![]() |
ИСТИНА |
Войти в систему Регистрация |
ИПМех РАН |
||
The workshop addresses the issue of how to combine the right base with the right affix. Consider, for instance, the English plurals box-es and ox-en. Many contemporary models of grammar propose that speakers do not need to store complex words like boxes and oxen as such, but that they can store only their parts (i.e., box and -es are stored as separate units). This allows us to construct complex words by assembling these smaller parts, while at the same time, reducing the number of stored items. However, the challenge for such models has always been to connect the right plural (-es or -en) with the right root, so that we do not end up assembling ‘incorrect’ words such as *box-en or *ox-es. Similar issues arise for the allomorphy of roots. Once we segment words like smart-er into two parts, the question becomes how to block the productive rule so that it does not derive incorrect forms like *good-er instead of bett-er. In the current literature, the issue goes under various names, such as ‘selection,’ ‘allomorphy,’ ‘suppletion’ and others. The standard approach to this problem is to rely on creating arbitrary classes of roots (‘irregular plurals,’ ‘strong verbs,’ ‘class I/II/III,’ etc.), and making the rules of exponence sensitive these classes, so that -en ultimately attaches only to ‘Class II irregular plurals’ (or whatever the label). However, the need to postulate such arbitrary and meaningless classes defeats its own purpose, since the process of assembling the words (that we had broken apart) can only proceed correctly if we somehow provide brute-force lists of what combines with what.