Mythology and hypothesis in English Language Teaching

Ignorance more frequently begets confidence than does knowledge

Darwin, 1871:3

Few users of this site believe in flying unicorns (presumably).
It is slightly surprising, however, just how many teachers of English believe in only slightly less credible ideas when it comes to teaching and the English language.
This guide is an attempt to see us back on the straight and narrow and to stop believing in things for which there is no evidence or for which the evidence is, in fact, almost wholly negative.
To be entirely fair, calling all these views myths is not always fully accurate because some are simply assertions and nobody knows if they are true or not. Others are real myths which have been proven to be untrue and to which some people still adhere often with confidence arising from ignorance.
Some of the ideas that are discussed here are hypotheses based on intuition alone so they aren't myths in the strictest sense of the word, but they slip between the status of hypothesis to assertions of fact rather too easily.
Some, indeed, have become received wisdom on teacher training courses and many otherwise quite helpful websites but they should be treated with a certain amount of caution.

How do we decide what's a myth and what's believable?

Well, we start by looking at the evidence.
It is an ancient aphorism that we should not be sure that we have the right answer until we are sure we have asked the right question.
Here, we need what Carl Sagan called a baloney detection kit (others have used a less respectable description of what is meant by baloney). Sagan's rules for the construction of a baloney detection kit are quite elaborate (but well worth reading). Here, we'll just focus on a few of them.
There are some rules to follow when anyone tells you about a new approach to teaching English or just tells you the right way to do it (with the implication that all the others are wrong).

Claims are too good to be true
If an approach, a methodology or whatever makes exaggerated claims for its efficacy and effectiveness, suspect it.
If something is too good to be true, it probably isn't true.
An example is some of the outrageous claims made for neurolinguistic programming.
Vague or inadequately hedged assertions
As soon as you hear the words generally, usually, in most circumstances and so on, be prepared to think when the assertion does not apply.
Equally, if you read or hear words like obviously, unquestionably, always and find that the message is couched in the present simple with no hedging or expression of argument rather than simple assertion, pass on by.
Dressed-up science-like terminology
If something is proposed in terms which appear scientific, check them out. Learning style theories are a good example, replete as they are with terms like activist, kinaesthetic and so on.
Appeal to intuition and common sense
If an assertion or approach is based on what seems or just ought to be true rather than evidence that it is true, it's probably wrong.
It can't hurt
Many ideas and approaches are couched in terms which imply that they can do no harm and may well do good. That is not much of a recommendation.
Falsifiability
If an idea cannot be falsified, it isn't worth considering.
Appeal to ignorance
The other side of this coin is the claim that anything which is not proven to be false, must be true. We cannot prove that there isn't a chocolate teapot orbiting the sun at a distance of six light years but that does not mean there is one.
The exclusion of the middle
If an idea or admonition is couched in terms which exclude a middle way, suspect it. A good example is the inductive vs. deductive debate which seems to imply that there is no middle way and, indeed, that one is better than the other with no consideration of how the processes work.

In what follows, we'll identify some of the most prevalent myths in two areas and see what's in them.

Myths about the language

A number of language myths circulate freely in English language classrooms and include:

English is a bastardised mix of Latin and French

Around 29% of the words in the Modern English Lexicon can be traced to Latin. Of the rest, another 29% or so are of Norman or Parisian French provenance but that leaves 26% derived from Germanic languages, 6% directly from Greek and another 10% of various origins.
The grammar of the language is, however, overwhelmingly Germanic in origin despite having lost nearly all the case and gender inflexions which existed in older forms of the language and owes almost nothing at all to the influences of Latin or French grammar.

English does not have a case grammar

While it is true that English, unlike some of the other members of the Germanic language group, does not distinguish, nominative, accusative, genitive and dative cases by alterations to endings on nouns, adjectives and determiners, case considerations are very important in the pronoun system and also affect how, for example, relative clauses are formed.
Case considerations are also very important in the analysis of transitivity (obviously).
The genitive system in English is complex and dative shifting for the indirect object is frequent and frequently misunderstood.

More people speak English than any other language

English is third on the list well behind Mandarin Chinese (which is spoken by almost three times as many people) and Spanish.
The difference is that English is by far the most frequently used language as a second language.
It may also not be spoken by as many people as Spanish and Mandarin but it is spoken much more widely than either of those languages.
English is an official language in over 50 countries but not necessarily the first language of most of the inhabitants.

English grammar is difficult

Most people assume that the grammar of their own language is more difficult than that of other languages.
The truth is that English is extraordinarily simple in many ways:

Verb forms for regular verbs require the learning of very few rules of spelling and pronunciation because there are so few inflexions. There are no irregular -ing forms at all.
Irregular verb forms require the learning of only one or two forms for all persons.
Nouns are only ever inflected for number and plurals, with a few exceptions, are almost universally simple and predictable.
No changes are made to determiners for case and few for number and countability.
No adjective or adverb forms change in any way for number, case or gender.

Nevertheless, there are obvious difficulties, not least the aspectual and tense structure of English which is complicated.

Making matters much worse in this regard is a regrettable tendency to make up rules where no rule exists. Three examples will suffice:

The past perfect is used when one event precedes another. That would make:
She locked herself out and climbed in through the bathroom window
wrong. There is no doubt that
She had locked herself out and climbed in through the bathroom window
is correctly formed but it's no better.
When we set two events in the future, we use the future perfect form to show the ordering. That would make:
I'll finish this by the time the boss gets back from lunch
wrong. Again, there's no doubt that
I'll have finished this by the time the boss gets back from lunch
is correctly formed but it's no better.
have to vs. must is fertile ground for the invention of non-rules:
1. The modal auxiliary verb have to applies to external obligation placed on a person and must is used for internal senses of obligation and duty. It follows that:
  I have to write to my mother
  is an obligation placed on me by another and
  I must write to my mother
  is a sense of duty I am imposing on myself to commit to an action.
  Not true. Most people use the verbs interchangeably (they are in free variation).
2. The verb must refers to a specific obligation and have to refers to more general obligations so, it is averred:
  I must tell her the truth
  is correct and so is:
  We have to tell the truth at all times
  but the alternative formulations:
  I have to tell her the truth
  and
  We must tell the truth at all times
  are somehow wrong.
  That is nonsense, of course.
3. We use must to refer to duties we impose on others and we reserve have to to imply that the obligation comes from elsewhere (a similar but slightly subtler idea than the external-internal distinction). So, it would follow that we should prefer:
  You must be careful (because I require it)
  to
  You have to be careful (because other authorities require it).
  Whether this distinction exists is doubtful and whether it is worth troubling most learners with it is even more doubtful. It is unlikely that most native speakers would wince if the modal auxiliary verbs were used in reverse.
  (For a little more, track down the guide to central modal auxiliary verbs.)

English has more words than any other language

It is notoriously difficult to arrive at an accurate figure for the number of words in any language. However, the latest version of the Oxford English Dictionary suggests around 170,000 currently used words.
The equivalent figures for other languages are often in excess of this with, e.g., more words in Italian, Slovak, Tibetan, Persian, Chinese, Romanian, Urdu and a host of other languages. Korean, by some estimates has over 1,000,000 words and Portuguese over 800,000.
A lot depends how you count.

There are lots of differences between American and British English

There are obvious pronunciation differences and some spelling disparities but, in fact, the number of different words which are used in the two varieties is very small. Mostly the issue is one of frequency of one form over another in the two varieties. The number of words affected by differences in the spelling systems is probably fewer than 400 (i.e., 0.25% of the language).
The two varieties are, for all intents and purposes, 100% mutually comprehensible.

Myths about teaching the language

There are even more myths about teaching.
Here's a selection.

Communicative Language Teaching is the best way to teach

Well, it rather depends on what the learners need, doesn't it?
There's plenty of evidence to show that people who have been taught the language using a communicative methodology are better at communicating than those who have been taught through a form-based methodology.
On the other hand, of course, there's just as much evidence to show that people who have been taught through a form-based methodology produce more accurate and acceptable language than those who have not.
What there is not is any evidence to show that communicative methodologies are more effective overall.
Many people who have learned the language through very form-based traditional approaches have nevertheless attained an impressive mastery of the language forms and are perfectly capable of using the language accurately and fluently to communicate their ideas.
It has been suggested that Communicative Language Teaching is popular because:

It absolves many teachers of the need to learn grammar.
It feels right that people will learn the language in their efforts to communicate in it although how one would acquire the case structures of German or Russian or the 15 or so cases in Finnish that way is obscure. Other structures in English, such as the passive and relative clauses are unlikely to be acquired at all in a purely communicative context because they are more rarely used and have complex grammar.
It fits well with Anglophone cultures (from which many teachers come) which prize risk taking and faulty but effective communication above accuracy and potential loss of public face.
It allows teachers to justify an unnecessary and wasteful amount of time being spent on skills work.
It absolves people of the duty to construct a logical syllabus progression and reinforces the idea that language structure can be taught in unconnected piecemeal gobbets as and when the need arises.

The corollary to this argument is that we should teach function not structure because that's the communicative, and, therefore best, way to teach.
The problem with the teaching of functions is that it quickly degenerates into a phrase-book approach to teaching anything. It needn't, if it's well handled, but it does.
Then we get to the stage at which a learner may be told that, for example:
    I'm sorry
    Please forgive me
    I regret that
    Excuse me
and so on are all ways to perform the function of apologising (and they are, of course, although they vary in intention and style).
Unfortunately, however, learning those four simple phrases and clauses tells the learner nothing which is of any further use and it is not possible for any learner to generate, for example:
    I apologise
from any of them or even from a combination of them because there is no overt pattern to emulate and manipulate.
On the other hand, teaching that:
    Forgive me, please
and
    Excuse me, please
are transitive verbs in the imperative and the structure is simply the base form of the verb followed by a direct object (hence me not I) which is softened by the use of please, does allow any learner to generate:
    Help me, please
    Tell me, please
    Show me, please
and thousands of other perfectly correct imperative-form requests which can perform multiple functions. The learner then has to supply the correct structure (not too hard) and know the meaning of the lexis. That's all.
With a small nudge, learners may also be able to generate:
    Tell Mr Jones, please
    Take Mary to school, please
    Open the box carefully, please
    Write me a letter
and we are still proceeding from the simple pattern of the clause.
It is not beyond the wit of most learners to see that the meanings of the verbs and the settings in which one might use these clauses will determine the function which is performed.
The reason is simple: functional language abjures a focus on pattern but structural considerations insist on it.

People have different learning styles and we should cater for them

The whole idea of attributing learning styles to individuals has been soundly debunked by every serious investigator. There is no truth in the theories and they are harmful and potentially discriminatory practices which detract from the core of teaching and waste both teachers' and learners' time.
This includes VARK, NLP, Multiple Intelligence Theory and a host more (over 70 at the last count and probably rising).

Teacher talking time is A Bad Thing and should be kept to a minimum

The idea that TTT is A Bad Thing pervades training courses at all levels.
The issue, if there is one, is that poor, uncontrolled, unfocused and interminable teacher talk is, naturally, a bad thing but well focused, clear, targeted and comprehensible teacher talk is a precious resource for learners, especially those in a non-English-speaking environment or those at lower levels of competence who are simply overwhelmed by authentic language.
In many cases, teacher talk is the only even quasi-authentic communication in English that some learners can easily access.
To dismiss it all as damaging is both harmful and wasteful of a very valuable resource.

There is a natural order of acquisition of language items

The central problems with natural order hypotheses are:

They apply to Spanish speakers
Although there are some studies concerned with speakers of other languages learning English, half the most often cited studies were focused on Spanish speakers learning English.
The orders that were found even with that narrow focus also varied.
There are lots of them
Of six documented studies we have a variation in findings. In some, for example, it was found that the article system was the first item to be securely acquired, in others the third, in two others the fourth and in one it was the eleventh item to be acquired.
One may be excused for thinking that learners of languages which do not have an article system are unlikely to acquire one very easily compared to, e.g., Spanish speakers who have a system akin to English. We should not be surprised to discover that the study which placed articles in eleventh place was conducted with Japanese speakers.
The focus of such studies is usually on easily identifiable morphemes in the language: the articles system, with no distinction between definite and indefinite, incidentally, the -ing form of the progressive aspect, copula be, regular past tense forms, plurals, contractions, possessive markers and so on.
No information in the studies was available or looked for concerning other structures such as conditional forms, aspects of tenses, passive clauses, relative pronoun clauses, causative structures and so on.
It is much less likely that these sorts of structures will show the same standard progression and the outcomes are much more likely to be based on issues concerning the learners' first-language structures.

Structural synonymy exists

This refers to the assumption inherent in much grammar practice material that, for example:
    Mary was arrested by the police = The police arrested Mary
    If you come to the party you will meet my sister = You will meet my sister if you come to the party
and so on (add in the synonym forms of your own).
Sometimes, indeed, learners are required to convert one form to the other.
What this assumption ignores in the fact that speakers of no language light on the grammar at random. We put things first or second in a phrase or clause or choose a particular structure for a reason. Usually, that reason has to do with markedness: what the speaker / writer wants to emphasise, so it is not true that:
    On Thursday Mary arrived
is the same as
    Mary arrived on Thursday.
because the sentences are very differently marked.
Using synonymy of any kind to explicate meaning is a procedure fraught with perils.

New language should be drilled for pronunciation

There is no evidence at all that drilling has any benefits. Learners may like it and, even if they don't, they may expect it, but its benefits have not been shown in any research into learning.
It is a classroom ritual rather than a useful procedure.
Drilling is based on the theory that learning a foreign language is one of making new habits and automatising the language. This hypothesis sits very uneasily within more cognitive and social approaches to language learning and teaching.
It can be shown empirically, however, that asking learners to think about where they are placing the stress on words or how they are forming sounds in the target language is effective but that is not a question of habit formation, it is asking learners to apply cognitive processes.

Pair- and group-work is beneficial

This is another classroom ritual. The usual theoretical underpinning of the practice of getting learners to work with other learners is twofold:

It is assumed, taking a social constructiveness theory of language acquisition, that learning is primarily a social activity. Proponents of such theories of learning have failed so far to explain the mechanisms of learning in a persuasive manner. It remains in the status of a hypothesis. Furthermore, it seems to exclude the possibility of learning anything by working entirely alone and that is not a hypothesis which bears much scrutiny.
It is assumed that learners will get more speaking practice if they work in groups or pairs without the direct intervention of a teacher and that, thus, teacher talk will be diminished and learner talk increased.
What is often not mentioned is that learner talk to which other learners are exposed is rarely a good model so learning from one's peers may not be a useful way of acquiring accurate and appropriate language. Indeed, it is unlikely that most people's first choice of a teacher would be someone at the same level of mastery of the language as themselves.

While in theory it may be motivating and interesting to work on a task with others, in practice what happens is that one student will often dominate the exchange and complete the task while others do very little or the task will not be adequately completed by anyone.
Some learners may actually need a little peaceful space to think and manipulate the language alone and pair- and group-work precludes this possibility.

We should work to lower learners' affective filters

While Krashen's affective filter hypothesis is a popular and pervasive influence on teachers' behaviours, there is, as yet, no empirical evidence to suggest that it even exists. It, too, remains a hypothesis, not a fact.
It may be the case that a relaxed and confident learner who feels valued and cherished in the classroom is better able to learn but this has never been demonstrated experimentally.
Indeed, some learners may well benefit from having a little stress to sharpen their wits and increase their motivation to do well.

Translating dictionaries and translation itself are Bad Things

Much teacher advice and, alas, time in the classroom has been spent encouraging learners to use an English-only dictionary but quite why this should be seen as more efficacious and accurate than using a translating dictionary remains obscure.
Modern bilingual dictionaries are extremely accurate and helpful in translating both words and phrases across languages and very few people would feel comfortable approaching the task of operating in a foreign language without one. Even teachers who have been convinced on an initial training course that English-only dictionaries are somehow better than the other sort are often seen clutching translating dictionaries when they first get a job working in a foreign country whose language they know only imperfectly, if at all.
The huge success of published translating dictionaries and phrase books should alert us to the fact that they are actually seen by millions as very useful things. That is, of course, not evidence that they actually are useful.

Nowadays, of course, we are also in the age of machine translations and some of the software is very impressive indeed, being able to translate the spoken word and also to note issues of style and appropriacy. None of this technology is going away any time soon and it will probably continue to get better and better, so teachers have to learn to use it to enhance learning rather than prohibit its use based on an assumption that cannot be supported by the evidence.
Many learners will also, when using their knowledge of English, be asked to translate for others and that, of course, is a skill worth acquiring for those people. Not to focus on it would seem slightly perverse.
Thousands of hours of classroom time may be being wasted by people struggling to explain concepts in English only when a 1-minute excursion into translation would clear a matter up very adequately.

Language should be heard before it is spoken, before it is read before it is written

This is a hangover from the reform movement of the 19th century and is no more true now than it was then.
It all depends, of course, on what sort of language, used for what purposes and by whom.
There is no obvious flaw in the idea that it is easier and probably more useful to hear something spoken as a model before being asked to say it and also that reading the language in a text may serve as a model for the learners' written production.
That this is always and inevitably the case is much less arguable.
Some forms of language are, in fact, only ever encountered in a written form and other items may be confined to the spoken language (or something written in a spoken style). It is unlikely, for example, that any learner would hear or have to say:
We look forward to reading your response
but they may well encounter it in a written text and may, indeed, be asked to write it. On the other hand, it is unlikely that one would read or have to write something like:
Pass the salt, please
because that kind of request is confined to the spoken language.
All this means that the medium in which language is presented and practised depends on the kind of language it is, not on some assumed correct ordering.

Language skills work is as important as language systems work

This is a myth greatly encouraged by well-known teaching qualifications in the profession but one which is hard to maintain in the face of argument.
The assumption behind it starts from the unproven assertion that learners are incapable of transferring social and written skills from language one to language two.
It is, naturally, demonstrable that the way texts are organised and the way oral interactions and transactions are conducted does vary across languages. That is primarily a cultural issue, of course, not a linguistic one per se because the culture determines the appropriate language to use, and the appropriate way to structure texts, not the other way around.
Some time spent, therefore, on the analysis and construction of conventional written texts and focusing on social appropriacy and oral competence is arguably very useful.
However, the assumption that such a focus is as important as a focus on increasing learners ability to use the structures and lexis of the target language is much more difficult to sustain.
The same arguments apply to receptive skills with an often unspoken assumption that learners are incapable of transferring skills such as listening or reading for specifics or gist. Most learners are, in fact, perfectly capable of scanning or skimming, monitoring listening or listening intensively for information in their own languages and there is little reason to suppose that the skills cannot be deployed in a foreign-language setting.
We may need to alert learners to what they should be doing (because operating in a foreign language is cognitively very demanding and people forget) but the presumption that we should be teaching these skills to adults is pushing at the limits of credibility.
It seems unarguable that without a sound grasp of the systems of the language, learners will be unable either to speak appropriately or write accurately and conventionally. Knowledge of systems, then, has to precede knowledge of speaking, listening, reading and writing.

Furthermore, it is often quite difficult to separate skills from systems work in a principled way. For example, a focus on how textual cohesion is maintained in writing or on how turn-taking and turn-passing opportunities in spoken interactions are signalled is a systematic endeavour allied to skills development. They cannot be sensibly separated.

The focus on grammatical and pronunciation accuracy is outmoded

This is a myth much favoured by those who cannot be bothered to learn about the systems of the language they are teaching because, they aver, it is unnecessary to do so. Accurate use will flow from the effort to communicate, they say, and learners will, by a process of induction, come to form internal rules which will allow them to speak and write with acceptable accuracy and in conventional form.
This may well be true for some of the simpler aspects of the systems of English, such as, for example, the formation of plurals or the pronoun system (both of which are transparently simple) but it is much more difficult to maintain this position when it comes to more complex language such as the formation of passive-voice clauses, causative expressions, relative pronoun clauses, transferred negation, verbal aspects or the use of modal auxiliary verbs.
The amount of time and effort that would be needed for a learner to work out the rules for any of those structures is simply too great to contemplate. The teacher's job is, therefore, to know the rules and systems better than her students and be able to transmit that information accessibly to her learners. To do that, she has to know how the language works.
There are times when a focus on teaching the systems of the language is vital. Swan, 2002, put it this way:

Knowing how to build and use certain structures makes it possible to communicate common types of meaning successfully. Without these structures, it is difficult to make comprehensible sentences. We must, therefore, try to identify these structures and teach them well.
...
In some social contexts, serious deviance from native-speaker norms can hinder integration and excite prejudice – a person who speaks ‘badly’ may not be taken seriously, or may be considered uneducated or stupid. Students may, therefore, want or need a higher level of grammatical correctness than is required for mere comprehensibility.

In sum, we need grammar to communicate our meanings accurately (to which we could also add the ability to make the sounds of English acceptably and spell it accurately) and we need decent control of grammar (and pronunciation and orthography) to maintain some social credibility.

Guided discovery and/or inductive learning is to be preferred

Over what?
Yes, it is, given the data, possible to figure out what the rule is for the use of any language (regarding structural, phonological accuracy or communicative appropriacy) but the key lies in the quiet given the data in this sentence.
A simple example is the plural and past-tense systems in English. Overwhelmingly, we make a plural by adding -s if the word ends in an e and adding -es if it ends any other way. There are about a dozen exceptions.
We pronounce it as /s/ if it ends in /t/, /p/, /f/, /k/ or /θ/ and a /z/ in all other cases.
The system is so simple that given enough examples of the correct forms, most learners can figure out the rules for themselves (providing they aren't allowed to become confused with dwarves, wives, wolves and leaves under the eaves).
Figuring out the rules for nouns which end in f is much less easy because most of them are quite rare and the data are not readily available. It is much easier to be told that the majority of words ending in f form the plural with ves and it's pronounced as /vz/. In other words, even in this simple case, it is arguably easier to supply the rule and let the learners deduce the forms.
The past-tense forms of regular verbs follow equally simple rules for their formation and pronunciation, as you are aware, and they can also be figured out by most people without recourse to a grammar book, simply by looking at a few dozen examples such as
    offend-offended
    point-pointed
    depart-departed
    smoke-smoked
    hope-hoped
    provoke-provoked
and so on.
Sooner rather than later, the rule will be figured out and retained.
However, the grouping of the forms of irregular verbs in English by their phonemic characteristics is a much more difficult rule to figure out from the few examples that are likely to cross a learner's path in a month or so. So, it is much easier to be told that, for example, /iː/ often changes to /e/ and /ɪ/ to /ʌ/ than to try to work that out from single instances of verbs such as bleed, feed, cling, spin and swing. There are, in fact only 10 or so verbs in each class and encountering them in a single lesson is very unlikely (helpful though it might be for the learners' inductive reasoning processes).

Even if it is possible to figure out all these rules by a process of inductive reasoning (aided by a bit of judicious guided discovery), there are clearly other rules that are just too complex for that to be done successfully in a way that doesn't take a dozen lessons.
For example, the rules for forming a passive-voice clause in English are decidedly complex and, even with monotransitive verbs it requires a six-step routine:

Identify the subject and the object
Raise the object to the subject position
Remove the erstwhile subject
Select the correct tense form of the verb be and insert it after the first noun phrase
Discover the past participle form of the verb and insert it after the verb be
Decide if the clause needs an agent (another multi-step process requiring consideration of communicative and semantic issues) and then insert it after a by-phrase

Doing all that for a sentence such as:

The committee which meets under the chairmanship of the Vice-Chancellor has been deciding the issue of whether to have a new wing added on the building

would be nearly impossible if all one had to go on were a few forms such as:
The window was broken by the stone
Mary was surprised by the ghost
and so on.
Even with multiple exposures to the correct formulation for a passive clause, it is unlikely that any learner would be able to arrive at the six-step process without considerable help. The help, naturally, would be to supply some rules so that the learners could deduce the correct forms rather than induce the rules.
Added to all this, there are many complications and constraints on passive-clause use in English (some quite rare events) and it is vanishingly unlikely that anyone would be able to figure out what they are without exposure to enormous and unwieldy amount of data. We have, therefore, to resort to the old-fashioned and deeply unpopular deductive reasoning processes by supplying some rules and guidelines for when the structural choice is appropriate and then ensuring that the learners get some real communicative practice in applying them.

Finally, of course, there is the issue of the middle way. Proponents of inductive approaches often forget to mention that even when learners have figured out a rule from the data supplied by the teacher or the materials, a process said to enhance the chances of the rules staying in the long-term memory and eventually becoming automated, they then have to apply the rule deductively. There seems little point in putting learners to the trouble of figuring out the rules for themselves if they cannot then deduce what is accurate and well-formed or appropriate to say in the circumstances and that involves deductive reasoning, of course.

Any more myths?

This little guide could have been greatly extended and may well be added to as new myths arise or come to prominence. They do all the time.
In the meantime, try the articles index (link on the left) for more on some of these myths.

If you have come across a myth that you think should be debunked, let ELT Concourse know and this page may be updated.

References:
Darwin, C, 1871, The Descent of Man and selection in relation to sex, London, John Murray
Sagan, C, 2000, The Demon-haunted Universe, Le Verne, Tennessee: Ingram International Inc.
Swan M, 2002, Seven bad reasons for teaching grammar - and two good reasons for teaching some, in Methodology in Language Teaching, ed. Richards and Renandya, Cambridge: Cambridge University Press, pp.148–152