Corroborating Corpus Data with Elicited Introspection Data: A Case Study

The last decades have seen an exponential growth of corpus sizes. This development has been driven by a desire to investigate rare syntactic phenomena, but issues remain: Corpora are by definition finite samples, but language is by definition infinite, leading to the negative data problem (‘absence of evidence is not evidence of absence’). One solution is corroborating corpus data with elicited introspection data that is obtained in a reliable, valid, and objective way. I present a case study to show how this can be done using the Magnitude Estimation Test (MET) method (Hoffmann 2013). Analyzing elicited data from 37 L1 English speakers, I show that introspective data can complement corpus data and lead to interesting new findings.