A Facebook post by our old pal Slavomír Čéplö/bulbul pointed me to Ahmad Al-Jallad’s All lines lead to Proto-Arabic: a review article on Jonathan Owens, Arabic and the Case against Linearity in Historical Linguistics (Bulletin of the School of Oriental and African Studies [2025], 1–16), in which Al-Jallad (seen here in 2018 and 2022) does a thorough demolition job on Owens’ book, which sounds like a classic case of an expert in a limited field trying to extend his expertise too widely (cf. John McWhorter). It begins (I omit the footnotes):
The past century or so has witnessed what one might call a “documentary turn” in the study of Arabic’s history. The full range of modern Arabic’s dialectal diversity came into focus as linguists began to produce descriptions of peripheral dialects, spanning from Central Asia to the Yemeni highlands and from Cypress to Chad. Sociolinguistic approaches to the dialects have advanced our understanding of language change and dialect formation in real time. The epigraphic exploration of Arabia revealed a “Jāhiliyyah” with stunning linguistic diversity, even when compared to the rich materials compiled by the Arabic Grammarians. Contrary to the commonly held belief, Arabic was not alone in Arabia, but was rather a part of a rich linguistic landscape, lost to history until recently. The discovery and study of papyri from the early Islamic period afford a unique view into the written register of Arabic before the rise of the grammatical tradition, and both pre-modern Christian and Jewish Arabic materials attest to writing traditions that existed parallel to normative Classical Arabic, and shed valuable light on the pre-modern dialectal landscape. The combination of these new sources of data and approaches have rendered the traditional view of Arabic’s past obsolete, and so the time is ripe to synthesize this material into the writing of new linguistic histories of Arabic.
The work under review is the latest monograph by Professor Jonathan Owens, a renowned authority on the Arabic dialects of Nigeria, Libya, and Chad, who has made significant contributions to the field of Arabic sociolinguistics and dialectology at large. Owens should be congratulated for the great effort put into this work, which spans over 500 pages. In this book, he builds on the case made in Owens (2006/9) that the field of Arabic historical linguistics has been fundamentally misguided, giving undue weight to older attested stages of the language when it comes to reconstruction. Here, he attempts to build a new, “non-linear” paradigm with a focus on the history of the modern Arabic vernaculars, but draws also on other sources such as epigraphy and papyri. […]
This model, however, is not necessarily wrong. Proto-Semitic, as any other Proto-language, is the hypothetical, reconstructed ancestor of all attested members of the Semitic language family. Every Semitic language must descend, by definition, from Proto-Semitic; Old Arabic is no exception. Yes, there are intervening stages, splits and sub-groupings – no Semiticist, to my knowledge, has ever claimed that Arabic was an independent branch of Proto-Semitic – but as a simplistic model, this illustrates an uncontroversial fact, not only in comparative Semitics but in the field of historical/diachronic linguistics in general. Now, his next issue seems to be with understanding Neo-Arabic, a concise way of saying modern vernacular Arabic, as a descendant of Old Arabic. It is hard to understand what the problem here is either. Old Arabic, at least the definition used by Van Putten and me, is not a single attested variety, but rather a chronological term that refers to all varieties of Arabic spoken/written prior to the rise of the Medinan state – we sometimes have evidence for these varieties in writing but the vast majority of the linguistic diversity of this period has been lost to history. Since Old Arabic encompasses all Arabics prior to, say, the mid-seventh century ᴄᴇ, it is obvious that the modern vernaculars are later, changed forms, in other words, descendants, of these varieties, whether attested or otherwise. To my knowledge, nobody has ever claimed that the modern Arabic dialects descend linearly from any attested Old Arabic epigraphic variety. If Old Arabic here is meant to substitute for Classical Arabic, it is safe to say that such an assumption has been long abandoned, and most recent work in Arabic historical linguistics does not operate within such a model. A final way to interpret this model would be that older attested forms of a language are by default more linguistically archaic than later attested ones. But even this is not a position any professional linguists hold. No Semiticist, for example, has ever argued that Akkadian phonology is more archaic than Modern South Arabian because it is attested over four-thousand years earlier. Indeed, it has been long recognized that Arabic is more conservative than Hebrew and Aramaic in terms of nominal morphology despite its being attested later. Indeed, there is interesting work being done on how modern vernacular Arabic can be in some respects more archaic than normative Classical Arabic. As such, the objections here do not seem to be directed at any real positions held by contemporary Semiticists or Arabists.
There is no overarching narrative guiding the book’s argument. Instead, the subsequent parts consist of a series of correctives and case studies, of varying detail, some of which are new and others previously published, on pre-modern forms of Arabic and the modern dialects. These studies are meant to show how Arabic is a “composite”, “non-linear” language. Owens understands this and is “unapologetic” for not developing a history of Arabic, which he seems to consider to be impossible because, according to him, “Arabic language history is inherently contradictory of classic comparative historical linguistic concepts” (p. 432). I would maintain that Arabic is, in fact, a normal human language and there is nothing inherently exceptional about its development, but rather it is the present methodological approach to its history that is responsible in large part for this apparent conundrum, as we shall see throughout the course of this review article. Since the monograph lacks a unified structure and does not argue the thesis presented in its introduction in a coherent way throughout the work, the only way to engage with it is to scrutinize the case studies, their data and argumentation. While some of these studies offer some interesting insights regarding the development of individual Arabic dialectal features, those that go beyond this are often highly problematic, both in terms of facts and method. The remainder of this review will focus on a number of these critical issues, ultimately demonstrating that Owens’ attempt at re-imagining Arabic linguistic history is based on a faulty foundation and is ultimately unsuccessful.
I don’t know much about the history of Arabic, but it’s a pleasure to read someone who knows what he’s talking about, even if I sometimes get lost in the details. That quote “Arabic language history is inherently contradictory of classic comparative historical linguistic concepts” made me roll my eyes hard. And I like the fact that Al-Jallad gives credit when appropriate:
While many of Owens’ unconventional assumptions are certainly open to debate, the data he presents are of much interest and will be a good starting point for future discussion and elaboration. It is important to emphasize that none of these case studies undermine Arabic’s classification as a Central Semitic language, nor do they challenge a conventional understanding of language development. The final part of this section addresses the issue of speech community and its relevance to historical reconstruction, with a lengthy and interesting case study on the varieties of Arabic spoken in Nigeria and the Lake Chad area.
Those varieties are, of course, Owens’ specialty; ne sutor ultra crepidam!
I am not a linguist, but I like to read popular books about language, including those of John McWhorter. This post briefly mentions him, in a way that suggests disagreement with McWhorter’s ideas. Ought I to steer clear of his stuff?
He writes well and is worth reading as long as you remember that his specialty is creoles, and when he gets beyond that and starts pontificating about wider linguistic matters you should add increasing amounts of salt.
Yay, open access!!!
I see the proofreading was done by autocorrupt.
Cypress, California might well have a few Arabic speakers. And a few Chads.
On creoles, too, McWhorter has some ideas which are controversial (i.e. most others don’t agree with him, because they are not good ideas.)
I [@Hat] don’t know much about the history of Arabic, … then perhaps I’ll be excused for asking a dumb question …
Is “Medinan” there code for the rise of Islam in general? Or does it specifically mean First Islamic State 622 – 632? From a historical linguistic point of view, that wikip says
I understand the importance of Medina politically/religiously. But why linguistically? Did Medina somehow ‘terminate’ Old Arabic? Or ‘fix’ some new form that persisted albeit only orally at first?
What became of all the varieties of vernacular Arabic as at 622?
Did Medina somehow ‘terminate’ Old Arabic? Or ‘fix’ some new form that persisted albeit only orally at first?
None of the above …
Pending a better answer from Lameen and other Hatters who Actually Know:
Classical Arabic is a puristic and regularised version, codified by Muslim grammarians, of a literary koine that antedates the composition of the Qur’an. So, despite the fact that Classical Arabic stricto sensu arose quite a bit later than the beginning of Islam, it is (paradoxically) more archaic in some respects than the language the Qur’an was actually composed in.
The particular dialect of Medina (and of the Qur’an) was different from Classical Arabic in several respects; for example, it had lost non-initial glottal stops, and perhaps some aspects of the case system. The Qur’an text was subsequently vowelled and provided with various other diacritics to enable a Classical Arabic reading; but, because the consonantal text was not tampered with, and its orthography was taken as definitive for Arabic in general, this introduced some complications which have survived to annoy learners right up until today.
I just recently discovered that the diacritic dots which distinguish so many Arabic letters from one another were effectively optional for a long time:
https://en.wikipedia.org/wiki/Rasm
(I’d heard of rasm, but didn’t realise that it was once basically the default way of writing Arabic.)
As I was probably supposed to learn in Hebrew school, diacritics are for the weak.
Examples abound of Israeli radio and TV announcers wingin’ it with unpointed foreign names. One example I heard about, dated to the late ’60s or early ’70s: “This was Bolero, by Maurice Roval — pardon, Revol.”
(Ravel is spelled רוול as was Rowal, pron. Roval, a famous Tel Aviv café named for its founders, Rosenberg and Waller.)
Classical Arabic …
Thank you for trying to explain. But I’m now more confused.
The term Al-Jallad uses/I quoted is “Old Arabic”. “Classical Arabic” is something else ??
The particular dialect of Medina (and of the Qur’an) …
The Prophet was born and spent his formative years in Mecca, only getting to Medina around age 50. He was there less than a decade before re-conquering Mecca.
The Qur’an text was revealed incrementally from when the Prophet was aged 40; only part of it when he was in Medina. Then wouldn’t it be in Mecca dialect (if that’s a thing)? What does Medina dialect have to do with it? (wikip on Classical Arabic doesn’t mention Medina.)
I thought the text was sacrosanct as the words of the Prophet hisownself?
Classical Arabic stricto sensu arose quite a bit later than the beginning of Islam, it is (paradoxically) more archaic in some respects than the language the Qur’an was actually composed in.
So …? Scholars hijacked the vernacular (oral?) text of the Qur’an; took it as a draft; back-cast it into some dubiously attested archaic form?? Something like casting the language of the KJV in some deliberately archaic form?
the diacritic dots which distinguish so many Arabic letters from one another were effectively optional for a long time:
I thought that was the norm for Semitic scripts(?) I remember from travelling in Israel, the public signage used unpointed forms most of the time. Entertaining.
It sounds like one of Owens’ issues is with a shorthand or conflation that you also find in the history of English. We say (accurately) that modern English descends from Old English (unless you’re Emonds and Faarlund, anyway). And in citing Old English, we tend to turn to the best-attested dialect, which is also what we teach to students when they take a class on Old English: West Saxon. But West Saxon is *not* the ancestor of any modern standard variety. An easy example is the word for “cold”, which was *ceald* in West Saxon, with an initial palatalized affricate [tʃ]. The modern word comes from Anglian *cald*, with the (more archaic) velar left unchanged.
The actual history of the language is one thing, variation in attestation and dialectal prominence quite another. There are very often gaps, and not only in prehistory, where you have to do a bit of linguistic reconstruction (either into a complete vacuum, or to flesh out only very sparse dialectal material) to get a linear history. It doesn’t mean that that linear history is wrong, though.
And of course dialect mixing and contact and all that are real too, meaning that the simple lines on a tree are simplifications. But I’ve never understood why that’s a bad thing. The whole point of a model is to be an abstraction that can help you understand a messier reality. Some people seem to think that a model that isn’t fully as complex as reality is flawed and should be jettisoned. Which I guess is an internally consistent position, but I wish those people the best of luck in wrapping their minds around all aspects of reality in all fields.
“All models are wrong, but some are useful” — Karl Lagerfeld
Some people seem to think that a model that isn’t fully as complex as reality is flawed and should be jettisoned.
Chapter XI
It sounds like one of Owens’ issues is with a shorthand or conflation that you also find in the history of English.
Precisely. There’s a vaguely Oedipal urge in Arabic historical linguistics to make a big point of demonstrating that Fusha isn’t the ancestor of the dialects after all. Well sure, it isn’t, in the same way that West Saxon isn’t the ancestor of modern English; there are several points of detail where we can show that the dialects happen to preserve a form older than Fusha. But it’s a much better approximation to that ancestor than any otherwise well-attested variety modern or ancient – and, in many respects, it’s even a better approximation to it than anything you could get by painstakingly reconstructing proto-Arabic from modern varieties.
I thought that was the norm for Semitic scripts
Leaving out (short) vowels, yes. But Arabic uses dots to distinguish different unrelated consonants as well; b from t, f from q, r from z… Without those, a text can get very ambiguous indeed.
Scholars hijacked the vernacular (oral?) text of the Qur’an; took it as a draft; back-cast it into some dubiously attested archaic form??
As far as I’m aware, no one has attempted to identify dialectal differences between 6th c. Mecca and Medina (though there are stylistic differences between Meccan and Medinan suras); the term van Putten uses for the language of the text is “Old Hijazi”, covering the whole region. The extant reading traditions all agree with the consonantal text, but differ in details of pronunciation, such as which glottal stops are realised as such (and, once in a long while, in dotting). This has traditionally been interpreted as reflecting differences among the dialects of the Companions who transmitted this (such differences in pronunciation being explicitly endorsed as acceptable by a hadith); van Putten interprets them slightly differently, as reflecting aesthetic decisions rather than purely dialectal variation, and argues that the realisation of short final vowels (marking case/mood) reflects influence from the later (but more conservative) Classical Arabic standard. (On the latter of these points, I don’t find his arguments compelling.)
But Arabic uses dots to distinguish different unrelated consonants
As does Hebrew: dots distinguish b from v, p from f. פלפל: falafel or pilpel (pepper). (Ok, you don’t need get very close to a veggie stall to realise it’s not a falafel stand.)
Your “unrelated” is doing a lot of work there. Only somebody who already speaks Hebrew would think that b, v are related.
Anyone who does speak Hebrew can’t miss the relation between b and v or p and f; if they did, they wouldn’t be able to conjugate their verbs. Any connection between, say, b and t is no less obscure to Arabic speakers than to non-speakers (because the relationship is purely accidental – two letters that originally looked completely different ended up the same in excessively cursive handwriting.)
Thanks, Nelson and Lameen — extremely helpful comments!
To call Arabic a ‘composite’ and ‘non-linear’, language makes sense as it makes sense to say that it contradicts ‘classic comparative historical linguistic concepts’. The lame objection ‘Arabic is, in fact, a normal human language and there is nothing inherently exceptional about its development’ says nothing. A language can be both a normal human language (as opposed to, say, an arbitrary utopian-linguistic construct like Esperanto or a philosophical-linguistic construct like Ithkuil) and very, very extraordinary in some respects, and Arabic is precisely that.
Owens has the expertise to show and does show that Arabic evinces traits characteristic of different Semitic sub-branches (e.g. South Semitic and Northwest/West Semitic), making it a ‘composite West Semitic language’. That implies that various influences (genetic, contact, substrate, dialect mixture) combine in both pre-Islamic and post-Islamic era Arabic, resisting classification as a single linear line of descent from a uniform ancestor.
In some domains (e.g. certain morphosyntactic structures, discourse-pragmatic patterns), Arabic seems strikingly stable across centuries and vast geography. Other features – especially among dialects – show dramatic variation and independent developments or contact-induced change. Owens argues that both patterns must be part of the historical picture. This mixture – parts stable, parts evolving differently across communities – doesn’t fit cleanly into a ‘tree-model’ of linguistic descent. Instead, it resembles a network of interacting sub-communities, where divergence, convergence, diffusion, innovation, and conservation all play simultaneous roles.
Owens rightly warns against oversimplifying Arabic’s past by focusing only on a few ‘attested stages’ (e.g. old epigraphy, Classical Arabic, modern standard). Doing so ignores the complex interplay of dialects, written vs spoken varieties, literacy traditions, and social usage. Historical linguistics must pay attention not only to change but also to continuity and contact, to see how features survive, shift, or spread – which conventional linear models often ignore. In other cases this matters little because the evolutionary mechanisms were simpler: change violently does away with continuity. But in the case of Arabic change doesn’t displace continuity but merges with it in unexpected and atypical ways.
In sum: Arabic’s humongous geographic spread, long history, extreme dialect and genre diversity, and evidence for both conservation and innovation across different varieties make its evolution more like a fabulous tapestry than a less impressive ‘family tree.’ That’s why Owens labels it ‘composite’ and ‘non-linear.’
Such a language is truly extraordinary and could be said to carry a divine spark, exactly what the ancients called ‘genius’. Gabriel did well to channel his divine message into the medium of Arabic. Glory to Arabic, glory to God.
I’m tempted to delete your typically pompous, ignorant, and belligerent comment, but I’ll leave it as a sample of your style. But please let it be and don’t continue this discussion, which is outside your realm. I don’t want a good conversation derailed by your bullshit.
If one focuses on any given subset of current Arabics (wikipedia tells me it’s now conventional to lump the dozens of variants into five larger regional clusters), is there a relatively established conventional narrative for each of them for how that subset got to be the way it is? They could be different from each other (probably not much Berber-substrate impact on Levantine varieties …) but if there are let’s say five different coherent stories for five different regional groups the fact that they aren’t exactly parallel to each other and don’t combine into one simple macro-story with variations on a theme shouldn’t be troubling.
Owens’ King Charles’ head seems to be the idea that all forms of modern Arabic descend from a language or languages that never had the final short-vowel case or mood endings in the first place.
It reminds me of the idea (which I have seen seriously proposed) that Akkadian divine names (which lack case endings) preserve the original state of affairs, and that the case endings of normal nouns are therefore a later innovation. Presumably those clever Akkadians were pioneers of comparative linguistics, which would explain how their choice of invented endings recreated the proto-Semitic vowel qualities. (Alternatively, the divine names are older than proto-Semitic.)
Lighting the Étienne signal – Romance seems similarly complex: huge spread, innumerable dialects that keep interacting with their neighbors and with regional prestige variants and also innovate in unpredictable directions.
Day saved.
Your “unrelated” is doing a lot of work there. Only somebody who already speaks Hebrew would think that b, v are related.
You find labiodentals unrelated to bilabials? I think my Hebrew school classmates (as far as I could tell) and I just had to be told once, and from then on it was obvious. The same is true for k and kh (which I learned as what Wikipedia calls a voiceless uvular trill, [ʀ̥], speaking of dots). On the other hand, “b” and “t” strike me as unrelated.
Oh, that – that the case endings, which are largely redundant, were borrowed into the poetic register from Akkadian. Here’s a convincing rebuttal.
(Also read the rest of Al-Jallad’s Academia page if you have 30 hours per day to spare.)
Owens’ King Charles’ head
Owens is an excellent and insightful linguist in most domains – his work on Nigerian Arabic is particularly worth reading – but the fatal mistake that afflicts his historical linguistic work is his rejection of the possibility of accidental parallel developments (“drift”). Applying the same approach to Indo-European would lead us to the conclusion that, since English, Spanish, and Persian all lack case marking on nouns, there must have been at least some PIE dialects that did too.
Romance seems similarly complex
And Iranian, and Berber, and Chinese, and… really quite a lot of larger families.
And then what? These dialects all copied loss of casemarking from each other? Or would that be “drift”, so they have to have a single common ancestor, and therefore English, Spanish and Persian have to form a branch of IE that does not contain Baltic or Greek or German?
@LH
Thank you for being so merciful. Can I please respond to what Lameen wrote? I’ll keep it as short as possible so as not to irritate you.
Owens has addressed Lameen’s specific accusation many times. His response, in essence, is not to deny that drift exists, but to say that his critics use ‘drift’ as a black box to avoid confronting patterns that are too systematic, too widespread, and too selective to be explained by coincidence.
From the (excellent) paper that DM linked to:
Applying this principle …
In Oti-Volta languages, trees belong to one particular noun-class pairing (“gender”), with clearly cognate class suffixes across the family. However, there are languages in which they belong instead to the gender of which the prototypical member is “child.” *
Specifically, these are: all of Western Oti-Volta except Nõotre; Konni (closest relation, Buli); the Gurma language Moba (closest relation, Gulimancema); and the Eastern Oti-Volta language Byali (closest relation, Mbelime.)
These exception languages are mostly geographically adjacent, but Byali is not. They belong to all three of the primary branches of Oti-Volta. Accordingly, proto-Oti-Volta both did, and did not, have a “tree” gender. QED.
* This actually is genuinely odd, especially with Byali, where diffusion seems impossible as an explanation. It’s not that the “tree” gender has been lost altogether in these languages (it hasn’t), and its members more or less randomly assigned to other genders. Why “child”?
In fact, it’s distinctly odder than the bog-standard loss of final short vowels, seen in such exotic languages as English.
@Y: I always assumed the Tel Aviv cafe was called Ravel…
>Presumably those clever Akkadians were pioneers of comparative linguistics, which would explain how their choice of invented endings recreated the proto-Semitic vowel qualities. (Alternatively, the divine names are older than proto-Semitic.)
Perhaps it’s wildly unlikely that the divine names preserved older case endings for a variety of reasons. But the idea that such a theory would require those names to be “older” than proto-Semitic seems wrong to me. I would think they would merely need to be as old, or even younger, coming from any language or dialectic older than attested Akkadian that kept the endings. And divine names in the setting of a monarchic religion do seem somewhat more likely to carry conservative forms lost elsewhere. One does say Dios mio in Spanish without raising doubts about one’s monotheism.
(Well, the stress doesn’t work for a plural either, but you’ll get the point that it conserves an ending, though I also need to concede it doesn’t conjugate.)
At any rate, I thought I’d make a plug for Arabs and Empires before Islam, mentioned in al-Jallad’s article and available free from the Internet Archive.
My own theory (which is mine) is that Akkadian lost all original final short vowels in non-clitics, and the divine names were originally diptotic, and thus lacked the mimation which preserved the case endings of common nouns.
I ran this idea past Benjamin Suchard once, who liked it; though he said that there is disagreement about whether Akkadian verb forms (for example) have lost final proto-Semitic short vowels, or never had them in the first place.
(Thanks for that link. Looks interesting.)
And then what?
Obvs: Proto-Germanic, Proto-Indo-Iranian, and Proto-Italic all had conservative caseless dialects coexisting with innovative case-marking ones, just like Proto-Arabic and proto-Semitic both did in Owens’ scenario. Super-stable sociolinguistic variation.
divine names
A lot of Akkadian divine names were loanwords, right? Not so unusual for loanwords to be invariable in otherwise consistently case-marking languages, and that might have been copied to inherited divine names by analogy. Though I think I like David Eddyshaw’s idea better.
Why “child”?
Something to do with the usual West African colexification of CHILD = FRUIT, maybe? There are several Berber and Arabic varieties where the feminine gender of a plant species name yields both its tree and its fruit.
No wonder I forgot it! (I read the paper I linked to, but years ago.)
@Lameen
There was never a single, uniform, isolated ‘Proto-Arabic’ language the way the classic family-tree model imagines it.
Yes, true proto-languages in the classical comparative sense have existed and they are the norm but for them you need:
First, a relatively cohesive speech community – people interacting mostly with each other. Second, limited dialect mixing – enough isolation to develop a recognizable shared system. Third, clear divergence afterward as people spread out and evolve new dialects.
That model works best for: small populations in one region, with limited long-distance mobility. That was not Arabia. Arabia before Islam was: highly mobile (nomads, trade routes, migrations), interconnected by tribal networks, multi-dialectal, even at small scale, influenced by Aramaic, Northwest Semitic, South Arabian, and local substrates, showing local innovations AND shared conservative features, diverse in script traditions, religious cultures, and social organization. That combination creates network evolution, not branching-tree evolution.
This means that features spread horizontally (tribe-to-tribe), not just vertically (ancestor to descendant) so there’s simply no clean break between ‘proto’ and ‘post-proto.’
I’ll keep it as short as possible so as not to irritate you.
The shorter you keep it, the less you irritate me.
CJ’s point about how spread-out the early range of Arabic (that we know about …) was reminds me that by contrast we know, because we have unusually good historical info, that the Urheimat of “Proto-Romance” was once quite compact indeed – not even taking up the entire current territory of Latium/Lazio. But I wonder how many other situations there are where we actually have good enough data to know with any confidence how compact or non-compact the range of any of a long list of proto-languages actually was. Even the fairly modest stretch of (continental-side) North Sea coastline where proto-English first diverged from other Ingveonic tongues might have been less compact than Latium?
Something to do with the usual West African colexification of CHILD = FRUIT
That’s a thought; and also Mooré bíiga and Kusaal biig can both mean “fruit” as well as “child” (though it’s not the usual word for “fruit” in Kusaal.)
However, the colexification doesn’t seem to apply across Oti-Volta as a whole, and Oti-Volta puts words for particular fruits in with “body parts” in gender assignment. Mbelime actually uses the same stem for “child” and “fruit”, but in different genders: biìkɛ̀, biìdè respectively.
As is typical of Volta-Congo genders, particularly larger ones, what I’ve called the “child” gender has more than just one semantic focus.
It’s specifically diminutive in several languages, in cases where the same stem also appears with other genders, but in Western Oti-Volta that’s no longer really salient. It is the other major gender beside the specific “human” gender for human-reference nouns (as with “Kusaasi”), and also includes a lot of words for animals, especially familiar and/or domestic larger animals. It also includes many words for tools, and in Kusaal, all deverbal Instrument nouns.
The “animals” semantic focus may be the key: trees are “animate” in the traditional worldview, certainly in Kusaal, and probably much more widely in those parts.
In fact, given that Kusaasi trees can be witches, and that the win (“soul” – kinda) of a tree can be a human being’s spiritual guardian, just like the win of a person can, there could even be a semantic assimilation to “human” (but not close enough to end up in the original specific “human” gender.)
David M.: Actually, Romance is simpler than most other language families in that Classical Latin clearly is the direct ancestor of all living and attested Romance varieties. “Proto-Romance” certainly was a daughter and no sister language of Classical Latin.
Ryan: Spanish “Dios” is not a conservative survival. Instead, it owes its final /s/ to the influence of Church Latin. That is why the phonologically expected descendent form, /dio/, survives in Judeo-Spanish.
David E.: There is one Romance language, Friulian, where /frut/ (from FRUCTUM) is the word for “child”. Studies on the role played by a West African substrate in the Eastern Alpine area might prove a most…fruitful (No, of course I couldn’t resist*) area of scholarly inquiry. Feel free to share the idea with any Africanists of your acquaintance.
All: There seems to be this widespread notion that early work in historical linguistics involving the family tree model somehow abstracted away all variation and dialect diversity and consistently postulated uniform proto-languages. This is emphatically not true (indeed, much sociolinguistic and dialectological work by many of these early historical linguists is to my mind much better than what most sociolinguists produce nowadays).
In his “Le slave commun” (1924), to quote just one example, Antoine Meillet cautioned his readers that language change takes place gradually and spreads even more gradually over different social and geographical varieties of a given language, all too often over a prolonged time period. Thus, he concluded, one should never make the mistake of believing that, because a sound change can be presented today as a clear and simple rule, this tacitly implies that the sound change took place as clearly and simply in the speech community studied.
*To quote (once again, I quoted him once here at Casa Hat) a comic strip character, “Si l’envie bête d’y résister me prenait, je suis sûr que ça me rendrait très malade”.
Etienne,
>DIos… owes its final /s/ to the influence of Church Latin.
I’m curious the exact distinction you’re making. Was dios completely lost, then borrowed back by learned Spanish speakers?
Generally, I still think it’s the sort of thing I was suggesting. I’m not sure it mattered to my point whether it’s an influence that imposes conservatism in a certain register or an influence that allows for the restoration of forms and words after they’ve been lost in the vernacular. And for that matter, there must have been centuries for which it was difficult to say definitely that Spanish had completely pulled away from Iberian Church Latin or whether they were simply different registers.
Classical Latin clearly is the direct ancestor of all living and attested Romance varieties
How? I always understood the modern varieties to be descendants of Vulgar Latin, which was distinct from but coexisted with Classical Latin, the literary idiom of Cicero etc.
@DE
We touched on case endings in Akkadian proper names four years ago, in a post about Benjamin Suchard’s work no less. Tout est dans tout…
@Etienne
Variation exists in all languages but that’s not the same as saying that a language is not highly uniform i.e. compact. Compactness depends on the degree of uniformity which is essentially the degree of mutual intelligibility. All varieties of English spoken worldwide except AAVE and various creoles like Jamaican Patois are easily mutually intelligible though minor difficulties may exist. As such, English can be said to be highly compact despite the extreme geographic disparities of its native speakers.
And by the way there was no ‘Proto-Romance’ language distinct from Latin. Proto-Romance is Latin by definition. The distinct regional varieties of the Roman language conventionally called Romance only developed around the beginning of the Middle Ages when cross-regional communication and urban life declined precipitously as the Germanic warrior put a glorious end to Roman despotism*
*Sorry boss, I couldn’t help it.
this tacitly implies that the sound change took place as clearly and simply in the speech community studied
There’s a really neat illustration of the contrary in Farefare, where *k͡p *g͡b have become the plain velars /k g/*: the language now just lacks labial-velars altogether, but Rapp’s Die Gurenne-Sprache in Nordghana, from 1966, shows that the change had still not worked its way through the whole vocabulary: he still has e.g. kpenge “strengthen” (Kusaal kpɛ’ŋ) beside keema “elder” (Kusaal kpɛɛnm): modern Farefare kéŋé, kẽ́emá.
The differences don’t seem to be conditioned by anything obvious; possibly less-common words held out longer, but that’s just my vague impression from looking at the data.
*This change is also seen in Mooré, but it cannot be a shared common innovation (quite apart from Rapp’s data), because the Nankare dialect of Farefare, unlike the Gurenne, has palatalised original *k *g to /s z/ before front vowels, while /k g/ from original labial-velars are not affected: Ninkare si “millet”, ki “die”, Gurenne ki “millet”, ki “die”, Kusaal ki “millet”, kpi “die.”
Of course, to a European, this seems like an eminently “natural” change anyway, but /k͡p g͡b/ show no sign of being even a little bit unstable in many West African languages. I think it might be a diffusional or substrate thing.
Aren’t “fruit” and “child” similar concepts, since the fruit contains the seed that becomes the literal child of the plant? And wouldn’t you expect them to be connected in many cultures? Be fruitful and multiply; Blessed is the fruit of your womb Jesus.
@Etienne: Spanish “Dios” is not a conservative survival. Instead, it owes its final /s/ to the influence of Church Latin.
Likewise espíritu, I take it.
Aren’t “fruit” and “child” similar concepts
Yes indeed; but I think that works against the idea of actual colexification to some extent, because it opens the door to metaphor.
What I mean is: in Mooré, bíiga, besides meaning “child”, does indeed seem to be the normal unmarked word for fruit: cf Genesis 2:16
La Wẽnnaam Sẽn-Ka-Saab kõo ninsaal noor n yeele: «Fo tõe n dɩɩ tɩɩs nins fãa sẽn be zẽedã zĩig pʋgẽ wã biisi.»
“And God Who Has No End gave the human being a command, saying: ‘You can eat the fruits of all the trees in the garden'”
But in Kusaal, it’s not the usual word:
Ka Zugsɔb Wina’am bas dau nɔɔr ye, “Fʋ tun’e di tiis wala banɛ bɛ lɔmbɔn’ɔgin la wʋsa
“And the Lord God left the man a command: “‘You can eat all the trees’ fruits which are in the garden'”
So I think that in Kusaal, biig “fruit” is a metaphor rather than an actual colexification.
In fact, Mooré, judging by my trawl of the dictionaries, seems to be an actual outlier in Oti-Volta in this respect: I wonder if the colexification might even be something that Mooré has picked up from its neighbours or from substrate languages (Mooré has pretty certainly expanded widely at the expense of other languages comparatively recently.) Dyula deen is both “child” and “fruit” … Kasem bu likewise.
Cf. Lat. planta > Ir. clann.
For a younger me, AAVE was a lot easier to understand than the Glasgow variety I often heard on and around the soccer pitch. The sonic qualities of AAVE and most of the grammatical twists are of a piece with various rural and southern accents. I subbed in a classroom on the west side of Chicago and never had any trouble understanding the kids.
Whereas the lore on our college soccer team included the time two teammates were walking along, the Glaswegian talking about his plans for spring break. When he finished, the American said, that’s cool. What are you doing for spring break? He had no idea what his Scottish teammate had said. Another Glaswegian on the team would shout geesi’ when he was open. I thought it was an unusual Scottish word he used, till on reflection I realized i was a contraction of “give us it”, which even uncontracted is ungrammatical and foreign to me — wrong word order, elision of consonant in a VCV sequence where it wouldn’t be lost in my dialect and a glottal stop with a different quality and oral shape than mine.
For DE’s sake, I’ll also give the story of the primarily Irish soccer team I played with just out of college. At a banquet, someone gave a toast to the team that had brought together people from all over the British Isles, England, Ireland, Scotland and Wales, as friends and teammates. Note that any Scottish or English on the team were Glasgow and Manchester Irish. But anyway, a guy from Glasgow next to me said “I never me’ a Wełshman.”
Interestingly, in Mooré, bíiga “child” does not use the expected regular plural biisi, but has the suppletive plural kamba; biisi just means “fruits” (and chess pawns, according to Niggli’s dictionary.)
Farefare has kɔma “children” as the plural of bíá “child”, which is obviously cognate, but beyond that, I don’t know where the word comes from.
Cf. Lat. planta > Ir. clann
Via Brythonic: cf Welsh plant “children.” But all Hatters know this, so the ellipsis was natural.
For a somewhat less scrutable metaphor, Kusaal tuŋ “calabash plant, Lagenaria siceraria” also means “lineage, clan.” Can’t see it, myself. Maybe it’s just one of those accidental homophony things (though none of the segments involved commonly results from historical mergers in Oti-Volta.)
“I never me’ a Wełshman.”
Alas, what unfortunate and impoverished lives some people lead.
It occurs to me that if, as Glasgow Irish, he was a Catholic, I probably would not have met him in my Glaswegian childhood. Catholics went to a different school somewhere.
(But I do know a Welsh-speaking Catholic, so the thing is not logically impossible.)
You find labiodentals unrelated to bilabials? I think my Hebrew school classmates (as far as I could tell) and I just had to be told once, and from then on it was obvious. The same is true for k and kh
In the Masoretic text, the bgadkpat thing is very nearly non-contrastive, so much so that the very few exceptions should probably be explained away rather than obscure the overall picture. The objections to this are based on failure to follow through on the implications of the elaborate and beautiful Masoretic recording of stress distinctions and stress sandhi, which not only shows that fricativisation at the beginning of words is regular and predictable, but that (at the time the pointing was devised) schwa was still /ə/, contrasting with zero, in cases where fricativisation follows (and in other cases where it derives historically from a full vowel.)
[Stress cannot be thrown back onto a closed syllable in stress sandhi, but is retracted over “vocalic” schwa to a preceding open syllable. That is to say, CVCˈCV and CVCəˈCV differ in their stress sandhi behaviour when the following word has initial stress, and the difference correlates nicely with whether the third C is fricativised.]
Aside: which scholarly tradition uses the acronym bgadkpat? I only know it as beged kefet.
That was Pre-Proto-Romance. Actual Proto-Romance was later – probably at the beginning of the Crisis of the Third Century, when Latin was spread over the whole empire and the geographic barriers weren’t up yet.
It’s still an important point, however, that Romance is descended from the Latin dialect of Rome – not the dialect of Praeneste, and not the dialect of the ager Faliscus. Just as the modern Standard Englishes are not descended from the entirety of Old English, Romance is not descended from the entire Latin language. If you reconstruct a protolanguage with dialectal variation in it, you haven’t reconstructed the proto”language”, the last common ancestor before any innovations of individual branches. That’s the point of Le slave commun at least as the term is used today – Common Slavic, the state after the great expansion, with about 6 dialects distinguishable by a grand total of maybe 5 features, was not Proto-Slavic; Proto-Slavic was the state during the great expansion (around the year 600), a monolith that had only undergone half of the Great Slavic Vowel Shift yet.
Conversely, of course, it’s generally impossible to tell from reconstruction alone if a reconstructed proto”language” was actually just one dialect among several that would be classified as all belonging to the same language if we had any idea of them. Praenestine and Faliscan Latin are only known from their own inscriptions and maybe from one loanword in Roman Latin (nihil, evidently from *ne hilum with an unexplained hilum, makes sense as *ne filum “not even a thread” once you apply the Faliscan f > h change known from inscriptions).
This used to be the accepted wisdom, but there doesn’t actually seem to be evidence for it. Tomorrow I’ll dig up the paper that says the basic vocabulary of attested Classical Latin and reconstructed Proto-Romance are actually identical. Romance may not be descended from the literary register, but there doesn’t seem to be an obstacle to the register of Cicero’s private letters being what he, and the rest of Rome, actually spoke.
Are you sure these different words aren’t taken from different speakers? Or perhaps a register distinction is involved (compare Tahitian, or maybe someone wanted to give citation forms in a “more careful” pronunciation)?
Violently replaced one despotism with another. Less bureaucracy and lower taxes, though.
Tahitian
The link, correctly, is to Sāmoan.
By “Proto-Romance” I meant merely what is commonly called “Latin.” Call it “Old Proto-Romance,” if you must.
Oops, yes, Samoan. It’s bedtime.
Are you sure these different words aren’t taken from different speakers?
Good question.
I’m going entirely on the lexicon part of Rapp’s book. I’d deduced that the simplification of the labial-velars couldn’t go back to proto-Farefare from the contemporary dialects, but hadn’t previously encountered evidence that it was in-living-memory recent.
But either way, it bears out my point: a change captured in progress (and nearly complete: I estimate that about 90% of words he records have already undergone it.)
Rapp’s work is based entirely on Bolgatanga Gurenne, so he wasn’t collecting words from quite different geographical dialects, at least. He was only actually in the field for some three months, and he seems to have relied on just one principal informant, Nicholas Aboya; he’d been Rattray’s interpreter thirty years previously, and so was presumably no teenager. The lexicon part names him specifically as co-author, and I presume that words are from his idiolect unless given a specific text reference. Such words include both kp gb and simplified k g types.
The proverbs which constitute the text material were contributed mostly by a local Catholic priest, with contributions by a teacher and various other people whose professions are not given. Unfortunately Rapp gives no ages, though 22 proverbs were contributed by the schoolboy son of Aboya.
Proverbs are a rather formalised genre*, so you’d expect the language to tend conservative if anything.
Rapp (excellent man) sources the proverbs and gives references to thrm in his lexicon. This suffices to show, at least, that forms both with kp gb and those with simplification to k g can all be found in proverbs from the main contributor, but I haven’t looked into it in more detail yet.
(I’ve been more interested in the distribution of the two imperfective forms: Nõotre, Mooré and Farefare, unlike the rest of Western Oti-Volta, have two distinct imperfective forms in the one regular conjugation, but unfortunately for the poor comparativist, not only do no two of the languages seem to agree regarding their use and distribution, but the two main descriptions of Farefare don’t agree with each other either.)
* I meant “proverbs in Oti-Volta.” But one or two Welsh proverbs preserve the ancient insular Celtic absolute versus conjunct thing, which seems otherwise to have broken down already in the Old Welsh period. Tyfid maban, ni thyf ei gadachan “A baby grows; its swaddling-cloth does not grow.”
which scholarly tradition uses the acronym bgadkpat? I only know it as beged kefet.
Joüon and Muraoka use it, as does Gesenius (as bᵉgadkᵉphath); it seems pretty usual in the Christian tradition of BH study. Dunno where Gesenius got it from (Delitzsch, most likely), but on first principles it’s likely to have been ultimately from a Jewish source.