Lesson 10.3: PHONEMES AS CLASSES OF SOUNDS

To clarify the concept of a phoneme with several allophones further, let's look at an example from English. In American English, the phoneme /t/ has five different pronunciations. This means that /t/ has 5 allophones. Imagine how confusing this must be for a language learner! Each pronunciation depends on the phonetic context in which /t/ occurs. As discussed in the book (pp. 89-92), the allophones and their contexts appear in the words top, stop, little, kitten, and hunter. The differences in the pronunciation of each of these /t/'s are subtle to native ears; you may not discern them right away. But they may be substantial for speakers of other languages. Let's look at the pronunciations one at a time.

I will demonstrate /t/ with its five allophones in American English by using another visual aid. Allophones will be represented by K'Nex pieces I found in my son's toy box several years ago. Each allophone has a different color. The allophone is represented by a joint and its unique phonetic context by a link of the same color. Here are the five allophones and their contexts (please listen to sample 21 on the companion tape for a demonstration of these allophones):

1) /t/ at the beginning of words, as in top. The allophone of /t/ that appears in this context is an aspirated t with a puff of air after the release of the complete stoppage of the airstream. You can dangle a sheet of paper or better yet a tissue in front of your mouth when you say this word. You will notice that the puff of air from the aspiration moves the tissue. Aspirated t is shown as a red joint (the wedge part). The red attached link represents the context, i.e., "at the beginning of words".



2) the /t/ in st combinations at the beginning of words, as in stop. This allophone has the same pronunciation as /t/ in top but without the puff of air. To my ears, it is very similar to a d. Say deal and steal in a row several times, and then try saying s+deal and steal several times. What do you think - are the two stops in s+deal and steal much different? Unaspirated t is shown as a yellow joint (the half circle). The yellow link represents the context "after s in st combinations at the beginning of words."



3) the third allophone of /t/, the flap [24], occurs when /t/ is pronounced after a stressed vowel and before an unstressed vowel or syllablic consonant in words like little or city. In a flap, the tongue is quickly brought up to the alveolar ridge and then dropped down again; it "flaps" up and down. To us non-Americans, this allophone is one of the obvious characteristics of American English as opposed to British English. The flap allophone of /t/ is shown as a blue joint (the almost complete circle), with its context "between stressed vowel and unstressed vowel or syllablic consonant" shown by the blue link.



4) another pronunciation of /t/, the glottal stop [7], occurs in words like button, mitten, or kitten. In these words, the two phonemes /t/ and /n/ meet, both of which are pronounced with an alveolar place of articulation. To make /t/ distinct from /n/, most speakers in casual situations stop the air between their vocal folds as they do in "uh-oh", without moving the tongue. The glottal stop is represented by a white joint (the full circle). The phonetic environment "a following nasal with alveolar place of articulation" is demonstrated by the white link.



5) finally, there often is no pronunciation of /t/ at all in words like winter, hunter, or center. This zero pronunciation occurs in the context of a preceding alveolar nasal, which has the same place of articulation as /t/. The pronunciation is often perceived as very informal, and some of you may not be aware that you use it. In fact, many times one or two of my students claim that they do pronounce the /t/ in some fashion when in fact they don't. This non-pronunciation carries a negative connotation in the U.S. but is nevertheless very common. Zero pronunciation is shown as a green 145-degree joint, the phonetic environment "preceding alveolar nasal" as a green link.

The situation is different, however, for speakers of some other languages, such as Hindi. In Hindi, the aspirated and unaspirated t of 1) and 2) above each forms its own phoneme. That is, in Hindi there are words that are only different because one has an aspirated and the other an unaspirated t in it. This means that Hindi speakers are very aware of the difference between aspirated and unaspirated t. To them this difference is psychologically real because it matters when meaning distinctions are made (=distinctive). Since these two sounds are not tied to particular phonetic contexts but can appear in any context in Hindi, Hindi speakers will pay attention to the difference between [t] and [t^h]. Since they are not tied to particular environments, we can represent unaspirated t and aspirated t without links when they are Hindi phonemes, as shown below:
 

aspirated t phoneme		unaspirated t phoneme

There are consequences when sounds don't have links because they are not tied to environments. Without the link, there is no particular phonetic context in which a sounds may only occur. It can occur any place and is therefore not predictable from its context. For example, in English one can predict that unaspirated t will always occur at the beginning of words after s. This is not possible in Hindi. Unaspirated t can occur anywhere; it is non-predictable. Similarly, all other t allophones in English are "reserved" for particular contexts and are thus predictable.

All five contexts for all five allophones together form one complete whole, the phoneme. It is as if one cuts a pie into five pieces. Each piece makes up only part of the pie, but together they complement each other and form a single whole. In English, the pieces are in complementary distribution. In Hindi by contrast, aspirated and unaspirated t each forms a whole pie by itself. Their contexts don't complement each other, but rather each pie overlaps the other. The two phonemes are in overlapping distribution.

Now let's consider another linguistic example that is parallel to the above. Unaspirated k, [k], and aspirated k, [kh] occur in the following English and Hindi words (Note: a coot is a type of bird):

 

English	Hindi
[sk,36,t] - scat	[kh,43,l] - wicked person
[kh,36,t] - cat	[k,43,l] - yesterday, tomorrow
[skut] - scoot
[khut] - coot
[sk,33,d] - skid
[kh,33,d] - kid

1) decide whether English /k/ has more than one allophone and, likewise, whether Hindi /k/ and /k^h/ have more than one allophone.

2) decide whether each phone (=sound) can only occur in certain phonetic environments or in any environment.

3) if a sound is restricted to an environment, describe this environment phonetically (e.g. at the beginning of a word, after 's', between vowels etc.). In terms of our K'Nex model, determine whether sounds in English or Hindi would have a "link" attached.

4) determine which of the phones are predictable because of the environment they occur in. Hint: predictable means that there is information about an environment attached to the phone.

5) determine whether each phone occurs in complementary or overlapping distribution. Hint: if the phone already has a link connected to it, this means that it only occurs in the environment that the link stands for; it can not appear in an environment that overlaps with the environment of another allophone; it is a piece of a larger pie, so to speak.

6) determine whether each phone is distinctive/contrastive or non-distinctive/non-contrastive. Hint: if the phone is distinctive it is not restricted to certain environments, and you must be able to find minimal pairs.

For this exercise, go to FirstClass. Again, find your groups Phonology 1-4, post your answers, and discuss each others' answers. Let me know if there are issues or questions you can not resolve. Good luck with this more challenging discussion!

This is the end of Lesson 10 and also the end of phonology. You will find lesson 11 on Semantics by going to the list of lessons on the home page (http://web.hamline.edu/personal/aschramm/linguistics2001/index.html) next week and clicking on "Lesson 11".

Updated last: 3/24/01       © Andreas Schramm and Hamline University