A Reader's Guide To SOV langs

[chris-gr]

Lately I've noticed some questions about SOV langs, so I thought it would be helpful to gather SOV characteristics in one place.
Please, post your comments and criticism

Notice: SOV langs are not necessarily agglutinative; some of them (like Turkish, or Mongolian) are, some others (like Hindi) aren't. The following are to be found in the majority of SOV langs.

1. Inflectional affixes or aux. elements are placed after the verb in SOV langs.

2. SOV langs have postpositions (instead of prepositions).

3. In SOV langs, comparatives of inequality place the 'standard' before the adj. ("The dog is bigger than the cat" is rendered as " Dog cat-from big/more")

4. Generally, adjectives (and genitives) come before the noun they modify.

5. Corollary: 'Titles' are placed after the name. Surnames are usually placed before given names.

6. Relative clauses precede nouns in SOV langs. Subordinate clauses precede the principal clause.

7. The equivalent of a relative pronoun in SOV langs may be a topic marker within the relative clause.

8. SVO langs use conjunctions to introduce complements. SOV langs may use nouns instead.

9. Infinitives are rare in SOV langs. Verbal nouns are common.

10. SOV langs express modality and tense/aspect after verb stems.

11. The interrogative marker (if any) follows the V in SOV langs. Sentence negation markers, too.

12. SOV langs usually do not have a passive voice.

13. In SOV langs' numerals, the word for 10 precedes the lower numeral (since it is treated as a 'standard' as in comparative constructions.)

That's all I could think of right now.

[The Rt. Hon. Vlad Dracula]

2. SOV langs have postpositions (instead of prepositions).

Latin has prepositions: sub abore "under the tree", in casa "in the house", etc.

12. SOV lang's usually do not have a passive voice.

Latin:

Active: Marcus caseum edit. "Marcus eats the cheese."
Passive: Caseum a Marco editur. "The cheese is eaten by Marcus."

Japanese:

Active: Amanda ga wain o nonda. "Amanda drank wine."
Passive: Wain ga Amanda ni nomareta. "Wine was drunk by Amanda."

These may just be exceptions, though, so does anyone else have any other counterexamples?

[chris-gr]

in english, adjectives are usually placed before the nouns they modify, but english is not a SOV lang, is it?
Anyway, as linguists say, "grammars leak", and "virtually all languages include patterns that do not accord with the patterning expected in languages of their type." I merely gave some general principles on SOV langs, that's all...

[Nuntarin]

I'm very grateful for the list; this will give us something to think about in Shaja!

One question -- if SOV languages tend not to have a passive, how would English passive sentences best be translated?

Incidentally, while vlad is right that Latin was SOV with prepositions, as I understand it that is an exception to the usual rule.

[Miekko]

IIRC the tendancy for SOV langs to have postpositions isn't as strong as the tendancy for VSO to have prepositions, but I might remember wrong.

in english, adjectives are usually placed before the nouns they modify, but english is not a SOV lang, is it?

SVO langs are considered a wild-card between VSO and SOV, where almost anything that goes in either of them goes.

8. SVO lang's use conjunctions to introduce complements. SOV lang's may use nouns instead.

I didn't quite understand this one. Could you explain or provide examples?

Anyway, Bryatesle does actually fit a few of those that I didn't know about, which is nice.

[chris-gr]

Miekko wrote: "I didn't quite understand this one [i.e that SVO lang's use conjunctions to introduce complements and SOV langs may use nouns instead]. Could you explain or provide examples?"

I had in mind that english or french will normally use a conjunction (like that) in order to introduce an object clause. SOV languages may use nouns (meaning something like fact, situation). That's because it would be illogical to put a conj. at the end of a [preposed!] object clause, long before the verb is expressed.
(I could give specific examples, but not now -I'm still at work!)


AhribarII wrote: "-- if SOV languages tend not to have a passive, how would English passive sentences best be translated?"

The passive is needed in order to topicalize the action (of the verb), and dowplay the subject. SOV langs do not assign the subj. a special status, so such a voice would be needless. The japanese 'passive' in fact had honorific and potential meanings before it was influenced by western languages. I think something similar happened with turkish 'passive' too, which is a relatively late development. Khanty has a passive conjugation as an extension of the earlier topicalization system.

[civman2000]

Hmmm...this list seems heavily influenced by Japanese, though it may be that features I associate with Japanese are actually common to most SOV languages...

GOD

[Echobeats]

It's better to talk about "head-initial" and"head-final" languages, since being OV and VO is just another item on that list of characteristics that tend to go together (typologists don't appear to worry about the subject when it comes to this: it's the relative position of the object and the verb that are thought to be important). Allow me to redo your list for you:

0. Head-final languages usually have OV word order.

Examples: Turkish, Japanese, Latin, Ancient Greek.
Counterexamples: ?

1. Inflectional affixes or aux. elements are placed after the verb in head-final languages.

Largely true.
Examples: Turkish, Japanese, Latin, Ancient Greek.
Counterexamples: ?Navajo

2. Head-final languages have postpositions (instead of prepositions).

Examples: Turkish, Japanese.
Counterexamples: Latin, Ancient Greek.

3. In head-final languages, comparatives of inequality place the 'standard' before the adj. ("The dog is bigger than the cat" is rendered as " Dog cat-from big/more")

Examples: Turkish, Japanese.
Counterexamples: Latin (often), Ancient Greek (often).

4. Generally, adjectives (and genitives) come before the noun they modify.

Examples: Latin (often), Ancient Greek (often), Turkish, Japanese.
Counterexamples: ?
Counterexample to the converse: English.

5. Corollary: 'Titles' are placed after the name. Surnames are usually placed before given names.

Counterexample to the converse: Chinese (both titles and surnames).

6a. Relative clauses precede nouns in head-final languages.

Examples: Turkish, Japanese.
Counterexamples: Latin, Ancient Greek.

6b. Subordinate clauses precede the principal clause.

Examples: Turkish, Japanese, Latin (often), Ancient Greek (often).

7. The equivalent of a relative pronoun in head-final languages may be a topic marker within the relative clause.

Er, what?

8. Head-initial languages use conjunctions to introduce complements. Head-final languages may use nouns instead.

I don't think nominalisation vs. complementisation has anything to do with word-order typology, but I could be wrong. People sometimes claim that Turkish has no complementisers, but that isn't true: although ki "that" is a borrowing from Persian, they use it all the time.

9. Infinitives are rare in head-final languages. Verbal nouns are common.

What is the difference between those?

Examples: ?
Counterexamples: Turkish, Latin, Ancient Greek, ?Japanese.

10. Head-final languages express modality and tense/aspect after verb stems.

This is the same thing as 1.

11a. The interrogative marker (if any) follows the V in head-final languages.

Examples: Turkish, Japanese.
Counterexamples: Latin, Ancient Greek.

11b. Sentence negation markers, too.

Examples: Turkish, ?Japanese
Counterexamples: Latin, Ancient Greek.

12. Head-initial languages usually do not have a passive voice.

Y'what?

Examples: ?
Counterexamples: Turkish, Japanese, Latin, Ancient Greek.

13. In head-initial languages' numerals, the word for 10 precedes the lower numeral (since it is treated as a 'standard' as in comparative constructions.)

Examples: Turkish. (I have no idea what Japanese does.)
Examples of the converse: English.
Counterexamples of the converse: Latin, French, Italian, Spanish.

------------

Can anyone supply me with more data from languages other than those four?

Latin has prepositions: sub abore "under the tree", in casa "in the house", etc.

And look what happened: it switched word-orders! It's actually been found that the most reliable indicator of word-order typology is the adpositions: they're least likely to be the "odd one out".

By the way, casa is Spanish, and the Latin word for "house" (domus) takes a locative.

Yours, Tim.

[Zhen Lin]

0. Head-final languages usually have OV word order.

In modern Japanese prose, it is near impossible for the object, or for that matter, anything but mood particles to come after the main predicate. This is unusually restrictive, in SOV languages.

4. Generally, adjectives (and genitives) come before the noun they modify.

6a. Relative clauses precede nouns in head-final languages.

6b. Subordinate clauses precede the principal clause.

There are few other possible constructions in Japanese.

7. The equivalent of a relative pronoun in head-final languages may be a topic marker within the relative clause.

I don't know what is meant here, but the topic marker in Japanese is usually bound to the main predicate. There is a nominalising particle which you can attach the topic marker to, for such things as "he who went was ...".

9. Infinitives are rare in head-final languages. Verbal nouns are common.

Japanese has a sort of infinitive, the ren'youkei stem of verbs, and a sort of gerund as well, from the koto nominaliser. There a large number of verbs that are morphologically nouns.

11a. The interrogative marker (if any) follows the V in head-final languages.

Old Japanese had a free-floating interrogative marker, which survives to today as the -ka in dareka "someone", dokoka "somewhere", itsuka "sometime", as well as the true interrogative marker.

11b. Sentence negation markers, too.

Only if the negator is a verbal. In Ainu, another fairly typical SOV language, the negator is adverbial and precedes the verb. Korean has both verbal and adverbial negators.

12. Head-initial languages usually do not have a passive voice.

Japanese has a passive voice and a causative voice. The passive voice also has a number of other functions.

The passive voice may also be related to the intransitive forms of certain paired verbs.

13. In head-initial languages' numerals, the word for 10 precedes the lower numeral (since it is treated as a 'standard' as in comparative constructions.)

The Sino-Japanese numerals are completely consistent here. Old Japanese numerals follow this convention as well.

[chris-gr]

Echobeats wrote: "People sometimes claim that Turkish has no complementisers, but that isn't true: although ki "that" is a borrowing from Persian, they use it all the time. "

One objection. Turkish 'ki' does not mean that Turkish has complementisers; 'Ki' is widely used in everyday speach, but not when you want to write something formally.

Echobeats wrote: "What is the difference between those {i.e. inf. and verbal nouns}? Examples: ? Counterexamples: Turkish, Latin, Ancient Greek, ?Japanese.}"

Infinitives, as complements to finite verbs can only be understood only after the principal verb has occured. Verbal nouns have a specific meaning, thus they can be used when they are preverbal.

Echobeats wrote: "12. Head-initial languages usually do not have a passive voice.
-- -- Y'what?"

I've already answered that one.


Zhenlin wrote:
"Japanese has a passive voice and a causative voice. The passive voice also has a number of other functions."

I've already answered that one too.

ALSO: The fact that some langs present different characteristics has already been answered. As for ancient Greek passive and infinitives (since I've been studying it for the past 12 yrs.), just take a look at Homeric Greek.


Finally, I should thank you both guys (Echobeats and Zhenlin) for your scientifically excellent criticism!
I hope I answered to your objections satisfatorily!

[Echobeats]

People sometimes claim that Turkish has no complementisers, but that isn't true: although ki "that" is a borrowing from Persian, they use it all the time.

One objection. Turkish 'ki' does not mean that Turkish has complementisers; 'Ki' is widely used in everyday speach, but not when you want to write something formally.

That is completely irrelevant. We're doing linguistics here, not prescriptive pedagogy. If it's widely used in everyday speech, it's a part of the language: full stop. Facts about language are not changed by what people say about them.

What is the difference between those {i.e. inf. and verbal nouns}? Examples: ? Counterexamples: Turkish, Latin, Ancient Greek, ?Japanese.}

Infinitives, as complements to finite verbs can only be understood only after the principal verb has occured. Verbal nouns have a specific meaning, thus they can be used when they are preverbal.

I think the distinction between the two is quite language-specific, so there isn't an awful lot of point talking about it: perhaps I shouldn't have brought it up. So let's talk about what you said about infinitives. What on earth do you mean by "Infinitives can only be understood after the principal verb has occured"? Are you saying that if a Turkish person says bir kebap yemek istiyorum, his listener doesn't know what yemek means until he hears istiyorum? (For those of you who don't know Turkish, that was "I want to eat a kebab".)

The passive is needed in order to topicalize the action (of the verb), and dowplay the subject. SOV langs do not assign the subj. a special status, so such a voice would be needless. The japanese 'passive' in fact had honorific and potential meanings before it was influenced by western languages. I think something similar happened with turkish 'passive' too, which is a relatively late development. Khanty has a passive conjugation as an extension of the earlier topicalization system.

Let's take that one by one.

"The passive is needed in order to topicalize the action (of the verb), and dowplay the subject.The passive is needed in order to topicalize the action (of the verb), and dowplay the subject."

Half-right. The passive topicalises the patient and downplays (or eliminates) the agent. The patient, which in an active sentence would be the object, becomes the subject in a passive sentence. So far so good.

"SOV lang's do not assign the subj. a special status, so such a voice would be needless."

What on earth are you talking about? What kind of "special" status do they not give the subject that other languages do give it, and how is it reflected in the system? I can honestly say I have not a clue what you are on about here.

"The japanese 'passive' in fact had honorific and potential meanings before it was influenced by western languages."

I have no idea whether or not this is true, but if it is it is completely irrelevant to a discussion of whether or not Japanese currently has a passive. Some languages consider it rude to make a second-person the object, and so use passives instead to make "you" the subject. Are you getting confused with something like this?

"I think something similar happened with turkish 'passive' too, which is a relatively late development."

Again, this is not relevant, even if it is true. Turkish has a passive, and that is that. Furthermore, the fact that it shows some phonological irregularity* when Turkish has so little irregularity overall suggests that it is nothing like a late development.

*I don't mean that there aren't rules to establish whether a Turkish verb will have a passive in -il or in -in, because there are, but it's hard to find any phonetic grounding for an l/n alternation caused by the number of syllables in the stem.

"Khanty has a passive conjugation as an extension of the earlier topicalization system."

This I can well believe; in fact, it is very probable that a great deal of passive constructions come from this. Essentially, subjects are grammaticalised topics, and if (in a language that deals with subjects) it so happens that the patient is the topic, then the patient is the subject, and you have a passive.

ALSO: The fact that some langs present different characteristics has already been answered. As for ancient Greek passive and infinitives (since I've been studying it for the past 12 yrs.), just take a look at Homeric Greek.

What feature of Homeric Greek would you like me to look at, precisely? I have a good knowledge of Classical Greek and have access to lots of material about Homeric Greek. Please tell me what you are referring to, and I'll be glad to look it up.

Finally, I should thank you both guys (Echobeats and Zhenlin) for your scientifically excellent criticism!
I hope I answered to your objections satisfatorily!

Thank you . I hope you appreciate this second round of it, then.

Yours, Tim.

[Zhen Lin]

I have no idea whether or not this is true, but if it is it is completely irrelevant to a discussion of whether or not Japanese currently has a passive.

I think this is more a feature of pro-drop languages than SOV languages, or topic-marking languages. In Japanese, To put emphasis on the object, simply move the object to initial position, and optionally substitute the object marker for a topic marker.

The passive can also do more than just deleting the subject (which is easy enough to do in an active clause, just drop the subject!) - it can also become a benefactive construction of sorts.

[Radius Solis]

The passive is needed in order to topicalize the action (of the verb), and dowplay the subject. SOV langs do not assign the subj. a special status, so such a voice would be needless.

I second Echobeats' comments above, but I must also add that topicalization is only one of several functions of the passive voice; even if it's not needed, that doesn't make a passive voice useless. The first and foremost function of the passive voice (in accusative languages) and the antipassive voice (in ergative languages) is to allow normally transitive verbs to detransitivize in the "wrong" (i.e. non-default) direction. That is to say, a verb can thus be used intransitively without mentioning the argument of the verb that's normally required - an extremely handy technique without which one would be forced to include a lot more information in many sentences than is important to express. It's nearly universal to all human languages. Only a handful lack it.

[chris-gr]

1. Echobeats wrote that "If it [i.e. turkish Ki] is widely used in everyday speech, it's a part of the language".

Of course it is; but it is a relatively new pattern that has nothing to do with native turkish [i.e. SOV] syntactic patterns stock. (In fact, ki is usually omitted in everyday speach --with some rare exceptions of course.) This "SOV characteristics list" just implies that these are the things we can expect from a SOV language, but we shouldn't be surprised if there aren't there anymore.

2. Echobeats wrote:
"What on earth do you mean by "Infinitives can only be understood after the principal verb has occured"? Are you saying that if a Turkish person says bir kebap yemek istiyorum, his listener doesn't know what yemek means until he hears istiyorum? (For those of you who don't know Turkish, that was "I want to eat a kebab".)"

Well, the said listener will surely have any idea what yemek means before he hears the verb. Otherwise, yemek can (and will) be understood as a noun: "bir kebap yemek" ([a] kebap is food). Yemek will be understood as an infinitive only after the verb has occured.

3. Echobeats wrote:
"Turkish has a passive, and that is that. Furthermore, the fact that it shows some phonological irregularity* when Turkish has so little irregularity overall suggests that it is nothing like a late development."

Once again, turkish passive came from neighboring languages. It's not as much used as it's English equivalent, but it's widely used to convey impersonal meanings.

4. Echobeats asked: "What feature of Homeric Greek would you like me to look at, precisely?"

The occurence and use of the passive was rare in homeric greek, but increased after the SVO structure was established. That's what I meant

[Peacekeeper]

Could someone please explain the following statements somewhat more?

"1. Inflectional affixes or aux. elements are placed after the verb in SOV langs."
(This one I don't get at all.)

"4. Generally, adjectives (and genitives) come before the noun they modify."
(The problem here is that I am not sure which part you mean by the genitive, the owning part, or the owned part?)

"9. Infinitives are rare in SOV langs. Verbal nouns are common."
(I know this one have been explained already, but I still don't get it.)

Thanks!

[chris-gr]

1. Inflexional affixes (i.e. suffixes) are placed after the verb stem: STEM+AFFIX (affixes may be about person/number, tense/aspect, or other functions). Auxiliary verbs are usually postpositioned, as in turkish: yardIm ediyorum
HELP AUX. ("I help")

2. Although the term genitive has a broader meaning, let's say that I meant the possessor.

4. What exactly is it that you don't get?

[Peacekeeper]

Ah, sorry, I get the last one now. And thanks for helping with the other ones!

[Echobeats]

I'm still awaiting an example of an SOV language without a passive.

Tim.

[chris-gr]

before I do that, wouldn't it be nicer if you gave an answer to my answers? (the old action-reaction pattern, you know )

[Echobeats]

Fair enough:

If it [i.e. turkish Ki] is widely used in everyday speech, it's a part of the language.

Of course it is; but it is a relatively new pattern that has nothing to do with native turkish [i.e. SOV] syntactic patterns stock. (In fact, ki is usually omitted in everyday speach --with some rare exceptions of course.) This "SOV characteristics list" just implies that these are the things we can expect from a SOV language, but we shouldn't be surprised if there aren't there anymore.

The point about the generalisation "SOV languages typically lack complementisers" is that there's something about having an SOV grammar in your head that makes you averse to having complementisers. Since Turkish is still very definitely SOV, we should be expecting it not to start using borrowed complementisers. If it had had them before, we'd expect it to lose them. So I still claim that Turkish is a counterexample.

A good example would be Latin, actually. Classical Latin uses an infinitive construction: quod only became a complementiser later, when Latin was already a long way down the road to SVO-ness. Ancient Greek is not such a good example: it has both, but hoti is used quite a lot.

"What on earth do you mean by "Infinitives can only be understood after the principal verb has occured"? Are you saying that if a Turkish person says bir kebap yemek istiyorum, his listener doesn't know what yemek means until he hears istiyorum? (For those of you who don't know Turkish, that was "I want to eat a kebab".)"

Well, the said listener will surely have any idea what yemek means before he hears the verb. Otherwise, yemek can (and will) be understood as a noun: "bir kebap yemek" ([a] kebap is food). Yemek will be understood as an infinitive only after the verb has occured.

OK, so I managed to choose the one* example of an infinitive that's also a common noun . Take any other example though: İstanbul'a gitmek istiyorum. Can gitmek in the phrase İstanbul'a gitmek be understood as a noun?

If by "Infinitives can only be understood after the principal verb has occurred", you mean that infinitives depend on main verbs, and so the sentence isn't complete until the main verb (which will be after the infinitive) appears, then of course that's absolutely true; but it's no different from saying "infinitives that depend on main verbs come before that main verb in head-final languages", i.e. that main verbs are heads. If this is what you mean, then we're on the same level .

*"One" is not meant to be taken absolutely literally, but I don't think there are that many. Please correct me if I'm wrong.

Turkish has a passive, and that is that. Furthermore, the fact that it shows some phonological irregularity* when Turkish has so little irregularity overall suggests that it is nothing like a late development.

Once again, turkish passive came from neighboring languages. It's not as much used as it's English equivalent, but it's widely used to convey impersonal meanings.

But it still has one. You said that SOV languages tended not to have passives. I also want proof that the Turkish passive came from neighbouring languages: do you have a book or a paper you can cite? I've never heard that claimed before (though of course that doesn't mean it hasn't been), and from the look of the way it's formed I find it unlikely. The passive is formed by a morphological affix to the root, not by some sort of analytic or auxiliary construction as it is in English or French. Borrowing inflectional morphology is very rare.

"It's not as much used as it's English equivalent": OK, maybe so, but then Latin is SOV and uses the passive far more than English does. Low use of the passive may be a trait of Turkish, but I remain to be convinced that it's a trait of SOV languages in general.

What feature of Homeric Greek would you like me to look at, precisely?"

The occurence and use of the passive was rare in homeric greek, but increased after the SVO structure was established. That's what I meant

Ah, OK. I'll have to take your word for that ? I was supposed to read Homer last year but never actually bothered because I couldn't understand it .

Yours, Tim.

[chris-gr]

Response to Echobeats
sorry it took me 24 hrs to respond, but it is easter holiday in greece, so i prefer to drink some beers than type

As for the first 2 points, i don't have something to say, i believe we're saying the same things.

As for the passive thing: I still claim that SOV languages didn't have originally a passive voice. Wherever this voice developed, I could say that it was a late development. PIE itself hadn't a passive (Lehmann, 1993 & Beekes, 1995), until it shifted to SVO.

You say you want proof that the Turkish passive came from neighboring languages. I think (and excuse me if I didn't make that clear) that its the use of the passive that passed into turkish from neighboing languages --not the suffix -l. But consider the following facts:

⇒ -l applies to verbs of different syntactic and semantic classes (not only to transitives);
⇒ -l allows multiple occurrence;
⇒ -l is not in complementary distribution with other ?voice? morphemes;
⇒ -l frequently produces idiosyncratic syntactic and semantic effects;
⇒ -l applies not only to verbal, but also to nominal stems.

These facts, the many uses and meanings of this suffix, lead me to the conclusion that the passive meaning is not what this suffix was originally for.

Once again, I was talking about characterisics that SOV languages should have ab origine; Not 2 or 3 millenia later

[Blackhand]

As for the passive thing: I still claim that SOV languages didn't have originally a passive voice. Wherever this voice developed, I could say that it was a late development. PIE itself hadn't a passive (Lehmann, 1993 & Beekes, 1995), until it shifted to SVO.

It may not have had a passive, but it did have a middle (*-mai/-moi *-sai/-soi *-tai/-toi, *-medha (1pl)), which is used for both reflexive and passive meanings, among others.

Once again, I was talking about characterisics that SOV languages should have ab origine; Not 2 or 3 millenia later

But if you say that they had them at the beginning, isn't this one of those things that can't be proved? After all, except for pidgins and creoles (which are resolutely SVO, and really they've been evolving as well, it's just a more intensive type of language interaction than simple juxtaposition), every language on the earth has been evolving for at least 2 to 3 millenia. So where are these, as you say, ab origine languages? about 2 to 3 millenia back, where we really don't have any examples except for the endlessly confusing Sumerian.

Damn, I wish there were more mainstream sources on it that I could get access to (the local university library for some odd reason sorts its language books according to the language name, not in one general language section, and I haven't been able to find anything on Sumerian outside of a LANGUAGE entry).

[chris-gr]

blackhand wrote:

1. "It [i.e. PIE] may not have had a passive, but it did have a middle..."

2. "But if you say that they had them at the beginning, isn't this one of those things that can't be proved?"

ANSWER: so, PIE had a middle voice, or is this one of the things that can't be proved?

[Blackhand]

PIE has what we reconstruct as a middle voice, given that's what the forms mean in Ancient Greek and Sanskrit. All the evidence we have is that it was a middle voice (at least that's what I read the authors to mean), that varied in meaning from reflexive to full passive.

I'm not really sure what you're trying to say.

Also, PLEASE use quote boxes. it is really very simple. just hit the "quote" button on the upper right of the post you want to copy, and then write what you want to.

[NTiOzymandias]

Time to dispute stuff.

7. The equivalent of a relative pronoun in SOV langs may be a topic marker within the relative clause.

12. SOV langs usually do not have a passive voice.

Nope, these apply to topic-prominent languages. There's no signifigant relation between topic prominence and SOV-ness. The end.