[...] in speech, precise phonetic stations can be abstracted only by patient observation and frequently at the expense of a direct flouting of one's phonetic (one should say 'phonemic') intuitions.- Sapir (1933b), a pag.47