Morphological complexity is mostly due to element order rather than to complexification

In this provocative talk, I point out that the prevailing view on “morphological complexity” has ignored the crucial role of element order and may thus have gone seriously astray. While earlier work (e.g. Greenberg 1960) merely observed differences in “morphological complexity”, more recent work has tried to identify exteral causal factors. In particular, McWhorter (2003; 2011; 2016), Lupyan & Dale (2010) and Trudgill (2011) have argued that particular sociolinguistic situations favour the development of “morphologically simple” languages, while others tend to lead to “complexification”. The causal mechanism of simplification through adult acquisition seems plausible, but is it specifically “morphology” that is hard to acquire, or grammatical markers in general (including clitics)? What exactly is “morphology”, and thus “morphological complexity”? Most straightforwardly, a morphologically complex word is one that has many constituent elements or requires many rules to be generated. But how are word boundaries drawn? There seems to be fairly widespread agreement that affixes exhibit host-class specificity (Zwicky & Pullum 1983; Haspelmath 2023), so that, e.g., the English definite article the is not a prefix (as it can occur not only before nouns, but also adjectives: the+cup, the+new cup), while the Swedish definite article -en is a suffix (as it occurs only after nouns: kopp+en ‘the cup’, den nya kopp+en ‘the new cup’, kopp+en där borta ‘the cup over there’). If this is adopted as the key criterion, then the affix status of a grammatical marker is co-determined by element order in a wide range of contexts. All of the following pairs of possibilities (where the marker is boldfaced) are richly instantiated in the world’s languages. affix non-affix (A) flag + noun (+ adjective) flag (+ adjective) + noun (B) (adjective +) noun + flag noun (+ adjective) + flag (C) (adjective +) noun + plural noun (+ adjective) + plural (D) aspect + verb (+ adverb) aspect (+ adverb) + verb (E) (object +) verb + relativizer verb (+ object) + relativizer (F) subject + verb + complementizer complementizer + subject + verb Flags and number markers can be affixes only if adjectives do not intervene, aspect markers can be affixes if adverbs do not intervene, and so on. Thus, affixhood is dependent on element order, and if “morphological complexity” is determined via affixhood, then it depends on element order. One might try to adopt other tests for distinguishing affixes from non-affixes, or hope to rely on test batteries. But none of these have been successful, and it is only for root changes that we have (near-)consensus about their morphological status. Thus, unless one wants to drastically restrict “morphological complexity” to root-change patterns, element order remains a crucial ingredient. The claims of sociolinguistic typology (which do have some prima facie plausibility) thus need to be approached more rigorously in the future, and it seems advisable to assess “complexity” also in terms of the presence or absence of grammatical markers regardless of their affix vs. clitic status.