Table 2. Power-law fitting results for words and lemmas, denoted respectively by subindices w and l.

V is the number of types (vocabulary size), n _m is the maximum frequency of the distribution, N _a is the number of types in the power-law tail, i.e., with n ≥ a, a is the minimum value for which the power-law fit holds, and γ and σ are the power-law exponent and its standard deviation, respectively. 2σ _d, the double of the standard deviation σ _d is also given. σ _d is the standard deviation of γ _l−γ _w assuming independence, which is ${\sigma }_d=\sqrt{{\sigma }_w^2+{\sigma }_l^2}$. The last column provides ℓ₁, the number of lemmas associated to only one word form. Notice that the lemma exponent is very close to the one found in Ref. [29] for the tail of a double power-law fitting, except for Moby-Dick and Ulysses.

Title	V w	n mw	N aw	a w	γ w ± σ w	V l	n ml	N al	a l	γ l ± σ l	2σ d	ℓ1
Clarissa	20492	38632	1514	51	1.83±0.02	9041	41679	838	101	1.83±0.03	0.07	5750
Moby-Dick	18516	14438	2658	8	1.97±0.02	9141	14438	1548	13	1.90±0.02	0.06	6157
Ulysses	29450	14934	4377	6	1.95±0.01	12469	14934	1024	26	1.97±0.03	0.07	8670
Don Quijote	21180	20704	939	40	1.93±0.03	7432	31521	936	32	1.83±0.03	0.08	3812
La Regenta	21871	19596	1196	26	2.01±0.03	9900	32300	993	32	2.00±0.03	0.08	5308
Artamène	25161	88490	936	200	1.86±0.03	5008	119016	641	200	1.79±0.03	0.08	2178
Bragelonne	25775	26848	3173	16	1.84±0.02	10744	45577	1382	40	1.84±0.02	0.06	5391
Seitsemän	22035	4247	22035	1	2.13±0.01	7658	4247	474	26	2.13±0.05	0.10	4246
Kevät ja	25071	5042	8660	2	2.05±0.01	8898	6886	699	20	1.96±0.04	0.07	5060
Vanhempieni	35931	5254	6523	3	2.09±0.01	13510	7526	571	32	2.05±0.04	0.09	7837