This work was supported by the National Programme for the Development of Humanities (NPRH) [0036/NPRH2/H11/81/2012, 0413/NPRH7/H11/86/2018].

I would like to thank two anonymous reviewers for their careful reading of my manuscript and their suggestions and comments.

1. Introduction

4In this article, I will focus only on the three mentioned inflectional forms (Nom, Acc and Voc pl m), whose inflectional endings fluctuated and changed in the 17^th and 18^th centuries. The connection between the choice of a specific inflectional ending and the denotation of the noun (a person or not a person) was being formed only then. I will examine these changes on the example of nouns. The authors of the 17^th and 18^th century text corpus developed a special method of annotating the changing inflectional forms of nouns. Our method of annotation in this case differs from that used in the corpora of modern Polish, because the modern method of annotation would blur the variance of the endings of the forms Nom, Acc and Voc in the plural, marking forms with different endings with the same tag. The adaptation of the tagset helps to research the formation of this specific variance.

2. Plan

5The third section of the article contains a short review of previous work on the formation of the masculine personal category in Polish. In the fourth section of the article, the masculine personal category will be explained. Tagset solutions adapted to the phenomenon of the formation of this grammatical category in the 17^th and 18^th centuries will be presented in detail, including a comparison to the current state in contemporary Polish and to the solutions adopted in the National Corpus of Polish (Pl. Narodowy Korpus Języka Polskiego, hereinafter NKJP; nkjp.pl and nkjp.nlp.ipipan.waw.pl; see also Przepiórkowski et al. 2012). The fifth section of the article presents an experiment using this annotation system in research. The conclusion summarizes the research results and outlines possible future research.

3. Related work

6The problem of the formation of the masculine personal category was of many researchers’ interest; here I present only the chosen works. The changes in Acc pl in the 16^th century investigated Kuraszkiewicz (1961) and in the 17^th century – Rzepka (1975). Works of Kucała (1978), Bajerowa (1964), Ostaszewska (2002), Kępińska (2006, 2011), Klemensiewicz (2009), and – with comparison with other West Slavic languages – Rappaport (2009) present a more general overview of the problem. The Nom pl of contemporary Polish was the subject of the articles of Turska (1953) and Szałkiewicz (2010).

7For the Nom pl forms, we are primarily interested in the change in the repertoire of inflectional endings for nouns denoting animals. The first documentation of this change comes from the turn of the 14^th and 15^th centuries, but in the 17^th century researchers still note the dominance of the old forms (Kucała 1978: 112–117). In the 18^th century, new forms began to dominate, and in the middle of the century the old forms disappeared (Kucała 1978: 118, Bajerowa 1964: 74, Ostaszewska 2002: 83–84). As for nouns denoting persons, new endings appeared sporadically in the 17^th century, and they became far more common in the 18^th century, but they were used mainly as a stylistic device (e.g. emotional or pejorative) and never displaced the old endings (Turska 1953: 136 and 154).

8The change in Acc pl applies primarily to nouns denoting persons. In this case, the use of new endings is documented from the early 16^th century; new endings gained popularity in the 17^th century and became the only ones in the second half of the 18^th century (Bajerowa 1964: 116–117, Rzepka 1975: 48, Kucała 1978: 137 and 145–146). In the 18^th century, old forms sometimes returned in artistic language (Bajerowa: 117).

4. The declension of masculine nouns in the 17^th and 18^th centuries and their morphosyntactic annotation

4.1. Changes in declension patterns of masculine nouns in the 17^th and 18^th centuries

10In modern Polish, masculine nouns in Nom and Acc pl (and Voc, which in plural has forms identical to Nom forms; further in this section I will mention only Nom pl) have different grammatical endings depending (slightly simplifying) on their meaning – masculine personal nouns take one type of endings, while the other masculine nouns take another type of endings.

Table 1. State in modern Polish

11In contrast, in Old Polish (before 1500), the grammatical endings of Nom and Acc pl forms of masculine nouns had no connection with semantics at all.

Table 2. State in Old Polish (up to 1500)

	m
Nom pl	syn-owie ‘sons’; chłop-i ‘peasants’; lw-i ‘lions’; dom-owie ‘houses’
Acc pl	syn-y; chłop-y; lw-y; dom-y

13In the Acc pl forms of personal nouns, the Gen pl ending was taken over, which for the here described types of declension in Middle Polish is -ów (syn-ów ‘son.GEN.PL’), while the Acc pl forms of impersonal nouns remained unchanged (lw-y ‘lion.ACC.PL’, dom-y ‘house.ACC.PL’).

14No linguistic change occurs overnight: for some time, old and new forms coexist, and the inflection of some nouns changes earlier than others. In the Middle Polish era, this was the case with masculine personal and animate impersonal nouns, such as syn ‘son’, chłop ‘peasant’ or lew ‘lion’ (the declension of inanimate nouns had stabilized earlier). Thus, there was a considerable variation in the endings of the Nom and Acc pl forms of animate masculine nouns (both personal and impersonal) –sometimes they were inflected according to the old patterns, and sometimes according to the modern ones. Moreover, for some time the development of some forms went even further and exceeded the limits outlined above: a masculine personal noun could take the ending -y in both Nom and Acc pl, as did impersonal masculine nouns (e.g. syn-y ‘son.ACC=NOM.PL’, chłop-y ‘peasant.ACC=NOM.PL’ like lw-y ‘lion. ACC=NOM.PL’) and impersonal animate masculine nouns could take the Gen ending ‑ów in Acc pl similarly to personal nouns (e.g. lw-ów ‘lion.ACC=GEN.PL’ like syn-ów ‘son. ACC=GEN.PL’). As mentioned above, in Nom pl the ending -owie, which may have coexisted with -i as a variant, also spread (e.g. biskupi-i, biskup-owie ‘bishop.NOM.PL’).

15Therefore, in texts from the 17^th and 18^th centuries, we observe a variation of the endings ‑i/‑owie vs. -y in Nom pl of masculine personal and animate impersonal nouns (e.g. syn-owie vs. syn-y ‘son.NOM.PL’, chłop-i vs. chłop-y ‘peasant.NOM.PL’, lw-owie, lw-i vs. lw-y ‘lion.NOM.PL’) and variant endings -ów vs. -y in Acc pl of masculine personal and animate impersonal nouns (e.g. syn-ów vs. syn-y ‘son.ACC.PL’, chłop-ów vs. chłop-y ‘peasant.ACC.PL’, lw-ów vs. lw-y ‘lion.ACC.PL’), which is rather unrelated to any specific lexemes. It is also not clear in which sentences the choice of one of these alternative endings was related to the meaning of the noun, as it is today, and in which it was caused by other factors. Researchers agree that at some point the users became linguistically aware of the difference between endings and associated them with another meaning of the noun –perhaps the study of a large data set would bring us closer to determining this moment. That is why it is so important to annotate this variation in the corpus of 17^th and 18^th century texts, to enable the study of the origin of this association and the actual emergence of the masculine personal category in the Polish language.

4.2. The tagset of KorBa

16For the above-mentioned reasons, the tagset of the National Corpus of Polish, which was adapted to annotate the KorBa corpus, had to be slightly changed. In the grammatical descriptions of contemporary Polish, there are three masculine (sub)genders, and a gender is always assigned to a specific lexeme (e.g. syn m1, lew m2, dom m3). Such a solution would be useless in the KorBa corpus, because, e.g. if both forms Nom pl syn-owie and syn-y were annotated as m1 gender, it would not be possible to separate them in a corpus query, which is why the subgender of masculine nouns (and their modifiers) is associated in KorBa with a specific grammatical form, not with a lexeme.

17As for the whole noun paradigm, in most grammatical cases (e.g. Gen pl or Dat pl), the inflection endings of masculine nouns are not related to the meaning (personal or impersonal). In these forms in the KorBa corpus they are assigned the gender m as the generalized masculine gender. Only for those grammatical cases where there is a difference which later developed into a distinction between personal and impersonal forms, the marker mamim1 is used for the characteristic endings and m for the non-characteristic ones. For example, in Nom pl the forms syn-owie, lw-i and lw-owie are annotated with the manim1 tag, and syn-y and lw‑y are annotated with the m tag.

Table 4. Comparison of masculine noun paradigms in 17^th–18^th-centuries and contemporary texts, and ways of annotating them in terms of gender in KorBa and NKJP

	17th and 18th centuries		21st century
	form(s)	gender tag in KorBa	form(s)	gender tag in NKJP
Nom pl	syn-owie, chłop-i syn-y, chłop-y lw-owie, lw-i lw-y dom-y	manim1 m manim1 m m	syn-owie, chłop-i lw-y dom-y	m1 m2 m3
Gen pl	syn-ów, chłop-ów lw-ów dom-ów	m	syn-ów, chłop-ów lw-ów dom-ów	m1 m2 m3
Dat pl	syn-om, chłop-om lw-om dom-om	m	syn-om, chłop-om lw-om dom-om	m1 m2 m3
Acc pl	syn-ów, chłop-ów syn-y, chłop-y lw-ów lw-y dom-y	manim1 m manim1 m m	syn-ów, chłop-ów lw-y dom-y	m1 m2 m3
Inst pl	syn-ami, chłop-ami lw-ami dom-ami	m	syn-ami, chłop-ami lw-ami dom-ami	m1 m2 m3
Loc pl	syn-ach, chłop-ach lw-ach dom-ach	m	syn-ach, chłop-ach lw-ach dom-ach	m1 m2 m3
Voc pl	syn-owie, chłop-i syn-y, chłop-y lw-owie, lw-i lw-y dom-y	manim1 m manim1 m m	syn-owie, chłop-i lw-y dom-y	m1 m2 m3

5. Use of the tagset in research

18This study does not pretend to be an exhaustive analysis. Its goal is to verify some of the conclusions appearing in the existing literature (see section 3. Related work) on the subject on a new and larger material basis. On this occasion, I will show how users can apply the KorBa tagset to study the phenomenon of masculine personal category formation.

5.1. Dataset and method

21The experiment proceeded in the following steps:

• • 12 The queries were:
    - [tag="subst:pl:nom:manim1" & !base=".*[cćjlńśźż]" & !base=".*d[źż]" & !base=".*[ (...)

  query for Nom, Acc and Voc pl forms of nouns, that are tagged as manim1 and m12; each query was run twice, limiting the timeframe to 1601–1700 and 1701–1800 using metadata;

• export the results to 12 XLS files;

• divide the forms in each file into four groups: people, animals, things and supernatural beings;

• analyze the results of the queries.

5.2. Experiment and Analysis

5.2.1. Nominative plural

22In all tables, the innovative forms are greyed out.

23The m endings among personal nouns (an innovation, that was almost withdrawn till the 19^th century) are slightly more frequent in the 18^th century than in the 17^th century, but in both centuries, they are in the minority (which is not entirely consistent with Turska’s and Kucała’s statements that they were popular in the 18^th century, see Turska 1953: 154; Kucała 1978: 128).

24Among animate impersonal nouns (whose forms are relatively few) we observe a big change from dominance of the old endings manim1 to the dominance of the innovative endings m; this confirms the results of previous studies (e.g. Kucała 1978: 118; Bajerowa 1964: 74; Ostaszewska 2002: 83–84).

25Among inanimate nouns the new m ending is almost the only one present throughout the whole epoch, which again confirms the results of the previous studies (e.g. Kucała 1978: 110); the most of a few examples of the manim1 ending, coming from the 17^th century, are the form narodowie ‘nation.NOM.PL’, which may be considered as an archaism or as a personification; anyway even in the 17^th century more common is the new, the m form narody.

26Among nouns denoting supernatural beings the old manim1 ending is much more popular than the new m ending during the 17^th century and even more popular in the 18^th century.

5.2.2. Vocative plural

27The vocative plural forms are much less numerous than nominative forms, so the percentages will not be calculated.

28Among personal nouns the old manim1 endings were more popular during the whole timeframe, but the proportion changed in the 18^th century and the m forms became more popular than they were earlier; this may confirm the observation of the previous researchers that from the 18^th century the new m forms became used especially as a stylistic device –the vocative case is more likely to appear in emotionally marked contexts than the nominative case (see Turska 1953: 154).

29Among (very few) inanimate nouns only the new m forms appear.

30The nouns denoting animals and supernatural beings are too few so that some conclusions can be drawn.

5.2.3. Accusative plural

31Among personal nouns in the 17^th century the distribution of the new manim1 and the old m endings are even, and it definitely leans in favor of the new manim1 endings in the 18^th century (which is consistent with previous studies, see Bajerowa 1964: 116–117; Rzepka 1975: 48; Kucała 1978: 145–146).

32Among animate impersonal nouns (though not so numerous) there are a few new manim1 endings, which is also consistent with previous studies (see Kucała 1978: 149; Ostaszewska 2002: 87), but the use of them decreased, although not totally, in the 18^th century.

33All occasional manim1 forms of inanimate nouns are somehow dubious. There are the cases of personification (Wszystkich Talentow zawołam na radę, lit. ‘all.PL.ACC=GEN talent.PL.ACC=GEN call.1.SG.FUT for council.SG.ACC’, ‘I will summon all the talented people to a council’) and the dubious accusative case (the context allows them also to be interpreted as the genitive case, e.g. áby Biszkotow gotowano lit. ‘to cake.PL.GEN/ACC=GEN cook.IMPS’, ‘to bake (some) cakes’). There are also forms of the noun tysiąc ‘thousand’ defining a noun denoting people (e.g. we czterech tysięcy ludzi ‘in four.ACC=GEN.PL thousand.ACC=GEN.PL man.GEN.PL’).

34Among the nouns denoting supernatural beings the old m ending strongly dominates in the 17^th century and this domination seems to be weaker in the 18^th century, but the amount of the examples is too scarce to draw firm conclusions. Nevertheless, these results undermine Rzepka’s statement that these nouns took on almost exclusively masculine personal endings in Acc pl in the 17^th century (see Rzepka 1975: 66).

6. Conclusions and Future Work

35The data from the manually annotated subcorpus of KorBa confirm that the differentiation of the personal and impersonal inflection endings in masculine nouns in Polish was already advanced in the 17^th century and almost finished by the end of the 18^th century. The forms we are interested in in the final part are those where innovative endings appeared and have been preserved to this day, i.e. Nom pl of nouns denoting animals and Acc pl of nouns denoting people. In the 17^th century, in Nom pl of nouns denoting animals, the new forms did not prevail over the old ones (36% of them), while in the Acc pl of nouns denoting people, innovative forms were already dominant, albeit slightly (there were about 52.5%). Thus, the changes in Acc pl were slightly ahead of changes in Nom pl. In the 18^th century, there already was a significant advantage of innovative forms: there were 87.1% of them in Nom pl of nouns denoting animals, and 89.9% in the Acc pl of nouns denoting people. It can, therefore, be assumed that the process ended at the beginning of the 19^th century, which is consistent with the results of previous studies.

36It is interesting how innovative endings entered groups of nouns, from which they later finally withdrew. In Nom pl, nouns denoting people in the 17^th century accounted for 1.9% of the new endings, and in the 18^th century 3.2% –very few, as Bajerowa also notes (1964: 75–76), which, however, does not confirm the observations of Turska and Kucała (Turska 1953: 154; Kucała 1978: 128). As for Acc pl of nouns denoting animals, new endings appeared earlier and were slightly more numerous, and also disappeared earlier: there are 12.9% such forms with new endings in the 17^th century, and only about 7.4% in the 18^th century (cf. Bajerowa 1964: 117–118).

37Some observations can be made about nouns denoting supernatural beings, although recorded examples are relatively scarce. Both in Nom pl and in Acc pl, if we consider the percentage of innovative forms, they are –unsurprisingly– between nouns denoting animals and nouns denoting people. This does not confirm Rzepka’s observation about the dominance of innovative endings in Acc pl of these nouns. It is worth mentioning that also today this group is not uniform when it comes to the choosing the personal or impersonal ending in the discussed inflectional forms.

38The Voc pl forms are relatively the least recorded, but some observations can be made about them. In the 18^th century, for personal nouns, there were relatively many forms with the innovative ending, later withdrawn. This reinforces the thesis that in this century these innovative forms in Voc pl (and, maybe, Nom pl) of nouns denoting persons occurred primarily in emotionally marked texts.

39These studies were easy to perform due to the accuracy of KorBa’s annotation system and the ability to use metadata. The use of a manually annotated corpus gave high reliability of data. In the future, it is worth repeating this research on the entire 25M KorBa corpus if automatic tagging appears to give data of sufficient quality. However, additional semantic tagging will also be needed. Thanks to the analysis of such extensive material, it will be possible to determine the formation time of the masculine personal declension type more accurately and to examine its dependence on style or genre.