European languages before Indo-European

[TaylorS]

8. The Hesperic languages of the British Isles shifted from a head-final to a head-initial word order, and exerted a substratum influence on the Insular Celtic languages. The British Isles were the site of the highest cultural development in the Hesperic sphere, spawning the Germanic and Celtic traditions of Elves, and the Greek traditions of Hyperborea and Atlantis.

I like this little detail a lot. How do you get the head-final > head-initial shift going, though? Is it just internal, or is it from yet another substrate that was there before Hesperic that you don't even bother with?

I haven't yet decided on the cause of that shift, though I lean towards an internal explanation. The language has verbs agreeing with both subject and object, so the relevant NPs were at one point reanalysed as appositions, which were moved to the end of the sentence, with the verb now being placed first. This reordering of the clause prompted a corresponding reordering of the NP. There are traces of the old head-final order in Old Albic, though. Compounds are head-final, and some case endings evolved from postpositions - and as is well known, according to Thomas Givon, "today's morphology is yesterday's syntax".

I did indeed consider yet another substratum, even an Afroasiatic one, but that turned out to be a horrible kludge. That the British Isles were Hesperic-speaking is indicated by the presence of the Old European Hydronymy; that the language that was spoken there before Celtic was head-initial is indicated by the substratum influence on Insular Celtic, which underwent a total restructuring of its syntax within a few centuries. So there is evidence for a head-initial Hesperic language in the British Isles. There may have been a kind of cline in pre-IE Europe, with head-final languages in the south and east and head-initial languages in the northwest.

IMO the Head-Initial syntax IMO is a very old NW European areal feature and survives today in many features of the syntax of Standard Average European. But that is just speculation on my part.

[Octavià]

Octavia > STOP TALKING

Other people > STOP TALKING TO OCTAVIA

This looks like you're applying me lynch law.

How nasty must be I!

[WeepingElf]

IMO the Head-Initial syntax IMO is a very old NW European areal feature and survives today in many features of the syntax of Standard Average European. But that is just speculation on my part.

This is very well possible. The British Isles may have been VSO since time immemorial, and the nearby mainland countries may have been SVO. We find SOV in the south (Basque, Celtiberian, just about every language of ancient Italy, etc.) and in the east (PIE, Caucasian languages and also the more easterly Uralic languages), but the northwest may indeed have been head-initial for a very long time.

[zompist]

There's already a medium for that; it's called a blog. Go and use it rather than abusing a discussion forum.

I take your advice seriously.

Sometimes an imperative is advice; sometimes it's an order. I've applied a little administrative weighting to help Octavià understand.

[Rodlox]

like the rabid beast

Can you believe I actually read "rabbit beast", and wondered what the poor bunnies had done to you to be compared with Octavia?


JAL

too much Monty Python?

[TaylorS]

IMO the Head-Initial syntax IMO is a very old NW European areal feature and survives today in many features of the syntax of Standard Average European. But that is just speculation on my part.

This is very well possible. The British Isles may have been VSO since time immemorial, and the nearby mainland countries may have been SVO. We find SOV in the south (Basque, Celtiberian, just about every language of ancient Italy, etc.) and in the east (PIE, Caucasian languages and also the more easterly Uralic languages), but the northwest may indeed have been head-initial for a very long time.

It would not really be that odd, either. in pre-contact NW North America there were Head-Initial (Salishian) and Head-Final (Athabaskan) language families right next to each other.

[WeepingElf]

IMO the Head-Initial syntax IMO is a very old NW European areal feature and survives today in many features of the syntax of Standard Average European. But that is just speculation on my part.

This is very well possible. The British Isles may have been VSO since time immemorial, and the nearby mainland countries may have been SVO. We find SOV in the south (Basque, Celtiberian, just about every language of ancient Italy, etc.) and in the east (PIE, Caucasian languages and also the more easterly Uralic languages), but the northwest may indeed have been head-initial for a very long time.

It would not really be that odd, either. in pre-contact NW North America there were Head-Initial (Salishian) and Head-Final (Athabaskan) language families right next to each other.

Yes. And the British Isles are separated from the rest of Europe by a swath of sea; you cannot walk there but need a boat. This makes them into a residual zone where languages survive that have disappeared elsewhere (case in point: Celtic), and may develop features that are not found on the continent (case in point again: Insular Celtic).

[alice]

Yes. And the British Isles are separated from the rest of Europe by a swath of sea; you cannot walk there but need a boat. This makes them into a residual zone where languages survive that have disappeared elsewhere (case in point: Celtic), and may develop features that are not found on the continent (case in point again: Insular Celtic).

That said, it used to take six days to travel from Cornwall to London, but only one to Brittany.

[WeepingElf]

Yes. And the British Isles are separated from the rest of Europe by a swath of sea; you cannot walk there but need a boat. This makes them into a residual zone where languages survive that have disappeared elsewhere (case in point: Celtic), and may develop features that are not found on the continent (case in point again: Insular Celtic).

That said, it used to take six days to travel from Cornwall to London, but only one to Brittany.

Yes. The sea does not merely separate; it also connects. Nevertheless, people without ships (and that characterized most inhabitants of Neolithic and even Bronze Age Europe) could not reach the British Isles, and it is an observed fact that Celtic survived there longer than on the Continent, and developed features no other Indo-European languages ever had.

It is uncertain whether such features are VSO word order are stable in a region to such a degree that newcomer languages adopt them. We know that English didn't; we don't know whether pre-Celtic languages of the British Isles were VSO or not. There are several instances of newcomer languages adopting indigenous word orders, such as the Ethiopic Semitic languages becoming SOV under the influence of the Cushitic languages, or Akkadian under the influence of Sumerian.

[WeepingElf]

Yes. The sea does not merely separate; it also connects. Nevertheless, people without ships (and that characterized most inhabitants of Neolithic and even Bronze Age Europe) could not reach the British Isles, and it is an observed fact that Celtic survived there longer than on the Continent, and developed features no other Indo-European languages ever had.

An afterthought: There are some kinds of contact (e. g., trade) which the sea facilitates, but others (like marriage between people from neighbouring villages) which the sea inhibits. And it seems that the latter kind of contacts are important for the diffusion of languages and linguistic features. The sea sets a boundary beyond which dialect continua cannot expand; while it is very possible that a language traverses the sea (especially if it as narrow as that between the Continent and Britain), the languages on bot sides of the sea will begin to diverge. Hence, islands function as residual zones, especially if they are located between the mainland and the outer edge of the known world.

[Jetboy]

I recall reading in several places that PIE's ancestor was probably active-stative; commonly cited evidence are the lack of case distinction in inanimate/neuter nouns, and differing roots for things that can seem either animate or inanimate, like fire and water. However, how exactly do these features reflect sometimes treating experiencers like patients, and sometimes treating them like agents, the defining property of Active languages?

[WeepingElf]

I recall reading in several places that PIE's ancestor was probably active-stative; commonly cited evidence are the lack of case distinction in inanimate/neuter nouns, and differing roots for things that can seem either animate or inanimate, like fire and water. However, how exactly do these features reflect sometimes treating experiencers like patients, and sometimes treating them like agents, the defining property of Active languages?

I am not really an expert on these matters, but perhaps a better indicator of earlier active-stativeness than the peculiarities in the noun case (to which I shall return below) is that Early PIE had two sets of verbal personal endings, one associated with active semantics and the other with stative semantics, though the matter already was quite blurry, and in Late PIE, the latter were used with perfects and middles (which are categories associated with stative meanings). If the thematic conjugation was in origin a transitive conjugation (which may or may not have been the case), the thematic vowel may have been the same morpheme as the 3rd person stative ending. That's a trait typical of active/stative languages; have a look at Lakota which does just that, only with prefixes.

Now, what regards cases, active-stative languages tend to distinguish between animate and inanimate nouns, of which the latter may not be used as agents. Hence, in an active-stative language which has cases, the inanimate nouns have a defective paradigm, lacking an agentive case. This would explain the case idiosyncrasies of IE neuters nicely. The active-stative case marking system with defective inanimate paradigm developed into a split-ergative system in Early PIE with nom/acc animates and erg/abs inanimates (still preserved in Hittite), and from there to a nom/acc system with case syncretism in neuters in Late PIE.

BTW: Using the term "experiencer" for intransitive subjects is WRONG. "Experiencer" is a semantic role; by far not all intransitive subjects are experiencers, and by far not all experiencers are intransitive subjects. The error AFAIK goes back to Justin B. Rye who wrote a web page about morphosyntactic alignments without understanding it.

[8Deer]

Just wanted to post this list here of proposed non-Indo-European words in Proto-Celtic.

I don't have anything to add other than to say that *andera "young woman" looks a lot like Basque andere "lady", although the Basque etymological dictionary I'm looking at right now says it may have been borrowed from Celtic into Basque. I hope I don't sound too much like Octavia.

[TomHChappell]

Just wanted to post this list here of proposed non-Indo-European words in Proto-Celtic.

I don't have anything to add other than to say that *andera "young woman" looks a lot like Basque andere "lady", although the Basque etymological dictionary I'm looking at right now says it may have been borrowed from Celtic into Basque. I hope I don't sound too much like Octavia.

I once heard that fully a third of proto-Germanic's vocabulary was not related to the more general Indo-European vocabulary. Anyone know any details? Could there be a list of such lexemes or proto-lexemes lying around somewhere? Could someone post a link or just copy it in?

[Jetboy]

Now, what regards cases, active-stative languages tend to distinguish between animate and inanimate nouns, of which the latter may not be used as agents. Hence, in an active-stative language which has cases, the inanimate nouns have a defective paradigm, lacking an agentive case. This would explain the case idiosyncrasies of IE neuters nicely. The active-stative case marking system with defective inanimate paradigm developed into a split-ergative system in Early PIE with nom/acc animates and erg/abs inanimates (still preserved in Hittite), and from there to a nom/acc system with case syncretism in neuters in Late PIE.

Ah, Thanks for the explanation. I wonder why that's a characteristic of Active languages, as opposed to ergative, or even nominative, though. So things like fire, water, etc., would have had an animate and an inanimate stem?

BTW: Using the term "experiencer" for intransitive subjects is WRONG

Sorry about that, I was in a bit rushed, and I couldn't think of any other terms.

[WeepingElf]

Now, what regards cases, active-stative languages tend to distinguish between animate and inanimate nouns, of which the latter may not be used as agents. Hence, in an active-stative language which has cases, the inanimate nouns have a defective paradigm, lacking an agentive case. This would explain the case idiosyncrasies of IE neuters nicely. The active-stative case marking system with defective inanimate paradigm developed into a split-ergative system in Early PIE with nom/acc animates and erg/abs inanimates (still preserved in Hittite), and from there to a nom/acc system with case syncretism in neuters in Late PIE.

Ah, Thanks for the explanation. I wonder why that's a characteristic of Active languages, as opposed to ergative, or even nominative, though.

The idea is that inanimate things cannot act out of themselves. A stone may smash a window, but you will never see a stone jumping at a window smashing it out of itself. It must have been set in motion by some other agent, e.g. a person throwing it. Hence, the stone is an instrument and not an agent. Therefore, in an active-stative language, you'll expect inanimate nouns to have defective case paradigms. In an accusative language, in contrast, all nouns need a nominative, because an inanimate noun, while not eligible for agency, can at least be the subject of a sentence. In an ergative language, the ergative case is often also restricted to animates, and syncretisms with other cases (usually the genitive or the instrumental) are common.

(Of course, the grammatical category of "animacy" needs not coincide with modern notion of what is a living being or not. Usually, natural forces such as water, wind or fire are considered animate, and some languages with an animate/inanimate distinction place machines in the animate class, as they can appear to act out of themselves.)

So things like fire, water, etc., would have had an animate and an inanimate stem?

At least some linguists say that - and PIE appears to have had them!

[Jetboy]

Ah, okay, that makes sense. Thanks!

[merijn]

Now, what regards cases, active-stative languages tend to distinguish between animate and inanimate nouns, of which the latter may not be used as agents. Hence, in an active-stative language which has cases, the inanimate nouns have a defective paradigm, lacking an agentive case. This would explain the case idiosyncrasies of IE neuters nicely. The active-stative case marking system with defective inanimate paradigm developed into a split-ergative system in Early PIE with nom/acc animates and erg/abs inanimates (still preserved in Hittite), and from there to a nom/acc system with case syncretism in neuters in Late PIE.

Ah, Thanks for the explanation. I wonder why that's a characteristic of Active languages, as opposed to ergative, or even nominative, though.

The idea is that inanimate things cannot act out of themselves. A stone may smash a window, but you will never see a stone jumping at a window smashing it out of itself. It must have been set in motion by some other agent, e.g. a person throwing it. Hence, the stone is an instrument and not an agent. Therefore, in an active-stative language, you'll expect inanimate nouns to have defective case paradigms. In an accusative language, in contrast, all nouns need a nominative, because an inanimate noun, while not eligible for agency, can at least be the subject of a sentence. In an ergative language, the ergative case is often also restricted to animates, and syncretisms with other cases (usually the genitive or the instrumental) are common.

(Of course, the grammatical category of "animacy" needs not coincide with modern notion of what is a living being or not. Usually, natural forces such as water, wind or fire are considered animate, and some languages with an animate/inanimate distinction place machines in the animate class, as they can appear to act out of themselves.)

So things like fire, water, etc., would have had an animate and an inanimate stem?

At least some linguists say that - and PIE appears to have had them!

The idea that inanimate nouns cannot be subjects of transitive verbs (or perhaps more technical but more correct the external argument of a verb) has always amazed me. What about "the storm ruined my house." "His love gave me strength to carry on". "her idea started the debate" "their victory humiliated him" "your name escapes me"? It may be rarer than animate nouns, but "that they cannot act out of themselves" does not make it impossible for them to be a cause or even be a metaphorical agent. If a language cannot have inanimate subjects of transitive verbs I would expect that they have alternative ways of expressing those concepts. The correct way of saying is not "inanimate nouns cannot act out of themselves and are therefore never subjects of transitive verbs" but "inanimate nouns cannot usually act out of themselves and are therefore rarer as subjects of transitive verbs than animate subjects. Also they are less likely to be topics and in many languages subjects are inherently topics. This has led in some languages to a ban on inanimate subjects of transitive verbs".
Also, it is more common in split ergative languages that inanimate nouns form an ergative pattern and animate nouns form a nominative pattern.

[merijn]

An equally good way of explaining the distribution of case markers in PIE is assuming that 1) animate nouns were marked accusative but not inanimate nouns (crosslinguistically quite common for example Spanish a) and 2) a topic marker was generalized as an animate subject marker since both subjects and animates are more likely to be topics.

[jal]

The idea that inanimate nouns cannot be subjects of transitive verbs (or perhaps more technical but more correct the external argument of a verb) has always amazed me. What about "the storm ruined my house." "His love gave me strength to carry on". "her idea started the debate" "their victory humiliated him" "your name escapes me"?

As Joerg wrote:

Of course, the grammatical category of "animacy" needs not coincide with modern notion of what is a living being or not. Usually, natural forces such as water, wind or fire are considered animate

So the storm case is covered. Love, idea and victory are all non-tangeable nouns, and may very well be treated differently in such languages, or gerunds may be included in the "animate" category and the above examples are encoded as "him-loving-me gave me (...)", "her-having-an-idea started (...)", "them-being-victorious humilated him" etc. That, or gerunds cannot act as subject (which is afaik quite common in the world's languages) and rephrasing is needed (e.g. "he loved me, so I had the strength (...)", "when she got an idea, the people started the debate", "he felt humiliated after being defeated" etc. etc.

It may be rarer than animate nouns, but "that they cannot act out of themselves" does not make it impossible for them to be a cause or even be a metaphorical agent.

It may very well make it impossible to be a cause. Don't be confused by the fact that in Western, IE languages, subjects can have a myriad of semantic roles, which is not the case for many other languages.

If a language cannot have inanimate subjects of transitive verbs I would expect that they have alternative ways of expressing those concepts.

Of course, see above.

The correct way of saying is not "inanimate nouns cannot act out of themselves and are therefore never subjects of transitive verbs" but "inanimate nouns cannot usually act out of themselves and are therefore rarer as subjects of transitive verbs than animate subjects."

I think you are confusing your definition of animacy with grammatical animacy of those languages. If "inanimate" is defined as "not being able to act", then your definition doesn't make much sense, does it? Also, the fact that they cannot act doesn't mean they cannot cause. It's just very likely that such causing is not described as the causee being the subject, in languages that do not permit inanimates as subject.


JAL

[WeepingElf]

More grist for the mill. The Etymological Dictionary of Proto-Celtic by Ranko Matasović contains an appendix listing 85 Celtic words without plausible IE etymology, which are listed here. These words may come from various substratum languages. Those that are attested in Gaulish probably were borrowed into Celtic on the Continent; those found only in Insular Celtic may have come from a substratum language of the British Isles. Some appear to have cognates or parallels in Italic and/or Germanic.

[merijn]

I think that I have to clarify myself. It wouldn't surprise me if there are languages where inanimate nouns cannot be subjects of transitive verbs, and although I am skeptical about Proto-Indo-European being such a language, I don't exclude the possibility. I am amazed tho by the assumption that there are nouns that for semantic reasons cannot ever be agents or causes. That assumption seems to deny the flexibility that meaning has.
I do have a problem with some of your reasoning. For a start, I think if you exclude all nouns that can act are excluded from inanimate nouns the group of inanimate nouns will probably end up being an empty set. I think that for any noun you can make scenarios in which it acts without any obvious outside cause, for example:"the rock started to move and crushed a bug" (come to think of it, "crush" is probably a better examples than the ones I gave in my previous post of a verb that doesn't require an animate subject, because it is a bit harder to think of a way to paraphrase it). My problem is that you are defining the problem away. For any inanimate noun that I give an example of acting out of itself, and I think that I can give examples of any inanimate noun, you say "well if it can act out of itself it must be an animate noun". The animacy hierarchy seems to me more closely related to mental powers and the ability to experience things than to the ability to act out of itself. Crosslinguistically the animacy hierarchy is pretty robust, and only the cut off point is different for different languages that make the distinction and there are occasionally exceptions that are originally culturally motivated (for example fire is seen as something that has its own will). Usually when a language makes the distinction between animate and inanimate nouns abstract nouns are considered inanimate (in fact they are lower on the animacy hierarchy than objects like rocks), so it would surprise me if "love" was animate but "rock" wasn't.

As for as I know, the problem with the animacy distinction in PIE is that there very few if any animate nouns that are neuter, but there are quite a few inanimate nouns in the common gender. I don't want to be that guy that drags his pet language (in my case Zulu) in every post he makes, but that reminds of a situation in Zulu. Zulu has a set of noun classes, singular noun class 1 and plural noun class 2, that only have animate nouns. There is however a closely related noun class called noun class 1a, which has a plural called noun class 2a, where the situation is a bit different. Originally noun class 1a was only used for personal names and family members like father. Nowadays it has a lot of loans (though it is not the noun class used for new loans anymore) as well as quite a few new nouns made by derivation (mainly by prefixing ma- or no-). These nouns don't have to be inanimate; an example is "umabonakude" which means television (lit: see far). These inanimate nouns in noun class 1a trigger noun class 3 agreement in the singular, which is a noun class used exclusively for inanimate nouns, but trigger noun class 2 agreement, which is otherwise only used for animate nouns, in the plural.
My point is that a noun class that was originally used only for animate nouns has now also inanimate members due to loans and derivation and something similar may have happened to prePIE. The common gender may have been enriched by loans (maybe because it had more predictable morphology when compared to the neuter gender) or some kind of derivation, previously used for animate nouns, may have been starting to produce inanimate nouns that because of its morphology nonetheless belonged to the common gender

[Soap]

Animacy is pretty counterntuitive in languages that grammaticalize it. I would actually say that a good way to define it is "anything that can be the subject of a transitive verb". So ocean, rain, fire, wind, etc can all be animates. I dont know about rocks, so I think a language in which rocks are not animate would just say "the rock was fallen on the bug" or use a type of "fall" which is intransitive (if you think about it, "on the bug" is just a prepositional phrase even in Englsih, it isnt a grammatical object, so the verb is still intransitive.)

[merijn]

Animacy is pretty counterntuitive in languages that grammaticalize it. I would actually say that a good way to define it is "anything that can be the subject of a transitive verb". So ocean, rain, fire, wind, etc can all be animates. I dont know about rocks, so I think a language in which rocks are not animate would just say "the rock was fallen on the bug" or use a type of "fall" which is intransitive (if you think about it, "on the bug" is just a prepositional phrase even in Englsih, it isnt a grammatical object, so the verb is still intransitive.)

The vast majority of languages grammaticalize it: English: The boy who vs *the car who, Dutch: naast hem vs ernaast, in mainland Scandinavian languages the choice of pronouns is dependent on it, in Bantu languages it interacts with the noun class system. In all those languages the concept is pretty clear and far from counterintuitive.
You probably mean that it is counterintuitive an all languages where there is a ban on inanimate subjects. But if you're saying that those nouns that are banned from being subjects of transitive verbs are those nouns that cannot be subjects of transitive verbs you are reasoning in circles and saying that inanimate nouns cannot be used as subjects of transitive verbs is not very informative. A better way to say it would be that all animate nouns can be subjects of transitive verbs as well as a subset of inanimate nouns.
There is a crucial difference between "fall on a bug" and "crush a bug", only in the latter the bug becomes crushed. That is in my opinion the underlying reason why the bug is the object of crush and not the object of fall. A better paraphrase would be "the rock fell on the bug and now the bug has been crushed.

[TomHChappell]

More grist for the mill. The Etymological Dictionary of Proto-Celtic by Ranko Matasović contains an appendix listing 85 Celtic words without plausible IE etymology, which are listed here. These words may come from various substratum languages. Those that are attested in Gaulish probably were borrowed into Celtic on the Continent; those found only in Insular Celtic may have come from a substratum language of the British Isles. Some appear to have cognates or parallels in Italic and/or Germanic.

Has anyone found anything like that for Germanic instead of Celtic?
How about Greek? (And btw is Greek groupable with one or more of Armenian or Anatolian or Albanian? or indeed any other language-subgroup of I-E smaller than all of I-E but larger than all the various Greeks?)