Improve German morphology

manuelaugustin commented 5 years ago

The following parts of German morphology need to be improved:

Stemming

[ ] When a word ends in -ln, the -n should be removed to form the stem, see https://trello.com/c/7UA6WwVW/3-when-a-word-ends-in-ln-the-n-should-be-removed-to-form-the-stem
[ ] If an -er suffix is preceded by i (but not ei), it should not be removed, see https://trello.com/c/reG8065O/5-if-an-er-suffix-is-preceded-by-i-but-not-ei-it-should-not-be-removed
[ ] Add exceptions for nouns that are treated as compounds of irregular nouns, even though they're not (e.g. "Reisepass" is a compound of "Reise" + "Pass", but "Kompass" isn't a compound based on "Pass")
Don't stem verbs -ern from verbs ending in -rn (e.g., "schildern") https://trello.com/c/dD3oja2P/15-find-solution-for-stemming-issue-for-verbs-ending-in-rn

Verbs

[x] Add functionality for German participles #2190
[x] Make sure that German words that have a vowel+e/i+vowel combination are stemmed correctly https://github.com/Yoast/javascript/pull/146
[ ] Make sure that all irregular verbs are stemmed correctly https://github.com/Yoast/javascript/issues/163
[x] Make sure that German verbs ending in -t are stemmed correctly https://github.com/Yoast/javascript/issues/189
[ ] We can't correctly stem 3rd person singular verb forms ending in -t/-et for internal linking, see https://github.com/Yoast/javascript/pull/199
[ ] Stem past participles with separable + inseparable prefix (e.g., "wiederentdecken") https://trello.com/c/3iGq0lgt/24-german-past-participle-stemming
And more to come (see Lingo Research Trello board)
Add affixes for verbs ending in -l or -n, see https://trello.com/c/tp6AW8cj/13-review-some-rules-for-possible-verb-suffixes

Adjectives

https://github.com/Yoast/javascript/issues/236

Parent of #2190

Requires https://github.com/Yoast/YoastSEO.js-premium-configuration/issues/14

nataliashitova commented 5 years ago

While acceptance testing the basic functionality introduced in https://github.com/Yoast/YoastSEO.js/pull/2153 I've noticed a bunch of cases that don't work. In consultation with @manuelaugustin we decided that they should be addressed separately.

sein

sein > Personal forms, imperatives, past participle not recognized. I failed to receive a full paradigm for any of the input forms. I realize that this word would not likely to be used as a keyphrase, but for consistency and future Linguistic Tree usages it can be more than beneficial to have a nice sein.

haben

Verb forms are now generated using default (regular) rules. We need to add irregular forms, too.

verbs with separable prefixes

Verbs like einkaufen need special processing rules, which have not been implemented so far.

schenken and similar

Not all verb forms are recognized. schenkende > Personal forms, imperatives, past participle not recognized. geschenkt > No other forms are recognized.

nataliashitova commented 5 years ago

While working on https://github.com/Yoast/javascript/pull/210 I've noticed that the function generateVerbExceptionFormsSpec fails to generate the form iss from any of the stems of the verb essen. We should investigate how local this problem is (just for this verb, only for verbs whose stems end with s, only for verbs of class 4, etc.).

manuelaugustin commented 4 years ago

All open points moved here: https://yoast.atlassian.net/browse/LIN-577

Yoast / YoastSEO.js