The origin and progress of linguistic norms

« previous post | next post »

Last Monday's post "Progress and its enemies" resulted in a vigorous exchange of views in the comments section. Reading over the comments, it seems to me that people were to some extent talking past one another. Such misunderstanding seems especially common in discussions of linguistic norms. So in a few paragraphs below, I've tried to explain how I, at least, think about these issues.

We start with discussions of linguistic style, of linguistic form, of pronunciation, of word usage, and so on. No one objects to this (although the topic has become become oddly rare among literary scholars, suggesting some covert objections). Certainly,  linguists spend most of their time engaged in such discussions. And inevitably, these discussions are relative to the norms of some speech community or some written tradition.

Then we have explicit comparisons and evaluations of different norms. Again, no one objects to this, in general, unless the discussion involves falsehood (such as the notion that regional or social norms are anarchic masses of chaotic mistakes caused by failure to learn the rules of a standard language) or prejudice (such as unsupported claims that particular regional or social norms are objectively ugly or inefficient).

There are apparently a few people out there who object to the whole notion of a standard written language, or believe that it's wrong to urge people to learn one;  but none of these people post on this site, as far as I know. (Some of our commenters may feel this way, I'm not sure.) At the other end of the spectrum, there are apparently some people who believe that all linguistic varieties that deviate from (someone's idea of) a  standard should be ruthlessly suppressed.  But most purveyors of usage advice, of whatever species, simply believe that the formal written language should be learned and used in certain contexts; and the disagreements are mostly about what those contexts are, and what the norms of that language should be taken to be.

There are at least four different kinds of arguments that can be used in discussions of what linguistic norms are or should be: empirical arguments, logical arguments, arguments from authority, and experimental arguments. Nearly everyone involved in these discussions invokes arguments of all four types at one time or another.

1. The empirical investigation of the norms of usage, by reference to the way that people actually talk or write, has been the foundation of linguistic scholarship for centuries. The advent of computational methods in recent years has made this method easier and more powerful. It can be applied with equal ease and appropriateness to determining the norms of formal written English in 18th-century America, or the norms of the spoken French in contemporary working-class Montreal, or the norms of contemporary scientific English in refereed publications in computer science or epidemiology.  The instinct of most linguists, given a question about usage, is to use this method to see what the facts are.

This sometimes annoys certain people,  especially if the results disagree with their own instinct, or the pronouncements of some authority figure. So even if the cited evidence comes from the canonical texts of literature, politics, science and so on, you sometimes see silly arguments about the irrelevance of polls to truth,  or the license of great writers to break the rules prescribed for lesser beings.

2. Arguments from logic depend on developing a theory of what the patterns of (some variety of) language are, and applying this theory to contested cases.  If the theory is wrong, then the argument is nonsense.   If the theory is apparently correct, then further discussion ensues.  Is this an area where logical consistency should be expected to apply, or do we expect free variation? Are the deviations mistakes, or idiomatic exceptions, or instances of a different pattern entirely? Some of the arguments that we ridicule as "prescriptivist poppycock" are of this form, for example here and here; on the other hand, I've often made arguments of this form myself, for example here.

3. Arguments from authority are the easiest to construct. They're valid if the authority's statements are relevant and trustworthy; if not, not.  We often make arguments based on reference to dictionaries, because we trust their scholarship — but this trust is provisional, not absolute, and we sometimes point out omissions, or apparent errors of other sorts, even in the best dictionaries.  Grammars and usage manuals are often a good source of evidence of types (1) and (2) above, and linguists often use them in that way, and cite them when we do. On the other hand, some grammars are mistaken or out-of-date, and some usage manuals are full of made-up or over-interpreted or absurdly vague "rules", contradicted by valid evidence of other kinds.  Most instances of what we call "prescriptivist poppycock" are of this kind — a couple of examples are here and here.

4. The fourth category of argument, involving a claim about the psychology of perception or production, most often appears in the form of an empty assertion, for example that something is perniciously ambiguous. These assertions are almost never based on, or checked by, actual experiments, though they could (and in my opinion should) be.

There's a particular variant of the argument from authority — maybe it's better to call it a meta-argument from authority — which starts with an observation that everyone agrees with, and then draws a conclusion that's much more controversial. There's an especially clear instance of this on pp. 57-58 of Mark Halpern's book Language and Human Nature:

By rules I don't mean laws, such as those that linguists find in the history of sound changes. A grammatical rule is an edict intended to regulate writers (and to a lesser extent speakers) or English, not to discover what they have done or are doing; its aim is prescriptive, not descriptive. I'll use laws when discussing the regularities found by linguists, rules for the judgments laid down by grammarians. Some protest against rules, calling them arbitrary; I can accept that description, but only with the understanding that arbitrary does not in such a context mean mindless, capricious, or oppressive, it simply means a convention, something that does not claim to have been dicovered in nature. It is instead a judgment or decision taken by some authority, declaring some form of behavior to be acceptable or not. it is almost always the result of much debate and balancing of equally desired but mutually exclusive goods — if it is arbitrary, it is so in a sense closely related to arbitration: a way of resolving disputes on complex matters in such a way as to preserve civilization, which cannot survive with them.

The classic example from outside the linguistic world of a rule is the injunction to keep to the right when driving. We know that it is only a convention, but it is none the worse for that — it does sort us out on the road, and virtually eliminates the head-on collisions that would result from allowing drivers to decide for themselves, as free-born citizens, which side of the road they would keep to. If a critic were to protest this rule on the grounds that it was — in the bad sense — arbitrary, and that no objective grounds could be found for preferring the right to the left here, he would find little support; it is accepted universally that while it matters not at all whether we all keep to the right or the left, it matters a great deal that we all keep to the same side — and the rule accomplishes that, while not, so far as we can see, infringing on anyone's rights.

The part of this that everyone agrees with is that linguistic norms are arbitrary, or at least have a large arbitrary component — for example, there's no reason to think that the English Subject-Verb-Object order is better or worse than the Japanese Subject-Object-Verb order,  but it's obviously useful for everyone to pick a word-order pattern and stick to it.

The part that's controversial, to say the least, is the idea that establishing and maintaining such norms motivates or perhaps even requires the linguistic equivalent of legislatures, police, and courts. In fairness to Mark Halpern, he explicitly denies that he means for state power to be applied to these questions — he just wants lexicographers, editors, teachers, and others to try to enforce (what he feels are) the proper standards.

Among different groups at different places and times, we can see linguistic norms developing as various mixtures of emergent or natural order, explicit analysis and discussion, and formally constituted authority. Some people seem to think that no worthwhile linguistic system is possible without the systematic intervention of explicit authority — but it's hard to square this view with examples like the development of Elizabethan English. Other people seem to think that no attempts to impose explicit authority over linguistic norms are ever at all effective — but this is inconsistent with various documented successes of language planning, not to speak of the many examples of extreme diglossia created by the long-term maintenance of admired antique norms.

I think that the people who post here generally have a sensible understanding of the issues, and a reasonable perspective that avoids the absurdities of various extreme views. But then, you'd expect me to think that, wouldn't you?

(I'm happy to say that after I wrote this but before I posted it, I got a long response from Mark Halpern to last week's discussion, which I'll put up as a guest post shortly.)


  1. Sarang said,

    February 22, 2009 @ 3:35 pm

    "On the other hand, some grammars are mistaken or out-of-date, and some usage manuals are full of made-up or over-interpreted or absurdly vague "rules", contradicted by valid evidence of other kinds."

    If you think of usage rules as a way for the educated to distinguish themselves from the illiterate (or the private-school educated from the public-school educated, etc.), surely there's some utility to having them be arbitrary? The more the in-group's rules deviate from the emergent patterns of the language, the harder it is to pick them up by observation, and the more liable the uninitiated are to slip into natural but "lower-class" (etc.) phrasing, and give themselves away. To the extent that this is true, it seems adaptive for a social class with cultural prestige to develop arbitrary usage rules — especially when it isn't the class with the money — in order to maintain its exclusive character.

    Of course, I don't know how far this sort of _political_ argument excuses bad linguistics.

    [(myl) Well, there's plenty of perfectly accurate linguistic description involved in cataloguing the naturally-evolved complexities of dialect differences, including the distinguishing features of prestige dialects. The politics is a separate matter from the linguistics.]

  2. Brian Johnson said,

    February 22, 2009 @ 9:41 pm

    Most, if not all, the discussion on this blog regarding linguistic norms has, not unreasonably, concerned their status in the English language. I'm curious as to what sources one might turn to in order to find this topic discussed in relation to other languages.

  3. Sridhar Ramesh said,

    February 23, 2009 @ 4:53 am

    Just out of curiosity, since the post mentioned them towards the end: what sort of things are you referring to when you mention various documented successes of language planning? Can you give some examples?

  4. Stephen Jones said,

    February 23, 2009 @ 7:38 am

    Catalan is a good example, Ramesh. Between 1975 and now all State education has changed from Spanish to Catalan in Catalonia. This obviously required a large amount of top down standardization that had been carried on since the beginning of the 20th century.

  5. Stephen Jones said,

    February 23, 2009 @ 7:51 am

    Shifting the goalposts is Halpern's speciality. See my post on his thread on the way he tries to pretend 'usage' is what we should use, not what we do use.

    I think the main problem with prescriptivists such as Halpern is that they are terminally confused. They don't see the difference between things such as spelling, where arbitrary prescriptivism works just fine and is the only way, punctuation which is still largely arbitrary and only rarely susceptible to a descriptivist analysis, substandard and dialectical registers, and their own often wacky preferences. They just talk about 'misuse' of the language.

  6. Picky said,

    February 23, 2009 @ 10:48 am

    Mark Liberman: "No one objects to" discussion of linguistic style (and from the comment about literary scholars I take it this might be science-aided but need not be science-based; ie that it might be based on literary aesthetic judgments).

    Do you think "a discussion of what (written language) norms are or should be" could also acceptably include a discussion based on literary aesthetic judgments – ie informed but subjective judgments of what is attractive and fitting – and would this fall outside your four types? (I'm just trying to look around the edges of your definitions.)

  7. marie-lucie said,

    February 23, 2009 @ 11:47 am

    A survey of literary styles across centuries (eg, as I mentioned somewhere, the difference between novels in English in the 18th, 19th and 20th centuries) shows that "literary aesthetic judgments" of "what is attractive and fitting" have varied widely during that period. Not only that, but some of this variation correlates with variations in taste in other fields such as clothing or music, and those can also be correlated with social mores and changes. A language does not exist by itself but is affected by the lives of its speakers. This is why trying to preserve its older forms at all costs regardless of circumstances is doomed to failure. The only unchanging languages are those no longer spoken, like Sumerian or Classical Latin.

  8. Picky said,

    February 23, 2009 @ 12:46 pm

    marie-lucie: I think at first glance I agree with all of that. I also think ML's post is full of meat, and I, for one, am grateful for the chance to learn from it. My question was posed in the hope that an answer would help me refine my understanding of his position, that's all.

  9. marie-lucie said,

    February 23, 2009 @ 12:59 pm

    Sure, Picky, I know that you were addressing ML, not me.

    Determining "what is attractive and fitting" is not usually considered part of a linguist's purview. Linguists study many languages, not just their own, so they can't give recommendations, only state facts: "In language X such and such is considered (un)acceptable, (in)elegant, etc". It often happens that forms or features which are praised in some languages are frowned upon in others, or that the same things are valued differently in the same language in different periods, as my English examples suggest.

  10. S Hawkins said,

    February 23, 2009 @ 2:31 pm

    The metaphor of road regulations is an awfully slippery one for him to use. I think the point is that, yes, these regulations are arbitrary, but that they are very important to have. Further, the implication is that they are fairly immutable. Once we've all agreed to drive on the right (or left) we can't very well change that. But of course, road regulations do change. Not only are there different regulations in different areas, but they are greatly affected by, surprise, surprise, usage. When the US government lowered the maximum driving speed to 55 MPH, there was a great deal of public resistance and the limit was routinely flouted. Years later, the limit was lifted. In other words, even in his chosen example, road regulations, there is far more change and conflict between usage and regulation than he seems to realize

  11. Stephen Jones said,

    February 23, 2009 @ 2:40 pm

    there is far more change and conflict between usage and regulation than he seems to realize

    Go to India. Only country I know where you need to read the regulation to see which side you should be driving on. because there's no way you could ever find it out from empirical observation.

  12. Picky said,

    February 23, 2009 @ 4:01 pm

    S.Hawkins: And it opens him to the charge that he wants to impose his views by law or similar, and therefore forces him to deny it … leaving at any rate the impression of a high-handed authoritarianism of the kind he says he opposes. Yep, not the best-chosen analogy for him.

  13. marie-lucie said,

    February 23, 2009 @ 5:13 pm

    Road regulations such as which side of the road to drive on are a poor metaphor for the rules of language, because these regulations are entirely man-made as well as very simple. Language is an innate faculty of human beings, actualized in multifarious ways by various groups of people, normally with little "guidance" except the necessity of communicating with others and therefore of sharing common ground. The most fundamental "laws" or "rules" of any language are ones that do not need to be "imposed" on speakers because they have alreay internalized them in the first few years of their lives, long before they were able to reflect on their language use.

  14. bianca steele said,

    February 23, 2009 @ 7:40 pm

    Okay, in fact, I have no issue with the "side of the road" analogy (other than that it seems arbitrary, but then most illustrative examples are). The basic issue is that everybody has to drive on the same road at the same time, they are all doing the same thing (they are all driving, each in his or her own car), and therefore they all follow the same rules about where to drive, when, and in what direction. In the absence of the rules, something would have to give.

    In the end, however, the example breaks down. Why? Has one trivial detail been overlooked? Maybe a different example would have been better.

    Politics is involved in the decision how to set up the rules of the road. It would be politics in the sense that all government involves politics somehow or other, or in the sense that we call the science of government "political" science. (I'm ignoring the possibility of a pun on "politics" and "policy," because no native English speaker could take that pun seriously, unless he or she were a little weird.) We hardly decide the rules of the road based on some kind of binding arbitration — not even by a jury trial :). Where are the judges supposed to be, for one thing?

  15. marie-lucie said,

    February 23, 2009 @ 8:35 pm

    If politics is involved in setting up the rules of the road, then we might expect those rules to change with a change in politics. People will take to the streets in support of their political opinions and aims, but hardly in support of changes in the rules of the road. Neither do they build political campaigns for or against some rules of grammar.

    Ability to learn the rules of the road is not hardwired in the human brain and manifested within the first two years of life, as ability to acquire language is. Nor are the rules of the road anywhere near the complexity of any human language. I find the comparison totally unhelpful.

  16. S. Hawkins said,

    February 23, 2009 @ 10:13 pm

    But people do politically engage with language use, both in pushing for, and resisting, changes. The elimination of "thou," for example, was deeply connected to the Quakers' refusal to use "you" when addressing a single person (in a manner parallel to the French usage of "vous"). To vastly oversimplify, to avoid being associated with some of the radical ideologies of the Quakers, people avoided using "thou" and it was gradually abandoned. Similarly, the debate over whether or not "he" refers only to men or not is not simply an issue of grammar, but also of politics.

    On a different note, yes, we are hardwired to acquire language, but that does not got terribly far in explaining why any given language takes the form it does and not some other. To understand a language as it is used, one needs to understand the social and historical forces that helped shape it.

  17. marie-lucie said,

    February 24, 2009 @ 12:07 am

    S. Hawkins,

    I absolutely agree with you that "to understand a language as it is used, one needs to understand the social and historical forces that helped shape it," but how those forces operate on a language are rarely obvious, and conscious attempts to reshape some of its details are rarely successful.

    What you call "political" I would call "cultural" in a broader sense. For instance, the question of "generic he" caused some small groups to consciously try to create a new, gender-neutral singular pronoun, but those efforts have not met with any response from the public and have not been either pushed or denounced by any political organizations, or become a subject of heated public debate, unlike other societal concerns. In fact such a pronoun already exists, in the form of "singular they" which has been used for centuries (see several older posts on Language Log) in spite of being condemned by prescriptivists and therefore kept out of most modern written sources.

    You say that antagonism towards the Quakers caused the demise of "thou", but that pronoun has survived to the present day (often as just "thee") in a variety of English rural dialects, including in parts of Newfoundland. It is more likely that its abandonment was associated with rurality than just with the Quakers, since "you" was already becoming the norm at the time. Conversely, during the French Revolution everyone was supposed to be addressed as "tu" (same as "thou"), and it was dangerous not to follow this rule in public, but once the revolution was over people reverted to using "tu" and "vous" as they had done before, and in subsequent political upheavals the imposed "tu" was never revived: universal "tu" just went too much against the grain, no matter what the political opinions were.

    Creation of new vocabulary (eg "chairperson") is one part of language where societal changes and concerns can be directly reflected, but again such creation is cultural rather than political. Political parties do not make creation and acceptance of specific words part of their platforms.

  18. Mark Liberman said,

    February 24, 2009 @ 12:19 am

    S. Hawkins: The elimination of "thou," for example, was deeply connected to the Quakers' refusal to use "you" when addressing a single person (in a manner parallel to the French usage of "vous"). To vastly oversimplify, to avoid being associated with some of the radical ideologies of the Quakers, people avoided using "thou" and it was gradually abandoned.

    According to evidence summarized in a Linguist List post by Alan Firth in 1996, this is false. He quotes Larry Trask:

    I find the idea that the loss of English `thou' was due to the Quakers to be inconceivable — though I note that Dick Leith takes this notion seriously in his book A Social History of English.

    All the sources I have seen, including Leith, who gives a very good account, agree on the main conclusions. English-speakers began to use 'you' as a respectful singular in the 13th century, probably under French influence. Except in conditions of intimacy, 'you' quickly became established as the ordinary way for an upper-class speaker to address an equal, as well as a superior, and by the 16th century 'thou' was all but non-existent in upper-class speech, except in addressing obvious inferiors. Naturally, this usage began to be copied by the middle class, and by the 16th century 'thou' was likewise rare in middle-class speech, except in addressing obvious inferiors.

    On this account, the Quakers resurrected, for ideological reasons, a pattern that had already been basically dead in their speech community for a century.

  19. marie-lucie said,

    February 24, 2009 @ 1:16 am

    Thank you, Mark, I was pretty sure that was the case but did not have the dates at my fingertips.

  20. Stephen Jones said,

    February 24, 2009 @ 5:53 am

    Another thing the prescriptivists fail to realize is the sheer volume of grammatically and/or semantically incorrect utterances there are, and how utterly irrelevant their rules are for avoiding them.

    They get round this by only considering utterances that are in fact grammatically or semantically correct and labelling those incorrect. They deal with the almost infinite number of stable doors that need shutting by only doing it if the horse has already bolted.

  21. thomas g said,

    February 24, 2009 @ 9:08 am

    Comparing language "rules" to road regulations makes complete sense, under Halpern's definition of what "rules" are— they are the complement of what a linguist would call grammar. "Rules" for him are those that are made up by people and intended to shape their behavior.

    Granted, road regulations are simpler than even prescriptivist rules, which better resemble fashion designers' conventions of dress. For instance, there is no reason why pink should be the new black this spring, except that some fashion authority proclaimed it to be so.

  22. marie-lucie said,

    February 24, 2009 @ 10:30 am

    But fashion authorities come up with new "rules" every year, because constant change is the essence of fashion, while language regulators or "activists" try to impose unvarying standards (in spite of what Halpern says), some of them 200 or more years old and never adopted in general speech. In this respect, fashion is comparable to teenage slang, which becomes obsolete with amazing speed.

  23. bianca steele said,

    February 24, 2009 @ 3:18 pm

    thomas g,
    "'Rules' for him are those that are made up by people and intended to shape their behavior."

    Exactly – institutions institute, teach and enforce rules. In other words, they are socially constructed. “Social constructionists about X tend to hold that: (1) X need not have existed, or need not be at all as it is. X, or X as it is at present, is not determined by the nature of things; it is not inevitable. Very often they go further, and urge that: (2) X is quite bad as it is. (3) We would be much better off if X were done away with, or at least radically transformed. . . . but they need not do so.” (Ian Hacking, The Social Construction of What?, 6-7). Yet, in my experience, writers who position themselves as Mark Halpern has done will reject any talk of “social construction” as a form of relativism or postmodernism. “Odd” really is the appropriate word here.

  24. dr pepper said,

    February 25, 2009 @ 3:00 am

    Question: Why is God "thou"? He's not our social inferior, a child, an animal, or an apostrophed object. So my guess is it's because he's supposed to be everybody's friend. Am i right?

    And is God called "tu" in French or Spanish? Is he called "du" in German?

  25. marie-lucie said,

    February 25, 2009 @ 8:50 am

    And is God called "tu" in French?

    That depends. Traditionally God was called "Vous" in the Catholic Church, but Protestantism introduced "Tu". With the reforms in the Catholic Church, God was increasingly addressed as "Tu", but I guess most of the older faithful must still be using "Vous".

  26. Aaron Davies said,

    February 25, 2009 @ 11:02 am

    yes, speech to god is intimate. come to think of it, i wonder if that was the english usage pre-reformation? the whole concept sounds rather protestant.

  27. Aaron Davies said,

    February 25, 2009 @ 11:12 am


    there's no reason to think that the English Subject-Verb-Object order is better or worse than the Japanese Subject-Object-Verb order, but it's obviously useful for everyone to pick a word-order pattern and stick to it.

    <nitpick>well, in (relatively) analytic languages, anyway. even without considering things like latin, japanese has a rather less fixed order than english.</nitpick>

    more seriously, and @Brian Johnson as well:

    Other people seem to think that no attempts to impose explicit authority over linguistic norms are ever at all effective

    how widespread is this attitude outside the english-speaking world? as an american, i was always vaguely aware of the french academy, but was astonished to discover just how many languages (spanish, norwegian, mandarin, japanese) are practically conlangs, especially compared to the almost perfect anarchy that is english.

    @Sridhar Ramesh: israeli hebrew?

  28. marie-lucie said,

    February 25, 2009 @ 11:36 am


    Are you fluent in the languages you mention? your impression that English is more anarchic could be due to the fact you are aware of all the different registers and at least some of the regional or social varieties of English, while in other languages you might only be familiar with a neutral register. In French, for instance, "Standard French", a conscious, mostly written register is not what you will hear in most conversations, no matter where in the world you hear it.

    Explicit authority over linguistic norms: this seems to be effective in cases where language planning is needed to standardize a wide variety of dialects, or to increase the vocabulary of a language which until then was only used in traditional local contexts and therefore lacked words dealing with modern developments. Where there is already a long-established written tradition, as in the major countries of Europe, norms may be established by educational authorities (eg minor spelling reforms in France). I wrote earlier about this (and the rather anachronistic role of the Académie Française), but I don't remember when that was. Apart from spelling though, teachers of the language are the ones that set themselves up as authorities. Whether the training they draw on was suitable is another matter, in particular whether they are aware of the difference between spoken and written styles.

  29. David Marjanović said,

    March 3, 2009 @ 6:40 am

    Question: Why is God "thou"? He's not our social inferior, a child, an animal, or an apostrophed object. So my guess is it's because he's supposed to be everybody's friend. Am i right?


    Though, of course, the King James Bible simply translated the pronouns literally, and there were no separate polite forms in Biblical Hebrew or Koine Greek…

    And is God […] called "du" in German?

    Yes — regardless of confession.

  30. David Marjanović said,

    March 3, 2009 @ 6:41 am

    Denomination, I mean.

RSS feed for comments on this post