Annotation Tools for Historical German

These tools have been developed in cooperation with Ercong Nie.

Tools for Early New High German

• Annotations: part of speech, lemma
• trained on: Potsdam ENHG treebank ReF.RUB
• uses: pretrained BERT model dbmdz/convbert-base-german-europeana-cased
• uses also: lemmatizer component of the RNNTagger
• The POS tags of the training data have been semi-automatically extended with morphological features and lemmas using ReF.RUB and ReF.MLU.

• Annotations: constituent parser trees
• trained on: Referenzkorpus Frühneuhochdeutsch ReF.UP
• base parser: Berkeley Neural parser
• uses: pretrained BERT model dbmdz/convbert-base-german-europeana-cased

Tools for Middle High German

• Annotations: part of speech, normalized form, lemma
• trained on: Referenzkorpus Mittelhochdeutsch
• also part of the RNNTagger distribution

• Annotations: constituent parser trees
• trained on: Indiana Parsed Corpus of Historical High German (Middle High German part)
• base parser: Berkeley Neural parser
• uses: pretrained BERT model dbmdz/convbert-base-german-europeana-cased