Over the last decade, I have often argued that grammatical coding patterns can be explained by frequency of use. In this blogpost, I provide a short summary of the claims for those who are not familiar with the argument.
What I’m claiming is not that I can explain language-particular patterns – the claim is entirely at the level of general linguistics, i.e. I am proposing an explanation of cross-linguistic tendencies. The tendencies that can be explained in this way are coding asymmetries, i.e. pairs of grammatical meanings that are in paradigmatic opposition and where one of the members shows a strong cross-linguistic tendency to be expressed by a longer form (which often means that the shorter form is zero). A few examples are:
singular – plural (cf. book-Ø/book-s)
nominative – accusative (cf. he-Ø/hi-m)
present – future (cf. Ø do/will do)
positive – comparative (cf. warm-Ø/warm-er)
referential noun – modifying noun (cf. Ø the dog/of the dog)
It is not just English that shows overt plural, accusative, future, comparative and modifier marking, but the world’s languages are very similar in this regard. For these (and many more) pairs of grammatical meanings, we can formulate implicational universals of the type: “If a language has a singular marker, then it has a plural marker”, or even more generally: “In all languages, the plural form is at least as long as the singular form” (see Greenberg 1966; Croft 2003: Chapter 4 for earlier discussion). In the past, such coding asymmetries were sometimes treated under the heading of “markedness”, but this is not an explanatory notion (Haspelmath 2006).
The explanation of these strong cross-linguistic tendencies comes from predictability: The grammatical meanings with shorter (often zero) coding are expressed much more frequently than their opposites, and as a result they are easier to predict and hence need no coding (or coding with short forms). This can be seen in any corpus of any language: Singular nouns are more frequent than plural nouns, nominatives are more frequent than accusatives, present-tens verbs are more frequent than future-tense verbs, and so on.
In a number of publications (some of them quite recent), I have made the case for frequency-induced predictability in some detail:
Haspelmath (1999): definite possessed nominals are more frequent than indefinite ones, and hence the article tends to be omitted (e.g. Kim’s (*the) house/a house of Kim’s)
Haspelmath (2008b): disjoint anaphoric forms are more frequent than reflexive forms, and hence universally shorter (e.g. her-Ø/her-self)
Haspelmath (2013): same-subject ‘want’ clauses are more frequent than different-subject ‘want’ clauses, and hence universally shorter (e.g. German sie will gehen/sie will, dass er geht ‘she wants to go/she wants him to go’)
Haspelmath et al. (2014): causal use is more frequent with verbs like ‘break’ (compared to verbs like ‘melt’), and hence these verbs tend to lack an overt causative marker for causal use (while ‘melt’ verbs tend to have one, e.g. French Ø casser ‘break (causal)’ vs. faire fondre ‘melt (causal)’
Haspelmath (2016): more generally, causal use is more frequent the higher a base verb is on the “spontaneity scale” of verbs exemplified as follows: ‘cut (tr.)’ < ‘talk’ < ‘melt (intr.)’ < ‘break (intr.)’ < ‘be cut’, and as we move up on the scale, overt causatives become less likely for causal use, while anticausatives become more likely for noncausal use.
Haspelmath & Karjus (2017): multiplex (“plural”) use is more frequent with nouns like ‘pea’ or ‘bee’, and hence these nouns tend to lack an overt plurative (“plural”) marker in some languages like Welsh, which have singulative markers for uniplex use instead (e.g. Welsh pys ‘peas’, pys-en ‘pea’).
Haspelmath (2017): possessed occurrence is more frequent with body-part terms and kinship terms (inalienable nouns) than with other nouns, and hence these nouns tend to lack an overt marker in an adpossessive construction (e.g. Mandarin ta-de chenshan ‘his shirt’ vs. ta mama ‘his mom’).
For some of these cases, an alternative explanation in terms of iconicity has been proposed, but I argued against such iconicity explanations quite some time ago (Haspelmath 2008a). There is a long tradition in functional linguistics to explain language form through isomorphism with linguistic meaning, and such explanations are of course plausible a priori. But when one looks more closely, they work only if one makes additional assumptions for which there is little evidence. For example, one could of course claim that Welsh pys ‘peas’ is really a collective noun (from which the singulative pys-en ‘a single pea’ is derived), not a plural form, and that it doesn’t have quite the same meaning as English ‘peas’. One could even go on to make the same claim for German Eltern-teil (‘parent’, lit. ‘parents-part’), which is derived from Eltern ‘parents’ (a plurale tantum in German, i.e. *Elter ‘parent’ doesn’t exist). But if there are semantic differences, they are very subtle, while the cross-linguistic coding asymmetries are robust and the frequency differences are quite striking. Moreover, languages are full of length differences that are hardly attributable to meaning differences but that follow from predictability: For example, ellipsis generally occurs when a meaning is predictable (“recoverable”) from the context, and frequent words are short (Zipf 1935) because they are predictable (there is no semantic difference between horse and hippopotamus that could explain the latter’s greater length). Thus, the predictability-based explanation is superior, it seems to me.
The idea is thus that language systems are adapted to their users’ needs in that preditable grammatical meanings are not expressed with a long form. This adaptation comes about through diachronic change which is guided by the “desired” result (though it is not claimed that the change is purposeful, of course). One further issue is whether the causality is perhaps the reverse, i.e. the observed cross-linguistic patterns might be due to usual paths of change (“mutational constraints”, as discussed earlier on this blog and in a forthcoming book edited by Schmidtke-Bode et al.). In other words, the cross-linguistic patterns might be explainable in a source-oriented fashion rather than through the adaptive results of the changes (as has been advocated in particular by Sonia Cristofaro). I do not want to exclude this as a possibility, but for most of the patterns listed above, it seems to me that the evidence clearly favours the result-oriented (adaptive) view over the source-oriented view.
Croft, William. 2003. Typology and universals. 2nd edition. Cambridge: Cambridge University Press.
Greenberg, Joseph H. 1966. Language universals: With special reference to feature hierarchies. The Hague: Mouton.
Haspelmath, Martin. 1999. Explaining article-possessor complementarity: Economic motivation in noun phrase syntax. Language 75(2). 227–243.
Haspelmath, Martin. 2006. Against markedness (and what to replace it with). Journal of Linguistics 42(1). 25–70.
Haspelmath, Martin. 2008a. Frequency vs. iconicity in explaining grammatical asymmetries. Cognitive Linguistics 19(1). 1–33. doi:10.1515/COG.2008.001.
Haspelmath, Martin. 2008b. A frequentist explanation of some universals of reflexive marking. Linguistic Discovery 6(1). 40–63. doi:10.1349/PS1.1537-0852.A.331.
Haspelmath, Martin. 2013. On the cross-linguistic distribution of same-subject and different-subject “want” complements: Economic vs. iconic motivation. SKY Journal of Linguistics 26. 41–69.
Haspelmath, Martin. 2016. Universals of causative and anticausative verb formation and the spontaneity scale. Lingua Posnaniensis 58(2). 33–63. doi:10.1515/linpo-2016-0009.
Haspelmath, Martin. 2017. Explaining alienability contrasts in adpossessive constructions: Predictability vs. iconicity. Zeitschrift für Sprachwissenschaft 36(2). 193–231.
Haspelmath, Martin, Andreea Calude, Michael Spagnol, Heiko Narrog & Elif Bamyacı. 2014. Coding causal-noncausal verb alternations: A form-frequency correspondence explanation. Journal of Linguistics 50(3). 587–625. doi:10.1017/S0022226714000255.
Haspelmath, Martin & Andres Karjus. 2017. Explaining asymmetries in number marking: Singulatives, pluratives and usage frequency. Linguistics 55(6). 1213–1235.
Schmidtke-Bode, Karsten, Natalia Levshina, Susanne Maria Michaelis & Ilja A. Seržant (eds.). 2018. Competing explanations in linguistic typology. To appear.
Zipf, George Kingsley. 1935. The psycho-biology of language: An introduction to dynamic philology. Cambridge, Mass.: M. I. T. Press.
Do Welsh verbs have number agreement?
Except, artificially, as a technical term in Mendelian genetics. Of course it’s masculine.