We Should Let Spoken English Shed Some Rules

I know where “red” belongs in “big red car” before I can explain why. Reverse the adjectives and I hear something wrong, although the car has not changed. That intuition is knowledge. I would rather give an intelligible spoken variant room to take hold than correct it merely for breaking an inherited rule. That means accepting some speech that sounds wrong to me. An unfamiliar pronunciation should have to obscure something before I call it evidence of decline.

Gregory Scontras, Judith Degen and Noah D. Goodman tested native English speakers’ adjective-order preferences in a 2017 paper,^[1] asking which of two orders sounded more natural. Across 26 adjectives, subjectivity scores accounted for 85% of the variation in average ordering ratings; less subjective adjectives tended to sit closer to the noun. In the authors’ “big blue box” example, judgments about bigness leave more room for disagreement than judgments about blueness. In a second experiment with 78 adjectives, subjectivity still accounted for more than half the variation. Ordering preferences weakened when the adjectives were similarly subjective.

In the English 2019 corpus of Google Books,^[2] “big red” exceeded “red big” and “old brick” exceeded “brick old” in every decade from 1900 to 2019. The gaps widened and narrowed, but both usual orders were several hundred times as frequent as their reversals in the 2010s. That comparison uses equal-year means of unsmoothed annual frequencies among all two-word sequences in each year’s books.^[3] The queries count literal lowercase strings, without checking their grammatical role or surveying speakers.

Two line panels compare lowercase word pairs in the English 2019 book corpus, using decade means from 1900 to 2019 on a shared logarithmic scale in occurrences per million bigrams. In the 2010s, big red is roughly 350 times as frequent as red big, and old brick roughly 500 times as frequent as brick old. Both usual orders lead in every decade despite varying gaps; every plotted decade mean is positive, although some annual rates are zero. — In both word pairs, the usual order leads by at least 170 to one in every decade from 1900 to 2019: occurrences per million bigrams in English books, decade means of unsmoothed annual rates, log scale. Source: Google Books.^[2]

Laurel MacKenzie’s 2018 study of three spoken corpora of American English^[4] found that younger speakers almost always kept the “v” in “wolves.” For “houses” and its derivatives, however, the stem’s final “s” was pronounced as “z” in 65% of tokens from speakers born in the 1940s, compared with 38% from those born in the 1980s (139 and 92 tokens, respectively). MacKenzie suggests that “house,” alone in its class, lacks support from similar irregular forms, and that variation may begin in children’s learning errors. She also offers spelling as a possible reason that “wolves” keeps its “v.” An error might be tomorrow’s systematic innovation.

The strongest case for correction is that learners need a dependable shared standard, not reassurance that today’s mistake might become acceptable. Being understood is a low bar; conventional order is part of what an English learner is trying to master. That warrants teaching the default. It does not warrant treating every intelligible spoken variant as a failure. Give learners pairs to compare and the subjectivity principle to test against them, accepting a judgment call where a checklist would be easier. I can live with two pronunciations of “houses.” Put “big red car” beside “red big car” for an English learner and explain the preference instead of handing down a prohibition.

/bibliography