GPC analyses

People can have valid differences of opinion regarding GPC analysis. Different people even have different opinions as to how many salient sounds, or phonemes, any given language has. There are splitters and there are lumpers. Different regional accents can account for some of these differences. Borrowed words can sometimes add extra pronunciations that are “foreign” to a language. For English, the opinions range between 38-44 sounds. Some argue that r-coloured vowels don’t exist in their framework and others might include more or less of these vowels. Some words can also have more than one legitimate GPC analysis. For example, the word “yacht” could be analyzed as “y_y,o_a,t_cht” or “y_y,o_ach,t_t”. Because English has many words with so-called “silent letters” there can be more than one valid analysis for these words. SynPhony will not dictate any one analysis but will allow for multiple analyses per language.

How do I analyse a language into its GPC components from scratch?

Analyzing a language’s GPC inventory is an iterative process. That means you can refine it over time. You can start by collecting the words used in the language. Then you need to list all the letters and digraphs used in a language. (Digraphs are more than one letter used to represent a single sound. Eg. 'th' in English.) Then you also need to make an inventory of the number of sounds used in the language. Sometimes you can find this information on the web, or a dictionary. Or you might find linguists who are working in that language who can help you. Once you have this basic information you can begin your GPC analysis. It consists of rewriting the orthographic form of the words into their GPC form. I recommend using a database program for this task. Currently I am using a program called Toolbox to assist with my analysis, but it’s possible that SynPhony will produce tools to accomplish the same thing. If you are interested in starting a GPC analysis for another language, please read this page.

What about regional dialects?

Regional differences in pronunciation of a language are a fact of life. The beauty of analyzing a language into its GPC components is that you can capture this variation and thereby easily adapt a reading curriculum to accomodate regional accents. SynPhony will allow for multiple GPC analyses per language to allow a database to be customized to regional accents. The original English analysis in SynPhony was done by Norbert Rennert, a researcher at the Canada Institute of Linguistics, using a general Canadian pronunciation. It is much easier to adapt an existing analysis to another accent than starting from scratch.

How do I adapt an existing GPC analysis to another dialect?

A GPC category identifies all the words which share a sound and which use the same letter(s) to represent that sound. Dialects vary in their pronunciations of certain words. Sometimes there are systematic sound shifts, other times only some words are pronounced differently. To determine if some or all words in a GPC category must be changed it is best to review a list of words from one category and ask yourself the question “Do the words all contain the same sound as the phoneme they have been assigned to?” For example, for English you can look at the GPC chart here, select one category’s vocabulary and ask “Do all these words contain the same sound in my dialect as other GPC categories in the same row?” If not, you are faced with the task of adapting the existing database to another regional accent. Don’t compare the accents of other dialects with your dialect or with the IPA symbol which has been assigned to it. Just because there is a difference in pronunciation between dialects doesn’t mean that the GPC analysis must be changed. What is important is checking to see if all the words in one row on a chart have the same sound in them. This kind of work should be carried out by a native speaker of the target dialect who has a good understanding of the GPC notation method used in the SynPhony system and a good ear for the sounds used in their dialect.

There are various kinds of changes that are possible:

1. Moving a whole GPC category to another row because all the words in this category have the same pronunciation for the phoneme portion. Eg. for the GPC “a_au” as in “aunt”, a Canadian pronunciation pronounces the “au” like the “a” in “cat”, while a British person pronounces the “au” like the “a” in “ball”. The GPC category for that would be different, perhaps “o_au”. This kind of change can be done by a search and replace function with the computer. The person doing this must only make sure that all affected words are pronounced the same way as the new target phoneme.
2. Moving only some of the words in a GPC category to another GPC category. This type of change is more time consuming than the previous change as each word must be changed manually.
3. Its possible that one dialect adds sounds where others don’t. Certain British dialects insert /r/ sounds where there is no “r” orthographically. In these cases, the affected words will have to be identified and given another GPC category. The same process holds true if a dialect drops certain sounds.
4. Its possible that in one dialect certain letters are “silent” and in others they are pronounced separately. Thus, using a hypothetical example, the GPC “d_ld” as in the word “would” could be pronounced differently in 2 different dialects: one pronounces the “ld” as one sound, the other pronounces each letter separately. In this senario, one of the dialects must either split or lump this analysis as a group. Again, care must be taken that all affected words are really all pronounced the same in that dialect. But this kind of change can also be done with a computer, resulting in a quick adaptation.