A century ago this year, just before the First World War began, the grammarian George O. Curme published a short but remarkable paper entitled “Origin and Force of the Split Infinitive” (Modern Language Notes 29 (2), 41–45). It has deep roots in the 19th-century tradition of critical analysis of English grammar. And it is sobering to compare his paper’s meaty content with the thin gruel that passes for discussion of English grammar today.
Curme is following up works such as the splendidly acrid 1882 article by Fitzedward Hall (American Journal of Philology 3, 17–24), which was the first scientific defense of the “split infinitive” construction against uninformed critics like Dean Alford. Hall “laid bare the glaring ignorance of the opponents of the construction” (as Curme puts it). Curme was aiming at an improved account of the historical development, but along the way he offers some interesting descriptive insights supported by compelling arguments.
He rejects careless references to “emphasis” in discussions of grammar. Thomas Lounsbury, in The Standard of Usage in English (1908), notes that Lord Macaulay rewrote the phrase “in order fully to appreciate the character of Lord Holland” (in an 1841 essay) to make it “in order to fully appreciate the character of Lord Holland” (in an 1843 revision), and assumes he was striving for “additional emphasis.” Curme calls that a “blunder”: The best place for extra emphasis on the adverb would be after the whole complement (“in order to appreciate the character of Lord Holland fully”).
Curme makes one observation that I had thought much more recent: that in some cases infinitives must be split (unless you simply abandon the attempt to use the adjunct). He cites an example containing the phrase sufficient to more than offset the losses. It cannot be recast as *sufficient more than to offset the losses or *sufficient to offset the losses more than : These are ungrammatical.
Curme also addresses a hypothesis due to the great Otto Jespersen: that the split infinitive became permissible because “the linguistic instinct now takes to to belong to the preceding verb” rather than the verb of the infinitival complement it introduces. “This theory is destroyed,” says Curme, by data such as To almost succeed is not enough, where there is no preceding word for to to depend on.
Curme’s own story about how the construction evolved involves:
- the change to the present practice of having most of a verb’s complements and modifiers following it (present-day “to expound true love”) from the earlier pattern of having them before the verb (early 14th-century “to trwluf expoun”);
- the increasing popularity of infinitival complements (“I intended it to be seen”) over subjunctive complements (“I intended that it be seen”); and
- ambiguities in negated clauses (once upon a time They learned not to blaspheme could mean either “They learned to avoid blasphemy,” which is now the only possibility, or “They didn’t learn to blaspheme”).
Curme closes by predicting that the innovative form (the so-called split infinitive) “will probably in due time become established to the exclusion of the other form.” A century later we are broadly on track toward the fulfilment of this prophecy, but Curme did not fully foresee the degree to which bossy copy editors and opinionated idiots would slow things down by continuing to spread the myth that the split infinitive is some kind of mistake. (The editors for Nature and The Economist insist on “correcting” split infinitives. Nevile Gwynne’s mendacious pothering on the topic is beautifully dissected and rebutted by Tom Freeman in this post on The Stroppy Editor. And Simon Heffer’s execrable book Strictly English, which I review here, is yet another source of atavistic advice.)
What I want to stress about Curme’s article is not undercut by the fact that his 100-year forecast has not fully come true yet. It is that Curme thinks about grammatical analysis the way a scientist thinks about natural phenomena. He doesn’t just reiterate dogma, or voice emotional reactions, or offer intuitions about invented sentences; he tries to draw sensible conclusions from facts about sentences that have actually been used.
His article is a model of modern rational argumentation about grammar. Comparing it with the ignorant emotionalism we find in popular discussions of grammar a century later reveals a lot about how the study of the English language has retrogressed.
One other thing: There is a completely different aspect of Curme’s article, quite separate from its insightfulness and modernity, that also astonished me. It will be instantly apparent to anyone who downloads the article from JSTOR. But it’s a whole different story, which I’ll write about next week.Return to Top