Consider languages whose case-systems allow the order of arguments to be changed without changing the arguments’ grammatical relations. (Note the 189 languages noted as having “no dominant word-order” at The World Atlas of Language Structure Online: Feature 81A: Order of Subject, Object and Verb .) In such languages, what information is word order typically used to convey?
As a native speaker of Polish (and a linguist), I can say this: in Polish, word order is almost absolutely free; the only constraint I can think of is that prepositions (e.g. w 'in', z 'from') can't be placed after the word they refer to. Apart from that, pretty much any ordering is always understandable. It is true, however, that in some sentences some orders will sound more natural, while others will be considered emotional or emphatic, as Aspinea explained above for Latin. An example:
Jaś poszedł do kina. Johnny went to cinema. The first association: natural, affirmative, simply stating a fact. But could be a question if so intoned, or perhaps anything else, too. Poszedł Jaś do kina. The first association: half a sentence, expecting the second part (e.g. ... and he saw a ghost there.) But could as well be a confirmation that he did indeed go, a question, or perhaps anything else, too. Jaś do kina poszedł. The first association: answer to the question "Where did John go?". But again, could be used in plenty of other contexts, too. Poszedł do kina Jaś. A tiny bit odd. The first association is that "Jaś" was added as an afterthought (could have been without the tiniest pause, though). Do kina Jaś poszedł. The first association: answer to the question "Who went to the cinema?" But again... Do kina poszedł Jaś. The first association: expect an enumeration who else went to the cinema. But again...
That's all the possibilities, right? You might think SVO would be the most frequent one. Well, that's the first thing that comes to one's mind when they hear the English sentence "Johnny went to the cinema", but I really don't think that in real life it is so much more frequent.
At any rate, take a look at the above examples and imagine that essentially any sentence in Polish can be reordered in an equally free way, with next to no consequences at all. In more complex sentences with subordinate clauses and such, certain orders will be more natural as easier to understand, but that's it: more natural, definitely not the only ones. I think this goes a long way to explain why the majority of linguists who work with languages with more inflection than English, do not share the Anglo-Saxon obsession with syntax and object to using 'syntax' and 'morphology' interchangeably. This includes generativism. I'm quite convinced the only reason it won any following at all in continental Europe is the power of the US and anything coming from there being automatically fashionable.
PS. A very distant afterthought: a mediaeval Polish shibboleth to distinguish Germans was miele młyn [ḿjεlε mwɨn] lit. ‘grinds mill’. In theory, it could be SV just as well but somehow this sounds just fine in modern Polish, too.
As Cerberus hints in his comment on Aspinea's answer, it's quite common to use word order to convey information structure. (Sadly, Wikipedia doesn't have an article on the relevant sense of the phrase "information structure," and I can't find a good nontechnical introduction elsewhere. The article "Basic Notions of Information Structure" by Manfred Krifka is a good survey of the subject for linguists.)
Information structure is a pretty broad subject, but there are two basic ideas that are especially important when you're talking about word order: topic and focus. The topic of a sentence is, roughly speaking, "what the sentence is about". Usually, a sequence of sentences will have the same topic, or at least closely related topics. But sometimes you'll want to change topics between sentences, or to contrast two topics with one another — and speakers will tend to mark the new topic in some way when they do this.
English actually does mark topic changes using word order, at least some of the time. In passages like these...
Mary likes beans. Me, I can't stand them.
Mary likes beans. Lentils, she doesn't like so much.
...the left-dislocated word in the second sentence ("me" or "lentils") is marking a change in topic: "Okay, we were talking about beans before, but now we're gonna switch gears and talk about lentils for a while."
The other important idea is focus. Focus actually has a bunch of different uses (see the Krifka article I linked up above for the gory details) but one of the biggest uses is to mark the answers to questions. In English we use prosody to mark focus, like so:
"Who likes beans?"
"MARY likes beans."
"How does Mary feel about beans?"
"Mary LIKES beans."
"What does Mary like to eat?"
"Mary likes BEANS."
But some languages use word order to mark focus too.
So let me give some examples from K'ichee', which is a "free word order" language that I've studied for a while. Default word order in K'ichee' is VOS, so the standard way of saying "Mary likes beans" would look like this:
Karaj kinaq' ra'l Mari'y "Mary likes beans" (literally "Likes beans miss Mary")
In K'ichee', both topics and foci can move to the left of the clause. After a topic, you get a pause; there's no pause after a focus; and if both the topic and focus move, the topic goes first. You can also move words to the end of the clause, which makes them sound a bit like an afterthought. So the "real" word order in K'iche' is more like "Topic (pause) focus V normal-O normal-S (pause) afterthought," though of course you won't get all those slots filled in a single clause.
Anyway, this lets you say things like these:
Kinaq' karaj ra'l Mari'y "Mary likes BEANS"
(This answers the question "what does Mary like?")
Ra'l Mari'y (pause) karaj kinaq' "As for Mary, she likes beans."
(This changes the topic: we were talking about someone else, now we're talking about Mary)
Ra'l Mari'y (pause) kinaq' karaj "As for Mary, she likes BEANS."
(This changes the topic and answers a question. For instance, maybe you asked a minute ago about what someone else likes, but now you're asking what Mary likes, and I'm answering the second question)
Karaj ra'l Mari'y (pause) kinaq "Mary likes 'em.... likes beans, that is."
(This makes "beans" an afterthought. You might use it when you've been talking on the topic of beans for a while, but you want to remind your listener that that's what you're talking about.)
And so on. The upshot is, all six permutations of subject, verb and object are possible in K'ichee', given the right combination of topic, focus and afterthought.
For what it's worth too, I don't speak Polish but it looks like something similar is going on in Kamil Stachowski's Polish examples. Note that sometimes he says that a particular word order answers a specific question. This suggests to me that those word orders are being used for focus-marking in Polish — and a little Googling turns up this article on Polish focus-marking claiming that word order is indeed relevant.
In Slavic languages (I am a native speaker of Russian) the relatively free words order is often used to convey the information that otherwise would have been rendered using articles (which we don't have).
Here are some examples:
Мальчик вошел в комнату.
(literal) Boy entered room.
(meaning) The boy entered a room.
В комнату вошел мальчик.
(literal) Room (object) entered boy (subject).
(meaning) A boy entered the room.
So, as you can see, we use the word order as a sort of mild emphasis that is rendered in other languages using definite and indefinite articles.
I can think of the following effects Latin can achieve with its word order:
- Intelligibility. When words that syntactically belong together stand close to each other, the sentence is easier to understand than when they are far apart.
- Emphasis. A word placed in a prominent position within a sentence -- such as the first word of the sentence -- is emphasized.
- Formation of Rhetorical Figures. An alliteration, for example, is easily formed if you can freely rearrange the words in a sentence.
- Poetic Prosody: Latin poetry doesn't rhyme, it builds rhythmical patterns instead. In my experience, Latin poetry is rather hard to translate, because word oder serves the purpose of language rhythm rather than that of intelligibility.
In the accompanying text to chapter 81, Dryer notes that some languages are treated as lacking a dominant order because two different word orders are conditioned by syntactic factors. German and Dutch are thus treated as lacking a dominant word order because main clauses containing an auxiliary have SOV order, while main clauses without an auxiliary have SVO order. (In languages where two different word orders are conditioned by the main/subordinate clause distinction, the language is assigned the value found in main clauses)
To take another example, Idoma (Nigeria) is treated as lacking a dominant order because (according Dryer's source, Abraham 1951: 18) in the past tense the object may come either before or after the verb.
I suspect that a modern-day analyst might try to investigate whether the word order difference in terms of information structure (where the form with the object before the verb would show focus on the verb rather than on the object). When certain word orders can be tied to focus-like properties, Dryer does not treat these as competing for dominant order (they are said to be "pragmatically marked"). Aghem (Cameroon), for example, is treated as SVO because even though various word orders are attested, Dryer's source (Watters 1979) gives detailed arguments for why word order changes are tied to focus. Since the non-SVO orders are all "pragmatically marked," they are treated as non-dominant orders, and the language is treated as being SVO-dominant.
Like a main clause/subordinate clause split, word order changes conditioned by a polarity split also do not appear to cause a language to be treated as lacking a dominant order. Leggbo (Nigeria), Bafut and Mbili (both Cameroon) are treated as SVO-dominant, even though they have non-SVO word orders in negated clauses (see chapter 144C). Apparently, negated clauses are also treated as "pragmatically marked."
In Turkish (I am a native speaker) word order, or more correctly the order of the constituents of a sentence (because, e.g., postpositions still have to follow their complements) is more or less free. The neutral order is SOV and other orderings are mainly used to emphasize an element. Generally, the element that immediately precedes the verb is emphasized. (Ali = a Turkish first name; dün = yesterday; okula = to (the) school (dative of "okul"); gitti = went (3rd person singular, past tense))
Ali dün okula gitti. (Ali went to school yesterday.) Dün okula Ali gitti. (It was Ali who went to school yesterday.) Ali okula dün gitti. (It was yesterday that Ali went to school.)
Please note that in spoken Turkish, stress alone can be used for emphasis with any word order.
Non verb-final orderings are also possible and frequent in colloquial and poetic language. Within the latter, even constituents of phrases can be broken apart and moved around. So, word order can also be a stylistic device.
Hungarian is a language with completely free word order.
Let's have an example: I was in the house.
In English, you could say something like "In the house I was..." but it's more poetic and generally not used in common speech. In Hungarian, all 6 permutations can be used. (I = én, was = voltam, in the house = a házban)
- Én a házban voltam = I was in the house, I don't know where the others were
- Én voltam a házban = It was me in the house, not someone else.
- A házban voltam én = I definitely visited the house, maybe other locations, I didn't. (used with an exclamation mark)
- A házban én voltam = In the house it was me, not someone else. At other places, maybe not.
- Voltam én a házban = these last two are rare, but can be used as a defense against an accusation that I wasn't in the house, or that I was somewhere else and not in the house.
- Voltam a házban én
Note, that if the person is not that important, or not accentuated, it is usually omitted, and some of the above sentences may sound more natural without the pronoun. They are correct, however, even with the pronoun.
Now you can add a time (like yesterday) to have 24 permutations, and the name of a person (who was there with me) to have 120 permutations! I did not check all of them, but did a mental check on a number of them and they are correct.
I'm no linguist, but I learned a bit of Hungarian a few years ago, and discovered that word order is pretty free, and that it can be used to convey emphasis. English lacks word-order freedom, so emphasis is conveyed by inflection in the spoken language and typographical conventions (e.g italics) in the written language. My Hungarian friends tell me this is largely unnecessary in Hungarian because of the extra semantic dimension provided by the word-ordering.
First, let us define what'a free word order' is and how the word order fuctions in different languages. E.g. Russian or Polish are reputed for being 'languages with free word order', but you never find a preposition at the end of a sentence in those languages.
Within the atlas, Russian is classified as a SVO language (together with Mandarin).
The word order in Russian and Polish carries the meaning of theme/rheme, or the same meaning is conveyed with Japanese -wa/-ga nominative markers.
Speaking on Japanese, its word order is not quite 'free', either (unless we consider phrases like 'watakusi-mo' (me, too) or 'heya-ga!' (the house!) being sentences. Basically, an elaborate Japanese sentence ends up with a predicate. Hence, we need a universal definition for a sentence or, for that matter, for a predicate. Japanese is classified within the same atlas as a SOV language.
In Mandarin Chinese, the word order is used to define what part of speech a given 'word' is. Hence, we should also define 'a word'. According to the atlas, Mandarin falls under the same category as Russian, Finnish or Swedish (SVO).
The two latter languages show a 'free word order', but never in sentences with subject-object clauses (hence, the examples with 'Ali went to school' and 'malchik voshel v komnatu' are wrong examples; the right ones are like 'Ali gave a present to a friend' or 'On arendoval komnatu na paru let').
Thus, the only language on the atlas list with a proper 'free word order' is Hungarian, where the word order conveys the meaning of theme/rheme dichotomy.