Free Word Order Languages: How Much Freedom?

Question

Persian, Russian, German, Turkish and Czech are generally described as free-word-order languages, but do you know any quantitative, corpus-based, or information theoretic definition of word order? Is it a matter of zero/one (to be free-word-order vs. not to be free-word-order) or it is indeed measured on a spectrum? If so, what is the exact measure?

Answer 1

As many pointed out, 'free word order' is a complete misnomer. I don't know of any language where the label would apply without at least some qualification. As TKR mentioned word order is motivated by information structure and pragmatics but also can display certain tendencies of iconicity and be subject to language default. That's why a language like Russian can be called free-word-order but also SVO at the same time.

There is certainly no way to quantify the free-word-orderedness of languages. Ultimately word order can be twisted in any language in almost any way for a good purpose - for instance poetry. Context and other grammatical elements then function as disambiguators.

The standard explanation for the relatively fixed word order in English is that it is used to encode grammatical information. So in John loves Mary vs. Mary loves John the agent and patient are expressed by position in the sentence whereas in a language like Czech, they would be expressed by a case marker. This is certainly correct in the broad strokes. But even in morphologically rich languages there is a huge amount of homonymy (expecially in the plural) so ultimately, it is the context that differentiates between the agent and patient (there's a Czech sentence that has 126 possible parses because of this). Equally, in English there are all sorts of raising phenomena that make the picture more complicated, as well. Even in English context and semantic defaults can override the word order rules. Whereas in John loves Mary, reversing the word order gives a sentence with a different meaning. When you compare John ate an apple and An apple ate John you see that it's more plausible to imagine both sentences with the same meaning but used in different contexts rather than a scenario where John would have been eaten by an apple.

On the other hand, in a context where order order is supposedly completely free such as coordination, iconicity (i.e. word order mimicking real world features such as sequence of events) can remove a lot of your freedom. So, you're more likely to say 'bread and butter' than 'butter and bread' when you say preceded with 'I ate some' whereas the word order is more free when you precede it with 'I bought'. And there's no freedom at all in 'He opened the window and shouted out obscenities.'

Finally, to illustrate how the word order matters in a 'discourse-configurational' language, let me tell you a little joke that appeared shortly after the velvet revolution. It's about two Moravians (who have a healthy disdain for Czechs). One says to the other. I have good news and bad news. Good news is "Češi udělali revoluci" (Czechs started a revolution) bad news is "revoluci udělali češi" (it's the Czechs who started the revolution). So you can see that although the word order appears to be free at the propositional level, it actually encodes completely different communicative meanings.

Now, when you look at this range of phenomena (and there are more), you can see that it would be impossible to quantify. In any language, the situations where you have a completely free choice in how you order the words are very rare and on the other hand, no language completely bars you from twisting the word order for a particular purpose. You could certainly measure the defaults and prevailing orders but that won't show you freedom but rather a distribution of the different configurations each of which have very limited freedom of order.

Answer 2

I tried to add it as a comment, but I couldn't so I post it here. I think there is a misunderstanding between free word order and scrambling languages. Free word order languages are those that do not have any order. It is a property of non-configurational languages. (Of course there is a big discussion whether these languages really non- configurational and whether they really do not have any order, but at least it looks like this). Warlperi is one of these languages. On other hand Turkish is not a free order language, its neutral order is SVO. It is a scrambling language. It may scramble objects and adjuncts to a higher position for topic, focus and etc. (there are some other scramblings as well) however, there are restriction on scrambling in Turkish. Not all possible word order combinations are grammatical in Turkish. That's why it is not free word order language, it is a scrambling language.

Answer 3

Thank you all for giving examples of word order in your mother language. As a typical example, the Persian sentence "من علی را دیدم دیروز"[I saw Ali yesterday] has exactly 4!=24 grammatical permutations with different frequencies and pragmatic meanings though. To have an interesting cross-lingual comparison:

• - Please add translations in your language (Czech,German,Turkish,Russian,etc).

• - How many grammatical permutations are possible in your own language?

Grammatical English reorderings are marked by '+':

1. من علی را دیدم دیروز /I Ali saw yesterday/
2. من علی را دیروز دیدم /I Ali yesterday saw/
3. من دیروز علی را دیدم /I yesterday Ali saw/
4. من دیروز دیدم علی را /I yesterday saw Ali/ +
5. من دیدم علی را دیروز /I saw Ali yesterday/ +
6. من دیدم دیروز علی را /I saw yesterday Ali/ +
7. دیدم علی را دیروز من /Saw Ali yesterday I/
8. دیدم علی را من دیروز /Saw Ali I yesterday/
9. دیدم من علی را دیروز /Saw I Ali yesterday/
10. دیدم من دیروز علی را /Saw I yesterday Ali/
11. دیدم دیروز من علی را /Saw yesterday I Ali/
12. دیدم دیروز علی را من /Saw yesterday Ali I/
13. علی را من دیروز دیدم /Ali I yesterday saw/ + {TOPICALIZATION}
14. علی را من دیدم دیروز /Ali I saw yesterday/ + {TOPICALIZATION}
15. علی را دیروز من دیدم /Ali yesterday I saw/ + {TOPICALIZATION}
16. علی را دیروز دیدم من /Ali yesterday saw I/
17. علی را دیدم من دیروز /Ali saw I yesterday/
18. علی را دیدم دیروز من /Ali saw yesterday I/
19. دیروز من علی را دیدم /Yesterday I Ali saw/
20. دیروز من دیدم علی را /Yesterday I saw Ali/ +
21. دیروز علی را من دیدم /Yesterday Ali I saw/
22. دیروز علی را دیدم من /Yesterday Ali saw I/
23. دیروز دیدم من علی را /Yesterday saw I Ali/
24. دیروز دیدم علی را من /Yesterday saw Ali I/

In Russian all the above variants are totally grammatical:

1. من علی را دیدم دیروز /Я Али видел вчера/ +
2. من علی را دیروز دیدم /Я Али вчера видел/ +
3. من دیروز علی را دیدم /Я вчера Али видел/ +
4. من دیروز دیدم علی را /Я вчера видел Али/ + {DEFAULT}
5. من دیدم علی را دیروز /Я видал Али вчера/ +
6. من دیدم دیروز علی را /Я видел вчера Али/ +
7. دیدم علی را دیروز من /Видел Али вчера я/ +
8. دیدم علی را من دیروز /Видел Али я вчера/ +
9. دیدم من علی را دیروز /Видел я Али вчера/ +
10. دیدم من دیروز علی را /Видел вчера Али я/ +
11. دیدم دیروز من علی را /Видел вчера я Али/ +
12. دیدم دیروز علی را من /Видел вчера Али я/ +
13. علی را من دیروز دیدم /Али я вчера видел/ +
14. علی را من دیدم دیروز /Али я видел вчера/ +
15. علی را دیروز من دیدم /Али вчера я видел/ +
16. علی را دیروز دیدم من /Али вчера видел я/ +
17. علی را دیدم من دیروز /Али видел я вчера/ +
18. علی را دیدم دیروز من /Али видел вчера я/ +
19. دیروز من علی را دیدم /Вчера я Али видел/ +
20. دیروز من دیدم علی را /Вчера я видел Али/ +
21. دیروز علی را من دیدم /Вчера Али я видел/ +
22. دیروز علی را دیدم من /Вчера Али видел я/ +
23. دیروز دیدم من علی را /Вчера видел я Али/ +
24. دیروز دیدم علی را من /Вчера видел Али я/ +

Answer 4

Yes there is recent research that provides a "quantitative, corpus-based, or information theoretic definition of word order". See for example Levshina 2021: 10.3389/fpsyg.2021.648200