MIT Press

Table 2:

Main results on Gothic in a variety of settings using A@10 scores. All scores are reported in the format of triplets, corresponding to base / partial / full models. In general, more phonological knowledge about the lost language, more segmentations improve the model performance. The choice of the known language also plays a significant role as Proto-Germanic has a noticeably higher score than the other two choices.

WR^{^†}	Known language
WR^{^†}	Proto-Germanic (PG)	Old Norse (ON)	Old English (OE)	avg^{^††}
0%	0.820 / 0.749 / 0.863	0.213 / 0.397 / 0.597	0.046 / 0.204 / 0.497	0.360 / 0.450 / 0.652
25%	0.752 / 0.734 / 0.826	0.312 / 0.478 / 0.610	0.128 / 0.328 / 0.474	0.398 / 0.513 / 0.637
50%	0.752 / 0.736 / 0.848	0.391 / 0.508 / 0.643	0.169 / 0.404 / 0.495	0.438 / 0.549 / 0.662
75%	0.761 / 0.732 / 0.866	0.435 / 0.544 / 0.682	0.250 / 0.447 / 0.533	0.482 / 0.574 / 0.693
avg^{^‡}	0.771 / 0.737 / 0.851	0.338 / 0.482 / 0.633	0.148 / 0.346 / 0.500	0.419 / 0.522 / 0.661

WR^{^†}	Known language
WR^{^†}	Proto-Germanic (PG)	Old Norse (ON)	Old English (OE)	avg^{^††}
0%	0.820 / 0.749 / 0.863	0.213 / 0.397 / 0.597	0.046 / 0.204 / 0.497	0.360 / 0.450 / 0.652
25%	0.752 / 0.734 / 0.826	0.312 / 0.478 / 0.610	0.128 / 0.328 / 0.474	0.398 / 0.513 / 0.637
50%	0.752 / 0.736 / 0.848	0.391 / 0.508 / 0.643	0.169 / 0.404 / 0.495	0.438 / 0.549 / 0.662
75%	0.761 / 0.732 / 0.866	0.435 / 0.544 / 0.682	0.250 / 0.447 / 0.533	0.482 / 0.574 / 0.693
avg^{^‡}	0.771 / 0.737 / 0.851	0.338 / 0.482 / 0.633	0.148 / 0.346 / 0.500	0.419 / 0.522 / 0.661

^†

Short for whitespace ratio.

^††

Averaged over all whitespace ratio values.

^‡

Averaged over all known languages.

View Large

This Feature Is Available To Subscribers Only