A general feature exhibited by human languages has now been alighted upon

Earth is a veritable Tower of Babel: Up to 7000 languages are still spoken across the globe, belonging to roughly 150 language families. And they vary widely in the way they put sentences together. For example, the three major building blocks of a sentence, subject (S), verb (V), and object (O), can come in three different orders. English and French are SVO languages, whereas German and Japanese are SOV languages; a much smaller number, such as Arabic and Hebrew, use the VSO order. (No well-documented languages start sentences or clauses with the object, although some linguists have jokingly suggested that Klingon might do so.)

Yet despite these different ways of structuring sentences, previous studies of a limited number of languages have shown that they tend to limit the distance between words that depend on each other for their meaning. Such “dependency” is key if sentences are to make sense.

...

The results, published online today in the Proceedings of the National Academy of Sciences, demonstrate that all 37 languages [surveyed], including German, minimize dependency lengths to degrees greater than expected by chance.

Read more via the above link.

From the article:

Nevertheless, he stops short of concluding that it is a “universal” or “hard-wired” feature of language, rather than a strategy that humans have developed over time to make themselves better understood.

So what is the difference?

Limiting dependency length is advantageous, Futrell says, because convoluted sentences require more memory processing—and thus more energy—for both listeners and speakers who are trying to understand and be understood. Thus it makes sense that short dependency lengths became a universal feature in human language. “As language users, we have a choice of many ways of expressing ourselves,” Futrell says. “What languages don’t do is force you” into inefficient and energy-wasting use of memory stores.

What is "hard-wired" however is physics. "Universal" is .. physics also.

tuco wrote:From the article:

Nevertheless, he stops short of concluding that it is a “universal” or “hard-wired” feature of language, rather than a strategy that humans have developed over time to make themselves better understood.

So what is the difference?

This may have nothing to do with what he means, but I would guess that:

- it may not be universal feature because, because in other potential contexts languages may be more useful if you had a different order. Research into gibbons suggests they have a fairly complex vocabulary that tends to focus on the objects - words for predators, for example. While gibbons might not be said to have a real language, depending how strictly you define that, it's plausible to imagine that a language evolving under a context like that may put higher value on different grammatical structures.

- it may not be hard-wired, mostly because it doesn't need to be hard-wired to arise. Although some studies suggest there are hard-wired grammatical categories - and this may well be a universal constraint as per your remark - the ordering of these categories is not fixed. One extremely important element of language as observed in humanity is how plastic and adaptable it is. I remember some studies on deaf Brazilian street kids who had developed their own sign language, complete with novel grammar, syntax, and idioms. All the hallmarks of a typical human language, but arising and developing spontaneously. The proclivity for language may be hard-wired while its composition may be undetermined.

tuco wrote:

Limiting dependency length is advantageous, Futrell says, because convoluted sentences require more memory processing—and thus more energy—for both listeners and speakers who are trying to understand and be understood. Thus it makes sense that short dependency lengths became a universal feature in human language. “As language users, we have a choice of many ways of expressing ourselves,” Futrell says. “What languages don’t do is force you” into inefficient and energy-wasting use of memory stores.

What is "hard-wired" however is physics. "Universal" is .. physics also.

I'm not really clear what the strict line from physics to language is... but I am also not particularly convinced of the above notion by Futrell. Firstly, I really don't think that this is adaptive - the quantities of energy used between 2 different language compositions are hardly going to present selection pressures, and language is in far greater flux than genes. Secondly, what we consider to be convoluted sentences is an artifact of the languages we are familiar with. Presumably, a creature with a complex language using obj-sub-prep-vb would find our grammatical structures equally taxing.

I'm with Spearthrower on this.

The first thing I noticed in the article was that:
" For most of the languages, the researchers used written prose from newspapers, novels, and blogs, although for ancient Greek and Latin they relied on poetry."

Written language is a very different thing from spoken language. Traditionally linguists have dismissed the spoken form of the language as 'degenerate' and 'corrupt' (These are Chomsky's words, from 1965) As such it was not deemed a worthwhile thing to study.
Fortunately, with the advent of reliable, portable, unobtrusive audio and video recording equipment, and good computers to process millions upon millions of words of transcribed spoken text in corpora we are now on happier ground. The fact is that the written form of the language is a narrow and brittle form of language, and the insights gained from studying the written form of the language are by no means always applicable to the spoken form, which is the basic form of human language.

So, in particular, what can be said about dependency length minimization?
Well, first off, the basic unit of language in speaking is not the sentence, so beloved of the generative grammarians, but the turn. And it turns out that turns at talk are usually quite short. Now one reason why turns are quite short might be because we don't want to place too much strain on the cognitive processes of our interlocutor, but the main reason is that we are are observing interactional, not syntactic rules. That is, we limit the amount of floorholding we do during interaction, so that it achieves a balance of speakership. (Research has shown that speakers do a remarkable job of aligning with each other to produce turns of similar length.)
Short turns which contain reference to the previous turn before proceeding with small incremental additions to the ongoing discourse are the norm in spoken interaction. Huge long multiple sub clause sentences with layer on layer of relativization are not the norm. For social rather than cognitive reasons.

The participants in spoken interaction co-create meaning by complex processes of repair (self and other initiated) backchanneling, discourse marking, repetition, reformulations, restarts and so on. The cognitive load of maintaining interaction (for example, achieving precision timing at turn transitions points, recognizing when other speakers are about to finish a turn, or signalling by linguistic and paralinguistic means that your turn is coming to an end, or not) is the main cognitive business of spoken interaction, not parsing sentences.

Also, in spoken interaction, people talk within context so they achieve their communicative goals by embedding their turns in the broad stream of ongoing, unfolding talk, not springing epistemic surprises on their unsuspecting listeners. In much of talk, the interlocutors already have a pretty good idea of what is going to come next, so it doesn't need that much cognitive load bearing to figure out the precise meanings of the particular grammar construction. People recognize schema and skim along, rather than struggle from scratch every turn.

Here's a complex sentence I came up with.

The people who were in the building that was hit by a plane which had been hijacked by terrorists who hate America all died.

I'm guessing that most people reading this will not have struggled too much with comprehension, despite the multiple relative clauses, due to familiarity with the content of the statement. ('Died' comes 21 words after 'people', but its no big deal once you get to 'hit by a plane'. Tell me if I'm wrong, of course.)

Lastly, corpus studies have revealed the ubiquity of chunking in spoken language. Fixed expressions such as 'You know' 'You know what I mean', 'At the end of the day', 'The thing is' are extremely common in spoken English. Because of their familiarity and fixedness, it has been proposed that the mind perceives them as 'chunks' that is, single items that only fill one 'slot' in the ongoing processing of utterances, thus allowing the brain to short cut when processing turns at talk. Multiple chunks would allow seemingly great distance between dependent words, that was not actually, cognitively that great. This data would be absent from studying only the written form of the language.

So, whilst it might be interesting to look at SVO and SOV languages and find differences in how they create meaning and then hypothesize about linguistic and cognitive universals, it all takes a back seat, in my mind at least, to the social universals that underpin all natural language use.

Oh, and to finish...
Mark Twain quipped that Schiller's history of the Thirty Year's War was entirely contained between the two parts of a German phrasal verb...

don't get me started wrote:I'm guessing that most people reading this will not have struggled too much with comprehension, despite the multiple relative clauses, due to familiarity with the content of the statement. ('Died' comes 21 words after 'people', but its no big deal once you get to 'hit by a plane'. Tell me if I'm wrong, of course.)

I am pretty sure that this would cause some horrendous difficulties in other languages!