Phoneme Inventory

The phoneme inventory of a language indicates the number and types of phonemes which it has. Phonemes are the distinctive units of sound in a language. One method to determine phonemes is minimal pairs.

If two words share the same number and order of segments but one and those different segments create different words, the words are minimal pairs and the sounds are phonemes. For example, the English words "bell" and "sell" are minimal pairs. The only segments which are different are the word-initial "b" and "s". Since they create different words, they are phonemes of English.

The UCLA Phonological Segment Inventory Database, also known as UPSID, contains the phoneme inventories of more than 400 languages. In the UPSID database, Rotokas, a language spoken in Papua New Guinea, has only 11 phonemes. In contrast, Xu, a language spoken in southern Africa, has 141 phonemes. Many of those phonemes are click sounds common to the languages of southern Africa.

It is often difficult to determine the precise number of phonemes in a language. This is due to several factors. One is that diphthongs are often classified as phonemes in certain languages but not in others. For example, the diphthong which most English speakers have in "take" is classified as a phoneme but the diphthong of the Spanish word "reina" (queen) is not. The Spanish diphthong is phonemic because "reina" (queen) contrasts with "rana" (frog). However, most Spanish speakers do not view the diphthong as a single sound but as two due to the orthography which represents it with two letters.

The short and long vowels of Hungarian are classified as phonemic but the short and long vowels of Finnish are not. The difference is that the Hungarian short and long vowels are considered qualitative and quantitative but the Finnish vowels only quantitative. However, for many Hungarian speakers the high vowels /i/ and /u/ are merely quantitative. In some speakers, the short/long distinction in these vowels is actually non-existent. However, it is traditional to classify the Hungarian short and long vowels as phonemic.

In Finnish, though, the difference is merely considered one of length, so the long vowels of Finnish are excluded from the Finnish phoneme inventory. The phonemic length of Finnish vowels can be considered chronemic. Likewish, consonants can also be chronemic. The Finnish word "kuka" (who) differs from "kukka" (flower) merely in consonant length but phoneme inventories exclude long consonants. This is merely convention.

The number of phonemes in English ranges from 36 to 48. Certain inventories classify the rhotic vowels such as those of "car" and "ear" as phonemic and certain ones classify them as separate segments. In non-rhotic varieties of English such as Received Pronunciation, also known as RP, words such as "ear" are analyzed as a high front vowel followed by a schwa, a diphthong which certain linguists classify as a phoneme and others do not. Thus, the standards for determining the number of phonemes in a phoneme inventory vary.

Phoneme inventories are useful for comparing the number and types of sounds of different languages. It is often the case that languages of a particular area share common features. For example, click sounds are common in languages of southern Africa, consonant clusters are common in Germanic languages, and the existence of only one liquid such as an alveolar lateral or alveolar tap is common in many languages of eastern Asia.