PIE gender
PIE gender
All the reconstructions of PIE I've seen have the two genders of Hittite. But it seems to me that by Occam's Razor it would make more sense to say that PIE had the masculine, feminine, and neuter, and that Hittite merged the feminine with the masculine, since the other old IE languages-Sanskrit, Latin, Greek, Gothic, and on and on and on-differentiate them. So what evidence do we have that it's two genders and not three?
I think it's the fact that the three genders of PIE proper look like they have been derived from a 2 gender system, and the 2 gender system looks like it's old. For example the plurals of the neuter are the same as the singulars of the feminine, and the neuters can never be the subject of a sentence in many early IE languages, so it makes sense that they were originally "inanimate" rather than neuter. And if you have an inanimate gender it's common to have only one other gender, the animate. It's really a mystery though how PIE managed to suddenly segregate almost all the nouns denoting male things into the masculine gender and almost all the nouns denoting female things into the feminine, so maybe there's more to it than meets the eye.
Sunàqʷa the Sea Lamprey says:
There's not a trace of the neuter in any Brythonic language. Should we therefore assume that proto-Celtic had no neuter?Etherman wrote:I would think that Occam's Razor says the exact opposite. Since there's not a single trace of the feminine in any Anatolian language, it's simpler to assume that it never had it than to assume it had it and lost it.
(I am not coming down on one side or another of the debate here, I'm as orthodox an IEist as the next man, but I don't think that bringing Occam's razor into it is necessarily convincing.)
Salmoneus wrote:(NB Dewrad is behaving like an adult - a petty, sarcastic and uncharitable adult, admittedly, but none the less note the infinitely higher quality of flame)
According to Wikipedia, the Anatolian languages were probably the first to split off from PIE; they may have done so before the gender switch.
"A positive attitude may not solve all your problems, but it will annoy enough people to make it worth the effort."
–Herm Albright
Even better than a proto-conlang, it's the *kondn̥ǵʰwéh₂s
–Herm Albright
Even better than a proto-conlang, it's the *kondn̥ǵʰwéh₂s
The Masculine and Neuter forms obviously look derived from an Animate and Inanimate forms, respectively, with the Inanimate forms having the same endings in the NOM and ACC, while the Animate NOM.SG *-s seems derived from an agentive definite article of some sort.
The Feminine evolved out of the Neuter/Inanimate plural in *-eh2, IIRC, though I don;t really know the specifics of the diachronics
The Feminine evolved out of the Neuter/Inanimate plural in *-eh2, IIRC, though I don;t really know the specifics of the diachronics
I remember reading that is was a diminutive or collective marker of some kind, IIRC.TheGoatMan wrote:And forms in -ih₂ so it seems that *-h₂- was some kind of morpheme.TaylorS wrote:The Feminine evolved out of the Neuter/Inanimate plural in *-eh2, IIRC, though I don;t really know the specifics of the diachronics
A collective marker seems to be the scholarly consensus (at the moment)TaylorS wrote:I remember reading that is was a diminutive or collective marker of some kind, IIRC.TheGoatMan wrote:And forms in -ih₂ so it seems that *-h₂- was some kind of morpheme.TaylorS wrote:The Feminine evolved out of the Neuter/Inanimate plural in *-eh2, IIRC, though I don;t really know the specifics of the diachronics
Salmoneus wrote:(NB Dewrad is behaving like an adult - a petty, sarcastic and uncharitable adult, admittedly, but none the less note the infinitely higher quality of flame)
Occam's Razor has to apply to all the relevant data. There are a number of indications that Proto-Anatolian split off first so it's quite possible that PA never had the feminine because it was a common development after PA split. If we likewise assume that Proto-Celtic didn't have a neuter then we'd have to explain why the non-Brythonic Celtic languages do (I assume they do, but I don't really know) and why that neuter is derivable from the neuter found in other IE languages (again, assuming that it does; but I don't really know). I would think that one would be forced to assume there were independent parallel developments in Continental Celtic and Goidelic. That would be more complicated than assuming that the Brythonic languages lost it.Dewrad wrote:There's not a trace of the neuter in any Brythonic language. Should we therefore assume that proto-Celtic had no neuter?
(I am not coming down on one side or another of the debate here, I'm as orthodox an IEist as the next man, but I don't think that bringing Occam's razor into it is necessarily convincing.)
I honestly think (particularly since the birth of the Internet) that Occam's razor has to be one of the most abused principles to be trotted out on a regular basis. As a heuristic, it's fine. It is neither irrefutable nor a conclusive argument: unfortunately it seems at times that Occam's razor is to the armchair expert what "common sense" is to the man in the street.Etherman wrote:Occam's Razor has to apply to all the relevant data. There are a number of indications that Proto-Anatolian split off first so it's quite possible that PA never had the feminine because it was a common development after PA split. If we likewise assume that Proto-Celtic didn't have a neuter then we'd have to explain why the non-Brythonic Celtic languages do (I assume they do, but I don't really know) and why that neuter is derivable from the neuter found in other IE languages (again, assuming that it does; but I don't really know). I would think that one would be forced to assume there were independent parallel developments in Continental Celtic and Goidelic. That would be more complicated than assuming that the Brythonic languages lost it.
Claiming that it has to apply to "all the relevant data" is all well and good, but in this field we can never be entirely sure that we actually have all the relevant data at our fingertips (as my Brythonic example shows).
Now, for a further example. It has become something of a truism that the Anatolian languages lacked a feminine gender, which leads us to not actually look for it. However, while synchronically this might be the case, why must we also assume it diachronically? Could we not simply be looking at a case of merger- which is far from unattested in the languages for which we have an extensively documented history?
As an example, I ask you to join me in a small thought-experiment. Let us first assume, contrary to received wisdom, that Hittite has innovated by eliminating the feminine gender by merger, as a consequence of certain phonological developments. Let us imagine that the Indo-European feminine *-ah2 (underlying *-eh2) lost its final laryngeal by regular soundchange (for the loss of final laryngeals is indeed regular in Anatolian). The subsequently undercharacterized nominative *-a (like the *-a resulting from unstressed nominative *-ō of the n-stems) was supplied by analogy with the highly salient nominative -s from the other declensions. I'm sure we all recall Zwicky's hypotheses of markedness, which state that the more "unexpected, uncommon, and striking" structures (as an underspecified nominative would be) are those which are most likely to be eliminated, so this isn't too great a leap to take. Therefore we have a resulting nominative singular -aš, formally identical to the regular reflex of the PIE masculine o-stems *-os.
But no, entia non sunt multiplicanda! I hear you cry. But what if we back this up with evidence from Lycian? Perhaps it is simply the case that we didn't actually have all the relevant data to hand.
Last edited by Dewrad on Tue Jun 01, 2010 4:12 am, edited 1 time in total.
Salmoneus wrote:(NB Dewrad is behaving like an adult - a petty, sarcastic and uncharitable adult, admittedly, but none the less note the infinitely higher quality of flame)
I agree. Occam's razor is a heuristic for investigation and research; it doesn't have anything to say directly about truth. It simply says that the most useful explanation is the one that includes the fewest unproven assumptions. This is intuitively obvious: how can you verify a theory if it relies on lots of unverified assumptions? You'd have to verify them all first. So it is more useful to propose the most easily testable theory, the truthfulness or not of which is found from the testing, not from Occam's razor. One of the theories with the unproven assumptions may well turn out to be the truth, but this will show up after the otehrones have been eliminated. There is obviously a link between the likelihood of it being true and the number of unproven assumptions in the theory, but that is more due to statistics and the way science/logic works, and is a secondary correlation, not the main point of the razor and not always or predictably true. People using it as if we must accept as true the explanation selected by the razor does annoy me a lot.Dewrad wrote:I honestly think that (particularly since the birth of the Internet) that Occam's razor has to be one of the most abused principles to be trotted out on a regular basis. As a heuristic, it's fine. It is neither irrefutable nor a conclusive argument: unfortunately it seems at times that Occam's razor is to the armchair expert what "common sense" is to the man in the street.Etherman wrote:Occam's Razor has to apply to all the relevant data. There are a number of indications that Proto-Anatolian split off first so it's quite possible that PA never had the feminine because it was a common development after PA split. If we likewise assume that Proto-Celtic didn't have a neuter then we'd have to explain why the non-Brythonic Celtic languages do (I assume they do, but I don't really know) and why that neuter is derivable from the neuter found in other IE languages (again, assuming that it does; but I don't really know). I would think that one would be forced to assume there were independent parallel developments in Continental Celtic and Goidelic. That would be more complicated than assuming that the Brythonic languages lost it.
- "But this can be stopped."
- "No, I came all this way to show you this because nothing can be done. Because I like the way your pupils dilate in the presence of total planetary Armageddon.
Yes, it can be stopped."
- "No, I came all this way to show you this because nothing can be done. Because I like the way your pupils dilate in the presence of total planetary Armageddon.
Yes, it can be stopped."
I think that part of the problem in understanding what happened is that the traditional picture (literally, in many cases) is that all IE languages diverged from PIE at the same time; there are many diagrams which suggest this.
In reality, the splitting was strongly binary. At some point, PIE split into Anatolian and "Common IE", and then Tocharian split off from that, and so on. Apparently there is not a lot of consensus in regard to what happened; some models say that Indo-Iranian and Greco-Armenian were two of the earlier branches; others suggest that Greco-Armenain and Balto-Slavo-Indo-Iranian were the last groups to diverge.
In reality, the splitting was strongly binary. At some point, PIE split into Anatolian and "Common IE", and then Tocharian split off from that, and so on. Apparently there is not a lot of consensus in regard to what happened; some models say that Indo-Iranian and Greco-Armenian were two of the earlier branches; others suggest that Greco-Armenain and Balto-Slavo-Indo-Iranian were the last groups to diverge.
True. Scientific theories don't tell us what is true, though lots of people would like to think so (this is not, in any way, shape, or form a slam against science).eodrakken wrote:Well said, Tengado. Occam's razor does not mean "the simplest explanation is always true". The misuse of it is one of my pet peeves too.
But let's consider two hypotheses to explain Grimm's Law.
1) The voiceless stops of PIE underwent the common lenition process of frictivization, inducing a chain shift on the voiced and voiced aspirated stops.
2) An invisible sky pixie confused the language of the Proto-Germans so that the voiceless stops underwent the common lenition process of frictivization, inducing a chain shift in voiced and voiced aspirated stops.
Who in their right mind would consider #2 a superior explanation to #1?
Melchert has repudiated his earlier views on the subject. He now views the feminine as deriving from an earlier collective, in the non-Anatolian languages.Dewrad wrote:But no, entia non sunt multiplicanda! I hear you cry. But what if we back this up with evidence from Lycian? Perhaps it is simply the case that we didn't actually have all the relevant data to hand.
:sighs: You're missing the point. Allow me to quote myself:Etherman wrote:Melchert has repudiated his earlier views on the subject. He now views the feminine as deriving from an earlier collective, in the non-Anatolian languages.Dewrad wrote:But no, entia non sunt multiplicanda! I hear you cry. But what if we back this up with evidence from Lycian? Perhaps it is simply the case that we didn't actually have all the relevant data to hand.
What I am addressing here is not, sensu stricto, whether the variety of PIE ancestral to the Anatolian languages had a feminine gender or not. (My own views on the subject are, for what it's worth, that it's not really a very interesting question.) Instead, I am attempting to make the point that if you wave Occam's razor around too much you'll cut yourself. Particularly if you try to support its use by a reductio ad absurdum as above.I wrote:I am not coming down on one side or another of the debate here.
Salmoneus wrote:(NB Dewrad is behaving like an adult - a petty, sarcastic and uncharitable adult, admittedly, but none the less note the infinitely higher quality of flame)
Occam's Razor is about removing unnecessary hypotheses. "Unnecessary" being relative to the extant data. So yeah, if people use it as if it's a a tool to getting towards the "truth" then they're going to cut themselves.Dewrad wrote: What I am addressing here is not, sensu stricto, whether the variety of PIE ancestral to the Anatolian languages had a feminine gender or not. (My own views on the subject are, for what it's worth, that it's not really a very interesting question.) Instead, I am attempting to make the point that if you wave Occam's razor around too much you'll cut yourself. Particularly if you try to support its use by a reductio ad absurdum as above.
Whatever you meant by putting "truth" in quotes, your definition doesn't bring the discussion to anything definitive.Etherman wrote:Occam's Razor is about removing unnecessary hypotheses. "Unnecessary" being relative to the extant data. So yeah, if people use it as if it's a a tool to getting towards the "truth" then they're going to cut themselves.
For there are at least two possible uses of Occam's Razor here.
(1) The hypothesis which projects a relatively complex feature (an extra class involved in non-semantic-based class agreement) onto the ancestral state is unnecessary.
(2) The hypothesis which implies that a relatively complex feature (an extra class involved in non-semantic-based class agreement) emerges from nothing in practically no time is unnecessary.
Some people (e. g. me) prefer the second, which also illustrates some problems with practical use of Occam's Razor other than intensity of waving.
Basilius
Is this a hypothetical or is this what you think the idea that feminine gender was an innovation happening after the branching off of Anatolian implies? That would depend on the time frames involved, wouldn't it?Basilius wrote:(2) The hypothesis which implies that a relatively complex feature (an extra class involved in non-semantic-based class agreement) emerges from nothing in practically no time is unnecessary.
(BTW, you made some good points in that old discussion. I basically didn't respond because checking and reviewing my own POV would require me to do a lot more reading than I'm currently able to do, due to a lack both of time and of access to the necessary literature.)
The latter. I think that quick loss of the feature in question is more probable than its quick emergence.hwhatting wrote:Is this a hypothetical or is this what you think the idea that feminine gender was an innovation happening after the branching off of Anatolian implies?Basilius wrote:(2) The hypothesis which implies that a relatively complex feature (an extra class involved in non-semantic-based class agreement) emerges from nothing in practically no time is unnecessary.
That is, I prefer the hypothesis that feminine is older than Indo-Hittite (although I don't usually think of such matters in terms of Occam's Razor, at least explicitly).
Well, yes, but I think it's a different order of magnitude. Anatolian langs just don't look genetically distant enough.That would depend on the time frames involved, wouldn't it?
(EDIT) And there are the two Tokharian langs, which look about as deviant but have feminine...
Thanks! It would be interesting to hear your comments when you have more time!(BTW, you made some good points in that old discussion. I basically didn't respond because checking and reviewing my own POV would require me to do a lot more reading than I'm currently able to do, due to a lack both of time and of access to the necessary literature.)
Basilius
-
- Sanci
- Posts: 25
- Joined: Thu Mar 16, 2006 12:27 pm
- Location: London
Gender in IE is clearly very late, and the agreement patterns are anomalous in the light of other languages groups that possess it (e.g. North Caucasian, Afro-Asiatic, Bantu, Australian, etc).
1. Gender is unmarked in the verb.
2. Not all adjectives show the masculine/feminine distinction, and the masculine/neuter distinction seems to have been confined to the nominative-accusative singular.
3. Nouns have no distinctive markers: -h₂ is not exclusively feminine in any language.
4. The pronouns sensu strictu have no gender.
Lehmann argued that a lot of so-called PIE morphology, reconstructed from just two or three languages, was probably non-existant; that it's just a case of different languages getting the same results by building on inherited material. Gender is so marginal, that it's absence from Anatolian doesn't seem to be such a big deal that we need to regard those languages as something special.
1. Gender is unmarked in the verb.
2. Not all adjectives show the masculine/feminine distinction, and the masculine/neuter distinction seems to have been confined to the nominative-accusative singular.
3. Nouns have no distinctive markers: -h₂ is not exclusively feminine in any language.
4. The pronouns sensu strictu have no gender.
Lehmann argued that a lot of so-called PIE morphology, reconstructed from just two or three languages, was probably non-existant; that it's just a case of different languages getting the same results by building on inherited material. Gender is so marginal, that it's absence from Anatolian doesn't seem to be such a big deal that we need to regard those languages as something special.
Gender in IE is clearly very old.David McCann wrote:Gender in IE is clearly very late, [...]
(BTW, I can be less vague: older than Proto-Indo-Hittite.)
... i. e. one cannot even resort to an adstrate influence.[...] and the agreement patterns are anomalous in the light of other languages groups that possess it (e.g. North Caucasian, Afro-Asiatic, Bantu, Australian, etc).
... i. e. one potential source for its hypothetized recent emergence (e. g. via a cliticized relative form of some auxiliary) is absent.1. Gender is unmarked in the verb.
... i. e. gender agreement in PIE is too tricky for a recently acquired morphosyntactic feature.2. Not all adjectives show the masculine/feminine distinction, and the masculine/neuter distinction seems to have been confined to the nominative-accusative singular.
... i. e. gender marking in PIE is too tricky for a recently acquired morphological feature.3. Nouns have no distinctive markers: -h₂ is not exclusively feminine in any language.
(Not a valid point actually, but the one I respond to is worse: not all words ending in -er in French are verbs - clearly verbs as PoS are late, see?)
... i. e. another possible source is lacking. (I assume "pronouns sensu strictu" = personal pronouns of 1st and 2nd person.)4. The pronouns sensu strictu have no gender.
Also, this means that gender is lacking exactly in those forms where its assignment would be purely semantic (i. e. would be easier to explain away).
(Also, there is gender agreement in 1st person singular pronoun in Tokharian, but this I prefer to consider an innovation, without real evidence.)
Sorry... any connection to the subject of this thread?Lehmann argued that a lot of so-called PIE morphology, reconstructed from just two or three languages, [...]
Gender is so central in IE morphosyntax that the absence of feminine in Anatolian doesn't seem to be per se a reason to regard those languages as something special.Gender is so marginal, that it's absence from Anatolian doesn't seem to be such a big deal that we need to regard those languages as something special.
Basilius