Dear Cecil: A speaker at a recent school board meeting claimed the vocabulary of the average American grade school student was 25,000 words in 1945 and about 10,000 today. This is pretty disturbing if true. What do you think? Dave Evans, Bellingham, Washington
I’ll tell you what I think: with nonsense like this spreading unchecked, we should be worried about our kids’ critical reasoning, not their vocabulary. A 20 percent decline, maybe, but 60? No chance.
Despite being plainly absurd, your factoid has appeared in newspapers, magazines, and even science journals, each typically citing other such appearances as backup. At times it’s been attributed to Gallup polls or even entomologists. I’d nearly despaired of unraveling it when I happened on a Harper’s from 1990. Bingo: it claimed the average written vocabulary of 6- to 14-year-olds was 10,000 words, down from 25,000 in 1945; named as sources were a book by Henry Rinsland for the latter figure and a report by Gary Ingersoll for the former.
Soon I found much the same information in a contemporaneous New York Times item. On examining the original research, however, it became clear someone had misconstrued it. Rinsland’s 1945 study tabulated words used by kids in grades one through eight. But the number Harper’s called an “average” was actually a total. Reviewing 100,000 student compositions containing six million words altogether, Rinsland recorded 25,632 distinct word forms.
The 1984 monograph by Ingersoll and Carl Smith was similar, but its sample was much smaller: 5,000 students, half a million words. With a corpus only a 12th as large, of course, you’d expect a far lower count of distinct forms. Using rough estimates derived from the American Heritage Word Frequency Book, a 500,000-word sample should have around 60 percent fewer unique words than a six-million-word sample independent of the writers’ vocabulary. As it happens, Ingersoll and Smith reported 10,265 forms – just about 60 percent fewer.
The authors never called this figure an average, nor did they pair it with Rinsland’s tally of 25,000. But some nameless journalist, not thinking too hard, stuck the two numbers together, and voila: further proof of society’s decline. (Contacted recently, Ingersoll recalls trying in vain to set the record straight.) It’s not apples and oranges, but no one paying attention could compare these apples without noticing (a) one is 12 times bigger and (b) both of them are rutabagas.
OK, so the stat is bunk. But is the average vocabulary shrinking? Maybe; it’s unclear.
The General Social Survey, conducted regularly among U.S. adults since 1972, includes a ten-question vocabulary test. Experts dispute the test’s validity; that said, the high scorers through 1990 (with 6.5 correct answers) were born around 1945, the low scorers (at 5.1) around 1975. Evidence of a sizable vocabulary loss? Hang on. First, the test words come from a list originally compiled in the 1920s. (This longer list has been published, but the subset of words actually used is secret.) Some have doubtless grown more rarefied over time; conversely, you get no credit for knowing words coined in the last 80 years.
More important, it’s hard to separate the effects of age (i.e., the subject’s at the time of GSS testing), cohort (his or her birth year), and period (the year testing took place). What looks like a cohort effect – i.e., people born in 1945 know more words than people born in 1975 – may well be an age effect: We know that vocabulary knowledge peaks in middle or old age. Hence (goes the reasoning) the aging boomers of the 1945 birth cohort test better (for now) than the still-callow cohort of ’75. This seems to be supported by subsequent GSS stats: subjects born between 1969 and ’76 averaged 4.9 when tested at ages 18 to 20; others from these cohorts scored 5.9 a decade later.
Researchers arguing for a cohort effect don’t automatically conclude worsening education is to blame. People who frequent school board meetings, showing less restraint, generally assume schools have gone to hell since that 1945 cohort passed through. Though belief in a bygone golden age of education is widespread, there’s little to support it: federal reading-assessment scores have held steady since the first test in 1969; an Indiana study from 1976 showed virtually no change in reading skills since 1945. (Not that this is anything for schools to crow about.)
Admittedly, experts who see no cohort effect in the GSS data acknowledge some period effect – the later the vocabulary test is conducted, the lower the scores across all groups, if only slightly. Whatever it’s called, researchers variously attribute this small drop to less reading overall, the dumbing down of reading material, the demise of intelligent conversation, or the ascent of TV.
But even if our vocabulary is dwindling, so what? English, having by some counts the largest vocabulary of any language, surely contains more words than we really need. We’d be no poorer if desuetude, for one, fell into a state of itself.
Send questions to Cecil via firstname.lastname@example.org.