phonetics
vs.
phonology
Physical description of sounds
 
Description of sound interrelation and function
presence vs. absence of sounds
 
complementary vs. contrastive distribution of
presence vs. absence of features
 
redundant vs. contrastive (phonemic) features
narrow transcription in square brackets
 
broad transcription in slanted brackets
transcr. symbols the same across langs.
 
transcription symbols generalized, unique to each lang.

**go over the difference between a phone, an allophone and a phoneme

Recap of the phoneme theory

Phonemic transcription is often convenient shorthand for getting down to the essentials in transcribing a language. But some linguists hold that all sounds in all languages can be broken down neatly into a small group of phonemes. Each phoneme would then consist of a sound or set of sounds which contrasts with other such sets. Here are the main tenets of the phoneme theory.

1) All sounds in a language can be grouped together into abstract units called phonemes. The phonemes of the language constitute the psychological, or underlying deep structure of the sound system of a language.

2) Each phoneme consists of one or more allophones, or positional variants (the actual phones of the language). Rules can be written to show how a given phoneme is realized as this or that allophone depending on the phonetic environment.

3) Allophones of the same phoneme are in complementary distribution. Conversely, sounds in complementary distribution are allophones of the same phoneme. On the other hand, sounds in contrastive distribution belong to different phonemes. Each allophone--that is, each sound actually pronounced--belongs to one and only one phoneme.

Dividing the sounds of a language into phonemes can be a useful way of looking at language, since it focuses attention of the interrelationship between sounds and not merely on the presence of sounds in a language. And the idea of phoneme often does seem to correspond to speaker intuition (fulfilling the idea of being a psychological reality). Since the concept of the phoneme was invented by speakers of European languages, dividing the sound system up into separate units called phonemes works quite well for European language such as English. However, the phonemic theory of phonology that we have just examined is not a perfect way to describe the function of all sounds in all languages. But it is taught in every university and discussed in every textbook, including your own, so you will have to be familiar with it. Please keep in mind, however, that since the phoneme theory--useful though it often might be--does not fit language behavior perfectly, the theory will always be slightly confusing to you. This is not your fault--it is the fault of the theory.

Be aware that the following problems may arise to plague any attempt to describe all the sounds of a language as allophones of separate phonemes.

The First Problem. Sometimes sounds found in complemetary distribution are too different to be considered part of the same phychological reality. Since the phoneme is an abstract psychological unit, the allophones of a given phoneme would be expected to be phonetically similar. Usually, allophones differ in terms of only one single phonetic feature, as is the case with the three English p's. Sometimes, however, two speech sounds are in complementary distribution, yet differ completely in their phonetic composition. A classic example is the case of English /h/ and /N/. These two sounds are in complete complementary distribution; they do not form a single minimal pair. Yet it is difficult to claim that they are part of a single unit in any speaker's mind. Therefore, linguists usually consider them to be separate phonemes, since allophones must be similar enough phonetically to be considered a single psychological reality in the mind of speakers. The problem remains in defining the point at which sounds in complementary distribution should be considered to belong to phonemes or to be allophones of the same phoneme.

The Second Problem has to do with the phenomenon of free variation. Occasionally, sounds occur in contrastive distribution without causing any change in meaning. When a sound said to belong to one phoneme can replace a sound said to belong to another phoneme without any change in meaning we call the phenomenon free variation: ay vs. i -- bite, beat, might, meat but either; neither. In English, free variation is a very limited phenomenon, but in some languages it is much more extensive: cf. Cherokee words ending in a or i. Cases of extensive free variation erode the distinction between phoneme and allophone, making it hard to define phonemes in terms of their ability to produce differences in meaning.

The Third Problem involves cases where one and the same sound functions as the positional variant of more than one phoneme. Phonemic analysis lists the inventory of phonemes in a language as if each phoneme was an entity separate from the others. For instance, the English phoneme /p/ contains entirely different allophones than the English phoneme /k/. Sometimes, a phonemic contrast is entirely neutralized in certain phonetic environments. In Russian, for instance, voicing in obstruents is phonemic just as it is in English; however, voicing is neutralized in word-final position by a rule that devoices all obstruents in this position. Consequently, the sound t may belong to the phoneme /t] or /d]. The same problem exists in English when tt, dd become a flapped r in the environment between vowel and r: ladder/latter. If the same phone can belong to more than one phoneme, then it is also difficult to talk of phonemes as entirely separate units.

The Fourth Problem could be called sounds in quasi-complementary distribution. Some sounds never contrast and therefore cannot belong to separate phonemes, yet their distribution in completely random, dependent upon each individual word rather than on any phonetic rule. The Cherokee sounds ch, ts, j occur in different words in essentially the same phonetic environment: chunela, Tsalagi, jo-i. They are perceived as separate sounds. Although they produce no minimal pairs and are similar to one another phonetically, no phonological rule can be written to define their distribution: cf. junsdi-i babies; jakaha you have (something dense). The distribution of these three sounds seems to depend upon the meaning of the words, just like sounds that belong to different phonemes. And yet these sounds produce no minimal pairs or change the meaning if they are switched. Are sounds found in quasi-complementary separate phonemes, or are they allophones of the same phoneme? The randomness of their distribution from a phonetic point of view makes it impossible to call them predictable positional variants of one and the same phoneme. Yet, since they produce no minimal pairs and therefore never directly contrast, they cannot be said to belong to different phonemes, either. The distribution of Cherokee [chh], [ch] and [ts], [ts-aspirated] is only quasi-complementary from a phonetic point of view.

The Fifth Problem deals with instances where one phoneme is in complementary distribution with sounds from two different phonemes. This is exactly the situation with the Chinese palatals j, q, x, which occur only before i or . Before other vowels, the retroflex consonants zh, ch, sh occur. However, so do the alveopalatals: z, c, s. Are the first three allophones of the second three or of the third three? Speakers perceive them as entirely separate sounds, not like the English p's. Also, all three sets of sounds derive from historically different sources; they do not alternate in the grammar the way allophones usually do (the three p's in English alternate phonologically in fast speech: pick vs. Bill's pick.) Linguists usually call j, q, x separate phonemes and ignore the fact that they are in complementary distribution with two other phonemes.

A final problem involves rare or semantically restricted phonemes. sounds exist below the level of specific meaning. As a rule, phonemes have function but no specific meaning. Mention the case with [T] and [D] in English, which have a very low functional yield. Phonemes are supposed to be able to build new words and have no semantic restrictions. But note the [g] in Quileute, which occurs only in a single word, hagaj, frog. Is it a phoneme or not. Also note the occasional sounds in expressive words and sound gestures in English (clicks, voiced [h], glottal stop.

Although dividing the sounds of a language into neat little functional units called phonemes is an endeavor that is only partly realizable, the distinction between redundant phonetic features and contrastive or phonemic features is nevertheless completely valid. The problem remains as to how best to discuss this difference in a concise and elegant way. For English phonology, the phoneme theory works quite well in most instances.