Vai and the Evolution of Writing.

Tessa Koumoundouros at ScienceAlert writes about how “a rare script from a language in Liberia has provided some new insights into how written languages evolve”:

“The Vai script of Liberia was created from scratch in about 1834 by eight completely illiterate men who wrote in ink made from crushed berries,” says linguistic anthropologist Piers Kelly, now at the University of New England, Australia. “Because of its isolation, and the way it has continued to develop up until the present day, we thought it might tell us something important about how writing evolves over short spaces of time.” […]

“There’s a famous hypothesis that letters evolve from pictures to abstract signs,” says Kelly. For example, “the iconic ox’s head of Egyptian hieroglyphics transformed into the Phoenician [aleph] and eventually the Roman letter A,” the team explains in their paper. “But there are also plenty of abstract letter-shapes in early writing. We predicted, instead, that signs will start off as relatively complex and then become simpler across new generations of writers and readers,” Kelly notes.

The eight Vai creators set out to design symbols for each of their language’s syllables, inspired by a dream. Their chosen symbols represented physical things like a pregnant woman, water, and bullets, as well as more abstract traditional emblems. It was then taught informally by a literate teacher passing their knowledge of the script to an apprentice student (with 200 individual letters that must have been a challenge to remember!). This practice is still used today to teach the written language, which is now even used to communicate pandemic health messages. Kelly and colleagues from the Max Planck Institute analyzed the 200-syllabic alphabet of the Vai people from 1834 onwards using archives across several countries.

Over the first 171 years of its history, the Vai script did indeed become increasingly compressed. The simplification occurred over generations of users; symbols with the highest complexity were simplified the most. These changes are far from random, the research team explained. Languages pass a kind of natural selection process via memory and learning, where the hardest to recall features do not survive. “Visual complexity is helpful if you’re creating a new writing system. You generate more clues and greater contrasts between signs, which helps illiterate learners. This complexity later gets in the way of efficient reading and reproduction, so it fades away,” says Kelly.

As the letters became less complex, Kelly and team found they also became more uniform. This is despite the language never having been adopted for mass production or for bureaucratic needs. These uses are what seemed to help standardize other languages – for example, Mesopotamia’s writing standardization coincided with the implementation of state-wide systems. Changes in tools, from new writing implements and the invention of paper to use with computers also likely play a role in simplifying languages. “However the fact that Vai continued to compress over the entire length of the 19th century, at a time when there was little change in writing media, indicates that shifts in writing technology cannot be the full story,” the researchers write.

While the rapidness of this writing system’s evolution seems rather remarkable, the researchers suggest it occurred because its inventors and users already knew what writing is capable of, because they knew of its use in other cultures. This may have encouraged Vai people to quickly optimize their system.

There are images at the link; I posted about the effects of literacy among the Vai in 2008.


  1. David Eddyshaw says

    symbols with the highest complexity were simplified the most

    Well, yes.* The only thing that surprises me about this is that the authors of the paper seem to have thought it was surprising that this simplification should happen despite there not being much change in the media used for writing. That seems to be a thumping non sequitur to me. To be slightly fairer, they seem surprised that it happened so rapidly. But how rapidly would you have expected it to happen, exactly?

    * As opposed to “symbols with the lowest complexity were simplified the most”? Astonishing. Who’da thunk it?

  2. It’s interesting to try and imagine a scenario where the symbols of a writing system start to get more rather than less complex. Maybe if its primary uses are ornamental? I seem to recall that hieroglyphic Meroitic may actually have been developed from cursive to satisfy Egyptianising aesthetics, rather than vice versa.

  3. Try Chinese, where the steady growth in the number of characters spans millennia. This was at times achieved by adding disambiguating semantic indicators, which necessarily created more complex characters.

    The First Emperor famously unified the characters across the disparate states that made up his new empire, but the proliferation of characters continued till modern times. It has largely stopped, although the naming of chemical elements in very recent times involved the creation of a large number of new characters.

  4. Stu Clayton says

    Languages pass a kind of natural selection process via memory and learning, where the hardest to recall features do not survive.

    This is not true toot curt. Proof: the orthography of English and French.

    The masses simplify it in everyday life, but keeping it intact in all its gory is a life project of intellectuals.

    I don’t know what the hardest to recall features of German and English might be. I must have forgotten.

  5. with 200 individual letters that must have been a challenge to remember!

    Languages pass a kind of natural selection process via memory and learning, where the hardest to recall features do not survive.

    Why is that second quote talking about ‘Languages’, when the ostensible subject is scripts? And is it true of (spoken) languages? I rather thought spoken languages trade simplification in one feature for increased complexity in another.

    Anyhoo, perhaps someone could tell the Chinese about humans’ 200-glyph memory limit. Also about the simplifying bit. They could perhaps observe that Sinitic spoken languages have simplified in the sense of having many more homophones than historically; but not having correspondingly collapsed together the glyphs.

  6. Proof: the orthography of English and French.

    Also Irish “mainly based on etymological considerations” [wp], despite the C20th “simplification”.

  7. David Eddyshaw says

    As a very broad principle, creoles seem to start unusually simple and then progressively revert to the human mean by getting more complicated (though they mostly still seem to have a fair bit of catching up to do.)

    And if complexification were not a force in human language development, presumably all our languages would have become beautifully simple by now. This does not seem to be borne out by experience.

    As AntC points out though, script is a quite different issue.

    The Hieroglyphic script got much more complicated after the Middle Kingdom, with many more individual signs and many more unsystematic spellings. Part of this was no doubt driven by language change spoiling previous sound-symbol correspondences, as with English, but mostly it seems to have happened because scribes as a class had no reason to favour simplicity as a goal of the writing system at all. (This is not simple cynicism: efficiency is indeed not the only virtue. Full employment and job security for scribes are obvious virtues too*; but aesthetic considerations are not to be undervalued.)

    It would be perfectly possible for the Japanese to write their language entirely in kana (like Murasaki), or at the very least to only use kanji for Sino-Japanese morphemes …

    * I think we can all agree on this. We are all members of the scribal class here.

  8. Script can become more complex, by very similar mechanisms to spoken language: to wit, positional forms in Latin scripts (namely capital letters), positional forms in Arabic and Indic scripts, or the abundant medieval manuscript abbreviation forms. In the case of the latter, the development parallels that of linguistic complexification in sound systems, adding complexity to make communication faster and easier.

  9. There is a distinction to be drawn between scripts as a whole getting more complex and their individual symbols getting more complex. The former, as illustrated by the comments above, seems much more prevalent than the latter. I suppose the creation of dots in Arabic would be a case of the latter, but not a very big one.

  10. PlasticPaddy says

    I would suppose there would be two main drivers for changes in the writing of symbols, i.e., (1) changes in technology and (2) changes in the variety/purpose of texts (or literacy). If writing was executed with a stylus on clay or chisel on stone, then curves would be implemented as angled segments. And if writing was only for the purpose of unambiguous record keeping, shapes of letters would change to avoid ambiguity, i.e., one letter being used for another. Alternatively you can see bold or italic font used in writing to convey more sophisticated messages.

  11. @Lameen

    An example of the characters getting more complex is the so called gothic script, and its most recent incarnation, the German Sütterlin handwriting.

    Same thing for various Indian writing systems that developed from Brahmi.

  12. David Eddyshaw says

    The problem with Sütterlin handwriting as far as I’m concerned is that it’s not complicated enough: too many letters look just the same.

  13. I assume, complexity of Russian г is evaluated based on computer-made г and not hand-made 𝓰?

  14. Professor Helgason used Sütterlin letters for the names of some Lie algebras when he was writing on the board. In his textbook, they were in Gothic, as is pretty standard—e.g. 𝔞, 𝔤, 𝔥, or 𝔰𝔲(2). However, although he still referred to them as “Gothic,” for the first two (𝔞 and 𝔤), his handwritten versions were basically lower-case Sütterlin, and his 𝔥 was close to upper-case Sütterlin (although that’s not that different from the lower-case Gothic). I don’t remember what he did for the other letters, which came up a lot less frequently, since we didn’t talk much about the semi-simple groups in that class.

  15. I assume, complexity of Russian г is evaluated based on computer-made г and not hand-made 𝓰?

    I write 𝓰 for д, not г!

  16. Amanda Adams says

    Nothing about making letters with a chisel on stone favours rectilinear forms over curved ones. Sorry.
    Unless maybe you are thinking of an electric chisel, which I know nothing about.

  17. @David E

    They look the same to someone not familiar with it. But in reality it’s a bit like Q and O in ‘regular’ Latin. The letterforms in Sütterlin are more comlex then their ‘regular Latin’ counterparts: we have the situation where eg. 2 straight strokes for ‘v’ become something resembling a straight stroke with a 6 added to it.

    On reflection, i should have said Fraktur rather than Sütterlin.

    However you bring up a good point about the evolution of writing systems: sometimes the characters are simplified so much that they become identical to eachother. This brings in another type of complexity: reading recognition. A good example is Pahlavi. Arabic has also had some of this, but at least the Arabs introduced points to differentiate otherwise identical letters.

  18. PlasticPaddy says

    You are right. However, what I observe, e.g. here is
    The u’s are uncurved.
    The Os have pointed tops and are more elliptical or even n-gonal than round.
    Flourishes or ornamentation are linear (bars)

  19. I write 𝓰 for д, not г!
    I felt that something is wrong (tha 𝓰 is a letter that means different things in English and Russian…).
    You are right:(

    The point was that “X-ish alphabet” makes one thing of a table in a book. And that is not right.
    Russian /d/ is {𝛛, 𝓰, д, …}.

  20. jack morava says

    @ bathrobe: the naming of chemical elements in very recent times involved the creation of a large number of new characters…

    can you say more about this? do you mean transuranic elements?

  21. Europeans at a certain moment developed Capital letters (Y noted this, and I liked his other observations too).

    Now, your initials can be arbitrarily ornamented to any unreasonable degree, with flowers, birds, boobs, faces.
    Depending on how often you use capitals, those also can be ornamented.

    I see some conceptul issues with defining coplexity. Complexity of what we measure?

    The artist who embedded a naked lady in every initial, he did not try to “copy” anyone’s shape. He clearly based it on a certain criteria that would allow the reader to recognize the letter — and a space of variation that allowed him to embed the lady. Else: the subspace of the space of all images such that every element of this subspace is recognized as б is large enough to allow for elementes that contain ladies. Readers can still recognize б and also recognize a person, and that it is female and that it is naked.

    I assume, “the letter б” is the space of shapes that are б OR an algorythm that we think (incorrectly maybe!) exists in your brain that makes brain say “yes, this image is an instance of б” rather than a pixel-precise image of a person who bent her leg and arms.

    (1) when we compare complexity of scripts we need (we have to) account for the fact that some Russian б’s have legs and arms and breasts and hair.
    Do we say: “…and thus the script is infinitely complex”? I think, no.

    (2) Next we maybe postulate two levels: the Platonic ideal б, and the ornament. We say: this О is simple, all these hooks are just decorations. They do not “belong” to the ideal O.
    And how do we say, what in 𝔜 is Platonic and what are decorations?

    The second issue is the question whether this “Platonic” б exists.

  22. January First-of-May says

    Europeans at a certain moment developed Capital letters

    I think you might be conflating two different innovations: 1) fancy initials, 2) script bicamerality. Those are closely related but not perfectly correlated; there are bicameral texts without fancy initials, and unicameral texts with them. AFAIK the initials came first.

    IIRC the development of bicamerality is usually described as the addition of lower case, i.e. non-capital letters. I’m not sure to what extent this is true. It does sound plausible that the added letters were in fact the capitals (possibly descended from the aforementioned fancy initials).
    The capitals we use now are almost identical to the non-cursive Roman letters (as seen on monumental inscriptions and on 1st century coins), which are consequently perceived as capitals [though Roman cursive is not much like anything in modern script]; but this is only true due to a deliberate archaising decision in 15th century Italy (which then spread into other places because of the printing press), and Germany, in particular, kept using their very much non-Roman Gothic capitals.

    (We’ve discussed on LH before – though I don’t recall offhand exactly where [EDIT: found it] – that it’s highly likely that the modern “Roman” and “Italic” font styles, which look very much the same in most modern fonts [especially the former], each ultimately descend from the handwriting of one specific 15th century Italian scribe.
    I looked it up a few months ago, and for the “Italic” font style, the name usually provided is Bartolomeo Sanvito, but for the “Roman” font style – the one I’m writing in right now – the case is open as far as I can tell, though it was probably someone in 1450s Padua. It might well have been the same Bartolomeo Sanvito.)

  23. David Marjanović says

    the development of bicamerality is usually described as the addition of lower case, i.e. non-capital letters. I’m not sure to what extent this is true.

    AFAIK, Roman cursive died out with the western empire (which is why it’s so hard to read if you haven’t studied it). Carolingian minuscle developed out of the capitals afresh. Today’s lowercase letters are fashioned after Carolingian minuscle in the erroneous belief that the latter went straight back to the western empire. The capitals have remained basic this whole time.

  24. From what i understand the complexity, referred to in this research, was about how each character in the script looks. Eg. Number of strokes involved and the ease of writing each character.

    In the New Scientist article that described this research, they described how the number of zig zag lines in a particular character diminished over time. They also gave as an example the Latin letter A, which ultimately derives from a picture of an ox head.

    Fraktur letters, especially the capitals, are more complex in that regard than their regular Latin counterparts. Eg. Fraktur A requires 4 or 5 strokes instead of 3; Fraktur P, Q, T, V have 3 strokes rather than 2.

  25. Rather I think they are related.

    I think at first there were beautiful initials and then they began using “beutiful” letters more often. But the amount of beauty you can invest in each letter depends on its frequency.
    I am not sure here.

    But my point was that complicated Fraktur capitals like 𝔜 (Y) are cheap in terms of your efforts because they are printed. You pay for cutting them, but you do not have to reproduce this shape with your hand each time in every Noun when you are copying yet another boring Latin grammar. Their extra curves may have the same nature as extra curves of initials.

  26. @drasvi
    “cheap in terms of your efforts because they are printed” Huh? Fraktur / Gothic was also written by hand.
    If you’re interested, there’s heaps of handwriting manuals on google books that show you how.

  27. @ jack morava

    Not just transuranic. Even elements like hydrogen (氢 qīng) or fluorine (氟 ) were, as far as I know, created for the purpose of having a single character to represent each element. Qīng uses the radical 气 , used for gases, combined with part of the character for ‘light’ (轻 qīng), indicating the lightness of hydrogen. Fluorine is fairly clearly a borrowing from a European language. If you can get hold of a Chinese period table, with pronunciations, you can often pick out the ones that were specially created.

  28. David Marjanović says

    Fraktur A requires 4 or 5 strokes instead of 3

    …or 2: ^ –

    Fluorine is fairly clearly a borrowing from a European language.

    …and the phonophore in the character, under the steam, is the Buddha ().

  29. Which Fraktur A?

    The vase-shaped simple thing that modern reader will read as U, but which looks prickly and is different from V in that V looks like B (and a bit like P)?

  30. David Marjanović says


  31. jack morava says
  32. That gives some insight into why Glagolitic was designed like this in the first place, and also why it was replaced by Cyrillic that soon.

  33. @ jack morava

    Unfortunately I’m completely snowed under at the moment, otherwise I could have done a better job (by actually looking up the Chinese periodic table, for a start).

    The naming of elements in East Asia is rather interesting and has been covered here at LH before. The Japanese used traditional names where possible but were forced to calque a large number of names (principally from German, e.g., 水素 suiso from Wasserstoff) or borrow the name directly, also generally relying on German-style pronunciations (e.g., ベリリウム beririumu for ‘beryllium’, チタン chitan for ‘titanium’).

    This approach to naming was adopted into Korean, but Jeongsong Park has indicated that the Korean education system has in recent years abandoned Japanese-style transliterations and switched over to transliterating from English as being more ‘international’. (Perhaps they also decided they’d had enough of Japanese-influenced names…)

    The Chinese eventually adopted a different tack. A decision was made that all elements were to be given single-character names. To do this, new characters generally needed to be created, using semantic radicals appropriate to the element, i.e. 气 for gases, 金 (metal) for metals, 石 (rock) for non-metallic solids. With over 100 elements and a limited number of syllables available in the Chinese inventory, you can appreciate that some elements ended up with the same pronunciation, differentiated only in writing (I’m writing from memory; I don’t have time at the moment to check this).

  34. The naming of elements in East Asia is rather interesting and has been covered here at LH before.

    Names of the Elements in Chinese (2015).

  35. This Quora answer might be useful for understanding the Chinese names of elements.

  36. Fraktur capital A: 1 stroke on the left, 2 on the bottom and 1-2more on the right.

    @V: Glagolitic “replaced” by cyrillic:

    Well, it depends on *where* you’re talking about.

    Anyway in Croatia glagolitic has a cursive version which made easier to write. But the bookhand went through other changes of its own. Eg. The old ’round glagolitic’ cruciform A gained an extra 2 strokes.

  37. Trond Engen says

    David E.: To be slightly fairer, they seem surprised that it happened so rapidly. But how rapidly would you have expected it to happen, exactly?

    Well ,I’ve never seen an alphabet simplify before, so …

  38. Well ,I’ve never seen an alphabet simplify before, so …

    Does anyone still say “common or garden…”? I dearly wish they would.

  39. @ Y

    “Common or garden” is out of date. What might once have been referred to as “common” in biology has probably been renamed “European” or “Eurasian”, or something else, in order to rectify the Anglocentrism of species names.

    Best Regards,


  40. AFAIK the expression spread out of biological nomenclature, to mean ‘common’ in general. I believe it’s the British equivalent of U.S. “garden variety”.

  41. David Eddyshaw says

    I say “common or garden.” Of course.

  42. I’m glad to hear it. Don’t stop!
    Would you ever tell someone, say, “what you have is common or garden macular degeneration…”?

  43. I say “common or garden.”

    As do I! Old fossils forever!

  44. David Eddyshaw says

    Would you ever tell someone, say, “what you have is common or garden macular degeneration…”?

    Certainly not. It is very bad form to suggest to a patient that his/her disease is of no great interest to a physician as experienced as oneself.

    However, I might well say to a colleague that I thought something was common or garden macular degeneration (in line with the fact that unusual presentations of common diseases are commoner than rare diseases, even in medical retina work, where you see one or another rare disease all the time, because there are so many of them.)

  45. I was a little flip in my choice of ailments. Macular degeneration is nothing to get light-hearted about. But some doctors might use that kind of language when referring to more minor ailments. I imagine hearing, “You are not having a heart attack. What you have is common or garden indigestion.”

  46. @zyxt, to avoid confusion: I was responding J1M. He quoted me, but your posted your comment as I was typing mine answer. (Maybe there is no confusion, but just in case).

    I was comparing printed Fraktur books to manuscripts of boring textbooks made before.

    I use a noun “book”, because writing a letter is not the same as copying a whole long book. It is a lot of work (consider doing this…). I use “boring grammar text[book]”, because Textura/textualis is named after THE “textus”. The Bible. It is a liturgical hand. There is a reason to expect other books to be simpler. But who knows. St. Gallen’s Priscian (one of the important sources on Old Irish) has beautiful initials, as it happens to Irish books.

  47. Modern letters (paper notes you sent to someone) can have very complex shapes compared to medieval manuscripts. Even Russian handwriting as it is taught in school is complex.

    But these shapes can be produced easily with a pen. Pens provoke it. The movement is pleasant, the result is pleasant… bored schoolchildren sometimes make such patterns as they are listening to the teachers. E.g. 888888-like pattern (cf. 𝓰).

    Compare also this:

    When such curves became common, people used a quill, not a pen, so I do not understand why that happened: quill is old. Why medieval books have thick (not thin) lines and no such curves?
    Maybe medieval ink was different?

  48. David Marjanović says

    It depends on how you hold the quill, and how you cut it…

    Curves like those in the signature were deliberately avoided to keep the letterforms standardized and legible (…assuming the intended readers knew all the abbreviation symbols and didn’t lose count of the mınıms).

  49. Yes, but for several hundred years (I am not sure since when: here they appear in 18th century) they are characteristic for “beautiful” European handwriting.

    Maybe another factor is “cheap paper”, by the way.

  50. This may be the most disturbing set of LH comments I’ve ever read.
    Politics & Script by Stanley Morison, & probably Bernhard Bischoff’s Latin Palaeography (translated by Daibhm O. Cróinin & David Ganz, but some of you can read it in the original, I am sure) will sort you all out. I couldn’t begin to without getting everyone into a room with the contents of my photo files, a projector, & a lot of quills for about a week. It would be fun!

  51. This may be the most disturbing set of LH comments I’ve ever read.

    My work here is done!

  52. David Eddyshaw says

    Not at all! Excelsior!

  53. @zyxt: Glagolitic surviving in Croatia.

    Your point is? I know it did. I know it developed cursive forms. I was talking about the Slavic-speaking (and writing) territories that had constant contact with Greek-speaking (and writing) ones.

  54. @drasvi
    No wuckers cuz 🙂

    Chill cuz

  55. @David Eddyshaw: I seem to recall we’ve had this discussion before, about sharing with your patient rather than with fellow practitioners.

    @Amanda Adams: is that what put you off? It did put me off myself.

  56. John Cowan says

    I assume, “the letter б” is the space of shapes that are б OR an algorythm that we think (incorrectly maybe!) exists in your brain that makes brain say “yes, this image is an instance of б” rather than a pixel-precise image of a person who bent her leg and arms.

    Quite so. And this, by the way, is what the frequently-dissed-on-this-blog Hofstadter (not Iscariot) was actually interested in. His truly (as opposed to culturally) important book is Fluid Concepts and Creative Analogies (not all of which is by him). It was D.H. who introduced me to the Stop font, which combines the virtues of ugliness with those of unreadability. Capital A is always closed at the top, right?

  57. David Marjanović says

    That looks a lot like the Star Wars font!

    Capital A is always closed at the top, right?

    Not in Fraktur, where it’s almost U-shaped.

  58. Majuscule A (long before there even was a minuscule version) was originally closed at the bottom and open at the top.

  59. John Cowan says

    I’m now consulting with a graphic-design student who wishes to develop a syllabary for Ga-Dangme visually based on Egyptian hieroglyphs (apparently the Dangme have a tradition that their people migrated from Egypt). We began by discussing the script direction. His original intention was to have it read right-to-left because leftwards is the direction of increasing honor in Ga-Dangme culture, but I pointed out that African right-to-left scripts are generally used by Muslim peoples. I then explained the difference between syllabaries and abugidas, and he concluded that he preferred an abugida. It isn’t yet clear to me whether there is a natural implicit vowel, as in Brahmic scripts, or not, as in Ethiopic ones. His intention is to use diacritics for nasalization, labialization, and tone, which go unrepresented in the standard romanization.

    I’ll update the Hattics as things develop.

  60. Trond Engen says

    Udresfeabifen, Norway’s oldest newspaper.

  61. David Eddyshaw says


    According to (the late) Prof Kropp Dakubu, Ga and Dangme are not mutually comprehensible (though they certainly are very similar.) She cites a lexicostatistical study (FWIW) that suggests that they began to separate about a thousand years ago.

    Lots of people in those parts like to think that their forebears came from Egypt. I suspect that some of this age-old tradition is recently imported from America:

    I think that the Ga-Adangme tradition itself talks of an origin around Lake Chad (also, as it happens, the supposed origin of (the grandfather of) the founder of the Mamprussi kingdom and its offshoots the Dagbamba, Mossi and Dagaaba. Lake Chad seems to have been a regular officina gentium.)

    The Akan prefer to nominate as Glorious Ancestors the ancient empire of Ghana (hence the name of the modern country.) Aesthetically, I’m with the Akan on this: if you’re going to have Glorious Ancestors, why not go with Glorious West African Ancestors?

  62. David Marjanović says

    Lots of people in those parts like to think that their forebears came from Egypt. I suspect that some of this age-old tradition is recently imported from America:

    That’s for the English-speaking side. The French-speaking side has the same thing, around the same time, in Cheikh Anta Diop.

  63. John Cowan says

    Ga and Dangme are not mutually comprehensible (though they certainly are very similar.)

    Indeed. From what I can make out, Ga has /ʃ/ and /h/ and the labialized phonemes /tʃʷ/, /dʒʷ/, /kʷ/, /ɡʷ/, /ʃʷ/, /hʷ/ that don’t exist in Dangme, as well as [ŋʷ], an allophone of /w/ before nasals that is romanized separately. In addition, Ga has three phonemic vowel lengths and coda consonants (nasals only), all lacking in Dangme. Per contra, Dangme has three tones, whereas Ga has only two. None of these are IMO significant barriers to a unified abugida. An interesting property of both languages is that all initial consonant clusters have /l/ as the second consonant, and that in Ga at least this can bear a tone distinct from the main tone.

    I’ve found Kropp’s thesis and will be working my way through the phonological parts of it.

    Here’s what the student sent me on the tradition: “The Gā-Dañgme started off in Egypt in a city called Goshen. Then they began to leave Egypt during 672-525 B.C. On that journey along the side of the Nile River, they crossed Sudan to Ethiopia, and moved west into Yoruba land Nigeria.”

  64. David Eddyshaw says

    Kropp’s thesis

    Good find. I’d missed that (partly because Ga/Dangme is far from my own area of particular interest.)

    Kropp Dakubu was responsible for a great deal of high-quality linguistic work over the years, touching on many Ghanaian languages (she wrote the best existing grammar of Farefare, for example. In French.)

    I gather she was also a very helpful correspondent with interested amateurs on questions of language.

  65. Ga has three phonemic vowel lengths

    The Estonian of Ghana?
    Where is this discussed in Kropp? I haven’t found it.

    In Kropp’s isogloss map, one of the two “features peculiar to Gã” is the word for ‘pineapple’, blɔfóŋmè. It seems odd choice.

  66. David Eddyshaw says

    Kusaal has three phonemic vowel lengths too, though you never get the full three-way contrast in identical environments, so you can magic it away with a bit of judiciously applied gratuitous abstraction if you want to, and reduce it to a respectable non-Estonian two.

  67. I couldn’t find it discussed in your grammar, either.

  68. David Eddyshaw says

    Pages 7-8.

    The only three-mora monophthong is in ma’aa “only”, which has a glottalised three-mora /a/; the exception is due to the fact that this word was stil *maɣaa up until about 1960. Otherwise, in contexts where you get three-mora diphthongs, you get an ordinary two-mora long vowel. Diphthongs, though, can have one, two or three morae. With just a handful of exceptions (like vuaa plural of “red kapok fruit”) three-mora diphthongs appear only word-finally before “prosodic enclitics” (p26.) You could thus get rid of most of them by calling this a prosodically-induced prologation of an ordinary two-mora diphthong, but this would involve ignoring the obvious common thread in how words are affected by these clitics (the clitics in fact suppress the usual apocope of an underlying word-final vowel mora.)

    The shortening of three-mora monophthongs must be a fairly recent development, though, as the tone system hasn’t quite caught up with it. As I describe tone in my grammar, a vowel, short or long, carries just one tone, but overlong diphthongs carry two. The two-mora vowels which are contracted from three-mora sequences also only carry one tone, but they break some of the otherwise neatly applying tone sandhi rules because the contraction, historically, was fairly recent (p26, on the tonal behaviour of da’a “market”, which comes from proto-WOV *dà̰:gá); because there actually aren’t all that many words like this, I scoot over it rather lightly in the grammar. Blink and you’ll miss it.

  69. David Eddyshaw says

    (Page numbers refer to the latestest version of the grammar on academia,edu: it’s a page shorter than previous versions because I’ve taken out the pretty map of the Oti-Volta languages now that I can refer teh hardcorz who care about such things to Toward proto-Oti-Volta.)

  70. Section numbers are good, too. They don’t depend on page numbers, and they give you an excuse to use the word silcrow (says WP, quoting Monotype).

    Are the versions on Zenodo and a.e synchronized?

  71. David Eddyshaw says

    I update the Zenodo versions monthly: as Zenodo archive all previous versions, it seems ungrateful to chew through their storage by updating for every minor edit, whereas I have no scruples about doing that for the academia version. So they’re only guaranteed to be synchronised on the first of the month.

    I don’t envisage any major changes for the Kusaal grammar now, but TpOV needs a lot of work yet. Not least, correction of sheer careless slips: but quite apart from the fact that I hope to make substantial progress yet on things like tone, the actual exposition could do with improvement in a lot of places. I originally wrote it for my own benefit, and I too often assume that things which are very familiar to me personally are somehow intrinsically obvious and don’t need elaborate explanations. Not so … (furthermore, you’re deluding yourself if you think you understand something yourself when you can’t explain it lucidly to someone who doesn’t already know it.)

  72. *maɣaa

    Surely you don’t need the Hypothetical Asterisk on a form that was still current in living memory. Abduce ‘abduct’ is last recorded 1920 by the as-of-2011 OED, and it is labeled archaic, but I would hardly write *abduce.

    you’re deluding yourself if you think you understand something yourself when you can’t explain it lucidly to someone who doesn’t already know it.

    Up to a point, Minister. Many of the mathematical articles on WP are incomprehensible to me, but I don’t think that means their authors don’t understand them, it just means they don’t think it appropriate to provide the amount of background explanation that I would require.

  73. He didn’t say “when you don’t explain it lucidly,” he said “when you can’t explain it lucidly.” (Emphasis added.)

Speak Your Mind