If the accepted answer stands, that's remarkable. I wonder how one could measure a language efficiency. Maybe syllable count ? But one would need a sort of assembly to translate to and verify that a sentence computes the intended information.
Translation is a process which both erases information and introduces new information. Any comparison of languages which tries to evaluate which languages are more compact has to work with some assumptions about what information should be conveyed. A statistical distribution of language-independent messages. But when you choose a distribution, you’re encoding your biases.
Not saying that language efficiency is a bunk concept, just that it’s a thorny, difficult concept to quantify. Same is true of data compression algorithms—there is no such thing as an absolute scale for Kolmogorov complexity, for the same reasons.
When translating for fun, I've often run into a choice between:
- Preserving the author's meaning as literally as possible.
- Preserving the author's style.
A translation can be literally very accurate, while destroying everything that made the original work charming. Or it might preserve the feel and the flavor of the original work, but skim over a lot of the details. A really good translation captures more of both, with fewer trade-offs.
Jorge Luis Borges encouraged his translators to improve upon his original work, if possible. He worked extensively with Di Giovanni, one of his translators, debating the best way to capture certain phrases in English: https://medium.com/@michael.marcus/dear-mr-borges-which-tran... His preference was almost always to capture the "feel" of the work, rather than a strictly literal translation.
I have an odd book, which contains three copies of the same story: An original in English, a French translation, and then a translation back into English by a new translator. The French version definitely loses something, and the second English version loses a bit more. But in the second English version, there is an occasional delightful turn of phrase, something that's briefly better than the original version. Translation is hard.
"Translation is like a woman. If it is beautiful, it is not faithful. If it is faithful, it is most certainly not beautiful."
I'm not sure if the Latin translation as it is captures that, or if you'd go for something more like "valedīxisse nimium valedīxit", "she said goodbye having said goodbye too much"; I kind of like that because then you're saying goodbye twice in the same line.
Exactly (or at the time). Many submissions here attain shortness by eliding this important precision.
https://en.wikipedia.org/wiki/Breathless_(1960_film)#Closing...
That’s why the translations of Asterix are so impressive.
There's an essay I enjoyed by Douglas Hofstadter which is all about this, though from an artistic POV without much (any?) information science. Translator, Trader. The title itself is a fun bit of translational wordplay on "traduttore, traditore."
An area of active research:
* https://www.theatlantic.com/international/archive/2016/06/co...
* https://www.degruyter.com/document/doi/10.1515/lingvan-2020-...
A constructed language:
* https://en.wikipedia.org/wiki/Ithkuil
* Via: https://www.reddit.com/r/linguistics/comments/zyi8q/what_is_...
Doing a search for "language information entropy" also gives back a number of results.
In an optimally breve language the meaning of a text could completely flip when a single letter/syllable/phoneme is changed. That, in turn, means listeners have to hear every letter/syllable/phoneme perfectly.
Interestingly, natural languages already have a bit of both.
As an example, if you skim-read a text and restart at “He said she wasn’t there anymore”, there are 3 ‘back references’ in that sentence that require you to look back in the text to find the meaning of.
Also, a paragraph’s meaning can change by adding the sentence “Just joking.” Or even a simple “Not.”.
then don't these other langs do a 'bad job' at compression ?
> That doesn't seem feasible
In general maybe not. But for some restricted 'assembly' ?
"The cat is on the table" has no ambiguity. And in some langs like Polish it compresses better : "Kot jest na stole" (Cat is on table), same info, better syllable-wise compression (5 vs 7).
For instance, your assumption that the definite "the cat" is being used idiomatically like so: this sentence, used in the manner you offer, might be used in conversation might occur in a farmhouse somewhere between an old man and woman who have lived together in this house for a long time, i.e. American Gothic. There's a vast amount of shared information and a perception of very little ambiguity held by both the speaker and the listener (whether correct or mistaken!). Any of those might fail. Furthermore, to use this sentence in English unadorned by context requires that both the speaker and listener have a shared reference to _what_ cat is being referred to by the definite article, "the". This very well might come with an unambiguous default in other languages!
Translation only gets more complicated from this.
Kanji look very compressed, and can convey a lot in a single character, but if there isn't one for your needs things can get ugly. Whereas English speakers find it much easier to borrow, shorten, abbreviate, or make up words for convenience.
In Turkish if you say kedi you mean the cat. The copula is in the third person is conveyed by an empty word at the end of a sentence.
Masada bir kedi var.
There is a cat on the table. A cat is on the table. A cat is said as one cat = bir kedi.
*Bir kedi masada.
Doesn’t work. This says a cat on the table. It’s an ellipsis.
“Darmok and Jalad at Tanagra” begs to differ
However, traditional way of computing efficiency of compression would not be useful for a meaningful analysis of the efficiency of a language. Barring issues like having an ideal encoding to bits, or even having the concept of "efficiency" being rigorously defined, there are problems just from the outset.
Take context for example.
All useful compression methods have some sort of decompression key involved. This could be the dictionary, or the bitmap or the know-how (for cases like RLE). In natural langauges, the compression/decompression key is stored in a distributed fashion across the minds of a society.
"Darmok and Jalad at Tanagra" is a VERY efficient compression for what is presumably a very long story about two hunters who met at an island and fought a beast together, but it is only efficient to the people who speak that language. The "local" efficiency (to the population who speak the language) is very high, but the "global" efficiency isn't.
So we must account for efficiency in terms of the size of the compressed concept as well as the compression key. And from my experience, it's a sorta lumpy kinda world out there.
This has been done! The answer is about 39 bits a second.
https://www.science.org/content/article/human-speech-may-hav...
I suppose image macros/memes are the modern equivalent. Social context enables readers to "decompress" the meme.
[Drake top]: "Two hunters who met at an island and fought a beast together"
[Drake bottom]: "Darmok and Jalad at Tanagra"
ISTM a similar principle would need to apply here: learning the "Darmok and Jalad at Tanagra" language would involve absorbing many volumes of history and mythology where for the usual sort of language a dictionary, grammar reference and maybe a book of common idioms would suffice.
Whatever metric is used to compare languages for efficiency should reflect this.
Similar thing I’ve noticed with the south indian language - Malayalam, just try to pronounce the name of the city - Thiruvananthapuram, local speakers would pronounce it with roughly the same speed as “London”, and would enunciate every syllable - its crazy.
No magic but plain agglutination, and I am sure this should be possible in languages like Finnish too...
https://www.science.org/content/article/human-speech-may-hav...
For example, mandarin or japanese can be very short on the character count. However, this increases character complexity and makes the languages harder to learn. On the other hand, large parts of english tend to be simple to learn.