In the English, we say:

Red apple

Red is an adjective. apple is a noun. Red tells us that, well, the apple is red.

In other languages, such as Arabic, it is the other way around. I.e.:

تفاحة حمراء

The word that looks like حمراء means red. The word that looks like تفاحة means (as you guessed it) apple. Now, you may say that isn't it the same order? خمراء (red) is appearing first before تفاحة (apple). The answer is no, because in Arabic you read from right to left (feel free to laugh for a few seconds, then come back when done).

Now, my question is: under which scenarios is which approach better? E.g.

  • Why did the English-speaking people evolve in such a way to put the adjective before the noun? Were the English-speaking people initially righting from right to left, but one day woke up and read it from left-to-right, and thought that the adjective is the 1st?
  • Or, the other way around, why did the Arabic-speaking people evolve to put the adjective second? Were the Arab-speaking people initially writing from left to right, but one day woke up and read it from right-to-left instead and thought that the adjective is the 1st?
  • Can English and Arabic swap the order of adjectives in this example without changing the meaning? Hence achieving the best of two worlds?

What I've Done Thus Far (plus some side questions)

Information theoretically, I'd argue that the goal of language is to deliver information with highest bit rate per second (this also implies error detection and correction). So I guess it may relate to another question:

  • Which approach allows for the transfer of a higher amount of information bits per second?

To answer that question, I'd argue that it depends on the probability model the reader has about the current state of the universe. E.g.

  • If the person thinks that most likely we are talking about apples, but he is unsure about its colour, then perhaps we better tell him first the adjective (red) then that it is an apple (because he already sort of knows with high probability that we are about an apple).

    In this case the person in English says "red apple" and he achieves highest bit rate. While the person in Arabic says "تفاحة حمراء" and gets lower bit rate.

    • Side question: is it possible to swap the order of adjectives in Arabic without changing the meaning here?

    Of course, if the person is sure that we are talking about apples, and he is unsure about the colour, then we will just say "give me the red one", and drop the "apple" altogether. But in this case the person is not sure about it being an apple (he just thinks that most likely it is an apple, but not certainly, hence telling him hat it's an apple is still informative).

    • Side question: can we say in English "give me the red", without "one", in case the person is sure it is all about apples? In Arabic we can "أعطني الحمراء", where word that looks like "أعطني" is "give me".
  • If the person thinks that most likely we are talking about red things, but he is unsure about which red one (e.g. red car? red pen? red apple?), then we better say the objective first (i.e. the apple), and then its colour (because he already sort of knows that we are talking about red things).

    In this case, the common Arabic approach of saying "تفاحة حمراء" works optimally (reminder: don't forget to read from right to left), but the common English approach of saying "red apple" doesn't.

    • Side question: Can English swap the order of adjectives without changing the meaning?

    Similarly, if the person is 100% sure that it's about red stuff, but unsure which red one, he can just say "give me the apple". Arabic can do this too "أعطني التفاحة". But in this case the person is not 100% sure that it's about red stuff, only maybe 90% sure, so he still needs to mention that it is red, hence the order.

This takes me to these thoughts:

  • I think the best language is one that allows swapping order of adjectives around depending on the probability model of the universe that the reader is expected to have. This way we can pick the optimum order per context. Can English do this? Can Arabic do this? I don't know.
  • But if the language does not allow such flexibility of moving the adjective around, without changing the meaning, then I think it mentioning the noun first and the adjective 2nd is usually more optimum, because I think usually we are more interested in what the thing is (i.e. noun) as opposed to its adjective (e.g. colour). So I think putting the adjective 2nd would maximize the average bit rate.
  • 1
    I’d say that they are both absolutely equal information-theoretically. However, you might find the discussion in wals.info/chapter/97#3._Theoretical_issues interesting.
    – bradrn
    Commented May 3, 2021 at 14:18
  • 1
    Your question ignores the recursive structural aspects of grammar (pace Daniel Everett): noun phrases are just one kind of structure in a language, and adjectives just one kind of modifier in a noun phrase. A given language tends to show a preference for either head-initial or head-final structures throughout its grammar.
    – Colin Fine
    Commented May 3, 2021 at 14:29
  • 2
    No @caveman. What I'm saying is that your question is a bit like asking about whether it's theoretically better for the + operator to be prefix, postfix or infix, while ignoring that it is embedded in a larger syntax.
    – Colin Fine
    Commented May 3, 2021 at 16:14
  • 1
    What about prosody and "highest bit rate per second"? In some natural languages, e.g. in Russian, both [red apple] and [apple red] are possible, with a substantial difference (the so called rheme-theme distinction)
    – Alex B.
    Commented May 3, 2021 at 18:40
  • 2
    Maybe the part about writing direction is a joke, but writing is irrelevant here -- most languages have never been written down, and the rules for adjective-noun order in English and Arabic predate the invention of writing systems for those languages.
    – TKR
    Commented May 4, 2021 at 2:52

1 Answer 1


Which approach allows for the transfer of a higher amount of information bits per second?

This is, as it turns out, a question that can be answered experimentally: neither. Coupé, Oh, Dediu, and Pellegrino (2019) showed that the information rate (bits per second) of different languages is roughly consistent all across the world; if the speakers of a particular language encode more bits per syllable, they speak fewer syllables per second, and vice versa.

(The information rate isn't perfectly constant, but notably it varies significantly less than bits per syllable or syllables per second do, and most of the variation between languages is well within the uncertainty caused by people speaking at different rates.)

Given this consistency, it's likely that there's some biological or psychological reason behind it. In other words, this value is the amount of information that humans can reliably absorb per second, and (linguistic) evolutionary pressure causes speech rate to change to match it.

Since we see plenty of head-initial and head-final languages all across the world, it seems clear that neither provides a significant advantage in this domain. Coupé et al 2019 and Oh's 2015 thesis also investigate other measures of linguistic information density (such as SDIR), coming to the conclusion that the biggest differences come from syllable structure and phoneme inventory, not syntax; even if it did, though, speech rate would be able to change to compensate.

  • Thanks, that answers my question. I have 2 comments. (1) Yes, it depends on biology/psychology, but I'd like to add that it should also depend on the distribution of concepts that need to be communicated. I.e. language encodes information describing some concepts that we see, and if time changes so that we talk about different concepts, then our languages will evolve to optimise for those concepts.
    – caveman
    Commented May 4, 2021 at 4:43
  • (2) I don't think that seeing similar numbers of head-init vs. head-final is a clear observation to support that their bit rate is similar. I rather think that it means that they are similar in terms of helping their speakers to survive against natural selection (not necessarily implying similar information bit rate), which leaves the room open for the possibility that maybe nature didn't pressure us strong enough to make us need to pick the more optimum one. For a similar reason we have many inefficient foods (e.g. cup cakes) simply because nature didn't force us to optimise better yet.
    – caveman
    Commented May 4, 2021 at 4:43
  • @caveman The trick is, the fact that we see a similar bit rate all across the world implies that linguistic evolution is constantly pushing language up against that limit. If there wasn't this sort of strong evolutionary pressure, we would expect to see a lot more variation. But it seems like, as information content changes for whatever reason (on the order of centuries), speech rate changes too to keep up.
    – Draconis
    Commented May 4, 2021 at 4:50
  • 1
    @caveman Or to put it differently, we've seen head-initial languages change to head-final and vice versa and everything in between. If there was an evolutionary benefit to one of those options, we'd expect to see that one retained, given that there's evidence of significant evolutionary pressure in this area.
    – Draconis
    Commented May 4, 2021 at 4:50
  • 1
    @caveman The key is, we see very different amounts of bits conveyed per syllable (and likewise per word) across languages—and yet other factors always change to compensate for this. To me, this indicates that there is significant evolutionary pressure pushing for a higher bit rate (up to the limit imposed by biology/psychology), and that this pressure hasn't had any effect on syntactic order.
    – Draconis
    Commented May 4, 2021 at 15:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.