Specializations > Typology and Descriptive Linguistics
"Original" Word Order?
Basic word order (e.g. SOV/SVO) in the world's languages has long fascinated scholars and laymen, and generated numerous scholarly articles.
As a layman, I'm attracted to Gell-Mann/Ruhlen's 2011  thesis of an "ancestral" SOV with a tendency to evolve into SVO, and disappointed their conclusions offer no suggestion why that was or why this happens.
Maurits et al 2019?  make some interesting information-theoretic arguments from English- and Japanese-language experiments that Uniform Information Density (UID) could have a bearing on this question.
My intuition, for what it's worth, is that the latter may be on the right track in regard to evolution but why does that fail to explain an original SOV (if true)?
As we are forced back to the origin of language itself, clues might be found in non-verbal communication.
My limited experience with a serious language barrier, and recourse to inexpert signing, suggests that a "natural" order starts with a concrete topic (e.g. subject) and proceeds to the more abstract action. In case an object is also important, being concrete, that comes second.
The signing order is gesture/picture/symbol, gesture/picture/symbol, then mime.
Just asking ;D
These are interesting questions, but possibly unanswerable.
Regarding the first topic in general, SOV is the most common order around the world, followed by SVO, then others. The idea that SOV was "original" and others evolved from it doesn't make much sense because SOV is still so common. There have been at least tens of thousands of years for languages to shift, so why aren't all languages SVO or something else by now? This is somewhat similar to why ideas of languages "becoming more complex" or "simplifying" over time don't really work, because the history of languages is too long for it to just operate in one direction. It is interesting that SOV>SVO seems like a typical change (although my intuition may be biased by well-known cases, especially in Europe). But it might be more helpful to look at examples of other changes. My guess is that the change might not be SVO>SVO directly, but through some intermediate steps: consider Arabic for example, with an ancestor that was probably SOV (if Akkadian is representative) then either VSO or topic-prominent with apparent SVO order. This is probably related to the fact that some languages have optional subjects (the verb, often agreeing with the subject, can appear alone). So SOV>OV/VO>Topic-VO>SVO. You could also look at how other word orders appear (the less common VOS, OSV, OVS types) and see if that can give a clue about how word orders change beyond SOV>SVO. More generally, there are a number of other statistical correlations between word order and other typological properties. SOV languages tend to be heavily suffixing, have postpositions or case markers (instead of prepositions), and so forth. There are also some other features less often discussed but still statistically correlated: one topic I've included in my research is the relationship between clause-chaining with converbs (i.e. adverbial dependent verbs, roughly like English "-ing" as in "waking up, eating breakfast, driving to work, he started the day") is extremely common in SOV languages but rare elsewhere, so the grammatical organization of discourse may be different as well. Finally, depending on your syntactic theory of choice, but regardless of which one, there are actually multiple ways to analyze surface "SVO" (etc.) order, so that could be another relevant direction to consider, because two languages that are both SVO on the surface may have very different properties, as well as whether there is any flexibility to that ordering.
Regarding the second part, there are a lot of ideas out there about functional explanations for word order. Most importantly, remember that all languages function just fine, and all word orders are attested (although the OVS and OSV orders are extremely rare and maybe not robustly attested or the only possible order in any language). So SOV and SVO (and others) cannot be "explained" through abstract functional (or other) motivation. There's no need for one to switch to the other, either can work, and does work in various languages around the world. One interesting approach to this is to try to explain not why those are basic orders, but why there are then typological correlations with the orders. Why postpositions and SOV but prepositions and SVO? That has been called "head-ordering" in general (where verb-final languages also have postpositions at the end of the noun phrase, for example), and it tends to be mostly consistent in a language. One proposed explanation for this is called dependency length minimization (see work by Hawkins and more recently many others), whereby in SVO languages most of the important syntactic relations are clarified at the beginning of a sentence, while in SOV languages they're mostly clarified at the end, but either way they're clustered together rather than randomly dispersed throughout a sentence. This is thought to give some consistency and efficiency to the parser, so either order works well, but a mix in the middle would mean more potential ambiguity in a sentence and less communicative efficiency/effectiveness. But as you say, these ideas don't clearly translate into telling us about an "original" order.
Regarding signed languages, there actually is some research about this, and interestingly SOV seems to be very common in signed languages (but very often word order is quite flexible: for example, it is somewhat controversial whether "SOV" is an accurate description of ASL, although that's a common description), probably even more common than in oral languages (and yes, they tend to be relatively "young" languages). I think some studies of homesign (that is, makeshift sign systems (but not full languages) used in a household to enable basic communication where there is no other established signed language) may have also suggested SOV is common (but I'm forgetting where I read that or if I'm accurately representing it, so you should confirm my comment here).
In the end, looking back past about 10,000 years is generally unreliable in historical and comparative linguistics. That figure is thought of as something of a "wall" (we can't cross) by some researchers. And while I'm optimistic that we may be able to sometimes think about things a bit earlier than that, it would make sense that passing 20,000 would be very unlikely, and then extremely unlike we could ever understand something about "original" languages. In fact, I think every explanation that attempts to explain that goes against the Uniformitarianism hypothesis [ https://en.wikipedia.org/wiki/Uniformitarianism#Social_sciences and search online for various other discussions about Linguistics specifically]. That is, some researchers claim to have understood something about the "original" language(s) by finding some property or distribution that has changed since that time, but many linguists would disagree with that premise.
 Message IndexGo to full version