There is a tradition of dividing all of (non-compounding) morphology into inflection and derivation in European languages, and like other European traditions, this one has been carried over to languages around the world. But on what basis?
There are clear cases of inflection and derivation where nobody would argue (e.g. instrument nouns like German Schlüss-el [close-INSTR] ‘key’ are derivational, and gender agreement forms like Italian buon-a [good-F.SG] are inflectional). But there are many cases, even in European languages, where linguists would be divided – for example in the case of adverbial -ly in English (e.g. Zwicky 1995).
Some people have argued that there is a continuum between inflection and derivation (Bybee 1995: ch. 4) or that there are more or less prototypical instances of the two categories (Dressler 1989). Plank (1994) applies a wide range of criteria and finds that there are many mismatches of the criteria. It seems that we have not found a way of dealing rigorously with this distinction.
Now Bauer & Bauer (2012) have examined Maori morphology from the point of view of the inflection-derivation distinction. Laurie Bauer is a distinguished morphologist and Winifred Bauer is a leading Maori expert, so their paper looked interesting to me. But the first disappointment was that they don’t even raise the possibility that Maori might have a lot of inflection that is generally not recognized as such because the elements are written separately. Take an example such as (1) (p. 10):
(1) He uaua te mohio-tia o nga whakaaro o nga kaihaina
TNS difficult ART know-PASS GEN ART thought GEN ART signatory
‘It is difficult to know what the signatories had in mind.’
Here only the passive suffix -tia is recognized as morphological, while the tense-marker he-, the articles te- and nga-, and the genitive marker o- are treated as particles. However, it seems that they could equally well be treated as inflectional prefixes. So their notion of inflection seems to be spelling-based.
AS a result, they consider only eleven morphological elements that are written jointly with the base: nominal plural marking (e.g. tangata/tāngata ‘person’), singular marking on determiners, the multiplicative numeral prefix hoko-, the nominalizing prefix -(C)anga, the agent noun prefix kai-, the causative prefix whaka-, the passive suffix -(C)ia and a few others. They consider eight different criteria, and again find a lot of mismatches, like Plank (1994). But like most earlier authors, their conclusion is not that something is wrong with their methodology. Instead, they decide to reduce the number of criteria that are allowed to count: When only productivity and agreement are considered, it turns out that only the nominalization and the passive suffix are inflectional, while all others are derivational. On such a definition of inflection and derivation, it turns out that Maori inflection is always suffixing, while Maori derivation is always prefixing.
But have we learned anything from this exercise? I’m not sure. It certainly is not rigorous, because there are no good reasons for choosing these rather than other criteria (this is a very clear example of what Bill Croft has called “methodological opportunism”). The comparison with inflection and derivation in English, Warrwa and Ura does not really tell us anything, because inflection and derivation have been identified quite differently for these languages.
The very question would seem to make sense only if one assumes that all languages must (potentially) have a distinction between inflection and derivation that is of the same nature (even if the manifestation may differ). But why should this be the case? Is there anything apart from the Western tradition that makes us think that this is how language works? It seems that the primary reason for distinguishing between inflection and derivation is descriptive convenience: Language description is generally divided into grammar and lexicon, and while derived forms such as German Schlüss-el ‘key’ of course need to be listed in the dictionary, inflected forms like Italian buon-a should of course be treated in the grammar.
But this only has to do with linguists’ descriptive habits – there is no reason to think that speakers have grammars and lexicons that are totally divided, and the lexicon-grammar continuum is now a widely accepted view (e.g. Jackendoff 2002: ch. 6). Thus, it seems best to stop worrying whether an element is inflectional or derivational, and to concentrate on its properties. Bauer & Bauer note that “it is often difficult to determine from descriptive grammars precisely what should be considered inflection and what derivation in that particular language” (2012: 20). This must be because descriptive grammars often do the right thing: They describe the language with categories that are needed for it, not with categories inherited from Western tradition.
Bauer, Laurie & Bauer, Winifred. 2012. The inflection-derivation divide in Maori amd its implications. Te Reo 55: 3-24.
Bybee, Joan L. 1985. Morphology: A study of the relation between meaning and form. Amsterdam: Benjamins.
Dressler, Wolfgang U. 1989. Prototypical differences between inflection and derivation. Zeitschrift für Phonetik, Sprachwissenschaft und Kommunikationsforschung 42(1). 3–10.
Jackendoff, Ray. 2002. Foundations of language : brain, meaning, grammar, evolution. Oxford: Oxford University Press.
Plank, Frans. 1994. Inflection and derivation. In R.E. Asher (ed.), Encyclopedia of Language and Linguistics, vol. 3, 1671–1678. Oxford: Pergamon.
Zwicky, Arnold. 1995. Why English adverbial -ly is not inflectional. Chicago Linguistic Society 31
Ok, but couldn’t this be boiled down to a distinction between ‘descriptive categories’ and ‘comparative concepts’? Your criticism of B&B (2012), and by extension, a principled cross-linguistically applicable distinction between inflection and derivation, makes sense if these notions are taken as categories of universal grammar or ‘cross-linguistic categories’ (as you note in the next-to-last paragraph).
But if inflection and derivation are thought of as comparative concepts, then maybe this isn’t a question of wrong or right but rather of ‘helpful or unhelpful for the purposes of comparison.’ As you point out in your paper on the topic, this can be evaluated by the kind of testable predictions the comparative concept allows one to make. So however one defines ‘inflection’ as a cross-linguistically applicable comparative concept, one could say whether a specific phenomenon in a given language matches it or not. In practice, this is what linguists like Anderson and Heine did in their terrific books about auxiliary verb constructions, and Schmidtke-Bode (explicitly) for purpose clauses. Incidentally, this would open the door for things like ‘Maori Inflection’ or ‘Coptic Derivation’ as language-specific descriptive categories, which would not only allow one to use the terms as one needs to describe the facts of Maori or Coptic (or to not use them if unnecessary), but it would also be in line with proposals about the ultimately language-specific nature of other allegedly universal notions like grammatical relations, etc.
I see two possible counter-arguments here. One is that the terms ‘inflection’ and ‘derivation’ are already too tainted for this sort of distinction, but we often recycle terms by redefining them, so that isn’t too serious. Another is that ‘inflection’ and ‘derivation’ are composite notions, and we shouldn’t be looking at them as single parameters but rather as aggregates of properties that should be compared individually across languages, like Bickel & Nichols 2007 did for some common morphological notions). That seems like the more substantive counter-argument.