The raw articulatory productions of a language recorded as phones constitute only the first level in the process by which sound is transformed into meaning. On the next analytical level phones are grouped into larger categories termed phonemes. A phoneme is describable as a minimal class or category of phones, which possess shared features that clearly contrast to those of other phonemes and form the basis of distinguishing one utterance from another. This complex concept is best understood by the realization that not all speech sounds mark the difference between one word or word component and another. For example, there are different ways of producing an "r" in English. One involves the simple release of air from a rolled tongue on the hard palate, the other, a more energetic trilling of the tongue in the same position. The trilled "r" is not common, but may be used in special circumstances, such as in a song or dramatic performance where the rather muddy standard enunciation must be sharpened. These different forms of "r" are "nonphonemic," i.e., they are not used in contrastive positions to produce different words. As such, they belong to the same phoneme and are known as allophones, or variant forms of a phoneme. In Spanish, however, the situation is completely different, since the trilled vs. nontrilled distinction is phonemic. This contrastive function is apparent in the presence of minimal pairs, words that are distinguished from each other through the use of the two phonemes. For example the Spanish word pero (but) is distinguished from the word perro (dog) by the respective presence or absence of trilling.

"perro"  (dog)

"pero" (but)

"Tengo un perro blanco, pero esta sucio hoy". (I have a white dog, but he is dirty today.) 38k

Another difference between English and Spanish phonemics rests on a contrastive system that is present in English and absent from Spanish. In English the difference between "d" and "th" is phonemic as evident in the minimal pairs "den" vs. "then." In Spanish these two sounds are allophones of one phoneme usually represented as d in the written language, that appear in complementary rather than contrastive positions. The true "d" plosive always occurs at the beginning of a word as in the word:

"donut" (doughnut)

The "th" fricative form occurs between two vowels as in:

"bebida   (drink)

Observe the two different pronunciations of the d sound in the following sentence:

"De me una bebida helada y un donut". (Give me a cold drink and a doughnut)
Phone English 
r R R
rr RR
d D D
th TH
  • trilled and untrilled r's are allophones of one phoneme
  • plosive d and fricative th form separate phonemes
  • trilled and untrilled r's form separate phonemes
  • plosive d and fricative th are allophones of one phoneme

Table to demonstrate the difference in English and Spanish phonemics

The grouping of a language's phones into phonemes must be accompanied by the inclusion of additional phonemic patterns that arise from the influence of suprasegment of phonemes, usually patterns of stress, juncture, and tone that occur in combination with more overt articulations. These features are present in all speech systems but are not necessarily phonemic. For example, the pitch or tone of a speech sound does not usually create a meaningful distinction between two words in English. In tonal languages such as Chinese or many African languages, however, the pitch of each syllable is crucial for understanding word meanings. In Twi, a Ghanaian language, there are two words made up of the exact same sequence of phones but differing in their tonal patterns and consequently their meanings.

Akonta (brother-in-law: low-low-high)

akonta (mathematics: low-high-low)

These are not homonyms; they are completely different words because of the differences in intonation. (Note that the word for mathematics is derived from the English word "account," and its tones corresponds to the English stress pattern, substituting a high tone for a stress.)

The relation of pitch to meaning in tonal languages can be appreciated in an incident about some culturally and linguistically insensitive missionary work in Nigeria. The Christian churches tried to deal with their new communicants in native languages and translated religious texts for use during worship. Hymn translation usually involved keeping the original melody but substituting words from the local dialect. However, when tonal languages are set to music, care must be taken to match the melody to the words' pitch sequences, or their meaning becomes garbled or drastically changed. The missionaries were unaware of this problem and made several disastrous mistakes. One hymn in Igbo, a Nigerian language,  was intended to glorify Christ by acknowledging his great power, translated as ike. However, the melody changed the low-high sequence of the intended word to high-low and transformed it to mean backside.
talking drums

Talking Drum
Photo Credit:
University of Wisconsin-Madison Libraries.
Africa Focus © 2000.

The tonal patterns of African languages form the basis for an interesting non-verbal form of communication known as drum language. This communication medium is based on two drums, or sometimes bells, one of which has a low and the other a high tone. Messages are produced by drumming a sequence of lows and highs that matches the spoken syllable patterns. The following drum message in Twi is transmitted by Radio Ghana to introduce its evening news broadcast:

Ghana mo tie (Ghana listen: high-low-high-high-high)

Drum signals were used to communicate with people beyond the range of the human voice in Africa long before the electronic age.

Intonation is actually phonemic in some situations in English. The sentence "This is living?" is distinguished from the very different statement "This is living!"  by the placement of a rising tone on the last syllable. Other suprasegmental phonemes in English are based on patterns of stress and juncture. Stress is very frequently used to mark the difference between noun and verb forms, as in "record" (stressed-unstressed) and "record" (unstressed-stressed). Juncture refers to the length of pauses between two syllables and distinguishes the term "greenhouse," with a short pause, from "green house," with a long one.