Incorrect lemmatization for German at beginning of sentence

[karelin]

How to reproduce the behaviour

import spacy
nlp = spacy.load('de')
doc = nlp('Gemeinsame Erklärung von Singapur: Was genau steht drin?')
[(t.lemma_, t.pos_) for t in doc][0]
[('Gemeinsame', 'ADJ')]

BUG: expected lemma is gemeinsam, this lemma is produced e.g. in sentence

US-Präsident Donald Trump und Nordkoreas Machthaber Kim Jong Un haben bei ihrem Gipfel in Singapur eine gemeinsame Erklärung unterzeichnet.

Note, in contrast with #2166, the POS tag is correct (for both sentences).

Info about spaCy

• spaCy version: 2.0.11
• Platform: Windows-10-10.0.17134-SP0
• Python version: 3.6.5
• Models: de, en, xx

[ines]

Thanks! This is likely related to the case-sensitivity and possibly an underlying bug in the lemmatizer (see #2368). Will investigate – we definitely want to sort this out for v2.1.x.

[ines]

Merging this with #2486!

[lock]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.