Principles to Follow
2.2 Always go back to the primary sources
It is a joy to peruse a first-rate description of a language. However, not every grammar achieves a reasonable level of reliability. Someone who has themself written a grammar knows what to look for in another’s work—consistency, plausibility, clarity, explicitness.
When pursuing typological studies—when essaying a further contribution to basic linguistic theory—the linguist will consult a wide variety of sources. One must take care to use only good and reliable descriptions, to read them carefully in order to understand exactly how a particular category works in the language under consideration, and to quote accurately.
Not everyone follows these principles. The ‘armchair typologists’—who have not themselves undertaken fieldwork and written a grammar—tend to be the poorest. To quote one of very many examples, I was recently reading a linguistic monograph put out by a high-profile publisher and encountered two references to my work, both erroneous. First, I was cited as using a terminology which I abhor and have always argued against; and then there was a quotation which cited the right page numbers but the wrong book. Checking a little further, I followed up a phrase quoted from Maori, ‘the chief ’s pig’. There were two errors in this one phrase, an accent missed off a noun and the subordinate possession marker given as o in place of a (the point being made in the original is that this type of possession uses marker a rather than o). And so on. No doubt many things are quoted correctly in this and similar works; but the fact that there is a significant batch of errors makes the work unreliable as a source of typological generalization.
Some of the grammars I have written get referred to a good deal. About30 per cent of citations involve an error of some kind. Quite often what happens is that someone doesn’t bother to go back to the original monograph but instead copies information from a secondary source which may have included one or more mistakes. The errors are thus perpetuated, and perhaps compounded.
When Winston Churchill (1951: 616) was writing The Second World War, he found that his memory of events did not always exactly correspond with the actual record of what had happened. ‘It was only when I got home and searched my archives that I found the facts as they have been set out here. I am reminded of the professor who in his declining years was asked by his devoted pupils for his final counsel. He replied, “Verify your quotations”. ’
In scientific work, one should never rely on secondary sources. If Gikam- Faker quotes Matisoff as saying something about Lahu which is relevant to a topic you are working on, do not just rely on what Gikam-Faker says; go to the library and consult Matisoff’s 1973 grammar, to get full and correct information from the horse’s mouth, as it were.
2.2 always go back to the primary sources 65 My policy of always checking every secondary source yields some interesting results. Joseph H. Greenberg was revered by his students as someone who had an enormous number of facts about languages in his head. Ah!, but were they correct? In discussing coincidental similarities between languages, he states: ‘man means “man” in Korean.’ A nifty example, which I wanted to use. But one should always check in a primary source, so I phoned a Korean colleague. Sorry, he didn’t recognize it. I photocopied the page from Greenberg and sent it over. Definitely not in the modern language so he looked up dictionaries of Middle Korean. No, nothing at all. (There is a Sino-Korean bound root nam- ‘man’, which requires a suffix; this may possibly relate to the error.) People continue to quote this piece of nonsense; those who don’t follow the golden rule of always checking back in a primary source. The more this—and other—errors appear in print, the more veracity they appear to obtain.
As incorrect statements in the secondary literature get further propagated, the error may intensify. In1952, Archibald A. Hill examined the oft-repeated tenet that Cherokee has a dozen or more terms for washing parts of oneself, or of something else, but no general term ‘wash’, this lack showing that in a ‘prim- itive language’ there are a multiplicity of specific but no general terms. Hill demonstrates that there are in fact two verbs each with a general meaning— -wo ‘bathe’ and -e ‘wash’—and these can take an incorporated noun (e.g. ‘face- wash’, ‘clothing-wash’), or a classifier, or a reflexive marker, yielding complex (but analysable) verb forms which were taken as unanalysable by an 1820s observer, with this dictum being repeated many times since—even in linguis- tics texts of good repute—as an instance of ‘primitiveness’. The people quoting didn’t think to check with any of the good publications on Cherokee since the 1820s!
Possibly the most pervasive myth is that the Eskimos don’t have one general term ‘snow’ (as do speakers of ‘sophisticated languages’ like English) but instead dozens, or hundreds—depending on the degree of generosity of the reporter—of specific terms. In fact, the number of terms for ‘snow’ in Eskimo is two (there are many complex forms based on these, as indeed in English we have snowdrift, snowstorm, snowflake, snowfall, snowbank, snowball). In 1986, linguist Laura Martin published a scholarly rebuttal of the popular belief about the myriad words for snow in Eskimo, and Geoffrey K. Pullum has shouted her dictum as loud as only he knows how (in a1989 article and a 1991 book, entitled The Great Eskimo Vocabulary Hoax, and other irreverent essays on the study of language). But the myth is now in the public domain and surely nothing will arrest its piquant appeal. However, it does concern language, and it is important that anyone who calls themself a scholar of linguistics should recognize this and other fables for what they are.
Misrepresentation can proceed further, in ways that would be unbelievable if not attested. In 1978, a linguist by the name of Janice Jake published a short paper ‘Why Dyirbal isn’t ergative at all’ (in fact, the identical paper was published in two different journals). The argument hinged on a sentence she provided, the meaning she assigned being ‘man told woman to scrape beans’. All of her information on Dyirbal came from my 1972 grammar in which this sentence does not appear; Jake had simply manufactured it. It is, in fact, not a grammatical sentence in Dyirbal; it could be interpreted—as many ungrammatical sentences can be—but could then only mean ‘man told woman [to do something] and [he] scraped beans’. (And there were other things made up by Jake in her paper, besides various errors of copying.)
As has been shown, a good deal of published work on linguistics is sloppy and unreliable. The Jake paper is an extreme example; but in well over half the papers and books coming out nowadays there are substantial errors of fact and/or interpretation. Yet the discipline is important and worthwhile. Lin- guists who themselves undertake fieldwork and write comprehensive gram- mars learn the value of accuracy and achieve the ability to distinguish the reliable from the unreliable in other people’s work. And, as a matter of habit, they follow Churchill’s advice always to check back in primary sources.