||-Tripping over tramlines-||

“Tramlines”, icydk, are those upright parallel bars that OED1 and OED2 editors used to indicate that a word was “alien or not fully naturalized”. So, for instance, zeitgeist you may recognize as a word of German origin, not infrequently heard in English. In OED1 (1928) it appeared as ||Zeitgeist, and this mark was preserved on revision in 1986.

Is zeitgeist still alien to English? Or is it naturalized, like menu? Should it be spelled with its German capital initial, Zeitgeist, as my checker insists? To my ear it is relatively unexotic, but there’s room for disagreement, and I’m not sure I’d want to write the word without italicizing it. I’d write plain “zeugma” though, and probably “zeta” too, though these are also given with || in OED1 and OED2.

OED3 doesn’t use tramlines, no doubt a sign of the more cosmopolitan spirit of our age, and also the best way of avoiding disputes over what is basically a matter of linguistic and cultural intuition. Arnold’s “Zeit Geist” is Shaw’s “Zeitgeist”, Auden’s “zeit-geist”, and Piper’s “Zeitgeist“, each indicating a different level of domestication. There’s a temporal dimension to this, too, obviously–in 1928 yoghurt appeared in OED as “||Yogurt”, but when Burchfield revised in 1986 it became just plain “yogurt”.

What the lexicographer does with such borrowings can be read in different ways, e.g. as a generous extension of the range of the language, or as a sequestration of “foreign sounding words” away from the core of the language. This too can be controversial. When it comes to the global Englishes of the former British colonies, what sounds perfectly natural to an English speaker in Delhi may sound awfully foreign to a lexicographer in Oxford. So, for instance, in OED2 longeur, pourparler, roulade, soupcon and vraisemblance do not receive tramlines, while chupatti, dhoti, pukka (lately of Jamie Oliver fame), and yogi do.

Sarah Ogilvie wrote a book about the OED and “World Englishes” (Words of the World: A Global History of the OED) which I have been going over somewhat closely this last week. There’s a chapter there on tramlines, and one on loanwords, borrowings, and World Englishes. In a future post I’ll discuss the latter. Here I want to correct a mistake or two about tramlines in Ogilvie, and see what these corrections might mean for her analysis.

On p.81 Ogilvie presents us with a graph of the percentage of tramlines per letter in OED1. Since all the letters but S and W were edited by only one of OED1’s four editors (Murray, Bradley, Craigie, and Onions), this leads Ogilvie to some conclusions about each editor’s receptiveness to including “foreign” words (but also, at the same time, their tendency to mark words as “foreign”). I’ve reproduced her graph here, tidied up a bit from the original:

% Tramlines, per letter, per Ogilvie

These numbers, which Ogilvie draws from OED1 prefaces, lead her to conclude that:

  1. Proportionally, Murray included the most loanwords (mean 0.05) [i.e. 5%], followed by Bradley [4%]; Onions and Craigie included the same proportion [3%].
  2. The letter K, edited by Murray, included the most loanwords (13.4%) which was nearly double the proportion of loanwords in the second-highest letter P (6.8%), also edited by Murray.

Unfortunately, both of these conclusions are wrong. OED1 prefaces published tallies of Total Entries and Main Words, as well as words marked Alien or not Fully Naturalized (= tramlines). The proportion of tramline words was always calculated as a percentage of Main Words, not including subordinate and variant forms. But Ogilvie divides the tramline count for XYZ by the total number of entries, not the number of Main Word entries. Thus the graph should look like this:

% Tramlines per letter, corrected

XYZ, the only volume edited by Onions, is 6.6% tramlines here – more than double what Ogilvie reports. This makes Onions by far the most-tramlining editor on average, a fact worth thinking about, since he stopped using them, controversially, for the First Supplement (1933), which he co-edited with Craigie. And, it makes XYZ the second-most-tramliny alphabetical range after K, rather than the 18th-most, as in Ogilvie’s accounting.

But why should we lump XYZ, just because that Preface presents only total counts from the range? The volume is only 105 pages long, and takes about a half-hour to scan for tramlines. I did this, then got two of my RAs to do it as well, and then I did it again just to be sure. Then, I re-counted the number of Main Words per letter, by looking at the <ST> tag in the background OED data (subordinate entries are marked “xref”). The result is somewhat illuminating, on several counts:

% Tramlines, per letter, including X Y Z individually

So not only does Onions have proportionally more tramlines than Murray, K is now the third-most tramliny letter, after X and Z, and not by a little–24% of X is alien, and 21% of Z, as anyone who has looked through the seven pages of X or 20 pages of Z would immediately appreciate. [See, e.g., p.103 ZOSTER to ZWINGLIAN, [click here for image], which counts nine tramlines for nineteen headwords.]

This affects Ogilvie’s argument because much of it turns on the kinds of foreign words that are likely to receive tramlines. She quotes Murray’s note that both J and K ‘contain a very large number of words adopted from Oriental, African, American, Australian, and Oceanic languages’, and points out that ‘the letter K in English may have the largest number of loanwords because most languages of the world display the voiceless velar stop and choose to represent the sound orthographically with the letter K, not C, G, or Q.’ Indeed, the top source languages for tramlined K-words in OED1 are Arabic, Persian, Greek, French (transmitting its own borrowings, e.g. kirsch, kiosk, kourbash), Hindi (or “Hindustani”), German, Maori, Dutch, and Turkish.

But the story doesn’t work so well for X or for Z: all but one X tramlines are either originally either Greek or Latin, whereas Z words are 58% Latin or Greek, and 88% Western European (Y words, fwiw, are from Sanskrit, Hindi, Russian, and Chinese). I think this picture is much more intriguing, and would like to know what Ogilvie might make of it.

Ogilvie wants to make an argument about individual editors, but the very feature that lets us know who edited one entry makes any comparison between them almost meaningless.  Because they each edited different alphabetical ranges, there’s no way of knowing if the ratio of tramlines to entries is due to the editor or the inherent features of the alphabetical range.

Looking at X, Y, Z individually illustrates this very well: they’re all by the same editor (Onions) and all fairly exotic, but Y has a very low proportion of tramlines comparatively, even though it has roughly the same raw number as X. This is entirely due to the large number of obsolete Old-English y- words, like ymet, ymixt, ymolten, ymorþred, ympliȝeþly, crowding out the newcomers like yaboo, yagé, and Yamato. The fact that letter ranges behave so differently in this respect means that you really can’t compare OED1 editors at all in this way. [I’m working on another method, which I hope to follow up on in a future post].

The Second Supplement, however, we can compare to OED1, since Burchfield was responsible for all the letters. Elsewhere in her book Ogilvie reports that, based on a case study, ‘Burchfield deleted 17% of all neologisms, adaptations, and loanwords in the 1933 Supplement’ (p. 181). I distrust the case study method, and will have more to say about this in a future post on this subject, but for now let’s just assume this is true or nearly so. His deletion rate would have to be taken together with his inclusion rate. But Ogilvie doesn’t rank Burchfield alongside the editors of OED1, presumably because his Supplement didn’t publish counts of tramlines as the original volumes did.

Here then are the those same counts, in percentage terms, for the entries Burchfield updated, and those he added. They suggest a story that requires a lot of elaborating, especially when read alongside the OED1 graphs [and this explains the previous odd Y axis scaling, btw]. The headline is that a far greater proportion of Burchfield’s additions carried tramlines, for most letters right across the alphabetical range, with the notable exception of X.

The elaboration I leave to you…

% Tramlines per letter, Entries updated in 2nd Supplement

% Tramlines per letter, Entries added in 2nd Supplement

No Comments

Leave a Reply

Your email is never shared.Required fields are marked *