The aim of the phonological analysis is, firstly, to determine
which differences of sounds are phonemic and which are non-phonemic
and, secondly, to find the inventory of phonemes of the
language.
As it was mentioned above, phonology has its own methods of
investigation. Semantic method is applied for phonological analysis
of both unknown languages and languages already described. The
method is based on a phonemic rule that phonemes can distinguish
words and morphemes when opposed to one another. It consists in
systematic substitution of one sound for another in order to find
out in which cases where the phonetic context remains the same such
replacing leads to a change of meaning. This procedure is called
the commutation test. It consists in finding minimal pairs of words
and their grammatical forms. For example:
pen [pen]
Ben [ben]
gain [gain]
cane [kain]
ten [ten]
den[den]
Minimal pairs are useful for establishing the phonemes of the
language. Thus, a phoneme can only perform its distinctive function
if it is opposed to another phoneme in the same position. Such an
opposition is called phonological. Let us consider the
classification of phonological oppositions worked out by N.S.
Trubetzkoy. It is based on the number of distinctive articulatory
features underlying the opposition.
1. If the opposition is based on a single difference in the
articulation of two speech sounds, it is a single phonological
opposition, e.g. [p]-[t], as in [pen]-[ten]; bilabial vs.
forelingual, all the other features are the same.
2. If the sounds in distinctive opposition have two differences in
their articulation, the opposition is double one, or a sum of two
single oppositions, e.g. [p]-[d], as in [pen]-[den], 1) bilabial
vs. forelingual 2) voiceless-fortis vs. voiced-lenis
3. If there are three articulatory differences, the opposition is
triple one, or a sum of three single oppositions, e.g. [p]- [ð], as
in [pei]-[ ðei]: 1) bilabial vs. forelingual, 2) occlusive vs.
constrictive, 3) voiceless-fortis vs. voiced-lenis.
American descriptivists, whose most zealous representative is,
perhaps, Zellig Harris, declare the distributional method to be the
only scientific one. At the same time they declare the semantic
method unscientific because they consider recourse to meaning
external to linguistics. Descriptivists consider the phonemic
analysis in terms of distribution. They consider it possible to
discover the phonemes of a language by the rigid application of a
distributional method. It means to group all the sounds pronounced
by native speakers into phoneme according to the laws of phonemic
and allophonic distribution:
1. Allophones of different phonemes occur in the same phonetic
context. In this case their distribution is contrastive.
2. Allophones of the same phoneme(s) never occur in the same
phonetic context. In this case their distribution is
complementary.
There is, however, a third possibility, namely, that the sounds
both occur in a language but the speakers are inconsistent in the
way they use them, for example, калоши-галоши, and [‘ei∫э - ‘егжэ].
In such cases we must take them as free variants of a single
phoneme. We could explain the case on the basis of
sociolinguistics. Thus, there are three types of distribution:
contrastive, complementary and free variation.