The Dong language was surveyed from early 2002 till the middle of 2004. The purpose of the survey was two fold. The primary purpose was to determine the level of intelligibility between the different dialects of the Dong language. The secondary purpose was to obtain wordlist data from all the dialects for continued analysis by linguistic researchers in years to come.
In the years since the survey concluded, many of us have been interested to see the results, particularly of the intelligibility survey. Since the results have not yet been published and so many of us are interested to see them, the survey manager has agreed to provide the basic results for DongTeam.org for the time being, while the publication process continues.
The survey project manager has provided the actual testing scores of the intelligibility survey and a simple analysis of those scores to help us understand them.
A bit of a primer is needed to begin to read and understand these results. You may wonder why such a long survey covering so many locations and taking so much time would do no more than half fill a small chart with numbers. Think of it as grades at school: a semester of work, daily study, and more homework than you want to remember are all summed up in one grade for one class. These results are much the same.
The scores are on a 100% scale. Though "intelligibility" is hard to pinpoint on a scale, 70% seems to be about the level where another dialect is intelligible. So, just for explanation, the 60% range will likely not be understandable, but a listener could pick out some things, and the 90% range would mean almost perfect understanding.
Here are the survey results. The town or village names along the left side are the areas listening to an intelligibility test (recording). The initials along the top correlate to the names along the side, and you will notice there a smaller number of places (meaning a few places listened to tests from other areas but did not have a recorded test made from their own area). The top initials are the tests themselves, or said differently, they are the recordings of speech from a particular area. So, when the Chejiang folk listened to the Shangzhong (SZ) recording, they scored at 72.
So, these data can answer one of two questions: how much a location can understand others and how much a location can be understood by others. If you want to know how well Shuidong understands other areas, you find "Shuidong" on the left side and read the scores they received listening to other areas' tests. If you want to know how well other areas understand Shuidong speakers, find "SD" along the top and look down through their scores.
Though other linguistic data had long suggested such, this intelligibility survey confirms the centers of the northern and southern dialects as Shuidong and Chejiang. For instance, look at the Shangzhong scores. Though geographically closer to both Shuidong and Xiudong speakers and itself in Liping, which would place it within the Guandong sub-dialect area, it actually scored higher on the Chejiang test. Though digging through these numbers and comparing the places' results will yield much similar information, a quick look at a couple places will give you a good starter. Shuidong is clearly the strongest northern dialect when you look at the Zhongzhai scores, and Chejiang is clearly the southern dialect center when you look at the Linxi scores. And Qimeng seems to be the dividing line between northern and southern speakers.
In the raw scores above, you will notice in the yellow boxes that no area actually scored a 100% on their own test. Curious. Surely they understand themselves perfectly. Less than perfect scores could be for one of two reasons. One, the listeners did not quite understand how to do the test and did not test well. That is unlikely, because each listener had to pass pre-testing, proving their understanding of how to take the test. The second, and more likely, reason is that the test itself could have questions that were too difficult. Assuming this to be the cause, we have adjusted the scores based on each test's score in its own home location.
Technically, these adjusted scores will be higher and more accurate, but the change is so small that in most cases the change really does not make much practical difference. Still, the raw scores are probably the most trustworthy.
Two different methods were used to adjust the scores, because it is difficult to determine which is best: multiplicative (multiplying the percentage) and additive (adding points). The multiplicative adjustment is most likely the better and more accurate option.