E is for Error

9 05 2010

“It’s self-evident,” wrote an MA student of mine recently, in an online forum, “that most learner errors are caused by mother tongue interference”.  Is it really self-evident? It was certainly self-evident in the mid-twentieth century, when the notion of interference reigned supreme.  But the advent of interlanguage studies put paid to that.  The new science of error analysis (as distinct from contrastive analysis)  suggested that many – some would say most – errors are the effect of developmental  processes and performance demands, and have nothing to do with the learner’s L1. This is evidenced by the fact that many errors are shared by learners from different language groups, and occur in a similar developmental sequence and under comparable processing conditions.

Büyük Han, Nicosia

Nevertheless, the idea that errors are caused by negative transfer is a persistent one and is still invoked in order to justify proscribing translation activities in the classroom (see T is for Translation). I was intrigued, therefore, to find that the case against L1 interference in fact predates the work of Pit Corder and Jack Richards in the 1960s and 70s, judging by a book I found in a second-hand bookshop in Nicosia this week. (The photo shows Büyük Han, the restored Ottoman inn in one corner of which the bookshop was nestled). The book is called Common Errors in English: Their Cause, Prevention and Cure (!), by F.G. French (published by OUP in 1949).  The author states his case thus:

The argument here presented is that if errors are due … to cross-association, then the Japanese form of error should be one thing and the Bantu form quite another…. But that is not the case. .. The collection of ‘common errors’ … proves that the errors which exasperate teachers of English are indeed ‘common’.

French adds: “In seeking the source of error in the vernacular, the teacher is searching in the wrong field. The fact that the errors are common indicates that they have a common cause”.

This ‘common cause’, according to French (although he doesn’t use the term) is false hypothesizing, including over-  and under-generalising.  (The antidote that the author suggests, by the way, is much more typical of its time: he recommends the ‘drilling-in’ of correct forms, and the ‘drilling-out’ of errors, all of which involves “considerable trouble and constant vigilance”).

In discussing this topic on the bus from Nicosia to Kyrenia en route to the conference dinner, Nick Jaworski pointed out, that if transfer were the explanation, why is it that his Turkish students willfully produce errors like *I went Antalya, when the analogous verb + prepositional phrase exists in Turkish (even if the preposition is attached as a suffix)?  The same might be asked of the commonly attested *I working, *the boys playing etc, by speakers of languages, like Spanish, that have a matching auxiliary construction: estoy trabajando, los niños están jugando…

But is the case for interference  really dead and buried?  Isn’t it a fact that many (if not most) learner errors are – as my student suggested – directly traceable to L1 influence?  

About these ads



25 responses

10 05 2010
Alastair Grant

I’m not sure about the traceability of all errors to L1 influence/interference (in five years’ time we’ll probably be calling it “L1 indigestion”).

For example, in my context of Argentina (where the L1 is Castellano), the very typical erroneous production of verbs following modals with the ‘to infinitive’ is, I think, overgeneralization/carelessness by the teacher not specifying bare infinitive for the form. In Castellano, there is no “to” form for infinitives: no puedo hacer la tarea (I can’t do the homework) doesn’t include any kind of prepositional “prefix” (as it were) for the infinitive hacer.

Certainly we can predict, year on year, the types of challenges that our learners will meet/errors that they will make. But, as per the above example, this doesn’t seem necessarily linked to the L1, although this can provide a great rough guide (Smith and Swann’s ‘Learner English’ springs to mind), it’s not going to cover all the bases. I think taking these things on a lesson-by-lesson basis is more human, too.

Actually, I think of more importance is the influence of error on the learner’s interlanguage and how it will provide learning opportunities for them/their classmates. What I like about the idea of interlanguage is that it suggests there is a state where there is no “incorrect” language, more like opportunities for development.

Steven Pinker reckons that that there isn’t really such as thing as “ungrammatical” anyway, and that “designating a sentence as ‘ungrammatical’ simply means that native speakers tend to avoid the sentence, cringe when they hear it, and judge it as sounding odd” [The Stuff of Thought: 2007].

This seems a good starting point for me. I think it’s about helping our learners’ interlanguages evolve. In terms of intrinsic motivation, I’m pretty sure that’s what they’d want too.

10 05 2010
Scott Thornbury

“What I like about the idea of interlanguage is that it suggests there is a state where there is no “incorrect” language, more like opportunities for development.”

Nicely put, Alastair. It would be nice if we could inter (as in ‘bury’) the term ‘error’ -and its negative connotations, once and for all – and talk instead about non-standard forms (only slightly less pejorative, admittedly) or simply ‘interlanguage forms’.

10 05 2010

Hi Scott,

When I think about students’ errors/mistakes that are heavily influenced by L1, what springs to mind is vocabulary. If a learner doesn’t know a certain word in English, it seems they naturally “anglicise” the word and hope it’s correct. This is the most evident source of L1 interference I can relate to right now off the top of my head. Oh, and pronunciation tends also to be heavily influenced by L1, especially in beginners’ language.

However, I believe I agree with Nick and what the studies on interlanguage say when it comes to grammar. I guess most grammar error/mistakes come from the language that’s being learned rather than the L1 of the learner.

I’ll keep my comment short and wait to see what others have to say.



10 05 2010
Scott Thornbury

Yes, my original post didn’t distinguish between grammar, lexis or phonology errors, and maybe it should have – since the influence of the L1 is different in each case – and probably increases at each level of ‘delicacy’ – i.e. most influential at the level of phonemes, least influential at the level of discourse.

10 05 2010
Nick Jaworski

Wow, quick turn around on reading the book and then posting! Our conversation and this post have really got me thinking. There are a lot of sources for errors. Perhaps we need a more nuanced approach where the source of errors is distinguished and then a method of combating them is applied.

On my site I cataloged a large amount of very common Turkish errors http://turklishtefl.com/for-students/common-mistakes-made-by-turkish-students/ . You’ll notice at the top that I attribute these to L1 transfer error and that’s true for 99% of them. These are errors that are avoided more easily when translation is not encouraged. L1 can create a lot of resistance to L2 acquisition because we naturally fall back on L1 structures and these sound more correct to us. I remember even when learning Turkish if I didn’t have a rule to guide me or if I couldn’t recall what others were saying I would fall back on L1 structures. Even then, very often, if an L2 structure seemed really weird to me when thought of in L1 I had a much much harder time remembering it and often felt like it wasn’t correct.

This is just as obvious when Turkish students refuse to say “I’ll be there in 10 minutes.” They want to say “I’ll be there 10 minutes later.” I have had so many students fight with me again and again about it and many refuse to believe it’s true or they simply can‘t get their head around it, even when done as a translation exercise. This is where minimizing translation in the classroom is beneficial because it helps students think more in English and accept the differences between the languages. Whenever I have students that, for whatever reason, rely a lot on translation, they are much more resistant to these language differences than students who learned by talking to tourists or listening to music.

Minimizing L1 also breaks away reliance on L1-L2 dictionaries, which are just really poor quality in Turkey and return false results probably 70% of the time.

Then there are the errors that occur with all learners through developmental processes or generalizations like “I did go yesterday.“ I think with these errors we just have to be patient. Eventually they will work themselves out. It’s good to be able to recognize them though because it makes you aware that we shouldn’t expect students not to make them or correct them quickly depending on where they are in their learning.

Sometimes errors occur in discourse communities. At my last school I constantly heard students say “I go went to the store” yet I can’t remember ever hearing it at my current school. I’ve noticed other errors that are the same. There really is a transference of errors within a classroom, school, or wider discourse community. These errors simply have to be pointed out and squashed if possible so they stop spreading.

Some errors are also perpetuated by teachers. Many Turkish teachers teach “I haven’t a car” as correct. In addition to that, I’ve even had native speakers teach the same mistake as American English. They heard it so often from students that they just assumed it was a form of English they were unfamiliar with. I remember doing the same thing; I used to think the British pronounced ski like sky because I heard it mispronounced so much.

For all errors, even fossilized ones, I’ve noticed persistent encouragement to self-correct is really effective. I think drilling, too, has its place here as the correct form begins to sound familiar and comfortable, but I think the encouragement of self-correction is the most effective. I think we simply need to recognize that this will take a lot of time and also that we don’t need perfection. I think the burden of perfection is one too many students and teachers see as a goal.

10 05 2010
Scott Thornbury

Yes, I did look at your collection of Turkish-speaker errors (and what a great resource for teachers!) but I have to admit that quite a few of them could have come from Arabic- or Spanish-speakers, in my experience. E.g.

I’ll wait my friend.
I went to home last night.
I asked to him to help me.
We really enjoyed.
I went to the store for buying some food.


Now, if these mistakes are common to speakers of other languages, it might be because these languages have shared grammatical features that English doesn’t (possible), or that they derive from some kind of universal grammar, as posited by the Chomsky-ists, (unproven) or that they derive from some universal cognitive processes of simplification, overgeneralisation, redundancy, etc. My bet is on the latter.

10 05 2010
Nick Jaworski

Hmm, interesting. I know that these reflect direct translations from Turkish, but of course I can’t say if they reflect the same in other languages.

I’m not saying the list is evidence of the majority of mistakes being tranfer errors, those are just the ones I cataloged as they are going to be different from common errors across languages and more useful for students.

As an aside, I actually created the list for students, but I have to admit that, after their initial excitement at seeing it and pouring over it, they almost never refer back to it and I haven’t seen corrections result from exposure to it. Interesting.

10 05 2010
Jason Renshaw

Certainly some errors are predictable based on transference from the L1, but in my experience variability is common and often unpredictable even with groups of learners from the same L1 background with pretty much the same level of exposure to the L2, and this variability often appears to have nothing to do with negative (or even positive, for that matter) transfer.

Something I have begun to take an interest in is the predictability of errors based on teaching methods/emphases and learning materials. These errors are as – if not more so – predictable than L1 transfer in some cases.

By way of a simple example, the huge (over) emphasis on the *be* copula and present progressive structures at the beginning of language courses (and the insistence that these be mastered perfectly) tends to account for a range of errors that persist for quite some time, and usually have nothing at all to do with L1 transfer. I know this resonates with some of the things that are written about interlanguage, but I do believe teaching itself could in fact exacerbate the potential for future systematic errors.



10 05 2010
Scott Thornbury

Thanks Jason. Yes, there is a class of errors that are ascribed to the instructional process itself – e.g. a tendency to over-use some forms that have been recently taught, or to ‘borrow’ chunks of teacher talk without making the necessary modifications:

T: Ask Juan what he does for a living.
S: What he does for a living?

In my experience these effects are not very durable, though. To these errors might be added those that are picked up from fellow students, although, again, these are probably a lot fewer than the behaviourists used to warn us about.

10 05 2010
Karenne Sylvester

This is a really interesting discussion and I agree with most of the comments above and have, from experience as both learner and teacher, seen the L1 interference in vocabulary as mentioned above but then I have also heard same structural errors (and the same easily grasped grammatical concepts) made by Chinese/Ecuadorian/German students – namely in things like adding an s- on verbs in 3rd person/ he-she mixed up (isn’t that weird) /present perfect.

When I hear these, I wonder… dare I say this… about the “natural” grammar and whether or not some of our English structures fall outside brain logic -the way the brain processes communication (I don’t actually know how to express this properly so hope you will understand what I mean…) because, grammar is to some extent man-made: we defined a set of rules a couple of centuries ago and then made them popular by spreading them through a printing press but that doesn’t mean that they are easily learned, grasped, absorbed, regurgitated…


10 05 2010
Scott Thornbury

I wonder… dare I say this… about the “natural” grammar and whether or not some of our English structures fall outside brain logic -the way the brain processes communication ..

This is a valid point, Karenne, and is supported by proponents of “cognitive grammar” who argue that there are fairly universal ways of thinking about things, e.g. that animate beings can affect inanimate objects, and that grammar is mostly easily learned and applied when it maps on to these cognitive schemata, as in The boy kicked the ball. But it is (relatively) harder to produce and interpret an utterance that runs counter to this schema, as in The ball was kicked by the boy or even The ball kicked the boy! When constructions fall outside this ‘brain logic’, as you nicely put it, it may be more error-prone. Would be interesting to research.

10 05 2010
John Rogers

Great posts so far (that makes me miss the discussion boards).

Two thoughts that occurred to me when reading this entry:

Firstly, in reference to the comment made by the MA student, I wonder if negative transfer is self-evident due to just how ‘evident’ we, as teachers, make it out to be.

In my current context, where I teach monolingual classes, I find myself on the lookout for mistakes which are caused by the learners’ L1 so that I can highlight and allow the learners to notice this difference.

But when errors occur that are obviously due to overgeneralization, for example I’ve had students use “goed”, then I’m not sure if it really occurs to me as a teacher that the error is one which merits special attention, though perhaps it should.

What I’m getting at is I wonder if the ‘self-evidence’ is simply due to teachers noticing and giving more importance to errors which can be attributed to negative transfer.

Secondly, I agree with the posts above that certainly some of the errors can be attributed to the learner’s L1. I was thinking this morning about Krashen’s monitor hypothesis and was curious to what extent a learner’s L1 might function as a monitor when speaking or writing in the L2, in the absence of learned rules about the L2.

I’m not sure if I can use “monitor” in that way (maybe someone with more knowledge about L2 acquisition can clear that up if not), but is it possible that there might even be a link between when/how learners make errors due to over-/under-generalization and their L1?

10 05 2010
Scott Thornbury

I wonder if the ‘self-evidence’ is simply due to teachers noticing and giving more importance to errors which can be attributed to negative transfer.

I think this is a very valid point. It assumes, of course, that the teacher is familiar enough with the learner’s L1 to recognise the transfer when it occurs. Is it part of learning to be a teacher that we become increasingly attuned to L1 transfer errors?

10 05 2010
Jessica Mackay

Great discussion.
Rick mentions above that when Cross Linguistic Influence occurs it tends to be in the domain of vocabulary. This is borne out by empirical research (e.g Olsen 1999) CLI researchers tend to classify Lexical transfer as misspellings, borrowings, coinage and calque. The first two are the direct influence of the L1 while the second two are the L1 and L2 interacting. Ringbom also suggests that misspellings, borrowings and coinage are transfer of form while calque is transfer of meaning.

Interestingly, one of the reasons that the title CLI is being used more often nowadays rather than L1 influence and (negative!) transfer is that it also allows for the influence of languages other than the L1. Most of the world is bilingual, within Europe most speakers have knowledge of other languages other than their mother tongue. Acquisition of an L3 implies interaction of two other linguistic systems.

“ The additional language complicates the operations of the process” (Clyne, 1997)

In effect this means that a learner activates a ‘speak foreign’ function or as Grosjean (1995) suggests a ‘bilingual mode’ and errors occur which clearly come from an additional language which has a ‘supplier’ role. Most of living outside the UK in a bilingual setting or having lived in more than one country can probably identify with this.

As far as both structures and vocabulary are concerned Jasone Cenoz (2002) lists 6 factors that determine the relative weight of CLI;
Linguistic Typology – influence is more likely to occur if the languages are typologically similar (Cenoz studied the acquisition of English in Spanish – Basque bilinguals and found that CLI in L3 English invariably came from Spanish).
Proficiency in TL – the greater the proficiency, the lower the influence
Specific context – interlocutor, setting, level of formality
‘Foreign Language Effect’ of L2 status (the learner wants to avoid appearing ‘foreign’)
Age – the effects of cognitive and metalinguistic development
Recency – learners are more likely to be influenced by a language they are actively using.

Does this seem to ring true in classroom practice?

10 05 2010
Scott Thornbury

Thanks Jessica – I should have read your post first, before replying to Josh (below) and citing Swan and Smith. Your comment that “influence is more likely to occur if the languages are typologically similar” echoes their viewpoint exactly, but is backed up with refeernces to research.

And this would also suggest that – given the typological dissimilarity between English and Turkish – Nick’s Turkish errors are less likely to be transfer than developmental, no?

10 05 2010
Nick Jaworski

I think Jason’s contribution is pretty important. That would be an excellent research project to see how teaching style affects learner error. I really see the overemphasis on be as an auxiliary early on contributing a lot to learner errors later on. It also works in reverse when there is no correction at early levels, especially on things like articles and the learners end up in upper-int and never use articles correctly.

I also agree with John’s observation that we as teachers often pay more attention to transfer errors. “Nick want me to go” is an expected mistake while “Nick wants my going” comes across as quite alien and noticeable, at least to me.

10 05 2010

Quick comment re. errors like *I working, *the boys playing etc (I wish I had more time to read and digest properly, sorry)
In this particular instance, speakers of Spanish are not just trying to get lexis and grammar right (well, morphemes stuck on lexis, really), there’s a pronunciation issue. ‘I’m working’ or ‘I’m speaking or whatever’ involves – to their mind – an /m/ at the end of a word, which, apart from in conversations about the Bardem clan or ex-Argentinian presidents, is a feature that doesn’t happen in Castellano (although it does in Catalan). If you get a consonant in there at all, it’s often the ‘ng’ one (no IPA on this interface). The 2nd person ‘you’re playing/speaking’ etc involves either an intercalated ‘r’ or nothing, in which case what they say is ok, and the third person means an ‘s’. These may all imply less familiar consonantal clusters, final sounds, etc so you have the plasticity of the mouth coming into play too. Students may simply focus on ‘getting the words right’ to get their message across, before perfecting their pron.
Just a thought, in a hurry, so probably not clearly expressed. But either way, progression of errors ie the natural order they occur in, may not just be related to some internal simplification/overgeneralisation of grammar; what your mouth can comfortably cope with at any given point in the learning process may also play a role.


10 05 2010
Scott Thornbury

Thanks, Fiona, for the comment. And thanks for reminding me of the pron issue. It’s curious that so much grammatical information is invested in morphemes that — for many learners — are both difficult to perceive and reproduce. This supports the view that learners often over-use the -ing suffix — when for example they want to put a verb in the past tense: yesterday I working — because, being an entire syllable, it makes more of a grammatical ‘splash’ as it were. It’s like saying: “Here be grammar!” And it’s a very common feature of what some researchers call the ‘basic variety’.

10 05 2010

It makes a grammatical splash, but not a literal one, fortunately. Unlike some phonemes that carry grammatical information in the form of morphemes. Some students I’ve encountered here in Spain actively avoid some morphemes because they’re ‘spitty’ (eg the /t/ -ed endings). /s/ endings or the -th on ordinals can also fall prey to the affective filter in some areas, because pronouncing /s/ sounds is ‘what they do in the North’ and the ‘th’ is what they say on the Peninsula…. Regional identity issues.
Syntax, tense error, word form etc is one thing, but morphemes/phonemes is a whole world. The errors could have a whole number of causes, including perception/reproduction difficulties and emotional/affective issues…

10 05 2010
Josh Lange

Great post, Scott T but a little too general in order to refute your student. Pronunciation errors are just as often due to the learner’s L1 as to English spelling or confusing words like WORSTESHIRE.

Jason’s point about teaching methods is good, however many languages don’t have a difference between “simple” and “progressive” tenses so the individual learner does whatever he or she remembers, independent of what he/she is taught, because there is no rule or internalized speech pattern in L1 to compare to, and his/her goal is communication over accuracy.

The book LEARNER ENGLISH by MICHAEL SWAN is an excellent reference tool for understanding groups that share a common language, as well as being able to speculate on the cause of an error, transference or otherwise.

Josh Lange
Dresden, Germany

10 05 2010
Scott Thornbury

Thanks for that comment Josh. It prompted me to pull down Michael Swan off the shelf (actually Learner English is edited by Michael Swan AND Bernard Smith) and re-read what he had to say about L1 influence. The following observation is relevant to the discussion, but no sources are cited, so I’m intrigued to know if this is based on any research:

Since transfer mistakes arise where the systems of two languages are similar but not identical, they are most common (at least as far grammar and vocabulary are concerned) in the interlanguage of students who speak languages closely related to English. Speakers of unrelated languages such as Chinese or Arabic have fewer problems with transfer, and correspondingly more which arise from the intrinsic difficulty of the English structures themselves. (p. xi)

11 05 2010

Looking at this from the teaching standpoint, if you are in a monolingual course and have experience of the learners’ language, then picking up on L1 interference is likely to offer some benefit to the majority of the class. Having said that, identifying the mistake’s cause does nothing to signal whether this is a disturbing or non-disturbing error.

The challenge for the teacher, particularly those who have a good command of the learner’s language, is to be clear on which errors are globally important. A German listening to another German giving a presentation in pretty bad English will probably be able to extract a fair bit of information because the L1 errors of the presenter are understandable. The Italian giving the next presentation in better English will have a far harder time being understood by the German because the Italian’s L1 errors are “foreign” to the German.

The problem is exacerbated in schools where the the teacher and learner share the same language – mistakes that are “non-disturbing” are sometimes only understandable by someone with a common L1. Non-native speaking teachers often run this danger in classes.

Personally, I find that L1 interference is a bigger issue at higher levels. At lower levels the errors come from too many sources – from poor learning through simple mistakes to ignorance – for any one source to claim primacy.

It’s easy for the teacher to jump all over the L1 errors, and that may have a benefit for the majority, but for me the acid test is to ask the question, “Would a (insert suitable nationality) of a similar standard of learning understand this clearly?” L1 errors may well be part of the solution, but they distract from defining the problem.

12 05 2010
Scott Thornbury

Thanks, Olaf, for the comment. You’re right – the source of an error is of less importance than its effect on intelligibility. I suppose, though, that many teachers feel, rightly or wrongly, that L1-influenced errors are (a) more easily dealt with, since it’s simply a matter of flagging the lack of match between L1 and target, and (b) prejudicial to L2 identity-construction, since interference errors – like a strong accent – are indicators of a failure to “cross over”. This may account for the fact that a learner who persistently says “I have twenty years” (borrowing directly from Spanish, French, etc) is more likely to irritate his/her teacher than a learner who says “I are twenty”.

29 06 2010
antalya haber

Thank you for this great and interesting article!

22 03 2012

Interesting discussion Scott and the rest even though I am two years late. Here in the UAE there seem to be a lot of fossiled mistakes with Emirati learners. One of the classics being ” I am from in the UAE” . Some of my learners range in their L1 proficiency (Arabic). Some of the learners speak Urdu to their mothers, some form of English with the filipino housemaids. I found Learner English by Swan and Smith a useful reference as well as Pit Corder, but will check out JC Richards argument on interlanguage transfer. I like the term managing mistakes rather than error correction.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s


Get every new post delivered to your Inbox.

Join 1,927 other followers