Conclusions Introduction e book 66 Castro Sui Dialect

Figure 7.12. MDS plot of Sui and Kam LDs. Our main question here is the positioning of Yang’an dialect, represented here by TN, in Kam-Sui. The clustering algorithms all indicate that TN belongs to the Sui cluster as in figure 7.11. However, TN’s grouping with the other Sui varieties is far less certain in the MDS plot figure 7.12. Out of the four Sui dialects, it is clearly the closest to the Kam dialects. The position of Yang’an in relation to Kam is far clearer for phonetic distance than for lexical difference see chapter 6, section 6.3.1. Working from the hypothesis that Yang’an historically belongs to the Kam branch, this indicates that pronunciation may be more resistant to change than lexicon, and that LD, especially using phonemicised transcriptions, may be more useful to the historical linguist than lexical similarity counts.

7.6 Conclusions

In this chapter we have examined phonetic similarity among the Sui dialects through application of the Levenshtein distance algorithm. We have shown that broad dialect groupings indicated by clustering algorithms applied to LD calculations are largely in agreement with dialect divisions indicated by lexical similarity. In the case of Sui, LDs reveal four clear clusters: 1 Yang’an; 2 Pandong; 3 Sandong Central, Western and Eastern; and 4 Sandong Southern. Delineation of clusters within group 3 is not so clear due to higher levels of phonetic heterogeneity in these areas. We have also shown that LD applied to phonemicised data is at least as useful, if not slightly more so, for indicating dialect clusters as LD applied to narrow, phonetic data. Overall dialect clusters revealed using both methods are the same. However, the position of SW within the Southern Sui cluster is far clearer when using phonemicised data. Furthermore, clustering based on the use of phonemicised data indicates that Yang’an is an outlying cluster, which is backed up by historical comparative work chapters 3, 4 and 5. The use of narrow, phonetic transcriptions shows that Pandong is the outlier, which is more consistent with clusters based on lexical similarity chapter 6. Finally, by applying LD to both Sui and Kam data, we have shown that Yang’an TN is marginally closer to a variety of Southern Kam Yandong than it is to most other Sui dialects. Despite this, clustering algorithms still indicate that Yang’an belongs to Sui rather than Kam, bolstering the notion that Sui, as a whole, is phonetically distinct from Kam and should be viewed as a distinct dialect group. 155 8 Interdialect Intelligibility Catherine Ching-yee Castro, Emily Wang Yongzhen, Andy Castro

8.1 Introduction

A key question which we sought to answer as part of this survey was how well speakers of different Sui dialects can actually understand each other. This very practical question is of value for several reasons. Firstly, levels of intelligibility are another measure of interdialectal similarity and relatedness, thus they help to inform our overall understanding of the Sui dialect landscape. Secondly, intelligibility is a key factor in defining the nature of a speech variety. According to ISO 639-3, lack of mutual intelligibility between two speech varieties is often a decisive factor in classifying them as two separate “languages” SIL 2013. 1 Thirdly, the answer to the intelligibility question is vital for language development purposes. By necessity, a dialect must be chosen on which to base an orthography, literacy primers, teaching materials and a corpus of literature audio or written for use by the speech community. The dialect which is best understood by the greatest proportion of the community is usually the best choice for such a standard. Knowledge of which speakers cannot understand this chosen standard is also important. Special attention must be given to such people either by empowering them to develop their own materials based on their own dialect, or by helping them to bridge the gap between the standard and their own dialect. Until a few years ago, publications relating to the Sui language universally agreed that there are high levels of intelligibility between the three main Sui dialects, Sandong, Yang’an and Pandong Zhang Junru 1980:75, Weng 2001:563, Zeng 2004:42, Edmondson et al., 2004:48. Pan and Wei 2004 summed up the prevailing view in this way: The internal differences between Sui dialects are very small …, particularly in terms of Sui phonetic tones, which are practically the same for all tone categories… therefore, Sui is not divided into ‘dialects’, only into ‘vernaculars’, and speakers of Sui from all places are able to communicate with each other. Pan and Wei 2004:385, translated by the author With respect to Sandong dialect, although some scholars have acknowledged some phonetic and lexical differences between Sandong speech varieties, all have claimed that levels of mutual intelligibility within this dialect group are high Stanford 2007:19, Pan and Wei 2004:390. In 2009, this view was challenged by Zhang Zhenjiang 2009 who interviewed Sui speakers in Shuiyao township SY, Southern subdialect of Sandong, Libo county. He wrote that during our survey we found the situation to be a little different [from that described in previous literature]. The Sui in Shuiyao consider that their own language is clearly different from Sandong Sui and they say they can only understand 60–70. They often cite their language differences as evidence to show that they are a separate community from Sandong Sui. Zhang Zhenjiang 2009:52, translated by the author He then quotes a retired first school teacher from Shuiyao who said, “Sui people from Shuiyao who have never been to Sandong before find Sandong Sui extremely difficult to understand.” The authors also report difficulties in communication between speakers from Libo county Southern and Sandu county Central and Western varieties as well as between Pandong and Sandong dialects. Partly in order to investigate these conflicting claims, as well as to address the issues described earlier, we decided to conduct intelligibility testing across the Sui region. We used a type of Recorded Text Test RTT based on translated sentences, described in detail in section 8.4. We tested four speech 1 According to ISO 639-3, “Two related varieties are normally considered varieties of the same language if speakers of each variety have inherent understanding of the other variety that is, can understand based on knowledge of their own variety without needing to learn the other variety at a functional level.” SIL, 2013 varieties: Central GC, Southern SY, Pandong PD and Yang’an LW. Our results visualised in section 8.6, figure 8.6 below show that, as expected, Central Sui was by far the most widely understood variety, with intelligibility levels of over 90 in most areas. Speakers of Southern Sui in Libo county, however, generally understood less than 75 of the Central Sui text. Pandong dialect speakers did even worse, on average understanding less than 50. Comprehension of the Southern Sui test was mixed. Speakers living in SD Central and Southern Sui areas understood it well. Sui speakers from Pandong, Western and Eastern dialect areas did not understand much more than 60. Pandong dialect showed low levels of intelligibility in most areas. Only two places understood more than 70: Jiaoli JL, also Pandong dialect area and Dujiang DJ, Eastern. Finally, the Yang’an dialect test only showed high intelligibility over 80 in geographical regions close to Yang’an, presumably due to acquired intelligibility see section 8.2 below. Speakers of Pandong, Eastern and Southern dialects mostly only achieved scores of around 50 or lower on the Yang’an test. These results confirm that inherent intelligibility among the Sui dialects is mixed and generally not as high as Sui scholars had previously thought. Central Sui is the most widely understood variety and is thus an excellent lect on which to base a standard. However, Sui speakers living in Duyun municipality Pandong dialect area and Libo county Southern Sui area appear to struggle with the standard lect and may need materials in their own dialects. Our results come with the caveat that they only estimate levels of “initial contact intelligibility”. The fact that Sui dialects are very closely related genetically with the exception, perhaps, of Yang’an dialect suggests that cross-dialectal intelligibility could be acquired fairly quickly. The authors’ own observations of Southern Sui speakers learning to communicate with Central and Western Sui speakers confirms this hypothesis. An interesting avenue for further research would be to test rates of acquisition of some dialects compared to others. Such a measure might give a more accurate picture of inter-dialect relatedness than a simple measure of initial contact intelligibility.

8.2 Measuring intelligibility