Persian: historical sound changes

< Previous | Next >

PersoLatin

Senior Member
UK
Persian - Iran
I believe به and با are different and distinct from one another, in meaning and therefore in use.

In English به can mean:

1 - To, examples:
سر بسر، پا بپا، در بدر، به انجا، به شهر، به ديوار، به تو/من چه، بپاى او افتاد

2 - In, examples:
به اين روش، بحالت عادى، بزبان فارسى، بجاى من رفت

3 - Also in, at or on, examples:
بوقت، به روز، بزودى، بتنهايى ،كلاه به سر، كاكل به سر
به بيمارى/مريضى مبتلا شد
سرش بدار رفت، بفروش رسيد
به چه معنى/دليل/حساب

(please note, this thread is not about the exact meaning or variety of meanings of به in English)

In the above examples, you cannot replace به with با, at least not without changing the meaning, or making them sound wrong altogether.

However in the following example types, با and به seem to be interchangeable:
4 - بخوبى، بخوشى، بسختى، بنام خدا، به زور، به لطف شما، اميدوارم بسلامتى برسى، به مرور زمان
To me it looks like these started off with با and over time, due to closeness of sounds, changed to به.

As an example, do these have the same, or subtly different meanings? And if they are different, has that difference developed because of accidental misuse of به?
به سختى/زور این کار را کرد and با سختى/زور این کار را کرد
جشن به خوبی/خوشی انجام شد
and جشن با خوبی/خوشی انجام شد
به مرور زمان
and با مرور زمان
به لطف شما and با لطف شما


what is your view on this?
 
Last edited:
  • farasso0

    Senior Member
    پارسی
    این کلمات در گذشته با به استفاده می شدند.

    من با آچار با سختی بازش کردم............من با آچار به سختی بازش کردم.
    جشن با حضور آقای شهردار با خوبی خوشی به پایان رسید.........جشن با جضور آقای شهردار به خوبی و خوشی به پایان رسید.
    با مرور زمان با هم واکنش دادند.....به مرور زمان با هم واکنش دادند.
    با لطف شما توانستم با مادرم صحبت کنم....به لطف شما توانستم با مادرم صحبت کنم.
    با سلامت، با سلامت برسید(که من نشنیدم تا حالا)....به سلامت برسید، به سلامت

    من فکر می کنم پیشینیان ما دلیل خوبی داشتند برای اینکه از حرف اضافه متفاوتی برای اینها استفاده کنند.
     
    Last edited:

    Treaty

    Senior Member
    Persian
    Historically, با (Middle Persian abāg) mainly meant "[together] with", while "به" (Mid. Pers. pad) had a wide range of meaning as explained by PersoLatin. For example, in Shahname, a person usually hits another person به a weapon not با it. However, we can see that later (1200AD onwards), با is also used to connote "by". Meanwhile, using به for this meaning has become less and less frequent (it will be confusing if I say من درخت را به تبر زدم).

    Considering your examples, it is possible to picture some of them with both با and به. For example با خوبی و خوشی به پایان رسید implies an end accompanying joy, while به خوبی و خوشی may imply the result of ending. Same goes for با مرور and به مرور. In the context, they become so semantically close that they are used interchangeably without notice. It indeed matters how the semantic context is defined. For example if I see روش as a process, I may be comfortable to use با (I do actually say با این روش as I also use با این راه).
     
    Last edited:

    PersoLatin

    Senior Member
    UK
    Persian - Iran
    For example, in Shahname, a person usually hits another person به a weapon not با it. However, we can see that later (1200AD onwards), با is also used to connote "by". Meanwhile, using به for this meaning has become less and less frequent (it will be confusing if I say من درخت را به تبر زدم).
    So what I said below is the opposite of what has actually happened? Although 'closeness of sound' as a reason for the change, may still be valid (?).
    To me it looks like these started off with با and over time, due to closeness of sounds, changed to به.
     

    Treaty

    Senior Member
    Persian
    I can only say the change from به to با for when it means "by". I don't know the reason (though closeness of sound makes sense, especially considering that this semantic change happened long after very different abāg and pad changed to more similar and ba in Dari).
     

    PersoLatin

    Senior Member
    UK
    Persian - Iran
    especially considering that this semantic change happened long after very different abāg and pad changed to more similar and ba in Dari
    That's if you believe that Persians ever pronounced the final 'g' or 'k', in this and many similar words, you will find in Pahlavi and Avesta text.
     

    Treaty

    Senior Member
    Persian
    I think this has already been discussed in another thread. Anyway, an evidence for the pronunciation comes from the ج (e.g., نموذج برنامج طازج) or ق (e.g., فستق خندق ابستاق) at the end of Persian loans in Arabic; although, I have no idea if this k/g pronunciation applied to all k/g endings and if all Middle Persian accents had it.
     

    PersoLatin

    Senior Member
    UK
    Persian - Iran
    I think this has already been discussed in another thread.
    You are right, I raised that thread, unfortunately, at the time, it was not thoroughly discussed, one forum member repeatedly took the focus away from the main point.

    Anyway, an evidence for the pronunciation comes from the ج (e.g., نموذج برنامج طازج) or ق (e.g., فستق خندق ابستاق) at the end of Persian loans in Arabic; although
    I'm sure you'd agree, one of the most likely routes for these into Arabic (& other languages), is via the written, rather than the spoken form. In fact, more of this word group survived in Arabic than in Persian, which may well be proof that, the transfer aoccurred in a relatively short period of time, between the time of Arab invasion and adoption Arabic script, during which time, text books in Pahlavi were translated & used by the Arabs.

    I have no idea if this k/g pronunciation applied to all k/g endings and if all Middle Persian accents had it.
    There are many hundreds of these words, just consider the past participle of around 800-900 Persian verbs. The point of my thread was, that the k/g (ending) applied only to words that ended in a long vowel i.e. ā, u, i & é/a (eh), and that k/g, very likely, was a marker to tell the reader, the word ended in a strong vowel, and was never intended to be pronounced, e.g. āhug (āhu), bināg (binā), ruzig (ruzi), satārag (setāré/satāra).

    I don't think anyone can convincingly argue that, for example the original ستاره, for a period of a few hundred years (duration of MP) was pronounced ستارگ, then it changed back to ستاره, without leaving any trace in spoken Persian, not even in one village, in the outer reaches of Iran, and also, it mysteriously disappeared in written Persian too, as soon as, or soon after, the Arabic script was adopted, it doesn't make sense.
     
    Last edited:

    Dib

    Senior Member
    Bengali (India)
    I'm sure you'd agree, one of the most likely routes for these into Arabic (& other languages), is via the written, rather than the spoken form. In fact, more of this word group survived in Arabic than in Persian, which may well be proof that, the transfer aoccurred in a relatively short period of time, between the time of Arab invasion and adoption Arabic script, during which time, text books in Pahlavi were translated & used by the Arabs.
    I am curious to know why you think so. I wouldn't guess that Pahlavi literacy (which is essential to understand its spelling conventions) was ever common among Arabs. On the other hand, if it was native Persians who introduced these words into Arabic, they would certainly know the proper Persian pronunciation of these words and spell them in Arabic accordingly? Yes, some residual orthographical conventions may get transplanted through transliteration, like writing setareh with a silent h in the end also in English, but it works only because an "h" in that context in English would also be silent. Using j/q as silent letters in Arabic spelling sounds really far-fetched to me.

    My guess is that the standard theory would suggest chiefly pre-Islamic borrowings from (Middle) Persian to Arabic (mostly through Aramaic).

    There are many hundreds of these words, just consider the past participle of around 800-900 Persian verbs. The point of my thread was, that the k/g (ending) applied only to words that ended in a long vowel i.e. ā, u, i & é/a (eh), and that k/g, very likely, was a marker to tell the reader, the word ended in a strong vowel, and was never intended to be pronounced, e.g. āhug (āhu), bināg (binā), ruzig (ruzi), satārag (setāré/satāra).
    For the (early MP?) endings -ik/-ak (> ig/ag > ī/a), there are clear Sanskrit parallels in -ika/-aka as well. So, New Persian setāre parallels the extended Sanskrit tāra-ka- = star (also unextended tārā- = star), the initial s- is a so-called IE s-mobile which does not show up in this Sanskrit form, though it does show up in other attestations (e.g. Rigveda 2.2.5 stṛ-bhiḥ = with stars), proving that we are indeed dealing with the same root.

    Also, remember, the unextended past participles in -ta of OP evolved into the simple past of Modern Persian having lost the final -a (and voicing t to d in specific contexts, like intervocalic among others). This form still acted as past participle in Middle Persian. So, to take fdb's example from the other thread (zāta-), its exact reflex in modern Persian is zād (Sanskrit jāta). Similarly OP kṛta- (Cf. Sanskrit kṛta-) > modern Persian kard, etc. The new Persian past participle (zāde, karde,...) was obviously an extended form. I am curious, does this extended past participle occur at all in MP? I expect their precursors (i.e. adjective formations not yet interpreted as past participles) in -ag to appear though (Cf. Sanskrit adjective derivations in part participle+ka-, e.g. jātaka- new-born, kṛtaka- artificial).

    This does not cover āhug (āhu), bināg (binā), for which I don't know the etymology to suggest one way or the other.

    I don't think anyone can convincingly argue that, for example the original ستاره, for a period of a few hundred years (duration of MP) was pronounced ستارگ, then it changed back to ستاره, without leaving any trace in spoken Persian, not even in one village, in the outer reaches of Iran, and also, it mysteriously disappeared in written Persian too, as soon as, or soon after, the Arabic script was adopted, it doesn't make sense.
    Could you provide a reference to the Old Persian form? I suppose that's what you mean by pre-MP "original"? I couldn't locate any attestation.
     

    Treaty

    Senior Member
    Persian
    ... without leaving any trace in spoken Persian, not even in one village, in the outer reaches of Iran, and also, it mysteriously disappeared in written Persian too, as soon as, or soon after, the Arabic script was adopted, it doesn't make sense.
    The simplification of a language, especially dropping of endings, usually occurs first in the spoken and local variants and then is reflected in the formal and written language. The Arab invasion effectively ended the formal Middle-Persian and paved the way for local variants (e.g., Dari). When Pahlavi writers used g/k endings, many Persian dialects had probably already lost them (like in Sogdian which has words with both k and y endings).

    For example, when Latin was still the official language, people had long departed from it (or its related proto-local-dialect) and spoken with local dialects. Almost no trace of Latin s/m endings is left in modern Romance languages. Would you consider words like German zentrum or English stimulus only took m/s endings because of the written but not pronounced in Latin?

    By the way, I consider forms like zendegi and zendegan as possible traces of that g ending.
     
    Last edited:

    PersoLatin

    Senior Member
    UK
    Persian - Iran
    The Arab invasion effectively ended the formal Middle-Persian and paved the way for local variants (e.g., Dari). When Pahlavi writers used g/k endings, many Persian dialects had probably already lost them (like in Sogdian which has words with both k and y endings).
    Or you could see it as: "many Persian dialects had never had them in the first place, or not fully" (like in Sogdian which has words with both k and y endings)"

    By the way, I consider forms like zendegi and zendegan as possible traces of that g ending.
    That is absolutely correct, and if someone who is not totally familiar with M and N Persian looks at these exact examples, s/he'd assume the noun form of zendegi is zendeg, because s/he knows the formula says, 'to get the noun/singular, drop the i or the plural marker ān'. I believe that same thing has happened. Other examples, setaregān becomes setareg, hamagi and hamag, and many many others of this type. So the g/k was for liaison purposes only, but someone got it wrong.


    There are far too many of this group of words, in Persian, to ignore or explain away with one formula (not suggesting you did). I believe this is an error, made by modern scholars, and they used the presence of those few words, found in Arabic (as well as, a few actual Persian words like tajik, tarik), as proof that they are right, and ignored that New Persian lacked any such form.

    Let’s look at this from a different angle; do you not think that Ferdôsi (& his contemporaries) would have been aware of this g/k ending, even if it had completely disappeared, from spoken Persian, by his time? Ŝāhnamé is full of references to history of Sāsāsniān, which, relatively speaking, was contemporary to Ferdôsi, and all(?) written in Pahlavi, he must have been an expert in reading Pahlavi, why did he not use setarag, hamag or zādag etc., in his poetry? He claims and we believe that he revived Persian; wouldn’t keeping the g/k ending be more consistent with this ideal?
     
    Last edited:

    PersoLatin

    Senior Member
    UK
    Persian - Iran
    Thanks Dib.

    I am curious to know why you think so. I wouldn't guess that Pahlavi literacy (which is essential to understand its spelling conventions) was ever common among Arabs. On the other hand, if it was native Persians who introduced these words into Arabic, they would certainly know the proper Persian pronunciation of these words and spell them in Arabic accordingly?
    I am trying to convey the idea, but let's take your suggestion that those few words entered Aramaic first, and in that case, it is reasonable to assume that the text books would have been translated without intervention/presence of Persians, therefore g/k ending were assumed to be correct.

    For the (early MP?) endings -ik/-ak (> ig/ag > ī/a), there are clear Sanskrit parallels in -ika/-aka as well. So, New Persian setāre parallels the extended Sanskrit tāra-ka- = star (also unextended tārā- = star), the initial s- is a so-called IE s-mobile which does not show up in this Sanskrit form, though it does show up in other attestations (e.g. Rigveda 2.2.5 stṛ-bhiḥ = with stars), proving that we are indeed dealing with the same root.
    Do -ik/-ak endings mean anything, for example, are they diminutive makers, or do they function in exactly the same was as g/k ending in MP? And do you have examples other than setāré/setāra, in both Persian and Sanskrit?

    Could you provide a reference to the Old Persian form? I suppose that's what you mean by pre-MP "original"? I couldn't locate any attestation.
    Yes I did mean pre-MP & I will look for it, however fdb's zāta example is good enough to prove the g/k was not present on this word which is a past participle, and maybe by extension, we can say, not present on words of similar category.

    What PersoLatin writes is partially, but only partially correct.
    In fact, I never asked (fdb) or worked out, in what part I was correct or incorrect (thread,).
     

    Treaty

    Senior Member
    Persian
    Or you could see it as: "many Persian dialects had never had them in the first place, or not fully" (like in Sogdian which has words with both k and y endings)"
    The issue is, as Dib explained, that Persian had this k/g ending in many of these words. The suffix(es) had a range of meaning connoting hypocorism, belonging, relation, and "instance of". It is natural to assume the ancestor of all Persian dialects had it as well and among them the formal Persian was the last Persian dialect to submit to its omission. You can compare it to ی-ending in words like فرمای نمای جوی بوی شوی etc. They have been disappeared from Persian dialects apart from the more conservative literary register. If it wasn't because of poems it would have been easy to argue that the ی in بفرمایم is only for liaison.

    As for fdb's post, I think he meant that your hypothesis (of mute k/g) is true for post-Islamic Pahlavi writings but not true for Sassanid Pahlavi writings (this was why it was "partially" true).
    ... and they used the presence of those few words, found in Arabic
    And in Aramaic and Armenian. It is a well-established practice in etymology because the borrowing languages don't go through the same sound changes as the source language does. For example, 'margharet' or خزن are closer to the original words than مروارید and گنج.
    Let’s look at this from a different angle; do you not think that Ferdôsi (& his contemporaries) would have been aware of this g/k ending, even if it had completely disappeared, from spoken Persian, by his time? He claims and we believe that he revived Persian; wouldn’t keeping the g/k ending be more consistent with this ideal?
    You have a great faith in him. I don't believe he revived Persian. Persian was already the lingua franca of the Samanids and Ghaznavids. I'm not sure he believed so either. I don't believe he was a linguist. He knew what he knew and spoke his own Persian (that is Dari, very different from Sassanid Middle Persian). I'm not sure he had access to or knew Pahlavi language or sources.
     

    PersoLatin

    Senior Member
    UK
    Persian - Iran
    The issue is, as Dib explained, that Persian had this k/g ending in many of these words. The suffix(es) had a range of meaning connoting hypocorism, belonging, relation, and "instance of".
    But you know by now that the g/k we are talking about, represent none of these general connotations: "hypocorism, belonging, relation, and "instance of"", even if we believe they were actually pronounced, they had no meaning, I'm more than happy to know what you think these connoted, other than what I believe, i.e. markers and liaisons.

    And in Aramaic and Armenian.
    The number of borrowing languages is not important, in fact they could be used to prove my theory, the same error copied faithfully by many others, and in a similar period.

    As for fdb's post, I think he meant that your hypothesis (of mute k/g) is true for post-Islamic Pahlavi writings but not true for Sassanid Pahlavi writings (this was why it was "partially" true).
    But wouldn't that be stating the obvious? At least that's not how I read it but are you (or fdb) actually saying there were two flavours of Pahlavi script, a Sassanid and a post Islamic one? I'm asking as I don't know.

    You have a great faith in him. I don't believe he revived Persian. Persian was already the lingua franca of the Samanids and Ghaznavids. I'm not sure he believed so either. I don't believe he was a linguist. He knew what he knew and spoke his own Persian (that is Dari, very different from Sassanid Middle Persian). I'm not sure he had access to or knew Pahlavi language or sources.
    His claim, as I understand it, whether true or not, is not that he made Persian the lingua franca of the Samanids and Ghaznavids, or rescued it from certain destruction, but that he took it back to the way it was used and spoken, before the influence of Arabic. Whatever our views of Ferdôsi, you can not know for sure what his knowledge of Pahlavi was, some accounts of his historical facts, despite being fanciful (Herodotus comes to mind, here) are accurate and although some came from oral traditions, the rest must have come from Pahlavi sources, and by the way, the spoken Dari and MP could not have been different, at least not in pronouncing this group of words. And you don't have to be a linguist to work out what the g/k ending represented and question why only one set of words have them and others don't.
     
    Last edited:

    Dib

    Senior Member
    Bengali (India)
    [...] and in that case, it is reasonable to assume that the text books would have been translated without intervention/presence of Persians, therefore g/k ending were assumed to be correct.
    Why is that? Achaemenids were one of the key figures in spreading Aramaic in the middle east. I don't know the status of Aramaic scholarship in Sassanid Persia, but I doubt it died out, given that they wrote half their roots in Aramaeograms (huzvarishn). Could you throw some light on that?

    Do -ik/-ak endings mean anything, for example, are they diminutive makers, or do they function in exactly the same was as g/k ending in MP?
    Quoting from Whitney's Sanskrit Grammar section 1222:
    "[ka] has been, on the one hand, specialized into an element forming dimunitives; and, on the other hand, and much more widely, attenuated into an element without definable value, added to a great many nouns and adjectives to make others of the same meaning..."

    The -ika was commonly used to derive adjectives, just like Middle Persian -ig > New Persian -ī. It is abundant in classical Sanskrit, but admittedly it was rarer in the older stages of the language, but existent.

    And do you have examples other than setāré/setāra, in both Persian and Sanskrit?
    Of course, like I already mentioned: Modern Persian zād~zāde / Sanskrit jāta~jātaka; Modern Persian kard~karde / Sanskrit kṛta~kṛtaka; etc. As a matter of fact, it seems this past participle in -ta + -ka formation was similarly generalized in both Persia and India in the last 1500 years or so (coincidence?), e.g. Punjabi past tense kitā goes back to "kṛta-ka" or something similar.

    however fdb's zāta example is good enough to prove the g/k was not present on this word which is a past participle, and maybe by extension, we can say, not present on words of similar category.
    You probably missed my comment on this:
    "... the unextended past participles in -ta of OP evolved into the simple past of Modern Persian having lost the final -a (and voicing t to d in specific contexts, like intervocalic among others). This form still acted as past participle in Middle Persian. So, to take fdb's example from the other thread (zāta-), its exact reflex in modern Persian is zād (Sanskrit jāta). Similarly OP kṛta- (Cf. Sanskrit kṛta-) > modern Persian kard, etc. The new Persian past participle (zāde, karde,...) was obviously an extended form."

    EDIT:
    So, yes, the standard past participle forms of Old and Middle Persian had no -ka: OP zāta- > MP zād. Obviously, New Persian past participle (zāde) is an extended (i.e. suffixed) version of it. To find what the source of that suffix was, you have to take recourse to middle Persian philology. I am, of course, unqualified to pass judgement on that, but the standard suggestion that it is a descendent of OP -ka makes sense to me because:
    1) I'd expect -ka extension to be common in this context, given how it behaved in Sanskrit
    2) How attested Old Persian -aka ending evolved, e.g. OP bandaka (cuneiform spelling: b(a)-d(a)-k(a)) > Modern bande.
     
    Last edited:

    eskandar

    Moderator
    English (US)
    His claim, as I understand it, whether true or not, is not that he made Persian the lingua franca of the Samanids and Ghaznavids, or rescued it from certain destruction, but that he took it back to the way it was used and spoken, before the influence of Arabic.
    Where did Ferdowsi ever say that? Or are you reading all that into the utterly ambiguous عجم زنده کردم بدین پارسی ? That's not how the line is typically interpreted, as far as I know. Serious studies of the work have contended that he did not deliberately avoid using Arabic words, and in any case, عجم here refers to the Persian people, not the Persian language.
     

    Treaty

    Senior Member
    Persian
    But wouldn't that be stating the obvious? At least that's not how I read it but are you (or fdb) actually saying there were two flavours of Pahlavi script, a Sassanid and a post Islamic one? I'm asking as I don't know.
    Script systems reflect the spoken language at the time when they were invented. Over time, the spoken language changes while the written form remains intact. For example, we used to pronounce و in خور (xwar-) and خواب (xwāb-). However, we haven't dropped it from the spelling despite we no longer pronounce it. Some other examples are "gh" in English, or "h" in French or Spanish.
    ..., the spoken Dari and MP could not have been different, at least not in pronouncing this group of words.
    The hubs of Dari and formal Middle Persian are separated by 1000Km and 300 years. Would you say so about, for example, 1700s Tajiki and modern Tehrani?
    the utterly ambiguous عجم زنده کردم بدین پارسی ?
    This line is not even by Ferdowsi.
     

    PersoLatin

    Senior Member
    UK
    Persian - Iran
    Hi Dib,
    Thanks for the contributions in post #9 & 15, I am trying to understand and digest the information about -ika/aka relating to past participle, they make good sense. Maybe we should separate past participles from the rest, because so far there hasn't been anything convincing said about those, I will provide a list of the latter.
     
    Last edited:

    PersoLatin

    Senior Member
    UK
    Persian - Iran
    Where did Ferdowsi ever say that? Or are you reading all that into the utterly ambiguous عجم زنده کردم بدین پارسی ? That's not how the line is typically interpreted, as far as I know. Serious studies of the work have contended that he did not deliberately avoid using Arabic words, and in any case, عجم here refers to the Persian people, not the Persian language.
    Yes عجم refers to Persians, that's fairly obvious. It is amazing that how much is read into what I said. I never said he avoided Arabic words in his works, or quoted عجم زنده کردم بدین پارسی.

    This line is not even by Ferdowsi.
    Ok, no problem, taking credit away from Ferdôsi automatically makes عجم زنده کردم بدین پارسی more credible, but so what, anyway. Whoever wrote عجم زنده کردم بدین پارسی, perhaps you can tell everyone, to me it means: "I made Persians aware of this (correct) Persian", again before I am misunderstood, 'correct Persian' doesn't mean 'void of Arabic' but maybe more authentic, and yes, less Arabic.

    Script systems reflect the spoken language at the time when they were invented. Over time, the spoken language changes while the written form remains intact. For example, we used to pronounce و in خور (xwar-) and خواب (xwāb-).
    You can count words with xwar- spelling, on fingers of one hand, words ending in g/k in Pahlavi script, amount to many hundreds, yet xwar- spelling survived but none of the group with g/k spelling did (in NP), apart from a handful in Arabic/Armenian etc.



    Contributors to this diverted thread are knowledgeable in language matters, so my question is, are we happy saying "languages change...the same happened to language x or y .. this is a normal pattern, etc", and still not expect to see any residual effect of g/k words in spoken NP? After all, some people, in/outside Iran, still pronounce that handful of xwar- words, (i.e. as they are written), but not even one reliable word ending in g/k has survived in the script or the spoken form. You need a cataclysmic event to cause this.
     

    Treaty

    Senior Member
    Persian
    Ok, no problem, taking credit away from Ferdôsi automatically makes عجم زنده کردم بدین پارسی more credible,
    This verse doesn't exist in most and oldest Shahname manuscripts. The couplet could have been made by a "nationalist" for propaganda, to make you presume (or simply reflect his own view) not only Ferdowsi's intent but also the linguistic outcome of his work. This makes your logic circular.
    You can count words with xwar- spelling, on fingers of one hand, words ending in g/k in Pahlavi script, amount to many hundreds, yet xwar- spelling survived but none of the group with g/k spelling did (in NP), apart from a handful in Arabic/Armenian etc.
    I'm not sure which of us can't understand the other. The number is not really important. It is about principle and common sense. There are three simplified principles:
    1. If the sound [X] existed when the script Y was adopted, then the letter X would be included in Y.
    2. If the sound [X] disappeared while the script Y was still in use, then it is likely that the letter X continues to remain in Y.
    3. If the sound [X] has disappeared before the script Y is adopted, the letter X will not be in the script Y.
    Basically, if you see mute و in xw- words, it is a case of principles [1] and [2]. If you don't see و in xw- words (e.g., xwadāy خدای or xwurd خرد) it would be a case of [3]. As for g/k words, if you see them in Pahlavi script it is a case of [1]. However, if after a while (mainly post-Islamic) they don't read it in Pahlavi but still write it, it is a case of [2]. Now, if you don't see g/k in Arabic script, it is a case of [3]. Same procedure has happened in a number of other languages including Persian itself (for t/θ letter before [r]: it was pronounced when Pahlavi was adopted and so was included in the script [1], it was gradually disappeared but still fairly remained in the script [2], however, by the time of Perso-Arabic adoption it had totally disappeared (from all Persian dialects) and so not included in the script [3] and replaced by هـ).
    are we happy saying "languages change...the same happened to language x or y .. this is a normal pattern, etc", and still not expect to see any residual effect of g/k words in spoken NP?
    This is not the only evidence. As I said, we have another similar change happened in Persian (t/θ in above example). But I doubt you'd say this was never pronounced. We have a number of loanwords in other languages which for some reason you keep dismissing (there are much more than just a handful). We have some of them in related languages (including older stages of Iranian, and I assume Balochi with all those -g endings, though an expert should verify this). The issue of not existing in any extant Persian dialect (if true) is not necessarily a problem because as I said, there are many other instances where a feature completely disappeared from child or related languages.
     

    Dib

    Senior Member
    Bengali (India)
    are we happy saying "languages change...the same happened to language x or y .. this is a normal pattern, etc", and still not expect to see any residual effect of g/k words in spoken NP? After all, some people, in/outside Iran, still pronounce that handful of xwar- words, (i.e. as they are written), but not even one reliable word ending in g/k has survived in the script or the spoken form. You need a cataclysmic event to cause this.
    As far as sound-changes are concerned, more often than not, they apply across the board in a given speech community at a given time. So, residuals are indeed not usual to find, not to imply that they are never found(1). There are instances of what is called "sporadic" sound changes, but they are typically far less common in the long term than "regular" sound changes. This whole principle holds true strictly for close-knit speech communities only. However, roads and empires (and now mass media) lead to large scale homogenization of languages by spreading linguistic norms of important trade/administrative centres by bringing people together on the market places, in armies, etc. It is hard to predict to what geographical extent any given sound change would end up travelling under such conditions. Making things even less predictable in case of cross-linguistically common sound-changes, the same change may start at different geographical centres quite independent of each other as well. In the long term, these factors can play together to make certain sound changes (and other linguistic changes in general) quite universal over large geographical tracts.

    As far as I understand the political history of New Persian, it is basically the Khorasani dialect which eventually displaced all the other dialects from urban areas, e.g. the would-be Larestani in Pars, the so-called Median dialects in Esfahan-Tehran stretch, Bactrian, etc. in the East, Sogdian, etc. in the North. So, any sound-change that would have occurred to Khorasani Persian before this spread (i.e. around/just after the Islamic conquest?) would be expected to be shared all over the Persian speaking urban world, and the loss of word-final -g seems to fit that bill; xw- > x- is a post-spread change, and hence not shared everywhere. If you do want to look for survival of word final -g, you'll need to look for it in the rural dialects, hopefully isolated enough from the spread of "Standard" Persian for the past 1000 years or more. So, my favourite hunting ground would be the rural dialects in the Larestani belt, not urban Persian variants. But I know virtually nothing about them to inform you whether the feature indeed survives in any of them.

    ---------

    (1) Apart from sporadic changes, there may be other "apparent" residuals. Regular sound changes are often phonetically conditioned, i.e. they take place in certain phonetic conditions (e.g. -g at the end of the word, in our case) and not others (e.g. -g- between vowels in our case). They are still "regular" because they seem to follow this "rule" of phonetic context. Within this framework, "zendengī" is taken to retain the "g", because it didn't fulfill the required phonetic condition (being word-final) for its deletion, but "zende" lost it because its g fulfilled. Historically, this sort of sound changes are an important source of the so-called "liaison" consonants. French liaison "t", for example in "Y a-t-il ...?" (is there ...?), is, historically speaking, the retention of an older "t" which was lost otherwise. Same for the 'n' in 'an apple'. Brand-new liaison consonants (i.e. not retained or derived from an older consonant) between two vowels are otherwise usually restricted to glides like y, w and maybe glottal stop and h.

    Another irregular source of apparent residual is analogical influence from related forms. If modern zende was replaced by a back-formation in *zendeg from the form zendegī, that would look like a residue. Such things do occur from time to time, but they are very unpredictable even when they do occur in terms of which items they would affect. By the way, if Middle Persian was something similar to Modern Persian in the relevant respects, it should have allowed a formation like "zende-ī" which is allowed in Modern Persian, when the -ī is the indefinite marker.
     
    Last edited:

    PersoLatin

    Senior Member
    UK
    Persian - Iran
    I'm not sure which of us can't understand the other. The number is not really important. It is about principle and common sense. There are three simplified principles:
    1. If the sound [X] existed when the script Y was adopted, then the letter X would be included in Y.
    2. If the sound [X] disappeared while the script Y was still in use, then it is likely that the letter X continues to remain in Y.
    3. If the sound [X] has disappeared before the script Y is adopted, the letter X will not be in the script Y.
    Life is complicated enough so simple is very good, thank you, this makes sense.

    However, if after a while (mainly post-Islamic) they don't read it in Pahlavi but still write it, it is a case of [2].
    What tangible evidence is there that in this period g/k was NOT pronounced? (Of course we already know it was written)

    If the sound [X] existed when the script Y was adopted, then the letter X would be included in Y.
    If we apply this principle to OP, when cuneiform was adopted, we should then expect to see g/k in OP text, but is that the case?

    I'd expect the answer to be yes, because, based on principle 1, MP must have included this in its writing system (Pahlavi), because spoken OP (replaced by MP) already had these sounds. This, by extension, implies that g/k was part of most, or all flavours of Persian from the start. (You might at point say, hallelujah, he's finally got it, but...)
     
    Last edited:

    PersoLatin

    Senior Member
    UK
    Persian - Iran
    As far as sound-changes are concerned, more often than not, they apply across the board in a given speech community at a given time. So, residuals are indeed not usual to find, not to imply that they are never found(1).
    There are certainly some around e.g. تاریک ,نزدیک or تاجیک and some others, but I don't know who's side of this argument, they help. But they clearly don't follow the rules that Pahlavi had in place, (if you believe in those, of course). I explain, the ik in nazdik, is an adjective marker, acting on nazd (close) to give 'closeness', but in NP, nazdik means 'close' and 'closeness' is nazdiki, i.e. overloading it with the NP rule for adjective making, and completely ignoring the 'ik'. The same applies to the other two.

    So, my favourite hunting ground would be the rural dialects in the Larestani belt, not urban Persian variants.
    What will be really interesting, and maybe helpful to find, is a dialect that uses a word like 'nazdik', in its Pahlavi sense, i.e. 'closeness' and not 'close'.


    (1) Apart from sporadic changes, there may be other "apparent" residuals. Regular sound changes are often phonetically conditioned, i.e. they take place in certain phonetic conditions (e.g. -g at the end of the word, in our case) and not others (e.g. -g- between vowels in our case). They are still "regular" because they seem to follow this "rule" of phonetic context. Within this framework, "zendengī" is taken to retain the "g", because it didn't fulfill the required phonetic condition (being word-final) for its deletion, but "zende" lost it because its g fulfilled. Historically, this sort of sound changes are an important source of the so-called "liaison" consonants. French liaison "t", for example in "Y a-t-il ...?" (is there ...?), is, historically speaking, the retention of an older "t" which was lost otherwise. Same for the 'n' in 'an apple'. Brand-new liaison consonants (i.e. not retained or derived from an older consonant) between two vowels are otherwise usually restricted to glides like y, w and maybe glottal stop and h.

    Another irregular source of apparent residual is analogical influence from related forms. If modern zende was replaced by a back-formation in *zendeg from the form zendegī, that would look like a residue. Such things do occur from time to time, but they are very unpredictable even when they do occur in terms of which items they would affect.
    To me it looks like, 'g', as a liaison (e.g. in zendengī), has been wrongly used in words ending in strong vowels e.g. āhūg آهو (gazelle), bānūg بانو (lady) or ārzōg آرزو (desire, wish). brūg ابرو (eyebrow) -- āsyāg آسیا (mill), āŝkārāg آشکارا (obvious, evident), as this group don't generally have adjective forms to need 'g' for liaison, nor to use for plurals. There's something odd about these.


    Does Aramaic (or also Arabic) have any word(s), of any form, that ends in a strong vowels, like ā آ ('a' in English car) and u و ('oo' in English moon)?
     

    Dib

    Senior Member
    Bengali (India)
    There are certainly some around e.g. تاریک ,نزدیک or تاجیک and some others, but I don't know who's side of this argument, they help.
    I don't know their etymology, but if they really retain MP -ik/-ig, rather than gaining the ending from another source, then that would need some explaining under the standard principles of historical lingustics.

    But they clearly don't follow the rules that Pahlavi had in place, (if you believe in those, of course). I explain, the ik in nazdik, is an adjective marker, acting on nazd (close) to give 'closeness',
    You certainly mean a "noun marker", because "closeness" is a noun. But, in the sense you are looking for, the MP suffix was -ih, not -ik, as we have already learnt in the other thread:
    I'm sure that Middle Persian had two suffixes for converting adjectives and nouns to each other: īg made relational noun or adjective and īh made essential nouns. For example, if you add īg to noun rōz it gives you rōzīg meaning "daily" or "of day", while adding īh to rōz makes rōzīh that (if existed) means "day-hood" or "quality of being day".
    However, either way, I am not sure how this evidence fits together with your claim that the word-final k/g was purely orthographic and not pronounced in MP.

    What will be really interesting, and maybe helpful to find, is a dialect that uses a word like 'nazdik', in its Pahlavi sense, i.e. 'closeness' and not 'close'.
    In light of what Treaty said, and I have verified independently, about the endings -ig/-ih, your assumption seems to be wrong.

    To me it looks like, 'g', as a liaison (e.g. in zendengī), has been wrongly used in words ending in strong vowels e.g. āhūg آهو (gazelle), bānūg بانو (lady) or ārzōg آرزو (desire, wish). brūg ابرو (eyebrow) -- āsyāg آسیا (mill), āŝkārāg آشکارا (obvious, evident), as this group don't generally have adjective forms to need 'g' for liaison, nor to use for plurals. There's something odd about these.
    As you yourself acknowledge the oddity of these words in your framework, how would you explain that (i.e. use of a "liaison" spelling, where liaison never occurs)? Note that in the standard framework, these words are no oddity. They are simply proposed to have lost the final once-pronounced g through a regular sound-change. A much smaller group of words (nazdīk, etc.) would be the odd ones requiring a special explanation, why they haven't lost it (if it is indeed a case of retention, rather than a different derivation).

    Does Aramaic (or also Arabic) have any word(s), of any form, that ends in a strong vowels, like ā آ ('a' in English car) and u و ('oo' in English moon)?
    Final -ā was very common in Aramaic. I believe, at some point, almost all nouns in Aramaic ended in an -ā (I faintly remember, it evolved from an earlier definite article or something, spelt by "alif"). For the exact timelines of this evolution, you have to consult someone else. Classical Arabic also has enough words in final -ā, e.g. those in "alif maqṣūra", though they may have evolved from earlier diphthongs -ay/-aw. I don't know about final -ū's.

    EDIT: I just remembered some final ū's in Arabic, e.g. construct state of plurals in -ūna and some other words like 'abū (father). There are also 3rd person masculine plural perfect verb forms in -ū but they are spelt in a weird way with a "silent" alif after the wāw.

    EDIT: I made a quick search in MacKenzie's Pahlavi dictionary. It seems the derivation goes like this:
    tār (darkness) > tārīk/g (dark) > tārīkīh (darkness)
    nazd (near) > nazdīk (near) > nazdīkīh (nearness)

    Indeed, I am curious to find out the standard explanation of tārīk, nazdīk surviving with the final k into Modern Persian.
     
    Last edited:

    Treaty

    Senior Member
    Persian
    What tangible evidence is there that in this period g/k was NOT pronounced? (Of course we already know it was written)
    I don't know. I reiterated fdb's implication.
    If we apply this principle to OP, when cuneiform was adopted, we should then expect to see g/k in OP text, but is that the case?
    OP texts are very rare. There are however a few instances, discussed here:
    Outcomes of the Indo-Iranian suffix *-ka- in Old Persian and Avestan
     

    Dib

    Senior Member
    Bengali (India)
    What tangible evidence is there that in this period g/k was NOT pronounced? (Of course we already know it was written)
    I don't know. I reiterated fdb's implication.
    fdb provided some evidence in the other thread for the closely related Manichaean script. Let me quote:
    And we even have a few texts in the New Persian language, but written in Manichaean script. In these we have a purely graphic –g not only in Persian words like rēša (written ryšg), but even Arabic words like jumla (written jwmlg, where the –g is totally un-etymological). You can read about it here: Dictionary Of Manichaean Vol 2 : Francois de Blois, Nicholas Sims-Williams : Free Download & Streaming : Internet Archive pp. 89 sqq.
     

    PersoLatin

    Senior Member
    UK
    Persian - Iran
    You certainly mean a "noun marker", because "closeness" is a noun
    Thanks Dib, yes I did.

    In light of what Treaty said, and I have verified independently, about the endings -ig/-ih, your assumption seems to be wrong.
    Yes, I have seen some words which have both -ih and -ig endings and having different meanings, but does that then mean, whilst we have kept -ih in NP with-i, we have lost the function that -ig provided in MP, so it is not just the case that the word-final k/g was purely orthographic and not pronounced in MP?

    EDIT: I made a quick search in MacKenzie's Pahlavi dictionary. It seems the derivation goes like this:
    tār (darkness) > tārīk/g (dark) > tārīkīh (darkness)
    nazd (near) > nazdīk (near) > nazdīkīh (nearness)

    Indeed, I am curious to find out the standard explanation of tārīk, nazdīk surviving with the final k into Modern Persian.
    Also, we should expect to see tārih (darkness), and nazdih (nearness, closeness) but they seem to be missing in MP.
     

    PersoLatin

    Senior Member
    UK
    Persian - Iran
    If this is true:
    EDIT:
    So, yes, the standard past participle forms of Old and Middle Persian had no -ka: OP zāta- > MP zād. Obviously, New Persian past participle (zāde) is an extended (i.e. suffixed) version of it.
    then can this be?
    I think this has already been discussed in another thread. Anyway, an evidence for the pronunciation comes from the ج (e.g., نموذج برنامج طازج) or ق (e.g., فستق خندق ابستاق) at the end of Persian loans in Arabic;
    At least one, maybe two, of these words is a past participle; خندق is the Arabized of کنده (pp of کندن, to dig/excavate) which must have existed as 'kandag' to give خندق.
     

    Dib

    Senior Member
    Bengali (India)
    Under the standard disclaimer:

    So, yes, the standard past participle forms of Old and Middle Persian had no -ka: OP zāta- > MP zād. Obviously, New Persian past participle (zāde) is an extended (i.e. suffixed) version of it. To find what the source of that suffix was, you have to take recourse to middle Persian philology. I am, of course, unqualified to pass judgement on that, ...
    and the observation:

    Also, remember, the unextended past participles in -ta of OP evolved into the simple past of Modern Persian having lost the final -a (and voicing t to d in specific contexts, like intervocalic among others). This form still acted as past participle in Middle Persian. So, to take fdb's example from the other thread (zāta-), its exact reflex in modern Persian is zād (Sanskrit jāta). Similarly OP kṛta- (Cf. Sanskrit kṛta-) > modern Persian kard, etc. The new Persian past participle (zāde, karde,...) was obviously an extended form.
    I'd speculate:

    I am curious, does this extended past participle occur at all in MP? I expect their precursors (i.e. adjective formations not yet interpreted as past participles) in -ag to appear though (Cf. Sanskrit adjective derivations in part participle+ka-, e.g. jātaka- new-born, kṛtaka- artificial).
    which would then be the source of forms like خندق.
     

    PersoLatin

    Senior Member
    UK
    Persian - Iran
    Thanks Dib.

    Any thoughts on?
    Yes, I have seen some words which have both -ih and -ig endings and having different meanings, but does that then mean, whilst we have kept -ih in NP with-i, we have lost the function that -ig provided in MP, so it is not just the case that the word-final k/g was purely orthographic and not pronounced in MP?
     

    Dib

    Senior Member
    Bengali (India)
    Yes, I have seen some words which have both -ih and -ig endings and having different meanings, but does that then mean, whilst we have kept -ih in NP with-i, we have lost the function that -ig provided in MP, so it is not just the case that the word-final k/g was purely orthographic and not pronounced in MP?
    No. You still have both the functions performed by one single suffix, -ī, in New Persian (e.g. bāzār > bāzārī, that's the function of MP -īg; xūb > xūbī, that's the function of MP -īh). If the thesis of dropping of final MP -g is correct, then it's a simple case of phonetic merger of the two MP suffixes -īg and -īh through regular sound changes. In Modern Iranian Persian, there is also the third -ī suffix (or probably a clitic?) that marks indefiniteness, which derives from early New Persian -ē (you may ignore this, since you are majhul-sceptic, without the loss of the general overview) from Middle Persian -ēw from Old Persian aiva (one), spelt a-i-v(a) in cuneiform.
     
    Last edited:

    PersoLatin

    Senior Member
    UK
    Persian - Iran
    ^ Thank you Dib.

    Persian -ēw from Old Persian aiva (one), spelt a-i-v(a) in cuneiform.
    Very interesting, this is just a thought, is it possible that the NP for one, i.e. yek, has also been subjected to -ik?


    I think there are three groups of MP words ending g/k:
    1 - Words that in NP, end in eh(é) and in OP, most probably end in 'a', example: zendé/zendag, setāré/setārag etc.
    2 - Words ending in -ig, examples: ruzig, arzānig (worthy) or angustarig (ring).
    3 - And these: āhūg آهو (gazelle), bānūg بانو (lady) or ārzōg آرزو (desire, wish). brūg ابرو (eyebrow) -- āsyāg آسیا (mill), āŝkārāg آشکارا (obvious,evident), etc.

    For groups 1 & 2, you have provided very good arguments, enough to keep me out of trouble for a good while, but for the third group, there hasn't been much said. What functional role, if any, does g/k play in these?
     

    Treaty

    Senior Member
    Persian
    We have some of the group 3 words in New Persian ending with ی (e.g., آهوی) even when not followed by an ezaafe. So, it is possible that -g has changed into -y and then disappeared (I know only one word with final g>y: bag>bay "god" but there are other examples of disappeared g in favor of a y-like vowel: tigr>tīr, dagr>dēr). The final y is disappeared even when it doesn't correspond with g (e.g., روی>رو, پای>پا).

    Of course the question is how those g in group 3 got there in the first place. The OP suffix is -ka that can attach to all types of ending (including even consonants, like huška خشک). In this case, there wouldn't be any difference between 1 and 3 (that is if your examples in 3 are made by the same suffix).

    Nevertheless, it should be noted that apart from loanwords and a few conjunctions and pronouns, there is almost no word in MP ending with a vowel. There is always a [g], [k] or [y] at the end of what we now pronounce with a vowel ending. While this brings suspicion (i.e. possible compensating the lack of [a]-letter by final [g], or differentiating two ī suffixes by and [g], though this pedanticness betrays the looseness of the script), the [g] ending didn't make sense for words final [ā] and round vowels as there were letters representing them.
     

    PersoLatin

    Senior Member
    UK
    Persian - Iran
    1 - Words that in NP, end in eh(é) and in OP, most probably end in 'a', example: zendé/zendag, setāré/setārag etc.
    2 - Words ending in -ig, examples: ruzig, arzānig (worthy) or angustarig (ring).
    3 - And these: āhūg آهو (gazelle), bānūg بانو (lady) or ārzōg آرزو (desire, wish). brūg ابرو (eyebrow) -- āsyāg آسیا (mill), āŝkārāg آشکارا (obvious,evident), etc.
    I think in all above groups, g represented i/y/ی and nothing else, but the function was slightly/subtly different.

    We have some of the group 3 words in New Persian ending with ی (e.g., آهوی) even when not followed by an ezaafe.
    Exactly, g represents i/y/ی, and never pronounced as g, only by misreading or misunderstanding of its function by later people. And when used with i, i.e. in -ig, it just stretched i as in your very good example ( tigr>tīr, dagr>dēr) and many more like them. It looks like that -ig can really be represented by the modern ē.


    With reference to the following well put-together excerpt, from Dib's post:
    You still have both the functions performed by one single suffix, -ī, in New Persian (e.g. bāzār > bāzārī, that's the function of MP -īg; xūb > xūbī, that's the function of MP -īh). If the thesis of dropping of final MP -g is correct, then it's a simple case of phonetic merger of the two MP suffixes -īg and -īh through regular sound changes. In Modern Iranian Persian, there is also the third -ī suffix (or probably a clitic?) that marks indefiniteness, which derives from early New Persian -ē (you may ignore this, since you are majhul-sceptic, without the loss of the general overview) from Middle Persian -ēw from Old Persian aiva (one), spelt a-i-v(a) in cuneiform.
    In NP, we currently have 3 i's, which act as; indefinite marker (bāzārī - a market) , noun marker (xubi - goodness/good deed) and associative marker (bāzārī - from/of the bāzār (-ig)). An indefinite 'person from the bāzār', is bāzārī-i (an indefinite 'good deed', is xubi-i). It is not too hard to believe that Pahlavi provided a g mechanism, for MP speakers, to make a distinction between these, purely as a marker to tell the difference, especially, when such constructions were used (MP) arzānigih.

    Also,
    Same for the 'n' in 'an apple'. Brand-new liaison consonants (i.e. not retained or derived from an older consonant) between two vowels are otherwise usually restricted to glides like y, w and maybe glottal stop and h.
    What Dib said above is correct and very evident in NP , i.e. y as liaison, xodāyā, pāye man, beguyam etc. How is that relevant to this group? So if we accept g was really a different i/y, we can see that it provides a liaison function in this group, but again as a i/y. As for its persistence in NP (zendegi, setāregān), I can image we can site the same reason we use for those few words that have survived (nazdik, tārik, tājik, خندق).
     
    Last edited:

    Dib

    Senior Member
    Bengali (India)
    As for its persistence in NP (zendegi, setāregān), I can image we can site the same reason we use for those few words that have survived (nazdik, tārik, tājik, خندق).
    Wait a minute. In your scheme of things, these are not survivals. You are proposing (something like) y where these words have g/k. So, you need a mechanism to explain how y became k/g. That is not a common sound change. You need to propose the intermediate stages involved and produce historical evidence of their presence. Or, am I missing something?
     

    Treaty

    Senior Member
    Persian
    I think in all above groups, g represented i/y/ی and nothing else, but the function was slightly/subtly different.
    Exactly, g represents i/y/ی, and, in my view, nothing else, i.e. not ever pronounced as g, only by misreading or misunderstanding of its function by later people
    It is true that in Book Pahlavi, the letters for [g] and [y] were similar and so your theory would have made sense. Indeed in Manichaean texts, most of these endings were written by this dubious letter. However, the problem with your suggestion is that in other Book Pahlavi texts (as well as Sogdian texts), they used the letter [k] for these endings. In other terms, it seems they deliberately used [k] in order not to misread it as [y] or they genuinely pronounce the [k] not [g] (and definitely not [y]). In addition, in Inscriptional Pahlavi, [g] and [y] had separate letters. In early Sassanid inscriptions, they used the letter [k] for those endings, not [y] or [g].
     

    PersoLatin

    Senior Member
    UK
    Persian - Iran
    Wait a minute. In your scheme of things, these are not survivals.
    They act as evidence that g/k was read as g/k, rather than what was the intended sound, i.e. i/y.

    So, you need a mechanism to explain how y became k/g.
    If this was a convention, chosen by the experts of the time, it will not be subject to the normal sound change rules.

    You need to propose the intermediate stages involved and produce historical evidence of their presence. Or, am I missing something?
    Well, this is not yet a fully functional and working proposal, the missing parts will hopefully be filled.
     

    PersoLatin

    Senior Member
    UK
    Persian - Iran
    However, the problem with your suggestion is that in other Book Pahlavi texts (as well as Sogdian texts), they used the letter [k] for these endings. In other terms, it seems they deliberately used [k] in order not to misread it as [y] or they genuinely pronounce the [k] not [g] (and definitely not [y]). In addition, in Inscriptional Pahlavi, [g] and [y] had separate letters. In early Sassanid inscriptions, they used the letter [k] for those endings, not [y] or [g].
    The current accepted theory about g/k, explains a general idea but at detail level, it falters, I refer you to words in group 3. Given time I can extract scores of these, of course you may say, sheer numbers don't prove anything and you'd be right but with a high number of group 3 words, OP conversations must have sounded a little odd, as every few words ended in g, and peppered with scores of common words with -ig. In fact it is not too hard to reconstruct a piece of contemporary text, in that style. Of course, old languages will sound odd to us but...

    The MP word anōŝag (hamiŝé - immortal/always), has the obligatory g, but it seemingly goes missing in anoushirvan (immortal spirit, title of a Sassanid king) but at closer inspection, you will see it as an i, i.e. anōŝēravān or anōŝīravān. Of course as ravān doesn't start with a vowel, there's no need for g as a liaison. However when you add the noun marker i, to anōŝa, you'd see the g, which was meant to be i/y, but it has remained in the word as g, so hamiŝegi.

    The MP word for Avestā, i.e. abestāg is interesting, I would imagine if g was pronounced, there was no way later priests would drop it, whatever the reason.

    These g in these is unexplained: ŝagr (ŝir, lion), sagr(sir, satiataed), dagr (dir, late), tigr (tir, arrow). but this explanation fits: 'ag' represented ē and, as said before, 'ig' represented ī. I have seen this list of words, in the discussions about classical Persian pronunciation.
     
    Last edited:

    Treaty

    Senior Member
    Persian
    These g in these is unexplained: ŝagr (ŝir, lion), sagr(sir, satiataed), dagr (dir, late), tigr (tir, arrow). but this explanation fits: 'ag' represented ē and, as said before, 'ig' represented ī. I have seen this list of words, in the discussions about classical Persian pronunciation.
    In fact, your examples are evidence that 'g' was a part of the word and pronounced at some stage (apart from sagr which I don't know its etymology):
    - as for dagr and šagr, it is a metathesis of rg>gr (e.g., Av. darǝɣa, Parth. šarg, Sogd. šarɣu).
    - as for tīr and sīr (garlic) [g] was a part of their etymology (OP ti-ga-ra cognate of تیغ, and θa-i-ga-ra).

    Basically, there is a pattern of [g] being disappeared (e.g., nagan>nān نان) or turned into a gliding sound [y] or [w] in Persian (e.g., raga>ray ری). The questions are whether, where and when letter 'g' was pronounced /g/ with stop or something more liquid on the way to /j/. However, saying ag represented [ē] is more problematic as most of words with [ē] are represented by letter 'y'. There was no reason to use a longer and confusing 'ag' for this function.
    The MP word anōŝag (hamiŝé - immortal/always)
    anōšag comes from an (not) ōš + (death) + ag, while hammēšag is ham (for emphasis like همینک?)+ mēšag (always). If you accept that there was a g>y shift MP as shown above, then anōšagruwān would have naturally turned into anōšēruwān overtime.
    The MP word for Avestā, i.e. abestag is interesting, I would imagine if g was pronounced, there was no way later priests would drop it, whatever the reason.
    Just remember that the name of their god was changed from ahūramazda to hurmuz in Persian. I don't think a final [g] really mattered to them.
     

    PersoLatin

    Senior Member
    UK
    Persian - Iran
    - as for tīr and sīr (garlic) [g] was a part of their etymology (OP ti-ga-ra cognate of تیغ, and θa-i-ga-ra).
    Please don't forget, evidence like this, is exactly the type that's under question here, and doesn't help neither side of the argument, least of all, the opposing one.

    If you accept that there was a g>y shift MP as shown above, then anōšagruwān would have naturally turned into anōšēruwān overtime.
    I don't believe it is a shift, I believe g was a y in the first place. Of course you are going to see a g, in anōšagruwān, in MP, but isn't that the whole point of this discussion. The work of correction (not shift) of g to y happened, as evidence shows in post Islamic Pahlavi, and around the time of Perso-Arabic adoption, which is probably when the discrepancy between the pronunciation such words by of the masses, and Pahlavi script was spotted.

    anōšag comes from an (not) ōš + (death) + ag, while hammēšag is ham (for emphasis like همینک?)+ mēšag (always).
    Although this is incidental in this discussion, it is possible that two words that sound and mean (almost) the same, come from two completely different sources.

    ahūramazda to hurmuz in Persian. I don't think a final [g] really mattered to them.
    Of course it would, they were the definition of 'conservatism', ok in time, they might have accepted it, but a footprint of the old name, will always be recorded somewhere.

    If you drop the short vowels (a) from ahūramazda, you'd get hurmuzd and you can't prove those vowels were ever there, so hurmuzd is much closer to the original pronunciation. You just have to look at bozorg (great) in cuneiform, v-z-r-g, we are told it was pronounced vazaraga, which is just as valid as a pronunciation as ahūramazda, is.


    - as for dagr and šagr, it is a metathesis of rg>gr (e.g., Av. darǝɣa, Parth. šarg, Sogd. šarɣu).
    The MP word dagrand (long, derived from dagr/dēr) which I am sure is the modern dērang (derang), has gone through more than the above shift which to me looks like a correction, I'm not doubting the shift, in other words like hagriz to hargez, but tigr, sagr, dagr and šagr can't be explained by it.
     
    Last edited:

    Treaty

    Senior Member
    Persian
    Please don't forget, evidence like this, is exactly the type that's under question here...
    Can you please expand on the etymology of tēx تیغ and tigr تیر.
    This is the answer for that question. Check here. Some other examples are "Av. taēɣa-, taēža- `sharp ... sharpness', Germ. *Þī̆hstila- `thistle'" (from Pokorni II 1015-6).
    The MP word dagrand (long, derived from dagr/dēr) which I am sure is the modern dērang (derang)
    I'm not sure. We have دیرند in NP that has the same meaning as MP's dērind/dagrind. I'm not sure if it is the same word as NP derang.
    I'm not doubting the shift, in other words like hagriz to hargez, but tigr, sagr, dagr and šagr can't be explained by it.
    Why? Because it doesn't fit your theory even if they have "be recorded somewhere" evidence which you emphasize on?
    Of course it would, they were the definition of 'conservatism', ok in time, they might have accepted it, but a footprint of the old name, will always be recorded somewhere.
    Would you explain this conservatism about the following religious terms and figures: spanta-dāta>esfandyār, štāspa>guštāsp, warhrān>wahrām, zaraθuštra>zartušt? Or did those magi have some special attachment for that [g] (which was naturally fading according to my theory) so that they were expected to declare jihad for this one but not to all of the other terms?
     
    Last edited:

    PersoLatin

    Senior Member
    UK
    Persian - Iran
    I think one point at a time is the best way:
    Why? Because it doesn't fit your theory even if they have "be recorded somewhere" evidence which you emphasize on?
    Please explain how changes in sagr, dagr and šagr, fit the main stream/your shift theory?

    I'll try answering it in my way & I am hoping you will correct my mistakes: I would switch g and r (1), take out the vowels (2), and get srg, drg and šrg, then I remove g (3), and I get sr, dr, šr, I then have to add at least one vowel, I add an 'i' (e, o, a & their variants should work too)(4), I get sir, dir and šir, these look and sound like the NP versions so the theory must be correct.
     
    Last edited:

    Treaty

    Senior Member
    Persian
    I recommend to ask it in etymology forum to get scholarly answers. My answers are based on a limited corpus of articles and dictionaries available to me (this handy article focuses exclusively on sound change. Examples for ag>ē are presented on page 9). The problem is that for every single word in your question, there is a post-long story. But the most important thing about etymology is not that how many times a word is found in a language, but how many languages have that word (or cognates). For example, regarding šagr, only one old language (Middle Persian) has the sequence šgr while four others (both western and eastern Iranian, in addition to, rarely in Persian itself) have the šrg(w) sequence. This points to a significantly higher probability of šrgw to be the original Iranian word (I wonder if šarza شرزه, if not Arabic, is a cognate but saved from metathesis because of an earlier g>z). This would explain OP *šargu- in two personal names written in Elamite (to mean شیرزاد and شیرکش). Why this particular [rg] became [gr] (only in šarg and darg- but not sagr and others which were [gr] originally), while others (marg, barg, gorg) were retained, is another thread-worthy question (maybe it was just a sporadic change, or maybe it was because the origin of the others was, if I'm not wrong, [rk] not [rg]).

    I'll try answering it in my way & I am hoping you will correct my mistakes: I would switch g and r (1), take out the vowels (2), and get srg, drg and šrg, then I remove g (3), and I get sr, dr, šr, I then have to add at least one vowel, I add an 'i' (e, o, a & their variants should work too)(4), I get sir, dir and šir, these look and sound like the NP versions so the theory must be correct.
    If you only had these three words, and only in Persian, then your proposal would have been intriguing.
     

    PersoLatin

    Senior Member
    UK
    Persian - Iran
    As I said, one point at a time, and I'm certainly not ignoring your contributions in the above post (#44):
    3 - And these: āhūg آهو (gazelle), bānūg بانو (lady) or ārzōg آرزو (desire, wish). brūg ابرو (eyebrow) -- āsyāg آسیا (mill), āŝkārāg آشکارا (obvious, evident), etc.
    How is this group of words explained by the the main stream/your theory, please? (I appreciate you have already made some comments on these, so a short answer will be sufficient)
     

    Dib

    Senior Member
    Bengali (India)
    Please explain how changes in sagr, dagr and šagr, fit the main stream/your shift theory?
    I am not familiar with these etymons. But it looks like a simple sound change of MP -agr- > early NP -ēr-. We need more examples to verify, strengthen or refute this proposition.

    I'll try answering it in my way & I am hoping you will correct my mistakes: I would switch g and r (1), take out the vowels (2), and get srg, drg and šrg, then I remove g (3), and I get sr, dr, šr, I then have to add at least one vowel, I add an 'i' (e, o, a & their variants should work too)(4), I get sir, dir and šir, these look and sound like the NP versions so the theory must be correct.
    So, where are all these changes taking place? In writing, or in speaking? I doubt you can seriously "take out the vowels (2)" in speaking. So, I assume you mean in writing/reading. Now, tell me, how many 2 years old Persian children were reading books back then in order to learn to speak, that they would pick up specific characteristics of the spelling in their speech?

    The current accepted theory about g/k, explains a general idea but at detail level, it falters, I refer you to words in group 3. Given time I can extract scores of these, of course you may say, sheer numbers don't prove anything
    How is group 3 such a problem? Their modern Persian reflex can be explained by the same rule as the other 2 groups: word-final g after a vowel is lost. Simple. So, finding scores of counter-examples is certainly more significant than having one or two stray instances, but here there is no counter example. The real counter-examples are "nazdik", et al. If you find more of those, that would be a challenge to this theory.

    and you'd be right but with a high number of group 3 words, OP conversations must have sounded a little odd, as every few words ended in g, and peppered with scores of common words with -ig.
    Please be careful about the stages of languages when talking etymology and history of languages. OP and MP are very distinct stages. When you confuse them, it becomes all that more difficult to take your ideas seriously. These are MP, not OP. OP would have inflectional endings after those stems. As a matter of fact, as far as the pronunciation can be reconstructed, there were no words in OP that ended in a consonant other than m/n (+maybe glides y/w), or even if they did, those final consonants were not written.

    In any case, on the one hand we have your feeling of weirdness*, on the other we have actual written k/g's in the texts in 2 traditions - Pahlavi and Manichaean - using clearly distinct spelling conventions, i.e. not simply copying from the other. In addition, we have a good etymological theory, where this written k/g might have come from (PIA *-ka-). On top of that, we need only one single sound-change to explain why this word-final post-vocalic k/g does not survive in Modern Persian. Words like sagr/dagr also follow the same logic (written attestation, known etymology with -g-), but understandably their g has a slightly different evolutionary path to NP because of its phonetic environment.

    In fact it is not too hard to reconstruct a piece of contemporary text, in that style. Of course, old languages will sound odd to us but...
    So what?

    The MP word anōŝag (hamiŝé - immortal/always), has the obligatory g, but it seemingly goes missing in anoushirvan (immortal spirit, title of a Sassanid king), but at closer inspection, you will see it as an i, i.e. anōŝēravān or anōŝīravān.
    I don't really know its MP form. But this wiki page suggests it was Anōšagruvān:
    Anushirvan - Wikipedia, the free encyclopedia

    As a matter of fact, here also the MP -agr- > eNP -ēr- sound change seems to be working. If the g was not present in the MP form, then the bolded vowel in Anōŝērvān would be expected to be an /a/.

    ======

    * I suspect, you are grossly exaggerating the frequency of -vowel+g words in Middle Persian. Here is a piece I found from "Kār-Nāmag ī Ardašīr ī Pābagān" as quoted in wikipedia:
    pad kārnāmag ī ardaxšīr ī pābagān ēdōn nibišt ēstād kū pas az marg ī alaksandar ī hrōmāyīg ērānšahr 240 kadag-xwadāy būd. spahān ud pārs ud kustīhā ī awiš nazdīktar pad dast ī ardawān sālār būd. pābag marzobān ud šahryār ī pārs būd ud az gumārdagān ī ardawān būd. ud pad staxr nišast. ud pābag rāy ēč frazand ī nām-burdār nē būd. ud sāsān šubān ī pābag būd ud hamwār abāg gōspandān būd ud az tōhmag ī dārā ī dārāyān būd ud andar dušxwadāyīh ī alaksandar ō wirēg ud nihān-rawišnīh ēstād ud abāg kurdān šubānān raft.
     
    Last edited:

    Dib

    Senior Member
    Bengali (India)
    How is this group of words explained by the the main stream/your theory, please? (I appreciate you have already made some comments on these, so a short answer will be sufficient)
    Are you asking how their (āhūg آهو (gazelle), bānūg بانو (lady) or ārzōg آرزو (desire, wish). brūg ابرو (eyebrow) -- āsyāg آسیا (mill), āŝkārāg آشکارا (obvious, evident)) subsequent evolution to New Persian is to be explained, or their history in pre-MP? I think the first part has been amply explained. For the second, you should better start etymology thread for each of them in the etymology forum.
     

    PersoLatin

    Senior Member
    UK
    Persian - Iran
    How is group 3 such a problem? Their modern Persian reflex can be explained by the same rule as the other 2 groups: word-final g after a vowel is lost
    But it was not lost, it was corrected to its intended sound i.e. the misrepresented i/y. As for numbers, there are scores & scores of them, please search a Pahlavi dictionary for āg, ūg and ōg, you'll get tired of pressing the search button soon and you are still on words starting in h, you'll need many hours, if you also search for words ending ig and ag, and ones containing agr and igr (not many of these, 7-8), of course extra time is required as you'll need to know which qualifies which doesn't.

    I suspect, you are grossly exaggerating the frequency of -ag/-ig words in Middle Persian.
    please refer to above and judge for yourself.

    please be careful about the stages of languages when talking etymology and history of languages. OP and MP are very distinct stages.
    Yes, sorry, I meant MP.

    In any case, on the one hand we have your feeling of weirdness*, on the other we have actual written k/g's in the texts in 2 traditions - Pahlavi and Manichaean - using clearly distinct spelling conventions, i.e. not simply copying from the other.
    I'm not being difficult for the sake of it, but can we really be sure of this?

    The real counter-examples are "nazdik", et al. If you find more of those, that would be a challenge to this theory.
    I remembered more: Manichaean - Māni was the religious leader's name, g (y really) was added to make Manig/Manich and the Westernise version of the name, added another assocative suffix to arrive at Mani+ch+aean. We also have Mazdak, bārik (narrow, thin), doruq (lie, drō in MP), I'll add more as I come across them, maybe other members can help here.

    I don't really know its MP form. But this wiki page suggests it was Anōšagruvān:
    Its MP form is anōšagruwān, but isn't the presence of 'ag' the whole point of this discussion?

    As a matter of fact, as far as the pronunciation can be reconstructed, there were no words in OP that ended in a consonant other than m/n (+maybe glides y/w), or even if they did, those final consonants were not written.
    I don't understand this, can you please explain the bold part for me, with a couple of examples?
     

    Dib

    Senior Member
    Bengali (India)
    But it was not lost, it was corrected to its intended sound i.e. the misrepresented i/y.
    That's your theory. The standard theory believes the Middle Persian spellings in this case, because there doesn't seem to be any good reason not to.

    As for numbers, there are scores & scores of them, please search a Pahlavi dictionary for āg, ūg and ōg, you'll get tired of pressing the search button soon and you are still on words starting in h, you'll need many hours, if you also search for words ending ig and ag, and ones containing agr and igr (not many of these, 7-8), of course extra time is required as you'll need to know which qualifies which doesn't.
    Sure, but within the standard theory, they are no challenge. We simply acknowledge that Middle Persian sounded different from Modern Persian.

    I suspect, you are grossly exaggerating the frequency of -vowel+g words in Middle Persian.
    please refer to above and judge for yourself.
    The above tells us the "number of items" in a dictionary. It's not same as how frequently they actually appear in a discourse, and make that discourse "sound odd", as you put it. I am not claiming that the text-excerpt I quoted is representative, but it has around 70 words excluding the ī's and ud's, and around 95 including them. Of these only 9 end in vowel+g. If you have a more representative statistic, I'd be interested to see.

    we have actual written k/g's in the texts in 2 traditions - Pahlavi and Manichaean - using clearly distinct spelling conventions, i.e. not simply copying from the other.
    I'm not being difficult for the sake of it, but can we really be sure of this?
    They clearly have significant differences in spellings, which - within the standard theory - is consistent with the fact that the Pahlavi tradition starts around 3rd century BCE and the Manichaean around 3rd century AD. The Pahlavi spellings seem to be archaic in ways that are consistent with the reconstruction of the evolution of the language. There are also other differences, like usage of huzvarishn, etc. In our particular case of word final vowel+k/g, Pahlavi has -k (which is consistent as an older form, if Indo-Iranian *-ka- is indeed the source), while Manichaean has -g, which would then be the "classical" Sassanian Middle Persian value (3rd century AD onwards). I found one exception: our old nemesis - "nazdīk", which has -k in Manichaean as well.

    I remembered more: Manichaean - Māni was the religious leader's name, g (y really) was added to make Manig/Manich and the Westernise version of the name, added another assocative suffix to arrive at Mani+ch+aean.
    Source of your claim? Wiki seems to disagree. It says Greek Manikhaios is from Syriac "Mānī ḥayyā" (The living Mani).

    We also have Mazdak, bārik (narrow, thin),
    Good. So, now we have a potential counter-example list of nazdīk, tārīk, bārīk, Mazdak. What exactly is Mazdak though? Also, let me add, Modern Persian diminutive -ak. Though I doubt it is a survival from MP, it is worth investigating further.

    doruq (lie, drō in MP)
    It is a different breed. The MP form has no final -k/g. So, it doesn't belong here. Maybe discussed in a separate etymology thread, though.

    I'll add more as I come across them, maybe other members can help here.
    That would be great.

    Its MP form is anōšagruwān, but isn't the presence of 'ag' the whole point of this discussion?
    Okay, I reread what you said. Seems, I misunderstood you. Anyways, this word's MP>NP development is naturally explained in "my" theory (MP -agr- > NP -ēr-) as well as yours (MP g represents y in this word). I prefer mine because the MP spelling fits it naturally without any jugglery.

    As a matter of fact, as far as the pronunciation can be reconstructed, there were no words in OP that ended in a consonant other than m/n (+maybe glides y/w), or even if they did, those final consonants were not written.
    I don't understand this, can you please explain the bold part for me, with a couple of examples?
    Actually I forgot another consonant that OP words frequently end in: -š.

    Consider this text for example (Darius Behistun I.1-15):
    TITUS Didactica: Old Persian Text Sample

    I am making a couple of small changes to the phonetic interpretation given on the linked page to fit it to the way I prefer it, plus adding punctuation:

    adam dārayava(h)uš xšāyaϑiya vazr̥ka, xšāyaϑiya xšāyaϑiyānām, xšāyaϑiya pārsai, xšāyaϑiya dahyūnām, vištāspahyā puça, aršāmahyā napā, haxāmanišiya.
    ϑātiy dārayava(h)uš xšāyaϑiya:
    manā pitā vištāspa. vištāspahyā pitā aršāma. aršāmahyā pitā ariyārāmna. ariyārāmnahyā pitā čispaiš pitā haxāmaniš.
    ϑātiy dārayava(h)uš xšāyaϑiya:
    avahya-rādiy vayam haxāmanišā ϑahyāmahay. hačā paruviyata āmātā a(h)mahī. hačā paruviyata hyā a(h)māxam taumā xšāyaϑiyā āha.
    ϑātiy dārayava(h)uš xšāyaϑiya:
    (aštā) manā taumāyā tyay paruvam xšāyaϑiyā āha. adam navama. (navā) duvitāparanam vayam xšāyaϑiyā a(h)mahī.
    ϑātiy dārayava(h)uš xšāyaϑiya:
    vašnā a(h)uramazdāha adam xšāyaϑiya ahmiy. a(h)uramazdā xšaçam manā frābara.
    ϑātiy dārayava(h)uš xšāyaϑiya:
    imā dahyāva tyā manā patiyāiša vašnā a(h)uramazdāha, adam-šām xšāyaϑiya āham: pārsa, (h)uvja, bābiruš, aϑurā, ārabāya, mudrāya, tyaiy drayahyā - sparda yauna [māda] ārmina ...
     
    Last edited:

    Treaty

    Senior Member
    Persian
    As I said, one point at a time, and I'm certainly not ignoring your contributions in the above post (#44):
    How is this group of words explained by the the main stream/your theory, please? (I appreciate you have already made some comments on these, so a short answer will be sufficient)
    In Book Pahlavi (non-Manichaean), Inscriptional Pahlavi and Sogdian texts all of these ending were written by [k] despite they also had [g] and [y] letters. This means they were (or had been) pronounced as [k].

    Why they disappeared? Sound shift.

    Why some were left intact or with little change? Maybe sporadic change, considering how few these exceptions are (دروغ is not one of them). So, the list is so far:
    تاریک بزرگ تاجیک باریک and maybe خوراک پوشاک. Of these examples, تاجیک had lost it adjective construct and turned into a demonym, and بزرگ was probably saved by a metathesis of ra>ar (vazrak>vazark) and so the final [k] was saved. تاریک does actually have its other version تاری.
     
    < Previous | Next >
    Top