Linguistic mapping

I caught a link today from a friend to a dialect quiz, which takes either a 25 question or a 140 question test and uses that to produce a map of your linguistic similarity to the rest of the country. For me, the results are for the short quiz:

and for the long quiz:

Neither of these maps is a surprise for me -- I grew up in the suburbs of Buffalo (western New York), went to college in southern California, did my first grad program in northern California, and then moved Lansing (middle of the lower peninsula in Michigan) for my PhD. Interestingly, the site also gives you a list of the 5 American cities most similar and most distant from your results, and Buffalo and Lansing are 3rd and 5th most similar to me, respectively. My linguistic similarities do line up not only with these cities, but the other places I've lived, and the degree of similarity appears to correlate strongly with the amount of time I've spent in these locations (also, both formative periods and current time are high similarity for me, which I would further expect -- I can imagine both a primacy effect, where the first way you're taught to pronounce something/taught a word for something has a larger impact than later ways, and a recency effect whereby you shift your current versions to be closer to what you've been hearing most recently.)

I've since seen a few of my friends posting their own linguistic maps, and I've noticed something striking -- what degree of similarity is represented by a particular color varies wildly from one person to another, based on the dynamic range of an individual's results. My similar ranges from 56.8 to 47.6 on the longer quiz, while the map from one particular friend ranges from 50.1 to 29.4. Therefore, while more of her map looks to be red than is true for mine, if we were to rescale things so that the every map had the same mapping (ha) from score to color, my entire map would be markedly more red-shifted than hers. One interpretation of this is that my diction is closer to Standard American than hers is, and thus has less to differentiate it from its most extremely dissimilar dictions than is true of hers. I suspect that this is probably the case, given what I've been told in the past about where newscaster speech came from and the ways in which it differs from what most people use in my hometown. Another interpretation, if you're an astrophysicist, is that perhaps my map is retreating from the viewer at a non-negligible speed.

However, these sorts of self-assessed linguistic tests always strike me as potentially problematic. Many of the questions on this particular one are going to be easily scored by anyone -- for example, whether your pronunciation of the word caramel has 2 or 3 syllables, or whether you refer to non-alcoholic beverages generically as soda, pop, Coke, soft drinks, lemonade, etc. Those don't need an expert to evaluate. But some of them get into use of the International Phoenetic Alphabet, which is an attempt to classify all of the phonemes across many languages. There was a time period, early on in doing Supplemental Instruction or acting as a Teaching Assistant, in which I strongly considered learning IPA notation in order to ensure that I would pronounce my students' names correctly; they'd only have to correct me once, I'd note down the IPA version of their name, and be fine. Then I looked into the IPA, and found that it largely rests on referencing other words, rather than having a set library of recordings of people pronouncing particular phonemes. If I have the Northern Cities Vowel Shift (which I probably do to some extent based on geography), for example, my perception of the vowels of various words could well be different than is assumed under the IPA. That makes it seem markedly less universal, to me. I'm not entirely confident that my assessment of how I pronounce words would line up with the assessment of an expert evaluating a sample of my speech.

There's still a part of me that thinks I should probably learn IPA notation for personal use in getting names correct, but since my transliteration into IPA may not be the same as someone else's, it makes it much less useful as a way to communicate to other people. I suspect that people who actually work with it professionally have some sort of standard to ensure consistency, but it's not terribly obvious what this is, at least to an outsider like me.