What I intended as a throwaway post on H. W. Fowler's analysis of the as an adverb in the the more the merrier construction engendered a torrent inundation freshet of emails. John Cowan pointed out that the article in this construction is in fact the last survivor of the Old English instrumental demonstrative, and Russell Lee-Goldman noted that the construction was hotly debated a century ago and that the analysis of the article as a legacy of the OE demonstrative is defended in a Fall 2005 Linguistic Inquiry article by Marcel den Dikken called "Comparative Correlates Comparatively" (available here), where the determiner is treated as the head of a phonologically null degree phrase. And Justin Mansfield observed that "[Fowler's] analysis is exactly how this construction is done in Latin: e.g. quo plures, eo laetiores "by how much [we are] more, by so much [we shall be] merrier," and added that "calquing the grammatical analysis of English constructions off of their Latin parallels [is] a major vice of traditional grammarians."
Well, yes, to be sure. But then Fowler was playing by different rules from modern syntacticians. Where the game now involves sorting through a vast number of analytic alternatives until they can be reduced to the categorical repertory of Universal Grammar, traditional grammarians were obliged to take their categories from off the shelf and constrict or deform the sprawl of syntax until one or another of them could be made to fit. Questions of empirical methodology aside, you have to wonder which exercise provided for a more gratifying display of ingenuity. Our method is clearly better suited to turning out scientists, but you'd have to give theirs the edge when it comes to training lawyers.
Posted by Geoff Nunberg at April 7, 2006 12:58 PM