Kaktovik Numerals on Unicode.

Amory Tillinghast-Raby writes for Scientific American (archived) about an interesting story of numeral creation:

In the remote Arctic almost 30 years ago, a group of Inuit middle school students and their teacher invented the Western Hemisphere’s first new number system in more than a century. The “Kaktovik numerals,” named after the Alaskan village where they were created, looked utterly different from decimal system numerals and functioned differently, too. But they were uniquely suited for quick, visual arithmetic using the traditional Inuit oral counting system, and they swiftly spread throughout the region. Now, with support from Silicon Valley, they will soon be available on smartphones and computers—creating a bridge for the Kaktovik numerals to cross into the digital realm. […]

The Alaskan Inuit language, known as Iñupiaq, uses an oral counting system built around the human body. Quantities are first described in groups of five, 10, and 15 and then in sets of 20. The system “is really the count of your hands and the count of your toes,” says Nuluqutaaq Maggie Pollock, who taught with the Kaktovik numerals in Utqiagvik, a city 300 miles northwest of where the numerals were invented. For example, she says, tallimat—the Iñupiaq word for 5—comes from the word for arm: taliq. “In your one arm, you have tallimat fingers,” Pollock explains. Iñuiññaq, the word for 20, represents a whole person. In traditional practices, the body also serves as a mathematical multitool. “When my mother made me a parka, she used her thumb and her middle finger to measure how many times she would be able to cut the material,” Pollock says. “Before yardsticks or rulers, [Iñupiat people] used their hands and fingers to calculate or measure.” […]

The Kaktovik numerals started as a class project to adapt the counting system to a written form. The numerals, based on tally marks, “look like” the Iñupiaq words they represent. For example, the Iñupiaq word for 18, “akimiaq piŋasut,” meaning “15-3,” is depicted with three horizontal strokes, representing three groups of 5 (15) above three vertical strokes representing 3.

“In the Iñupiaq language, there wasn’t a word for 0,” says William Clark Bartley, the teacher who helped develop the numerals. “The girl who gave us the symbol for 0, she just crossed her arms above her head like there was nothing.” The class added her suggestion—an X-like mark—to their set of unique numerals for 1 through 19 and invented what mathematicians would call a base 20 positional value system. (Technically, it is a two-dimensional positional value system with a primary base of 20 and a sub-base of 5.) […]

Now support from Silicon Valley is helping to reignite the Kaktovik numerals. Thanks to efforts by linguists working with the Script Encoding Initiative at the University of California, Berkeley, the numerals were included in the September 2022 update of Unicode, an international information technology standard that enables the digitization of the world’s written languages. The new release, Unicode 15.0, provides a virtual identifier for each Kaktovik numeral so developers can incorporate them into digital displays. “It really is revolutionary for us,” Judkins says. “Right now we either have to use photos of the numerals or write them by hand.”

There are very useful illustrations at the link, as well as much more about history and use. Thanks, Eduardo!


  1. From the article:

    1. tallimat—the Iñupiaq word for 5—comes from the word for arm: taliq

    2. Maggie Pollock, speaking in English, said that the system “is really the count of your hands and the count of your toes”.

    This wording suggests to me that Iñupiaq might be one of those many languages in which there is no distinction between “arm” and “hand”. “Fingers”, of course, are a different matter.

  2. The general structure is very similar to Mayan numerals. One shape is used for up to three fives and another for up to four ones. Above 19, base-20 place value is used.

  3. This wording suggests to me that Iñupiaq might be one of those many languages in which there is no distinction between “arm” and “hand”. “Fingers”, of course, are a different matter.

    Apparently, it’s the other way around.

  4. David Eddyshaw says

    Those WALS maps always induce in me an irrational desire to go “BUT THEY HAVEN’T CONSIDERED …”

    The hand/finger one unfortunately gives me less scope than usual for this, as the Oti-Volta languages belong to what seems to be an extremely common group where “finger” is different from “hand”, but is still transparently derived from it, often as “little hand” (e.g. Kusaal nu’ug “hand/arm”, nu’ubil “finger.”)

    Still, I was interested to see “In no language among those sampled is there an overtly marked construction for ‘hand’ based on a word for ‘finger’, because I think there is a good argument that this happened, historically at least, in proto-Oti-Volta.

    The stem for POV “hand” was certainly *nù-, and it’s difficult to believe that this could be completely unconnected with the totally homophonous stem *-nù “five.” However, a problem I once mentioned before is that nù “five” is found outside Oti-Volta as well, in languages in which “hand” is clearly unrelated. However, I hadn’t then come across proto-Bantu *-nʊ̀è “finger”, which certainly looks like a cognate. Proto-Bantu “arm, hand” was *-bókò, and this etymon does actually turn up in Oti-Volta, sometimes actually meaning “arm”, as with Mbelime bākīhṵ̀, though more often as “shoulder” e.g Mooré bã́oko; it’s only “hand” in Yom (bākā.) [The Yoowa just had to be different. Show-offs.]

    So between the number “five” and the cognates outside Oti-Volta, it looks to me like the original word for “finger” in POV has elbowed out (sorry) the original word for “arm, hand” from the semantic range of “hand” completety, and has designs on creeping up past the elbow too.

  5. Stu Clayton says

    So, in these languages, to express disapproval of someone you point the middle finger, but to congratulate you have to hand it to them ? As a result of these subtleties, I suppose everyone must arm themselves against misunderstandings.

  6. David Eddyshaw says

    Ah, but in West Africa you give people the thumb rather than the middle finger (a thing which it is useful for foreign hitchhikers to bear in mind.)

    (With hitchhiking, I have seen it done in the European manner in Francophone countries, but I expect that was just people showing off how sophisticated they were. Évolué stuff. Either that, or they just didn’t like my face.)

  7. Stu Clayton says

    Then how do they express “thumbs up” ? Don’t they know about gladiators ?

    Clearly one must be a rocket scientist to get anywhere near multi-culti. That’s why I have always refused even to learn how to eat with chopsticks. It’s just pretentious, forget cultural appropriation. I am proud to be an unassuming person.

  8. David Eddyshaw says

    I presume that the attrition rate for gladiators in West Africa must have been particularly high. This was probably frustrating for both participants and spectators, and may shed light on Tertullian’s somewhat wet-blanket views on gladiator shows.

  9. Stu Clayton says

    Ah, de spectaculis ! How well I remember the impression he made on me at the age of five, with his long variety of affected and unfeeling witticisms. Here is a zippy summary for those of little Latin.

    A bit further down:

    # Tertullian’s closing passage has been quoted (out of context, and translated ‘with judicious laxity’ – T.R. Glover (p xi)) by Gibbon in Decline and Fall, ch. xv, to rubbish him as ‘the stern Tertullian’ with his ‘long variety of affected and unfeeling witticisms’, and following him Matthew Arnold, ‘the fierce African’ of the ‘unpitying Phrygian sect’, with a view to anti-Christian polemic in both cases. #

    Hardly anyone writes like that nowadays, more’s the pity. Of course people still write with judicious laxity, but no one uses such an expression to call them out for it.

  10. David Eddyshaw says

    I have to say that T doesn’t really enter into the spirit of the thing. You can’t really imagine having a pint with him.

    (Ovid, on the other hand, recommends gladiator shows as a good place for picking up girls. You can imagine having a pint with him. Though you’d probably end up paying for the round yourself. And he’d probably get maudlin after a few, banging on about his bloody carmen et error.)

  11. Stu Clayton says

    Much like Prince Edward, he claimed to be “bewildered” by it all. But E’s error was Virginia, not Carmen.

  12. David Eddyshaw says

    What you get for getting mixed up with Julia, the Ghislaine Maxwell of her day.

  13. Those WALS maps always induce in me an irrational desire to go “BUT THEY HAVEN’T CONSIDERED …”

    I’ve come to the conclusion that WALS must be treated with considerable caution.

    For example, Mathew Dryer’s characterisation of “relative clause” at https://wals.info/chapter/90:

    A construction is considered a relative clause for the purposes of this map if it is a clause which, either alone or in combination with a noun, denotes something and if the thing denoted has a semantic role within the relative clause.

    I challenge anyone to translate this intelligibly into another language, let alone make sense of it in English.

    WALS is, of course, only as good as its sources. There is a group of languages that supposedly have “internally-headed relative clauses” as a “non-dominant” type. WALS dutifully puts these on its map at https://wals.info/feature/90D#2/28.6/154.7. For Japanese, the phenomenon of “internally-headed relative clauses” was first raised by Kuroda (who actually called them “headless relative clauses”) in the 1970s, but the very status of this type of construction as a “relative clause” has been heavily debated since then. Unless you know your stuff, this glib reproduction of positions presented in the literature could be highly misleading.

  14. the Iñupiaq word for 18, “akimiaq piŋasut,” meaning “15-3,” …

    There was a thread recently (LangLog Chris Button and ff) hypothesising that in Tibeto-Burman, “Chinese”? unspecified the words for ‘six’, ‘seven’ could derive from ‘and first’, ‘and second’. This system didn’t extend to eight.

    Can anybody explain the comment:

    The thing is that the word for “seven” clearly contains “two” in Chinese and even more explicitly in other Tibeto-Burman languages

    “Chinese” is annoyingly vague. “Word”? Meaning spoken? How does either 七 or qī “contain” anything? Especially 二 èr ?

    In Taiwan (can’t speak for the Sinosphere at large), numerals up to four are shown by holding up that many fingers, excluding thumb. Five is all fingers+thumb. Six is thumb+little finger; seven is thumb+index finger; … up to nine is all fingers except the little finger. Ten is kinda flash all fingers twice.

    Your other hand is thereby freed to hold a shopping bag/small child/moto handlebars/etc.

  15. Many Austronesian coastal languages of New Guinea count along the lines of 1, 2, 3, 4, hand-half, otherhand 1, otherhand 2, otherhand 3, otherhand 4, handsboth. Twenty is ‘man one’.

    Many Papuan languages in the highlands lack any numerals over 2, so 3 is 2+1, 5 is 2+2+1, and yet they use very effective tally systems to keep track of the pigs and other items they owe or are owed. Some tally starting from the little finger of the left hand via thumb, wrist, forearm, elbow, etc., all the way to the little finger of the right hand.

    I remember one linguist giving a talk about one of the highland languages (Enga?) that was based on 15 and multiples thereof. Why? Because you raise one hand to count, then the other, then one leg, but if you raise the other leg you fall down.

    Languages in the Admiralty Islands (Manus) tend to have subtractive systems after 5 or 6, so that 7 is minus3, 8 is minus2, and 9 is minus1. The same subtractive system is found in Yapese, 7 is ‘and3’, 8 is ‘and2’, and 9 is ‘and1’.

  16. David Eddyshaw says

    internally-headed relative clauses

    Needless to say, no Oti-Volta data … (a significant omission, because Oti-Volta languages do some cross-linguistically very unusual things with noun modifiers and determiners.)

    Kusaal relative clauses are all internally headed; this is pretty obvious in cases like

    On yɛl si’el la ka’ sidaa.
    He.NOMINALISER say something the is.not truth.NEGATIVE
    “What he said is not true.”

    It’s also true in cases like

    bikanɛ nu ku’om la
    child.that.NOMINALISER drink water the
    “the child who drank the water”

    but demonstrating this entails a whole lot of subsidiary arguments about the exact nature of noun compounding in Oti-Volta languages, which is of a kind that those clever Chomskyans (may they be forgiven for their arrogance) have determined by logic to be impossible. In other words, you can’t just look at the word order and know the answer: it depends on deep syntactic analysis, and analysis, moreover, of a kind where perfectly good (non-Chomskyan) researchers might come to different conclusions.

    I just don’t believe that it’s possible for a WALS-style one-size-fits-all “analysis” to convey anything linguistically useful at all about internally-headed relative clauses. I’ve no reason at all to think that Kusaal is particularly unusual in the intricacy of the questions it raises here: it just happens to be the language I know most about.

    One of the few African languages WALS mentions here is Supyire. Relative clauses in Supyire are in fact all unembedded, i.e. simply adjoined (usually preposed) to the main clause, which has anaphoric pronouns picking them up in situ. Calling these “internally-headed relative clauses” is frankly an abuse of terminology (it’s a question worthy of discussion whether they are actually relative clauses at all.) At the very least, they are structurally quite different from anything in either Kusaal or Japanese, and lumping all these things together as WALS does is an utterly valueless exercise.

    The Dryer definition “A construction is considered a relative clause for the purposes of this map if it is a clause which, either alone or in combination with a noun, denotes something and if the thing denoted has a semantic role within the relative clause” is far too vague. In Kusaal, not only actual relative clauses fulfil this criterion, but many, many uses of clause catenation, which are constantly used where SAE has non-restrictive relative clauses:

    Dau daa bɛ n mɔr pu’a yiimir
    Man TENSE exist CATENATER woman unique
    “There was a man who had one wife.”

    As Slashdot polls say, “if you’re using these numbers for anything important, you’re insane.”

  17. David Eddyshaw says

    [Thanks to Bathrobe for providing me with the opportunity to rant about WALS. I was disappointed over the finger/hand thing.]

    The last example should go

    Dau da bɛ n mɔri o pu’a yimmir.
    man TENSE exist CATENATER have his woman unique
    “There was a man who had one wife.”

    An entirely idiomatic specimen (at least as far as the syntax is concerned) from the BIble translation goes

    Pu’a sɔ’ da bɛ mɔr o bipuŋ ka kikirig dɔl o.
    woman certain TENSE exist [CATENATER] have her girl and fairy follow her
    “There was a woman whose daughter was oppressed by a devil.”

    All “relative clauses” according to Dryer.* Except they’re not … just because something corresponds, more or less, to a SAE relative clause in meaning, that doesn’t make it a relative clause. This is on a level with finding six cases in English because Latin has six cases.

    * You can actually demonstrate that these catenated clauses are embedded, even those linked by ka “and” (more or less.) The clause-final negative particle, which has no form itself but inhibits the normal loss of the final short vowel of the preceding word, follows such a “relative clause”:

    M daa pʋ nyɛ dau la ka o an na’aba.
    I TENSE not see man the and he be chief.NEGATIVE
    “I didn’t see the man as a chief.” (na’ab “chief”)

    where the final negative enclitic is induced by the indicative-negative particle preceding the main-clause verb.

  18. David Eddyshaw says

    The Yom numbers go “one, two, three, four, five, six, five-and-two, two-not-being-ten, one-not-being-ten, ten”, after which they count in twenties, like all civilised persons, but unlike practically every other Oti-Volta language.

    I don’t know why they ditched the perfectly good proto-Oti-Volta forms for “seven, eight, nine.” Probably just showing off again (a curiously underestimated driver of language change.)

  19. @DE … just because something corresponds, more or less, to a SAE relative clause in meaning, that doesn’t make it a relative clause.

    Trying to be fair, I think even Chomskyans (of whom usually I’d agree with your observation) would insist that equivalence of meaning is nothing whatever to do with the case. There must be some syntactic/inflectional difference versus merely producing two sentences/clauses in sequence:

    “There was a woman who had a girl/daughter”. “There was a girl/daughter oppressed by a devil.”

    Is how it would go in Pirahã, if I’m following Everett. Pragmatics (Grice again) says a double-mention of “girl/daughter” means we’re talking about the same person.

    Your rubric has “[CATENATOR]”. Presumably that’d not appear if talking about two separate girl/daughters(?) But if omitted, would Kusaal use some other device to emphasise the unusual case of talking about separate girl/daughters?

    And presumably you might use [CATENATOR] when the catenated clause doesn’t repeat material from the first, but does have connected meaning:

    “There was a woman who had a girl/daughter.” [CATENATOR] “There was a husband oppressed by a devil.” — meaning the daughter’s husband.

  20. There must be some syntactic/inflectional difference

    Yes, Chomskyans are past masters at the art of finding that other languages that appear superficially different from English covertly feature exactly the same categories if you just use the right tests. For example, relative clauses in Hindi (correlative relative clauses), Japanese (no relativiser or movement), and Warlpiri (which Hale doubted were relative clauses at all) have been shown by acolytes of the Chomskyan school to be just the same as English underneath.

    (At a deeper level, of course, Chomskyans haven’t even reached agreement on how relative clauses are supposed to be generated at base.)

  21. David Eddyshaw says


    I may have misunderstood your question, but …

    No, the catenater is a necessary part of the construction. You still need it even if you omit the possessor o before bipuŋ “girl”, which you could, without changing the normal interpretation; it’s just more idiomatic to put it in (illustrating, incidentally, that pronoun possessors don’t automatically make a noun definite in Kusaal: if the daughter had already been referred to, you’d say o bipuŋ la ‘the her daughter.”

    The brackets round [CATENATER] are because the particle, which at some abstract level is n, is usually actually realised as zero, although the wonders of Kusaal sandhi mean that you can still demonstrate its presence without getting all metaphysical about it.

    This construction with n is the one which has been misanalysed as a serial verb construction in most accounts of Kusaal: in fact, the n introduces a formally subordinate clause with the same subject as the main clause. If the subject (or the polarity) changes, you use ka instead of n: that’s what’s going on in ka kikirig dɔl o “who was followed by a fairy.” Although ka is usually said to mean “and”, it’s actually much more often used as a different-subject catenater introducing a formally subordinate clause; you can actually demonstrate that such clauses really are subordinate and embedded, as I said above.

    The standing exception is narrative, where so long as the action is proceeding in sequence without asides or flashbacks or descriptions, clause after clause begins with ka and lacks tense marking. But this use of subordinate-style clauses in narrative is very common in African languages (Hausa does it, for example: it uses its so-called “relative perfective” as the narrative past form.)

    Although this is the usual way to express what English does with non-restrictive relative clauses, Kusaal has a quite distinct actual relative clause construction, using clause nominalisation, which is restrictive by default, but can be used non-restrictively by putting a headless relative clause in apposition to a preceding noun instead of incorporating that into the relative clause itself. In the Bible translation, Eve gives part of the apple to o sid onɛ da bɛ nɛ o la “her husband, who was there with her.” If it were restrictive, it would go o sid kanɛ da bɛ nɛ o la, where sid kanɛ “husband-that.NOMINALISER” is shown by its tones to be a compound, despite the standard orthography.

  22. >In traditional practices, the body also serves as a mathematical multitool.

    In the course of using a rivet gun to secure gunwales on 10 canoes one spring, I discovered that my handspan is precisely 9 inches, which really accelerated my boatbuilding.

    My wife refuses to acknowledge this and insists on confirming my reckonings with a tape measure.

    (Except that spell check is insisting she’s using a tapé measure?)

  23. David Eddyshaw says

    In general, what looks like parataxis in Kusaal is generally actually subordination: the idea that it’s paratactic is an illusion caused by mechanically translating ka everywhere as “and.” It’s not like Piraha, where the recursion is in the semantics but not the syntax (apparently); even quite everyday Kusaal loves formal subordination and recursive embedding. More so than English …

  24. Ted Applebaum says

    John Baez wrote his brilliant reactions on Mastodon

  25. David Eddyshaw says

    Nice pictures!

    It’s unfortunate in that the comments make a big deal over the “Eskimo” word being Bad. (Along with prissy asterisks to show how sensitive the writer is, the whole thing.)

    I’m a bit conflicted about this; historically, it’s a plain error: “Eskimo” does not come from any word meaning “eater of raw flesh.” It’s a myth, like Oprah’s “squaw/vagina” myth.

    On the other hand, if enough of the people in question mistakenly believe that this is the real etymology, well, it is offensive, like it or not, and should be avoided.

    To make life more complex, I believe that there are Alaskan Yup’ik who object to being called “Inuit” on the perfectly sensible grounds that they aren’t Inuit, and prefer to be called Eskimo.

  26. @ DE

    Well, that cleared things up nicely!

    I think I need a crash course in the Kusaal language — you know, how to speak it, and how to put sentences together in concrete terms. Linguistic terminology (Chomskyan or otherwise) only confuses the uninitiated unless you actually know the language in question.

  27. Stu Clayton says

    Linguistic terminology (Chomskyan or otherwise) only confuses the uninitiated unless you actually know the language in question.

    And when you do, you don’t need the terminology. Actually knowing a language saves so much time otherwise spent in discussing it.

  28. And when you do, you don’t need the terminology.

    Well, yes, but if you don’t speak a language natively then terminology and explanations are a handy way of learning it. Simple terms like “plural”, instead of “use this particular form of the word when you’re talking about more than one”, and even that explanation involves using linguistic concepts like “form” and “word”. Another way, of course, is to keep repeating examples to the student until the penny drops. “Oh, you mean that you say ‘dogs’ when you mean more than one dog! Wow! And the same for ‘rat’ and ‘man’ and ‘crumpet’! And all you have to do is add an ‘s’. Except, of course, that you don’t say ‘mans’, you say ‘men’. Hmmm. And when you’re talking about more than one, you don’t say ‘goes’ you say ‘go’. You know, ‘The rat goes’ and ‘The rats go’. And ‘man’ is exactly the same (except you don’t say ‘mans’, you say ‘men’): ‘The man goes’, ‘The men go’. And the same goes for all the other different, er, words, for, er, doing things or being or whatever.

    That might be how children learn, but I think I prefer grammar as a shortcut. The key is making sure the abstraction is always tied to the concrete. Linguistics is about finding greater and greater abstractions in order to account for larger sets of linguistic phenomena. If you don’t know the concrete manifestations, the abstractions simply evaporate into confused nothingness. That is when the rarefied world of the linguist, initiated into their abstruse theories and terminology, easily loses touch with reality.

  29. David Eddyshaw says


    I wish I had the skill set needed to write a decent paedagogical grammar of Kusaal, but unfortunately that’s far from being the case.

    I had a nice email a while back from somebody training aid workers in Ghana to speak Kusaal, asking if I minded them using a printout of my grammar (which I obviously didn’t), but it turned out that the only bit they were really interested in was the (brief) vocabulary section at the back. Oh well …

    Other communications I’ve had about it have been focused entirely on the placenames I cite (in one case) or on my throwaway and not very profound or extensive remarks about Kusaasi traditional culture (that one was from an actual Kusaasi; I was very tempted to suggest that he just ask his grandparents instead, and get back to me with their comments.)

    It’s basically the book I wish I’d had when I first went to Ghana, as somebody with abundant opportunities to hear the language and attempt to speak it, but absolutely no idea about its grammatical structure and nowhere to find any description of it.

    David Spratt wrote an introduction to Kusaal, but it’s a grand total of forty-two pages long, and I think I probably have the only copy in existence outside the offices of GILLBT in Tamale. It would have been a help if I’d come across it a couple of years earlier than I actually did, though. It’s heavy on the language-lab-style pattern-drill stuff that was so popular back in its day (early 70’s), so “forty-two pages” actually doesn’t cover a huge amount, though it is quite accurate in what it does cover.

    Anthony Agoswin Musah (an L1 speaker and actual linguist) wrote a full-length reference grammar of Agolle Kusaal, which was published by Peter Lang in 2018; his approach is very different from mine.

    AFAIK there’s no adequate teaching grammar (or reference grammar, come to that) of the Toende dialect (though there is a nice dictionary.)

  30. David Eddyshaw says

    Oh, and the indefatigable White Father André Prost wrote a teaching grammar of Toende Kusaal, which is not bad at all, but unfortunately ignores tone altogether, which is a pretty major drawback. It’s particularly interesting, though, because it dates from 1979, and it’s detailed enough to show that there have actually been some significant phonological changes in the language even over the past forty years. Language change before our very eyes …

  31. How do you pronounce Toende in Ghanaian English? Your grammar only gives the Kusaal pronunciation.

  32. David Eddyshaw says

    Tone-dee (with stress on the first syllable.) The French is Tondé, pronounced as you’d expect.

    In Toende Kusaal itself it’s Tɔɔn, Agolle Tuon [tʰuɵn], both low tone throughout. Literally, it means “in front”, i.e “West”, as opposed to Agolle “Upper.” It’s primary a geographical term; the name of the dialect itself is Tuonnir in Agolle Kusaal.

  33. Tone-dee

    Something like /ˈtoʊn.diː/?

  34. David Eddyshaw says

    Yup. I was too lazy to do it in proper IPA.

    The mismatch with the actual Kusaal forms is due to the local custom of citing proper names without the characteristic Kusaal apocope of final short vowels in citation forms: it goes back to the Brits’ use of Mamprussi interpreters when they first arrived on the scene in the early part of the 20th century, but has long since been established as a convention in its own right, and applied even to forms which are distinctively Kusaal and not Mampruli at all. The appropriate final vowels are still (as it were) accessible to speakers, because they are not deleted in absolutely all contexts: the Agolle Kusaal for “it’s not Toende”, where the negation counteracts the usual final vowel apocope, is Li ka’ Tuonnɛ.

  35. capra internetensis says

    Doesn’t everyone measure with body parts? Sweet numerals though.

    The natives of Northwest California used to go one better and get the standard length measures (of shell money and such) tattooed on their arms, which seems very practical. (Also were very big on accumulating money, staying slender, and visualizing the things they wanted to obtain. So Californians have apparently always been Californians, at least since the inception of the Gunther Pattern.)

    Inuit do eat raw flesh, and aren’t ashamed of it, so I don’t think it’s the pseudo-etymology. Heck, I like raw flesh too, and I have to pay through the nose to eat it. (Okay fine, I have to pay through the nose to eat it with a negligible risk of parasitosis.) One of our favourite Governor-Generals famously ate a raw seal heart, which épatered certain of the bourgeoisie. For whatever reason, in Canada at least, Eskimo for Inuit is now used only by the purposely offensive or hopelessly out of touch.

    Also I just learned that Kalaaleq “Greenlander” may well be the Inuit pronunciation of Skraeling.

  36. Trond Engen says

    Inuit eat raw flesh and aren’t ashamed of it, but that may be just what attracted the pseudo-etymology.

    As the Kalaaleq example shows, the offensiveness of a word has little to do with etymology, real or imagined, and all to do with how it’s identified with culturally ingrained bigotry and enforcement of social and economic subordination. This means that offence may be dependent on context and referent rather than the word itself. ‘Bantu’ became offensive to black South Africans during Apartheid. Still, I believe that speaking of Bantu languages and the history of Bantu peoples is fine, and that what is considered offensive is describing a person as “a Bantu” or black South Africans as “Bantus”*. The situation with Eskimo in Canada could be similar.

    * Warning: Some may consider a week doing tourist stuff in the Western Cape insufficient as field study.

  37. @DE No, the catenater is a necessary part of the construction.

    Ah, ok. You consistently put it in square brackets, whereas other upper-case terms not. So I assumed square brackets meant omissible. (Too used to reading Backus-Naur form.)

    The brackets round [CATENATER] are because the particle, which at some abstract level is n, is usually actually realised as zero, although the wonders of Kusaal sandhi mean that you can still demonstrate its presence without getting all metaphysical about it.

    It’s the “realised as zero” that worries me wrt the methodological/typological/”metaphysical” debate [you know who I mean]. I get it that pragmatically, marking a possibly-zero or always-zero element makes perfect sense, but it opens the gate to …

    English is SAE with six cases, it’s just that in many cases (hah!) — except pronouns and possessives, the case endings are “realised as zero”. We can look historically to demonstrate older English did have non-zero case markings. (If not SAE, at least Standard Average Germanic.)

    In Kusaal presumably the ka can’t get “realised as zero”, so the jobbing linguistician can concoct minimal pairs triples to elicit the legitimacy of the n/zero/ka alternants.

    So now we need to shore up our methodology to say always-everywhere realised as zero is not allowed.

  38. January First-of-May says

    although the wonders of Kusaal sandhi mean that you can still demonstrate its presence without getting all metaphysical about it

    I think this is supposed to be saying something to the effect of “if you made a Panini-style analysis of Kusaal, you’d need separate rules for the assorted realized-as-zero words/morphemes”.

    Vaguely similar example from a better-studied language: the default French plural ending -s, which is silent (= “realized as zero”) in most non-liaison contexts.
    (IIRC French verbal conjugation actually has a bunch of nominally-different zero morphemes, though I’m not sure if all – or any – of them can actually be distinguished by [e.g.] sandhi and not only in the written forms.)

  39. David Eddyshaw says


    I know what you mean, and as a Nominalist in good standing I actually share your metaphysical squeamishness …

    However, the Kusaal example is far from metaphysical.
    The square-brackets convention is actually just nicked from the industry-standard Leipzig glossing rules (Rule 6):


    where it figures because the issue turns up a lot in language description. But the Kusaal example is even less metaphysical, as I was hinting with my reference to “sandhi” (JfoM’s comment is on the correct lines.)

    Many Kusaal clitic words induce changes in the preceding word (even proclitics can do this, not only enclitics) which historically arose from partial suppression of the usual Kusaal apocope of word-final short vowels. The catenater particle is one such, It is also one of half a dozen words which themselves are of the underlying form V, and consequently are reduced to zero by this selfsame apocope process. So the particle itself disappears, but the effect on the preceding word remains:

    M nɔk sʋ’ʋgʋ kia’ nim la.
    I take knife {CATENATER] cut meat the
    “I’ve cut the meat with a knife.”

    “Knife” in isolation is sʋ’ʋg: the final vowel is a sandhi phenomenon, induced by the now-absent catenater.

    Because these various surface-zero words have different effects on the preceding word-final vowel, and also leave differing tone sandhi effects in their wake, you usually can even tell which word you’re dealing with without any guessing or appeals to context.

    If this all seems a bit of a stretch (as I concede it might), the personal pronoun objects provide a neat illustration of the whole process. All the non-contrastive personal pronoun objects do this partial suppression of preceding vowel apocope thing: thus with ba “them” (where the underlying form is baa, so it’s not reduced to b):

    M nɔŋi ba.
    “I love them” (nɔŋ “love.”)

    However, “you” (sg) has the underlying form , and therefore usually surfaces as f:

    M nɔŋi f.
    “I love you.” (Written M nɔŋif. in standard orthography, unsurprisingly but inconsistently.)

    Now (this is the interesting bit) “him/her” is underlyingly o [ʊ], and is deleted altogether by apocope; but before that happens, it rounds the preceding word-final vowel to o [ʊ]:

    M nɔŋo.
    “I love him/her.”

    Unsurprisingly, the standard orthography here mistakes the word-final [ʊ] for the pronoun itself, and writes this M nɔŋ o, but the tone on the vowel is exactly the same as the tone on the vowels that precede the other object pronouns, and with verb forms where the citation form has simplified a final consonant cluster after the apocope of the final vowel, the cluster reappears before all object pronouns, including “him/her”; thus (in standard othrography)

    M dɔlli ba.
    “I follow them” (dɔl “follow.”)

    M dɔllif.
    “I follow you.”

    M dɔl o.
    “I follow her.” actually pronounced [dɔllʊ]

    Lest this talk of “underlying forms” itself sounds too metaphysical: in Kusaal, they’re not theoretical, but actually turn up as real spoken forms, for example at the end of negated clauses:

    M pʋ dɔlli baa,
    “I don’t follow them.”

    M pʋ dɔlli fɔ.
    “I don’t follow you.” (So written, inconsistently, in the standard orthography, which is a poor guide to real word boundaries; though understandably so given the complexity of the issue in Kusaal.)

    M pʋ dɔl oo.
    “I don’t follow her.” [dɔllʊʊ]

    M pʋ dɔlla.
    “I don’t follow.”

    M pʋ nɔŋɛ.
    “I don’t love.”

  40. Thanks @J1oM, yes French verb conjugations show exactly the dangers of analysing as full SAE forms. Would anybody be positing those ‘realised as zero’ morphemes if it weren’t for the spelling? And because Latin, of course.

    And good point re the liaison: analysis needs to identify some of those underlying forms, to explain the liaison. Similarly the most parsimonious description of French adjectives is to posit the fem. form as underlying, with the last consonant realised as zero for masc.

    English ‘that’ complementiser “Mary believes (that) it is raining.” is usually omitted in casual speech (or MsWord wavy lines), but it’s never wrong to make it explicit.

    So we can point to non-zero realisations to motivate the systematic ‘reduced to zero’ in specific circumstances. As opposed to those [I will not name] who want to splatter elements all through the phrase structure to satisfy some theory-driven prejudice, but which always get realised as zero. (Hell-bent on showing this splatter as a ‘Language Universal’.)

    @DE, perfect explanation thank you.

    they’re not theoretical, but actually turn up as real spoken forms,

    Excellent! To keep the metaphysical at bay. Except …

    I’m wondering if ‘realised as zero’ is the most perspicuous way to describe it? ‘Realised in the previous word’ seems more felicitous, going by your description. And is it always and only the immediately previous _word_? Or rather the head of the preceding phrase? (Or does that amount to the same thing?) What if there’s an aside or interjection intruding? (Can that happen?)

    “I’ve cut the meat with that knife you sharpened yesterday.”

  41. David Eddyshaw says

    I’m wondering if ‘realised as zero’ is the most perspicuous way to describe it?

    True enough. I actually say “lack any segmental form” in my grammar.

    It is specifically the preceding word, not the head of the preceding phrase; in fact, the whole complicated thing is really a matter of phonology rather than morphology or syntax.

    The preceding word may well be the head of its phrase, but it doesn’t have to be, e.g.

    M nɔk sʋ’ʋkaŋa la kia’ nim la.
    I take knife.that the [CATENATER] cut meat the
    “I cut the meat with that knife.”

    Because Kusaal relative clauses are nominalised and internally-headed , “I’ve cut the meat with the knife you sharpened yesterday” does not actually differ in basic structure from this:

    M nɔk sʋ’ʋkanɛ ka fʋ sa pʋlis su’os la kia’ nim la.
    I take knife.that.NOMINALISER and you HESTERNAL.TENSE sharpen yesterday the [CATENATER] cut meat the

    Although in a serial-verb-like construction like this you wouldn’t expect an adverb to intervene before the catenater, for pragmatic reasons, part of the reason* for denying that Kusaal clause catenation is actually a serial verb construction is that you can indeed put adverbs (and even other subordinate clauses) before the catenator; an example from the 1976 Bible translation is

    Ka dau so’ due n zi’e la’asug la nidib sisoogin, n a Farisee nid ka o yu’ur buon Gamaliel, n a one pa’an Wina’am wada la yela
    and man certain arise CATENATER stand assembly the people among CATENATER be Pharisee person and his name call.IMPERFECTIVE Gamaliel, CATENATER be this.NOMINALISER teach.IMPERFECTIVE God law the matters
    “A man stood up among the people in the meeting, a Pharisee called Gamaliel, a teacher of the Law …

    It’s in the older orthography, and dates from a period when the particle was still often written, and presumably realised, as syllablic n. My own informants only did this in a few very restricted contexts.

    Farisee nid ka o yu’ur buon Gamaliel “a Pharisee called Gamaliel” is an instance of subordinating ka used to make the equivalent of an English non-restrictive relative clause, as I described above.

    * There’s also the fact that the catenater particle can follow a non-verbal clause, which can hardly represent verb serialisation.
    (Whole theses have been written on serial verbs in Western Oti-Volta languages – which goes to show that you can find anything at all if you look hard enough, even if it’s not actually there. I think people just assumed that e.g. Kusaal must have serial verbs Because Africa, and then never took a step back to go “But …”)

  42. David Marjanović says

    you can find anything at all if you look hard enough

    Rather the opposite: you can find anything at all if you don’t look hard enough at whether what you’ve found is what you think you’ve found.

  43. John Cowan says

    This is on a level with finding six cases in English because Latin has six cases.

    Alice’s brother’s Latin grammar textbook managed it: a mouse — of a mouse — to a mouse — a mouse — [by a mouse om. —] O mouse!” Note the now-American NGDA order.

    applied even to forms which are distinctively Kusaal and not Mampruli at all

    A morphophonological process referred to as artificial Mamprulication by one Eddyshaw some four years ago.

Speak Your Mind