Isomorphism-inspired theorising about optionality and variation is unsupported by data from English grammar

Language variation (specifically: optionality between different ways ofsaying the same thing, as in check out the places vs. check the placesout) tends to be considered abnormal, short-lived, and needlesslycomplex, especially in functional or cognitive linguistic circles. In thiscontribution, we are fact-checking these assumptions: does grammaticaloptionality increase the relative complexity (or: difficulty) of languageproduction? We use a corpus-based psycholinguistic research design witha variationist twist and analyse SWITCHBOARD, a corpus ofconversational spoken American English. We ask if and how grammaticaloptionality correlates with two symptoms of production difficulty, namelyfilled pauses (um and uh) and unfilled pauses (speech planning time).Our dataset covers 108,487 conversational turns, 22 grammaticalalternations yielding 57,032 optionality contexts, 589,124 unfilled pausesand 43,801 filled pauses. Analysis shows that overall optionality contextsdo not make speech more dysfluent – regardless of how many language-internal constraints are in operation, or how many variants there are tochoose from. With that being said, we show how some alternations in thegrammar of English are more prone to attract or repel productiondifficulties than others. All told, our results call into question old dogmasin theoretical linguistics, such as the Principle of Isomorphism or thePrinciple of No Synonymy.