Monday, August 7, 2017

Conlanger Lore: Free Word Order and Case

There is a very common notion bandied about on conlanging groups online that free word order and case go hand in hand. This leaves out a significant chunk of the truth. I did mention this as an example in the previous conlanger lore instalment, but this goes deeper into this particular issue, and looks at some of the things people do not often know.

1. There are languages with case that do not have particularly free word order. German is one of them.

2. There are languages without case that have rather free word orders. For instance, Swedish has more free word order than German despite lacking case on nouns.

More extreme instances exist, for instance among the overwhelmingly isolating languages of south-east Asia, but I picked Swedish and German as they are fairly familiar, SAE languages where the freedom of word order and presence of case contradicts the 'received wisdom' in the conlanging community.

However, there are a few particular reasons why this particular piece of wisdom annoys me: it ignores the wealth of variation there is among human languages. Several other strategies of disambiguation exist!
A. Noun Class Congruence
A strategy that is common in Sub-Saharan Africa is having a bunch of noun classes, and congruence with those on the verb:
Nounclass 1 Nounclass 2 Verb(subject congruence=class 1)(object congruence=class 2)
With such a system, it is clear how shuffling the location of the nouns and the verb do not affect understandability, except when the nouns are of the same class. Such things do occur, but will be discussed further down simultaneously with similar ambiguities in other languages.

B. Animacy Hierarchies
In a language where subject- and object-disambiguation is guided by animacy hierarchies, a noun that is higher on the animacy hierarchy is assumed to act on a noun lower in that hierarchy. This in part, I am convinced, guides the disambiguation when OVS word order occurs in Swedish. Usually, the hierarchy in most languages is something like 1 p > 2 p > 3p animate > 3p inanimate, but quirks exist: apparently, some languages have second person outrank first person.

Another important fact is that it's possible for verbs and nouns to somehow be associated: bears are, for instance, more closely associated with the subject position of 'roar' than bags are. These are not even necessarily lexical facts but rather physical facts we know about things. So, if we take a sentence like this key opens that lock, even if we cut it up into three slips of paper on which the phrases 'this key', 'that lock', 'opens' are, a random anglophone can with several nines of probability assign the nouns to their correct roles.

This seems to be a grammaticalized state of affairs in many languages - to the extent even that if the subject and object are the unexpected way around, even with explicit case marking you'll sometimes get people parsing it as though the speaker made a mistake in case marking rather than parsing it as referring to an unlikely situation.

Even in languages with case marking, situations where case does not help disambiguate the situation may exist - and sometimes in such languages, this does not negate free word order. In Finnish, the plural object and the plural subject take the same marker, -t, in telic, positive, (etc) clauses. Even then, I regularly hear and utter sentences where the subject and object are displaced from the canonical SVO order.
C. Animacy Hierarchies with Inverse/Direct Alignment
The previous system does not really permit for changing the subject and object except maybe with some complicated work-arounds. Inverse/Direct alignment simplifies this by having a 'voice-like' marker that simply tells whether the higher or lower noun in the hierarchy is the subject. This is not strictly speaking a voice, since it does not affect the transitivity of the verb.
The other part we have to look at is free word order. Sometimes when some speaker of some language points out that their language has absolutely free word order, they will, in one post deny the idea that their language's free word order expresses anything, while maintaining that it does express something that just can't be expressed in words in another post, etc. Generally speaking, word order tends to have some kind of significance, though this significance can be pretty difficult to express and maybe even harder learn to get right for a second-language speaker. However, difficulty of formalizing/expressing a thing is not the same as that thing not existing. Thus, it seems people are quite confused as to what free word order generally even signifies.

Conclusion: Don't buy the hype re: case enabling free word order, it's not the only option. Also, don't just go and say 'this language has free word order', plz tell us what the language does with its free word order.


  1. Just wanted to raise two ideas about the question of how to deal with animacy hierarchies and 'inverse' verbs.

    One way, as you say, is explicit direct/inverse marking on the verb.

    An alternative, however - which I've not used yet but would like to at some point - is to use word order itself as a direct/inverse marker. The easiest way to do this is to allow free word order (of whatever kind) only for 'direct' verbs, reverting to a fixed word order for inverse verbs. So, if a man eats a cow, you can say "man cow eat", "cow man eat", "man eat cow", etc - but if a cow eats a man, you have to say "cow man eat". Now, this makes "cow man eat" ambiguous (but all other word orders unambiguous) - but in situations where the either reading is possible, it can be expected that the direct meaning would avoid that specific order. So that although the word order isn't 'fixed', it is, as it were, 'biased'.

    Such a system (or another) may be reinforced by some other way of marking "whoah!" verbs - mirativity, for instance, can call attention to potentially surprising relations, without that being its sole or primary function.

    [note to self: do this!]

    Secondly, there can be a fixed subject/object allocation depending on the animacy hierarchy, in which case alternative methods are required to circumvent these rules.
    My Rawang Ata, for instance, restricts inverse verbs (well, for non-statives, at least), and forces detransitivisation. In Rawang Ata, that needn't mean valency reduction, but it does alter case marking. Other languages may go further and force full valency reduction. So, "the man eats the cow", but "the cow feasts upon the man" (well, if we treat 'feast' as an intransitive, rather than 'feast upon' as a phrasal verb, but you get the point - an intransitive with an oblique rather than a transitive). Or even passive "the man is eaten by the cow".

    Alternatively, animacy assignments may be altered - Rawang Ata can make use of vocatives to artificially raise animacy to enable inverse verbs - so, "the man eats the cow", but "O cow (you) eat the man" (vocatives acting like second person pronouns). Other languages may use replacement by pronouns ("The cow, she eats the man"), or may temporarily grant animacy through metaphorical appositives ("The woman, a cow, eats the man") or the like.

    Are there perhaps also languages in which case marking only occurs with 'inverse' verbs? It feels like there should be, but I don't know (and languages with case do seem to get quite fond of it even when it isn't needed).

    1. The question you ask in the last paragraph can probably be answered by 'nope', but this is due to the fact that the areas where morphological case is common and the areas where inverse alignment systems are common don't really overlap. I only know of one language that has both morphological case and an inverse system, and that's Chukchi.

      Caveats, though: it's conceivable there's more languages with inverse alignment and case, but that inverse alignment was so unknown among an earlier generation of field linguists that no one spotted that that's what was going on in some language?