Comment by yorwba | Hacker News Reader

Comment by yorwba | Hacker News Reader

Parent Full thread

I think if you wanted to construct a language for maximum learnability, you'd have to tackle it both from the perspective of a speaker/writer and a listener/reader. The language should allow a speaker to make any distinction they can in their native language, but not require them to make a distinction that is foreign to them. On the other hand, it should allow a listener to understand any distinction they could understand in their native language, but allow them to ignore everything else.

In practice, these requirements are obviously at odds with each other. You have to balance the freedom available to a speaker with the mental capacity of a listener. If you want to accomodate speakers of languages that allow free word order due to case marking, as well as languages that have a fixed word order, you might end up with case marking and marks to indicate a specific fixed word order. But then all of these markings would have to be understood by the listener, who'd have trouble keeping the various forms straight.

However, you might be able to quantify the learning effort required and optimize for that. This is obviously hard for grammar and semantics, but could be done for phonetics.

Instead of trying to intersect the phoneme sets in use, or giving up and choosing arbitrarily, you could collect statistics on the ability to produce and distinguish various phones, and then optimize for communicability. A speaker can pronounce a phoneme as well as the allophone that is easiest to pronounce for them, a listener can distinguish a pair of phonemes as well as the most similar combination of allophones from them. Then calculate the expectation over all pairings of speaker and listener, maximize the bandwidth accounting for the required error correction, and you get your perfect phoneme set. I'm actually interested what the result might look like, whether there will be a few, clearly separate phonemes, or a huge number with potential overlap, that needs to be compensated by unambiguous vocabulary.

It also occurred to me that this might somehow apply to constructing programming languages, but I'm hazy about the details. Maybe something like Python's "one way to do it" vs other languages supporting a variety of styles?

Esperanto is already far easier than natural languages for anyone to learn; it's about 10x easier than English or German for French speakers. I'm not sure how much easier it can get.

[1] https://en.wikipedia.org/wiki/Propaedeutic_value_of_Esperant...

Easier than natural languages for anyone French :)

It's not particularly surprising since Esperanto borrows a lot of vocabulary from Romance languages.

Even citing from the same Wikipedia page:

>According to the results, preliminary Esperanto study led to a 25% improvement in acquiring Russian, 30% for German, 40% for English, and even 50% for French.

It seems to clearly show a greater similarity with French (a romance language) than the others.

I'm French myself but I must say that I never really got the appeal of Esperanto. Sure, I might be able to learn it relatively easily. I could also learn Italian pretty easily and I'd gain access to around 64million native speakers and a very rich culture.

If you want an international lingua franca then there's already English and if you think that's too difficult to learn there's Spanish. Not as easy as Esperanto? Maybe, but call me when Esperanto reaches half a billion speakers.

Also the "10x easier than English or German for French speakers" surprises me because it seems to say that English and German are roughly equally difficult to learn for French speakers while it seems pretty obvious to me (having studied both languages at school for many years) than German is very significantly harder.

English has many cognates thanks to the large influx of French words in past centuries, a very simple grammar for the most part (trivial conjugations in particular), no real case system (only a few leftovers like in Romance languages), no gendered nouns. The only thing that's a bit more difficult in English as far as I can tell is the pronunciation.

The answers to your questions are in the wikipedia link.

> "10x easier than English or German for French speakers" surprises me

145 hours of Esperanto = 1500 in English = 2000 in German

> I could also learn Italian pretty easily

Can you learn Italian with less than zero effort? No. Yet in a number of studies listed, those who learned Esperanto and another target language (including Japanese) became better at the target than those who learned only the target.

Less than zero effort? Probably not. In 150hours? I'd say I'd get a pretty decent level. French and Italian really are very close, the main difficulty is probably adjusting to the phonology of the spoken language coming from French's rather barbaric pronunciation. I can already slowly decipher Italian newspapers without having ever studied the language.

And once I get fluent enough I have access to a wealth of resources to keep studying: movies, tv shows, radio, videogames, newspapers, philosophy books, plays, operas... I can go anywhere in Italy and practice with people in all kinds of situations. I can even get a job there where I'll speak Italian all day long.

Esperanto is an interesting experiment but I find it really hard to invest the time necessary to learn the language.

>Yet in a number of studies listed, those who learned Esperanto and another target language (including Japanese) became better at the target than those who learned only the target.

Did they check if people performed worse when taught, say, 150 hours of Spanish before the target language? It's not really a surprise that learning a language can make it easier to learn other languages, if only because it gives you a certain discipline and mental framework.

And if it does turn out that learning 150hours of Spanish has the same benefits, wouldn't it make more sense to do that instead even if the level of proficiency reached is lower? The practical uses of (even basic) Spanish dwarf that of Esperanto.

> Did they check if people performed worse when taught, say, 150 hours of Spanish before the target language?

The Australian study [1] that is cited in the Wikipedia article compared pupils that learned Esperanto, Indonesian, German or Japanese, respectively, in primary school and then went on to learn a possibly different language in secondary school.

Let me note a few problems with that study up front:

- There was apparently an additional class per week by a professional Esperanto teacher (the other Esperanto teachers were learning on the go).

- The Japanese teachers in the comparison rated themselves 0.25-0.5 on a number of 5-point scales (Esperanto: 1.6-2.4, German 4.5-5, Indonesian: 2-3). This is essentially a plus for Esperanto (easy to learn for the teachers), but taints the comparison since the Japanese teachers were basically incompetent.

- Due to attrition of study subjects, their evaluation after one year of secondary school has only 21 students of Esperanto. (306 students in total)

- They do not differentiate by (or even talk about) the languages learned in secondary school.

Now to the findings (on a 5-point scale [1,2,3,4,5], rated by the secondary-school teachers, given as percentages in each category):

- Motivation to learn the language (presumably the new one): Esperanto [0, 5, 10, 67, 19], others [ 13, 9, 22, 24, 31]. This one is difficult to analyze, since their motivation might change due to liking the new language better (or worse), or due to a general effect on motivation to learn any language (the authors claim the latter). I'll just point out that both the lowest and the highest scores are rarer among Esperanto students.

- "attainment level" (no idea what that specifically means): Esperanto [0, 4, 26, 61, 9], others [10, 9, 23, 26, 32]. Again, both the lowest and the highest scores are rarer among Esperanto students. The striking resemblance to the motivation results is interesting.

- speaking ability: Esperanto [0, 0, 48, 43, 10], others [12, 8, 46, 27, 6]. This is the only measure where Esperanto seems to be a uniform improvement.

In my opinion, this study mostly shows that this specific Esperanto course did not completely kill off the students ability to learn a foreign language, unlike apparently some other courses. The improvement in speaking ability could be caused by the focus on an immersive classroom experience, but it did apparently transfer to other languages, which is a plus.

The study definitely doesn't show anything like "those who learned Esperanto and another target language (including Japanese) became better at the target than those who learned only the target." (Since they didn't test any such thing.) In fact, my best guess at explaining the higher rate of perfect motivation/attainment scores for students of other languages is that they continued studying the same language they had already had in primary school. If this were the case, it would pretty clearly show that learning Esperanto before learning Japanese is not going to help someone learn Japanese faster overall.

[1] http://web.archive.org/web/20040215074307/http://www.educati...

That is a refreshingly honest evaluation of the study that you tend not to see when these studies get brought up (the Wikipedia page is particularly bad at this).

We already know that language acquisition is going to be driven in large part by motivation on the learner's part, as well as competency on the teacher's part, and it's all too easy to construct a study that self-selects motivated students and competent teachers for Esperanto and unmotivated students and marginal teachers for other languages.

Another major question that's completely unanswered to me is the effect of pedagogy on teaching (and, perhaps to a lesser degree, to assessing competence in the learned language). One of the studies Wikipedia cites claims that you can learn the grammar of Esperanto in ~5 hours, which is "impossible" for other languages. That instead screams to me that Esperanto is taught as a detached grammar course in early instruction, whereas other languages will tend to start with a phonological crash-course combined with basic phrase acquisition, instead spreading grammar out over several hundreds of hours of instruction.

Completely agree with you. In particular regarding the "motivated students and teachers" aspect I think it probably plays a role in favor of esperanto in these studies.

I don't think many people have to learn esperanto at school or for work, clearly it heavily biases towards people who are interested in learning languages and esperanto in particular. You have to be very careful while creating your samples to compensate for that I think.

>One of the studies Wikipedia cites claims that you can learn the grammar of Esperanto in ~5 hours, which is "impossible" for other languages. That instead screams to me that Esperanto is taught as a detached grammar course in early instruction, whereas other languages will tend to start with a phonological crash-course combined with basic phrase acquisition, instead spreading grammar out over several hundreds of hours of instruction.

That might be true but on the other hand imagine teaching English students a "detached" French grammar for instance. I think the students would give up within the hour, it's too dense and quirky to be palatable. If Esperanto's grammar is really simple and regular and can be integrated in 5 hours it's a good point in its favor.

"The only thing that's a bit more difficult in English as far as I can tell is the pronunciation."

Add in there a greater load of multi-word expressions compared to other languages, and painstaking word order rules (like if there's more than one adjective then those have to follow each other in a specific manner), and many other downplayed hardships.

It's true, I almost mentioned multi-word expressions because I recall that it was a sticking point when learning the language. That being said German loves compound words and those can get a while to get used to as well. Phrase structure is also pretty different in German, with a propensity to put the verb at the end of sentences in some forms.

Adjective order however is not really a problem IMO, it's irrelevant while consuming the language (unless a text tries to convey some meaning by purposefully miss-ordering the adjectives) and in general getting it wrong while producing English will make you sound foreign but shouldn't hurt comprehension too much. Saying "the red big house" instead of "the big red house" will raise some eyebrows but should remain understandable.

Reply on news.ycombinator.com