We could say, as is usually done, that repeated morphemes express concord, as in Latin feminine '–a' in 'mēnsa parva' ‘the small table’ or the modern Hebrew article 'ha' (and feminine '–a') in 'haiša haktana' ‘the small woman’. Alternatively, we could say that in each of these cases we have not a repeated word-suffix or word-prefix. But rather a single phrase-infix consisting, in the case of the Hebrew article, of the phonemes /ha/ before every noun-morpheme (including adjectives) in a noun phrase. This would mean that instead of our being given a morpheme 'ha' and having to state that it occurs only with certain syntactic selections, we are given a morpheme which we may write 'ha…ha…' and which has no further limitations of selection, but either occurs or does not occur in a phrase, just as do the other morphemes. If the phrase contains the morpheme for ‘man’ and ‘small’, it is 'iš katan' ‘small man’; if it also contains the morpheme for ‘the’, it is 'haiš hakatan'; if it also contains the morpheme for ‘feminine’, it is 'iša ktana'; if it contains both, it is 'haiša haktana'.
- Harris (1946), a pag.162