neologistic formation made by shortening has the homonym wren, n. "a small bird with dark brown plumage barred with black" (R. крапивник).
Words made by sound-imitation can also form pairs of homonyms with other words: e. g. bang, n. ("a loud, sudden, explosive noise") — bang, n. ("a fringe of hair combed over the forehead"). Also: mew, n. ("the sound a cat makes") — mew, n. ("a sea gull") — mew, n. ("a pen in which poultry is fattened") — mews ("small terraced houses in Central London").
The above-described sources of homonyms have one important feature in common. In all the mentioned cases the homonyms developed from two or more different words, and their similarity is purely accidental. (In this respect, conversion certainly presents an exception for in pairs of homonyms formed by conversion one word of the pair is produced from the other: a find < to find.)
Now we come to a further source of homonyms which differs essentially from all the above cases. Two or more homonyms can originate from different meanings of the same word when, for some reason, the semantic structure of the word breaks into several parts. This type of formation of homonyms is called split polysemy.
From what has been said in the previous chapters about polysemantic words, it should have become clear that the semantic structure of a polysemantic word presents a system within which all its constituent meanings are held together by logical associations. In most cases, the function of the arrangement and the unity is determined by one of the meanings (e. g. the meaning "flame" in the noun fire — see Ch. 7, p. 133). If this meaning happens to disappear from the word's semantic structure, associations between the rest of the meanings may be severed, the semantic structure loses its unity and falls into two or more parts which then become accepted as independent lexical units.