LINGUISTICS 4302W PHONOLOGY I

Example of how to solve a phonology problem:

Ganda Liquids
(Note! This is NOT an example of a homework write-up, this is an example of the process you should go through in
solving a problem.)

The sounds [] and [] are in complementary distribution in one variety of Ganda (Bantu, Uganda): they are members
of one phoneme. The environments each sound occurs in are illustrated by the following forms:

 ‘do’
 ‘fight’
 ‘tell’
 ‘eat’
 ‘sit’
 ‘bride’
 ‘sweet potato’
 ‘hospital’
 ‘Ganda language’
 ‘tongue’
 ‘hear’
 ‘help’
 ‘remember’
 ‘canoe’
 ‘fire’
 ‘whistle’
 ‘ship’
 ‘lightning’
 ‘accuse’
 ‘command’

Which of [] and [] represents the underlying element? State the appropriate phonological rule.

(Halle and Clements 1983)

Identify the problem

This part is relatively easy: the question you need to answer is provided right after the data:
“Which of [] and [] represents the underlying element? State the appropriate phonological
rule.” In order to go about solving the problem, there are several steps you must take.

(1) Break the data up into two classes, one for each allophone. To do this, it is helpful to
make a table listing the phonological environment in which each sound occurs. A phonological
environment simply describes the sounds close to the sounds in question. An important thing to
keep in mind is that phonological processes are very local, meaning that they tend to involve
alternations due to sounds that are very close to the sounds in question. For this problem, you are
asked to discover something about the two sounds [] and [] so you can structure your table like
this:

[] []
o_a i_a
#_w i_o
#_j i_a
#_u e_a
u_a i_a
o_e e_j
#_u i_o
a_i i_i
o_u e_i
u_i e_a
u_i i_a
i_a

What exactly does this table do for you? A crucial part of phonological analysis is discerning
patterns, and a table like this helps you find the patterns in the data. But what do all of these
things in this table mean? First of all, as you can see, the table is divided into two columns: one
for the sound [] and one for the sound []. As we are told at the beginning of the problem, these
two sounds are in complementary distribution – they are both related to the same underlying
sound, or the same phoneme. This phoneme, however, has two different possible pronunciations:
one pronunciation is [], and one pronunciation is [].

Find the pattern

The table here shows us the pattern of when one sound occurs and when the other sound
occurs. For instance, let’s look at the column for the sound []. The very first line in that column
says “o_a”. What does this mean? It means there is a word in the data that contains the sound []
between the sounds [o] and [a]. The blank line (‘_’) stands in for the sound in question - [] – and
the sounds on either side represent the neighboring sounds. So “o_a” is a way to abbreviate the
occurrence of the sound [] in the word [].
The rest of table works the same way. The second line in the [] column says “#_w.” The
symbol “#” stands for a word boundary, so “#_w” means that in the data there is a word in which
[] is at the beginning of the word, followed by [w] (this would be the word []). Spend a
few minutes now and examine the rest of the table to make sure you understand how this works,
and how this helps us divide the data up in the ways needed.
Once you understand how this type of table works, you can move on and continue
solving the problem. You are asked to determine which of the two sounds - [] and [] – represent
the underlying sound. Any time you are asked a question like this, your job is to figure out which
of the two sounds occurs in a predictable environment and which one does not. What exactly
does this mean? Recall that these two sounds are in complementary distribution – that
underlyingly, they are really a single sound, or phoneme, that happens to have two possible
pronunciations. Which of these two possible pronunciations actually occurs in a given word is
not random – it is something that is predictable, and your task is to be able to make the right
prediction.

Which sound is underlying and which is predictable?

Whenever you are searching for an underlying sound, what you are really searching for is
the sound whose pronunciation is not predictable. This is where the table comes in handy,
because by looking at the table you should be able to see which of the two sounds is predictable
and which is not. Let’s do this. Let’s first focus on the first column, for the sound []. What you
want to do is figure out if the sounds on either the right or the left side of [] have something in
common. Let’s first examine what we find to the left of []. We find the following: [o], #, [u], [a].
So then we need to ask the important question Do these form a natural class? The answer to this
is no – there is no way to characterize these things as a natural class. (The idea of a natural class
is something we’ll spend more time on over the course of the semester.) So now we need to look
at what occurs on the right side of []. To the right of [], we find the following: [a], [w], [j], [u],
[e], [i]. Again, this is not a natural class – which tells us something very important: the sound []
does not appear to occur in a predictable phonological environment, because we find no natural
class of sounds to its left or to its right. In order to be sure of this, however, we must also
examine the column for the sound [].
Let’s first look at what sounds occur to the right of []. To the right of [], we find [a], [o],
[a], [j], [i]. This is not a natural class, so we must keep looking – this time to the sounds that
occur to the left of []. To the left of [], we find [e] and [i]. Ah hah! A natural class! These two
vowels, [e] and [i], are both front vowels. It is impossible to give this type of label to the group of
sounds that occurs to the right of []; likewise, as we saw, it is impossible to give this type of
label to the groups of sounds that occur to the left and right of []. But the sounds to the left of []
do form a natural class. What this means is that the sound [] occurs in a predictable environment
– that is, following the group of sounds known as front vowels (the natural class consisting of [e]
and [i]), we only find [] – note that the sound [] never occurs after a front vowel. You can
verify this by checking your table – and you should verify this to make sure that we haven’t
jumped to the wrong conclusion. Doing so, we find that [] occurs after [o], #, [u], [a], but never
after a front vowel, so our conclusion seems to be correct.

So the underlying sound is…? And the predictable one is…?

One of the sounds occurs predictably – and we know that of the two sounds, the
predictable one is []: it occurs only after a front vowel. Of the two sounds, the one that is
underlying is the one whose occurrence is not predictable – in this case, the sound []. You can
think of the unpredictable, or underlying, sound, as the one that occurs everywhere, except in the
one predictable environment where the other sound occurs. Sometimes this sound is called the
‘elsewhere’ case – this sound is also known as a phoneme. The next step is to take your
knowledge about the sounds [] and [] in Ganda and formalize it through a phonological rule. A
phonological rule is part of the grammar of the language.

Formalizing a rule: the beginning

In the Ganda case here, we know that the phoneme // is pronounced as [] before front
vowels, and is pronounced as [] everywhere else.. Notice, by the way, the different use of
slashes (/,/) and brackets ([,]). Slashes are used to denote underlying, or phonemic, material, and
brackets are used to indicate the actual pronunciation of a sound. Don’t confuse the two!
Our rule, then, will start out looking something like this:

//  [] / V __
|
[-back]

The way we read a phonological rule like this is as follows: the phoneme on the left side of the
arrow (in this case, //), is pronounced as the sound on the right side of the arrow (in this case,
[]), in the following environment: following a vowel that is [-back] (i.e., a front vowel). The rule
predicts that any time the lexicon contains a word where // occurs after a front vowel, the word
will be pronounced with []. In all other cases (when the phoneme // is not preceded by a front
vowel), no change occurs, and // is pronounced as []. As far as the next step in rule
formalization, we’ll focus on that in the classes to come.