Talk:Word order

From Wikipedia, the free encyclopedia

This article is part of WikiProject Theoretical Linguistics, an attempt to build a comprehensive and detailed guide to theoretical linguistics and theories of language on Wikipedia. If you would like to participate, you can edit the article attached to this page, or visit the project page, where you can join the project and see a list of open tasks.
To-do list for Talk:Word order: edit · history · watch · refresh

Here are some tasks you can do:


    • Greenbergian universals wrt word order
    • Some statistics and maps
    • Section on parsing and John Hawkins
    • Generative accounts of WO
    • Antisymmetry and underlying order always SVO
    • Criticisms of Antisymmetry
    • Something on problems of linguistic sampling (Dryer)
    • Expand on non-clausal orders (GEN N, STD MARK COMP, REL N, etc)
    • Some stuff on WO frequency in actual texts
    • Change patterns between different WOs
    • Maybe find some claim that all lgs are basically free WO, the only exceptions being English and French. This to counter systemic bias not of wp but of linguistics.
    Priority 5  

    Contents

    [edit] Free word order

    Also, could someone clarify what "free word order" means? I know there are idiomatic conventions about Russian word order, though alternate orders are not necessarily considered ungrammatical. Do languages with "free word order" have no constraints on order other than convention? How does one explain that in light of X-bar theory? Do different orders have different semantic meanings or the same, and the differences are a matter of emphasis? -- Beland 17:38, 6 December 2006 (UTC)

    Most common order in Russian is SVO, like "мама мыла раму", "mom washed the frame". But all of the following sentences are grammatically correct and mean the same:
    • Мама мыла раму.
    • Мама раму мыла.
    • Раму мама мыла.
    • Раму мыла мама.
    • Мыла мама раму.
    • Мыла раму мама.
    SVO violation may be used for emphasis, in composite sentences, and to mark continuous action (VSO and VOS). -- Abolen 15:14, 14 December 2006 (UTC)
    Still, that's basically topicalisation. Under Transformational grammar it would be ascribed to movement, and the language would be considered underlyingly SVO. If there was an absolutely free-word-order language, there would still be other indications as to its underlying X-bar structure. BovineBeast 18:09, 4 February 2007 (UTC)
    Which order is picked also depends on the individual person, and in some regions certain ways of speaking are more popular that others. This is only a wild guess, but perhaps SVO became the most common order in the central regions because it was the closest to other European languages. After all, Russia's aristocracy spoke mostly French for several centuries (an SVO language), so they would have found the SVO order to be most natural when speaking Russian. I wouldn't be too surprised if someone did a study about this somewhere; the effect that it had on the Russian language when the most successfull members of society spoke a different language most the time. Esn 08:27, 17 June 2007 (UTC)
    According to generativism every language is underlyingly english :P Jest aside, in answer to the the question above a semantic difference would usually lead to the order being analyzed as not free. For example, in danish word order can be SVO (active declarative sentence), VSO (interrogative i.e. a question) or even OVS (same basic semantics as SVO but with marked topicalization of the object, usually emphatic).--AxelGrenger (talk) 22:23, 6 December 2007 (UTC)
    What has been said about Russian I can see it about Spanish too, constitutents cannot be broken, but otherwise word order is quite free and the only change caused is that of emphasis, as a Romance Language its word order would be the same as English (see it appears in the same list) but I see that the following are all equally valid:
    • Giovanni mató al veloz roedor
    • Mató al veloz roedor Giovanni
    • Al veloz roedor Giovanni mató
    • Mató Giovanni al roedor veloz
    • Al roedor veloz mató Giovanni
    • Giovanni al roedor veloz mató

    The basic idea is that there was a rodent that was very fast (Think of Speedy Gonzales) and that it was killed by Giovanni (The constitutents being "al veloz roedor", "Giovanni" and "mató", well the first example aint a constitutent as much as you could say "al roedor veloz" but it is as much as you change the meaning deeply if you say "al veloz mató roedor" or "al roedor mató veloz", indeed the later makes "veloz" [fast] adverb of "mató" [killed; fastly killed] while the former makes "roedor" [rodent] an adjective for the subject [Giovanni; rodent-like Giovanni] while making the sentence more poetic)... All forms are valid and have different positions for object and verb and subject... But, what about English?

    • Giovanni killed the fast rodent--->1
    • Killed the fast rodent Giovanni--->0,5
    • The fast rodent Giovanni killed--->0,5
    • Killed Giovanni the rodent fast--->XXX
    • The rodent fast killed Giovanni--->XXXX
    • Giovanni the rodent fast killed--->XX

    Well, certainly English aint a free order language... But Spanish does works in that way (note; "al" is properly translated as "to the" with "the" being masculine in this case since Spanish have no neutral The but two gendered The's, El is masculine and "a el" becomes "al" while "a la" stays the same, for phonetic reasons, but avoid contracting "a él" into "ál", this has a different meaning and is more akin to "a ella", "a él" is more like "to him", "al" is like "to the [masculine noun]" and in this case refers to the receiver of the action.

    Anyway... Anything to say about this?Undead Herle King (talk) 05:56, 30 April 2008 (UTC)

    Spanish is surely more free in its word order than English. I agree that all 6 example sentences you give are possible, but not all of them are pragmatically neutral in the way this is understood here. For instance, mató al roedor Giovanni is only possible if the killing of the rodent is already established in discourse, and the new information provided is the performer of this killing, Giovanni in this case. Whatever the pecularities of Spanish WO, I am not really sure in how far this has an import on the article Word order as it is now. Jasy jatere (talk) 19:04, 19 May 2008 (UTC)

    [edit] other matters

    The allocation of German to the languages with SOV word-order overlooks the fact that the most basic word-order in German in "stand alone", that is, main clauses making statements is SVO. The proviso that German is subject to the V-2 rule merely highlights this. The fundamental SVO word order is particularly clear in those instances where all the determiners are omitted, as in newspaper headlines, where readers have to rely on word order (and context) to indentify the subject and object, for example: "Mann biß Hund!" (German only uses SOV word-order in suborinate clauses) - JohnC 03:53 29 Jul 2005 (UTC)

    It is often unsatisfactory to refer to a language as a "SOV" language if it merely used that word order in some contexts. I endorse the comments made by JohnC and would add that in the case of German (and Dutch) it's standard practice, with compound verbs, to treat the auxiliary as "the verb". The whole logic V-2 is based on this. For example, in a sentence like "Gestern haben wir Ilse besucht" the V-2 position or "slot" is occupied by "haben" and there is no question of the verb following the object as SOV implies. Norvo (01:30 UTC, 31 Jul 2005)

    The verb in Gestern haben wir Ilse besucht is besuchen. It's true that the part of the verb complex that's inflected for person and number occupies V2 position in main clauses in German and Dutch, but that doesn't change the fact that the basic word order is SOV. The dominant theory in syntax for German/Dutch word order is that the person/number inflection is moved from final position to second position in main clauses, not that the infinitive or participle is moved to final position. --Angr/tɔk tə mi 22:24, 25 September 2005 (UTC)
    What standard theory? Inflections can be moved? If that´s transformational grammar then it is for good reason not generally used in europe. It is a theory developed "for any language, say english". Semantically speaking the participle may be the main verb but in no traditional grammar of these languages that I've heard of is there a notion of the finite verb not being the main verb. It is with the finite that there is agreement. There is no "movement" of anything. Proto-germanic is known in fact to have changed from SOV to SVO, and that's the standard.--AkselGerner (talk) 20:59, 4 March 2008 (UTC)

    [edit] Syntax

    How exactly does "word order" differ from "syntax" and isn't the former the more appropraite term? Would a merger be appropriate?

    Peter Isotalo 07:17, 25 April 2006 (UTC)

    Syntax involves much more. Word order is an abstraction; it is not concerned with all the possible different syntactic variations, rather it looks exclusively at the relative distribution of certain sets of sentence constituents: for example french differs from english in that adjectives follow their head rather than precede them. Examples "Eminence grise" (allow for errors in spelling, I am not a french speaker) compares to "Grey eminence". This article is mainly concerned with basic word order in which only the transitive verb and its basic arguments are taken into account. This whole discussion is rather poorly conducted and I would advise not trusting anything said here, instead read the relevant chapters in Bernard Comries classic textbook "Language Universals and Linguistic Typology: Syntax and Morphology" (1981), it should be found from any major university library. --AxelGrenger (talk) 22:02, 6 December 2007 (UTC)

    [edit] Why?

    What was the motive behind agent=>patient? [anon]

    It was subject => agent. The article explained this before you deleted the explanation. Not all languages have subjects (especially in the case of the rarer orders where the object occurs before the agent), and of those that do, the S of intransitive verbs often has a different word order than the A of transitive verbs, so it's a good idea to keep them distinct. kwami 02:27, 27 April 2006 (UTC)
    The agent is an unfamiliar term to common people,
    Thus the link.
    and is incompatible with the object.
    Incompatible how?
    it must be reverted to subject.Lehi 04:09, 27 April 2006 (UTC)
    "Must"? Are you suggesting we should be inaccurate in order to avoid a word that some people are unfamiliar with? kwami 05:40, 27 April 2006 (UTC)
    Agent must only be used in conjunction with patient. Any other use of it is incorrect, and must be reverted. It is not an innacuracy, it is a correction. Thousands of books use Subject without being innacurate.Lehi 01:59, 3 May 2006 (UTC)
    You have a remarkable insight into how the world 'must' be. Please correct the many linguistic works out there which use 'agent' with 'object'.
    If you wish to change 'object' to 'patient', be my guest. That term is more compatible, as you suggest. Meanwhile please stop adding erroneous material to the article. Are you doing this on purpose? 'Subject' is incorrect, as the article quite clearly explains. kwami 05:48, 3 May 2006 (UTC)
    As soon as you correct the papers that use subject. It is not incorrect, and changing the title of an article to suit just a few languages is.Lehi 06:36, 3 May 2006 (UTC)
    I don't know what you mean by "correct the ones that use subject". Also, this typology is not restricted to nominative-accusative languages, as you suggest on your talk page. kwami 07:20, 3 May 2006 (UTC)
    I was refering to your second paragraph in your third entry
    I still don't understand what you mean. kwami 10:19, 3 May 2006 (UTC)
    Read againLehi 06:23, 4 May 2006 (UTC)
    I assume from that response that your demand is of no importance to you, so I will disregard it. kwami 06:36, 4 May 2006 (UTC)
    He is refering to your request that he "Please correct the many linguistic works out there which use 'agent' with 'object'." He wants you to correct all papers that use "subject verb object" rather than "agent verb object." Linguofreak 14:56, 5 May 2006 (UTC)

    The change of "subject" to "agent" seems remarkably like original research to me, though since the article doesn't have any references either way it's hard to be sure. It is certainly true that SVO/VSO/SOV etc. are much better known and more widely used terms in English. I would strongly support changing all the "agent"s back to "subject" and then adding a sentence explaining that in some languages "agent" would be more appropriate than "subject" (assuming sources are provided to back up that claim). In languages like English, though, it is certainly the subject, not the agent, that comes before the verb in passive sentences like The ball was kicked by John. Angr (talk • contribs) 08:04, 5 May 2006 (UTC)

    As far as English is concerned, I think calling it SVO is appropriate; on the other hand for Hindi it is really difficult to come up with a satisfactory definition of "Subject". In case of Hindi the PaNinian concepts of अभिहित and कर्ता seem more appropriate. User:Vineet Chaitanya
    A quick google search turns up over 120,000 sites that use the term "Subject verb object" vs. 144 that use "Agent verb object." Of course, popularity is not neccesarily a measure of correctness, but the fact that references to SVO outnumber those to AVO 1000 to 1 does bear consideration. I'd use SVO, and say something to the effect of "in ergative-absolutive languages the terms "Agent" and "Patient" are used rather than "subject" and "object." (EDIT: Since Ergative-absolutive languages group the patient and the arguement of intransitive verbs under one case, I'm actually not sure that we should use "patient" for "object." How do you do a strikethrough?)
    When a nominative-accusative language is being described, and when the language the description is written in is also nominative-accusative, it is much, much less confusing and more accurate to use the terms "subject" and "object." If you're dealing with an ergative-absolutive language, it's much better to use the terms "Agent" and "Patient."
    I'll make the edit I described above to try and find some middle ground we can all agree on. However, the arguement is fairly pointless, so if it gets reverted, I'll let it go, and I suggest everyone else do so to. Linguofreak 14:56, 5 May 2006 (UTC)
    OK, I've made the edit to this page. If people are happy with it sombody else can make the appropriate changes to the pages on the individual word orders. I don't have enough experience with Wikipedia to rename the articles (for example, I couldn't rename "Agent Verb Object" to Subject Verb Object because the redirect page from SVO to AVO seems to be using the name "Subject Verb Object" and blocking the move back, and I didn't want to change the text without changing the name). If people aren't happy, then they can revert this page and I won't do anything about it. Linguofreak 15:34, 5 May 2006 (UTC)
    Moving a page to a name where there's already a redirect isn't a matter of experience, it's a matter of being an admin (because it involves page deletion). I'll move the pages back once there's been a little bit more discussion here. Angr (talk • contribs) 17:26, 5 May 2006 (UTC)
    Yes, 'subject' is better for English (which is why SVO is much more common than AVO), but 'agent' has nothing to do with ergativity, as the article now implies, and when discussing basic word-order typology we don't need to worry about passives or antipassives. Also, several of the example languages don't have subjects, so the description with S instead of A in the rest of the article is simply wrong. 66.27.205.12 19:33, 5 May 2006 (UTC)
    It's true "agent" has nothing to do with ergativity, so I've removed that sentence. But I've never heard of a language that didn't have subjects! Certain constructions in some languages may not have overt subjects, but is there such a thing as a language that has no subjects at all? I'm completely baffled by this edit summary where you say "the word 'subject' does not apply to ergative languages". Of course it does! Ergative languages have subjects same as accusative languages; it's just case marking that works differently. Verbs in ergative languages agree with their subjects, regardless of whether the subject is in the ergative or the absolutive. At least I've never heard of an ergative language with verb agreement along the lines of "The man arrives" (intransitive verb agrees with subject) vs. "The man see the boys" (transitive verb agrees with object). Angr (talk • contribs) 20:15, 5 May 2006 (UTC)
    In ergative languages the ergative case corresponds to the agent, just as the nominative case corresponds to the subject in nom-acc languages, so I thought that distinguishing SVO word order for nom-acc languages from AVO for erg-abs would clear things up. Was I overlooking anything? Linguofreak 21:39, 5 May 2006 (UTC)
    But in SVO, S is A, so that's rather confusing. Also, when people use A, they restrict S to intransitive verbs (and sometimes replace O with P).
    I made a mistake in using the term 'agent'. That is, of course, where the A comes from. But 'A' doesn't have the semantic connotations of the term 'agent' which you're objecting to. You're quite right to do so. Instead we should use A, S, and O as fundamental terms rather than as abbreviations. kwami 23:50, 5 May 2006 (UTC)

    In ergative languages, the ergative case is used for the subject of a transitive verb (which in active sentences will also be the agent), while the absolutive case is used for the subject of an intransitive verb (which will also often be the agent, depending on the semantics of the verb in question). The word order depends on the syntactic subject of the sentence, regardless of whether or not this is also the semantic agent, and regardless of whether the language is ergative or accusative. The ergative language/accusative language distinction is AFAICT a complete red herring to the SVO/SOV/VSO... issue. Angr (talk • contribs) 22:44, 5 May 2006 (UTC)

    Not quite. The ergative case is used for what would be the subject of a transitive verb in an accusative language, while the absolutive case is used for what would be the subject of an intransitive verb in an accusative language. But the term 'subject' is meaningless in ergative languages themselves. Neither case corresponds to topicality or other aspects of a subject, so there's no semantic, syntactic, or discourse equivalence, and of course there's no morphological equivalence. It's similarly meaningless in Philippine-type languages, which is why in the 1960s there was so much debate as to what the "true" subject was in these languages. There are even nom-acc langs which do not have subjects, like the nominative-marked Cushitic languages. Even in some languages with subjects, it's unhelpful to use the same term for the A of transitives and the S of intransitives, because they don't follow the same typology. In Russian, A precedes verbs, but S does not. kwami 23:50, 5 May 2006 (UTC)

    The biggest problem with ergative languages is that they generally have some nom-acc features (and often vice versa). Take the example of Russian, which from your description seems to be a nom-acc language with word order consistent with an ergative language.

    But the basic problem is that whether it is best to use "SVO" or "AVO" depends on the language. In English the subject is fronted, but the agent can be pushed to the end of the sentence (or even ommitted) in a passive construction. In Russian the agent is always fronted, but the subject is kicked to the end when it isn't the agent. There may well even be languages where neither subject nor agent is consistently before the verb. Linguofreak 02:18, 6 May 2006 (UTC)

    Well, yeah: VSO languages. Except that actually there is no such thing as a VSO language. The Celtic languages, for example, are usually called VSO, but in fact they're InflSVO. Kwami seems to be using a very different definition of "subject" from what I can find in all the linguistics I can find, which invariably refer to ergative languages as having subjects. To take the Basque examples from Ergative-absolutive language, every description I can find calls gizona the subject of Gizona etorri da and gizonak the subject of Gizonak mutila ikusi du. The fact that they're in different cases doesn't mean they're not both subjects. As I mentioned above, to the best of my knowledge, ergative languages still have subject/verb agreement regardless of whether the subject is in the absolutive or the ergative. Angr (talk • contribs) 07:51, 6 May 2006 (UTC)
    Ergative language transitive subjects are marked in absolutive, Gizona is subject of the first sentence, mutila is subject of the second. Beware optional topicalization in the word order context; an example may show the same word order for absolutive subject and ergative object simply because the word order can be modified for emphatic topicalization. The basic word order may be very different. In the basque transitive example the word order is OSV, the typologically least favored type, however this is not relevant if word order is flexible or free. (in short, you're both wrong)--AkselGerner (talk) 23:39, 28 February 2008 (UTC)
    First of all, my mistake in calling A the 'agent'. That was very sloppy of me. In word-order typology, A is the more active argument of a transitive verb, whereas S is the only argument of an intransitive. This is standard typological nomenclature. An English active clause is AVO (or AVP if you prefer; it makes no difference), whereas an English passive clause is SV. (Passives are intransitive by definition, and so play no part in the AVO order.)
    AFAIK, verbs in erg langs agree with the absolutive, which is why the absolutive case is often called the 'subject'. Actually, in all cases I can think of, the verbs agree with both arguments, but it's the O and S which take the same agreement markers. kwami 08:39, 6 May 2006 (UTC)
    The AOS terminology is not widely applied to word order typology. The AOS terminology is used especially in showing the differences between accusative(S/A vs O) and ergative (S/O vs A) languages (and tripartite: S vs A vs O). Subject is not the same as S in this terminology, instead subject equals the S/A conglomerate of accusative languages and the S/O conglomerate of ergative languages. Basic word order however is not concerned with the semantics involved, S of SOA is not S(ubject) of SVO and O of SOA is not O(bject) of SOA except in accusative languages where the termonologies just happen to overlap (for purely etnocentric reasons). Remember also that ergative languages frequently have nominative-accusative type subsystems for pronominal arguments.--AkselGerner (talk) 23:23, 28 February 2008 (UTC)

    [edit] Factual question on markedness

    That word order is unmarked. That is, it contains no extra information to the listener.

    That's a bit confusing. In languages with looser constraints on word order, it's my understanding that there are morphemes that indicate grammatical case, to distinguish subjects vs. objects, etc. In that sense, these languages are very much "marked". In languages without case-marking morphemes, word order does carry information. Is this what this passage is talking about, or is it trying to make a different point of some kind? -- Beland 17:38, 6 December 2006 (UTC)

    The point is that of markedness ([mar-ket-ness] not [markt-ness]) theory. In this context marking (as in morphological marking) is not the same as markedness, rather markedness refers to a member of a set being less default than another member (less marked). Usually the member of a set that is least marked is referred to as unmarked. The distinction can be traced in many directions, some linguists use a purely psycholinguistic measure others use a more typologically based measure. See the book Language Universals and Linguistic Typology: Syntax and Morphology" (Comrie, Bernard, 1981).--AkselGerner (talk) 23:09, 28 February 2008 (UTC)

    [edit] Word order vs transitivity in Russian

    Russian, for example, has SVO transitive clauses but free order (SV or VS) in intransitive clauses.

    Actually the Russian word order is free both for transitive and intransitive clauses, as Abolen mentioned. It is true that SVO (SV) is usually the unmarked order (so, Сел(V) Балда(S) на кобылу верхом - (he) sat(V), Balda(S), on the horse's back is marked: Pushkin emphasizes that Balda didn't tried to carry the horse with his hands).

    However, this can depend on many things, not only transitivity.

    For example, шёл(V) дождь(S) - (It) went(V), the rain(S), i.e. "it rained" is unmarked, whereas Светило солнце, но дождь(S) шёл(V) - the sun was shining, but the rain(S) went(V) is marked (the fact of rain is underlined). At the same time я(S) шёл(V) лесом - I(S) went(V) through a forest is unmarked and шёл(V) я(S) лесом is marked.

    Another example. наступила(V) ночь(S) - (it) come(V), the night(S) is unmarked and ночь(S) наступила(V) is slightly marked (it is emphasized that either (1) the night come (not finished) (2) it is night (not day) that come. The oral speech distinguishes between (1) and (2) via stress). On the contrary я(S) наступила(V) вам на ногу - I(S) stepped(V) over your foot is unmarked, вам на ногу наступила(V) я(S) is marked (It's me who hurt you).

    Thus the unmarked word order for the same verb depends on the kind of subject.

    [edit] Missing sections:

    It is absolutely necessary to add a section on constituent word order. There is much more to word order than basic word order and this should be shown in greater detail. Greenbergs universals should be mentioned. Languages without unbound nouns should be mentioned (polysynthetic and nounless). This should be done in the context of morpheme order which is also missing.--AkselGerner (talk) 22:20, 29 February 2008 (UTC)

    I agree that the article is far from complete, but might it be the case that the section on phrasal word order covers what you term constituent word order? Anyway, you can always be bold and add those sections, we really need more people working on coverage of linguisticsJasy jatere (talk) 11:51, 12 March 2008 (UTC)
    I am giving it a try. Thanks for fixing my typoes and wiki-formatting errors.--AkselGerner (talk) 21:08, 17 March 2008 (UTC)

    [edit] Extraordinaire?

    This is obviously a french loan, and is used specifically when wanting to be, shall we say, flamboyant. It can be analyzed as a single-word french paraphrase, especially since it has an english-proper version (also a loan of course) namely extraordinary. As such it is not the best possible example, are there others?--AkselGerner (talk) 21:19, 4 March 2008 (UTC)

    [edit] Problem with generative explanation

    [edit] Generative explanations

    This trend to have S precede both V and O can be accounted for by X-bar theory. In such an analysis a sentence is treated as an inflectional phrase specified by the subject of the sentence. As the specifier of any phrase almost always precedes the rest of the phrase in human language it stands to reason that the subject of any inflectional phrase would follow the same pattern, resulting in the observed preference for subject initial word order...

    The problem is that this argument is cyclic, hence void. The subject is treated as a specifier because it comes before the verb phrase in english. I am not saying that there is not a valid generativist explanation, I'm just saying that a cyclic argument is hardly it. Can someone fluent in Chomsky check for the real explanation?--AkselGerner (talk) 20:37, 4 April 2008 (UTC)


    I fear this is the real explanation advanced in most textbooks. Richard Kayne has the antisymmetry hypothesis, which goes one step further, but is still no "real" explanation, but a reformulation. Furthermore, the fact that most languages are not SVO but SOV makes this explanation seem unlikely. I think in Newmeyer 2006 there should be a good treatment of this, but alas I will not have the time to write that part myself. Jasy jatere (talk) 09:41, 7 April 2008 (UTC)
    Well, I can't check it either since I am allergic to generativism :(. Maybe the whole section can be reconstructed somehow. There's no real point in having the "explanations" like that, and especially not worded like they are somehow contradicting each other. I find it hard to believe that generativists would specifically object to the functionalist explanation as given, as it is too vague to contradict anything. I suggest that perhaps we cut out the generativist explanation, reword the "functional explanation" and merge it with the "Functions of WO"-section. That would give something like the following:
    ==Functions of sentence word order==
    A fixed word order is one out of many ways to ease the processing of sentence semantics. As we are exposed to a speech stream we must attempt to pin down the relevant relations of the things referred to in that stream. With no clues whatsoever of the relations we are less able to be certain that we have understood what is said, as a result of this the speakers might have to use simpler explanations of events or to clarify constantly what was just said, either preemptively or because the hearer fails to understand. One means of codifying the speech stream to be less open to ambiguity (complete removal of ambiguity is probably impossible) is to have a fixed order of arguments and other sentence constituents. This works because speech is inherently linear. Another means would be to modify the constituents in some codified way, say morphologically. These are not the only means[citation needed]. Fixed word order reduces expressiveness but added marking increases information load in the speech stream.
    Knowledge of word order on the other hand helps to identify the thematic relations of the NPs in a clause. If we can identify the verb in a clause, and we know that the language is accusative SOV, then we know that Grob smock Bzug probably means that Grob is the smocker and Bzug the entity smocked. Languages with no fixed word order cannot use this device for identifying the roles of the NPs and often (but not always) resort to other means, like case marking or agreement. Also see Marking.
    The mentioned functions of word order can be seen to affect the frequencies of the various word order patterns: An overwhelming majority of languages have an order in which S precedes O and V. Whether V precedes O or vice versa meanwhile is a different matter, with both arrangements commonly found in the world's languages.
    Observing discourse patterns, one finds that old information (topic) tends to precede new information (comment). Furthermore, acting participants (like humans) are more likely to be talked about (to be topic) than things simply undergoing actions (like oranges being eaten). If acting participants are often topical, and topic tends to be expressed early in the sentence, this entails that acting particpants have a tendency to be expressed early in the sentence. This tendency can then grammaticalize to a privileged position in the sentence, the subject.
    How's that? --AkselGerner (talk) 21:40, 7 April 2008 (UTC)
    Why don't you WP:be bold? For some more thourough feedback: The use of "we" seems somewhat strange to me, also it looks a bit like original research, so some refs would be good. I am personally fond of a lot of subsectioning, but other people have different opinions. Anyway, go ahead, the article needs you. Jasy jatere (talk) 08:52, 8 April 2008 (UTC)
    It's that consensus thing :p I would imagine that some would think I'm POV if I take out the generative explanation, but I really don't feel it adds anything of value. And the functionalist explanation is hardly controversial it's pretty much straight out of Comrie. None of that stuff is my own writing BTW, I've just tweaked some of it, like the changing of "jack hits tom" to something more enlightening, and more weird. That might be questionable, but an english example definitely doesn't show the details involved. Comrie has it all, but I guess I should find the page numbers, that'll take a bit of time.--AkselGerner (talk) 20:34, 8 April 2008 (UTC)
    Oops, now I see what you mean. Yes, I did write that we-thing... I'll look it over, it can probably just be truncated into oblivion.--AkselGerner (talk) 20:48, 8 April 2008 (UTC)

    [edit] POV problem

    I believe this article is biased towards a formalist generative understanding of word order. I don't think for example that there is a wide consensus among linguists that a concept such as "Basic word order" is ubiquitous. I think the article downplays languages in which word order is primarily used for encoding pragmatic information instead of grammatical relations and the article also seems to lack a discussion of non-configurationality in general. It also seems to take for granted that word order in all languages can be described according to the "subject, object, verb" distinctions - but this is an oversimplification since other distinctions such as agent, patient, recipient, theme, topic, focus, etc. can be equally basic concepts when describing word order in individual languages.·Maunus· ·ƛ· 20:55, 8 April 2008 (UTC)

    Greenberg however can hardly be described as a formalist and certainly not a generativist, and it has been added in several places that there are languages not conforming to the given patterns. I do agree that it is not enough, and also reduces clarity. It probably calls for a rewrite of several sections. Go ahead and contribute.--AkselGerner (talk) 21:10, 8 April 2008 (UTC)
    No, of course not - I was referring to the relative weight given to sections when I said it was biased towards a generativist formalist view. My reluctance to contribute to the article is mostly because I don't feel I know enough about the subject.·Maunus· ·ƛ· 21:21, 8 April 2008 (UTC)
    I wrote a long reply to that, but the read-only-daemon ate it :( To sum up; you can always read up on the matter using the general and specific sources given in the article, and/or sources of your own. And you can also point out exactly what you see as problematic, with a long multi-section article it is hard to keep track of what is wrong where, fresh eyes see better.--AkselGerner (talk) 21:24, 9 April 2008 (UTC)
    BTW, if you want to see an example of fer real generativist bias, take a look at Language_acquisition, it's a mess! It would take so much effort to clear it up that I don't want to even think about it. I hit it with the truncation stick a bit, but when an article is so POV it gets really difficult to know where to even begin. The same problem occurs when an article is jumbled and states the same thing in different ways several times, like the Word order does.--AkselGerner (talk) 21:41, 9 April 2008 (UTC)
    Could you point out where you see generativism creeping in? The only clear generativist passages I can find are the links to Branching and Head Directionality Parameter. Section 1 and 2 relativise the concept of BWO a lot, Section 3 gives functional explanations.
    I am certainly no generativist (God beware), but I do actually think that there should be more coverage of generative approaches to WO, possibly with criticisms, where applicable. Jasy jatere (talk) 13:06, 9 April 2008 (UTC)
    I agree, there should be adequate coverage of all theories with enlightening perspectives on WO. The problem is that if you give Chomsky an inch he'll take the whole nine yards, in other words, it's difficult to find an expert on generativism who isn't either fanatically for or against it, and neither give a good basis for an encyclopedic treatment. Specifically the generativist insider terminology is a problem, a generativist finding can be accurately described in generativist terms but in a way that is in no way accessible to non-generativists. Other theories can be extremely complex but still it is usually able to paraphrase to a human language without years of dedicated study.--AkselGerner (talk) 21:24, 9 April 2008 (UTC)
    I think I probably overstated the part about generativism. What I don't like is the matter-of-factish style in which the therms S V and O are used as if they're completely unproblematic and agreed upon as being the best description of wordorder in all languages. What I would like is forstatements to be duly qualified such as "some linguists believe that ", "linguists of the generativeist chool believe that", "Joseph greenberg proposed that" etc. So that points of view are stated as such, along with their contrary points of view.·Maunus· ·ƛ· 05:57, 10 April 2008 (UTC)
    I'm not sure that would be entirely WP, disclaimers make the articles unaccessible rather than informative. Matter-of-factishness is of course one of the many charms of generativism and of poor scholarship as well, and that can usually be reformulated by clarifying the definitions and providing the missing justifications. What could be added is a section "What it means when a language has Subject and Object:" in which can be clarified the implications and justifications of the terms. In such a section it can be explained that language might not have subjects and objects, and might not have arguments at all. Comrie has subject as simply a conventionalized correlation of Topic and Agent (in accusative languages), from which follows that non-agent subjects and non-topical subjects are justified as subjects through abstraction and associative principles, the archetypical subjectness is no longer tied to the parts that originally constituted it. Find some refs for what you have in mind, bring them here for discussion or WP:be bold! --AkselGerner (talk) 20:24, 10 April 2008 (UTC)

    [edit] How about we start with the sources?

    If we concentrate as a start on finding good reliable sources, preferably online, then it will be easier to organize the rewrite/reorganization. I'll try and find the sources on Dikian Functional Grammar, it's functionalist but quite formalist in the original sense of the word, they have some very astute observations on the layering of clause structure, not in terms of deep structure or underlying form (both of which are highly suspect because they usually synchronify as processes what can be proven to be diachronic "processes" of the past), but in terms of parts of speech. They have a lot to say on word and morpheme order.--AkselGerner (talk) 20:29, 9 April 2008 (UTC)

    Well, so far I haven't found exactly the ref I was looking for but this has a short description of what I mean: powerpoint presentation on functional grammar --AkselGerner (talk) 21:51, 9 April 2008 (UTC)

    [edit] Functions of word order problem

    The section currently reads:

    [edit] Functions of sentence word order A fixed word order is one out of many ways to ease the processing of sentence semantics and reducing ambiguity. One method of making the speech stream less open to ambiguity (complete removal of ambiguity is probably impossible) is a fixed order of arguments and other sentence constituents. This works because speech is inherently linear. Another method is to label the constituents in some way, say morphologically. These are not the only means[citation needed]. Fixed word order reduces expressiveness but added marking increases information load in the speech stream. Knowledge of word order on the other hand helps to identify the thematic relations of the NPs in a clause. If we can identify the verb in a clause, and we know that the language is accusative SOV, then we know that Grob smock Bzug probably means that Grob is the smocker and Bzug the entity smocked. Languages with no fixed word order cannot use this device for identifying the roles of the NPs and therefore often (but not always) resort to other means, like case marking or agreement. Also see Marking. The mentioned functions of word order can be seen to affect the frequencies of the various word order patterns: An overwhelming majority of languages have an order in which S precedes O and V. Whether V precedes O or vice versa however has been shown to be a very telling difference with wide consequences on phrasal word orders[7]. Observing discourse patterns, one finds that old information (topic) tends to precede new information (comment). Furthermore, acting participants (especially humans) are more likely to be talked about (to be topic) than things simply undergoing actions (like oranges being eaten). If acting participants are often topical, and topic tends to be expressed early in the sentence, this entails that acting particpants have a tendency to be expressed early in the sentence. This tendency can then grammaticalize to a privileged position in the sentence, the subject.

    The difference between the first and second paragraph is one of two different meanings of "function". The original version (which I heavily modified) focused only on the meaning "what is the purpose of the study of word order?", ignoring the more relevant meaning "what function does word order serve in languages?". I wished to highlight the latter and clarify the former (which was very vague and contained errors of fact). The fact tagged sentence in the first paragraph is a relic from that version because I couldn't find a source to deny that claim. I see the problem, but I would prefer the paragraphs to be merged rather than just the first one be taken out. I guess it's just my just reward for failing to WP:Be bold enough in the first place.--AkselGerner (talk) 20:27, 19 May 2008 (UTC)

    Tweaked it a bit more, merged the most aggravating repetitions to the more relevant paragraph, reordered most of the paragraphs. Is that better?--AkselGerner (talk) 20:45, 19 May 2008 (UTC)