I've been thinking about vocabulary a lot recently. I've tried to teach vocabulary in a "robust" way in recent years (offering friendly definitions, using the words in context, playing fun games, etc.), but I haven't had much success. Some kids knew the words already, some didn't; some studied, some didn't; most forgot the words by the end of the year. I was dispirited. Then, while following the common core debate, I noticed that in a recent letter in Education Week Linda Diamond defended the National Reading Panel Report and its emphasis on skills, "explicit instruction," and vocabulary instruction ("Common-core standards in reading not 'flawed,'" March 28). I was most interested in the question of vocabulary instruction, and I decided to try to figure out who was right, the Common Core Standards, the National Reading Panel, and the "What Works" Clearinghouse, or those who argue that explicit vocabulary instruction is probably less valuable than other activities, like reading itself.
My first step
was to look up the government's "What Works" publication on adolescent
literacy (http://ies.ed.gov/ncee/wwc/practiceguide.aspx?sid=8). I found
that this apparently authoritative, evidence-based publication said
there was "strong" empirical evidence for explicit vocabulary
instruction. When I followed up on their evidence, however, I was
surprised to find that there was almost nothing there--and that some of
what their strongest evidence seemed rather to question the value of
explicit vocabulary instruction.
The strongest evidence
cited by the What Works report seemed to be in the following passage:
"Children often learn new words from context. However, according to a
meta-analysis of the literature, the probability that they will learn
new words while reading is relatively low--about 15 percent. Therefore,
although incidental learning helps students develop their vocabulary,
additional explicit instructional support needs to be provided as part
of the curriculum to ensure that all students acquire the necessary
print vocabulary for academic success.
interesting, but a little obscure. If the probability of learning new
words while reading is "relatively low--about 15 percent," what exactly
does that mean? 15 percent of what? Does that mean that 15
percent of my students learn NO words at all in the course of their
reading? That would be terrible. Or does it mean that for any
occurrence of a word they don't know, there is a 15 percent chance of
their learning it? That doesn't sound so bad.
I followed the citation to a 1999 paper by a couple of researchers in
Amsterdam, researchers who wrote in perfect English, of course. (An
aside: how did they acquire their excellent English vocabulary? Not, I
imagine, from much explicit instruction in vocabulary, but I could be
wrong). These Dutch academics, Swanborn and De Glopper, had reviewed a
number of studies of vocabulary acquisition from what they charmingly
called "natural reading." They noted that some uncertainties remained,
because there seemed to be great variability among students in how many
words could be learned incidentally, and it was unclear also how many
unknown words students encountered in their "natural reading," but the
researchers concluded on a positive note: "What we do know, however,
from our meta-analysis, is that students have a fair chance of learning
unknown words from reading. Natural reading has the potential to make a
contribution to vocabulary growth." This is, strikingly, at odds with
the way the government publication interpreted their article. So I
decided to look more closely, at the 15 percent that the What Works
article had claimed was too low a number.
Swanborn and de Glopper, students encounter unknown words at a rate of
at least one percent. That is, in a thousand words of text, ten of them
will be unknown to a student who is reading a book that's comfortable
for him to read. Of those unknown words, 15 percent will be learned
without any conscious effort. The What Works authors deemed this too
low a number. But how many words would a student learn at this rate?
Say a student read ten pages a day, hardly impossible, and say each page
had three hundred words, also a low estimate. Then in a week the
student would have read 7x10x300 words, or 21,000 words. Of those words,
at least one percent, or 210, would be unknown. Of those 210 unknown
words, the student might be expected to learn 15 percent, or 31 words.
So, according to the meta-analysis that the What Works authors cited to
show the inadequacy of natural reading as a way of improving one's
vocabulary, students who are reading at the relatively slow pace of 70
pages a week could be expected to learn 31 new words a week. At my
school, we have had a big push in recent years to teach more vocabulary,
and many teachers are spending as much as 10 or 15 percent of their
class time to explicit vocabulary instruction. But even with this
extraordinary expenditure of time and energy, no teacher is teaching her
students more than 10 words a week, at the most, and few students are
actually learning all ten of those words. With the hour a week that we
are spending on vocab, our students could be reading another thirty
pages, thereby learning another 13 words, and also accruing all the
other benefits that reading brings.
It seems that
having a 15% chance of learning new words is far from "too low";
instead, it is wonderful and promising. So the main evidence cited by
the What Works authors does not support their argument that explicit
vocabulary instruction is needed.
Natural reading may
work to improve vocabulary. But what about explicit vocabulary
instruction? Maybe research shows that explicit instruction is very
effective--even more effective than natural reading, despite my own poor
results. So I looked at some of the research the report cited and I
looked at some papers I found elsewhere, and NOWHERE could I find clear
empirical evidence that explicit instruction in vocabulary would lead to
more word acquisition than just plain reading, nor that the word
acquisition that was achieved in any of the studies had actually
increased comprehension. As Baumann et al. say in their 2003 paper, "causality regarding vocabulary-to-comprehension relationships [...] remain [sic] murky."
is typical of my experience with educational research. The claims
people make about what is supported by the data are often strikingly at
odds with what the data actually support. There is no doubt that good
readers usually have good vocabularies, and there is no doubt that they
acquired their good vocabularies somehow, but it is very far from clear
how they did, and it is very far from clear what teachers can do to
help. In the absence of much clearer evidence that explicit instruction
is significantly better than just reading, I think we should mostly
stick with just reading. That said, I am still going to do some
vocabulary stuff in my classes next year.
This year, after I had my students learn vocabulary words drawn from the books we read
as a class, they didn't make much progress. Next year I am going to have them pay attention to words
in the books they read on their own and make their own vocab tests from
those words. I also hope to be very intentional about using a lot of
higher-order words in class myself. A few weeks ago I used the word "behoove"
a few times, and many of my weakest students loved it. I'm skeptical
about whole-class word lists, but I hope that modeling and encouraging
word-love (and upping the reading volume) can make a difference. We'll see.