[…] there may be sequences of morphemes which occur in environments where single morphemes do not occur, i.e. where they cannot be replaced by any single morpheme. Such sequences may or may not be useful as elements of the utterance structure. For example, Semitic roots plus verb patterns occur in environments in which no single morpheme occurs. They occur before verb suffixes, after verb prefixes, and in various sentence positions such as (in early Semitic) before an accusative noun (presumably with command intonation): thus, in classical Arabic, 'root ftḥ' ‘open’ and pattern '(i)--a-' ‘command’ in 'iftaḥ ilbāba' ‘open the door!’. In all these environments we always find the sequence of some root plus some pattern; we never find a single morpheme here. We may consider this sequence to constitute an element in the utterance structure, calling it, say, verb stem.
- Harris (1946), Pag. 166

A check of the preceding equations will show that all morpheme classes and all sequences of morphemes […] occur in in positions where they can be replaced by 'N⁴' or 'V⁴'. We can therefore state in terms of these classes what sequences of morphemes occur in English utterances. The great majority of English utterances are a succession of the following forms: 'N⁴' 'V⁴' with /./, /?/, or other intonations; with 'N⁴' ('= P N⁴ = Da'), independent morphemes, and successive repetitions introduced by '&', set off by /,/. Independent morphemes and almost all others except affixes (classes '–Nn' to 'Xx-' […] ), occurring singly or with affixes, with /./, /?/, /!/, and other intonations: 'Yes.'; 'Why?'; 'No!'; 'Come!'; 'John!'; 'English.'; 'Here'.
- Harris (1946), Pag. 175