Grammaticalization

[wakeagainstthefall]

So recently I made an isolating conlang with the purpose of being the proto-lang for another conlang. The goal is to derive a fusional language that develops over the course of 1000-1500 years and has an Indo-Europeanesque phonology (I like my conlangs pretty.) The sound changes are easy enough for me to take on, but the issue is that it seems much easier to develop an isolating language from a fusional language than the other way around (I don't know how to grammaticalize very well.) So I have several questions that may seem a bit broad.
How does a fusional language develop several common affixes in nominative forms? For example, how did -us and -a, or less common -ur and -ex come about in Latin? This seems simple from PIE to Latin because PIE was even more complex, but how do I get a similar result from an isolating language instead?
How do I believably derive a case system from an isolating language? One way I know is by making nouns into bound morphemes, but from what words would I derive those affixes? Or what other way can I produce those affixes?
How do I produce a more complex system of conjugation that was completely absent in the protolang? How do I get infinitive forms?
How can I mess with the syntax to change word order or change prepositions to postpositions, or how does a shift from head-initial to head-final occur? (E.g. Adjectives switching from following nouns to coming before nouns?)
Has anyone done something similar who would be willing to share some examples?

[clawgrip]

How does a fusional language develop several common affixes in nominative forms? For example, how did -us and -a, or less common -ur and -ex come about in Latin? This seems simple from PIE to Latin because PIE was even more complex, but how do I get a similar result from an isolating language instead? How do I believably derive a case system from an isolating language? One way I know is by making nouns into bound morphemes, but from what words would I derive those affixes? Or what other way can I produce those affixes?

As you say, make independent adpositions into dependent suffixes. The simplest way is just to make them adpositions.

How do I produce a more complex system of conjugation that was completely absent in the protolang? How do I get infinitive forms?

Same as above: have independent morphemes become dependent. Creating new tenses and so on is as easy as making a serial verb construction, and maybe throwing in some other things like a nominalizer, adverbializer, or whatever.

Once you have these, then use contractions and sound changes to obscure them. Don't be afraid to have sound alterations that only occur in grammatical endings, As an example of inconsistent and grammatically localized sound changes, Old Japanese to modern Japanese frequently drops /k/ in multiple grammatical endings, but not consistently and not in other places in the language, e.g. OJ kakitari → MJ kaita, but OJ yoroshiku is still yoroshiku in MJ, except in highly formal language, where the loss of /k/ does occur, i.e. yoroshū (note also the sound change).
Maybe the opposite happens: maybe the addition of the auxiliary causes a stress change in the verb phrase which induces a sounds change in the verb only when it is modified by auxiliaries, giving rise to something that could eventually become a participle or gerund or something.

How can I mess with the syntax to change word order or change prepositions to postpositions, or how does a shift from head-initial to head-final occur? (E.g. Adjectives switching from following nouns to coming before nouns?)
Has anyone done something similar who would be willing to share some examples?

Once adpositions have fused to nouns, it opens the doors for agreement and freer syntax. Dislocation for focus change can also become mandatory. Also, formal, ritual, or poetic language can alter languages for stylistic/aesthetic reasons, and if this is a prestige register, the general populace may adopt it over time. Let's also not discount the effect that a significant foreign language speaking population can have on a language (just look at Singlish).
Mandarin has verb-like prepositions and noun-like postpositions. I can imagine using something along these lines to create a smooth transition:

"The cat jumps onto the table."

cat jump go step.on table
↓
cat jump go table top
↓
cat jump table top position
↓
cat jump table topposition
↓
cat jump table onto
↓
table onto cat jump
↓
cat table onto jump

[k1234567890y]

I am not a professional linguist, I can just say what I have known or what I speculate, I can't guarantee that they are correct.

How does a fusional language develop several common affixes in nominative forms? For example, how did -us and -a, or less common -ur and -ex come about in Latin? This seems simple from PIE to Latin because PIE was even more complex, but how do I get a similar result from an isolating language instead?
How do I believably derive a case system from an isolating language? One way I know is by making nouns into bound morphemes, but from what words would I derive those affixes? Or what other way can I produce those affixes?

As far as I know, case suffixes evolve from adpositions, adpositions evolve from nouns and verbs(however, it seems that certain pronouns or demonstratives can also evolve into adpositions), and there are languages that mark direct objects with adpositions like Modern Hebrew.

For the accusative case, you can do differential object marking(DOM, it means some direct objects get a distinct marking while others have the same marking as the subject, usually it is the definite nouns or the more animate nouns that get a distinct marking, the marking can be adpositions or case affixes, you can read https://en.wikipedia.org/wiki/Different ... ct_marking to know more) first, then generalize the object marking to all direct objects.

As for the IE-style nominative case, it seems that Proto-Indo-European did not actually have masculine and feminine gender, and it had a distinction between animate nouns and inanimate nouns instead, and inanimate nouns evolved into the neuter gender, and animate nouns evolved largely into masculine and feminine gender later, but I am not sure about how gender evolved in Indo-European languages, anyways, Proto-Indo-European have the distinction between nominative and accusative forms in animate nouns, but not in inanimate nouns, I have read an article on a blog (the actual link of the article: http://paleoglot.blogspot.tw/2009/10/ni ... ht-in.html ) that the animate nominative *-s in Proto-Indo-European evolved from a demonstrative pronoun(so does in many languages with a marked-nominative), I am not sure about the *-m for the animate accusative case and the inanimate nominative-accusative case, it might have evolved from adpositions earlier before the animate nominative *-s started to emerge.

you can also develop adposiitions from verbs, you may consider to have a serial verb construction first, Chinese is known to have a serial verb construction, which allows many verbal phrases appearing in the same sentence, thus enabling several of them being served as something similar to adpositional phrases of other languages(and thus verbal phrases that function as adpositional phrases .

Also, plural affixes also evolve from words, but I don't remember that well on this, it seems that dual affixes can evolve from the word for "two", and I think plural affixes can evolve from the word for "many".

How do I produce a more complex system of conjugation that was completely absent in the protolang? How do I get infinitive forms?

first, pronouns often cliticized to adjacent words or words they modify, and cliticized words, not only pronouns, often become grammatical affixes later.

Also, I think you can have aux verbs, adverbs or particles(particles are independent words, they are more grammaticalized than adverbs and aux verbs though) expressing tenses, aspects, and something similar first, then let them be grammaticalized, becoming a part of the main verb, and I think many Tense-Aspect-Mood affixes in PIE were originally independent words, too.

How can I mess with the syntax to change word order or change prepositions to postpositions, or how does a shift from head-initial to head-final occur? (E.g. Adjectives switching from following nouns to coming before nouns?)

As the article shown in http://www.pnas.org/content/108/42/17290.full suggests, SOV word order does not occur naturally unless being influenced by other languages with an SOV word order, as far as I know; also, it seems that the sentential word order changes first then the place of adpositions and the modifiers of nouns.

Moreover, being analytic does not imply that the language has a SVO, prepositional, head-initial word order, you can just have a SOV, postpositional and head-final word order for your proto language instead, there are languages that have a SOV, postpositional and head-final word order but are isolating and don't have markings differentiating the subjects from the direct objects, like Yi and Ainu(however, Classical Ainu was polysynthetic, so Ainu might have dropped most of its inflections).

however, if you don't have appropriate sound changes, the descendant languages will keep being agglutinative languages, to make affixes fuse together with desired results, you need wise sound changes, too.

[Salmoneus]

As the article shown in http://www.pnas.org/content/108/42/17290.full suggests, SOV word order does not occur naturally unless being influenced by other languages with an SOV word order

Err...

OK, first thing to point out there is that that article is by a highly controversial linguist and a highly respected quantum physicist. Gell-Mann is of course one of the most important physicists of the 20th century (and a Nobel Laureate to boot) but that doesn't inherently make him an expert on historical linguistics. Ruhlen, meanwhile, as his wikipedia page notes, is generally regarded as "standing outside the mainstream", to say the least. Note, for instance, that the paper relies on analyses of such families as "Amerind", "Austric", "Nostratic" and "Dene-Caucasian" - i.e., families that are only accepted as families by the author the paper...

The second thing to point out is that the article doesn't say that. It says that Proto-Human was SOV, and that, in essence, the natural state of all languages is SOV, but that there's one-way change from SOV to SVO. So it says that SOV doesn't occur naturally unless by influence from another SOV language OR as a retention from a proto-language. It kind of has to say that, since your version would mean there could never be any SOV languages (where would the influencing SOV language itself come from?). That is, that SVO does not become SOV unless there's another SOV language nearby. This is both fatuous and probably untrue, but either way isn't what you said.

[Leaving aside the problem that there's always an SOV language nearby to produce the desired influence, there's also the problem that it's only really IE where the diachronics are well-enough established to give direct evidence of the direction of change. Everywhere else, they can just say "some languages in this family are SOV and some are SVO, so clearly some retained the original SOV and some changed it to SVO, therefore this is evidence that SVO never changes to SOV"

[k1234567890y]

As the article shown in http://www.pnas.org/content/108/42/17290.full suggests, SOV word order does not occur naturally unless being influenced by other languages with an SOV word order

Err...

OK, first thing to point out there is that that article is by a highly controversial linguist and a highly respected quantum physicist. Gell-Mann is of course one of the most important physicists of the 20th century (and a Nobel Laureate to boot) but that doesn't inherently make him an expert on historical linguistics. Ruhlen, meanwhile, as his wikipedia page notes, is generally regarded as "standing outside the mainstream", to say the least. Note, for instance, that the paper relies on analyses of such families as "Amerind", "Austric", "Nostratic" and "Dene-Caucasian" - i.e., families that are only accepted as families by the author the paper...

The second thing to point out is that the article doesn't say that. It says that Proto-Human was SOV, and that, in essence, the natural state of all languages is SOV, but that there's one-way change from SOV to SVO. So it says that SOV doesn't occur naturally unless by influence from another SOV language OR as a retention from a proto-language. It kind of has to say that, since your version would mean there could never be any SOV languages (where would the influencing SOV language itself come from?). That is, that SVO does not become SOV unless there's another SOV language nearby. This is both fatuous and probably untrue, but either way isn't what you said.

[Leaving aside the problem that there's always an SOV language nearby to produce the desired influence, there's also the problem that it's only really IE where the diachronics are well-enough established to give direct evidence of the direction of change. Everywhere else, they can just say "some languages in this family are SOV and some are SVO, so clearly some retained the original SOV and some changed it to SVO, therefore this is evidence that SVO never changes to SOV"

ok thank you for your opinion, looks like I am wrong (: although I think the reason that the theory I had posted is discredited is not because who wrote this(in my own opinion, knowledge itself is a form of power, professionals are powerful in their professions and they can abuse their professions like all other power abusers, they are human beings after all), but there are counterrexamples to their theory.

[Frislander]

Gell-Mann is of course one of the most important physicists of the 20th century (and a Nobel Laureate to boot) but that doesn't inherently make him an expert on historical linguistics.

I hate it that Linguistics is practically the only science where people who are completely unqualified to talk on the subject can just spout stuff and nobody save for linguists (along with a few enthusiastic amateurs) can be bothered to check up on what they are spouting, even when they just make stuff up. (see Language Log for lots of examples of this kind of thing).

there's also the problem that it's only really IE where the diachronics are well-enough established to give direct evidence of the direction of change. Everywhere else, they can just say "some languages in this family are SOV and some are SVO, so clearly some retained the original SOV and some changed it to SVO, therefore this is evidence that SVO never changes to SOV"

Another example of Indo-European biases skewing the picture, like when people say that their verbs "agree in person and number" when they mean specifically the nominative subject.

[Imralu]

Both Italian and German sign language are SOV in spite of the languages that they are most heavily influenced by being predominantly SVO.

[k1234567890y]

Both Italian and German sign language are SOV in spite of the languages that they are most heavily influenced by being predominantly SVO.

ok, thank you for your information (:

[Frislander]

Both Italian and German sign language are SOV in spite of the languages that they are most heavily influenced by being predominantly SVO.

ok, thank you for your information (:

Further to that, BSL is pretty strongly topic-comment, despite English definitely not being, and when the topic-comment structure isn't being used it's apparently OSV.

[Imralu]

Further to that, BSL is pretty strongly topic-comment, despite English definitely not being, and when the topic-comment structure isn't being used it's apparently OSV.

Is it really OSV? I'm curious for your source. There's a situation with ASL where learners are very often taught that OSV is the default order and this is snapped up ... essentially because it's different from English and sentences with English order are often assumed to be incorrect and a result of translating directly from English. Linguistic analysis actually shows it to be predominantly SVO in unmarked sentences. Auslan (a dialect, alongside BSL, in the BANZSL) is predominantly SVO or SOV in unmarked sentences (depending mostly on the type of verb). Googling BSL word order gives me sources that say it is SVO and the BSL that I've seen (it's mutually comprehensible with Auslan, so I understand it ... depending on the dialect) certainly never struck me as syntactically different from Auslan to any great degree.

In any case, my comment was a reply to the idea that SOV does not arise without influence from another SOV language.

Was this dazzling website your source? The sentence "the cat sat on the mat" as MAT CAT SIT is not an example of OSV but of the figure-ground-principle (where the "ground" appears before the "figure"). Essentially, MAT is a locative and sit is not transitive.

[Frislander]

Further to that, BSL is pretty strongly topic-comment, despite English definitely not being, and when the topic-comment structure isn't being used it's apparently OSV.

Is it really OSV? I'm curious for your source. There's a situation with ASL where learners are very often taught that OSV is the default order and this is snapped up ... essentially because it's different from English and sentences with English order are often assumed to be incorrect and a result of translating directly from English. Linguistic analysis actually shows it to be predominantly SVO in unmarked sentences. Auslan (a dialect, alongside BSL, in the BANZSL) is predominantly SVO or SOV in unmarked sentences (depending mostly on the type of verb). Googling BSL word order gives me sources that say it is SVO and the BSL that I've seen (it's mutually comprehensible with Auslan, so I understand it ... depending on the dialect) certainly never struck me as syntactically different from Auslan to any great degree.

Well somebody had better go and edit Wikipedia then, and probably put a more fleshed out grammar page while they're at it: the amount of info there compared to ASL is minute.

[Imralu]

Well somebody had better go and edit Wikipedia then, and probably put a more fleshed out grammar page while they're at it: the amount of info there compared to ASL is minute.

Ah, I'll probably do that then. I wrote the syntax section on the article on German Sign Language. I'll have to read a lot though ... so I don't know if I'll get around to it anytime soon.

[Imralu]

This site also gives an example of the figure-ground principle in locative-based sentences, moving from large immobile referents to small, mobile references (garden > tree > cat), not a transitive sentence. The source for the OSV claim on Wikipedia is from a book that I've seen before though, so there may be some weight in it, but honestly, there is a lot of crap that gets published that is not based on actual linguistic research.

[Frislander]

Well somebody had better go and edit Wikipedia then, and probably put a more fleshed out grammar page while they're at it: the amount of info there compared to ASL is minute.

Ah, I'll probably do that then. I wrote the syntax section on the article on German Sign Language. I'll have to read a lot though ... so I don't know if I'll get around to it anytime soon.

Thank you, I very much appreciate it.

[Imralu]

Thank you, I very much appreciate it.

Are you learning it yourself?

[Frislander]

Thank you, I very much appreciate it.

Are you learning it yourself?

No, though I sort of wish I was: I just would like there to be more information available.