- I've just had dinner with a friend who has recently moved from EFL to secondary; at his new school, they're being given a course by some educational psychologist who claims that the problem with pair or group work is that students tend to learn the errors made by their peers. The course isn't just for language teachers, but geography, history, maths.......teachers too.
He asked me if I knew of any published research that refutes this negative aspect of pair/group work, and we discussed Stevick, Krashen, Skehan, Thornbury and the obvious pluses of..........but
Is there a study that tends to prove that students don't necessarily pick up errors? My mate says he knows (we know) instinctively, and from years of experience and training, the benefits of p/g work, but we couldn't think of anything that might convince geography teachers.
Any ideas or recommendations?
[Non-text portions of this message have been removed]
There is all that great stuff in "Talking to Learn" (RR Day eds 1986,
Rowley MA: Newbury House): Porter and Rulon & McReary found that the
main differences between teacher led work and groupwork, besides the
obvious fact of increased learner output, were that the groups tended
to have a LOT more negotiation and questioning.
Craig Chaudron, in his 1988 "Second Language Classrooms" does a whole
review of studies on this question and decides that there is no
significant amount of error transfer learner to learner (CUP).
Here's Ellis (R.), p. 599 of his doorstopping brick of a book, "The
Study of Second Language Acquisition":
"Both Pica and Doughty, and Porter, show, not surprisingly, that
interlanguage talk is less grammatical than teacher talk. It is
possible, therefore, as Plann (9177) has suggested, that exposure to
incorrect peer input may lead to fossilization. However, two of
Porter's findings give reason to believe this may not be the case.
She reports that when learners corrected each other's errors, they
did so wrongly only 0.3 percent of the time and also that only 3
percent of the errors produced could be attributed to repetition of a
fellow student's error. In general, therefore, learners do not appear
to be unduly disadvantaged by exposure to deviant input from other
This is slightly incredible to me. How can Porter tell the difference
between an error that was learned from another learner and an error
that was shared to begin with? If language learning is basically a
matter of "pick up and practice", and "listen and repeat", how could
leaners avoid picking up and repeating each other's errors?
I think we can explain it this way. I argued last time that there are
TWO possible ways that "inter-psychological" language can
One is, as I suggested, when people summarize discourse as grammar,
by conglomerating constructions made collectively across many turns
as a single turn. Here's some of my video data again:
T (explains in Korean that Pinnochio is trying to use English
homework to fool Gepetto about the reason for his after-school
T: I studied. (in Korean: Right. What's next?)
S2: I homework.
S3: I did
T: I...(in Korean: Yeah, you got it, it needs a verb, doesn't it?)
T: (in Korean: Yeah, great. Let's hear it again!)
Ss: I did!
S1: I did!
S2: I did my...
Ss: I did my homework! Ya!!!
"Vertical becomes horizontal" and a discourse co-construction is then
summarized as a grammatical one, created collectively but now owned
individually. What was inter-psychological becomes intra-so--as the
students very properly say, Ya!!
You can see that this way of constructing is subject to at least
THREE kinds of checks on errors. First of all, two heads are better
than one. Secondly, the teacher is standing watchfully by. But
thirdly, and I think most importantly, the horizontal "summarizing"
of the vertical construction involves every child recreating the
grammar in his/her own head, according to his/her own rules and
subject to all the critical perfectionism that involves. Children who
are not ready to defy the teacher are very ready to vie with each
other in building a better version of a collectively authored
The second way is, as Scott suggests, by analysis of input. This is
frequency based, and only at the point where there are
enough "deviant" inputs to create a whole grammar are you going to
get consistent deviation, at which point deviation becomes the norm,
and a whole dialect, and not just an error, comes into being. Vive la
This may not be of great help, but I seem to recall reading an article
on the subject in the English Language Teaching Journal, probably late
90's. Someone else may recall the exact issue. If you have the CD_ROM
compilation, I believe it is fully searchable ... Unfortunately, I
haven't got it.
Sorry I can't offer anything more specific.
On Saturday, November 2, 2002, at 03:20 AM, Fiona M wrote:
> I've just had dinner with a friend who has recently moved from EFL to
> secondary; at his new school, they're being given a course by some
> educational psychologist who claims that the problem with pair or group
> work is that students tend to learn the errors made by their peers. The
> course isn't just for language teachers, but geography, history,
> maths.......teachers too.
> He asked me if I knew of any published research that refutes this
> negative aspect of pair/group work, and we discussed Stevick, Krashen,
> Skehan, Thornbury and the obvious pluses of..........but
> Is there a study that tends to prove that students don't necessarily
> pick up errors? My mate says he knows (we know) instinctively, and from
> years of experience and training, the benefits of p/g work, but we
> couldn't think of anything that might convince geography teachers.
> Any ideas or recommendations?
> [Non-text portions of this message have been removed]
> To Post a message, send it to: dogme@...
> To Unsubscribe, send a blank message to: dogme-unsubscribe@...
> Your use of Yahoo! Groups is subject to
- Fiona asked (11/2 ""Peer-work") re. pair and group work: "Is there a study
that tends to prove that students don't necessarily pick up errors? ...We
couldn't think of anything that might convince geography teachers.
I passed your question on to Marc Helgesen who has spoken and written on
pair and group work. His response is below.
I don't know of anyting recent (in part, perhaps) because this is
such a non-issue in elt.
This is an ancient reference but:
Long, M & P. Porter, June 1985 "Group Work, interlangauge talk and
second language acquisition" TESOL quarterly 19 (2) is sort of the
classic piece. It found peer talk superior to teacher-led "lockstep".
>Is there a study that tends to prove that students don't necessarily pickThis is an aspect of the errors and correction issue. To my
knowledge, there has never been any evidence that error correction
works (language planning, however, does increase accuracy. One
article is Foster, P and P. Skehan (1996), Studies in SLA, 18,
299-323. I don't know that this will be of any help but two
arguments I would use are:
a. If the ELT profession which specifically studies language
learning/acquisition doesn't worry about it, the teachers in other
subjects should follow that lead.
b. Little kids, still learning L1, talk to each other . I've never
seen one get stuck doing infantile language as they grew up.
All learners need continued input. Reading is especially good since
the learner controls the speed on input (i.e., can make use of
bottom-up and top-down skills.
Let me know if you (or the writer) needs copies of either of the
------end of Marc's response
- Peer-peer error transfer a NON-ISSUE, Marc? I don't think so.
Actually, the problem of "classroom dialects" is at the root of a lot
of very recent stuff, from Ohta (See "Pragmatics in Foreign Language
Teaching, Kasper and Rose, CUP 2001) to Jennifer Jenkins!
For this reason there's a great round-up on peer-peer interaction in
the latest issue of the Annual Review of Applied Linguistics by
Merrill Swain, Lindsay Brooks, and Agustina Tocalli-Beller. It's true
that the notion of error contagion doesn't get much discussion, but
this is from p. 179:
"Not all the peer interaction was error free, but Ohta found,
contrary to a previous study by Mackey, McDonough and Kim..."
(I'd like to get my hands on that, but it's not cited in the index--
anybody seen this?)
"that incorporation rates of incorrect utterances were very low.
According to Ohta, the benefits of peer interaction outweigh any
negative effects, as through scaffolding, learners build bridges to
Ohta claims that even LESS proficient people can help MORE proficient
people. Swain says this too--and argues that this is how performance
comes to lead competence.
Actually, we observed a fair amount of this learning, ye, out of the
mouths of babes and sucklings, when we did a study here in Korea
(desk-drawer publications, I'm afraid) that showed, rather
incredibly, that "Low Output Generators" (LOGS, for short) paired with
High Output Generators (call 'em HOGS) performed BETTER than two HOGs
in a pair.
I remember explaining this by using my old China "English Corner"
experience. In "English Corners" (the corner of a public park which
in many Chinese cities and almost all Chinese universities is set
aside for free English practice) you often find flocking behaviour
around the rare native speakers who show up.
These flocks soon become listening circles rather than speaking
circles, and the rare non-native speakers who dare to ask questions
are usually either boring or ungrammatical or both. On the other
hand, the non-native groups (which never create dense flocks) are
much more dialogic, and the same learner will often work harder and
better in a non-native group than in a dense flock around a native.
Maybe it's a symbiotic relationship. The LOG battens on the rich
input provided by the HOG, but the HOG has room and reason to stretch
his/her own abilities which would not occur in the presence of an NS
or even another HOG.
Another possibility occurs to me, though. Remember the old Donald
Duck cartoons, in which Donald's three nephews, Hughie, Louie, and
Dewey would co-construct a single sentence (frequently a rhyming one)?
HUGHIE: Don't look now but...
LOUIE: Use your noodle 'cause...
DEWEY: Your're being followed by...
(fierce looking dog taps Uncle Donald on the shoulder)
DONALD: Sergeant McPoodle! I wasn't doing anything...!
Well, Harvey Sacks picks up a lot of examples of this in NS-NS
JOE: We were in an automobile discussion
HENRY: discussing the psychological motives for
MEL: drag racing on the streets
(Lectures on Conversation, Blackwell 1995, p. 136)
Well, all right, it doesn't rhyme, but it adds up to something much
more grammatically complex and informative than the sum of its parts.
As stand alone utterances, actually, Henry's and Mel's additions are
ungrammatical, and Joe's "head" phrase is unhelpfully vague and
uninteresting to a teenager.
We do it too. That is, after all, what a thread is!