In part 2 we used the naturally occurring distribution pattern of the language to remove any bias towards the linguist’s mother tongue for sounds that could be phonetically written in different ways.
Now’s the fun part: we figure out what sounds should go into a written alphabet and which are just ‘flavours’ of other characters. A basic premise is this: we don’t speak characters in isolation, sounds flow into each other and can cause changes to their neighbours. If we can see that a sound is always caused by a particular environment we don’t need to write it because it’s not a phoneme, a significant sound for the alphabet. It would be an allophone, a special pronunciation of another phoneme in a certain environment.
If on the other hand, we have examples of two different sounds in the exact same environment — two words where the only way to tell them apart is that one character we need to include those characters as phonemes.
“zip” and “sip” for example demonstrate that in English we’ll need both /s/ and /z/ in the alphabet.
A partial example I can think of right now is basic plurals in English. The plural to both dog and cat is -s, making dogs and cats. Pay attention though and you’ll realise that when you say dogs you’re actually saying a “z” [dɔgz] while cats is an ‘s’ [kæts]. (We would actually call this an allomorph, a particular kind of allophone caused by grammatical bits coming together). The rule is that when your word ends in a voiced consonant you say “z” and when it’s voiceless you say “s”.
It’s a partial example because while we can explain the difference when it comes to the plural suffix, we can’t when it comes to “sip” and “zip”. If however, we could always explain why it’s an s or a z, we could “unite” the two phonemes. We’d just pick the one to go in our alphabet and we’d just use that one symbol everywhere an
knowing that’s it’s pronounced differently in certain (predictable, intuitive) places.
We want to find contrasts (like “zip” and “sip”) or find mutually exclusive environents (where we can say it’s always this sound, unless it’s in this place then it’s this) we can explain for our phones. We don’t need to do them all though! A “k” is obviously never going to be a variation of an “m” because they’re too far apart on our phonetic charts. We look for suspicious pairs – pairs of sounds that only differ by a single phonetic feature, like “s” and “z” which differ only by voicing.or like “s” and “ʃ” (an sh sound) which only differ only by where in the mouth they are made.
Sound complex? It’s not too bad as we have nice diagrams that tell us what sounds we’re looking for!
So we get to work in Kovol, here are some examples of our findings:
[i] ≠ [e]
#672 [sindi] “swing”
#323 [sinde] “inside”
[p] ≠ [b]
#609 [pɛlɛg] “not ripe”
#194 [bɛlɛg] “chest”
And on and on we go. Sometimes we don’t have perfect pairs like this and so we contrast sounds where words are sufficiently similar instead.
We found a few environmental changes where we could conclude that we don’t need extra alphabet characters for them (which means a shorter literacy course and easier reading).
[i] = [ɪ]
[ɪ] only occurs before [ŋ], [i] never does. Velar pressure is causing a more open vowel.
The [i] sound (like the ee in “jeep”) seems to be the main vowel and [ɪ] (like the i in “sing”) seems to be an allophone because it only occurs next to [ŋ] (an ng sound). This is similar to English actually.
That’s our hypothesis at least, now we need to gather more data to test that.
The other times we put sounds together were [k] and [ʔ] due to dialect variation, then [o] and [ɔ], and [l] and [r] both due to fluctuation. Fluctuation means that sometimes people use one and sometimes the other, seemingly freely. This one is a bear to prove!
So what do we end up with?
/p/ all environments
/b/ all environments
/t/ all environments
/d/ word medially and finally with /n/
/ʔ/ [k] as a dialect variation
[ʔ] all environments
/g/ all environments
/m/ all environments
/n/ word initial and medial
/ŋ/ word medial and final
/l/ [r] fluctuates freely with [l] word medial
/β/ word medially and finally
/s/ all environments
/j/ all environments
/w/ word initial and medial
/i/ [ɪ] preceding velar C
/e/ word medial and final
/ɛ/ all environments
/u/ all environments
/o/ [ɔ] free fluctuation in all environments
/ɑ/ all environments
/ɑⁱ/ word medial and final
Or in chart format:
Phonemics is a bit of science and a bit of art. As a linguist you have to make certain decisions and suggest hypotheses based on data. New data can come up to throw a spanner in your lovely analysis and some of your decisions on balance may not have been valid, but so far it’s been all about the data.
You’ll notice we’re still using the symbols from the phonetic alphabet. The next step is to pick the symbols we’ll be using – and that’s all political. More on that next time (if you’re still interested in reading that far).