Sound Patterns

Chapter 2 described the sounds the human voice can make. This chapter asks a different question: which sound differences does a given language actually use? Every language ignores some phonetic differences and exploits others, and what counts as "the same sound" depends on the language. The study of how a language organises its sounds into a system is called phonology. Phonetics asks what sounds exist; phonology asks what a language does with them.

Phonemes and Minimal Pairs

A phoneme is a sound that can change the meaning of a word. The standard test is the minimal pair: two words that differ in exactly one sound, in the same position. If swapping one sound for another gives a different word, the two sounds are separate phonemes of the language.

Phonemes are written between slashes, as in /p/, and actual pronounced sounds between square brackets, as in [p]. Slashes mean "this is a unit of the language's system"; brackets mean "this is a physical sound that someone makes". The distinction matters in the next section.

Allophones: Different Sounds, Same Phoneme

Now the reverse. Two sounds that are phonetically different but never distinguish words are allophones of one phoneme. A speaker normally does not notice the difference at all. English has a famous example. Hold a hand or a thin strip of paper in front of your mouth and say pin and then spin. In pin the /p/ is followed by a puff of air; in spin it is not.

The difference between [pʰ] and [p] never separates two English words: no pair of English words differs only in whether /p/ has a puff of air. So English treats them as one phoneme, /p/. Their occurrence is also completely predictable: the aspirated form comes at the start of a stressed syllable, the unaspirated one after /s/. Linguists call this complementary distribution: each allophone appears where the other cannot.

The same sounds can be separate phonemes in another language. In Hindi, aspiration does change meaning: पल [pəl] means "moment" and फल [pʰəl] means "fruit". A Hindi speaker hears [p] and [pʰ] as two different sounds, and an English speaker usually cannot hear the difference until it is pointed out. The pair of sounds is physically the same; the languages simply sort them differently.

More Allophones

The Japanese case explains a familiar observation: Japanese speakers learning English often find rice and lice hard to tell apart. It is not a lack of hearing ability. Japanese has no /r/ versus /l/ contrast, so listeners have never had a reason to attend to the difference, just as English listeners had no reason to attend to aspiration. Chapter 1 called the human ability to produce language productive; here is the flip side: our perception is trained by the language we grew up with.

Phonological Rules

A phonological rule states how a phoneme changes depending on the sounds around it. The English plural is the clearest everyday example. The plural ending is the same grammatical unit in every case, but it is pronounced three different ways:

Nobody teaches English speakers this rule, and yet nearly all of them apply it correctly to a made-up word. Given wug (the invented word of a famous experiment with children), speakers say wugs with [z]. That is evidence that the rule lives in the speaker's mind, not in a rulebook.

Syllables and Allowed Sequences

Sounds are organised into syllables. A syllable has a nucleus (usually a vowel), optionally preceded by an onset and followed by a coda. In strengths, the onset is str, the nucleus e and the coda ngths. Languages differ in how complex their syllables may be.

Each language also has rules about which sounds may follow which, called phonotactics. Consider blick, bnick and brick. English speakers accept blick as a possible English word (it just does not happen to exist), but reject bnick as impossible, because English does not allow /bn/ at the start of a word. The same sequence is fine in other languages: Polish, for example, begins words with consonant clusters that English does not allow. Phonotactic rules are also why loanwords are altered: Japanese adds vowels to English words to fit its consonant-vowel shape, so strike becomes sutoraiku.

Stress and Rhythm

Stress, the extra prominence of one syllable in a word, is also part of the sound system. In English, stress can distinguish words: REcord (a noun) and reCORD (a verb) differ only in stress, so stress is contrastive. In Esperanto stress is fixed on the next-to-last syllable of every word, so it is completely predictable and never separates two words. Hungarian puts stress on the first syllable and French on the last syllable of a phrase.

Rhythm also varies. English is traditionally described as stress-timed (stressed syllables come at roughly regular intervals and unstressed ones squeeze in between), Spanish and French as syllable-timed, and Japanese as mora-timed (a mora is a unit that is roughly a short syllable). These labels are a convenient first approximation; careful measurement shows that languages do not fall neatly into three groups, but speakers of different languages really do hear differences in rhythm.

Hands-On Exercises