Are some languages more "complex" than others?

[Jokodo]

I did find the multi-author book Language Complexity as an Evolving Variable on-line. It has 19 chapters, including "An interview with Dan Everett" regarding his controversial claim mentioned upthread. (This chapter is also available at daneverettbooks.com.)

Chapter 8 "Linguistic complexity: a comprehensive definition and survey" adopts a specific definition of grammatical complexity, and pursues this quantitatively. She concludes that complexity DOES vary, and ranks 68 languages: Basque is lowest at 13.0, Mandarin at 14.6, German at 16.0, Thai at 16.4, Russian at 17.2, Ingush highest at 27.9. I am suspicious of her strict arithmetic approach.

You are what??? The claim that language A has more complexity than language B is intrinsically a quantitative claim. It can either be demonstrated to be true or false with a strictly arithmetic approach, or not at all. Those are the only two possibilities. Pick one!

If you toss out arithmetic, you're leaping into the emptiness of outer space.


(What gave her away though? Was it the fact that your oh-so-simple Thai lands midway between German and Russian made you suspicious?)

You've got nothing but banal insults for me. Please just set me to Ignore, or I'll do the vice versa.

Thanks in advance.

I am unable to find a single insult, banal or otherwise, though i am sorry if you feel that way. I've pointed out holes in your arguments. If you don't like that, make better arguments!

 

[Jokodo]

@Swammerdami you started the thread asking, among others, by saying "We have 2 or 3 linguists here and I hope they can show me what I'm missing."

I came here, as a linguist, to point out some things I feel you're missing with the general theme being that you are missing the complexity involved in merely agreeing on what the question might even mean, let alone answering it.

If you feel "you're missing X" to be an insult, maybe don't ask "what am I missing?"

 

[Swammerdami]

You are what??? The claim that language A has more complexity than language B is intrinsically a quantitative claim. It can either be demonstrated to be true or false with a strictly arithmetic approach, or not at all. Those are the only two possibilities. Pick one!

Even if "language complexity" had a clear or agreed-upon definition -- it doesn't -- multidimensionality adds difficulties. Often there is NO clear way to weight different criteria. What will you do? Use a Mahalanobis distance?

Even assigning a count to a discrete set, as Nichols does in that chapter, is not easy. The incidence partitions {.25, .25, .25, .25} and {.33, .33, .33, .01} both have population 4, but the latter may behave more like a Threeness.

If you toss out arithmetic, you're leaping into the emptiness of outer space.

If you think arithmetic approaches are practical for all topics like this, you are ... mistaken. In another thread I criticize Richard Carrier who touts his absurd and wrong approach to calculate a specific probability that a Jesus of Nazareth was crucified by Pontius Pilate.

(What gave her away though? Was it the fact that your oh-so-simple Thai lands midway between German and Russian made you suspicious?)

This comment was false, banal, and ... an INSULT against me.

Hope this helps.

 

[Swammerdami]

@Swammerdami you started the thread asking, among others, by saying "We have 2 or 3 linguists here and I hope they can show me what I'm missing."

I came here, as a linguist, to point out some things I feel you're missing with the general theme being that you are missing the complexity involved in merely agreeing on what the question might even mean, let alone answering it.

Are you contradicting yourself?? A few minutes ago you insist that simple arithmetic was the path forward, and insulted me for suggesting otherwise. Make up your mind!

 

[Swammerdami]

I'm sure Thai has means to make both of these questions unambiguous with particles, adverbs, or some slight rearrangement. The ambiguity persisted not because Thai doesn't have the means to make it go away, but because the speaker didn't anticipate that it would be necessary, based on making unwarranted assumptions about the hearer's context frame.

Of course. Often the easiest way to make the ambiguity go away is with an extra question-answer. Some conversations end up requiring more TURNS than the English equivalent, but Thai's tersity leads to greater efficiency when the ambiguity is resolved by context and the Thai information exchange accomplished in fewer words or fewer seconds than English would use.

But "complexity" is not "efficiency" or "inefficiency." I thought the following relevant quote from my earlier post was interesting:

David Gil, "How Much Grammar Does It Take to Sail a Boat?" (pp. 19-33), argues that a minimal language (no morphology, no stem-class oppositions, semantic composition by "association") is generally adequate for human communication, and that some languages such as (semidiglossic colloquial) Riau Indonesian attain such minimality in conversational stretches and approximate it elsewhere. Other languages are much more complex (e.g., have cascading syntactic hierarchies), but such complexity is "hugely dysfunctional" insofar as such a language "forces us to say things that we don't want to say" (p. 32). The implication of the papers by Nichols, Dahl, and Gil is that languages can differ greatly in overall overt morphosyntactic complexity.

In many languages of the Middle East and to some extent Southeastern Europe, there are no neutral words for "aunt" or "uncle",

There are eight possible uncle/aunt relationships:

{older,younger} {male,female} sibling of {mother,father}.​

Thai makes do with just four words.
When the uncle/aunt is an OLDER sibling, Thais tell us his/her sex. But when younger, Thais tell us the PARENT's sex.
(Strictly speaking there are 16 possible uncle/aunt relationships since the uncle/aunt's spouse is also denoted. But Thai and English both treat this without additional words.)

Instead of "brother or sister" Thai speaks of "older sibling or younger sibling". Many nicknames are used by both men and women, and more than once, my wife relayed a conversation; but when I asked "was Noi a man or woman", she didn't know!

However, I think all this has little or nothing to do with "language complexity."

 

[Jokodo]

You are what??? The claim that language A has more complexity than language B is intrinsically a quantitative claim. It can either be demonstrated to be true or false with a strictly arithmetic approach, or not at all. Those are the only two possibilities. Pick one!

Even if "language complexity" had a clear or agreed-upon definition -- it doesn't -- multidimensionality adds difficulties. Often there is NO clear way to weight different criteria. What will you do? Use a Mahalanobis distance?

Hey, I'm not the one claiming that ranking entire languages by and global measure of complexity, you are. Your claim that English is more complex than Thai *is* another phrasing for the claim that the combined weight of the areas in which English is more complex is such that they outweigh the areas in which Thai is more complex. Of which you yourself have listed plenty, protestations of the "is that really complexity?"-kind notwithstanding.

That it's hard is a problem for you as much as it is for her. That she recognised that it is, and is at least trying, is a point in her favour.

Even assigning a count to a discrete set, as Nichols does in that chapter, is not easy. The incidence partitions {.25, .25, .25, .25} and {.33, .33, .33, .01} both have population 4, but the latter may behave more like a Threeness.

If you toss out arithmetic, you're leaping into the emptiness of outer space.

If you think arithmetic approaches are practical for all topics like this, you are ... mistaken.

There are a lot of questions that don’t require arithmetics to answer them. Many qualitative questions qualify. The fact that Thai has no case agreement, English has residual case agreement in pronouns only, German has fragmentary case on nouns and more systematic case on articles/adjectives/pronouns are not established by arithmetic formulae because the underlying question is qualitative.

The question which of two languages has a higher global complexity is a quantitative question that can only be answered with arithmetic. Asking for a non- arithmetic answer is like asking "which contains more water, the Black Sea or Baltic Sea? Please answer without referencing their surface area or average depth!"

It's confusing to say the least.

In another thread I criticize Richard Carrier who touts his absurd and wrong approach to calculate a specific probability that a Jesus of Nazareth was crucified by Pontius Pilate.

(What gave her away though? Was it the fact that your oh-so-simple Thai lands midway between German and Russian made you suspicious?)

This comment was false, banal, and ... an INSULT against me.

Hope this helps.

Wha? I was classifying that as light banter. Maybe it's a generation thing, maybe you have been contaminated by Thai ideals of Face and Politeness, or maybe English is insufficiently complex to distinguish these two very different speech acts linguistically and the whole situation could have been avoided in a language with a developed particle system, like Thai?

 

[Jokodo]

@Swammerdami you started the thread asking, among others, by saying "We have 2 or 3 linguists here and I hope they can show me what I'm missing."

I came here, as a linguist, to point out some things I feel you're missing with the general theme being that you are missing the complexity involved in merely agreeing on what the question might even mean, let alone answering it.

Are you contradicting yourself?? A few minutes ago you insist that simple arithmetic was the path forward, and insulted me for suggesting otherwise. Make up your mind!

I never said "simple". I said "strictly arithmetic". Of course, for any metric to be more than a sandbox demo, we need to define weights and argue why they are reasonable and get an estimate of how susceptible the results are to change from twisting any of our input parameters and many other things. Once all that is said and done, if and when we agree on a metric, the question of whether English or Thai, or Ingush or Basque, scores higher on global complexity by the metric we just agreed to is going to be answered by applying the formula to data about the language, or it will remain unanswered.

Nichols is aware of these difficulties. She isn't claiming to have shown that Ingush really is more complex than Basque. She is making a much more limited claim: that within her arbitrary and arbitrarily weighted sample of features no pairs or families of features emerge as being anti-correlated in their complexity. This could mean they're is no complexity trade off in a narrow sense, or that the features that tend to grow more complex to compensate for simplicity in the areas she included tend to be ones in areas she didn't include, maybe for systematic reasons because the compensation often happens at the grammar-pragmatics interface, an area that is often poorly documented in particular on overall less studied languages, and hers was admittedly in part a convenience sample, that is she picked features because she found data for them for a large range of languages, and tells us about doing so!

 

[Jokodo]

I'm sure Thai has means to make both of these questions unambiguous with particles, adverbs, or some slight rearrangement. The ambiguity persisted not because Thai doesn't have the means to make it go away, but because the speaker didn't anticipate that it would be necessary, based on making unwarranted assumptions about the hearer's context frame.

Of course. Often the easiest way to make the ambiguity go away is with an extra question-answer. Some conversations end up requiring more TURNS than the English equivalent, but Thai's tersity leads to greater efficiency when the ambiguity is resolved by context and the Thai information exchange accomplished in fewer words or fewer seconds than English would use.

But "complexity" is not "efficiency" or "inefficiency." I thought the following relevant quote from my earlier post was interesting:

David Gil, "How Much Grammar Does It Take to Sail a Boat?" (pp. 19-33), argues that a minimal language (no morphology, no stem-class oppositions, semantic composition by "association") is generally adequate for human communication, and that some languages such as (semidiglossic colloquial) Riau Indonesian attain such minimality in conversational stretches and approximate it elsewhere. Other languages are much more complex (e.g., have cascading syntactic hierarchies), but such complexity is "hugely dysfunctional" insofar as such a language "forces us to say things that we don't want to say" (p. 32). The implication of the papers by Nichols, Dahl, and Gil is that languages can differ greatly in overall overt morphosyntactic complexity.

In many languages of the Middle East and to some extent Southeastern Europe, there are no neutral words for "aunt" or "uncle",

There are eight possible uncle/aunt relationships:

{older,younger} {male,female} sibling of {mother,father}.​

Thai makes do with just four words.
When the uncle/aunt is an OLDER sibling, Thais tell us his/her sex. But when younger, Thais tell us the PARENT's sex.
(Strictly speaking there are 16 possible uncle/aunt relationships since the uncle/aunt's spouse is also denoted. But Thai and English both treat this without additional words.)

Instead of "brother or sister" Thai speaks of "older sibling or younger sibling". Many nicknames are used by both men and women, and more than once, my wife relayed a conversation; but when I asked "was Noi a man or woman", she didn't know!

However, I think all this has little or nothing to do with "language complexity."

Hey, I'm not the one who's been bringing up instances where I overheard misunderstandings that could have been prevented by another language's on always providing the providing that would have preempted them as anecdotal evidence that the language which allowed the misunderstanding to happen is less complex!

 

[Swammerdami]

The claim that Thai is a SIMPLE language shouldn't be taken as deprecation. Leonardo da Vinci, Newton and several top-notch 20th-century physicists all stressed the beauty and importance of SIMPLICITY in their models

What I find "simple" with the Thai language also gives it much elegance. I wonder if I'm correct that foreigners pick up conversational Thai MUCH faster than they learn many other languages. I LOVE the example sentence with 12 consecutive verbs. It's fascinating that they make do with one helping verb (/daai/) to serve multiple purposes, with differences in word order or intonation as cues. And, as quoted above, [some] complexity is "hugely dysfunctional" insofar as such a language "forces us to say things that we don't want to say."

(What gave her away though? Was it the fact that your oh-so-simple Thai lands midway between German and Russian made you suspicious?)

This comment was false, banal, and ... an INSULT against me.

Wha? I was classifying that as light banter. Maybe it's a generation thing, maybe you have been contaminated by Thai ideals of Face and Politeness, or maybe English is insufficiently complex to distinguish these two very different speech acts linguistically and the whole situation could have been avoided in a language with a developed particle system, like Thai?

Good points. I think we'd have been fine in a face-to-face conversation. Your expression would have helped me see that this was, indeed, just light banter.

I am proud that it's been years since I lost my temper in face-to-face encounters. It's too bad that written comments often infuriate me irrationally. Sometimes I go back to re-read an exchange that infuriated me and find that I was "turning molehills into mountains" or fantasising an insult that wasn't there at all. PLEASE accept an apology.

 

[Jokodo]

The claim that Thai is a SIMPLE language shouldn't be taken as deprecation.

I'm not taking it as depreciation. I'm taking it as a empirical claim that lacks empirical support and is indeed so vague that it is unclear what such support could even look like.

Repeating that you feel it's SIMPLE while presenting yet another way in which it's complex but which for some often undeclared reason you feel shouldn't count is not presenting support nor helping us hone in on a definition.

Leonardo da Vinci, Newton and several top-notch 20th-century physicists all stressed the beauty and importance of SIMPLICITY in their models

What I find "simple" with the Thai language also gives it much elegance. I wonder if I'm correct that foreigners pick up conversational Thai MUCH faster than they learn many other languages. I LOVE the example sentence with 12 consecutive verbs. It's fascinating that they make do with one helping verb (/daai/) to serve multiple purposes, with differences in word order or intonation as cues.

... and yet another way in which Thai is arguably more complex than English. In English, you have to memorise only one sentence template* to produce and interpret all of

"I do intenionally boil my eggs soft"

"I have intenionally boiled my eggs soft"

"I shall intenionally boil my eggs soft"
etc.

As the difference between them is encoded in the lexicon.

In Thai, according to what you just said, they are distinguished not by relatively straightforward differences in the lexical semantics of auxiliaries, but require separate templates or structural representations! How on Earth is that not (A) intuitively more complex overall, to the extent that comparing the two is straightforward enough that it doesn't require theoretical analyses of the processes involved and/or an assessment of the difficulties it presents to speakers through acquisition studies and the analysis of errors, ideally induced by distracting participants with o bogus task and having them think that's what's being tested, and (B) specifically complexity that resides in a more core aspect of grammar, rather than in the periphery of the lexicon?

By my count, you yourself have listed about twice as many ways in which Thai is complex than ways in which it is simple.

And, as quoted above, [some] complexity is "hugely dysfunctional" insofar as such a language "forces us to say things that we don't want to say."

(What gave her away though? Was it the fact that your oh-so-simple Thai lands midway between German and Russian made you suspicious?)

This comment was false, banal, and ... an INSULT against me.

Wha? I was classifying that as light banter. Maybe it's a generation thing, maybe you have been contaminated by Thai ideals of Face and Politeness, or maybe English is insufficiently complex to distinguish these two very different speech acts linguistically and the whole situation could have been avoided in a language with a developed particle system, like Thai?

Good points. I think we'd have been fine in a face-to-face conversation. Your expression would have helped me see that this was, indeed, just light banter.

I am proud that it's been years since I lost my temper in face-to-face encounters. It's too bad that written comments often infuriate me irrationally. Sometimes I go back to re-read an exchange that infuriated me and find that I was "turning molehills into mountains" or fantasising an insult that wasn't there at all. PLEASE accept an apology.

Apology accepted. And next time I'll follow the banter with a "", eh?


(* I speak of templates for simplicity; I do not wish to imply that templates are necessarily part of a psychologically realistic description of what goes on in speakers' brains, and indeed am more inclined to a procedural view analysing grammatical patterns as necessary outcomes of restricted processes. However, when comparing the complexity of a phenomenon across languages our preferred theoretical analysis of that phenomenon is a poor guide and we need to use the minimal model to capture it for a fair comparison. )

 

[Jokodo]

Incidentally going to Thailand very soon for a few weeks of vacation. I was there 5 years ago and enjoyed it. I only ever learnt a few phrases of common courtesies and don't remember one of them.

That must be because Thai is so hard! I remember "please" and "thanks" and "orange juice" in Greek and it's been 30 years since I was there!

I do want to learn to read an write this time around. I never tackled that last time, but how hard can it be when it's an alphabetical system. I can read Cyrillic, Arabic, Greek and (sort of - I'll eventually remember must letters) Hebrew and with the excepting of Cyrillic I have no more use for those than for Thai. I do remember noticing they have a wide variety of fonts, some of which almost fell like you're reading Latin after a stroke, ie it's hard to shed the feeling that you should be able to read them because the writing looks familiar until you focus on individual letters. I think in my head I had a nickname for a local soft drink based on what the logo could have read if it where Latin.

 

[Bomb#20]

if it where Latin.

Semper ubi sub ubi.

 

[lpetrich]

Or for an example closer to your Thai example: a Nurse might ask in English "where did you get your last shot?" And be surprised when the answer is "at my local health centre" when they expected something like "in the right shoulder".

That's a case of "where on your body" vs. "where were you at".

We normally use context to avoid being too verbose, but in this case, the context fails.

 

[Jokodo]

Or for an example closer to your Thai example: a Nurse might ask in English "where did you get your last shot?" And be surprised when the answer is "at my local health centre" when they expected something like "in the right shoulder".

That's a case of "where on your body" vs. "where were you at".

We normally use context to avoid being too verbose, but in this case, the context fails.

Yes, the nurse is making the assumption that the only relevant interpretation in context is the "where in your body" one, and the patient fails to understand its relevance and thus randomly picks between the two.

Of course some misunderstandings are harder to produce in some languages because the two meanings involved do not share a form even when you're being non-verbose and/or making unwarranted assumptions about the context.

 

[bilby]

Reminds me of one of my favourite jokes:

It's a quiet day at the casualty department of Kings College Hospital in South London. Then the door from the street bursts open, and an Essex Girl staggers in, covered from head to toe in blood.

The triage nurse rushes over. "Can you tell me where you're bleeding from?"

The patient replies, "Yeah, I'm from bleeding Romford, ain' I?"

 

[Jokodo]

Here's an example of "Latin with a stroke": with a bit of creativity, you can read "Aśaña", but the more you look, the wronger it gets. The "a"s are from the different fonts, and why does the "ñ" have a "m" as its base?

It actually reads "crystal"

 

[Swammerdami]

I didn't notice that this thread had been bumped!

Sentiment seems to be that language complexity is too ambiguous to quantize easily. (Of course this makes the "equal complexity" claim itself undecidable or even meaningless.)

A working definition might be based on how long it takes someone to become conversant in the language.

Incidentally going to Thailand very soon for a few weeks of vacation.

Did you go to Chiang Mai? I should have invited you up for a beer and some sight-seeing. January and (early) February are the best times to visit. Despite the high latitude and altitude, hot climate is a problem throughout much of the year. (And March is "smoky season" here. :-( )

I was there 5 years ago and enjoyed it. I only ever learnt a few phrases of common courtesies and don't remember one of them.

That must be because Thai is so hard! I remember "please" and "thanks" and "orange juice" in Greek and it's been 30 years since I was there!

Maybe there's a "start-up cost." My first two visits I picked up only a few words by chance. But then an acquaintance with very limited English skill extrapolated from my question ("What does /mong-arai/ mean?") to teach me /khit-arai/ = "what you think?"; /gin-arai/ = "what you eat?", etc and the simplicity of this template jarred me into realizing it might be easy to become conversant.

There are MANY two-syllable words formed nicely by two simple one-syllable words. Learn a few 1-syllable words and get several 2-syllable compound words almost for free. You mentioned "intentionally". In Thai that's /tang-jai/ -- a verb which is simpler than an adverb hahaha -- which decomposes into the two words "establish mind." It is one of 285 derived words that Wiktionary shows on its /jai/ page:

Noun​

ใจ • (jai)

1. heart.
2. mind; spirit.
3. breath.
4. desire; intention.
5. opinion; view.
6. emotion; feeling.
7. favour; kindness.
8. centre; middle.

Here's an example of "Latin with a stroke": with a bit of creativity, you can read "Aśaña", but the more you look, the wronger it gets. The "a"s are from the different fonts, and why does the "ñ" have a "m" as its base?

It actually reads "crystal"

Here's a piece of Thai script which was deliberately designed to look like its English equivalent! Doesn't it seem to spell out L A almost-Y ?
BUT the first symbol is a vowel, usually written like an I, not an L. And the letter that looks a bit like an 'A' is an 'L' consonant. The final symbol is a silent 'Y'. (The dot above is a silencing mark that looks a bit like the apostrophe in "Lay's." Omission of the final 'S in "Lay's" is appropriate: Thais NEVER pronounce an 'S' at the end of a word.)

 

[Politesse]

Sentiment seems to be that language complexity is too ambiguous to quantize easily. (Of course this makes the "equal complexity" claim itself undecidable or even meaningless.)

It's not just that it is ambiguous; it's that the very concept of social complexity is a deeply loaded one, with a grim history attached to it. In truth, you'll find few academics interested in going anywhere near it. It may be that, to you, a more complex language isn't necessarily a superior one, but is that the view taken by, say the US Supreme Court or the Academie Francaise when they are weighing whether to forcibly exterminate indigenous languages, or scrub dialectic speech from government publications?

But it is also fuzzy thinking even without that context, and I think this thread has well demonstrated why. While one could imagine concocting some sort of objective criterion of complexity, the fact would remain that no such proposal has yet achieved anything like consensus adoption; even among fans of the notion of linguistic complexity, there are almost more proposed definitions of it than there are authors writing them. That is not how, cannot be how, scientific terminology works, and one should always question conclusions that precede meaningful investigation. That many people are convinced of language stereotypes prior to any sort of meaningful study should be a red flag for us in terns of whether or not those stereotypes are subjective or not. The scientist proposes an idea, then tests it, and only after analyzing the results of that test decides whether the idea needs to be refined or discarded. She does not assume a conclusion based on "common sense" or "what's obvious" and start hunting for data to support it. That is the job of a lawyer, not a researcher.

 

[Swammerdami]

Sentiment seems to be that language complexity is too ambiguous to quantize easily. (Of course this makes the "equal complexity" claim itself undecidable or even meaningless.)

It's not just that it is ambiguous; it's that the very concept of social complexity is a deeply loaded one, with a grim history attached to it. In truth, you'll find few academics interested in going anywhere near it. It may be that, to you, a more complex language isn't necessarily a superior one, but is that the view taken by, say the US Supreme Court or the Academie Francaise when they are weighing whether to forcibly exterminate indigenous languages, or scrub dialectic speech from government publications?

But it is also fuzzy thinking even without that context, and I think this thread has well demonstrated why. While one could imagine concocting some sort of objective criterion of complexity, the fact would remain that no such proposal has yet achieved anything like consensus adoption; even among fans of the notion of linguistic complexity, there are almost more proposed definitions of it than there are authors writing them. That is not how, cannot be how, scientific terminology works, and one should always question conclusions that precede meaningful investigation. That many people are convinced of language stereotypes prior to any sort of meaningful study should be a red flag for us in terns of whether or not those stereotypes are subjective or not. The scientist proposes an idea, then tests it, and only after analyzing the results of that test decides whether the idea needs to be refined or discarded. She does not assume a conclusion based on "common sense" or "what's obvious" and start hunting for data to support it. That is the job of a lawyer, not a researcher.

It sounds like you agree with me that
(1) The simple language may be in many ways superior to the "more complex language" and that
(2) The common claim really made by real linguists (e.g. R.M.W. Dixon "It is a finding of modern linguistics that all languages are roughly equal in terms of overall complexity.") is unclear or almost meaningless.

I stand by my claim that the rapidity with which one can become conversant in a language, albeit very VERY crude and very VERY inexact, may be a useful criterion for one interpretation of simplicity.

 

[Politesse]

I stand by my claim that the rapidity with which one can become conversant in a language, albeit very VERY crude and very VERY inexact, may be a useful criterion for one interpretation of simplicity.

What data set are you referring to?

Also, we understand the field very differently than I do if you think a linguistic theory can be "crude and inexact" at the same time as being "useful". Why use an ineffective tool to understand comparative syntax when many excellent generative models of syntax already exist to make them obsolete?