The reality of the phoneme

We have already seen that the phoneme system of a speaker’s native language, and specifically the difference between pairs of sounds which contrast and pairs which do not, strongly condition her perceptions: the early twentieth century American linguist Sapir concludes that ‘What the native speaker hears is not phonetic elements but phonemes’. However, the phoneme is a psychologically real unit in other ways too, since it does not only condition what we hear, but also what we do.

First, alphabetic spelling systems are frequently based on the phonemes of a language: there are various reported cases of linguists teaching variants of the IPA to speakers of languages which lacked orthographies, and providing inventories of symbols which covered all the phones of the language, but where speakers subsequently made use of only one symbol per phoneme. In Old English, both [f ] and [v], which

were then in complementary distribution, were spelled >f<, whereas in Modern English contrastive /f/ and /v/ typically correspond to >f< (or >ph< ) versus>v< . Similarly, in Hungarian /k/ and /c/ are consistently distinguished as >k<and >ty<. The alphabet has several times been borrowed by speakers of one language from those of another, and has been remodeled in some respects to fit the borrowing phoneme system better. So, the first letter of the Semitic alphabet represents the glottal stop, [ʔ], which is phonemically distinctive in Arabic, for example: but when this alphabet was borrowed by the Greeks, that first letter, Greek alpha, was taken to represent the vowel which begins the word alpha itself. Although Greek speakers would commonly produce an initial glottal stop on a word like alpha (as would English speakers, especially when saying the word emphatically), they would not observe it or want to symbolize it, since [ʔ] is not a phoneme of Greek. We should not, however, as we saw , assume that we can simply read the phoneme system off the spelling system, since there is not always a one-to-one correlation. Hence, English does have two orthographic symbols for /k/, namely >k< and >c<, but these do not systematically signal two separate allophones: the spelling system simply has a redundant extra symbol here. Furthermore, some phonemes are spelled consistently, but not with a single graph, so the phonemic difference between the English nasals /m/, /n/ and /ŋ/ in ram, ran and rang, is signaled orthographically by >m <, >n <, and >ng< (or >nk <in rank).

More importantly, our native phoneme system tends to get in the way when we try to learn other languages. It is perhaps unsurprising that we should find it difficult at first to produce sounds which do not figure at all in our first language. However, it is just as difficult, and sometimes worse, to learn sounds which are phonemically contrastive in the language we are learning, but allophones of a single phoneme in our native system. For instance, there is no contrast between aspirated [th ] and unaspirated [t] in English; we can predict that the former appears only wordinitially. In Chengtu Chinese, however, /t/ contrasts with /th /, as we find minimal pairs like [tou] ‘a unit of dry measure for grain’ versus [th ou] ‘to tremble’; the same is true in Thai, where [tam] ‘to pound’ contrasts with [th am] ‘to do’, establishing a phonemic distinction of /t/ and /th/.

When a native English speaker tries to learn Chengtu Chinese, or Thai, she will find this distinction extremely awkward to replicate, despite the fact that she herself has always used both these sounds. The problem is that, whereas a totally new and unfamiliar sound simply has to be learned from scratch, an old sound in a new role requires further processes of adjustment: our English speaking Thai learner has to suppress her instinctive and subconscious division of the aspirated and unaspirated sounds, and learn to produce both in the same context. In perceptual terms, it is again easier to hear a completely new sound, which will initially be extremely easy to perceive because of its very unfamiliarity, than to learn to distinguish two sounds which have conceptually been considered as one and the same. Conversely, a Korean speaker, who has [r] and [l] as allophones of a single phoneme, with [r] produced between vowels and [l] everywhere else, will make errors in learning English, finding minimal pairs like lot and rot highly counter-intuitive, and tending to produce [l] at the beginning of both, but [r] medially in both lolly and lorry. A combination of unlearning and learning are needed to get those patterns right.