If you think twice, you might not forget: inhibition and memory-maintenance trade-offs in Russian single vs. Hausa bipartite negation

Abstract

Several approaches have attempted to explain the skewed distribution across languages and regions of bipartite compared to single negation, the latter unequivocally the most common, but still often subject to diachronic change. It is a paradox that double negation arises, among other things, to strengthen negation semantically or define its scope, while at the same time, this fortification may lead to its demise. Several grammaticalization pathways result in double, discontinuous negation typologies negating single events or constituents, but such constructions seem to be inherently unstable. Here, I look into the possibility that when negation-induced forgetting, or NIF, a cognitive phenomenon which presumably affects all types of negation, is applied in bipartite negation, it simply becomes too much for memory to handle. NIF research has established that when negated sentences are processed, the mind simulates the affirmative as well as the negative version in parallel, straining processing capabilities. Working memory limits the number of units to be kept actively in mind to about four chunks, which directly impacts on how many syntactic constituents, and how long and complex they are, may intervene between a first and a second negator, in the case of double discontinuous negation. In the case of Hausa, this has the effect of moving the second negator forward to the next syntactic landing site, even if this means bisecting a negated constituent. Such behavior is absent in Russian simple, contiguous, and compact negation, undoubtedly the most common negation type across languages and regions. The possibility that Russian, too, might be in the process of developing a second negator is considered, and rejected. However, Russian single negation also faces memory-related drawbacks; it is argued that focal negation and late negator placement in incremental syntactic processing logically leads to deeper NIF effects compared to distributed and early negator systems. Evidently, linguistic negation typology does not select or reject negation strategies based on how deeply a negated content is forgotten; instead, memory might index negation strategies given that the same general neurocognitive mechanisms that produce NIF: activation, inhibition, prediction, and working-memory allocation, are also the mechanisms that shape real-time language processing.