3

Is there a tool that will analyze my German sentence and tell me which is a dative case which is genitive, etc. And ultimately tell me if my sentence was properly written out to be grammatically correct.

This would help me a lot for fully understanding the German cases.

5
  • Possible duplicate of Automatische grammatikalische Satzanalyse Apr 6, 2017 at 20:16
  • 1
    @HubertSchölnast Your question is in German. As far as I know, an English question is no duplicate, right?
    – Arsak
    Apr 6, 2017 at 21:08
  • @Marzipanherz: This is discussed here: german.meta.stackexchange.com/q/14/1487 As far as I know, this discussion has not yet led to any official result. Apr 7, 2017 at 11:45
  • @HubertSchölnast Eben drum :)
    – Arsak
    Apr 7, 2017 at 15:23
  • 1
    The generic understanding of natural language is one of the hardest problems of computer science. Solutions today are usually based on neural nets, which are trained on HUGE corpi of sample texts. In turn, even a perfect neural net does not allow a rule based insight into the thing it specialized in, since the rules are implicitely coded in the purely numerical parametrization that characterizes the relationships of the cells that consititute the network. Apr 10, 2017 at 10:23

1 Answer 1

4

It is not so hard, because dative and accusative are always very different:

  1. With bestimmten Artikel, there is always difference (den-dem, die-der, das-dem, die-den). Similarly with unbestimmten Artikel (einen-einem, eine-einer, ein-einem, keine-keinen).
  2. In the case of pronouns, there is also a very visible difference (dich-dir, sie-ihr, etc)

On the first spot, this whole thing is not very hard. We have around 3 4x4 matrices (plus the pronouns, but also they are similar). The problem is that these to automatize... well... it can take years, even if your first language has much bigger ones. It is a big mistery of the language.

If I understand German sentences, I don't take care about what is dative or accusative. I simply decode the text from the meaning of the words. To identify Angehörigkeitrelationen ("of") is much more important, but it is also very characteristic.

To synthetise these structures, it is more easy to learn as realtime decoding them.

I think German is the best algorithmizable language I know, but this is also a reason, why is it a big win if you don't write code for that, instead develop this tool in your own mind. Look for complex sentences written by native speakers and decode some them.

15
  • 1
    "I think German is the best algorithmizable language I know." I fully agree, although I am not a computer scientist myself. But German is as close to being a "mathematical" idiom as natural languages can ever hope to be. Apr 8, 2017 at 8:33
  • 1
    Errr what?! German has a high degree of irregularity when it comes to things like prepositions, it has genders of imported words that are more or less randomly assigned based on comparable German words, it's impossible to flex many verbs without having a dictionary, there's HUGE freedoms on how to structure sentences (which comes with very different meanings)... German would be one of the harder languages to understand as a set of rules, tbh Apr 10, 2017 at 6:55
  • @MarcusMüller ABC die dieses in Reihenfolge Sätzes sind Wörter. :-)
    – peterh
    Apr 10, 2017 at 7:26
  • 1
    @peterh exactly, while many structures are valid, many aren't, and it's very hard to know the difference based on strict rules. Construct a valid sentence from your words. Switch a couple of words, and still form a valid sentence, with a different meaning. Rinse, repeat. Apr 10, 2017 at 7:57
  • I also disagree with the computer analysis statement. German has way too many ambiguities. If one takes a look at e.g. Lojban, which was designed to be understood by computer programs, one comes to realize how horribly unprecise natural languages can be.
    – marstato
    Apr 30, 2017 at 23:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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