Lectal contamination: How lexical biases in linguistic variation may originate from lectal differences

Grammatical variation is often driven by lectal differences, i.e. differences between regio-, dia , ethno- or sociolects (e.g. Szmrecsanyi et al. 2016). For instance, the optional -s ending in the Dutch partitive genitives, as in (1)-(2), is far more often used in the Netherlandic regiolect of Dutch than in the Belgian regiolect (Pijpops Van de Velde, 2018). Meanwhile, lexical variation also often exhibits lectal differences (Speelman et al., 2008). For instance, the adjective bijzonder ‘special’ as in (1) is more often used in the Netherlandic regiolect, while the adjective speciaal ‘special’ as in (2) is more popular in Belgium. (1) Is er iets bijzonder(s) dan? Is there something special(-GEN) then ‘Is there something special then?’ (2) Is er iets speciaal(s) vanavond? is there something special(-GEN) tonight? ‘Is there something special tonight?’ As a result, lexical items that are typical of Netherlandic Dutch, such as bijzonder ‘special’, will more often occur with the variant with -s ending, simply because they are more often used by people from the Netherlands. Now, there is of course language contact between the Netherlands and Belgium (van Agtmaal-Wobma et al., 2007). As such, Belgian language users will more often hear the variant with -s ending among partitive genitives with typically Netherlandic lexical items, e.g. iets bijzonder(s) ‘something special’, than among partitive genitives with typically Belgian items, such as iets speciaal(s) ‘something special’. If we then assume that language users are susceptible to lexical biases in the language usage around them (Bybee, 2013; Diessel, 2017), we should expect that Belgian language users themselves also start to prefer the variant with -s among partitive genitives such as iets bijzonder(s), compared to partitive genitives such as iets speciaal(s). In other words, lexical preferences of one lect would ‘contaminate’ the grammatical variation in another. This hypothesized effect is therefore named ‘lectal contamination’. We have investigated lectal contamination in two ways. The first is corpus research. Here, it was found that partitive genitives with typically Netherlandic lexical items indeed appeared more often in the variant with -s ending, both among the language use of Belgians and Dutchmen, while controlling for other factors that are known to affect the use of the -s. The second is agent-based simulation (Beuls Steels, 2013). Here, we built a computer simulation that incorporated the following assumptions, (i) there are two grammatical variants, with one group of agents initially preferring the one, and the other group relatively preferring the other; (ii) both groups of agents also prefer different lexical items, and these grammatical and lexical preferences are initially independent of one another; (iii) there is a limited degree of language contact between the groups; and (iv) the agents store concrete combinations of a lexical items with a grammatical variant in their memory, i.e. they are susceptible to lexical biases. The results of this simulation show that the effect of lectal contamination indeed emerges under these four assumptions.