This page contains examples of the citation tones of a speaker from Maodian 毛店, a town in the middle of the Chinese province of Zhejiang. Maodian is at the southern tip of Yiwu 义乌county, about equidistant from the cities of Yiwu 义乌 and Yongkang 永康, in Yongkang county to the south. The dialects in the Maodian area belong to the largely typologically-defined Wuzhou 婺州 subgroup of the Wu 吳 dialects. Wuzhou varieties are spoken in the central part of Zhejiang. Click here to view the location of Maodian in the Wuzhou subgroup . There is a characterisation of Wuzhou Wu, in Chinese, on pages 21 through 24 of Fu Guotong et al’s 1985 monograph Wu dialect subgroups of Zhejiang (傅国通, 方松熹, 蔡勇飞, 鲍士杰,傅佐之: 浙江吴语分区) published by the Zhejiang Linguistics Society. Information on Wuzhou may also be found in Cao Zhiyun’s 2002 book Studies on the phonetics of Southern Wu dialects (曹志耘: 南部吴语语音研究) published by the Commercial Press, Beijing.

The Maodian speaker's tones are interesting because they include both short stopped and long unstopped reflexes of the Middle Chinese Ru category, giving him nine separate tones. Read more about this here. The relationship between the citation tones and the syllable-Onsets is also interesting. Read about that here . Read about the recording of his tones here. Listen to the tones and inspect their acoustics here.


The recording, made by Professor W.L. Ballard in August 1988 at Ehime university in Japan, was part of his survey of tones and tone sandhi in southern Wu varieties. His generosity in making the data available for analysis is gratefully acknowledged. The speaker is a male who was 27 years old when recorded. He was born in Maodian in 1961 and lived there until going to Hangzhou University at 17.

Ballard compiled the corpus, and elicited the tones, according to their eight Middle Chinese tonal categories: Ia/Yinping 阴平, Ib/Yangping 阳平, IIa/Yinshang 阴上, IIb/Yangshang 阳上, IIIa/Yinqu 阴去, IIIb/Yangqu 阳去, IVa/Yinru 阴入, IVb/Yangru 阳入, in that sequence. Five examples of each of the eight cateories were included, each repeated three times after Ballard's numerical prompt. Click the following button to listen to Ballard's elicitation of the five tokens of the Ia/Yinping tone:

Ballard also recorded examples of the speaker's disyllabic tone-sandhi. This sandhi involves complex relationships between isolation tones and sandhi tones. You can listen to this disyllabic data, and inspect the disyllabic tonal acoustics, here: Maodian disyllabic tone sandhi.


The table below shows the speaker's citation tone acoustics (F0 as a function of absolute duration). It is arranged according to the conventional 2x4 matrix of Middle Chinese tonal categories (Yin Yang; Ping Shang Qu Ru. ). Each tone's Middle Chinese name is followed by the name of the Maodian tone in capitals, followed by Ballard's Simplified Character elicitation text, and glosses.

I measured the acoustics from the first of the three replicates elicited for each token. You can hear these tokens by clicking on the play/pause button (should work with Chrome, other browsers not guaranteed, sorry!). The dotted lines in the graphs belong to the individual replicates; their arithmetical mean is shown with a thick red line, or a thick blue line in the case of short tones.

Yinping/Ia: UPPER-MID LEVEL tone

飞 fly, 高 tall, 东 east, 关 shut, 风 wind


酒 wine, 手 hand, 广 factory, 火 fire, 好 good


货 goods, 四 four, 信 letter, 跳 jump, 太 too


north, snow, bone, put forth, go out

Yangping/Ib: LOWER-MID LEVEL tone

年 year, 田 field, 平 flat, 门 door, 长 long

Yangshang/IIb: DELAYED LOW RISING tone

被 blanket, 坐 sit, 马 horse, 肚 stomach, 象 elephant


用 use, 饭 rice, 电 electricity, 面 face, 病 ill


month, study, poison, white, ten


One of the characteristics of Wuzhou varieties is said to be the historical lengthening of the short stopped reflexes of the Middle Chinese so-called Ru 入 or entering tones (IVa & IVb). In some varieties the short tones have lengthened and merged with other tones; in others they have lengthened but remained separate by virtue of different pitch shapes. Interestingly, this Maodian speaker provides a further variation on this theme, in that he clearly shows a merger of etymological tone IVb with tone IIb (the delayed low rising tone), whilst keeping a lengthened version of etymological tone IVa separate (as the lower-mid rise tone). Note that in the table above two shapes appear in the column for the entering tone Ru/IV category- one short and one long. This reflects a situation again said to be typical for Wuzhou, whereby some morphemes with etymological IVa and IVb tones have alternative phonological shapes. One shape is conservative, preserving the short pitch ending in a glottal stop; the other is the innovative lengthened tone. For most IVa and IVb cognates the speaker has innovative long reflexes. For a few IVa and IVb cognates, however, he retaines a conservative short stopped tonal shape. Although this phenomenon is traditionally termed 文白異讀 "different colloquial and literary character readings", there was nothing in the linguistic structure of any of the morphemes involved that would serve as an obvious conditioning factor. Thus, for example, you can hear that he read the characters for the morphemes bone 骨, go out 出 and ten 十 with short stopped tones, but, in the same formal elicitation session, those for snow, put out and month were given long tones. Indeed, Ballard’s Maodian notes indicate some free variation, where bone is also transcribed with a long tone. Although the conditioning of such short forms remains elusive, therefore, it is clear that one has to deal with nine different tonal shapes.


In many Wu dialects there is a close relationship between the tone of a syllable and the segmental sound at the syllable’s beginning, called its Onset.  Click here to open up a table with the Maodian Onsets.  This table shows that the sounds that can occur at the beginning of Maodian syllables can be nearly all exponents (except rhotic) of the major classes of obstruent (i.e. stop, affricate, fricative),and sonorant (nasal , lateral, glide). It is also useful to recognise a zero Onset for cases where the syllable begins with a vowel.  The relationship between Onsets and isolation tones (which include citation tones of the type demonstrated on this page) is one of phonotactic co-occurrence. In other words, some Onsets can only occur with some isolation tones and some with others.  Let’s have a look at this relationship.

Looking at the table of Onsets, the first important thing to note is that three categories of stops or affricates are listed for each of the columns except glottal. So for example the bilabial stops have /pʰ/ /ɓ/ /b/. Looking across the columns, it can be seen that two of these stop/affricate categories are voiceless aspirated and voiced (e.g. /pʰ/, /b/), but the third differs. For bilabial and alveolar stops it is voiced implosive (i.e. /ɓ/, /ɗ/), otherwise it is voiceless unaspirated (e.g. /k/, /ts/). If you want to hear some nice implosives, click the following button to hear a word with the implosives /ɗ/ and /ɓ/: /ɗeɓi/ to compare 对比.

The nine Maodian tones divide into two natural classes depending on whether they can occur with the plain (i.e non-imploded) voiced stops and affricates (/b d dz g/ etc.), or with the two other sets (/pʰ ɓ, tʰ ɗ, tsʰ ts, kʰ k/ etc.). These two classes of citation tones are plotted separately in the figure to the left, so that you can see the some of the acoustic tonal properties that underlie the distinction into the two natural classes. This figure is a bit complicated so it needs some explanation before looking at the characteristics of the two classes of tones. In it, the speaker’s mean tonal F0 has been declination-adjusted and converted to semitones relative to the lowest F0 value in his low rising tone (for details of this transformation see my Interspeech paper on tone transcription). The semitone scales are on the left of each panel. The lowest value in the low rising tone was chosen as the reference point because the actual lower boundary of the speaker's modal F0 is indeterminate. It cannot be determined with reference to his offset values in the falling tones, as he prolongs phonation in these to fairly low levels without changing phonation type. The scales to the right are the well-known Chao 5-point values fitted to the highest and lowest semitone values. It can be appreciated that, transformed in this way, most of the speaker’s tones can be assigned a Chao tone-letter description without too much procrustianism. The depressed high fall could be [241], for example, and the low rise [13]. The dipping contour of the mid rising tone would be a problem though. Now we can go back to looking at the two classes of tones …

The citation tones that can occur with the plain voiced stops and affricates are in the figure's right-hand panel; the tones that can occur with the voiceless aspirated and voiceless unaspirated/imploded stops and affricates are in the left-hand panel. You can see that the F0 trajectories of the tones with the plain voiced Onsets are distributed mostly in the lower half of the speaker’s F0 range whereas those occurring with the voiceless aspirated and voiceless unaspirated/imploded Onsets are largely in the upper part, although there is quite an overlap. The clearest F0 feature that differentiates the two classes of tones is its value at onset. The tones occurring with the plain voiced Onsets have a value at about 2 semitones or below; the class of tones with the other two Onset categories onset at values above about 2 semitones. The Maodian tones therefore divide into two natural classes by virtue of both co-occurrence relationships with Onset stops and affricates, and F0 onset value and location of trajectory within F0 range. Quite often the terms upper and lower REGISTER are used to refer to these two classes. The classes correspond of course to the Middle Chinese tonal categories of Yin 阴and Yang 阳. Since there are other correlates of this distinction as well, it may be that REGISTER is more a feature of morphemes than tone. The systematic presence of three categories of stops/affricates at different places has been recognised since Chao Yuen Ren’s 1928 pioneering Studies on the Modern Wu Dialects 現代吳語的研究 as the main typological characteristic of Wu dialects, although, unlike the Maodian speaker, many have voiceless unaspirated values instead of voiced implosives for the third category, and many varieties have lost or are losing the three-way distinction. Note that the presence of voiced implosives in the categories that can occur with the Maodian upper register tones makes it impossible to use the feature [+/-voice] to characterise the two sets of obstruents that co-occur with the two tonal registers. One needs a feature that will classify voiced implosives, voiceless unaspirated and voiceless aspirated stops and affricates against voiced stops and affricates.

In many Wu dialects, the fricatives also participate in the relationship with the tonal register, a voiceless set, for example /f s ɕ /, co-occurring with the upper register tones and a voiced set - /v z ʑ / with the lower register tones. It is not yet clear whether this can be said of this Maodian speaker, as his morphophonemically voiceless fricatives have a tendency to show hold periodicity intervocalically. It may be therefore that he has only one series of fricative phonemes, or that he is losing the distinction between voiced and voiceless fricatives. That is why I have put the /voiced fricatives/ in brackets.

The last interesting aspect of the relationship between Maodian Onsets and tones concerns the nasals. In many Wu dialects, nasals and laterals occur very commonly with the lower register tones, but are only rarely found with the upper register tones. Because of an interesting historical change common to many Wuzhou varieties this is not the case for Maodian. Historically, voiceless unaspirated alveolar and bilabial Onset stops have become nasals before a tautosyllabic nasal coda. There is a nice example in the upper-mid level citation tone in the word for east 东/nʊŋ/, which Ballard, in his 1969 Phonological History of Wu p. 86, reconstructs as Proto-Wu *toŋ. Its /n/ Onset has come from Proto Wu *t which has changed to /n/ before the velar nasal coda. Other examples, from the speaker’s disyllabic data, are: /nɐŋ/ stool 凳 < *taŋ, /nia/ a scholarly book 典< *tien and /mɑ/ board 板 <*pan.

This historical change means that, unlike many other Wu dialects, Maodian nasals can be commonly found with the upper register tones as well as the lower. The last two examples lack a synchronic nasal coda, although cognates in other varieties make it clear they had one in Proto Wu. This shows that the nasalisation of the Onset occurred before the loss of the coda.

Given the division into upper and lower register tones, the speaker’s citation tones can be paired into upper and lower versions of the same or similar contours, and thus a separate contour representation can be extracted. Thus the upper-mid level and lower-mid level tones are examples of an upper and lower register LEVEL tone pair. The upper-mid and lower rising tones are examples of an upper and lower register RISING contour. The high falling and depressed high falling tones instantiate an upper and lower register HIGH FALLING contour. This pairing-by-contour arrangement, which is another characteristic of conservative Wu dialects, is shown in the table below, where the tones are named after their pitch, and their Middle Chinese tonal categories are also given in brackets. Note that the low-rising tone is a merged reflex of Middle Chinese categories IIb and IVb (Yangshang and Yangru). The pairing idea should not be pushed too far. It is not clear whether the upper and lower short stopped tones actually have a contour, for example, and the relationship between the upper and lower register versions of the FALLING contour is not the same as that between the upper and lower versions of the LEVEL contour. The lower-mid rising tone also lacks a lower-register counterpart. It is of course most similar in pitch to the upper-mid rising tone and I suspect that Ballard has caught this speaker in mid merger with the upper-mid rising tone, thus paralleling the merger between IIb and IVb.

PITCH CONTOUR level rise high fall short stopped rise

















(IIb Yangshang + IVb/Yangru)





Now, finally, back to the Onsets. I have described them above as phonemic units, and they are represented thus in the Onset table. But their realisation also needs a comment. The most important thing is that the plain voiced stops/affricates (the ones that co-occur with the low register tones) show variation word-initially between coincident and lead realisations, with the former more common. Click the following button to listen to an example of variation in VOT in plain voiced stop phonemes word-initially. You will hear two clear VOT lead realisations [d] and [b] in /di / electricity and /bɐn/ ill, followed by two coincident realisations [p] in /bɐn/ flat and /ba/ white.