11. Some Mysteries of Language Learning

this entry is also posted on two of my Wordpress blogs: Thor's Unwise Ideas and Thor's Language Teaching Notes


Thor May
2005 - 2012


An expert is a fool a thousand miles from home. Having successfully failed to learn about nine languages, I’m a veteran language learning imbecile, always a thousand miles from success, and an eternally hopeful beginner. I’ve also had the cheek to teach my native language to hopeful novices for over thirty years, which sometimes leads them and others to mistake me for a wannabe guru. The sheer hypocrisy of this dilemma should condemn me to embarrassed silence forever, yet I persist probing the reasons and remedies for my own language learning incompetence. After all, my exasperated search is surely shared by millions of others. The discussion which follows is informal, but makes serious points. It builds on an original e-mail exchange with a correspondent in 2005.


[illustration courtesy of Dr Phap Dam who unlike me made a successful transition into the world of another language.]


Hi Thor,

I would like to talk about your analogy of playing chess/football in connection to learning a foreign language;[ed: an allusion to my May 1995 paper. See references].

First let me tell you that I used to live among Indonesian overstayers in Wellington. One had managed to survive for many years (maybe 10 years) working at McDonald’s kitchen undetected by the immigration. So it is the equivalence of “having played chess/football” for many years in terms of foreign language learning. And yes, undeniably, he could make himself understood by speaking English, not Indonesian, in his working environment. 

However, it was totally different than I had worked in an IT gaming software auditing company and then in a technical call centre. Had he had the same technical skills as I had, it would still be very hard to cope with the tasks involved using that kind of English level.

I remember on one occasion, he left a note in the kitchen of our flat meaning: don’t be messy because there is a cat. But the written English was so funny, in the sense that it is riddled with any kind of mistakes, that the other overstayer wrote a joke comment on it.

When I discussed with other Indonesians who had a good mastery of English, the conclusion was that they simply did not learn. Esp, at lower strata of life, you don’t need to master English to survive. So the other way round, the fact that you can survive is not a proof that you have mastered a foreign language simply by having been in a language speaking environment.

One more point is that the analogy is not all true. If you play chess or football, what is wrong is absolutely wrong. But this does not apply in speaking a foreign language. For example: you can say “I have do” or “I have doing” and still make yourself understood as “I have done” and in most situation people won’t be bothered to correct you. So learning through making mistakes in this sense, even if it is true, is a very slow and very long process.

And the overstayer example also shows that there is a difference between a child learns his/her mother tongue and a non-native speaker adult learns a foreign language. It would take too long for an adult to learn a foreign language using child-learn-to-talk approach.

Cheers, Ming
[blog http://gradspot.blogspot.com]

Hello Ming,

The game analogy for language learning is only partly relevant, as you noticed, especially for chess (where you normally do learn the basic rules before trying to play). Touch football, with its learn-as-you-go pattern and loose interpretation of rules it probably closer.

You are also right that most games impose a penalty for rule-breaking, while in language the message is the important thing, and the message can often be inferred even when rules are “broken”. In fact there is something important going on behind that fact. Hardly any language teachers or text book writers seem to realize what it is.

The human brain works like a parallel information processor, not a linear processor as in your personal digital computer. Linear computing is very economical, it yields mathematically certain outcomes, but it is extremely fragile. A single rule violation will freeze the processing, which is why computer programs crash constantly. (There is an apocryphal story that a NASA Venus probe was lost in the early days because a single comma was missing in the computer program).

Parallel processing needs much more computing power than linear computation, and its outcomes are always probabilistic, not certain. One the other hand, it is extremely robust : SOME kind of useful answer or result can usually be inferred, although the probability of it being the right one varies. Human communication is like this. We are never quite certain that we understand exactly what the speaker means, but we are right often enough to get by.

That describes your Indonesian friend, doesn’t it. He could survive socially with noodle speech, and even hold down a simple job. He struck a problem with writing because the cultural rules attaching to formal patterns in writing are more demanding than the standards of informal speech. The stricter standards exist because writing is nearly always separated from the original context of the message, so there is nothing else such as body language or a social situation to help with decoding it. Therefore it is not surprising that writing is not “acquired” automatically by young children (let alone adults).

All human beings have to be TAUGHT writing AFTER they have acquired spoken competency in their mother tongue. In fact, only a minority of native speakers ever become fluent writers of their language. Beyond simple things like shopping lists and SMS electronic text messages, writing is a stilted and painful activity for the majority of people everywhere. They are a bit more at ease with reading, but the reading age of most populations decline after about 14 years of age (i.e. after puberty very large numbers of learners lose interest in this kind of learning). In Australia’s supposedly advanced society, 47% of people do not have sufficient literacy to read a medical prescription or a train timetable according to a recent census. The figures are roughly comparable in other OECD countries. It is true that functional literacy is a fuzzy and contested notion (see Larsen 2002, and Wikipedia 2012). Still, there is no doubt that vast numbers of native speakers struggle with literacy in a way that never occurs with speech.

Even faced with bad writing however, a well educated language user can generally extract useful information (an ability that illiterates may lack). In spite of the Indonesian’s crazy note Ming, you probably guessed the drift of his written message correctly. We humans are geniuses at “finding” meaning, sometimes even when there is none to be found. (Hands up if you have never “seen” faces or animals while you were looking at clouds in the sky ^_^ ).

The kind of explanations I have been offering here about the complicated world of language learning are not what you will normally hear from university teachers of Applied Linguistics. Why not? Well, for a moment let’s take a turn away from real people learning real languages, and look a little at the theories about mental language creation cooked up by academics, and taught in schools and universities ...

Firstly, at the moment the majority of text books and published models by linguists about how languages work assume a linear processing model for the human brain. Every era has its metaphors, usually coming from the wonders of that age. At present we are reveling in the enormous changes wrought by digital computing, which happens to depend mostly upon linear computation. Of course, we quickly draw on this as an analogy to explain our own thought processes. Added to this tendency is the powerful effect of peer pressure in professional groups. For academics, “published” is a key term, the survival word for a career. As with religions and ideologies, in academia it is usually a health risk to challenge orthodoxy. It is relevant to keep these pressures in mind when evaluating any contemporary research conclusions.

The “standard model” of language cognition for half a century now has come from Noam Chomsky’s essentially linear (though recursive) generative grammar models, which require a “deep structure” in the mind called UG, meaning (for him) universal grammatical rules common to all languages, and genetically programmed only for human language.

The UG model claims that children learning their first language don’t need huge speech input from parents or others because they already know the rules, and only have to switch the right ones on for their L1. However, after a certain age (“critical period”) this wonderful convenience of automatic language learning is no longer available, so it becomes hard to learn L2, L3 etc.

Well, we adults can certainly agree that foreign language learning is a tough challenge. However, the critical period hypothesis (which is controversial itself) is not necessarily linked to UG. I was personally skeptical about the UG story even as a student, and I have formally rejected it since the 1980s. This is the reason that I walked away without completing my first PhD on generative grammar in a university where it was dominant (a stupid career move).

Of course, researchers working through the prism of UG have still learned many useful things. You can get quite useful information even out of fairy tales, although you may have to do some heavy interpreting and editing later. For what it is worth (I come cheap nowadays ^_^ ) my unfashionable skepticism has been getting more support lately from research which challenges UG. For example, see Frank, Bod, & Christiansen (2012) in the reference list below, as well as Dabrowska (2012) and Griffiths (2011).

Anyway Ming, you want information that is actually useful to teachers and students. From the viewpoint of classroom language teaching, a heavy insistence on ‘rules’ (which may be fairy tales anyway) might be counterproductive for many learners. It is true that some students want more explanation than others. The logical types often ask for some kind of explanation. They are nearly always a minority, and not necessarily the best L2 learners (I know: I’m one of them). As an explanatory tool, I have found as a teacher that grammatical explanation works best as an “aha!” experience, clarifying language patterns which students already have some control of (see May 2007 for further discussion).

For classroom-type language learning, my general experience has been that an emphasis on fluency is more important than accuracy up to about intermediate stage (see May 2009). People need to get airborne, get some confidence in using L2. After intermediate level, depending on a student’s professional needs, the teacher might offer more guidance on formal explicit rules. Even at this later stage however, what a student needs most is curiosity and the learned skill to guess about rules (really patterns) themselves, before checking the opinions of text book or dictionary writers.

Which learning method actually works best for language learners? One man’s meat is another man’s poison. Learning efficiency depends a lot on the learner, and people vary greatly. Some people are allergic to classroom study (a huge problem in mass education systems generally) while others lack the personal organization for private study. Usually, too, both personal and classroom methods which might be just right for a beginner will have to be adapted or changed as the learner progresses to intermediate and advanced levels. This is hardly surprising. The learner’s brain is changing. Moreover he or she is developing new skills in how to learn. Therefore it is really disappointing that neither schools nor textbook writers usually adapt in any systematic way to the changing methods needed by their progressing students.

Reflecting on my own early-stage learning experience, probably the best series of non-classroom courses I have ever come across are the Pimsleur audio (only) language courses, originating in the 1960s. They are now published for many languages. Go to http://www.pimsleurdirect.com/languages/sample/ to try half an hour in a language which you don’t know for free. The publishing company retaining Pimsleur’s copyright, Simon & Schuster, have used the pioneering work of the late Dr John Pimsleur to produce sometimes brain-dead, committee manufactured course content, but the clever design of Pimsleur’s approach has been strong enough to survive even that. This teaching program works for me anyway, and I’m a dumb language learner. Sure, I am made aware of mistakes instantly, but it is my awareness, not someone telling me in a punitive way. I just become aware that I need to run the tape again.

A great advantage of courses designed purely for audio like Pimsleur is that you can learn on your feet, walking or running, which keeps you alert. Against that, a problem for listening-learning + running is that nobody has ever made an MP3 player with big, tactile control buttons and a programmable memory which could optimize adaptive learning.

Another critical factor that both language learners and language teachers generally handle in a very amateur manner is human MEMORY. Without language memory you not only can’t learn or use a language, you have lost your essential humanity. “Memorization” has become a code word in the West for boredom and abuse because the process of its management is so often handled foolishly.

This is a big topic. Acquiring the first memory traces for a new language, and cross-linking them is a subtle process. Reinforcing those first delicate mental associations needs skill and consistency, unless you are one of the blessed with wonderful memory chemistry. I could discuss theories about memory with a great flourish of academic references referring to fragmented experiments (usually by psychologists) without actually adding much to the daily understanding of teachers and language students. However, the practical management of memory processes is worth thinking about. There is some research which can be applied by anyone. Much of it has to do with graduated (but interesting!) repetition. Maybe the best known name associated with this concept is Sebastian Leitner. For a discussion of Leitner’s ideas, and a commercial extension of them, see the Supermemo web page at http://www.supermemo.com/. More recent developments of Leitner’s theory go by the general name of Spaced Learning Systems, which have become common in electronic flashcard programs.

Actually I find from personal experience that spaced learning needs to be combined with other tools for maximum advantage. MULTIPLE CHOICE questions might be one such tool. It is rather remarkable how little attention has been paid to multiple choice as a learning mechanism, as opposed to a testing tool. Where the MC learning option has been investigated, the evidence seems very supportive (Roediger et. Al. 2010; Kleeman 2010).

For testing, MC is often a lazy option, machine programmable, which explains its popularity. (Effectiveness in testing comes a poor last when human time and effort by testers is at stake... ).

However, as a relatively ungifted language LEARNER, I find that multiple choice, especially with immediate feedback, is one of the best ways to lay down the first faint memory traces of a foreign language - traces which can be built upon more robustly later in other ways. At the moment I am using an Android app’ for HSK Chinese (a bit like English TOEFL or IELTS) to do exactly that. It seems that the acts of guessing, if necessary, and choosing, maintain attention in a way that, say, merely flipping through flashcards cannot. This process works best when the multiple choice is combined with a Leitner-type spaced learning algorithm to revise the choices which were made incorrectly.


And now for the Far Side ….

Speaking of language learning mysteries, none may be so impenetrable as the babble of an AUTOMATIC TRANSLATION MACHINE. The following masterpiece claims to be an assignment from a Korean student of English who declined to admit that Google Translate had met his writing on the way to my desk. Actually it is a coded message from intergalactic white mice planning a terran invasion, but temporarily disguised as computer bots.

Hello . I am sorry to be late. I will speak about thousand sulfuric acids that I am famous today.

Cloth sulfuric acids are acid in Kyongsang-namdo Milyangsi makeup cotton acid inside and the Woolsan city upper facing north border. Height is 1, 189m.

While cloth sulfuric acids are high-pitched the north and the west, but east four sides one generation is slow assistant inspector, accomplish evenness side that is called lion easiness in height 800m neighborhood, and is connected to southeast four sides of Jaeyaksan (1, 108m).

Danyangcheon passes for the east of the mountain, and Sijeoncheon that is upper stream of makeup cloth southward passes and soaks lion review one generation.

Sannaecheon of the north of the mountain makes Gokjeopyeongya and is accomplishing agriculture rent.

It are clause of coming to help cancer Seosangam etc. and Chingchingpokpo “Herculean strength waterfall” tourist attraction of Hongryongpokpo and so on on the west of the mountain much cultural assets such as mission dialogue’s relics including Pyochungsa that is.

If is north edge Sannae of the mountain, because ice begins to freeze since pumpkin cattle, mid-June beside dolomite that hold traditional Korean rite to pray for rain in other people fame and wealth, because there is ice marrow (national monument No. 224) that melt in the fall, tourists’ kick does not run out.

Cloth sulfuric acids are connected southwesterly in the spontaneity mountain, and is linked by Jaeyaksan because acid is connected again southward.

Because cloth sulfuric acids are situated in Yongnam districts Alpine middle,

Yeolnamalpeuseu’s huge mountain range and eulalia field etc.. of lion review of view in normalcy enter and offer a grand spectacle at a look.

North four sides of cloth sulfuric acids is achieved to steep stony cliff zone and the cloth yellows, Jaeyaksan’s the mountain range west physical aspect of a mountain does Geupjun because accomplishing Cheongildanae everywhere or the west is forming steep slope again while slant is slow.

Also, west valley that live dander is deep and everywhere waterfall of being forming magnificent view taking thousand sulfuric acid one generations that had called is Herculean strength of three southern provinces from the example speak.

The cloth sulfuric acid wests and the north are steep slope, the east is forming gentle slope, specially huge rockwalls of normalcy neighborhood Cheongilbyeorang make that and there is ice marrow that ice freezes in the north summer, consist of steep stony cliff rockwall.

Yongnam districts Alps’ noted product is outspread wide eulalia field over ridge line during 8-9 minute.

This middle Jaeyaksan lion plain is counted to place that eulalia field is outspread most solemnly.

To 1 Million-pyeong as many as to lion plain Ireuneungo, if is the fall white personal appearance because eulalia grows widely toplofty climbers welcome .

Cloth sulfuric acids are called is Korea Peninsula’s Yeongsan or Herculean strength of three southern provinces to the Yongnam districts Alps center mountain being Milyangsi’s guardian mountain.

Next, I will send the other news. Be healthy.




Chomsky, Noam (1995) The Minimalist Program. Cambridge, Mass.: The MIT Press.

Frank S.L., R. Bod, M. H. Christiansen (2012) How hierarchical is language use? Proceedings of the Royal Society B: Biological Sciences, 2012; http://rspb.royalsocietypublishing.org/content/279/1747/4522.full.pdf+html

Griffiths, Thomas L. (2011) Rethinking language: How probabilities shape the words we use. Proceedings of the National Academy of Sciences of the United States: PNAS March 8, 2011 vol. 108 no. 10 3825-3826. Online @ http://www.pnas.org/content/108/10/3825.full

Kleeman, John (2010) Multiple choice quizzes help learning, especially with feedback. Question Mark (blog) @ http://blog.questionmark.com/multiple-choice-tests-help-learning-especially-with-feedback/comment-page-1#comment-63107

Larsen, Merrin (2002) Literacy Levels in Australia – Can Ozzies Reed and Rite?. Kingsley Educational Pty Ltd. Online @ http://www.kepl.com.au/literacylevels.html

May, Thor (1987) Verbs of Result in the Complements of Raising Constructions. [ed: an example of the writer’s earlier work within, but beginning to find limitations with the generative linguistics tradition]. Australian Journal Of Linguistics, Vol.7, No.1, June 1987: pp.25-42 Online @ http://www.academia.edu/1552308/Verbs_of_Result_in_the_Complements_of_Raising_Constructions

May, Thor (1994) Postsupposition And Pastiche Talk : Mediating Order And Chaos In Language. Working Papers In Linguistics, Vol. 14, 1994: 22pp. University of Melbourne. Online @ http://www.academia.edu/1550958/Postsuppositon_and_Pastiche_Talk

May, Thor (1995) Observations on the AMES Certificate in Spoken and Written English. [ed: An informal discussion paper for teachers intended to challenge an incoming ESL “competencies” curriculum. It cost the writer a job!]. @ http://thormay.net/lxesl/teach3.html

May, Thor (2007) When Grammar Doesn’t Help. Independent research, online @ http://www.academia.edu/1562624/When_Grammar_Doesnt_Help

May, Thor (2009) Fluency Vs Accuracy OR Fluency AND Accuracy for Language Learners? Independent research, online @ http://www.academia.edu/1544290/Fluency_Vs_Accuracy_OR_Fluency_AND_Accuracy_for_Language_Learners

Phap Dam (2010) Hindsight of an English language learner. Institute of Vietnamese Studies, Texas Womens University. Online @ http://www.viethoc.com/Ti-Liu/bien-khao/bai-giang/hindsightofanenglishlanguagelearner

Roediger, Henry L. & Andrew C. Butler (2010) The critical role of retrieval practice in long-term retention. Dept. Psychology, Duke University. Trends in Cognitive Science (in press). Online at http://duke.edu/~ab259/pubs/Roediger&Butler%282010%29.pdf

Wozniak, Piotr A. (2012) The Supermemo learning program, and memory research discussed. Online @ http://www.supermemo.com/

Wikipedia References


Note: Although I have included a number of Wikipedia references, Wikipedia itself, being crowd-sourced, is always open to severe challenge. However it is unsurpassed as a quick starting point to grasp essential concepts and pursue more thorough references. I have assumed that readers will use the Wikipedia references here in that spirit. The same principle applies to general blog discussion references.

Wikipedia (2012) The Pimsleur Method. @ http://en.wikipedia.org/wiki/Pimsleur_method

Wikipedia (2012) The Leitner System. @ http://en.wikipedia.org/wiki/Leitner_system

Wikipedia (2012) Spacing Effect (in recollection). @ http://en.wikipedia.org/wiki/Spacing_effect

Wikipedia (2012) Distributed Learning. @ http://en.wikipedia.org/wiki/Distributed_learning

Wikipedia (2012) Spaced Learning @ http://en.wikipedia.org/wiki/Spaced_learning


Wikipedia (2012) Functional Illiteracy @ http://en.wikipedia.org/wiki/Functional_illiteracy


Wikipedia (2012) Connectionism. @ http://en.wikipedia.org/wiki/Connectionism

Wikipedia (2012) Cognitive Model & Dynamical Systems. @ http://en.wikipedia.org/wiki/Cognitive_Model#Dynamical_Systems

Wikipedia (2012) Distributional Hypothesis. @ http://en.wikipedia.org/wiki/Distributional_hypothesis

Wikipedia (2012) Hierarchical Temporal Memory. @ http://en.wikipedia.org/wiki/Hierarchical_temporal_memory


Wikipedia (2012) Universal Grammar [UG]. @ http://en.wikipedia.org/wiki/Universal_grammar

Wikipedia (2012) Generative Grammars. @ http://en.wikipedia.org/wiki/Generative_grammar

Wikipedia (2012) Minimalist Program. @ http://en.wikipedia.org/wiki/Minimalist_program



Bio: Thor May's PhD dissertation, Language Tangle, dealt with language teaching productivity. Thor has been teaching English to non-native speakers, training teachers and lecturing linguistics, since 1976. This work has taken him to seven countries in Oceania and East Asia, mostly with tertiary students, but with a couple of detours to teach secondary students and young children. He has trained teachers in Australia, Fiji and South Korea. Many of his papers, essays and stories may be seen on his website at http://thormay.net ; e-mail thormay@yahoo.com .

All opinions expressed in Thor's Unwise Ideas and The Passionate Skeptic are entirely those of the author, who has no aim to influence, proselytize or persuade others to a point of view. He is pleased if his writing generates reflection in readers, either for or against the sentiment of the argument. Personal names have been changed when their use might embarrass the owners.

copyrighted © Thorold (Thor) May 2012,   all rights reserved

return to index
e-mail Thor May