Names and Rules

01 Nov 2018 Justin B Rye

This is a postscript to a trivial point of SF/linguistics that I mentioned in passing on my Fandom Mince page back in 1999.  Ever since, inquiries have flooded into my mailbox at a rate of more than one a decade, so I thought I'd better explain properly.

A lot of American readers seem to have terrible suspension‐of‐disbelief trouble with “The Dispossessed”, the classic ambiguous utopia by the late great Ursula K. Le Guin.  How is it possible to imagine a society founded on the principle of absolute individual liberty if none of them own any guns?  Me, I have a rather different problem, for a reason also hinted at on my SF Xenolinguistics page… though it's something that SF and Fantasy authors take for granted so much of the time that it has practically attained the status of an accepted genre trope, like the starships full of humanoid aliens with psi powers that they somehow get away with.

Seven generations before the time the story is set, anarchists inspired by the political philosophy of Laia Asieo Odo came from the planet Urras to settle on its twin world, Anarres, where they established a new society, speaking a new language, and adopting new names with considerably fewer vowels in them.  From that point on, every newborn baby was assigned an arbitrary, unisex, guaranteed‐unique Pravic name.  Most of the Anarresti characters we encounter – such as the hero, Shevek, or his childhood friends Bedap and Tirin – have names that fit the template CVCVC, where C means any consonant sound out of the set {b, ch, d, f, g, gv, k, kv, l, m, n, p, r, s, sh, t, v, z}, and V means any vowel out of the set {a, e, i, o, u}.  Yes, that list includes gv and kv as single items; we see several names like Kadagv and Kvetur.  On the other hand the list doesn't seem to include h, which is odd given that the Pravic language definitely has that sound; we hear a lot about holum plants.  (Then again, in a later story the main feature of an Anarresti foreign accent is that initial “h”‐as‐in‐HOW becomes a “kh” sound… so I don't know what's going on there!)  Anyway, the overall effect is reminiscent of Hebrew, which is thematically appropriate for a kibbutz‐like community eking out a living in a near‐desert environment.

The egalitarian nature of the society makes names very important.  Honorific titles like “Doctor” or “Madam” aren't used; two people meeting for the first time simply speak their own names as an introductory greeting.  Mind you, when I say “their own” names, it's important to remember that as far as they're concerned, names aren't items of private property – Odonians don't believe in ownership!  The name Shevek isn't even comparable to the favourite pencil that he's hanging onto while he uses it to make his notes on chronophysics.  No, it's a label that everybody else needs to use to refer to him, and the most practical way of organising that is to put the census‐keeping computers in Abbenay in charge of them.  Still, if you don't like the name you've been given, we know it's possible to get it officially changed, and of course there's no law to stop you always introducing yourself as “Ace”!

They've already run through the whole supply of valid names several times – for instance, a previous (female) Shevek was an engineer, the inventor of a kind of mechanical bearing that is itself now known as a shevek.  But it isn't clear whether that supply includes every possible string fitting the template – some of them may be reserved for one reason or another.  For a start, I wonder what Anarresti playwrights (such as Tirin) do when they need fake names for use in works of realistic fiction – is there some equivalent to 555 phone numbers?

If all the valid names are in use, this raises the question of whether there are unfortunate children being assigned names like “Loser”, “Moron”, and “Vomit”.  Since Pravic is a constructed language, it's imaginable that the dictionary could have been carefully designed to avoid using the same CVCVC template for ordinary words, so that all the randomly generated names were guaranteed safely meaningless.  Nouns such as ammar (“sibling/comrade”) and kleggich (“drudgery”) make this idea look as if it might fly; but then there are those holum trees again…  Besides, if there ever had been any such rule, it would at best have lasted only until people started coining new words from names, and some of those would be especially worth avoiding: when a particular name gets into the dictionary as a synonym for “psychotic killer” it should probably be retired from use.  But the straightforward way of handling this would be to say that it applies to any name that has become a common noun, and thanks to the shevek we know that no such rule exists.

So, is part of the namespace being systematically avoided?  If you examine the sample of names we encounter, it looks as if a substantial proportion of the theoretically possible names must be getting skipped.  If every possible CVCVC string was in use (including for instance Fofoz, Fozof, and Zofof), there should be no such thing as common or uncommon letters; and yet f, o, and z turn up much less frequently in our sample than a, r, or t.

And if there are any CVCVC combinations that aren't being used, that just makes things worse for the fundamental problem with this whole scheme: that template only allows for 18×5×18×5×18 = 145,800 possible names.  That's nowhere near enough UUIDs even for the original million or so Odonian settlers!  Surely, if you were taking the trouble to design a new language and a new naming system at the same time as you set up the rest of your utopian society, you'd try to do better than that, wouldn't you?

When I raise this issue, people usually point out that we do occasionally see some more complex patterns in names mentioned in passing: Sessur, Terzol, Trepil.  This has the minor disadvantage of making it harder to imagine drawing a line between personal names and ordinary words, but more importantly, the names we see just don't match what we'd expect if this was what was covering the shortfall.  The “anomalous” cases would need to outnumber the “normal” ones by such a margin that you'd expect most of the names we encounter to be things like Bredrap or Tlirink; but instead, nine out of ten of them are CVCVC, with digraphs like sh accounting for most of the ones over five letters long.  And for all we know, every case where a name has an actual CC sequence might be an ordinary word adopted as a nickname – Sessur being Pravic for “Shorty” (which is better than being known by an official name full of unfashionable letters).  At any rate, names like those aren't enough to explain the problem away.

Wouldn't it make more sense to assume that Le Guin's unaccented transcription of Pravic is losing distinctions, and there should really be diacritics on some of the consonants (retroflex , palatal ñ) and vowels (close‐mid , nasalised )?  Maybe syllables should even get tone marks to distinguish low‐tone from neutral la and high‐tone ?  If the name that appears on the page as Tirin might be the US‐ASCII version of an original Ṭíríṇ then this idea might even stretch things as far as ten‐million‐plus possible names (so we could assume that the reason a, r, and t are so common is that half of the cases we see are actually , , and ; and meanwhile Terzol might be a haphazard attempt at conveying the sound of a retroflex z in Teẓol).  Unfortunately, the figure we're given for the current population is twenty million; and wait, didn't the Odonians originally hope that everybody on Urras would join their movement?  You'd expect them to have planned for a good few billion distinct names.

The only way a CVCVC template could get into that sort of range is if it had a hundred‐odd kinds of C, a dozen kinds of V, and half a dozen distinct phonemic tones.  Everyone's names would look more like Shẹwhǿkʾ, Gbẹ̀ḍẓäpf, Tlʾıṛîñ; and conversations would always begin with five minutes of “Did you say Shẹwhǿkʾ or Shẹwhǿk?”

If Le Guin had waited to get my advice before sending her manuscript off to the printers then once I'd finished apologising for making it far too late to qualify for a Hugo or Nebula in 1974 I'd have advised her not to give her protagonist the name Shevek.  The characters already routinely abbreviate one another's names to forms like Shev and Dap; and if they aren't going to be using full names all the time, why not declare that the official form is something more like Shevekozuf?  That way we immediately get a billion or so possible names (or several times as many if we assume a light dusting of invisible diacritics); nicknames like Shev or Zuf would be likely to get into the dictionary from time to time as ordinary nouns, but anything matching CVCVCVCVC is unmistakably a unique personal name.

Okay, now that I've established why it is that I can't believe in Shevek, I suppose my next target is (the nine billion names of) Ged.  How did the Archmage end up with a True Name three letters long?