PHONEMIC
ANALYSIS
General assumption:
Speech is composed of phonemes
Whenever a speech sound is produced, it
is possible to identify which phoneme the
sound in question belongs to.
Still, there are various problems!
AFFRICATES
Phonetically, they are composed of a plosive
+ fricative
But, it is possible to treat each affricate as a
single consonant phoneme (one-phoneme
analysis)
They can also be treated as composed of two
phonemes each, all of which are established
as independent phonemes in English (twophoneme analysis)
/t∫з:t∫/ & /dζλdζ/
In the first case they would be considered as
consisting of 3 phonemes, in the second
case of 5 phonemes
Which analysis is preferable?
In the 2-phoneme analysis (no separtion of
affricates) the total number of phonemes is
smaller and should be preferred as more
economical (the most efficient codes do not
use unnecessary symbols)
Still,
1-phoneme analysis is generally chosen by
phonologists as preferable
The arguments for:
1. Phonetic/allophonic argument: phonetic
quality of /t/ and /∫/ in /t∫/ and /dζ/ is
different from realisations of the sounds
mentioned found elsewhere, e.g. Different
quality of /t/ in “watch apes” vs. “what
shapes”
- still, this argument is weak
2.
Distribution
The proposed phonemes have
distributions similar to other consonants,
while other combinations of plosive+
fricative do not: /t∫/ and /dζ/ are found
initially, medially and finally while no
other combination has such a wide
distribution.
However, there are several consonants in
English accepted as phonemes in spite of
not being free to occur in all positions
(think of r, w, j, h, ζ, ŋ/
3.
-
-
Combining with other cons.
Free combining to form clusters would
support the 1-phoneme analysis
Initially they never occur in clusters
Finally, they can be followed by t,d and
preceded by l, n
Another combination: pre-final l,n can
occur with post-final t,d: e.g. squelched,
hindged
So, /t∫/ and /dζ/ do not combine freely
to form clusters, particularly not initially
Two-phoneme analysis
Initial /t∫/ and /dζ/ would have to be
interpreted as initial t,d + post-initial ∫,ζ
(besides l,r,w,j) which can combine with
t,d, only
4.
Intuition of the native speaker
Rather difficult to discover what native
speakers (if untrained in phonetics and
phonology) think or feel
Other problems:
Sounds transcribed as hw, hj
Velar nasal ŋ (should it be treated as a
separate phoneme or an allophone of the
phoneme n occurring before g)
The English vowel system
Treating all long vowels and diphthongs
as composed of two vowel phonemes:
e.g. long vowels can be seen as
containing short vowels twice,
triphthongs would be composed of a
basic vowel + one of ι,υ + ə (which
makes three phonemes altogether)
Another way of treating long vowels
and diphthongs
As composed of a vowel + a consonant (j,
w, h, r), e.g. /eı/ - /ej/, /әυ/ - /әw/,
/υә/ - /υh/, /i:/ - /ıj/. /a:/ - /ah/, /u:/ - /uw/
Thus, inequality of distribution is corrected
for consonants that do not otherwise occur
finally in a syllable.
More about long vowels
Remember NEUTRALISATION of /i:/ and
/ı/ to /i/; i.e. Cases where contrasts
between phonemes which exist in other
places disappear in certain contexts
Syllabic consonants
A possibility is to add new cons.
Phonemes to the existing list: syllabic
l,r,n as in bottle, button, Hungary
Clusters of s + plosive
P, t, k in syllable initial position are
aspirated, but when preceded by s they
become unaspirated and could perhaps
be transcribed as b, d, g because
contrast between these two groups of
consonants become neutralised in this
context
Schwa /ә/
-
/ә/ can be treated as an allophone of
several vowels, not only of /λ/ in
unstressed syllables, e.g.
Economy vs. economic
German vs. Germanic
DISTICTIVE FEATURES
Distinctive feature analysis is one of
many different ways of treating the
notion of phoneme.
The principle: phonemes are not
minimum, independent, indivisible units
but combinations of different features.
In a table showing presence or absence
of features in different consonants there
would be no phonemes with the same
combination of +s and –s; otherwise,
they could not be treated as different
phonemes