions on the distribution of /e/ that are absent in Zhongjiang. In Standard Mandarin, /e/ generally does not combine with onset consonants, occurring only as an independent syllable in a handful of words, such as the [e31] ‘modal particle’. In Zhongjiang, in contrast, /e/ can occur with onset consonants, e.g., [xe31] 黑 ‘black’, [pe31] 白 ‘white’, [se31] 色 ‘color’.
The high back vowel, /u/, is typically less rounded in Zhongjiang than in Standard Mandarin, and this is reflected in higher F1 and F2. In the formant plot, relative to /o/, /u/ is only slightly higher (lower F1) and is more front (higher H2). We have kept the symbol /u/ to represent this vowel for two reasons. First, even though it is less rounded than /u/ in other language varieties, it is not entirely unrounded. Second, a hyper-articulated token of this vowel involves dynamic formants moving from a more central position in the vowel space towards the high back part of the vowel space, something like [ɵ͡u] or [ə͡u]. These two options suggest that, even though it is more centralized in the vowel plot below, there is a sense in which it is still a high back rounded vowel.
We have noticed that the vowel /u/ can also be produced with turbulent energy in certain environments. These include onsetless syllables, as in /u31/ 吴 ‘a Chinese family name’, and following a labial-dental fricative, e.g., /fu31/ 福 ‘happiness’ (for other examples of vowels produced with turbulent energy, see approximant description). All of these examples also have a falling tone. Our impressionistic observation is that turbulent energy produced with /u/ is also more likely when it is produced with a falling tone.
The quality of /o/ is a mid-back rounded vowel, but it is particularly back in its auditory quality. This auditory quality may be due in part to a high degree of rounding. Figure 8 shows the lips for our speaker during the production of /pu/ (left), /tu/ (middle), and /to/ (right). The figure shows narrow lip aperture for /u/ in both consonantal contexts and greater rounding for /o/.
Frontal video of the lips comparing rounding in /u/ and /o/.
The low vowel, which we have labeled as /ɑ/ is relatively back (low F2) in these tokens, although more front (higher F2) allophones are also readily observed.
The single contrastive nasal vowel in Zhongjiang is /ã/, which contrasts with /ɑ/ and /e/ in minimal pairs such as: [mã324] 慢 ‘slow’ vs. [mɑ324] 骂 ‘scold’ (c.f., *[mãn] vs. *[mæn]); [xã31] 含 ‘contain’ vs. [xe31] 黑 ‘black’. The nasal vowel developed from loss of alveolar nasal codas in the environment of /ɑ/. Consequently, /ɑ/ is no longer followed by an alveolar nasal—it can still be followed by a velar nasal—and, /ã/ can only occur in open syllables.
Nasal codas condition other patterns of vowel allophony as well. In the environment of the alveolar nasal coda, neither [u] nor [e] appears. Instead there is a centralized and lowered allophone in this environment, e.g., [sɐn34] 深 ‘depth’. At first blush, it might seem reasonable to consider that this surface vowel [ɐ] found before alveolar nasals derives from either /u/ or /e/ phonemes. Historically, some of the words produced with [ɐn] rimes derive from *e and some derive from *wɐ. As we noted above, nasals do not follow /u/ in Zhongjiang, so it seems more likely that [ɐ] is a surface realization of /e/. If this is the case, the nasal coda conditions lowering and centralization of /e/ to [ɐ], such that [sɐn34] 深 ‘depth’ derives from /sen34/. Relatedly, [ɐ] also occurs as the first member of a diphthong [ɐɯ], which could be analyzed as /eu/.
Of the five non-rhotic oral vowels, there is a four-way contrast before nasal codas. If we adopt the analysis of [ɐ] as deriving from /e/, then the four monophthongs that contrast before nasals are: /i/, /e/, /o/, /ɑ/. On this analysis, /u/ cannot be followed by a nasal.
The contrastive rhotic vowel is /ɚ/. This vowel typically occurs in syllables that lack onset and coda consonants, as in [ɚ31] 儿 ‘son’, [ɚ51] 耳 ‘ear’ and [ɚ324] 二 ‘two’. The other monophthongs can all also occur in such syllables, which make a seven-way contrast between syllables consisting only of a monophthong: /ɚ/ [ɚ] vs. /e/ [e] vs. /o/ [o] vs. /a/ [ɑ] vs. /ã/ [ã] vs. /i/ [ʝ̩] vs. /u/.[v̩].
The rhotic vowel can also occur in syllables with an onset consonant if derived through contraction, a common process in Mandarin varieties, referred to as Er-hua. For example, surface syllables such as [tʰɚ34], which have an onset consonant and a rhotic vowel, derive from contraction of an open syllable /tʰeu31/ [tʰɐɯ31] and /ɚ31/, so that /se31 tʰeu31 ɚ31/ ‘tongue’ surfaces as [se31 tʰɚ34] 舌头 ‘tongue’. In this contraction process, the vowel in the syllable preceding a rhotic vowel is replaced with the rhotic vowel, resulting in syllables that, on the surface, have both an onset consonant and a rhotic vowel.
F1 and F2 values of monophthongs produced in open syllables by Sp1 and Sp2.
Zhongjiang does have an apical vowel, [ɹ̺]. This vowel surfaces as an allophone of /i/ after /s/, /z/, /ts/, /tsʰ/ (see sections on fricatives, pg. 10, and affricates, pg. 5). In Standard Mandarin, when [ɹ̺] follows [ts tsʰ s], it is typically referred to as a syllabic apico-laminal or as a laminal denti-alveolar approximant. Both of these fall under the cover term of ‘apical vowels’ and occur only in open syllables in Standard Mandarin (Duanmu, Reference Duanmu2007:111). Zhongjiang shares this same distribution. [ɹ̺], as an allophone of /i/, occurs only in open syllables. The formants of the apical vowel allophone are more centralized than /i/. In the examples included in this article, there is an average F1 of 449 Hz, which is closer to /e/ than to /i/; the average F2 is 1165 Hz, which is between /i/ and /u/. The presence of aperiodic energy is also a dimension that varies across Chinese varieties (e.g., Ling, 2009; Faytak & Lin, 2015; Shao & Ridouane, 2023). In two of the examples included with this illustration [sɹ̺34] 丝 ‘silk’ and [zɹ̺31] 日 ‘sun’, there is clearly aperiodic energy throughout the apical vowel.
A final note about monophthongs, unlike Standard Mandarin, there is no /y/ in Zhongjiang Chinese. All the /y/ in Mandarin Chinese are pronounced as /i/ in Zhongjiang Chinese. Thus, /ji31/ 一 ‘one’ has the same pronunciation as 鱼 ‘fish’ in Zhongjiang, c.f., a /i/∼/y/ contrast in Standard Mandarin.
Diphthongs
There are four contrastive diphthongs. Phonemically, we characterize these as starting with a mid- /e/ or low- /ɑ/ vowel and ending with an offglide, either /i/ or /u/: /ɑi/, /ei/, /ɑu/, /eu/. However, the offglide portion of /ɑu/ and /eu/ is almost entirely unrounded. Additionally, the starting position of /e/ is quite central and low, more like [ɐ], a quality that is also seen before alveolar nasals: [sɐn34] 深 ‘depth’. Phonetically, /au/ and /eu/ are better characterized as [ɑɯ] and [ɐɯ], respectively.
A formant plot of the four diphthongs is provided in Figure 10. The figure plots the F1 and F2 of the representative tokens for each diphthong at three time points, 25%, 50%, and 75% of total vowel duration.
Formant tracks of diphthongs. F1 and F2 are plotted at 25%, 50%, and 75% of total duration.
In addition to the four diphthongs above, which can be characterized as off-gliding diphthongs, the high front vowels, /i/ and /u/, can proceed some other vowels, giving rise to on-gliding. The distributional restrictions, i.e., which vowels can follow /i/ and /u/, are described in the approximants section (Table 4).
Nasal onset consonants appear to condition some variation in diphthongs. In words beginning with /m/, the Zhongjiang diphthong /eu/ often corresponds to the Standard Mandarin diphthong /au/ or /u/. For example, /meu324/ [mɐɯ324] 茂 ‘luxuriant’ 贸 ‘trade’ in Zhongjiang is produced as [mau51] (SM) in Standard Mandarin. These words are different still in the Chengdu variety of Mandarin, produced as [moŋ324]. These variants have in common a tongue dorsum rising movement but they differ in the vertical extent of the movement, i.e., starting and ending points of the movement, and whether it co-occurs with rounding (Standard Mandarin) or nasalization (Chengdu Mandarin). Notably, in non-nasal environments many words maintain the same diphthong across varieties. For example, /xeu324/ [xɐɯ324] 后 ‘back’ in Zhongjiang is produced as [xɐu51] (SM) in Standard Mandarin, /kau34/ [kɑɯ34] (ZC) 高 ‘high’ in Zhongjiang is produced as [kau55] (SM) in Standard Mandarin. Possibly co-articulation with the initial nasal complicates the perception of vowel height by widening first formant bandwidth (e.g., Carignan, Reference Carignan2018), conditioning variation across varieties. This diphthong variation is primarily observed in the /m/ environment—the nasal consonant that occurs in onset position in both varieties (see section on nasals, pg. 6).
Example pitch tracks for each tone.
Tones
Citation form
There are four contrastive tones in Zhongjiang: high-rising (T1), middle falling (T2), high falling (T3), fall-rise (T4). Each tone is exemplified below, along with Chao numerals representing the pitch contour:
The labels T1–T4 have correspondences with tones in other varieties, including Standard Mandarin. That is, the syllables with T1, a high rising pitch contour in Zhongjiang, also pattern together in Standard Mandarin, where these syllables have a high flat contour, /55/. The same goes for T2, T3, T4. In this sense, the tone labels T1–T4 function to describe large swathes of the lexicon that pattern together in tone across Mandarin varieties, even as the phonetic realization of the tone category varies.
Pitch tracks of the four citation tones are shown in Figure 11. We characterize T1 as ‘middle-rising’, 34 in Chao numerals, which is consistent with past work on related Southwest Mandarin varieties. T2, we characterize as a middle falling tone, 31. T3 is a high falling tone. Oftentimes, there is an initial rise to the high before falling. This pattern is reflected in the pitch track in Figure 11. Lastly, T4 is a fall rise tone, 324. In the pitch track below, the end of the pitch contour reaches the level of the end of T1. Outside of single syllable citation form pronunciation, the pitch rise of T4 typically does not reach the height of T1; rather, it is realized as low tone, 32, only.
Tone sandhi
Some words of Zhongjiang, like other dialects of Mandarin, involve fully reduplicated syllables. In these words, T2, T3, and T4 undergo tone sandhi. As summarized in Table 5, T2 becomes T1 on the second of two identical syllables; T3 becomes T2; and T4 becomes T1. For completeness, we also provide an example of T1 reduplication. The second of two identical syllables with T1 does not undergo a shift in tonal category—it remains T1; however, the realization of T1 in the second syllable can be higher and flatter than the realization of T1 on the first syllable. This reflects a broader trend in Zhongjiang, whereby the second syllable of disyllabic words tends to be more prominent than the first, c.f., the opposite trend in Standard Mandarin, whereby the second syllable is sometimes reduced relative to the first (Duanmu, Reference Duanmu2007; Qu, 2013).
Tone sandhi patterns in fully reduplicated syllables
Tone sandhi patterns and exceptions
A different set of tone sandhi patterns is found outside of syllable reduplication. However, there are exceptions to many of the patterns. For example, T1 + T3 syllables sometimes surface as T1 + T2, e.g., [tʰi34tsɹ̺31] 梯子 ‘ladder’. But there are also examples of T1+ T3 compounds that do not undergo sandhi, e.g., [kʰai34swei451] 开水 ‘boiled water’. Exactly what conditions the variation in tone sandhi—possibly metrical stress, morpho-syntactic, or lexical factors—requires future research. One type of example suggesting a basis in the morpho-syntax or related phonological properties such as metrical stress involves N-V dyads. For example, the compound 热水 [ze324 swei451] is ambiguous between a noun ‘hot water’ and a verb ‘to heat water’. The compound is a combination of T4 and T3 syllables, which undergo tone Sandhi, surfacing as T4 + T2, only for the noun meaning ‘hot water’. The verb meaning resists Sandhi, which means it is still T4 + T3 for the verb meaning ‘to heat the water’.
Table 6 lists examples of various Sandhi patterns as well as some known exceptions to each pattern.Footnote 9 The table focuses on cases that involve a categorical shift in tone. There are also some cases of within-category phonetic variation, such as T1 + T1 combinations, which sometimes surface with a high-level tone for the second T1, e.g, [ɕɑŋ34 ku45] 香菇 ‘mushroom’, [fa34sɐn45] 花生 ‘peanut’, the same phonetic pattern described above for reduplicated T1 syllables.
Syllable structure and phonotactics
The syllable structure in Zhongjiang Chinese is (C)(G)V(C), where C stands for consonant, G stands for glide, V stands for vowel. We have already reported some phonotactic restrictions on consonant and vowel combinations (see Table 4 for coronal consonants and Table 8 for glides). All consonants can be in the onset except [n]. Glides [j]/[w] can also be a syllable onset, such as [wai324] 外 ‘outside’, [je214] 夜 ‘night’. The rhotic vowel [ɚ] cannot be immediately preceded by a consonant, except through contraction. It can occur independently, in vowel-initial syllables, as in [ɚ31] 儿 ‘son’ or it can follow other vowels, creating complex syllable nuclei (see, e.g., Lee & Zee, Reference Lee and Zee2003). For consonant codas, only [n] and [ŋ] can be codas, as in [ɕin34] 心 ‘heart’, [tʰaŋ31] 糖 ‘candy’. There are also some restrictions on consonant-glide combinations. For example, the glide [w] cannot follow the bilabial stops: [p], [pʰ], [m] (see also the descriptions of approximants on pg. 12--13).
Recorded passage
The IPA transcription below is a phonemic transcription of the North Wind and the Sun passage, translated into Zhongjiang Chinese. The Chinese orthography is given below the phonemic transcription. The phonemic transcription denotes the lexical tone of each syllable, as opposed to the surface tone resulting from tone sandhi. For example, the sequence of two T4 tones in tse324 (T4) ko324 (T4) undergoes tone sandhi, and is produced as T4–T1, tse32 (T4) ko34 (T1), by the speaker of the passage. Our transcription denotes the underlying tones, e.g., tse324 (T4) ko324 (T4), throughout. Following the Chinese Orthography we provide a table aligning the IPA, English gloss, and Chinese characters.
IPA transcription
pe31 foŋ34 xo31 ǁ tʰai324 jɑŋ31 ǁ sɹ̺324 xɑɯ451 pʰoŋ31 jɐɯ451 ǁ ǁ ji31 tʰjã34 ǁ tʰɑ34 mbɐn31 tsai324 ji31 tɕʰi451 ǁ tsɐn34 lɐn324 ǁ kʰã324 lɑ451 ko31 lɐn31 kã324 ɕi34 ǁ ǁ tsʰɐn34ko324 tʨʰi324 ǁ tsʰɐn34 ko324 lai31 ŋɡã34 ǁ mbo324 te31 ko324 so451 ji451 zã31 ǁ ǁ tse324 ko324 sɹ̺31 xɐɯ324 ǁ jɐɯ451 ko324 ko324 lu324 ti31 zɐn31 ǁ lai31 lo31 ǁ pe31 foŋ34 ŋɡã34 ǁ tʨɐɯ324 twei324 tʰai324 jɑŋ31 so31 ǁ ŋɡo451 mbɐn31 kʰã31 ǁ lɑ451 ko324 ǁ lɐn31 pɑ451 lɑ324 ko324 ǁ ko324 lu324 ti31 zɐn31 ǁ ji34 fu31 ǁ tʰo31 te31 ɕɑ324 lai31 ǁ ŋɡo451 mbɐn31 ǁ tʨɐɯ324 so31 lɑ451 ko31 ǁ lɐn31 kã324 ǁ ǁ tʰai324 jɑŋ31 ǁ tʰoŋ31 ji324 lɐ31 ǁ pe31 xoŋ34 tʨɐɯ324 ɕã34 tʨʰi324 ǁ twei324 tso31 lɑ324 ko324 ǁ ko324 lu324 ti31 zɐn31 ǁsɹ̺451 tʨin324 ji31 tsʰwei34 ǁ lɑ324 ko324 zɐn31 ã34 ǁ pɑ324 ji34 fu31 ǁ tsʰɑɯ34 to31 ǁ je31 tsʰɑɯ34 je31 tʨin451 ǁ tsʰwei34 te31 je31 tɑ324 tsʰɑɯ34 to31 je31 tʨin451 ǁ pɑ451 pe31 foŋ34 ɑ31 ǁ tʨʰi324 fai324 lɐ31 ǁ ǁ tʰai324 jɑŋ31 ã34 ǁ tsai324 ji31 pjã34 ǁ in34 to31 ɕɑɯ324 ǁ ǁ tʰɑ34 tʨɐɯ324 twei324 tʂe324 ko324 pe31 foŋ34 so31 ǁ kʰã324 ŋɡo451 ti31 ǁ ǁ tʰai324 jɑŋ31 tʨɐɯ324 tsɐɯ451 tʨʰi324 ǁ twei324 tso31 tʂe324 ko324 zɐn31 ǁ tsɑɯ324 tso31 ǁ tse324 ko324 zɐn31 ǁ pɑ451 tʰai324 jɑŋ31 ǁ tsã451 tʨin324 lo31 ji31 tsɑɯ324 ǁ tʂe324 ko324 zɐn31 ŋɡã34 ǁ ze31 te31 pu31 ɕin31 ɐn31 ǁ ze31 te31 mbo31 fɑ31 ǁ pɑ451 ji34 fu31 ǁ tɐɯ34 tɑ451 sɹ̺31 lo31 ǁ ǁ tswei324 xɐɯ324 ǁ pɑ451 ji34 fu31 ŋɡã34 ǁ je451 tʨɐɯ324 tʰo31 to31 ǁ kã34 kã44 tʨin324 tʨin324 ǁ ǁ pe31 xoŋ34 ŋɡã34 ǁ pu31 te31 ǁ pu31 so31 ǁ xai31 sɹ̺324 tʰai324 jɑŋ31 ǁ ɲɟi451 lɐn31 kã324 ɕi34 ǁ ǁ
Chinese Orthography
Alignment of IPA, English gloss and Chinese character
Glossary
1pS – 1st person singular pronoun
2pS – 2nd person singular pronoun
3pS –3rd person singular pronoun
1pPL –1st person plural pronoun
3pPL – 3rd person plural pronoun
CLASS – classifier
COP – copular verb
DV – directional verb
LINK – linking particle
MOD – modal particular
NEG – negation marker
REL – relative clause marker
PER – perfective marker
PREP – preposition
Acknowledgments
We acknowledge support from the Humanities and Social Sciences project “Sichuan Xiang dialect documentation and phonetical historical strata study” (24YJA740036) from the Ministry of Education, China, and the project “Study on historical strata of dialects in Chengdu Plain” (DFWH2023-002) from Xihua University. We would like to thank the participants of this study as well as two anonymous reviewers, the associate editor, Matthew Gordon, and the editor, Marija Tabain, for comments that greatly improved the final paper. All remaining errors are our own.
Appendix A. List of items that contributed to the measurements of VOT for aspirated and unaspirated plosives reported in Table 1
[pɑ451] 把 ‘handle’, [pɑ34] 爬 ‘climb’, [pu324] 布 ‘cloth’, [po31] 簿 ‘blotter’, [pu324] 步 ‘step’, [pei324] 贝 ‘shell’, [pai324] 拜 ‘bye’, [pai451] 摆 ‘put’, [pai324] 败 ‘fail’, [pi324] 币 ‘coin’, [pei34] 杯 ‘cup’, [pei324] 背 ‘back’, [pei34] 碑 ‘stele’, [pei324] 被 ‘blanket’, [pi451] 比 ‘compare to’, [pi31] 鼻 ‘nose’, [pɑɯ451] 宝 ‘treasure’, [pɑɯ324] 抱 ‘hug’, [pɑɯ34] 包 ‘bag’, [pɑɯ451] 饱 ‘full’, [pjɑɯ451] 表 ‘watch’, [pã324] 扮 ‘dress up’, [pã324] 办 ‘do’, [pɑ31] 八 ‘eight’, [pã34] 班 ‘class’, [pã451] 板 ‘board’, [pjã324] 变 ‘turn’, [pjã324] 便 ‘convenience’, [pje31] 别 ‘do not’, [pjã451] 扁 ‘flat’, [pje34] 憋 ‘hold’, [pã34] 搬 ‘move’, [pã324] 半 ‘half’, [po31] 拨 ‘to pull’, [pi31] 笔 ‘pen’, [pɐn451] 本 ‘notebook’, [pjɑŋ34] 帮 ‘help’, [po31] 薄 ‘thin’, [po31] 缚 ‘bind’, [pɑŋ451] 绑 ‘tie’, [pɑŋ324] 棒 ‘stick’, [po31] 剥 ‘peel’, [pe451] 北 ‘north’, [pin34] 冰 ‘ice’, [pje34] 逼 ‘force’, [pe31] 百 ‘hundred’, [pe31] 白 ‘white’, [pin34] 兵 ‘soldier’, [pin451] 柄 ‘handle’, [pin324] 病 ‘illness’, [pin451] 饼 ‘pancake’, [pi31] 壁 ‘wall’, [pʰo324] 破 ‘broken’, [pʰo31] 婆 ‘grandmother’, [pʰu451] 谱 ‘spectrum’, [pʰu34] 铺 ‘pave’, [pʰai31] 排 ‘row’, [pʰai324] 派 ‘delivery’, [pʰai31] 牌 ‘card’, [pʰei324] 配 ‘match’, [pʰei31] 赔 ‘compensate’, [pʰi31] 皮 ‘skin’, [pʰi324] 屁 ‘fart’, [pʰɑɯ324] 炮 ‘cannon’, [pʰjɑɯ324] 票 ‘ticket’, [pʰin451] 品 ‘sample’, [pʰjã324] 骗 ‘cheat’, [pʰjã324] 片 ‘piece’, [pʰã324] 判 ‘judge’, [pʰã31] 盘 judge’, [pʰo31] 泼 ‘spill’, [pʰin31] 贫 ‘poor’, [pʰi31] 匹 ‘be equal to’, [pʰɐn31] 盆 ‘basin’, [pʰɑŋ324] 胖 ‘fat’, [pʰoŋ31] 朋 ‘friend’, [pʰe31] 拍 ‘pat’, [pʰoŋ31] 棚 ‘shack’, [pʰin31] 平 ‘flat’, [pʰin31] 瓶 ‘bottle’, [pʰi31] 劈 ‘rive’, [pʰoŋ31] 蓬 ‘fluffy’, [to34] 多 ‘many’, [tɑ324] 大 ‘big’, [to451] 躲 ‘hide’, [tu451] 赌 ‘bet’, [tu324] 杜 ‘a last name’, [tai324] 袋 ‘bag’, [tai324] 带 ‘bring’, [ti34] 低 ‘low’, [ti324] 弟 ‘younger brother’, [ti324] 递 ‘pass’, [twei324] 对 ‘for or to’, [ti324] 地 ‘ground’, [tɑɯ34] 刀 ‘knife’, [tɑɯ324] 道 ‘way’, [tjɑɯ324] 钓 ‘fish’, [tɐɯ324] 豆 ‘bean’, [tjɐɯ34] 丢 ‘lose’, [tɑ31] 搭 ‘build’, [tã451] 胆 ‘courage’, [tã324] 淡 ‘light’, [tjæ451] 点 ‘spot’, [tjæ324] 店 ‘store’, [tje31] 跌 ‘tumble’, [tje31] 碟 ‘dish’, [tæ34] 单 ‘single’, [tɑ31] 达 ‘reach’, [tjã451] 典 ‘law’, [tjã324] 垫 ‘pad’, [twã34] 端 ‘hold’, [twã451] 短 ‘short’, [twã324] 断 ‘break’, [to31] 夺 ‘deprive’, [twɐn34] 墩 ‘block’, [tɐn34] 蹲 ‘crouch’, [tɑŋ451] 党 ‘Party’, [tɐn34] 灯 ‘light’, [tɐn451] 等 ‘wait’, [tɐn324] 凳 ‘chair’, [te31] 得 ‘get’, [tɑ451] 打 ‘fight’, [tin34] 钉 ‘nail’, [tin451] 顶 ‘top’, [tin324] 定 ‘fix’, [ti31] 笛 ‘flute’, [toŋ34] 东 ‘east’, [toŋ451] 懂 ‘understand’, [toŋ324] 冻 ‘freeze’, [toŋ324] 动 ‘move’, [toŋ324] 洞 ‘hole’, [tu31] 读 ‘read’, [toŋ34] 冬 ‘winter’, [tu31] 毒 ‘poison’, [tʰo34] 拖 ‘pull’, [tʰu451] 土 ‘soil’, [tʰu31] 图 ‘picture’, [tʰai34] 胎 ‘tyre’, [tʰai31] 台 ‘platform’, [tʰi34] 梯 ‘ladder’, [tʰi324] 剃 ‘shave’, [tʰɑɯ451] 讨 ‘mooch’, [tʰɑɯ31] 桃 ‘peach’, [tʰiɑɯ31] 条 ‘strip’, [tʰɐɯ451] 抖 ‘shake’, [tʰɐɯ34] 偷 ‘steal’, [tʰɐɯ31] 头 ‘head’, [tʰæ34] 贪 ‘corrupt’, [tʰæ31] 潭 ‘pond’, [tʰɑ31] 踏 ‘tread’, [tʰã451] 毯 ‘blanket’, [tʰɑ31] 塔 ‘tower’, [tʰjã34] 添 ‘add’, [tʰjã31] 甜 ‘sweat’, [tʰje31] 贴 ‘paste’, [tʰã324] 炭 ‘coal’, [tʰã31] 弹 ‘pop’, [tʰjã34] 天 ‘sky’, [tʰjã31] 田 ‘field’, [tʰje31] 铁 ‘iron’, [tʰo31] 脱 ‘take off’, [tʰɐn34] 吞 ‘swallow’, [tʰɑŋ34] 汤 ‘soup’, [tʰɑŋ31] 糖 ‘candy’, [tʰo31] 托 ‘lift’, [tʰɐn31] 藤 ‘vine’, [tʰe31] 特 ‘very’, [tʰin 34] 厅 ‘office’, [tʰin34] 听 ‘listen’, [tʰin31] 停 ‘stop’, [tʰin451] 挺 ‘very’, [tʰi31] 踢 ‘kick’, [tʰoŋ34] 通 ‘unblock’, [tʰoŋ451] 桶 ‘bucket’, [tʰoŋ324] 痛 ‘ache’, [tʰoŋ31] 铜 ‘copper’, [tʰoŋ451] 统 ‘all’, [ko34] 歌 ‘song’, [ko324] 个 ‘piece’, [ko451] 果 ‘fruit’, [ko324] 过 ‘pass’, [kwa34] 瓜 ‘melon’, [ku451] 古 ‘ancient’, [kai34] 该 ‘should’, [kai451] 改 ‘change’, [kai324] 盖 ‘cover’, [kai34] 街 ‘street’, [kai451] 解 ‘work out’, [kwai324] 怪 ‘odd’, [kwai451] 拐 ‘turn’, [kwai324] 挂 ‘hang’, [kwei324] 桂 ‘laurel’, [kwei34] 规 ‘rule’, [kwei324] 跪 ‘kneel’, [kwei34] 龟 ‘tortoise’, [kwei324] 柜 ‘cabinet’, [kwei451] 鬼 ‘ghost’, [kwei324] 贵 ‘expensive’, [kɑɯ34] 高 ‘tall’, [kɐɯ34] 钩 ‘hook’, [kɐɯ451] 狗 ‘dog’, [kɐɯ324] 够 ‘enough’, [kã451] 感 ‘feeling’, [ko31] 鸽 ‘dove’, [kæ34] 甘 ‘sweet’, [kã451] 敢 ‘dare to’, [kã34] 肝 ‘liver’, [ko31] 割 ‘cut’, [kwã34] 官 ‘officer’, [kwã34] 关 ‘close’, [kwã324] 惯 ‘be used to’, [kwɑ31] 刮 ‘scrape’, [kɐn34] 根 ‘root’, [kwɐn451] 滚 ‘roll’, [ku31] 骨 ‘bone’, [kɑŋ34] 钢 ‘steel’, [ko31] 各 ‘each’, [kwɑŋ34] 光 ‘light’, [ko31] 郭 ‘a last name’, [ko31] 角 ‘angle’, [kue31] 国 ‘country’, [kɐn34] 更 ‘even more’, [kɐn324] 梗 ‘stalk’, [ke31] 格 ‘grid’, [kɐn34] 耕 ‘plough’, [ke31] 隔 ‘partition’, [koŋ34] 公 ‘male’, [ku31] 谷 ‘grain’, [koŋ34] 宫 ‘palace’, [koŋ34] 恭 ‘respectful’, [koŋ324] 共 ‘together’, [kʰo451] 可 ‘can’, [kʰo324] 课 ‘lesson’, [kʰu34] 箍 ‘hoop’, [kʰu451] 苦 ‘bitter’, [kʰu324] 裤 ‘trousers’, [kʰai34] 开 ‘open’, [kʰwai451] 块 ‘piece’, [kʰwai324] 快 ‘quick’, [kʰwei34] 亏 ‘deficit’, [kʰɑɯ324] 靠 ‘depend on’, [kʰɑɯ34] 敲 ‘knock’, [kʰɐɯ451] 口 ‘mouth’, [kʰ ã324] 看 ‘look’, [kʰo31] 渴 ‘thirsty’, [kʰwã34] 宽 ‘width’, [kʰwe31] 阔 ‘broad’, [kʰwɐn324] 困 ‘sleepy’, [kʰɑŋ34] 糠 ‘chaff’, [kʰwɑŋ34] 筐 ‘basket’, [kʰwɑŋ31] 狂 ‘mad’, [kʰo31] 壳 ‘shuck’, [kʰɐn 451] 肯 ‘agree’, [kʰe31] 刻 ‘carve’, [kʰɐn34] 坑 ‘pit’, [kʰe31] 客 ‘guest’, [kʰoŋ451] 孔 ‘hole’, [kʰu31] 哭 ‘cry’
Appendix B. List of items that contributed to the plot of monophthong formants in Figure 9
[pa31] 爸 ‘father’, [pʰa31] 扒 ‘scrabble’, [ta31] 达 ‘reach’, [tʰa31] 塔 ‘tower’, [mba31] 麻 ‘hemp’, [la31] 拿 ‘take’, [pi31] 笔 ‘pen’, [pʰi31] 皮 ‘skin’, [ti31] 敌 ‘enemy’, [tʰi31] 题 ‘question’, [mbi31] 迷 ‘secret’, [ldi31] 梨 ‘pear’, [pu31] 不 ‘no’, [pʰu31] 仆 ‘servant’, [tu31] 毒 ‘poison’, [tʰu31] 涂 ‘paint’, [mbu31] 木 ‘wood’, [lu31] 奴 ‘slave’, [pe31] 白 ‘white’, [pʰe31] 拍 ‘pat’, [te31] 得 ‘get’, [tʰe31] 特 ‘very’, [mbe31] 麦 ‘grain’, [le31] 勒 ‘rein’, [po31] 钵 ‘bowl’, [pʰo31] 婆 ‘old woman’, [to31] 夺 ‘deprive’, [tʰo31] 坨 ‘lump’, [mbo31] 莫 ‘no’, [lo31] 罗 ‘a surname’, [pã324] 板 ‘board’, [pʰã31] 盘 ‘dish’, [tã34] 担 ‘carry’, [tʰã31] 弹 ‘bounce’, [mbã324] 慢 ‘slow’, [lã31] 蓝 ‘blue’, [ɚ31] 儿 ‘son’, [ɚ451] 耳 ‘ear’, [ɚ324] 二 ‘two’. [tsɹ̺31] 汁 ‘juice’, [tsʰɹ̺31] 吃 ‘eat’, [sɹ̺34] 丝 ‘silk’, [zɹ̺31] 日 ‘sun’, [sɹ̺451] 四 ‘four’
Open access
