Datasets:

Urdatorn
/

norma

@@ -12,7 +12,7 @@ configs:
 We introduce *Norma* as a common benchmark for the evaluation and comparison of NLP tools concerning markup of two tasks for Ancient Greek (grc): (1) **vowel length** of dichronic vowels (alpha, iota, ypsilon) in open syllables (where they impact syllable weight) and (2) **syllabification**, both boundaries and weight. This means that the benchmark also indirectly tests handling of sandhi phenomena and, for verse, metre.
-The benchmark includes 1228 lines of text by each of the below 14 authors, spanning many periods and genres of both verse and prose:
 - Alcman (*Louvre Partheneion* 36-49, mix of aeolic, trochaic and dactylic cola, NB: contains digamma, ϝ)
 - Bacchylides (*Epinikion 5*, anapaesto-iambic and dactylo-troachaic)

 We introduce *Norma* as a common benchmark for the evaluation and comparison of NLP tools concerning markup of two tasks for Ancient Greek (grc): (1) **vowel length** of dichronic vowels (alpha, iota, ypsilon) in open syllables (where they impact syllable weight) and (2) **syllabification**, both boundaries and weight. This means that the benchmark also indirectly tests handling of sandhi phenomena and, for verse, metre.
+The benchmark includes 1,916 vowels (323 long, 1,593 short) to disambiguate across 1,228 lines of text by each of the below 14 authors, spanning many periods and genres of both verse and prose:
 - Alcman (*Louvre Partheneion* 36-49, mix of aeolic, trochaic and dactylic cola, NB: contains digamma, ϝ)
 - Bacchylides (*Epinikion 5*, anapaesto-iambic and dactylo-troachaic)