Vous êtes sur la page 1sur 6

Week 2 Phonological Features

September 13, 2011

Oppositions
Consider the sounds [p, ph , b, bh , t, th , d, dh , k, kh , g, gh ] in Nepali. Below are some
minimal pairs. Assume there are minimal or near-minimal pairs for all combinations.
Explain why the pair illustrating the [p/ph ] distinction is more informative in a way
than the pair illustrating the [p/bh ] distinction. What phonetic dimensions must be
associated with the contrast? What about the English and Spanish sounds discussed
earlier?

[pir]
[ph ir]
[tal]
[th al]
[kal]
[kh al]

anxiety, pain
Turn on!
lake
plate
time, death
kind, skin

[bar]
[bh ar]
[dar]
[dh ar]
[gol]
[gh ol]

fence
burden
a kind of tree
edge
circle, ch arcoal
Mix! Stir!

Distinctive Feature Theory

2.1

Trubetzkoy

Nikolai Sergeevich Trubetzkoy (18901938), Russian linguist, used the term opposition to
refer to a pair of speech sounds that are distinctive, or contrastive. In other words, for those
pairs of sounds for which we can find a minimal or near-minimal pair. He classified these
oppositions in the following ways:
Bilateral oppositions are those where two members of an opposition have sufficiently
many phonetic properties in common which distinguish them from every other member
of an opposition.
Explain why /k,g/ are in a bilateral opposition in Nepali.

J. Heinz

Week 2: Phonological Features

Does this mean members of a bilateral opposition differ only in a single phonetic
dimension?

Multilateral oppositions are those which are not bilateral.


Give an example of a multilateral opposition from Nepali.

An opposition is proportional if and only if the relation between its members is


identical with the relation between the members of another opposition or several other
oppositions of the same system.
An opposition which is not proportional is isolated.
Provide some examples of proportional oppositions in Nepali.

Oppositions wherein one member carries some phonetic property that the other lacks
are said to be privative. The member carrying the phonetic property is said to be
marked. This is the origin of the term markedness in phonology, which today means
either less-common, dispreferred, or ill-formed.
Gradual oppositions are those where members of an opposition differ in some degree
of some phonetic property.
When members of an opposition differ in a way that is neither privative nor gradual,
it is said to be equipollent.
An opposition is neutralizable iff it occurs in certain contexts. Otherwise it is constant.
Explain why this paradigm from German establishes that the /t,d/ opposition is
neutralizable.
[rat]
[rat]

advice
wheel

[rE:t@]
[rE:d@r]

advices
wheels

J. Heinz

Week 2: Phonological Features

(Hyman, 1975, p. 29): With these notions, Trubetzkoy was able to reveal how the same
phonetic contrast may structure differently in different languages.
What does this mean? Can we apply this to the English and Spanish example we
discussed last week?

2.2

Jakobson

Jakobson introduced the notion of distinctive feature into phonological theory.


[NB: The phonetic symbols have been standardized to the IPA, JH]
While Trubetzkoys concern was to capture the phonological properties of such
frequent phonetic contrasts as voicing in consonants and height in vowels, the
concerns of Jakobson, another founding member of the Prague School, were somewhat different. Jakobson wanted to develop a theory of phonology which would
predict only those oppositions which could be found in languages. In particular,
he hypothesized that the presence of certain phonetic oppositions precludes the
presence of other oppositions. For example, in works such as Jakobson, Fant and
Halle (1952) and Jakobson and Halle (1956) it is maintained that languages do
not have contrasts between labialized, velarized, and pharyngealized consonants,
that is, /Cw /, /CG /, and /CQ /, respectively. Jakobson claimed that a given language will contrast only one of these three consonant types with a plain /C/.
Thus, while there can be an opposition between /C/ and /Cw /, /C/ and /CG /,
and /C/ and /CQ /, one cannot find an opposition between /Cw / and /CG /, /CG /
and /CQ /, or /Cw / and /CQ /. This mutual exclusiveness of these three kinds of
consonants led Jakobson, Fant and Halle to propose that they are merely surface phonetic realizations of the same underlying feature of flatness (see below).
They hypothesized that there are a limited number of such features, say 12 to 15,
which together account for all of the oppositions found in the worlds languages.
Since many more than 12 to 15 phonetic features are necessary to differentiate
the various sounds occurring in languages, it becomes apparent that some of
these phonetic features will be conflated into the more limited set of phonological or distinctive features. This represents, then, a major departure from
earlier phonetic studies of speech sounds. In the work of other phoneticians and
phonologists, there is an assumption that the same features are to be used to
characterize phonological contrasts in a language and to describe the phonetic
content of various speech sounds. Jakobsens position is that there are certain
phonetic distinctions, such as labialization, velarization, and pharyngealization,
which are not available per se as phonological features but rather are representative of the more basic phonological feature of flatness. Thus, for the first time,
3

J. Heinz

Week 2: Phonological Features

the possibility is entertained that the set of phonological features may not be the
same as the set of phonetic features.
(Hyman, 1975, p. 30)
Two other innovations of Jakobson: the use of acoustic features and the requirement that
all features are binary. The motivations again come from typological considerations.
With respect to binary features, it is logically possible that sounds could be voiceless,
barely voiced, somewhat voiced, somewhat fully voiced, and fully voiced. But in fact languages
only seem to make a 2-way distinction. Since Trubetzkoy considered voicing a privative
opposition, he was (at least implicitly) making a similar claim.
With respect to acoustic features, Jakobson was interested in determining which features
define natural classes of sounds. The set of sounds sharing a feature form a natural class.
These classes ought to be reflected in the phonological patterning of sounds across languages.
Binary features most naturally describe privative oppositions. How can binary features
describe gradual or equipollent oppositions?

If the phonemes are only identified by distinctive features then what determines the
phonetic realization of the phoneme? How could these language-specific instructions
be formalized?

2.3

Determining which features are distinctive

It is not obvious how to determine which features are distinctive in any given language.
There are at least two possibilities.
1. Pairwise Algorithm (Archangeli, 1988)
(a) Fully specify all segments.
(b) Isolate all pairs of segments.
(c) Determine which segment pairs differ by a single feature specification.
(d) Designate such feature specifications as contrastive on the members of that pair.
(e) Once all pairs have been examined and appropriate feature specifications have
been marked contrastive, delete all unmarked feature specifications on each segment.
2. The Successive Division Algorithm (Dresher, 2009)
4

J. Heinz

Week 2: Phonological Features

(a) Begin with no feature specifications: assume all sounds are allophones of a single
undifferentiated phoneme.
(b) If the set is found to consist of more than one contrasting member, select a feature
and divide the set into as many subsets as the feature allows for.
(c) Repeat step (b) in each subset: keep dividing up the inventory into sets, applying
successive features in turn, until every set has only one member.
Lets apply both methods to the following mini-inventory.

[voiced]
[nasal]

p b m
+ +
+

Now what happens when both are applied to a common inventory of vowels?

[high]
[low]
[back]
[round]

i
+

+
+

+
+

u
+

+
+

Dresher calls this the too many features problem. It is not always the case that
contrastive speech sounds differ along a single phonetic dimension.
Dresher concludes that the Pairwise Algorithm suffers from a logical problem, but the
Successive Division Algorithm (SDA) does not. On these grounds, further research into
distinctive features ought to proceed along the lines as outlined by the SDA.
Are there any weaknesses to the SDA?

Can you think of a third alternative to identifying distinctive features?

J. Heinz

Week 2: Phonological Features

The Emergence of Distinctive Features

Mielke (2008) conducts a cross-linguistic study of 500 languages to see to what extent the
phonological rules in grammars reflect natural classes, defined according to various feature
proposals. He finds that about 75% of rules target natural classes as defined by any of
the proposals he considers, and the other 25% do not. His follow-up studies show that
the unnatural classes are of varying sizes and types and do not neatly fall into any simple
description. On this basis, he claims that distinctive features may not be innate but instead
emerge, or be learned in some manner.

Hayes 2009 and Features

In this course, we will rely on Hayes (2009) for the feature system used in this course. It is
provided on the handout. Please read at your leisure.

References
Archangeli, Diana. 1988. Aspects of underspecification theory. Phonology 5:183208.
Dresher, Elan. 2009. The Contrastive Hierarchy in Phonology. Cambridge University Press.
Hyman, Larry. 1975. Phonology: Theory and Analysis. Holt, Rinehart and Winston.
Jakobson, Roman, C. Gunnar, M. Fant, and Morris Halle. 1952. Preliminaries to Speech
Analysis. MIT Press.
Jakobson, Roman, and Morris Halle. 1956. Fundamentals of Language. The Hague: Mouton.
Mielke, Jeff. 2008. The Emergence of Distinctive Features. Oxford: Oxford University Press.