structures. Their absence from lexical representations makes epenthetic segments strong candidates for unmarked features as insertion might be expected to provide the least marked features.
De Lacy (2006:79) also states that “markedness reduction (i.e. to be less marked) is often the driving force in determining the feature content of epenthetic consonants.” Coronal PoA is often viewed as the most unmarked, so it is reasona-ble for the inserted nasal suffix to be [-n]. At the present, why the unmarked PoA emerges in the nasalized diminutives has been definite; how can this markedness hierarchy be put into the POC analysis? The hierarchy can be expressed by the three PoA constraints in (16a) and the pair-wise interactions among them in (16b).
The interactions establish the substantial grammatical restrictions for [n]-suffixed diminutives in the Wu and Hui dialects. They can be placed into “…” in (15), and reproduced here as (17).
(16) (a) Constraints for PoA: *labial, *dorsal, *coronal
(b) Pair-wise Constraint Interaction for [labial > dorsal > coronal]
*labial >> *dorsal
*labial >> *coronal
*dorsal >> *coronal
(17) IDENT-SD[nas] >>
As for the second issue, if [n] is the least marked, why does [ŋ] emerge?
Note that the suffix [ŋ], to my knowledge, exists only in the Wu dialects. The ar-guments for [ŋ] can be expressed as follows. First, the suffix [ŋ] may bear a close
relation with the historical development of nasal rimes. As previously stated, nasal codas are merged predominantly into [ŋ] in Wu, which may motivate the diminu-tive suffix to surface as [ŋ]. Second, Cao (2002) states that nasals are key ele-ments for recognizing nasalized diminutives, and more nasality will surely lead to much easier recognition. As far as nasality is concerned, [ŋ] is stronger than [n] on account of more airflow in the nasal cavity during production. On the basis of Ohala and Lorentz (1977), [m], [n] and [ŋ] can be distinguished by their different size of the oral cavity, as shown in (18).
(18) Vocal Tract Shapes for [m], [n], [ŋ] (Ohala and Lorentz 1977:586)
a. [m] b. [n] c. [ŋ]
According to (18), when nasals are produced, the airflow goes primarily into the nasal cavity (as marked by ●) because of the lowered velum. However, when the oral cavity is larger, more airflow deviates into the oral cavity, and thus more consonantal constriction results. To put it differently, more airflow into the nasal cavity will result in stronger resonance and more salient nasality. If diminutives are to be identified by nasality, it is reasonable that a high degree of nasality in [ŋ]
will increase successful recognition. Third, Wang (1999) claims that the emer-gence of [-ŋ], at least in some dialects, is a later linguistic phenomenon, and is highly related to language contact. He expounds that the strong policy to enforce Standard Chinese brings to other Chinese dialects many linguistic characteristics of Mandarin, one of which is the difference between [n] and [ŋ]. Both [n] and [ŋ]
remain as syllable codas in Mandarin, but they have different phonological be-haviors. Take retroflex diminutives in Beijing Mandarin for instance. When retro-flex diminutives are formed, vowels in the stems are nasalized when [ŋ] is deleted (e.g. kaŋ + r → kẽr ‘a tub’), but vowel nasalization is not observed in the case of [n] (e.g. phan + r → phar ‘a dish’) (Lu 2001, Wang 1991, Xu1993). The survival of the [nasal] feature after [ŋ] deletion also implies that [ŋ] is stronger than [n] in terms of nasality. In addition, the gradual shift from [n] to [ŋ] in nasalized dimin-utives is also a change in progress, as shown in (19). DT = diminutive tone; ~ = nasalization
Zhengzhang (1980, 1981) investigates the nasalized diminutives of the Wu dialects in Zhejiang Province, and discovers a generational difference in the use of diminutive suffixes in the Wenzhou dialect. The old generation prefers [-n], while the young generation favors [-ŋ]. Such findings undoubtedly side with the viewpoint of the linguistic influence from Standard Chinese.
The third issue is how to represent the competition between [-n] and [-ŋ]? If [-n] has the least marked PoA, why does [-ŋ] emerge? On the contrary, if [-ŋ] has the strongest nasality, why does [-n] surface? In fact, the variability between [-n]
and [-ŋ] can also be expressed by a relation of crucial nonranking, as shown in (20).
(20) Transitional Stage *labial >> *coronal
*labial >> *dorsal
*dorsal >> *coronal
X. [-n]
Y.
*dorsal ～ *coronal
*labial >> *coronal
*labial >> *dorsal *labial >> *coronal
*labial >> *dorsal
*coronal >> *dorsal
[-n] or [- ŋ] Z. [- ŋ]
On the basis of (20), constraint rankings are totally ranked for [-n] in Stage X (i.e. *labial >> *dorsal >> *coronal) and for [-ŋ] in Stage Z (i.e. *labial >>
*coronal >> *dorsal). Nevertheless, the ranking of *coronal and *dorsal is unde-cided in Stage Y, so both [-n] and [-ŋ] are possible, like the case in the Wenzhou dialect. Additionally, the variability between [-n] and [-ŋ] can also be regarded as a conflict between the PoA markedness and the influence form language contact.
When the former beats the latter, [-n] dominates [-ŋ]. If the latter is stronger than the former, [-ŋ] surfaces. Besides, the [-n] suffix is widely observed synchroni-cally, but the gradual change from [-n] to [-ŋ] can be predicted as a result of the leading linguistic impact from Standard Chinese. This view can also be captured by traversing from Stage X, via Stage Y, to Stage Z, as schematically shown in (21).
(21) Transitional Stage
Y. X. *dorsal >> *coronal [-n]
*dorsal～*coronal
[-n] or [-ŋ] Z. *coronal >> *dorsal [-ŋ]
Amusingly, the shift from [-n] to [-ŋ] also arises two additional theoretical issues about the PoA markedness hierarchy of nasals. First of all, although coro-nals are considered less marked than dorsals cross-linguistically, the reverse view (i.e. velar unmarkedness), even if less popular, is also proposed (Harris 1990, Kaye et al. 1985, van OOstendorp 1999, Rice 1996, 1999a, 1999b, Williamson 1977). For example, Trigo (1988) proposes that velars are the least marked con-sonant in coda position. Nonetheless, velar unmarkedness is rejected by de Lacy (2006) who states that “the least-marked PoA feature – [glottal] – is usually pro-duced with glottal constriction, but will also be propro-duced with velar~uvular oc-clusion as a side effect of nasal production (de Lacy 2006:369),” resulting in a glottal nasal [N], so-called ‘placeless nasal stop’. [ŋ] and [N] are similar in both manner and place of articulation, and their difference lies in consonantal con-striction in the oral cavity. [N] is articulated without any consonantal concon-striction in the oral cavity (de Lacy 2006:37-41). The similarities between [ŋ] and [N] usu-ally cause descriptive confusion in the literature. That is, a placeless [N] is usuusu-ally transcribed as a velar [ŋ], which mistakenly gives rise to velar unmarkedness.21 If this is the case, the nature of [ŋ] in nasalized diminutives should be further scruti-nized articulatorily and acoustically. Probably, it may be a placeless [N]. If it is an [N], it can developmentally pattern with [] in glottalized diminutives, another kind of diminutives that is extensively observed in such southeastern Chinese di-alects as Min, Wu and Yue.22 There is also another possibility that [ŋ] is develop-ing towards a placeless [N], like the development of [k] into [] in glottalized
21 For detailed discussion on this issue, please refer to Chapter Eight of de Lacy (2006).
22 According to Chen (1992b, 1999), the glottal stop in glottalized diminutives derived historically from [k] which was, in turn, a degenerate form of the syllabic diminutive suffix [kiaŋ] in Min. [k]
and [] are phonetically similar except that [] lacks an overt supralaryngeal specification (i.e.
placeless). For more discussion about this issue in glottalized diminutives, please see Cheng (2009).
minutives (i.e. [k] → []). Obviously, whether this issue can be attested waits for more research endeavors in the future.
Second, markedness hierarchies are relative, depending on different lan-guages and linguistic contexts. Unmarked features can vary or be identified, de-pending on the system in which they occur (Rice 2007, Trubetzkoy 1969). As a result, according to de Lacy (2006), if there is a universal PoA markedness hier-archy (i.e. [dorsal > labial > coronal]), which PoA eventually surfaces as the least marked depends, to some degree, on other factors, such as language-specific and cross-linguistic occurrence frequency, perceptual salience, ease of articulation, and historical and socio-cultural conditions. Evidently, the case in this study (i.e.
[n] → [ŋ]) belongs to the last factor.
6. Conclusion
Chinese languages are often characterized as analytic languages and hence exhibit few alternations induced by affixation processes. This explains why most literature is descriptive, and little theoretical attention is paid to Chinese affixa-tional phonology. This study brings OOC and POC into the analyses of the nasal-ized diminutives in the Wu dialects, and establishes a connection between de-scriptive data and theoretical analyses. Not only does it bridge the gap between synchronic variation and diachronic change, but also captures the gradual transi-tions by two sets of crucial nonrankings (DEP-SD ～ IDENT-SD[nas] and IDENT-SD[nas]～MAX[nas]), both of which are empirically supported from the synchronic Wu dialects. Besides, whether [-n] or [-ŋ] surfaces as the nasal dimin-utive suffix results from the competition between the PoA markedness (i.e. dorsal
> labial > coronal) and language contact with Standard Chinese, and the variabil-ity between [-n] or [-ŋ] can also be caught by a set of crucial nonranking (i.e.
*dorsal～*coronal). On the whole, the transitional stages of the nasalized diminu-tives and the gradual change from [-n] to [-ŋ] are under the regulation of POC.
Besides, this study also shows that the variation of diminutives is neither random nor abrupt, but systematic and gradual.