Data or datum?

By K.N.C. | LONDON

MESSY data are a fact of life: so much are unstructured or in non-standard formats. But what of the very term data itself? There, too, messiness, the unstructured and a lack of standards reign: there is no consensus if it should be plural or singular.

North American readers will have certainly felt jarred by the opening sentence. "Data are" doesn't roll of their tongues smoothly. But for British readers the plural is more natural. (The argument boils down to this: data in Latin is the plural of datum—but Latin is a dead language and English evolves.)

A debate has been stirred up in recent days. In a blog post last week by the Wall Street Journal, its style sergeant Paul Martin wrote: "Most style guides and dictionaries have come to accept the use of the noun data with either singular or plural verbs, and we hereby join the majority." But the post's author, Phil Izzo, suggested that "the plural will continue to dominate in our prose."

Kevin Drum at Mother Jones took a more data-driven approach. He analysed the frequency of the expressions "data is" and "data are" in books published over the past century using Google Ngram viewer. (His chart is republished below, with permission.) It shows that after a growing gap throughout most of the 1900s in which the singular was by far less common, the plural has dramatically waned over the past two decades. Nowadays in books, the plural is still preferred, albeit barely.

How does the populist wisdom of the internet compare? A search on Google (both .com and .co.uk) showed that the frequency of "data is" is almost five times greater than that of "data are" (ranking below).

The Economist's style book lists data under the heading "Plurals" and later, expands in a section meant to instruct correspondents as much as to amuse them:

"Propaganda looks plural but is not. Billiards, bowls, darts and fives are also singular. Data and media are plural. So are whereabouts and headquarters. Teams that take the name of a town, country or university are plural, even when they look singular: England were bowled out for 56."

This has led to a few funny situations. In a 14-page special report called "The data deluge" in 2010, the term data appeared no less than 97 times. This (American) author recalls being asked to check and change all instances where the grammar wasn't right. It meant occasionally having to paraphrase American experts rather than quote them directly. (To strike a balance so as not to sound too odd to American readers, I went through the report and added things like "a handful of" or "a swath of" or "a mountain of" before the word "data" so as to keep the verb in its singular form.)

Meanwhile, the New York Times's stylebook, like the WSJ, accepts it both ways. Our friends at the Guardian avowedly favour the singular. "[I]t sounds increasingly hyper-correct, old-fashioned and pompous to say 'the data are'," opined its style arbiter, David Marsh.