A complete solution to a phonology problem goes beyond "finding the right answer". You also need to explain to your audience what you are proposing, and convince them that your proposal is worth taking seriously. Therefore, your argumentation is very important.
A phonology write-up or paper typically has the following structure:
- State, briefly, what your proposal or main idea is
- Make arguments to support your proposal
- If relevant, explain why your proposal is better than a relevant alternative
This web page focuses on item 2: making a phonological argument.
The best way to make an argument in phonology varies somewhat depending on what topic, model, or phenomenon you are working with. But here are some general guidelines that will help you make your argument clear, well-organized, and convincing.
- Describe the pattern and justify your description
- Present a formal analysis
- Exemplify your analysis insightfully
Describe the pattern and justify your description
Before you launch into a technical phonological analysis, you need to make the case for why you need it. Therefore, you need to begin your argument by describing what is going on in the data set.
Description before analysis
-
Did you find a pattern in the data? If yes, describe the pattern
you found, in a phonologically relevant and insightful way. If you see
no pattern, state what kinds of phonologically relevant patterns you
looked for but failed to find.
It is a good idea to include both a fairly concrete level of description and insightful generalizations about natural classes or larger patterns.
Justification — use evidence from the data set
-
Whenever you claim to have found a pattern, you must justify your claim
by showing all of the relevant examples from the data (or at
least a representative sample, for large data sets). The examples
should be organized in a phonologically insightful way, preferably in
list, table, or chart format for easy comprehension. The convention
is to list a data item along with its gloss (translation).
Conversely, whenever you claim not to have found some specific type of pattern, provide at least a few examples showing that the pattern does not hold.
Example (for a case of allophony or morpheme alternation; change as needed for making other types of arguments)
- Describe: What group of sounds changes? Into what? Under what
circumstances?
- Justify: Show lists or charts of examples to prove that...
- the sounds in the natural class you name actually do what you say they do
- the sounds you have excluded from your class need to be excluded, because they do not participate in the phenomenon in question
- any other restrictions you have placed on the phenomenon (such as the environment where it occurs) are indeed necessary
Present a formal analysis
What is a formal analysis?
-
A formal analysis uses the representations
and operations that we have included in our phonological model in order
to account for the patterns in the data set. Think of a formal analysis
as something like a computer program or a recipe that tells the
phonological system how to apply the analysis you have developed.
If you propose a rule, a new constraint, etc., be sure to clearly state or define it before using it. Also, remember that formalisms (rules, constraints, etc.) should always be paraphrased in ordinary prose as well as being stated in formal phonological notation.
Why include a formal analysis at all? Why isn't it enough just to describe the pattern?
-
We analyze phonological patterns in languages in large part because
we are trying to test and refine our model of phonology. Even if you
as a human linguist "see what is going on" in a data set, it is still
essential to determine whether our current model allows us to account
for the phenomenon, or what questions the phenomenon might
raise for our model.
Exemplify your analysis insightfully
Show how your analysis works, and confirm that it works as needed, by applying it to example items from the data set. The rules, parameter settings, or constraint rankings you have proposed should automatically apply in a way that is consistent with all of the information available in the data set.
When you exemplify your analysis, it is important to choose your examples insightfully — don't be redundant. Choose a small number of items that are representative of the patterns in the data. It is not necessary to show the analysis at work on multiple items of the same general type (the place to show that there are many examples that follow one general pattern is in the description/justification stage of your discussion).