Having read the first couple of chapters of The Turkish Language Reform: A Catastrophic Success (see this post), I thought I’d share some bits that I found interesting, enlightening, or amusing. From Chapter 2, “Ottoman Turkish”:

The mixture of Turkish, Arabic, and Persian, which Turks call Osmanlıca and we call Ottoman, was an administrative and literary language, and ordinary people must have been at a loss when they came into contact with officials. But while they must often have been baffled by Ottoman phraseology, they were capable of seeing the funny side of it. In the shadow theatre, the running joke is that Karagöz speaks Turkish while his sparring partner Hacivat speaks Ottoman. In the play Salıncak, Karagöz keeps hitting Hacivat. Hacivat asks him why, but receives only nonsensical answers sounding vaguely like his — to Karagöz — unintelligible questions. Eventually he asks, ‘Vurmanızdan aksâ-yı murad?’ (What is your ultimate object in hitting me?). To which Karagöz replies, ‘Aksaray’da murtad babandır’ (The turncoat at Aksaray is your father) […]. A rough English parallel would be, ‘Explain your bellicose attitude.’ — ‘How do I know why he chewed my billy-goat’s hat?
* * *
Even before the rise of the Ottomans there had been expressions of dissatisfaction with the dominance of Arabic and Persian. In 1277 Şemsüddin Mehmed Karamanoğlu, the chief minister of the ruler of Konya, decreed that thenceforth no language other than Turkish would be spoken at court or in government offices or public places. Unfortunately he was killed in battle a few months later.

Chapter 3, “The New Alphabet,” explains that -h- was very nearly adopted to indicate palatalization, on the model of Portuguese (“so khatip for what is now written kâtip“), and palatalized k was almost written q (“The explanation is to be sought in the name of the letter q, which Turks follow the French in calling , pronounced /kyü/. This letter, whose name had the requisite palatalized initial sound, seemed the ideal device for indicating /ky/.”). And in a discussion of the nearly one-to-one match of letters to sounds, Lewis says “the only word one can think of that [a well-programmed speech synthesizer] might fail to enunciate correctly is ağabey (elder brother), pronounced /ābī/,” which I’ll have to try to remember. I like annoying irregularities.


  1. “Irregularity” in this connection is a bit curious. I know it’s a standard term for such things as the pronuncation of ağabey as /ābī/, for “irregular verbs” and so on – I’m not claiming that this use of the term is “wrong”. But since the pronuncation of ağabey is always /ābī/, it’s a regular pronunciation. Similarly, irregular verbs are always conjugated as they are, according to the rules for those verbs.
    According to one rule system, a phenomenon is irregular, according to another it might be regular – if if some regularity can be identified. Only when the pronunciation of a word, or the conjugation of a verb, is unpredictable would it seem appropriate to call it irregular. But that’s not how the terminology panned out.
    In looking for systematic explanations of phenomena, there has historically been a tendency to look for a single “most important” or over-arching rule – “most important” in the sense that it covers most cases. Deviations are called irregularities, or placed under less important or “secondary” rules.
    Over the last fifty or so years, workers and theoreticians in several fields have found it more and more difficult to establish such a hierarchy of rules to organize the things they work with or investigate. This has happened in IT, physics, neurobiology and sociology. Thinking in hierarchies of rules is replaced by thinking in networks of rules, as a Denkstil. This has been pointed out by many people – it first became really clear to me from reading Luhmann.

  2. There are many factors that have contributed to this change in Denkstil. One is pretty simple, almost crass: due to the sheer amount of specialized, fragmented knowledge in any field, no one is in a position to grasp the whole. Yet the fragments are not completely unrelated to each other, and even display unexpected relationships with fragments from other fields.
    One is almost forced to think of nets, which connect many different nodes in a way that allows repair and reconnection. The question of whether you have the same net after repair and reconnection is not that important, so long as you can still catch fish. Hierarchies, in contrast, do not remain the same when you rearrange the dominance structure. They are too stiff, they do not morph into each other easily.
    And now back to Turkish language reform.

  3. In that sense there is no irregularity in language: there is nothing that varies completely randomly without underlying cause.

  4. I studiously avoided the notion of randomness. Instead, I wrote of unpredictability and deviations from a rule – and whether or not regularity can be “identified” in given cases.
    I’m not quite sure what you mean by “there is nothing that varies completely randomly without underlying cause”. I read it as a statement of intent (“I intend always to try to understand things in accordance with this principle”), but that may not be what you meant.
    You say it with respect to language, but why language in particular ? It seems general enough to apply to any field of inquiry. Indeed it seems to be a claim about how the universe works – past, present and future. Such statements are notoriously difficult to prove, or even to make sense of. That’s why I assume it is a research proposal.

  5. That is, a proposal for linguistic research.

  6. @John Cowan
    Let me say, first, that I have no idea what you mean by “there is nothing that varies completely randomly without underlying cause”. That said, what about bog-standard quantum mechanics? Whatever you mean by “varies completely randomly”, I suspect that I can construct a pure QM state that has that property, and “no hidden variables” seems to me to be a plausible translation of “without underlying cause.”
    Technical note: please define the ensemble that underlies what you mean by “randomly.”

  7. With a little help from Anselm, I think I can prove that a weaker statment is true: not all things vary completely randomly.
    For suppose that the statement A “all things vary completely randomly” were true. Since statements are things, A too would also vary completely randomly, and so would literally be untenable. From one moment to the next, at the pop of a quantum foam bubble, the statement could change and slip from your grasp. QED
    WARNING: this comment contains traces of nuts.

  8. First of all, I disclaim any extensibility of my remark beyond the realm of language behavior.
    Second, by random I mean ‘unpredictable’; see the first definition.
    I read your claim to be that mice as the plural of mouse ought not to be called irregular, for it consistently obeys the rule “in every case mice is the plural of mouse“. This is true at the level of tokens, but at the level of types, it is too singular to count as a rule. What is more, regular behavior at the level of tokens is how languages work, so it is not worthwhile within linguistics itself (as opposed to pre-theoretical talk about language) to use the term regular at the token level, whereas at the type level the regular/irregular opposition is very useful.
    But I must further qualify my statement by noting that some linguistic behaviors are only statistically predictable: there is no knowing in a particular instance whether a native New Yorker will be rhotic or non-rhotic, but the probability of an utterance being rhotic varies with the speaker’s age and class and the formality of the occasion among other things.

  9. What is more, regular behavior at the level of tokens is how languages work, so it is not worthwhile within linguistics itself (as opposed to pre-theoretical talk about language) to use the term regular at the token level …
    That makes perfect sense, thanks for the clarification.
    …, whereas at the type level the regular/irregular opposition is very useful.
    I’m wondering in what precise sense you are using “type” here, in connection with the SEP article. It contains only one section that makes sense to me: 1.1 “What the Distinction Is [between types and tokens]”. The rest of it is a coroner’s report on the history of ideas.
    I bet you could explain what you mean with a practical example, perhaps without using the word “type” – if you care to take the trouble, which I hope you do. Would “category in a classification scheme” catch the meaning, without getting tangled in metaphysics ?

  10. Athel Cornish-Bowden says

    I doubt whether ağabey would present much of a problem for a well designed speech synthesizer in practice, because it could just be programmed as a special case. It’s when the exceptions to a rule get to be more numerous than the instances of it that the real difficulties start.
    Before I started using French every day I wondered why they bothered to have a rule that nouns in -al form their plural in -aux, as cheval appeared to be the only common word to which it applied. Now, of course, I realize that even though nouns in -al are thin on the ground, adjectives in -al are abundant and follow the same rule.

  11. Today I had a revelation (more of an ear-opener, actually) as to the difference between the a’s in tache and tâche. marie-lucie has mentioned several times that the distinction has been disappearing in certain places in France, and even gave the different IPA pronunciations. I’ve never been able to understand (not knowing IPA) or hear what she was talking about – as far as I was concerned, tash is tash.
    Today I was dozing during a rerun of an arte film in French on wild cats, that I had seen before. Listening to it, I kept experiencing aural black holes at certain points, as still happens when I listen to French speakers. I knew the film, as I said, and wondered what it was that kept cropping up without my being able to understand it – something like “chaesovahje”.
    I looked at the scene, saw one of the wild cats, and immediately solved the problem: I had been “listening for châts sauvages” !! I have always pronounced chat (to myself) as if it were chât. It has taken me all this time until I could hear this crucial difference – which, I daresay, doesn’t count for much in English or German. Now, I wouldn’t even say that it’s “subtle”. All you have to do (!) is learn what counts.

  12. I feel a little nervous taking on Geoffrey Lewis here, but based on my year and a half in Turkey, ağabey is not always pronounced abi. That’s generally how it’s said but I believe one can also say it the ‘proper’ way. This appears to back me up:
    If I had to nominate a Turkish word that isn’t quite pronounced as its spelling would indicate it is estağfurullah, where the second ‘u’ is not pronounced, at least based on my experience.

  13. I love estağfurullah, or rather its Arabic original astaghfirullah, which seems to be as ubiquitous as inshallah but is far less well known outside of the Muslim world. There’s something very satisfying about muttering it.

  14. The article says: “When a Muslim … wishes to prove their innocence in an incident, they will use this expression.” Could it be said by a father about to punish his son, say as an equivalent to “God help me” or “This is going to hurt me more than it hurts you” ?
    In other words, can estağfurullah be used to wash one’s hands in innocence without waiting for Allah to do so, as if saying “I am only doing what I must do” ? People being people everywhere, I guess that it can, just like the English expressions – but I don’t know that.
    “When a Muslim wishes …”, in the article, is a pretty general and vague claim. You sure couldn’t claim that “when a Christian says ‘So help me God’, he is seeking forgiveness”. How could that be explained in a WiPe article addressed to a Muslim readership ?

  15. Re irregularities:
    As far as I can tell, the standard pronunciation of ‘herkes’ [everyone] is as though it were ‘herkez’.
    Also, it seems to me that ‘agabey’ is more often than not spelled ‘abi’ (like it’s pronounced) these days.
    Re Lewis’s excellent book:
    Can it really be true that Ataturk invented several of the commonly used terms of elementary mathematics?

  16. Can it really be true that Ataturk invented several of the commonly used terms of elementary mathematics?
    Apparently so!

  17. Re: ağabey, It seems to me that in haydi (roughly equivalent to Russian давай, meaning something like “come on”), the glide is usually elided, which does not happen in the same environment in other cases, like the name Aydın.
    For that matter, I can never really decide whether merhaba is pronounced with an h or not. No matter how I say it, I don’t really sound Turkish; but I tend to hear it as h-less.
    But! I suppose Lewis is not splitting hairs here, and these could well just be ordinary-speech corruptions of some idealized “correct” speech that Turkish teachers wish we were all using.
    One of the benefits of Russians’ obstinate prescriptivism is that from the beginning you’re drilled in the nuances of what belongs to proper литературная речь; what is merely “laziness” or stupidity on the part of native speakers, not to be emulated (e.g. надо with noun complements, звОнит); and what is unforgivably slangy (ни фига се!). Turkish pedagogy generally seems to lack that prescriptivist bent, and I have often felt the lack of it in my education.

Speak Your Mind