Table 5. Median errors, measured by $\overline{F_1}$, for individual language datasets and six estimation procedures.

Lang	xval(1.9 strat, block)	xval(9:1, no-strat, block)	xval(9:1, strat, rand)	seq(9:1, 20, equi)	seq(9:1, 10, equi)	seq(2:1, 10, semi-equi)
alb	0.026	0.016	0.137	−0.014	−0.007	−0.009
bul	0.020	0.024	0.047	0.003	−0.002	−0.019
eng	−0.019	−0.020	−0.015	−0.027	−0.027	−0.028
ger	0.056	0.058	0.072	0.025	0.028	0.014
hun	0.022	0.022	0.030	−0.006	−0.009	−0.005
pol	0.013	0.020	0.044	−0.001	0	−0.007
por	−0.050	−0.045	−0.040	−0.049	−0.056	−0.092
rus	0.008	0.010	0.025	−0.019	−0.018	−0.021
scb	−0.034	−0.037	0	−0.030	−0.032	−0.050
slk	0.005	0.008	0.025	−0.013	−0.015	−0.013
slv	0.003	0	0.029	−0.022	−0.026	−0.032
spa	−0.001	0.024	0.060	0.007	0.010	0.012
swe	0.030	0.037	0.071	0.008	0.006	−0.011
Median	0.008	0.016	0.030	−0.013	−0.009	−0.013