Recalcitrant problems of comparative alternation and new insights emerging from internet data

This paper adduces novel support for the claim that a theory of processing efficiency is best suited to explain the morpho-syntactic variation involved in comparative alternation, i.e. the choice between the synthetic comparative form fresher and its analytic variant more fresh. It introduces three studies investigating semantic, cognitive and pragmatic determinants that have hitherto not been subjected to empirical investigation. In order to arrive at a sufficiently sized database, conventional mega corpora comprising 600 million words will be supple¬mented by web data. The first study takes as its starting point independent psycholinguistic evidence showing that abstract concepts require a higher processing load than concrete ones. As regards the choice of comparative form, the analytic variant is favoured for abstract rather than concrete meanings. Thus, the comparative of a fresh approach is more often formed analytically than that of a fresh taste, etc. Secondly, the paper provides indi¬ca¬tions that even historical analyses can benefit from using web data. Cases of gradual increase (e.g. they became friendlier and more friendly vs. ?they became more friendly and friendlier) appear to reflect an iconic ordering in which more form encodes more meaning. And finally, the choice between comparative variants in compounds/derivatives such as broader-based vs. more broad-based will be shown to correlate with the entrenchment of a parallel ADJ N structure (e.g. a broad base)