[Warning : this is long.]
An email exchange with Kevin Marks a few weeks ago got me thinking more about one of the theories of linguistics that I’ve always taken for granted as a given. Only now as I am about to begin graduate level work in the subject am I realizing the degree to which various researchers in the field disagree about it. Of course, as is undoubtedly the case in most academic fields, there is disagreement about pretty much everything.
The following is probably of little interest to those not interested in linguistics (although may be of some small interest to those curious about the Korean language), and may best be skipped entirely. I am, however, keen to hear what people think, if they are interested in this field at all, so rather than keep my response restricted to email, I’ve decided to post it here. I suspect that it doesn’t even answer the question that Kevin put to me, which was ‘I’d like to hear a cogent argument for (the validity of linguistic relativism),’ if I understood it correctly. More of a wee survey for my own interest. Ah, well.
The Sapir-Whorf Hypothesis, which is variously referred to as the ‘Whorfian Hypothesis,’ ‘linguistic relativism,’ and ‘linguistic determinism’ (a description of the strong formulation meant by implication to be a bad thing, I think) concerns the relationship between language and thought, and suggests in its strongest form that the structure of a language determines the way in which speakers of that language perceive and understand the external world. This formulation is generally understood by many to be untenable, but the hypothesis also exists in a weaker form : that language structure and content does not determine a view of the world, but that it shapes thought to some degree, and is therefore a powerful impetus in influencing speakers of a given language to adopt a certain world-view.
A possible opposite claim, from a sociolinguistic viewpoint, is that the thought (and thus culture) of a linguistic group is mirrored in the structure and content of their language, that because they behave and understand things in a certain way, their language reflects those behaviours and understandings – the idea that language is molded, if not determined, by culture.
Two quotes from the linguists whose names are most closely associated with this idea, the first from Edward Sapir (Language, 1929b, p. 207) :

Benjamin Lee Whorf, who was a student of Sapir, went further than the ‘predisposition’ suggested by his teacher, and proposed that the relationship was a more deterministic one :

Whorf does not go so far as to say that language structure totally determines the world-view of a speaker here. He does add, though :

This last is where the argument runs off the rails for me, at least the argument in which I have any interest. It is also the portion of the idea upon which most critics focus, and which was fueled by the Great Eskimo Snow Silliness set off in great part by this :

and which has been discussed at length in many places, including, cogently here, for example.
To most people, particularly those with little knowledge of Hardcore Linguistics, including myself, the weaker form of Sapir-Whorf seems self-evident. Of course the words we use, the words we know, have some influence on the way we think! The very fabric of our cognition is language, it might well be claimed (but of course that would be a claim that would meet great opposition as well). There is, predictably, great argument about what constitutes ‘mentalese,’ the native language of our minds, as it were). Do words determine the shape of our thoughts? Well, it seems equally clear that that’s nonsense, and though it may and can be argued, it must be said most people don’t bother to try.
Steven Pinker, who was the entry point to the brief exchange between Kevin and I a few weeks ago, calls the idea ‘linguistic determinism,’ and argues as most do that the strong version is nonsense. A student of Noam Chomsky, he works from Chomsky’s idea of ‘Cartesian linguistics,’ that the brain has a ‘hard-wired’ built-in language acquisition device with an understanding of ‘universal grammar’, and suggests that language acquisition is an instinct. If we accept that language is an instinct, as Pinker and his mentor Unca Noam argue, it seems as if we must reject the proposition that language shapes thought. Some consequences of this :

In this, Pinker seems to be arguing not only against the idea that culture shapes language, but also the against idea that language shapes culture (by shaping thought). The use of the pejorative ‘insidious’ is a little unnecessary, but I’m not one who should poke people with sticks for using flowery language.
In his discussion of the idea, Pinker suggests three possibilities for interpretation:
(a) identicality: that language determines thought precisely, word-for-word;
(b) concept determinism: language determines (to an unspecified degree) what we
can think (doubleplus ungood!);
(c) linguistic relativity: that the form of our language (merely) influences what we tend to believe.
In Chapter 12 of The Language Instinct (quoted to me by Kevin), it seems that Pinker does concede the weak form :

Some commentators apparently do not take this as evidence that Pinker is admitting the weak formulation (c, above) of Sapir-Whorf. As I do not have access to a copy of The Language Instinct (no English language libraries and no damn money!), I’ll have to take their word for it.

The amount of time and energy that’s been expended on arguing about how vocabulary effects cognition surprises me, frankly. I think there’s a much more interesting discussion about grammar and deeper structures here that often seems ignored, at least in what reading I’ve managed to do.
The effect of such things on language users seems to me to be more pervasive and more subtle than simple differences in richness or breadth of vocabulary, on which most work and thought has seemed to focus.
One reason I believe this to be so is as a result of some of the fundamental differences in language structure between Korean and English (and to a great extent, the other European languages with which I have some familiarity). Please note that I neither claim to be a expert in Korean language (more of a lazy amateur), nor have I conducted any experiments or formal observations. First, some background. There are three ideas with some circulation about the earliest genetic relationship of Korean with other language families : 1) the traditional view that Korean is an Altaic language, sharing its origins with Manchu, Mongolian, and Turkish, amongst others; 2) the proposition that Korean has its origin in two language families, Altaic and Polynesian; and 3) the view that because of insufficient evidence to support a definitive relationship with other languages, Korean is a language isolate.
Regardless of its origins, Korean does share a number of features common to Altaic languages : words are built by agglutinating affixes, vowels within words follow certain rules of harmony, and articles, relative pronouns, explicit gender markers, and auxiliaries are not found.
Although Korean is not related to Chinese, as a result of history and geography more than 50 percent of the words in the Korean dictionary are of Chinese origin. Most legal, political, scientific, religious and academic vocabularies, as well as Korean surnames, and increasingly at present given names, are based on Chinese borrowings and can be written with Chinese characters, although meanings and pronuciations have often shifted as they have been adopted.
Although some basic words for body parts, clothing and agriculture are shared between Korean and Japanese, and other similarities exist, including grammatical structures similar enough that word-for-word translations between the languages is relatively easy, it is still uncertain whether the similarities are genetic or come as a result of historical borrowing between the two. Many features of Korean separate it from English and other Indo-European languages. Some of the most important of these (for my discussion here, at least) are the use of honorifics, relationship words, and different levels of speech (others include articles, plural markers, pronouns, adjectives, verb forms, demonstratives and so on).
Honorifics are markings for nouns and verbs that express the speaker’s attitude toward the addressee and the person who is being spoken of. Relationship words are blanket nouns denoting relationships between people that are commonly used in informal conversation between people, rather than given names – older brother, younger sister, uncle, auntie, grandmother and so on. (In the slummy, thin-walled building I used to live in in Busan, it was de rigeur on Saturday nights to hear sounds of passion and female cries of ‘Opa! Oh, opa! (older brother)’ from the playboy-next-door’s apartment.) These extend to the common practice of referring to a woman as ‘so-and-so’s mother,’ rather than using her given name.
There are four main levels of speech – polite-formal, polite-informal, plain, and intimate style – from which a speaker chooses, generally unconsciously, in everyday speech. The rules which determine the appropriate choice in conversation derive from the arcane art of knowing the ins and outs of the complex sociocultural fabric of Korean. It is equally inappropriate (in general) to address an older non-relative informally as it is to address a child with the polite-formal style, and mistakes like this may constitute a social breach (although it is generally understood that non-native speakers might make such mistakes). Depending on the relative status of the speaker, the person spoken to, and the person or thing that may be spoken about, the speaker can choose different words and forms to express intended meaning. For many basic verbs like eat, sleep, or give, at least two Korean words are available, each reflecting a different status of the subject or object of the verb. Each verb in Korean is further altered by a choice of grammatical affixes, adding not only grammatical information (such as tense), but carrying different levels of respect, deference, or politeness. Many nouns that refer to kinship or the household alsohave plain and honorific versions, the latter of which are used speak of another’s house or relatives, and the former of one’s own.
How does all of this relate to my earlier discussion of Sapir-Whorf, and considerations of how much and in what manner language may shape thought, and whether culture (loosely) determines language stucture, or vice versa? Don’t worry, I’m getting to that.
Korea is widely acknowledged to be the most Confucian nation in the world technically neo-Confucian, but there’s no need to split that particular hair here). Confucius focused on the need to maintain social order though willing or unwilling submission to the five primary relationships :
1) Ruler and subject
2) Parent and child (teacher and student)
3) Husband and wife
4) Older and younger person
5) Friend and friend
All of these relationships are explicity hierarchical, excepting, significantly perhaps, the last, although friendship of a Confucian bent is a considerably more meaningful proposition, it may be argued, than ‘buddies’ in North America might be.
Appropriate behaviour is expected for participants in each of these relationships, and the language used must be similarly hierarchical :

Although as in many nations, the strength of these traditional beliefs is fading, Confucian tenets still underly a great deal of the conscious and unconscious expectations of social behaviour, and deeply influence the relationships between the sexes and the generations.
The question that interests me, then, is this : do structures and forms like these in the Korea language shape the way in which Koreans think, particularly in terms of their relationships not so much to the world but to the people in it, to such a degree that we can say that language has given them a world-view substantially different than, for example, my own, as an English native speaker? It certainly seems so, to me.
Language is a tool for communication, a social construct, and it seems somewhat pointless to argue about what nouns one uses, and whether the presence or absence of a given bit of vocabulary in one language or another either permits and limits one’s ability to think about it. This may be so, but I don’t think it’s very interesting, except in the abstract.
More interesting to me is the idea that the structures of a language – in this case Korean – may expand or limit the way in which one thinks about something much more important than snow (for example) : how one fits into society, and how one interacts with other humans. That Koreans really do think differently about these things, and that this may spring (entirely, partially, as much or less so?) from their language.
Is this a valid argument for a weak form of lingustic relativism? Is it even something that comes under the Sapir-Whorf rubric? I’m not sure. An opposite, equally important question is this : is it the case that the language has come to have the form it does as result of culture and belief, rather than the opposite? Confucius was Chinese, after all, and from an entirely different language group!
Again, I’m not sure. The correct answer is usually ‘a little from column A, a little from column B’, I know. Like I said, though, I’m an amateur who hasn’t taken a single course in this stuff (yet!). So I’m curious about what you might think, dear reader, whether you’re a full-fledged linguist (like languagehat) or just, like me, an enthusiastic dabbler.