Determining The Distribution of Segments in A Language: Linguistics 200 Phonology

Linguistics 200 Phonology
Determining the distribution of segments in a language

In phonological analysis, we often want to determine whether two sounds:
1. belong to two separate phonemes

2. are allophones of a single phoneme
We can gather evidence about segments’ status with respect to the phonemes of a language by
examining their distribution in that language. That is, what are the environments in which
each sound appears? Are these environments predictable or unpredictable?
We begin by making a list of the environments for each sound and then examining the lists for
patterns.
1. Contrastive distribution
We can prove that two sounds belong to two separate phonemes if we can show that they are in
contrastive distribution.
Two sounds are in contrastive distribution, also known as overlapping distribution, when they
appear in (at least some of) the same environments. In other words, given information about
the phonological environment, it is unpredictable which of the sounds under investigation will
appear in that environment. Unpredictable information can’t be assigned by a phonological
rule — it must be stored in the underlying representations (URs) of morphemes in the mental
lexicon. On the other hand, when a phoneme has multiple allophones, they are assigned by
phonological rules. Therefore, two sounds that are in contrastive distribution can’t possibly be
allophones of the same phoneme (although see below on the distribution pattern known as
neutralization).
There are two ways to show that sounds are in contrastive distribution.
• Find at least one minimal pair of words distinguished only by the sounds in question. A
minimal pair is quick and easy proof that the two sounds are contrastive — after all, the
two sounds are used in the language to make contrasting words.
• Even if you can’t find a minimal pair, you may be able to demonstrate that the two
sounds appear in overlapping environments; that is, that nothing about their
environments can be used to predict which sound will appear. This requires more careful
discussion, showing that neither the preceding environment, nor the following
environment, nor both taken together, can be used to predict which sound will appear.
2. Complementary distribution
If two sounds are in complementary distribution, we can show that they are allophones of the
same phoneme. (To think about: When multiple pairs of sounds are in complementary
distribution with each other at the same time, how do we decide which of them to pair up?)
1
Two sounds are in complementary distribution when they never appear in the same
environment. That is, one sound appears in a certain set of environments, and the other sound
appears in the complement of that set of environments — so that between the two sounds, they
cover all of the possible environments of the language. (Extending this, if a phoneme has more
than two allophones, we may find three, four, or more sounds that all appear in distinct
environments, dividing up the labor of covering all the environments in the language.)
When two sounds are in complementary distribution, it is predictable which of the two will
appear in any given phonological environment. Thus, we will be able to write a phonological
rule to determine which allophone appears where. Some examples:
• Sound A appears only at the ends of words. Sound B appears elsewhere (i.e., in all other
environments that we know of).
• Sound A appears only when it precedes a fricative. Sound B appears elsewhere, including
before stops, vowels, liquids, nasals, glides, etc.
• Sound A appears only when it follows a sonorant and precedes a vowel. Sound B appears
elsewhere — whenever those conditions are not met.
These made-up examples illustrate some of the things that need to be considered when we look
for crucial environmental factors. Sometimes we need to look at word boundaries. Sometimes
we need to look at just the preceding sound, or just the following sound. Sometimes both the
preceding and following sounds are relevant. Also, sometimes the crucial natural class in the
environment is a very general one, like “vowels” or “fricatives,” while other times we will need
to be more specific, as in “high vowels” or “voiced coronal obstruents”.
Very often, the environment for one allophone can be stated in terms of natural classes, but the
other environment cannot. In such cases, we can refer to the uncharacterizable, “left-over”
collection of environments as “elsewhere” (a concept that is important in phonological
analysis).
• A note about long-distance environments:

As we saw in the Turkish problem, we can’t always look only at the immediately adjacent
segments to determine what phonological environments are relevant. Vowels, in
particular, can be affected by nearby vowels even if consonants intervene. But don't try
too hard to find long-distance environments for consonants; such cases are comparatively
rare and almost always involve assimilation or dissimilation of a specific feature (well-
known cases include [±nasal], [±lateral], [±anterior]). We would not expect to find, say, a
language where the voicing of a word-final consonant depends on whether or not the
word-initial consonant is a labial fricative.
As always, be as general as possible when stating an environment. Don't refer to “high front
tense vowels” when “high vowels” would be sufficient. And remember that the word or is a red
flag when you are trying to characterize an environment — it warns you that you may not have
correctly identified the appropriate natural class. Finally, be sure to formalize your analysis by
stating the natural class in terms of features.
2
3. Neutralization
Neutralization is a situation in which two sounds belong to separate phonemes — their
distribution is demonstrably contrastive — but, in some particular environment (context),
phoneme #1 has an allophone that is identical to phoneme #2 (or more accurately, to phoneme
#2's allophone in that environment).
An example of neutralization would be a language in which /p/ and /b/ are in contrastive
distribution in most environments, but /p/ has a [b] allophone when it follows a nasal. Thus,
following a nasal, both /p/ and /b/ surface as [b]: the /p/-/b/ contrast is neutralized following a
nasal.
Neutralization is most clearly seen in cases of morpheme alternations, where we can directly
observe two different phonemes “turning into” identical-looking allophones in a particular
context. We will discuss morpheme alternations in an upcoming class.
4. Free variation
In a case of free variation, two sounds belong to the same phoneme, but their distribution is not
determined by their environment. Instead, either one of the two sounds may be freely used in
any environment (although the meaning does not change when one of the sounds is exchanged
for the other — this is why we know the sounds are not distinct phonemes).
Often, free variation is not truly “free,” because the choice between the two options may be
determined by non-phonological factors like sociolinguistic context or speech rate. In any case,
because the focus of this course is on phonological factors, we will probably not discuss many
cases of free variation.

Determining The Distribution of Segments in A Language: Linguistics 200 Phonology

Uploaded by

Copyright:

Available Formats

Determining The Distribution of Segments in A Language: Linguistics 200 Phonology

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Determining The Distribution of Segments in A Language: Linguistics 200 Phonology

Uploaded by

Copyright:

Available Formats

Linguistics 200 Phonology

Determining the distribution of segments in a language

1. belong to two separate phonemes

• A note about long-distance environments:

You might also like