Linguistics Stack Exchange is a question and answer site for professional linguists and others with an interest in linguistic research and theory. It only takes a minute to sign up.
Sign up to join this community
What is the most precise/accurate/reliable way of determining whether a speech sound has been reproduced correctly by a person?
I am looking at something like perhaps an inspection of a person's speech organs while reproducing the sound through radiographic means, or comparison of waveform produced. But need more precise/detailed suggestions as I am not well-versed in this area.
First you need a standard of comparison as to how the sound should be produced, and even determining that is difficult. Let us say that the standard is how the average female between 20 and 40 in the village produces the sound. There are 3 main ways of measuring sameness: physiologically, acoustically, or perceptually. Physiology is the least accessible and informative, because there is much lower level variation in how people move muscles to make sounds. Something based on sound itself would be more informative. An acoustic study involves measuring known properties of the waveform, and that is usually how we look at sounds. However it would be better, if possible, to compare how people react to a sound, for example asking "does that sound like a native speaker". That does introduce subjective factors (irrelevant prejudices of the listener), but it does not require you to know in advance what all of the acoustic properties are that need to be measured.
A hypothetical ideal strategy would be to collect a sample from the pool used to define the standard and then find the 25% that are most agreed to be "good examples". Then attempt to find a property in the waveform that distinguishes those tokens. You could then investigate withe your specific individual the extent to which he has that property. However, the notion of accuracy is problematic since it assumes something that probably doesn't exist, namely One Correct Pronunciation. It might make sense when dealing with non-native attempts to produce language sounds.
Ask a linguist who speaks the language. Linguists have training in making such normative judgments. (This is similar to the position taken in SPE on phonological fact.) If you can't find such a linguist, then arrange for a native speaker to be trained as a linguist (I think Ken Hale has done this).
