Difference between phonetics and phonology

      Phonetics simply describes the articulatory and acoustic properties of phones (speech sounds).  Phonology studies how sounds interact as a system in a particular language.  Stated another way, phonetics studies which sounds are present in a language; phonology studies how these sounds combine and how they change in combination, as well as which sounds can contrast to produce differences in meaning (phonology describes the phones as allophones of phonemes).

Redundant and contrastive features

      Every language consists of speech sounds called phones.  (Give example of various English sounds.) These different sounds do not all have the same status in the system of English phonology. The occurance of certain phonetic features is entirely predictable, as is the case in English with voicing in sonorants, nasality of vowels, or length in vowels.  Features whose presence is entirely predictable based on the phonetic environment are called redundant phonetic features: give example of the three English p's.  Contrasts involving a redundant feature cannot be used to signal a change in meaning.  Adding or removing a redundant, predictable feature results merely in a mispronunciation, not in a new meaning: cf. interchanging the different p's of English. In a broad phonetic transcription, redundant features can be ignored, and only the elements of pronunciation that are important for distinguishing meaning listed.   

      The occurrence of other sounds and features in a particular language is not predictable based on phonetic context.  In English, for example, voicing in obstruents is never predictable: pat/bat/ tip/dip/ girl/curl; the contrast between fricatives and stops is also not predictable: send/tend.  Or the difference between central and lateral in liquids: red/led; ball/bar. Such features are called distinctive, or contrastive, phonetic features, Phonetic features whose presence or absence can alter meaning are called phonemic features.  The presence of a phonemic feature is not predictable according to phonetic context.  Adding or subtracting a phonemic feature normally results in a change of meaning as well as in a change in pronunciation.

Complementary and contrastive distribution

      Phonetic features that are redundant in one language can be phonemic in the other.  The two phones [r] and [l] are present in English and Korean but play an entirely different phonological role in each language.  They are phonemically different in English, always signaling a difference in meaning: list/wrist  war/wall.  In Korean, [r] is word initial and [l] is syllable final: rupi ruby; mul water.  The two sounds never contrast to produce a difference in meaning.  In English the two sounds are in contrastive distribution; in Korean they are in complementary distribution.  Similarly, features that are redundant in English may be phonemic in another language: aspiration in English and Mandarin Chinese: kha~n (to see) vs. ka~n (trunk, stem); tha# (pagoda)  vs. ta#  (beat, strike); phi@ng (a sound)  vs. pi@ng (soldiers, army).

Phonemic analysis

How does a linguist determine which phonetic features in a given language are phonemic and which are not?

      First, a phonetic inventory of speech sounds must be carried out.  One needs to determine which speech sounds are present in a particular language in the first place.  A linguist studying English for the first time, for instance, will soon discover that English has aspirated p, unaspirated p and non-released p, but no glottalized p, implosive p, or pharygealized p.  The first stage of phonological analysis simply involves an exhaustive phonetic analysis.

      Second, having determined which speech sounds occur in a particular language, the linguist must determine whether or not the phonetic difference between these sounds is redundant or phonemic.  Phonology is concerned with determining which speech sounds contrast with one another to produce differences in meaning and which speech sounds are in complementary distribution and never produce meaninful contrasts.  To find which sounds are redundant and which are phonemic, linguists usually try to find a pair of words with different meanings that differ formally by only a single sound.  A pair of words which are distinguished by a difference in only one sound is called a minimal pair: pit/sit cat/cought; law/raw.  In the case of a minimal pair, the two contrasting sounds are obviously capable of distinguishing meaning, so the difference between them is phonemic.  The linguist will find that some speech sounds in a given language help form many minimal pairs, others only a few.  The number of minimal pairs involving a particular sound is called the functional yield of that sound.  There are 429 minimal pairs involving English sound [d] and only 32 involving [D].  Some sounds contrast directly with one another in only a few minimal pairs: ether/either, thigh/thy, dilution/delusion, Confucian/ confusion.

      A third phase of phonological analysis is to try to remove from consideration all redundant phonetic features and focus only on the distinctive, phonemic features.  The most widely employed means of accomplishing this is to group together all the sounds that actually occur in a language into contrastive sets called phonemes.  For example, sounds which are in complementary distribution, such as the English sounds p, ph, non-released p, are treated as a single phonological unit, or phoneme, their redundant phonetic differences ignored.   The actual phones that act as positional variants of one and the same phoneme are called allophones of that phoneme; thus, the three English p's are allophones of a single phoneme.  The phoneme is an abstract unit.  We don't hear or pronounce the phonemes of a language; we hear and pronounce their allophones.

      The question then arizes as to how to symbolize such an abstraction as the phoneme that is manifested as the three English p's.  Since there are two ways of looking at the sound system of language: one phonetic and the other phonemic, there are also two types of transcriptions.  The one we have been using up till now is a narrow transcription intended to portray as much phonetic information as possible.  This type is called phonetic transcription and is enclosed in square brackets.  The same IPA symbols can be used in phonetic transcription to transcribe any language of the world.  Thus the phonetic symbol [l] represents the same sound in English and Korean.

        The other type of transcription is called phonemic transcription and is enclosed in slanted brackets.  Phonemic transcription ignores any redundant phonetic detail in a language and only portrays the distinctive and meaningful phonetic differences.  Phonemic transcription represents phonemes, not the sounds as they are actually spoken. 

Choosing phonemic symbols

      What symbol is chosen from among the phonetic symbols for the individual allophones is up to the linguist performing the phonological analysis and depends on several factors.  Some phonemes have the same or nearly the same pronunciation in every phonetic environment, as is true of English [s],  [m].  In such cases, the symbol for the phone can also be used as the symbol for the phoneme.  If there is only one surface manifestation of a phoneme, then the phonetic symbol for that sound becomed the phonemic symbol, as well.  If a phoneme is composed of several phonetically distinct allophones, however, then any of the following may be done:

a.) diacritics are removed from allophone symbols, simplifying the sound.

b.) the phonetic symbol for one of the allophones may be co-opted to stand for all the allophones (carrot instead of schwa, or o instead of O)

c.) the most common letter of the alphabet is chosen (t/T)

d) some compromise letter is chosen., perhaps not even a symbol from the phonetic alphabet (capital R for l/r in Korean).

      Obviously, the process of choosing a phonemic symbol is somewhat arbitrary and up to whatever linguist is performing the analysis.  Phonemic symbols are a type of phonetic shorthand that with specific value for one and only a single language; they are not universal like the symbols of the phonetic alphabet.

        Thus, the value of symbols used in phonemic transcription is idiosyncratic and differs from language to language.  Phonemic transcription depends upon the interrelationship of sounds in each particular language, whereas phonetic transcription depends simply on the pronunciation of each individual sound regardless of its function in the sound system of the given language.  A phonetic symbol stands for one and the same sound regardless of language, but a phonemic symbol often stands for any one of several actual sounds.  For example, the phonetic symbol [l] stands for the same sound in the phonetic transcription of English and Korean.  A phonemic symbol such as /l/ usually stands for quite different collections of sounds in different languages.  English, for example, has a phoneme that could be represented as /l/ with two different separate surface manifestation: velarized or non-velarized l;  In phonetic transcription these two l-sounds would be written each with their own separate symbol.  In phonemic transcription they would both be written with the same symbol /l/.  In Korean the phonemic symbol /l/ would represent the allophones [r] and [l].  Remember that phonetic transcription, enclosed in square brackets, attempts to express as much phonetic detail as possible, redundant or otherwise; phonemic transcription does not mark redundant features, but rather is intended to represent only those phonetic details of a given language that are distinctive.  Phonemic transcription, therefore, uses phonetic symbols in a way unique to each particular language.  The phonemic symbols chosen in your handout are essentially the same sounds used in phonetic transcription minus the diacritical marks.