Are these Spanish translations by computer accurate?

Ask specific questions about your target languages. Beginner questions welcome!
Dragon27
Green Belt
Posts: 267
Joined: Tue Aug 25, 2015 6:40 am
Languages: Russian (N)
English - best foreign language
Polish, Spanish - passive advanced
Tatar, German, French - studying
x 544

Re: Are these Spanish translations by computer accurate?

Postby Dragon27 » Fri Jan 15, 2021 5:48 am

AcademiaNut wrote:But where is a complete list of Spanish sentence patterns? I couldn't find such a list online.

Apparently, no one got interested with this idea enough to bother to apply it to another language.
Much of the linguistic stuff available for the English language is unavailable for most other languages (if not all of them). Languages are heavily unequal in this regard, you just have to get used to it.

AcademiaNut wrote:Since English speakers learning Spanish will naturally try to translate sentences they already speak into Spanish, why do Spanish teachers not realize this and explicitly explain how each English grammar pattern would be translated?

Because this idea is as questionable (i.e. it's rubbish) as trying to map each word in one's native language to a word in the target language. That's not how languages work. Spanish language is not an English language in different clothes. Words, phrases, sentence patterns, grammatical categories, etc. in different languages do not correspond to each other in general. Anyone who has managed to learn at least one foreign language to a fairly advanced level understands that.

It's one thing if you're playing around with these sentence patterns and grammar systems and vocabulary lists just out of purely academic linguistic interest. If you just want to know more about some technical subtleties of languages for its own sake. This is an interesting endeavor and can be a satisfying intellectual exercise. But if your goal is to actually learn languages in order to speak them and use them and undestand spoken or written texts, then you should be told from the get-go that you're on the wrong path. You will not be able to find a mechanical algorithm you're wishing for to construct meaningful and natural sentences. You cannot learn languages the way you learn to solve quadratic equations in school. Your "big picture" you wanted the teachers to see and teach doesn't and cannot exist (if it did, the teachers would happily teach it to students because it would significantly simplify their job; they already try to do as much, but fall short of the task, because it is impossible, after all). The sooner you realize this, the sooner you'll begin the actual process of acquiring the target language (if that is your actual goal).

But feel free to just dismiss all of that. Our brains are very good at resisting ideas we don't agree with until we learn from our own painful experience.
4 x

Cainntear
Black Belt - 1st Dan
Posts: 1859
Joined: Thu Jul 30, 2015 11:04 am
Location: Scotland
Languages: English(N)
Advanced: French,Spanish, Scottish Gaelic
Intermediate: Italian, Catalan, Corsican
Basic: Welsh
Dabbling: Polish, Russian etc
x 4652
Contact:

Re: Are these Spanish translations by computer accurate?

Postby Cainntear » Fri Jan 15, 2021 12:00 pm

AcademiaNut wrote:But where is a complete list of Spanish sentence patterns? I couldn't find such a list online.

Also: Why does English seem to be the only language ever listed with a complete list of sentence patterns? And why isn't that list for English better known?

I have a degree in English language, a Cambridge CELTA certificate in English teaching and a Masters degree in teaching, I can say I was never shown any such list.

Now, you've just repeated the description of it as "complete", despite the very long list of caveats that tell us the very many types of sentences that are not included.

Why isn't that list taught in English grammar classes long before students reach college, or before they learn a foreign language?

Because it is practically valueless. Without the claim of completeness, it might be mildly useful, but as it is, it is totally misleading.

Since English speakers learning Spanish will naturally try to translate sentences they already speak into Spanish, why do Spanish teachers not realize this and explicitly explain how each English grammar pattern would be translated?

Let's start with where you're right:
Language learners always fixate on their native language (L1). Many teachers work under the assumption that simply telling them not to and ignoring the L1 will result in them not doing. I personally have always found that understanding what the L1 does helps understand why the new language (L2) does it differently.
In short, teachers should make students aware of what they do, so that they can consciously and deliberately do something different. We agree on this.

But those 10 patterns do not describe why English does what it does, and therefore are totally insufficient to allow an informed translation. You are again talking about transformational rules, despite telling me explicitly in another thread that you were not.

I've already demonstrated why this type of superficial "pattern" is not enough for translation with the dog food example -- the information required to create a good translation is not encoded in this type of superficial pattern.

If Spanish has so few grammar patterns, then why wouldn't those be even easier to learn, and therefore why wouldn't a list of them appear in every Spanish grammar book, or at least somewhere online?

Let's go back to Computer Science and software dev for a minute.

Those 10 patterns, or "macropatterns" are combinations of many "micropatterns" (grammar rules), and these micropatterns are also the foundation of the great many sentence patterns that were hand-waved away from the "complete".

Now, if you're coding these up in a computer, should you be coding this incomplete complete list of macropatterns, or should you be generating these macropatterns procedurally from a set of micropatterns that can generate the genuinely complete list?
In CS terms, the answer is unequivocally procedural generation as it is logically correct and prevents reduplication of data; from a software dev perspectivethe answer is the answer to the question "Which is quicker?"

In English, with it's fixed word order "I shot him" can mean many things, but in Spanish, with its freer word order, it might translate to "le disparó", "le disparó a él", "le disparó yo". The information we need to disambiguate between these isn't encoded in the 10 patterns, but in a variable that has been disregarded: vocal emphasis.
"le disparé" = "I shot him" or "I shot him" (as in "I didn't kiss him, I shot him.)
"le disparé a él" = "I shot him" or "I shot him"
"le disparé yo" = "I shot him" or "I shot him"

And note that I've included the neutral, unemphasised form for every alternative in Spanish, because even including the emphasis data you still don't have enough information to know how to translate for absolute certain every time.

So you should hopefully be able to see that if we want to truly expand out the micropatterns into enumerated macropatterns, you're heading towards a combinatorial explosion -- we only expanded out one variable there, and each of our macropatterns is now replaced with two, three or even four. Adding in questions doubles. Adding in negatives doubles again. We're now at about, what, 120ish? Adding in imperatives doesn't double it, but should bring us up to about 180.

It seems to me that would be a much faster way to learn, as well: The student could see at a glance all the grammar patterns they would ever need to learn, in a short list, then master those

Memorising a list isn't mastering it. I agree, awareness of why and how examples take on the form they use, but these patterns simply do not do that.

With knowledge of all the patterns that can exist, language learning then reduces to almost a mechanical process of filling in the slots: 4 slots for the overall topics (script, pronunciation, grammar, vocabulary), 10 subslots for the grammar patterns + a few rules for any relative clauses hanging off of them, maybe 5 subslots for phonemes that don't exist in English that need to be mastered, a list of the 3 verb types and how to conjugate them (-ar, -er, -ir), and a few more miscellaneous rules, like for agreement of gender and plural. The student can then see at a glance the entire set of knowledge that needs to be tackled, and how it is organized, and plan accordingly. Sadly, I suspect not even the teachers themselves see the big picture, which is why they do not teach it.

This approach has been tried... it does not work.

The presentation of the patterns encodes the problem.

Someone came up with the principle of patterns and slots to simplify all the individual grammar rules. They convinced themself it was right.
They started expanding out. Soon they got hit by the combinatorial explosion. But rather than accept that the combinatorial explosion proved them wrong, they simply started putting up caveats and saying "complete... except for...", effectively saying "I'm right if you ignore all the evidence I'm wrong."

Why do you only see it in English? Because it's much harder to kid yourself on that it works in any language other than English, and because it is rejected by actual academics on grounds of being naive and insufficient so it's only spread by naive, uninformed English teachers: the English teaching industry is huge and as a result is full of people who are not properly trained and build incomplete frameworks that allow them to function as a teacher, then share them as though they are some kind of amazing truth.

[edited for errors in Spanish.]
Last edited by Cainntear on Tue Jan 19, 2021 9:56 am, edited 1 time in total.
4 x

caam_imt
White Belt
Posts: 46
Joined: Fri Oct 30, 2015 10:04 pm
Location: FI
Languages: Español (MX)

English (C1)
Suomi (C1)
Deutsch (~B1?)
Svenska (~A2-B1?)
日本語 (~A1?)
x 49

Re: Are these Spanish translations by computer accurate?

Postby caam_imt » Tue Jan 19, 2021 8:20 am

Small correction for the above post: in all three examples, it should be "disparé" instead of "disparó". Otherwise correct.
0 x

Cainntear
Black Belt - 1st Dan
Posts: 1859
Joined: Thu Jul 30, 2015 11:04 am
Location: Scotland
Languages: English(N)
Advanced: French,Spanish, Scottish Gaelic
Intermediate: Italian, Catalan, Corsican
Basic: Welsh
Dabbling: Polish, Russian etc
x 4652
Contact:

Re: Are these Spanish translations by computer accurate?

Postby Cainntear » Tue Jan 19, 2021 9:57 am

caam_imt wrote:Small correction for the above post: in all three examples, it should be "disparé" instead of "disparó". Otherwise correct.

:oops:
It's amazing what an entire year of isolation will do to your language skills.
DAMN YOU, COVID19!!!!!
2 x


Return to “Practical Questions and Advice”

Who is online

Users browsing this forum: No registered users and 1 guest