Conversion in Morphology
This is an advance summary of a forthcoming article in the Oxford Research Encyclopedia of Linguistics. Please check back later for the full article.
Despite its apparent formal simplicity, to define conversion as a word-formation technique is by no means a simple matter, even in respect of one language, let alone languages representing different typological groups or subgroups. The traditional claim that conversion is a derivationally unmarked word-class changing operation involving formally identical (homonymous) lexical items seems largely justifiable so far as English is concerned where this operation is exclusively word/lexeme-based (cf. to swap > (a) swap, clear > to clear). However, while this same claim is also true for Hungarian, a Finno-Ugric language (cf. esteadv ‘in the evening’ > esten ‘(an) evening’), it does not readily apply to inflecting-fusional languages in which conversion is primarily root- or stem-based, rather than word/lexeme-based; and in which, therefore, a conversion rule has to take care of the possibility of not only adding new inflectional material to the output bases (whether roots or stems), but also of deleting and replacing it with other inflectional material. This is illustrated in the following examples taken from randomly chosen languages; the underlined segments in the examples signal inflectional material affected by conversion: cf. Catalan abraçar ‘to embrace, to hug’ > abraçadan.fem‘(an) embrace, hug’, Dutch ketten ‘(a) chain’ > kettenen ‘to chain’, French garder ‘to guard’ > garden.masc‘(a) guard’, German verleichen ‘to lend, to grant’ > Verleichn.masc‘loan office’, Italian favore ‘(a) favour’ > favorire ‘to favour’, Old English slœpan ‘to sleep’ > slœp ‘(a) sleep’, Russian igrat’ ‘to play, to act’ > igran.fem‘(a) play, acting’.
To determine the linguistic nature of conversion and its place among other types of word formation is not a simple matter either, and, paradoxically, it is especially so in the case of the most extensively studied English conversion. The reasons for this to a great extent lie in the fact that practically each element of the traditional definition suggested in the previous paragraph has been called into question, giving rise to a diversity of interpretations of conversion not only in English, but also in a cross-linguistic perspective. Thus, if conversion is viewed as a kind of derivation, the assumptions can be made that being derivationally unmarked means either the presence of a zero formative, or, alternatively, the lack of any overt derivational marking on the converted item (consider for instance the English, Hungarian, German, and Old English examples above). Regardless of their long-debated justifiability, what these assumptions respectively suggest is that conversion after all should be treated either as a kind of derivation, namely zero derivation, or as a self-contained word-formation process different from derivation (affixation). In addition, being derivationally unmarked is also viewed in the corresponding literature as the absence of derivation altogether; and the suggestion is made that during conversion it is in effect the change in the inflectional paradigm that can only signal word-class shift. Because of this, so the argument goes, conversion should be seen as an inflectional and not as a derivational process.
The notion of word class itself and the uncertainties characterizing its understanding present further challenges to morphologists dealing with conversion. Concretely, it is a widely shared view that only the unmarked change of the entire word class can be recognized as conversion (see the examples above). However, there are opinions that insist that the change of a subclass or subcategory also qualifies as conversion, albeit partial or non-prototypical (cf. to runintransitive > to runtransitive). In some languages, but not in English, a shift from one inflectional paradigm to another within the same word class is also recognized in certain cases as conversion of the latter type (cf. Russian fizikan.fem‘physics’ > fizikn.masc‘physicist’). Moreover, the assumptions that words/lexemes are stored in the mental lexicon with underspecified categories or with no category specifications at all relegate conversion to the fields of syntax or pragmatics. Related to the latter, pragmatics, is the interpretation considering conversion as coinage, a kind of non-rule-governed word formation.
Finally, treatments of conversion that focus on underlying semantic or conceptual motivations further add to the diversity of views of conversion. These treatments draw on the fact that there is a strong semantic link between the input and the output in the sense that normally the meaning of the latter is semantically derived (predictable) from that of the former. It is argued that this semantic link between the pair words of conversion is based on various types of conceptual, predominantly metonymic shifts whereby extralinguistic entities such as actions, instruments, properties, natural kinds, etc., undergo cognitive reanalyses (cf. instrument as action, property as action, action as actor/place) driven by the communicative needs of interlocutors. Consequently, along with the interpretations mentioned in the previous paragraphs, conversion can also be considered a word-formation process motivated by different types of conceptual shifts between formally identical input and output items.