Lecture 9: Phonology II
Lecture 9: Phonology II
Phonemes: in this example English has two distinct phonemes, Turkish has just one, which can be pronounced in at least these two different ways.
[note: the colon notation after a vowel means that the vowel is long]
Phonemes: Serbo-Croatian has two phonemes, /a/ and /a:/; English, on the other hand, has just one phoneme /a/, and it can be pronounced either long or short.
The phonetic distinction is possible in both languages
The distinction is phonemic in just one of the two languages
One phoneme with two (or more pronunciations) &endash; each pronunciation is an allophone of the phoneme
Here is another rule for deriving allophones in English. Nasal and non-nasal (oral) vowels are not contrastive in English; they are derived from underlying phonemes. Vowels in English become nasal when they precede a nasal sound:
vowel -> nasal / ___ [nasal]
Compare: nasal and oral vowels contrastive in Akan (Ghana, Africa) (for the examples below, a = oral vowel and a = nasal vowel
nsa 'hand' vs. nsa 'liquor'
ka 'bite vs. ka 'speak'
Oral and nasal vowels must be separate phonemes in Akan; there are minimal pairs which differ only on the vowel sounds.
Question: does this feature of Southern American English have the effect that vowel length is phonemically contrastive in Southern US English, making it like Japanese and unlike Northern American English? The way to test this is to find out whether Southern American speakers make a systematic difference in the way they pronounce "pod" [pad] and "pied" [pa:d], or "trod" [trad] and "tried" [tra:d].
Some well-known dialects of English contain a phonological rule which deletes /r/ in syllable final position. We can, however, show that speakers do represent the /r/ phonologically in words like "car" and "dinner", because the /r/ reappears when these words are followed by words beginning with a vowel, with the consequence that the /r/ is now syllable-initial rather than syllable-final.
In uncovering the phonemes and phonological rules of an unfamiliar language we are simply doing what every child does -- with no apparent difficulty -- in learning his/her native language.
Learner's task is to figure out which sounds are:
Researchers do not yet agree on how children achieve this task so easily, but there are well established methods that linguists have used to analyze the phonological system of unfamiliar languages, and similar strategies may be useful to children in analyzing their own input.
Notice that one important difference between the task of the child faces and the task that you -- the Introduction to Linguistics student -- face in analyzing the phonology of a language: the phonology problems that you will tackle in this course all provide precisely the most relevant language data for you, in neatly organized fashion; on the other hand, children don't get this kind of help - they have to figure out from the stream of words and phrases coming at them which are the relevant data for figuring out the phonology of their language.
2 phonemes
or
1 phoneme + rule(s)
There are a number of tools we can use to look at phonological data to determine whether we have different phonemes or allophones of a single phoneme.
(A more detailed version of these hints is found in Test 1 Tips and Practice Questions)
1) Are there any minimal pairs? These are a dead giveaway that we have different phonemes.
2) If not, we want to look for allophones. Determine the environment that each possible allophone occurs in. Are these environments in complimentary distribution? If there are, then the sounds are probably (but not always) allophones.
3) Choose an allophone to be the underlying form (the form from which all the other allophones are derived). As a rule of thumb, this is the allophone that occurs in the greatest number of context. This limits the number of rules we have to write in order to account for all of the data.
4) Write the rules to derive all of the allophones. Check to make sure that all of the data is accounted for.
minimal pairs: only the pair of sounds that we're interested in differ, everything else about the two words is identical.
Are there minimal pairs in Japanese which alternate between vowel length? Yes (to vs. to:). Therefore vowel length must be contrastive in Japanese. It cannot be derived from the phonological context.
|
|
p |
ph |
|
syllable-initial |
no |
yes |
|
following /s/ |
yes |
no |
|
syllable-final |
yes |
no |
Is there a difference in the length of the vowels in the following pairs?
Yes! There is a difference in vowel length. Is it
contrastive like in Japanese (are there minimal pairs)? No! Vowel
length is therefore not contrastive in English; it is not phonemic.
It is allophonic. Vowels in the words above are long before voiced
sounds and short before voiceless sounds (for most speakers). We can
write a rule which will give us vowel length in English.
This is the Superman / Clark Kent explanation given in your text. Allophones are never found in the same environments, just like Superman is never seen in the same place as Clark Kent. They are the same thing in some underlying form; phonological rules turn the underlying phonemes into different allophones on the surface.
If so, then the sounds are in free alternation.
"A certain phoneme is realized as a certain allophone in a certain environment"
or
A --> B / C _____ D
(Note: either C or D may be absent if irrelevant.)
"A --> B": Some feature (or phoneme) A changes to some feature (or phoneme) B
/ = "In the environment of"
C ______ D: C and D represent some context (either before or after the sound to which the rule applies)
Here is a more concrete example from the vowel length distinction in English:
i --> i: / _____[+voice]
"Short vowels (i) change to long vowels (i:) in the environment of before voiced sounds)
When looking at phonological data, remember that complementary distribution test is just a detective strategy, not a foolproof rule.
For example, in English [h] and [ng] (the velar nasal) are in complimentary distribution. There are no words which begin with [ng] and there are not words which end in [h]. But are they in complimentary distribution? Are they derived from a single underlying form?
[ng] = voiced velar nasal
[h] = voiceless glottal fricative
These two sounds have nothing in common. It would be diffidult to write a rule that would change all of the features of [ng] to the features of [h]. This is a case where complimentary distribution is accidental. [h] and [ng] happen to occur in different environments, but they are not derived from a common source.
These are problems that were discussed in class, or discussion section, or will be discussed at the review sessions on Wednesday October 8th.
In this example, we're interested in whether /s/ and /z/ are allophones of the same phoneme, similarly for /t/ and /d/.
Here we're interested in the distribution of [v] and [b], and of [f] and [p] - when are [v] and [f] used? Also, can we state a generalization that accounts for both alternations with one rule?
State the distribution of [c], [t], and [ts] in Japanese.
The 4 sounds of Greek listed below are the pronunciations of two underlying phonemes. Which are the two pairs of allophones, and when are the different allophones used?
Mokilese uses voiceless vowels (these are marked by a small circle placed beneath the vowel). Can the voicing of a vowel be used to mark a meaning distinction in Mokilese, or are all voiceless vowels allophones of regular voiced vowels? If the voiceless vowels are systematically related to voiced vowels, then state a rule that accounts for their distribution.
The forms below are based on an exercise in your textbook, except that misprints have been fixed (e.g. word for 'a horn'). See the questions in F&R's textbook.
State the distribution of the (palatal) affricates in Amharic.
A number of rules are needed to account for the various forms of the progressive in Japanese, which is formed using a suffix with the basic form '-te'. Can you state a set of rules which will account for the forms below?
