1. Nouns
1.1."simple" BG declensions
1.2. CS declensions

2. Adjectives
2.1. "simple" BG forms
2.2. CS forms

3. Pronouns
3.1. personal
3.2. demonstrative
3.3. relative

4. Numerals

5. Verbs
5.1. present tense
5.2. aorist & imperfect
5.3. periphrastic tenses
5.4. other moods
5.5. participles
5.6. irregular verbs

6. Adverbs

7. Prepositions

8. Conjunctions

9. Particles

10. Related Developments

Introduction

The following lines serve two main purposes. One is to provide a contrastive grammar of the language varieties, which we can observe in Pop Punčov Sbornik and other 17th-18th century sources from the Balkan Slavic area. It might be perceived as a dialectological or historical source for the development of modern Bulgarian (BG) or Macedonian (MK), or for observing the adaptation of Church Slavonic (CS) literature to be more accessible to less-educated audience. Such adaptations first started with translations of Damaskēnos Stouditēs' Thēsauros (and other popular texts) in the first half of the 17th century - into variety called "vernacular-based" (Demina et al. 2012), "early modern" (Velčeva 1996:118) or - as it was common by then - "simple" Bulgarian (prostěišii jazykъ Bolgarskii by Krčovski). The other is to provide, alongside the lemma dictionary a comprehensive guide for using the annotation of the corpus provided at this website, either using the Search tool or token browser.

The Sbornik was written in 1796 in Mokreš, a village in Northwest Bulgaria. Punčo, its author, was likely a native of the village too. A fixed linguistic norm did not exist in the area, nor a systematic, mass education system. Writers often mechanically just copied older texts without really reflecting, whether they understand the contents or not. Punčo may have not only undestood, but also dared to edit them into what he thought would be comprehensible to his neighbors - still not a natural practice in the literature of the time and area. Still, the language of the Sbornik is not based on a systematic grammar with closed set of norms, like the ones created only later by professional grammarians - like Neophyte of Rila (1835), Xristaki Pavlovič (1836) or Ivan Bogorov (1844; followed e.g. by Todor Xrulev in his works). His set is eclectic, following the text models from Church Slavonic, school of Josif Bradati, and likely some damaskini sources as well. For this reason, it is not always clear, which forms were productive in Punčo's dialect, and which were foreign or archaic, but still comprehensible enough to escape Punčo's attention. Some of the texts were likely left (or composed) in (a sort of) Church Slavonic deliberately. Punčo's idiolect, or, rather, the doculect of the Sbornik shows itself first on the prevalent tendencies in morphosyntactic realm, as well as in comparison with other text from the given time and area.

The language of the Sbornik has been first described by Conev (1923:284f.) as "vernacular with influences from literary Russian and Serbian" (naroden sъs ruski i srъbski knižovni primesi), based on a dialect from the Northwest. Indeed, a Russian edition of Prolog from 1735 has been used as a source for multiple chapters in the Sbornik (Miltenova 2019). On the other hand, Šaur (1970:6) demonstrated, that many features previously considered foreign influences indeed were productive in Vidin-Lom dialects, like the distinctive 3pl.aor ending (ibid. 49). Unlike both damaskini and East Slavic editors, Punčo uses the l-participle in a narrative mood (with degrees of evidentiality marked by the presence of auxiliary) and the adjectival pronoun/numeral edin as a marker of indefiniteness (Sonnenhauser 2015:47). The use of a dependent form on both nouns and articles of masc. and fem. gender (Šaur 1970:39) is still attested in dialects on the Bulgaro-Serbian border (Stojkov 2002:165). Orthographically, Punčo followed examples from the area too. He used ь as the final jer; in this respect, he differs from Bradati, preferring ъ, more common in East Slavic redactions. We also find ligatures, graphic variants of t, letters like џ (instead of the digraph дж), which are not usual in East Slavic redactions, flowing to Bulgaria at the time by print. The same models were also preferred in earlier damaskini and apocryphal collections from the Bulgarian East. The language of the Sbornik also adopts the complex (even if chaotically applied) accentuation of damaskini and Church Slavonic.

As this description serves as a reference to the digital edition of the Sbornik and other texts, it may sometimes deviate from usual methods. The annotation of the text - PoS- and UD-tags - aims to capture the morphological variation of the original, without necessarily interpreting the function and meaning of the morphem. Each word has been coded using the MultextEast scheme designed for a corpus of historical Balkan Slavic texts, released earlier at Clarin repository, as well as by syntactic relations according to the Universal Dependencies scheme. The method of their application was adjusted to the specific needs of the project - the ability to reflect both older (CS) and more recent ("simple" and modern BG/MK) varieties. We will thus give an introduction to the morphology of both of these stages of Balkan Slavic development.

Citations from the corpus are mostly direct links to Syntax Browser showing separate sentences, showing also basic information about the source. Citations from Punčo's Sbornik indicate the chapter according to the structure of the corpus.

1. Nouns

1.1. "simple" BG declensions
1.1.1. masculine nouns
1.1.2. "new i-stems"
1.1.3. feminine nouns
1.1.4. old i-stems
1.1.5. neuter nouns

1.2. CS declensions
1.2.1. o-stems
1.2.2. jo-stems
1.2.3. ā-stems
1.2.4. jā-stems
1.2.5. i-stems
1.2.6. u-stems
1.2.7. r-stems
1.2.8. ū-stems
1.2.9. n-stems
1.2.10. nt-stems
1.2.11. s-stems
1.2.12. other C-stems

Nominal morphology shows some of the most characteristic contrasts between the modern Balkan Slavic varieties on the one hand, and its predecessors and the rest of Slavic on the other. Old Church Slavonic (OCS) has inherited a complex system of nominal inflection with 7 case rows (BG padež; with up to 5 distinctions per gender/number paradigm), 3 genders (rod; f/m/n), 3 numbers (čislo; sg/dl/pl) and marking of m.sg animacy (oduševnost). All of these are marked on a single ending following the stem. The stem class (osnova) is indicated in the corpus as a property of the lemma. The number of stem classes, traditionally classified by the shape of Proto-Slavic sg.nom ending, added another aspect to the morphological variety:

	
	stem class	genders		examples
	o-stem		m/n	 	gradъ 'city', sg.gen grada, sg.loc gradě
	jo-stem		m/n		konь 'horse', sg.gen konja, sg.loc koni
	ā-stem		f		žena 'woman', sg.gen ženy, sg.dat/loc ženě
	jā-stem		f		duša 'soul', sg.gen dušę, sg.dat/loc duši
	i-stem		f/m		kostь 'bone', sg.gen/dat/loc kosti
	u-stem		m		synъ 'son', sg.gen/loc synu
	r-stem		f		dьšti 'daughter', sg.gen dьštere
	ū-stem		f		crьky 'church', sg.gen crьkve
	n-stem		m/n		plemę 'tribe', sg.gen plemene
	nt-stem		n		telę 'calf', sg.gen telęte
	s-stem		n		nebo 'heaven', sg.gen nebese
	

On the other hand, modern BG/MK have discarded most case distinctions, marking only the gender and number in the case ending. Relations formerly indicated by the case variety are marked analytically. In most varieties, only a distinctive vocative form remains (zvatelna forma; e.g. Ivane). Unlike CS and the rest of Slavic, definiteness is marked too by an additional suffix (article, členna forma; mostly handled as a separate token in our corpus) marked for gender and number too. "Simple" BG texts represent a late stage in the process of change, where the most nouns are unmarked for case:

	
	pustynju ostavlьši kь otčьstvu vьzvrati se (Vuk.1536)
	'return to your fatherland as you leave the desert'
	
	da ostavišь pustinja+ta i idi si pakь nazadь na tvoe otčьstvo (Tixon.d.)
	'(you should) leave the desert'
	

Here, oblique cases in the CS text were edited by "simple" BG editors into forms based on old nominative - a "straight", case-unmarked, or "common" form (obšta forma). Standard MK and some BG dialects sometimes also show one distinctive oblique case, called "dependent form" (Lunt 1952), "agglomerative" (Stojkov 2002) or "general oblique" (Sobolev 1991) case. This appears on m.sg (rarely also f.sg) proper names, kinship terms or animate nouns following prepositions or used as direct objects of sentences. In dialects, it is often marked on both case ending and article, e.g. on vide čoveka+toga 'he saw the man' (Stojkov 2002:165). The marker etymologically continues the old gen/acc ending. The marking is optional, and in most varieties considered an archaism. However, such dependent forms of m.sg animate nouns are used regularily in most of our sources from the 17th-18th century. An exception is NBKM 1423 by Petăr Carski, reflecting a Paulician variety. In his text, the dependent form can be seen only on some names and theonyms, excluding the adjective:

	
	ze da dodeva tozi nevernik Sveti Antuna (NBKM 1423)
	'the heretic began to annoy Saint Anthony again'
	

Damaskini sources also show traces of the f.sg.dep marking. It is not very systematic: besides the unmarked case, they also show the old f.sg.acc ā-stem ending -u, used also in most CS redactions (usually reflecting CS original: e.g. našu ništetu 'our misery', Tixon.d.; but cf. original) and the ending -ь (e.g. dšь+ta 'the soul'; Ljub.d.). The latter likely denotes the middle vowel /ă/ (cf. below §10.6), which is attested in this function in the Gabrovo region (Stojkov 2002:110). The use in damaskini differs between individual editors (cf. Mladenova 2007:306). For example, the editor of Trojan d. sometimes used the OCS ending -ǫ as a f.sg.dep alongside other markers (e.g. tьzi čistǫ i xva golubica 'that pure dove of Christ', Ivanova 1967:77). In Punčo's Sbornik, the marking of f.sg.dep dependent forms is more systematic (in about half of the expected instances), regularily using the -u ending:

	
	poče avramь da ljubi robinju+tu (PPS 008)
	'Abraham began to love the slave'
	

It is notable, that the function of dependent form is not equal to that of old accusative. As in modern-day varieties showing inflection, it appears at most syntactic positions excluding the subject and predicate. But besides the syntactic function, the use of the form seems to be restricted to m/f.sg nouns denoting objects high in the animacy scale - not much differently from that of still-productive vocative. The apparent restrictions also resemble the differential object marking known from Balkan Romance varieties (cf. Makarova & Winistörfer 2020, Onea & Mardale 2020). Modern varieties also distinguish only two numbers (sg/pl), although the masc.dl.nom form has acquired a new function - marking nouns after numerals ("count form", brojna forma), which usually appears on nouns which are not animate. It is attested in our corpus too:

	
	i širokь e *n* lakta (PPS 013)
	'and he is 50 elbows wide'
	

In both present-day BG/MK standards and texts from pre-standardized era, morphological differences between nouns can be observed on distinctive plural forms (e.g. kon, pl. kone vs. sin, pl. sinove), and partly also on palatality of m.sg articles (e.g. konja+t vs. sinъ+t). The palatality can be often observed on dependent and dative forms too. Some of the differences reflect old stem-based paradigms, fading over time due to case syncretism and gender-based levellings. For example, the distinction between ā- and jā-stems seems to be fading in damaskini, where the possessor forms always use -i (e.g. ā-stem: mošti stěi petki 'relics of St.Petka', jā-stem: ikona prstěi bci 'icon of Most Holy Mother of God'; Tixon.d.). Other stems arise from newly appearing factors like semantics and word length. These include "ethno-stems", denoting confessional or ethnic groups of people with a sg. suffix -in (cf. Mirčev 2000:56 for emergence); monosyllabic masculina, using a generalized u-stem pl.nom ending; and also loanwords, among which new paradigms can be observed among masculina with the phoneme cluster -ii in stem auslaut (new i-stem), as well as abstract nouns, where no inflection is observable at all (indicated as uninflected, e.g. kabaxat 'misdeeds').

	
	stem class	genders		examples
	o-stem		m/n		človekь 'human', sg.dep človeka, pl človeci
	jo-stem	 	m/n		konь 'horse', sg.dep konja, pl kone
	ā/jā-stem 	f		žena 'woman', pl ženi
	i-stem	 	f/m		kostь 'bone', pl kosti
	n-stem	 	n		pleme 'tribe', pl plemena
	nt-stem		n		tele 'calf', pl teleta
	s-stem		n		nebe 'heaven', pl nebesa
	monosyllabic	m		sinь 'son', sg.dep sina, pl sinove
	ethno-stem	m		xristianinь 'Christian', sg.dep xristianina, pl xristiane
	new i-stem	m		axčii 'cook', sg.dep/dat axčiju, pl axčie
	

The number of nominal endings is thus smaller in "simple" BG than in CS, but the endings from the older system may still appear in fixed phrases, citations and unedited transcripts. Sometimes a productive ending is used for an archaic function, e.g. -i in zmii could be a f.pl ending, but in reče zmii sanьtailь 'Satanael said to the snake/dragon' (PPS 013) it indicates a jā-stem f.sg.dat; the context is thus also relevant to determine the case row and number. On the other hand, some case tags may denote various case rows (for the one gender/stem and number) due to syncretism, e.g. -a in bga can be both m.sg.gen and acc. Such homographic forms (given in square brackets below) are not tagged distinctively - the first (highest) form in the paradigm is used. Thus, bga would be always tagged "Nmsgy", even if it fulfills old accusative functions, like direct objects (e.g. dali ljubi bga 'if he loved God'; PPS 008).

The word category of nouns is indicated by an N in the PoS-tag. As the case row is the property with most options in CS, we will focus our explanations of the tag from its perspective. The tag used in the corpus indicates four properties of each noun: gender (f/m/n), number (s/p/d), case (n/g/d/a/v/l/i/o) and animacy (y/n). For example:

	
	člvekь - PoS-tag: Nmsny, lemma: človek (o-stem indicated in the dictionary)
	N	noun		property of lemma
	m	masculine	property of stem
	s	singular	property of ending
	n	nominative	property of ending
	y	animate		property of lemma
	

1.1. "simple" BG declensions

1.1.1. masculine nouns
1.1.2. "new i-stems"
1.1.3. feminine nouns
1.1.4. old i-stems
1.1.5. neuter nouns

In the first section we will introduce forms typical of nouns in two "simple" Bulgarian varieties - that of pop Punčo, as attested in his Sbornik, and of the togazi-editor, whose works are preserved in early damaskini. The marking of CS-only forms described below is complementary to the system used in texts of "simple" varieties.

1.1.1. masculine nouns

Case-unmarked form (n) of most masculines in singular shows a jer (ъ or ь, according to orthography in damaskini sources) or no graphic ending (as today, or if the last letter is written in a superscript) after the stem. The most common plural form is based on old o- and i-stems, -i: človekь 'human', pl človeci. Masculines with a monosyllabic stem (or sg.nom form) show pl -ove: sin 'son', pl sinove. Some nouns show the old jo-stem pl.acc ending -e in plural: carь 'king', pl care, which is also often, but not always used with "ethno-stems" (pl both xristiane and xristiani are attested in the Sbornik). The case variation shows dependent (marked mostly g: človeka, carja, sina, xristianina) and dative forms (marked d: človeku, carju, sinu, xristianinu). As ę (small jus) is often used for the sequence /ja/ (cf. §10.5), it may also denote sg.dep form of jo-stem masculines (e.g. pred crę+tokъ 'in front of the king'; Berl.d.).
Case tag o denotes a definiteness marker, arising from a hiatus vowel preceding a now-omitted demonstrative root: e.g. stlьpo 'the pillar' (PPS 020; < *stl̥po+t < *stlьpь+tь). A tag for a post-positional article (p_nom) is added to the syntactic annotation of the token as an extension. When the hiatus vowel is written as -a, the form is handled as ambiguous and marked as a "g" in the PoS tag. If the context hints at the function (e.g. on subjects like diavola discussed below in §2; Ned.1856), a secondary PoS tag and an UD tag extension "p_nom" is added.
For handling of old u-stems, cf. below §1.2.6. The count form is tagged as a dual. Vocatives (v) share some endings with other cases, and are tagged according to context.

	
	1.1.1. masc.nouns (Nm)
		sg		pl		dl
	n	-ь		-i/-e/-ove	-a
	g	-a
	d	-u
	v	-e/-u
	o	-o
	

1.1.2. "new i-stems"

Stems of many Turkish loanwords (mostly nomina agentis) and biblical proper names end in an -i. They show different handling across sources, also with different degree of regularity. The differences can be observed well in proper names of this class, like Georgi. In CS, an additional -i is regularily used in sg.nom (e.g. Georgii že někto ot nixь 'Georgie, one of them'; Vuk.1536). Tixon.d. adds commonly an -e, used also for vocative (georgie ime+to mu 'George [was] his name'; here). Punčo uses -a (e.g. ime mu beše georgia 'his name was George'; PPS 020), which is used also in non-proper nouns of this class (edin mu beše axčia 'one of them was [served] him as a cook'; PPS 022). In Punčo's texts this stem class seems to be influenced by the two-way declension of jā-stems, as the ending -u is used both for the dep.form (izvede axčiju 'he led out the cook'; PPS 022) and the dative (kakvo xorotuva staja paraskeѵa georgiju 'what St.Parascheva says to George'; PPS 020). On the other hand, the use of final -a may be an expression of definiteness: some Western dialects regularily add an -a to m.sg proper names (e.g. iovo 'Job' in NBKM 722 here). Still, Punčo himself chooses freely between -a (e.g. here) and -i (e.g. here) on some proper names of this class in subject positions.

	
	1.1.2. new i-stem nouns (Nm)
		sg		pl
	n	-i		-i
	d	-iu
	o	-ia
	v	-e/-o
	

1.1.3. feminine nouns

As a rule, under grammatical "feminines" we class nouns, which govern feminine articles. Thus, these include old ā/jā-stem nouns with the sg case-unmarked form (n) ending -a, e.g. žena 'woman', but also nouns denoting "male" objects like bašta 'father' (sg.def žena+ta, bašta+ta). As ę in most of our sources represents the sequence /ja/ (e.g. ispolni se zemlę sasь nepravdu 'the Earth was filled by injustice'; PPS 025), it is tagged in f.sg nouns with an "n" too, unless not indicated otherwise in the description of the text. Plural ending for all feminines is -i (e.g. ženi, bašti, radosti).
Oblique cases are marked differently across the sources. Endings -e and -i for old sg.gen and sg.dat are handled according to CS rules (cf. below §1.2.3/4), i.e. based on palatality of the stem; their productivity is questionable. The old ā/jā-stem sg.acc, which may have worked as the dependent form in the vernacular underlying early damaskini, is realized as -u (e.g. ženu) or -ь (or -ъ; e.g. ženь) in Tixon.d. and other sources. In some sources, also the OCS ending -ǫ appears (e.g. ženǫ; cf. §10.6). In our corpus, -u and -ǫ do not appear along each other as f.sg.dep (or f.sg.acc) markers (like in Trojan d.), so they use the same PoS tag a. In this way, they can be compared to CS f.sg.acc endings of Resava/Tarnovo redactions. The ending -ь, which is specific for damaskini, is tagged o.
Ending -e is mostly met in former duals like ruce 'hands' (e.g. here; tagged "Nfdnn"). The ending rarely appears also elsewhere, lexicalized for particular lemmas (e.g. dъštere+te crevi 'daughters of the King'; Berl.d.; tagged "Nfpny" as a jā-stem archaism; cf. below §1.2.4). Vocatives (v) have a specific ending and do not need analysis of the syntactic context.

	
	1.1.3. fem.nouns (Nf)
		sg		pl		dl
	n	-a		-i		-e
	a	-u/-ǫ
	v	-o
	o	-ь
	

1.1.4. old i-stems

Old i-stems show no overtly marked dep.forms - as they neither did in OCS. In singular they show the same ending as old o/jo-stems, i.e. silent jer or no graphic marking. Regular plural form is -i, e.g. tri puti 'three times' (PPS 010). This class includes many abstract nouns with no plural forms, like e.g. rados+ta tvoja 'your joy' (PPS 017). Feminine nouns of this stem class showed less morphological variety than e.g. ā-stems: the f.sg.acc did not differ from the unmarked (sg.nom) form, sg.gen/dat/loc had all the same form -i. In texts of our corpus, there does not seem to be any dependent or vocative form productive.
Masculine i-stems mostly adopt inflection of other stems according to the word size (e.g. črьvove 'worms'; PPS 006) or other factors, described in the lemma dictionary.

	
	1.1.4. i-stem fem.nouns (Nf)
		sg		pl
	n	-ь		-i
	

1.1.5. neuter nouns

Neuter nouns similarly show only one distinctive plural ending. There seems to be no distinctive dependent form (there was also no sg.nom/acc distinction in OCS too), nor a vocative. The unmarked sg ending is -e (old jo-, n-, nt- and some s-stems, e.g. loze 'vineyard', ime 'name', tele 'calf', nebe 'heaven') or -o (old o- and some s-stems, e.g. xoro, slovo). Plurals show an -a ending, which is added to the stem directly (xora), or following an extension in old consonantal stems (imena, teleta, nebesa, slovesa). Although the nt-stem ending is used also with some jo-stem neuters (lozeta), in our sources the ending -a seems to be still preferred (e.g. lozia prorodixu 'the vineyards grew with abundance'; PPS 023). Plural ending -i reflects lexicalized duals like oči 'eyes' (e.g. here; tagged "Nndnn").

	
	1.1.5. neut.nouns (Nn)
		sg		pl		dl
	n	-e/-o		-a		-i
				-ena
				-esa
				-eta
	

1.2. CS declensions

1.2.1. o-stems
1.2.2. jo-stems
1.2.3. ā-stems
1.2.4. jā-stems
1.2.5. i-stems
1.2.6. u-stems
1.2.7. r-stems
1.2.8. ū-stems
1.2.9. n-stems
1.2.10. nt-stems
1.2.11. s-stems
1.2.12. other C-stems

We have already seen that the complexity of CS nominal inflection reflects both the number of case rows and stem classes. Some of the endings are still used in "simple" varieties. Besides the productive ones like dependent forms, old case variety is still seen on adverbial instrumentals (e.g. denem 'all over the day'), lexicalized (vkъšti 'home') and fixed phrases (veki vekov 'forever'), untranslated instances (e.g. the Psalm 44:12 here) and similar material. The marking is designed to be usable for any CS variety - from OCS, through medieval Resava and Tarnovo redactions, to the recent standard used by Orthodox churches.
As a general principle, the paradigms for CS are followed, if a noun shows a form, which is not seen among the "simple" ones above. The paradigms themselves are based on OCS. Throughout the ages, paradigms were, of course, affected by many phonetic, analogical and orthographic developments. Changes leading to more homography and case syncretism will be shortly discussed for each stem class, as well as below in §10.

1.2.1. o-stems

Old o-/jo-stems show distinctions according to gender and animacy. Masculine o-stems have a m.sg.nom ending -ъ (taken from the u-stem class), a vocative as well as a different pl.nom and acc endings. Animate masculine o-stems have a different m.sg.acc ending (-a) than inanimate (-ъ), which both are homographic (and likely were also homophonic) with other cases. Animacy is handled in the corpus as a property of the lemma, so words denoting humans and other speaking creatures (e.g. the speaking camel in PPS 007) are marked as "animate". Also horses (CS konь) are marked as animate, although the texts oscillate often in this aspect, not only in Balkans (cf. Bratishenko 2003:95). Besides the loss of weak jers (§10.1) there were not many phonetic changes affecting o-/jo-stem singular paradigms. As already mentioned, sg.inst -om may still appear in damaskini in an adverbial function (e.g. redom po sički světь 'all over the world'; Berl.d.). Sg.loc rarely appears in later sources with the ending -e (due to §10.4; e.g. vь stlьpe 'in the pillar'; ODNB 1/112), which can be distinguished from the vocative only by context.
An extension -ov- starts to appear in some plural forms of monosyllabic stems (e.g. pl.gen gradovь 'cities'; NBKM 326) from the 14th century onwards, under the influence of u-stem class (cf. Mirčev 1978:162). However, also the older u-stem pl.loc ending -oxъ appears (e.g. vь epivatox 'in Epibates'; Vuk.1536; cf. Knoll 2019:106). The pl.acc/inst ending -y later may show irregularities (due to §10.3: e.g. mitropoliti [...] postavi 'he appointed metropolites'; Vuk.1536). In later sources, we can see sometimes the pl.inst ending -mi (e.g. pride sь korabmi 'he came with ships'; NBKM 667), likely taken over from u-stems too.

	
	1.2.1. o-stem masc.nouns (Nm)
		sg		pl		dl
	n	-ъ		-i		-a
	g	-a		-ъ		-u
	d	-u		-omъ		-oma
	a	[-a/-ъ]		-y		[-a]
	v	-e
	l	-ě		-ěxъ		[-u]
	i	-omь		[-y]		[-oma]

	masc. o-stems in later sources
		sg		pl	
	n	-ь		-i
	g	-a		-ь/-ovь
	d	-u		-omь/-ovomь
	a	[-a/-ь]		-i
	v	-e
	l	-e		-oxь/-ověxь
	i	-omь		[-i]/-mi

	o-stem neut.nouns (Nn)
		sg		pl		dl
	n	-o		-a		-ě
	g	-a		-ъ		-u
	d	-u		-omъ		-oma
	a	[-o]		[-a]		[-ě]				
	l	-ě		-ěxъ		[-u]
	i	-omь		-y		[-oma]
	

1.2.2. jo-stems

The jo-stem class is characterized by palatalized auslaut of the stem. This leads to shifts of the following ending vowel. The pl.acc ending -e (reflecting §10.5) has been later generalized; traces of it can be seen already in later CS (e.g. izide izь kona medna sion crь i *l* voe xrabri 'King Sion and 30 brave warriors went out of the horse'; NBKM 667). We distinguish pl.nom and acc endings only in CS texts. Punčo uses plurals cari and carove (PPS 003) for 'kings', but both kone (PPS 022) and koni (ibid.) for 'horses', no matter the syntax. Common jo-stem lemmas like carь and voinь also often show influence of other paradigms, especially i-stems. The ending -ě is used as sg.gen/acc likely denoting sequence /ja/ (e.g. nebsnago že crě 'of the Heavenly King'; Zogr.107). Nomina agentis with suffix -tel show commonly pl.gen ending -ei (e.g. ot roditelei 'by parents'; Zogr.107). As pl.acc, both -e (cf. §10.5) and -ie (an i-stem pl.nom, but used on objects too: e.g. poraziše vьse crie xanansti 'they defeated all the kings of Canaan'; NBKM 326) appear in later texts. Only Zogr.107 prefers -ie as sg.nom form (e.g. here). Jo-stem m.sg.dat and voc -u can be distinguished only by context.

	
	1.2.2. jo-stem masc.nouns (Nm)
		sg		pl		dl
	n	-ь		-i		-a
	g	-a		-ь/-ei		-u
	d	-u		-emъ		-ema
	a	[-a]		-ę		[-a]
	v	-u				
	l	-i		-ixъ		[-u]
	i	-emь		[-i]		[-ema]

	jo-stem neut.nouns (Nn)
		sg		pl		dl
	n	-e		-a		-i
	g	-a		-ь/-ii		-u
	d	-u		-emъ		-ema
	a	[-e]		-a		[-i]
	l	-i		-ixъ		[-u]
	i	-emь		-i		[-ema]
	
	lemma car 'king' in later sources (Nm)
		sg		pl
	n	crь		cri/crove/[crie]
	g	cra/crě		crei/crь
	d	cru		cremь
	a	[cra/crě]	crie
	v	cru		
	l	cri		crexь
	i	cremь (?)	crmi (?)
	

Jo-stems according to Derksen (all are, of course, to be understood as *): agnьcь, belenь 'henbane', bičь, bodъlь, bričь 'razor', carь (could be also a i-stem), čaroděi, čemerь 'hellebore', črěnъ 'handle', datelь 'giver, donor', dъždь, elenь, ězъ 'fish weir', gnoi, graždъ 'stable', grьměždь 'gramiae', gъbežь 'bend, joint', inii 'hoar-frost, rime', ključь, konь, konьcь, korьcь 'measures' (100 l/100 kg), košь, kotьcь, krajь, kъšь 'lot, fate', kyjь, lemešь 'plough', loi, lučь, mečь, medvědь (an i-stem according to Mirčev), měsęcь, mladenьcь, molь, mǫžь, netopyrь, nožь, netii 'nephew', ognь (sometimes an i-stem), otьcь, palьcь, pokoi, prijatelь, pьnь, stražь, strъi, strъžьnь 'core', stьblь, slavii, srьšenь, šurь 'brother-in-law', tatь 'thief', uboi 'murder', ulii 'hive', voi, voinь, znoi 'heat', želudь, žrьcь, zmii, žrěbii. The class also includes all the nomina agentis with -telь and -arь endings.

1.2.3. ā-stems

1.2.4. jā-stems

Old ā-/jā-stems denote grammatically feminine nouns. They differ only according to palatality of stem auslaut: e.g. žena is a hard ā-stem, but pustynja 'desert' or duša 'soul' are soft jā-stems. Semantically "male" nouns like voivoda do not use different endings. However, the paradigms of this stem class were affected by many changes. Sg.gen endings were -y (affected by §10.3) and -ę (by §10.5), while sg.dat/loc were -ě (§10.4) and -i. For this reason, ā-stems may show sg.gen -i (e.g. prosexa ju ženi sebě; ODNB 1/112) and sg.dat/loc -e (o ednoi žene 'because of one woman'; ODNB 1/112) in later sources. Inversely, jā-stems would use -e as sg.gen (dše mojee ozlobljenie 'evils of my soul'; Vuk.1536) and -i for dat/loc, as before. In such texts, also the jā-stem sg.voc can be distinguished only by context. It is also possible, that -i was in a certain moment (and area) generalized as a single sg.gen/dat/loc ending for both stem classes, as hinted by instances like otsěče ženě svoei glavu igiludi crci [sestri svoei] 'he cut off the head of his wife Queen Giluda [his sister]' (NBKM 326).
When a -ě is found on a soft stem, it is analyzed as the sequence /ja/, and thus annotated as a case-unmarked form: e.g. mlstině očistjuva grěxove 'alms wash away the sins' (Berl.d). Such examples can be found in CS texts too (e.g. jakože pustyně 'as the desert'; Zogr.107).
Also other case endings were affected by multiple phonetic changes. While sg.acc -ǫ was shared by both stem classes, its reflexes (cf. §10.6) differed dialectally, as already mentioned above. The reflex /ă/ was sometimes not distinguishable from the sg.nom ending (due to §10.7); the both could then appear confused in the script (e.g. togda i magdona izvědoše emu 'then they brought Magdona to him'; NBKM 667). Sg.inst was affected by multiple developments (§10.6/8/12), which led to confusion with sg.acc (rodi crca glavnoju ogьnnoju 'the queen gave birth to a burning firebrand'; ODNB 1/112).
Plural forms were less ravaged by phonetic shifts. In both paradigms, pl.nom and acc showed syncretism already in OCS. As they are identic to sg.gen, they were also affected by the same changes: thus, we can see pl.nom/acc -i for ā-stems (e.g. iže běxu im ženi 'who were their women'; NBKM 326), and -e for jā-stems (izvěstvoujušte i svoje dše otdati 'promising to give even their souls'; Vuk.1536). Although -i was later generalized in most dialects as the case-unmarked plural form, sometimes we find also -e on ā-stems, reflecting another development (naredi vьse voiske 'he ordered all the troops'; NBKM 326). The jā-stem pl.loc ending is sometimes written with a jat, which again denotes a /ja/ then (o domoxъ i rabyněx 'about houses and handmaids'; Zogr.107).

	
	1.2.3. ā-stem fem.nouns (Nf)
		sg		pl		dl
	n	-a		-y		-ě
	g	-y		-ъ		-u
	d	-ě		-amъ		-ama
	a	-ǫ		[-y]		[-ě]
	v	-o
	l	[-ě]		-axъ		[-u]		
	i	-ojǫ		-ami		[-ama]
	
	ā-stems in later sources
		sg		pl
	n	-a		-i
	g	-i		-ь
	d	-e		-amь
	a	-u		-i
	l	[-e]		-axь
	i	-oju		-ami

	1.2.4. jā-stem fem.nouns (Nf)
		sg		pl		dl
	n	-a		-ę		-i
	g	-ę		-ь		-u
	d	-i		-amъ		-ama
	a	-ǫ		[-ę]		[-i]
	v	-e
	l	[-i]		-axъ		[-u]
	i	-ejǫ		-ami		[-ama]

	jā-stems in later sources
		sg		pl
	n	-a		-e
	g	-e		-ь
	d	-i		-amь
	a	-u		-e
	l	[-i]		-axь
	i	-eju		-ami
	

Jā-stems according to Derksen: burja, cěvьnica 'lyre', dažda 'distribution', dalja 'distance', dirja 'crack', dolja 'fate', duša, ęźa 'disease', korica, košerja, koža, kǫšta, luča 'beam', luža, lъža, lьźa 'possibility', mežda, mrěža, moča, mъšica 'mosquito, myšьca 'muscle', nozdrja, nǫžda, onušta 'sandal', ovьca, polьźa, struja, stьźa, svěšta, věja 'branch', volja, vonja, zarja, zemlja, zorę, želja, žažda, . The class also encompasses nouns with feminine suffix -ica (e.g. pъtica 'bird'). Added should be also common stems ending in -št-/-žd- (< *-tj-/-dj-) like bašta 'father', nadežda 'hope'. Although they show oblique cases only rarely, they seem to be handled as hard ā-stems in damaskini sources: e.g. reče bašte si 'he said to his father' (PPS 022).

1.2.5. i-stems

As already mentioned, i-stems showed a lot of case syncretism already in OCS. In singular, they show only two oblique forms: one for sg.gen/dat/loc, other for sg.inst, which also differs between genders (feminine -ьjǫ, masculine -ьmь or -emь due to §10.2). Genders also differ in pl.nom (fem. -i, masc. -ie). OCS sources show various pl.gen endings like -ьi, -ii or ei (due to §10.13; cf. Lunt 2001:73). In our Balkan Slavic sources, -ei is used consequently (e.g. pečalei oblakъ 'cloud of sorrows'; Zogr.107), -ii is used only by Rostovski (e.g. here). There are irregularities especially when handling masculines. For example, vь različnyje zvěry 'into various beasts' (Vuk.1536) shows formally an o-stem ending, but this may be a mistake based on phonetic levelling (§10.3); however, Zogr.107 shows źvěrę, formally a jo-stem pl.acc (here).

	
	1.2.5. i-stem fem.nouns (Nf)
		sg		pl		dl
	n	-ь		-i		-i
	g	-i		-ii/-ei		-ьju
	d	[-i]		-ьmъ		-ьma	
	a	[-ь]		[-i]		[-i]
	l	[-i]		-ьxъ		[-ьju]
	i	-ьjǫ		-ьmi		[-ьma]
	
	i-stem masc.nouns (Nm)
		sg		pl		dl
	n	-ь		-ie		-i
	g	-i		-ii/-ei		-ьju
	d	[-i]		-ьmъ		-ьma
	a	[-ь]		-i		[-i]
	l	[-i]		-ьxъ		[-ьju]
	i	-ьmь/-emь	-ьmi		[-ьma]
	

Masc. i-stems according to Derksen: črъvъ, drъkolь, golǫbь, gostь, gvozdь, nogъtь, ognь (also jo-stem), lakъtь, ǫglь, pǫtь, stěnь, zętь, źvěrь. Mirčev adds: bolь, grъtanь, medvědь, tьstь, oušidь 'runaway'.

1.2.6. u-stems

Old u-stems represent a small, but influental group of words. As we have already mentioned, many of the endings were taken over by the o-/jo-stems in both written and spoken varieties. The pl.nom ending -ove was ultimately adapted by all monosyllabic masc. nouns. However, the (likely gender-based) mixing of paradigms was two-sided. The OCS sg.dat ending -ovi, sometimes used in OCS with other stems too ("personal dative" in Lunt 2001:56), is not attested in our sources; the -u is used instead (e.g. tvoemu edinorodnomu poslědovavšǫǫ snu 'having followed Your single-born Son'; Zogr.107). In CS texts, these instances are annotated with two PoS tags ("Nmsdy" as the primary one). The ending -a is abundantly attested in later ("simple") sources as the dependent form (e.g. vidě sina si 'he saw his son'; Berl.d.).
As already mentioned above, the extension -ov- appears later in masc. monosyllabic stems of other classes too. All CS u-stems are, indeed, monosyllabic. The ending -ovi appears in both later texts and dialects as a pl.nom ending (cf. Mirčev 2000:57). In our corpus, it is attested in a role of a pl.acc (i.e. as a direct object: obretoše domovi svoi pusty 'they found their houses deserted'; ODNB 1/112), thus likely reflecting an following an extended *-ov-y (with §10.3 at work). Extended endings can be seen in other cases too, e.g. pl.loc in vъ domověxь svoix 'in their houses' (ODNB 1/112), which appears alongside older forms like domoxъ (e.g. here); the latter are already common in OCS (cf. Lunt 2001:57).

	
	1.2.6. u-stem masc.nouns (Nm)
		sg		pl		dl
	n	-ъ		-ove		-y
	g	-u		-ovъ		-ovu
	d	-ovi		-ъmъ/-omъ	-ъma
	a	[-ъ]		-y		[-y]
	v	-e
	l	[-u]		-ъxъ/-oxъ	[-ovu]
	i	-ъmь		-ъmi		[-ъma]
	
	u-stems in later sources
		sg		pl
	n	-ь		-ove
	g	-a (?)		-ovь
	d	-u		-omъ/-ovomь
	a	-ь/-a		-ovi
	l	[-u]		-oxъ/-ověxь
	i	-omь		-mi
	

U-stems according to Derksen: borъ, darъ, domъ, medъ 'honey, mead', olъ, synъ, vrъxъ; Lunt adds: činъ, dlьgъ, dǫbъ, jadъ, glasъ, mirъ, polъ, rędъ, rodъ, sanъ 'rank', stanъ, volъ. We may also add stěgъ 'banner'.

consonant stems

1.2.7. r-stems

1.2.8. ū-stems

Consonant stems are archaic phenomena, showing different stems in oblique cases. Historically, their sg.nom forms were shortened by pre-Proto-Slavic loss of some consonants in word-final positions (§10.9). They occur in CS quite regularily, though only for a limited number of lexemes. Although the extensions differ, the distribution of endings (e.g. sg.gen -e) is mostly shared.
There are two feminine consonant stem classes. The first are r-stems, which in CS include only two commonly met lemmas: mati 'mother' and dъšti 'daughter'. The latter is in its archaic sg.nom still attested in some dialects in the Northwest (BAN I 467). Generally, the morphology of the lemma differs greatly between dialects. Plural with -e is used in Berl.d. (e.g. here), although -i is preferred in fem.nouns of other stem classes (of course, excluding duals). Standardized are ā- (MK ḱerka, pl. ḱerki) or jā-stems (BG dъšterja, pl. dъšteri) using the oblique stem.
The other stem class showed -y in sg.nom (PSl *-ū) and -ъv- in oblique cases (PSl short *-u-). Oblique extension was being generalized already in OCS: for example, the word for 'blood' is attested commonly with sg.nom krъvъ (cf. Lunt 2001:74). Further development likely shifted the class towards i-stems, as attested by examples showing sg.loc -i (e.g. vъ crkvi crьscěi 'in the royal church'; Vuk.1536). The later sg.nom/acc ending shows irregular developments. Sometimes the jer is lost (despite §10.1; blgočъstivaa větvь 'honorable branch'; Zogr.107), sometimes another vowel (vъ crkovь 'into the church'; Zogr.107). Another (basically ā-stem) ending -va appears in damaskini sources (u crkva xva 'into the church of Christ'; Tixon.d.). The corpus does not provide sufficient data about plural forms.

	
	1.2.7. r-stem nouns (Nf)
		sg		pl		dl
	n	-i		-eri		-eri
	g	-ere		-erъ		-eru
	d	-eri		-erьmъ		-erьma
	a	-erь		[-eri]		[-eri]
	v	-i
	l	[-eri]		-erьxъ		[-eru]
	i	-erjǫ		-erьmi		[-erьma]
	
	1.2.8. ū-stem nouns (Nf)
		sg		pl		dl
	n	-y/-ъvъ		-ъvi		-ъvi
	g	-ъve		-ъvъ		-ъvu
	d	-ъvi		-ъvamъ		-ъvama
	a	[-ъvъ]		[-ъvi]		[-ъvi]
	l	[-ъve]		-ъvaxъ		[-ъvu]
	i	-ъvjǫ		-ъvami		[-ъvama]
	
	ū-stems in later sources
		sg		pl
	n	-ovь/-vь	-vi
	g	-ve		-ovь (?)
	d	-vi		-vamь (?)
	a	[-ovь/-vь]	-vi
	l	[-vi]		-vaxь (?)
	i	-viju		-vami (?)
	

1.2.9. n-stems

The third class of consonant stems is characterized by stem extension -en-, which in sg.nom evolved into -ę (§10.9), as attested in OCS. They show two gender-based paradigms in OCS: a masculine and a neuter one. Masculines generalized the sg./pl.acc endings for nominative (already in OCS; Lunt 2001:74), and later as the case-unmarked forms (e.g. kamen 'stone', pl. kameni). In Punčo's Sbornik we also find old -e as the plural ending (e.g. uvixu u rizi+te kamene 'they wrapped stones in the shirts'; PPS 029). Most neuters of this class retain the nom/acc forms until today. The nasal in sg.nom developed into -e (§10.9), which is regular in Resava orthography (e.g. toje obrěšteši obnosimo ime 'you will find her famous name'; Vuk.1536). Strong jers in sg.inst and some plural forms also evolved to an -e- (imenem efimia 'by name Euphemia'; BAR 287).

	
	1.2.9. n-stem masc.nouns (Nm)
		sg		pl		dl
	n	-y		-ene		-eni
	g	-ene		-enъ		-enu
	d	-eni		-enьmъ		-enьma
	a	-enь		-eni		[-eni]
	l	[-ene]		-enьxъ		[-enu]
	i	-enьmь		-eny		[-enьma]
	
	n-stem neut.nouns (Nn)
		sg		pl		dl
	n	-ę		-ena		-eně
	g	-ene		-enъ		-enu
	d	-eni		-enьmъ		-enьma
	a	[-ę]		[-ena]		[-eně]
	l	[-ene]		-enьxъ		[-enu]
	i	-enьmь		-eny		[-enьma]
	

1.2.10. nt-stems

1.2.11. s-stems

Two consonant stem classes are used with neuters only: nt- and s-stems. The first denoted commonly diminutives and small animals. By the loss of final -t and nasalization (§10.9), the sg.nom ending became the same as that of n-stems; following the denasalization (§10.9), the ending -e was likely so widespread among neuters, that it was taken over per analogy by some s-stems too, e.g. nebe 'heaven, sky', replacing CS nebo in the damaskini (e.g. pod' nbe+to 'in the skies'; Tixon.d.). Pl.nom ending -eta (having undergone §10.9 too) appears in various former jo-stems today, but, as mentioned above (§1.1.5), not as much in the earlier literature. Some old s-stems may have first adopted jo-stem plural endings (e.g. rodixa drьvja+ta 'the trees bore fruit'; PPS 033) before taking -eta; the distribution may be also dialectal (cf. BAN I 458). In our sources, the common s-stem pl.loc endings seems to have been -esěxь (na nbsěxь vьsprijetь slavou 'she received glory in the heavens'; Vuk.1536; nbsexъ in Zogr.107), reflecting influence of o-stems. S-stem pl.inst sometimes shows a final -i (bь otkril bi čjudesi 'God would show [that body] by miracles'; BAR 287; curiously, the "mistake" is repeated by Rostovski too).

	
	1.2.10. nt-stem neut.nouns (Nn)
		sg		pl		dl
	n	-ę		-ęta		-ętě
	g	-ęte		-ętъ		-ętu
	d	-ęti		-ętьmъ		-ętьma
	a	[-ę]		[-ęta]		[-ętě]
	l	[-ęte]		-ętьxъ		[-ętu]
	i	-ętьmь		-ęty		[-ętьma]

	1.2.11. s-stem neut.nouns (Nn)
		sg		pl		dl
	n	-o		-esa		-esě
	g	-ese		-esъ		-esu
	d	-esi		-esьmъ		-esьma
	a	[-o]		[-esa]		[-esě]
	l	[-ese]		-esьxъ		[-esu]
	i	-esьmь		-esy		[-esьma]
	

1.2.12. other C-stems

Some words etymologically going back to consonant stems showed irregular paradigms already in OCS. The word den (< OCS dьnь, sg.gen dьne) 'day' historically was an n-stem (Derksen 2008:134), but it adopted an i-stem sg.nom. In OCS, it showed various pl.gen forms: usually i-stem dьnii or dьnei, but after numerals dьnъ or denъ (Lunt 2001:74). In our corpus, attested examples show a single -i (e.g. dva+dsetь dny puta 'twenty days of journey'; Kiev.d.). Modern varieties use the singular form going back to n-stem sg.acc den. The common pl.nom/acc dni (e.g. here) seems to reflect jo-stems, but the root is usually abbreviated in text. In fact, case-unmarked plural dini (reflecting an end-stressed *dení) appears in a non-Cyrillic source (tái si ispróvodi diní+ti sfói 'thus [St.Petka] passed her days'; NBKM 1064). Pl.nom dnie, attested in OCS (Mirčev 2000:64), does not occur in our corpus. The adverbial form denja shows an ending likely adopted from noštja, as both appear together often as a fixed phrase (e.g. here), but not always (e.g. eliko dnju želaxu večeru biti 'so that in day they wanted the evening to come'; NBKM 326). In sg.loc, both i-stem dni (e.g. here) and n-stem dne (here) can be seen, next to each other.
The word ljudie 'people' is historically an s-stem (Derksen 2008:282). No singular forms are attested in OCS. In plural, the attested forms are mostly those of the i-stem class, but there are traces of the o-stem pl.gen ending -ъ too, also in our corpus (e.g. osta sь sultanom i aleksandrom *r* ljud '100 men remained with Sultan and Alexander'; CIAI 1161). Punčo sometimes uses old pl.acc form ljudi, but does not distinguish it systematically from ljudie (e.g. here).
Finally, the theonym Gospod 'Lord' is originally an i-stem (Derksen 2008:180), but it shows a "great variation" (Lunt 2001:75) already in OCS. The word, usually abbreviated, oscillated between i- and n-stem in a similar way as den. Sg.gen forms include gda and gdne, the ending -a being commonly used until very late as a dep.form (e.g. here). The latter is handled as possessive adjective in our corpus (e.g. za uverenie gdne 'for the sake of faith in the Lord'; PPS 007). Sg.dat has both gdu and gdi - only the former is attested in our texts (e.g. here), although the form gdi is attested as a vocative (e.g. here).

	
	1.2.12. lemmas den 'day', Gospod 'Lord' and ljudie 'people' in later sources (all Nm)
		sg				pl
	n	denь		Gdь		dni		ljudie
	g	dne		Gda		[dni]		ljudь/ljudei
	d	[dni] (?)	Gdu		dnem (?)	ljudem
	a	[denь]		[Gdь]		[dni]		ljudi
	v			Gdi
	l	dni/dne		Gdi (?)		dněxь		ljuděxь
	i	dnemь/dnju	Gdemь (?)	dnemi (?)	ljudmi
	

2. Adjectives

2.1. "simple" BG adjectives

2.2. CS adjectives
2.2.1. short forms
2.2.2. long forms
2.2.3. pronominal stems
2.2.4. degree

Morphology of adjectives shows similar developments between CS and modern varieties as nouns. These include preference for analytic case marking, new definiteness marking, and discarding of a distinctive dual. Furthermore, adjectives also show syncretism of plural forms, formerly distinguished by gender. Analytic marking also replaces old comparative endings. Another difference is syntactic. Especially later CS tends to place verbs into the middle of noun phrases and adjectives behind the noun. In vernacular, noun phrases are compact, with adjectives placed mostly in front of the noun, as in the following example:

	
	kako dvamь sьlikьstvujetь moudrymь (Vuk.1536)
	'how could she liken herself to the wise maidens?' 
	
	kak' šte da sa prismesi sъs mdri+te dvcy (Berl.d.)
	'how she could get among the wise maidens?'
	

In Proto-Slavic, it would be likely hard to distinguish nouns and adjectives as two distinct word categories. The both used practically the same set of endings, and also developed the same distinction of paradigms according to the palatality of stem auslaut (Lunt 2001:54). Even the comparative could be analyzed as a stem extension, followed by a palatal (or "soft") suffix. The main difference was in the suffixing of a relative pronoun, which resulted in a distinction between simple (prosti) or short forms on the one hand, and complex (složni) or long forms on the other. Complex forms are usually interpreted as marked for definiteness (e.g. Flier 1974:65, Mirčev 1978:175, Lunt 2001:142). Unlike the later articles, developing from demonstrative pronouns in BG/MK, complex forms were used by all adjectives in thus marked noun phrase. It was likely a Common Slavic phenomenon, and possibly older, as a similar system can be seen in Lithuanian (Mayer 1978). With the exception of Serbo-Croat, most Slavic varieties have later discarded the distinction. In modern BG/MK, the short forms were generalized in plural and f/n.sg adjectives. The distinction between short and long forms can be found until quite late in the literature, but it is not very strictly followed:

	
	i sič'ka xytros diavol'skaa isčeznuvaše (Tixon.d.)
	i sič'ka xitrostь diavol'ska isčeznuvaše (Ljub.d.)
	'and all the devilish cunning disappeared'
	

It is possible, that the distinction of complexity was affected by animacy scale too. In the standards, majority of dialects, and also in damaskini, m.sg adjectives use long forms, when the article follows: e.g. njegovia+tь dxь 'His Spirit' (Ljub.d.). In some dialects, the pronominal stem was lost, and here the m.sg article is represented only by the long form ending and the surviving hiatus vowel; this is also common in Punčo's Sbornik (e.g. svičkija narodь 'all the people'; PPS 007); elsewhere the long form ending alone works as the article itself (e.g. in Ixtiman area: golemii 'the big'; Mladenov 1963:407). Elsewhere, the article is attached to short forms instead (e.g. neino+tъ životъ 'her life'; Berl.d.). The interaction between old long forms and new articles makes it actually easier to observe the development of the latter on adjectives than on nouns, where it is sometimes homographic with the dependent form:

	
	diavola sę prestruvaše na razni zvěrove (Ned.1856)
	'the Devil transformed into various beasts'
	
	koeto iska duša+ta i čistyę razumъ (ibid.)
	'what the soul and the pure mind want'
	

Of course, this also makes the analysis of the dependent form itself easier. As for the nouns, it is likely that an adjectival dependent form was productive until quite late. Punčo uses in his Sbornik both the dependent and dative forms, if the head noun is masculine and animate - and also in the same form, e.g. ednogo vernago člveka 'I called one loyal man' (PPS 029). Endings like -ago and -omu, commonly used in Sbornik, actually reflect different paradigms: the former reflects an adjectival m/n.sg.gen/acc long form, while -omu is originally a pronominal ending. Short forms are used rarely, only with possessives (e.g. do doma faraonova; PPS 022), short m.sg.dat appearing only scarcely on possessors based on theonyms (po obrazu bžiju; PPS 024). When in oblique cases, articles also follow adjectival short forms, but showing then a rather irregular ending -o-:

	
	i drugo+togo brata svoego nemu predadoxu (PPS 022)
	'they gave him the other brother too'
	
	u sakulju nai mlado+tomu člveku (ibid.)
	'to the bag of the youngest one'
	

The adjectival ending was likely assimilated to the vowel of the article stem. But articles are generally less common on adjectives in the Sbornik. The complexity distinction was unlikely a thing by then; the case, in the sense of dependency degrees, could be. Similar forms can be also seen in damaskini sources, e.g. na tvoigo+tokъ sina 'to your son' (Sv.d.). Damaskini also show a change in the anlaut vowels of long endings the of the ending to -i-, which is still preserved in some dialects using adjectival pronouns (Mirčev 1978:177; e.g. starimu cited in §1). Adjectival pronouns are also the only type showing dependent forms later too. They seems to remain productive in NBKM 1423 and 19th c. sources, as well as in today's standards:

	
	cato da i hodila sas drughigo (NBKM 1423)
	'as if she went with another one'
	

As the corpus includes archaisms and archaic varieties, the lemmas are classified according to categories relevant for OCS. Here, the primary distinction was between adjectival and pronominal stems. Adjectival stems could build both short and long forms; pronominal ones had only one regular form - similar to short ones in nominative, and f.sg.acc and inst, other cases resembling long forms. Both of these classes had a hard and a soft paradigm, used according to the palatality of the stem auslaut - as in nouns. The following table (based on Lunt 2001:64), showing adjectives novъ 'new', ništь 'poor', tъ 'that' and našь 'ours'. Lunt (2001:62) defines also a specific paradigm for mixed stems, which is used by some adjectives like vьsь 'all, whole', showing a set containing both hard and soft endings:

	
				m.sg.nom	m.sg.gen	f.sg.nom	f.sg.gen	pl.gen
	adjectival stems
		hard		novъ		nova		nova		novy		novъ
		- long form	novyi		novaago		novaa		novyę		novyixъ
		soft		ništь		ništa		ništa		ništę		ništь
		- long form	ništii		ništaago	ništaa		ništeę		ništiixъ
	pronominal stems
		- hard		tъ		togo		ta		toę		těxъ
		- soft		našь		našego		naša		našeę		našixъ
		- mixed		vьsь		vsego		vsa		vseę		vsěxъ
	

Two stem classes are relevant also for newer varieties. Some adjectives generalized the long-form as their m.sg form, thus having no overt m.pl marking (unless they have an article); they are classed as complex sg. Finally, some borrowed adjectives show no inflectional marking, i.e. neither for gender nor number, and can be handled as particles or classed as uninflected.

In the corpus described here, the category of adjectives (indicated by an A in the PoS-tag) is wide and encompassing many tokens, which could alternatively be analyzed as other categories too, e.g. pronouns, numerals and even verbs (participles):

	kato na někoi elenъ (Berl.d.)
	'as to a deer'
	
	vъ edna noštъ (Tixon.d.)
	'in one night'
	
	da mi ne bъdi zabranenъ pъte+tъ (Berl.d.)
	'may my path shall not be barred'
	

The corpus strives to reflect the morphological variety of the sources. For this reason, any token showing endings specific for adjectives is also annotated as such. Alternative or secondary PoS-tags are employed in the corpus to distinguish other properties, like verbal aspect in participles. When the token is used as a head of a clause (as zabranenъ in the example above), the adjectival tag is secondary. When it is used adnominally (like e.g. in stenania ne+prěstanna 'endless sighs' here), the primary PoS-tag is adjectival. The grammatical properties of an adjective are all reflected on the ending. Properties shared with nouns and pronouns are gender (f/m/n), number (s/p/d) and case (n/g/d/a/v/l/i/o); beside these, the tag also indicates complexity (y/n) and, optionally, degree (c):

	
	světlěišia - PoS-tag: Afsnnc, lemma: světъl (hard stem indicated in the dictionary)
	A	adjective	property of lemma and syntax
	f	feminine	property of ending
	s	singular	property of ending
	n	nominative	property of ending
	n	short-form	property of ending
	c	comparative	property of ending
	

2.1. "Simple" Bulgarian adjectives

Adjectival inflection is very straightforward in modern BG/MK. The ending agrees with the rest of noun phrase in gender (if singular) and number - thus, the full set would include only four endings (null for m.sg, f.sg -a, n.sg -o and pl -i), excluding articles. The old m.sg long-form -i is also used in ordinal numerals like pъrvi and relative adjectives with -sk- suffix (e.g. iudeiski 'Judaean'; PPS 018). However, the situation in our corpus is more complicated.
One aspect is the common presence of dependent and dative forms on the one hand, and of complexity distinction on the other. Case-unmarked forms appear abundantly in all sources, although often representing fixed phrases like crstvo nbsnoe 'Kingdom of Heaven' (e.g. here). Marking of oblique forms follows the same rules as for the nouns.
A distinctive n.pl form -a can be sometimes seen in Punčo's Sbornik (e.g. imenita čudesa 'notable miracles'; PPS 019). Only such ending is tagged as a marked plural neuter, i.e. "Anpnn". Plural adjectives ending in -i are marked for gender only when the head noun is a f/m.pl; gender remains unmarked, if the head is a neuter (e.g. negovi čeda 'his children'; PPS 007: tagged "A-pnn").
In the standards, the ending usually agrees with the semantic gender of the head noun, i.e. it shows a masculine form, even if the noun is an ā-stem, like in pъrvi voevodo 'o first commander!' (PPS 013). In our sources, the ending may reflect the grammatical gender too (e.g. junačka voevoda 'commander of the heroes'; PPS 032). This does not reflect itself on the annotation.
Finally, the "case" option o marks hiatus vowels, which arise by the attachment of a second pronominal stem to the m.sg ending (i.e. the article). This enables us to analyze article development even when the article stem is lost, and thus not reflected as a separate token (as in čistyę).

	
	2.1.1. masculine adjectives (Am)
		sg			pl
		long	short	
	n	-i	-ь		-i
	g	-ago	-a
	d	-omu
	o	-ia	-o

	2.1.2. feminine adjectives (Af)
		sg			pl
		long	short	
	n	-aa	-a		-i
	a	-uju	-u

	2.1.3. neuter adjectives (An)
		sg			pl
		long	short	
	n	-oe	-o		-a
	

2.2. Church Slavonic adjectives

2.2.1. short forms
2.2.2. long forms
2.2.3. pronominal stems
2.2.4. degree

While Church Slavonic adjectives show less stems classes than nouns, complexity distinction and local influences of analogy and phonetic contraction (§10.8) pose a challenge for a readable overview. CS sources in our corpus reflect different redactions (Tarnovo, Resava, Ruthenian), and also degrees of application of orthographic rules, by which later CS variants differ from older ones. The situation is further complicated by the divergence between vernacular and CS in the realm of syntax.
Already in OCS there were words like vьsь, which showed a mixing of paradigms - having a "soft" m.sg.gen vьsego, but "hard" pl.gen vьsěxъ (Lunt 2001:62).

2.2.1. short forms

Short forms represent adjectives unmarked for definiteness. They are used systematically in predicates, but also in comparatives (e.g. světlěišia 'brighter'; Vuk.1536), possessives (even if the rest of phrase is definite: siju dobruju čistouju goloubicu xvou 'this good, pure dove of Christ'; Vuk.1536), and participles (especially when used as the main verb: kako prěnesena byst 'how she was translated'; Vuk.1536).
As the endings are the same as in nominal declensions, phonetical changes have similar effects. F.sg.gen/dat forms may appear as -e (e.g. naše Petky 'of our Petka'; Vuk.1536) and -i (xvi mtere 'of the mother of Christ'; Vuk.1536) in later sources (§10.3-5), but differently for each stem. Both f.sg.acc and inst endings are affected by denasalization (§10.6), which leads to confusion already mentioned in comments to §1.2.4 (glavnoju ogьnnoju). Besides f.sg.gen, also f/m.pl.acc and m/n.pl.inst -y is affected by a shift to /i/ (§10.3), which sometimes leads to confusion (e.g. xromy xoždenie 'the crippled [receive the boon of] walking'; BAR 287).
In OCS, there was also one distinctive vocative form -e (e.g. farisěju slěpe 'o, blind Pharisee!'; Mt 23:26, cf. Lunt 2001:142). It would later become undistinguishable from sg.loc (due to §10.8; o srětenii ženixově 'about meeting the bridegroom'; Vuk.1536), but regretfully (or fortunately?) neither of them is common in our corpus.
In hard stems, the ending -omь may appear both as a sg.loc (e.g. na presvětlomъ Prestolě 'on a shiny throne'; Ned.1806) and inst (ęzykomь plьtěnom 'by the tongue of flesh'; Zogr.107), as it is common in Serbo-Croat today. As the ending goes back to OCS sg.inst, it is tagged "i" in singular. Similarly, also soft-stem -emь is sometimes found instead of a sg.loc ending (o vьsemь svoem domu; Zogr.107) too.
As the forms like -ago and -omu in the damaskini sources show, it is possible that long forms were generalized in oblique cases before the case marking became obsolete. The elimination of short forms may have also been due to influence of East Slavic redactions. In any case, oblique short forms are generally less frequent: in the corpus of Life of St.Petka, 948 oblique adjectives show long forms, but only 198 short ones.

	
	2.2.1. CS adjectival short forms
	
	hard stem
		sg			pl
		m	f	n	m	f	n
	n	-ъ	-a	-o	-i	-y	-a
	g	-a	-y	-a	-ъ	-ъ	-ъ
	d	-u	-ě	-u	-omъ	-amъ	-omъ
	a	[-ъ]	-ǫ	[-o]	[-y]	[-y]	[-a]
	l	-ě	[-ě]	-ě	-ěxъ	-axъ	-ěxъ
	i	-omь	-ojǫ	-omь	[-y]	-ami	-y

		dl
		m	f	n
	n	-a	-ě	-a
	g	-u	-u	-u
	d	-oma	-ama	-oma


	soft stem
		sg			pl
		m	f	n	m	f	n
	n	-ь	-a	-e	-i	-ę	-a
	g	-a	-ę	-a	-ь	-ь	-ь
	d	-u	-i	-u	-emъ	-amъ	-emъ
	a	[-ь]	-ǫ	[-e]	-ę	[-ę]	[-a]
	l	[-a]	[-i]	[-a]	[-ixъ]	-axъ	[-ixъ]
	i	-emь	-ejǫ	-emь	[-i]	-ami	-i

		dl
		m	f	n
	n	-a	-i	-a
	g	-u	-u	-u
	d	-ema	-ama	-ema

	
	short forms in later sources
		sg			pl
		m	f	n	m	f	n
	n	-ь	-a	-e	-i	-i/e	-a
	g	-a	-i/e	-a	-ь	-ь	-ь
	d	-u	-e/i	-u	-omь	-amь	-omь
	a	[-ь]	-u	[-e]	[-i]/-e	[-i/e]	[-a]
	l	[-omь]	[-i]	[-omь]	-exъ	-axь	-exь
	i	-omь	[-u]	-omь	[-i]	-ami	-i
	
	

2.2.2. long forms

As already mentioned, long forms are used in OCS to mark definiteness. They appear in nominal phrases marked for definiteness in the Greek original, but also denoting "generally known objects" (Lunt 2001:142), as well as anaphoric and deictic references (as described in our study).
Etymologically, long forms are created by attaching a pronoun based on the root *jь (used e.g. in relative pronouns like iže 'which' and personal pronouns) to a short form adjective. In the aforementioned example xytros diavol'skaa, the adjective can be analyzed into *diavolska + f.3sg.nom *ja. The actual pronounciation is obscured by the script already in OCS sources, which often does not reflect phonem *j clearly. Thus, we cannot clearly say, whether the pronounciation was long *diavolskā (as in Czech and Slovak) or *diavolskaja (as in East Slavic). From a Balkan Slavic perspective, both options are feasible as diachronic variants. According to Lunt (2001:65), long forms in OCS reflect an early stage of contraction (§10.8), in which the post-*j vowel is assimilated to the preceding one, as a phonotactic rule. Thus we can find forms like m.sg.gen velikaago 'of the great' (Mt 5:35), but not a form reflecting underlying *velika+jego. Doubled vowels were commonly shortened in later CS redactions into forms like velikago (e.g. here). The contraction could also result in confusion: endings based on etymological (single) -i (or -y, due to §10.3) are sometimes doubled (e.g. on adverbs like skoči junaš'kyi 'he jumped heroically'; Tixon.d.).
In the orthographic system employed in our texts, contraction affects mostly word-internal vowel sequences, e.g. -aago > -ago, -yixъ > -ixь. Only Zogr.107 tries to preserve them fully. Many hard stem oblique endings lose distinction from soft stems because of §10.3 (m.sg.nom/acc, m/n.sg.inst, pl.gen/loc, pl.dat, pl.inst); the same rule also affects f.sg.gen, m/f.pl.acc and f.pl.nom.
Shifts of nasals (§10.5-6) affect f.sg.acc ending, turning to -uju in most sources; Tarnovo-based Zogr.107 shows still -ǫǫ or -ǫę. Hard-stem f.sg.gen shows commonly -yje (e.g. podvizy prěpodobnyje petky 'deeds of the Reverend Petka'; Vuk.1536), less frequently as -ie (blizь těla stie 'next to the body of the saint'; BAR 287). Soft-stem f.sg.gen is then -ee (dše mojee ozlobljenie 'evils of my soul'; Vuk.1536). Sometimes, the gen/dat endings are confused (e.g. styjei prpdbnii paraskevii '[Life] of Holy Reverend Parascheva'; BAR 287). Typical for Zogr.107 is the writing -yǫ (e.g. ot mǫdryǫ seǫ dvy 'by this wise maiden'; here); the choice of nasal is quite free.
The mixng of m/n.sg.loc and inst is practically regular for the soft stem; in hard stems, sg.loc -ěmь is used quite regularily, although there are also examples like sъ mnoźěmь usrьdiem 'with much effort' (Zogr.107), but this can be due to the lemma oscillating between the adjectival and pronominal stem class. Sometimes, the ending is simply abbreviated to -m (e.g. prstmъ; 'most holy' Zogr.107), remaining ambiguous.
The influence of the pronominal paradigm is well seen in the m/n.sg.dat: only -omu can be seen in our sources (e.g. tvojemou strašnomou prěstolou 'to your awesome throne'; Vuk.1536). Long form f.sg.inst endings are not attested in our corpus at all - only the ones used originally as short and pronominal forms.

	
	2.2.2. CS adjectival long forms
	
	hard stem
		sg			pl
		m	f	n	m	f	n
	n	-yi	-aa	-oe	-ii	-yę	-aa
	g	-aago	-yę	-aago	-yixъ	-yixъ	-yixъ
	d	-uemu	-ěi	-uemu	-yimъ	-yimъ	-yimъ
	a	[-yi]	-ǫǫ	[-oe]	-yę	[-yę]	[-aa]
	l	-ěmь	[-ěi]	-ěmь	[-yixъ]	[-yixъ]	[-yixъ]
	i	-yimь	[-ǫǫ]	-yimь	-yimi	-yimi	-yimi

		dl
		m	f	n
	n	-ii	-ěi	-ěi
	g	-uju	-uju	-uju
	d	-yima	-yima	-yima


	soft-stem
		sg			pl
		m	f	n	m	f	n
	n	-ïi	-aa	-ee	-ii	-ęę	-aa
	g	-aago	-eę	-aago	-iixъ	-iixъ	-iixъ
	d	-uemu	-ěi	-uemu	-iimъ	-iimъ	-iimъ
	a	[-ii]	-ǫǫ	[-ee]	-ęę	[-ęę]	[-aa]
	l	[-iimь]	[-ěi]	[-iimь]	[-iixъ]	[-iixъ]	[-iixъ]
	i	-iimь	[-ǫǫ]	-iimь	-iimi	-iimi	-iimi

		dl
		m	f	n
	n	-aa	-ii	-ii
	g	-uju	-uju	-uju
	d	-iima	-iima	-iima


	long forms in later sources
		sg				pl
		m	f	n		
	n	-ii	-aa	-oe/-ee		-ii/-ie/-ee/-aa
	g	-ago	-ie/-ee	-ago		-ixь
	d	-omu	-ei	-omu		-imь
	a	[-ii]	-uju	[-oe/-ee]	-ie (m.pl)/[-ie (f.pl)/-ee/-aa]
	l	[-imь]	[-ei]	[-imь]		[-ixь]
	i	-imь		-imь		-imi	
	

2.2.3. pronominal stems

The hard pronominal paradigm is relevant for demonstrative pronouns tъ, onъ and ovъ, but also a number of adjectival stems like drugъ, inъ both 'other', samъ 'alone, self', edinъ 'one', eterъ 'such' or kakъ 'which', sometimes also mnogъ 'many'. The soft paradigm was used by the demonstrative sь, as well as possessive adjectives like moi 'my' and našь 'ours' (Lunt 2001:62). In both later CS and especially "simple" sources, oblique short forms and other adjectival endings may appear freely (e.g. kamto edinago bga towards the One God'; Berl.d.).
Besides the nominatives, as well as f.sg.acc and inst, the endings are tagged as long forms.

	
	2.2.3. pronominal declension of adjectives
	
	hard stems
		sg			pl
		m	f	n	m	f	n
	n	-ъ	-a	-o	-i	-y	-a
	g	-ogo	-oę	-ogo	-ěxъ	-ěxъ	-ěxъ
	d	-omu	-oi	-omu	-ěmъ	-ěmъ	-ěmъ	
	a	[-ogo]	-ǫ	[-o]	-y	[-y]	-a
	l	-omь	[-oi]	-omь	[-ěxъ]	[-ěxъ]	[-ěxъ]
	i	-ěmь	-ojǫ	-ěmь	-ěmi	-ěmi	-ěmi

		dl
		m	f	n
	n	-a	-ě	-ě
	g	-oju	-oju	-oju
	d	-ěma	-ěma	-ěma
	
	
	soft stems
		sg			pl
		m	f	n	m	f	n
	n	-ь	-a	-o	-i	-ę	-a
	g	-ego	-eę	-ego	-ixъ	-ixъ	-ixъ
	d	-emu	-ei	-emu	-imъ	-imъ	-imъ	
	a	[-ego]	-ǫ	[-o]	-ę	[-ę]	-a
	l	-emь	[-ei]	-emь	[-ixъ]	[-ixъ]	[-ixъ]
	i	-imь	-ejǫ	-imь	-imi	-imi	-imi

		dl
		m	f	n
	n	-a	-i	-i
	g	-eju	-eju	-eju
	d	-ima	-ima	-ima
	
	

As mentioned above some adjectival pronouns like vьsь 'all' and sicev 'such' show endings from both hard and soft paradigms in OCS. When we take later CS sources into account, the variety rises due to further levelling of paradigms. Aside forms traceable to productive adjectival ones, there are also some instances in Zogr.107, which reflect its orthographic peculiarities: vsě denoting a n.pl.nom/acc (vьsě jaže kь čъsti 'everything due for respect'; here), as well as vsę, denoting a f.sg.acc due to §10.12 (na tę vьsę moǫ vьzlagaǫ nadeždǫ 'I lay all my hope upon you'; here).

	
	mixed stem: vьsь 'all'
		sg			pl
		m	f	n	m	f	n
	n	vьsь	vьsa	vьse	vьsi	vьsę	vьsa
	g	vьsego	vьseę	vьsego	vsěxъ	vsěxъ	vsěxъ
	d	vьsemu	vьsei	vьsemu	vsěmъ	vsěmъ	vsěmъ
	a	[vьsь]	vьsǫ	[vьse]	vьsę	[vьsę]	[vьsa]
	l	vьsimь	[vьsei]	vьsimь	[vsěxъ]	[vsěxъ]	[vsěxъ]
	i	vьsěmь	vьsejǫ	vьsěmь	vsěmi	vsěmi	vsěmi
		
		dl
		m	f	n
	n	vьsě	vьsi	vьsi
	g	vьseju	vьseju	vьseju
	d	vьsěma	vьsěma	vьsěma
	
	
	forms of lemma vse attested in the corpus	
	m			f			n
	vьsь	Amsnn		vsa	Afsnn		vse	Ansnn/R
	vsa	Amsgn		[vsę]	Afsgn		vsa	Ansgn
	vsego	Amsgy		vseje/ę	Afsgy
	[vsě]	Amsln		vsei	Afsdy		[vsě]	Ansln
	vsem	Amsin		vsu	Afsan
	vsěm	Amsiy		vseju/ǫ	Afsin		vsěm	Ansiy

	vsi	Ampnn		vsę/e	Afpnn		vsa/ě	Anpnn
	vsěx	Ampgy		vsěx	Afpgy
	vsěm	Ampdy
	vsę/e	Ampan
	vsěmi	Ampiy
	

2.2.4. degree

Synthetic comparatives are regular only in CS, and thus rare in our corpus. They are annotated by an additional c in the PoS tag. Technically, this allows also adding a superlative marker in languages, where it is relevant. The marking of degree is mostly avoided on the adjectival level in "simple" BG texts: comparative po and superlative nai are handled as separate tokens (cf. §9). The tag is used, when the stem shows the specific suffix -ě(iš)- or, when the stem-final consonant undergoes jotation (Lunt 2001:78), -i(š)-. The distribution of the suffixes was likely lexical in OCS. The inflection of these forms is similar to that of present active participles (cf. below §5), using soft-stem short form adjectival endings (§2.2.2):

	
	2.2.4. comparative degree of adjectives
	
		unmarked	sg.nom		pl.nom
	fem	nova		nov-ěiši	nov-ěišę
	masc 	novь		nov-ěi		nov-ěiše
	neut	novo		nov-ěe		nov-ěiša
	
	fem	draga		draž-ьši	draž-ěišę
	masc 	dragъ		draž-ii		draž-ьši
	neut	drago		draž-e		draž-ьša
	

3. Pronouns

3.1. personal
3.1.1. "simple" BG personal pronouns
3.1.2. CS personal pronouns
3.1.3. reflexive pronouns

3.2. demonstrative
3.2.1. articles
3.2.2. CS short demonstratives
3.2.3. extended demonstratives

3.3. relative
3.3.1. simple "BG" relative/interrogative pronouns
3.3.2. simple "BG" marked relative (etc.) pronouns
3.3.3. CS relative/interrogative pronouns
3.3.4. CS marked relative (etc.) pronouns
3.3.5. pronoun iže

Pronouns are a heterogenous word category: some are syntactically used as nominal arguments, others as adverbs; some have the same root-stem-ending structure as nouns and adjectives, other show indeclinable suffixes or only one productive form. The borders between pronouns and adverbs, conjunctions and other particles are not always clear; such definitions are deeply embedded in grammatical traditions. The variation is also considerable from both dialectal and diachronic points of view. The corpus provides many examples, such as the following one:

	
	i abie posla kь iže tamo souštimь vь cьrigradě frougomь (Vuk.1536)
	i togazi pusti povelěnie na crigrad', što běxa tamo fren'ci (Tixon.d.)
	i tutak'si provodi na fren'ci+te deto běxa tam v' crigrad (Berl.d.)
	'and he immediately sent [a message] to the Franks, who were in Tsarigrad'
	

The temporal adverb abie was translated to togazi 'then' by the earlier editor; the Berl.d. editor uses another adverb, which is composed of multiple demonstrative stems. Relative pronoun iže 'who' shows here a general case, as it is usual in later CS (cf. §3.3.5). Tixon.d. uses the pronoun što instead, originally interrogative 'what' (cf. Demina et al. 2012:1131; CS čьto), while the later damaskin shows deto, going back to interrogative pronoun of location kъde 'where' and the later relative suffix -to. All three varieties use tamo (or tam) with a general meaning 'there'. In CS sources also tu could be used in the same meaning, as attested also in our corpus (e.g. here); tamo had originally an allative meaning (cf. §3.3.4). The systems sometimes mix up with each other:

	
	i onia čudotvorci iscelixu onuju ženu (PPS 007)
	'and those miracle workers cured that woman'
	

Both demonstratives are based on the old stem on, which is also common in personal pronouns, both in CS sources and in Punčo's Sbornik (e.g. here). The first likely represents pl.nom /oni+ja/, with an indeclinable suffix -ja attached after the ending showing the number (in script, Punčo seems to distinguish plural ѡниа used here from m.sg ѡнıа, written with a digraph for /ja/ like here). In the second pronoun, it seems to be reinterpreted as an adjectival long form suffix. Punčo handles the suffix -ja not differently from jā-stems and masculines described in §1 as "new i-stems" (like djukjandžija 'store owner'): as a case-unmarked ending, contrasting with dependent form -ju. M.sg distal pronoun is usually onogova (e.g. here).

Pronouns are tagged with an initial P. As already mentioned in §2, adjectival pronouns (used as "amod" dependent on nouns) are generally handled as morphological adjectives in the corpus, i.e. with initial "A" in the PoS-tag. Thus, in někoi elenъ the first token receives the tag "Amsny", while in da ne+doide nekoi 'so that nobody comes' (PPS 010) we could use pronominal tag "Pi---n". An exception from the rule are adjectival pronouns using the extensions -ja, -va or -zi, which include all demonstratives, but also other pronouns like the f.pl relative takviva 'such' (e.g. here). The second position in the tag denotes the semantic type. All of these are tagged according to relevant affixes - not according to the actual function in the text! For example, Punčo uses the "interrogative" form koga 'when' commonly as a relative pronoun (e.g. here).

	
	tag	type			morphem		examples
	Pp	personal		cf. §3.1.	az, ti, toi ...
	Px	reflexive		cf. §3.1.	sebe, si, se
	Pd	demonstrative		cf. §3.2.	tъ, sь ...
	Pq	interrogative		no affix	koi, kъde, kak
	Pi	indefinite		ně-		někoi, někъde, někak
	Pr	relative		-to		koito, kъdeto, kakto
	Pz	negative		ni-		nikoi, nikъde, nikak
	

The position after the type indicates the person (1/2/3), which is relevant for personal pronouns only. Following positions indicate gender (f/m/n), number (s/p/d) and case (n/g/d/a/v/l/i/o), if relevant. They can be left empty in adverbial pronouns like koga (tagged simply "Pq") or pronouns distinguishing only case/dependency, like nikogo (tagged "Pz---g"; e.g. here).

	
	éĭ - PoS-tag: Pp3fsd, lemma: tja
	P	pronoun		property of lemma and syntax
	p	personal	property of lemma
	3	3rd person	property of form
	f	feminine	property of form and lemma
	s	singular	property of form
	d	dative		property of form
	

3.1. personal pronouns

3.1.1. "simple" BG personal pronouns
3.1.2. CS personal pronouns
3.1.3. reflexive pronouns

3.1.1. "simple" BG personal pronouns

"Simple" BG literature shows a system of personal pronouns very similar to that of modern varieties. However, they also reflect dialectal differences. 1sg azь 'I' is preferred in most sources, but ja is frequently used especially by Punčo (e.g. here). Nominatives of 3rd person are built using the demonstrative stem t- (toi, tja, pl. tě) in the damaskini, while Punčo prefers the stem on- (onь, ona, pl. oni). F.3sg.nom clitic form is ja, while Punčo uses more frequently ju; this form sometimes appears even in constructions, where one could expect a dative (e.g. da ju gi dade 'to give them to her'; PPS 006). Very rarely, Punčo uses also the 1pl.nom form mie (e.g. here).
The forms reflect number, case (or, rather, degree of dependency), and, in 3sg, also gender. Moreover, both dependent and dative forms have long and clitic forms; the latter show more restricted placement and - very rarely in our sources yet - can fulfill other functions when placed next to the syntactic head of dependency, i.e. using so-called "clitic" or "object doubling" (e.g. ikona+ta i na presfetae bogorodica 'icon of the most holy Mother of God'; NBKM 1064). In our corpus, clitic dependent forms (e.g. me, gi) are tagged with an "a" on the case position; long forms (mene, těx) as "g", so that they can be searched for separately. Both dative forms, if distinguishable from dep.forms (like vamь, but unlike mene) are tagged with a "d". Plural forms like ni and vi are tagged "a", even if denoting a possessor or recipient.
Long dat.form těmь (cf. Mirčev 1978:183) is attested only in CS sources (e.g. here); Punčo rarely uses the form gimь (e.g. here). N.3sg.nom form tova (or onova) is tagged as a demonstrative ("Pd-nsn") because of the suffixed extension -va (cf. below §3.2.3).
In "simple" BG texts, as a rule, form ti is always tagged "Pp2-sn", is it is homographic with the dat.form. If the text regularily distinguishes 2sg.nom ty and 2sg.dat ti, it is indicated by a separate lemma. In the same way, CS sources may distinguish between dat forms ending in -ě like tebě.

	
	3.1.1. "simple" BG personal pronouns (Pp)
		sg
		1	2	3
				m	f	n
	n	azь/ja	ti	toi	tja	[tova]/[onova]
	g	mene	tebe	nego	neja	nego
	d	mi	[ti]	mu	i	mu
	a	me	te	go	ja/ju	go

		pl
		1	2	3
	n	nie/mie	vie	tě/oni
	g	nasь	vasь	těxь
	d	namь	vamь	imь/gimь
	a	ni	vi	gi/[n]ix
	

3.1.2. CS personal pronouns

Church Slavonic forms may appear in "simple" texts as archaisms or parts of fixed phrases too (e.g. here). The underlying morphems have already been analyzed above in §2.2.5-6. Lemmas for the most personal pronouns are identic with those for "simple" BG texts - i.e. az, ti, toi, tja, mie, vie, tě. Demonstrative stem can be used on oblique forms too (e.g. na toi 'on her'; Vuk.1536). If on- stem is preferred or /y/ is distinguished from /i/, other lemmas may appear too. Church Slavonic texts also distinguish forms like nego and ego according to phonetic context, not as "long" and "clitic" forms; both are tagged in the same way (with "g" on the case position). N.3sg.nom to is tagged as a demonstrative particle ("Qd"), as it can be used also as a conjunction (e.g. here).
As in "simple" BG, the gender is not tagged in most non-neuter plural (and also dual) forms. The form těxъ is tagged "Pp3-pg", even if used as a direct object, as it is usual in later CS sources (e.g. těxь pokryvaaše oblakь 'a cloud covered them'; Vuk.1536). The form ixь (and prepositioned nixь) is tagged as "Pp3-pa" for compatibility with "simple" BG paradigm. The original 3pl.acc ę (e.g. here) is already rare in newer sources. In letters of Wallachian princes, 1/2pl.acc forms appear as in the modern varieties both in form (reflecting §10.3.: as ni and vi) and function (denoting possessors and recipients: e.g. tuzi vi davam znati lit. 'I give this for you to know' in the letter № 210 by Vlad III); the feature is not found in the Tale and the Legend yet, nor in other CS sources (Mirčev 1978:180). 3sg.acc short forms go (or ga) and jǫ(written ę, ǫ or ju) are common in our sources (e.g. priǫsta go 'they have accepted him'; Vat.slav.2) too.
Homographic personal and demonstrative pronouns are distinguished in the corpus according to the context. The pronoun is marked as "personal", when it appears without any further elements in the noun phrase (e.g. na toi: "Pp3fsd", lemma tja; těxь pokryvaaše oblakь: "Pp3-pg", lemma tě). The pronoun is analyzed as a demonstrative, if it is used adnominally or enclitically within a noun phrase with more elements (e.g. vь toi straně 'in that land'; BAR 287; tagged "Pd-fsd", lemma tъ).
Generally, the annotation of pronouns tries to stick to the paradigm of "simple" BG.

	
	3.1.2. Church Slavonic personal pronouns (Pp)
		sg
		1	2	3
				m	f	n
	n	azъ	ty	tъ	ta	[to]
	g	mene	tebe	[n]ego	[n]eę	[n]ego
	d	mně/mi	tebě/ti	[n]emu	[n]ei	[n]emu
	a	me	tę	go	[n]jǫ	[to]
	l	[mně]	[tebě]	[n]emъ	[nei]	[n]emъ
	i	mnojǫ	tobojǫ	[n]imъ	[n]ejǫ	[n]imъ

		pl
		1	2	3
				m	f	n
	n	my	vy	ti	ty	ta
	g	nasъ	vasъ	[n/ixъ]	[n/ixъ]	[n/ixъ]
	d	namъ	vamъ	[n]imъ	[n]imъ	[n]imъ
	a	ny	[vy]	[n]ę	[n]ę	[ta]
	l	[nasъ]	[vasъ]	[n/ixъ]	[n/ixъ]	[n/ixъ]
	i	nami	vami	[n]imi	[n]imi	[n]imi

		dl
		1	2	3
	n	my/vě	vy/va	ja/i
	g	naju	vaju	[n]eju
	d	nama	vama	[n]ima
	

3.1.3. reflexive pronouns

All Slavic varieties show a similar system of reflexive pronouns - used when some non-subject dependents denote the same thing as subject. There are two lemmas denoting them in our corpus: possessive adjective svoi, handled as a soft pronominal stem adjective (§2.2.6) in the corpus, and the proper reflexive pronoun se. The pronoun has three distinct forms in modern varieties: long (sebe) and clitic (se) dependent forms and one clitic dative form (si). All are old, but showing some phonetic shifts in comparison to CS forms (§10.5-6). In some texts, the nasal is occasionally reflected with a jer (due to §10.12, e.g. i utrě sъ razvali 'and tomorrow it falls apart'; Tixon.d.). OCS long dative form sebě is not consistently distinguished in CS sources of our corpus (cf. sebě oudroučevajuštii 'making herself strong'; Vuk.1536). The tag marks only the type (Px) and case.

	
	3.1.3. reflexive pronouns
	
		"simple" BG	late CS		OCS		PoS tag
	g	sebe		sebe		sebe		Px---g
	d	si		si		sebě/si		Px---d
	a	se		se		sę		Px---a
	l			[sebe]		[sebě]
	i			soboju		sobojǫ		Px---i
	

3.2. demonstrative pronouns

3.2.1. articles
3.2.2. CS short demonstratives
3.2.3. extended demonstratives

There are multiple types of morphems indicating deixis, anaphoric relation, general knowledge and other shades of determination in our corpus - adjectival long-form endings discussed above, articles (e.g. žena+ta), adnominal pronouns (tozi, iže), and also particles like se or že. Most pronominal ones are tagged Pd. Function of a word as a syntactic determiner is indicated by the syntactic (UD) tag "det". Not all "Pd"-s act as determiners in the sentence (e.g. tova namъ potrebno estъ 'that is important for us'; NBKM 728; tova: "nsubj"), nor are all "det"-s handled as pronouns (e.g. umrě někoi korabnikь 'a sailor died'; Tixon.d.; někoi: "Amsny").

3.2.1. articles

A characteristic feature of modern BG/MK varieties are postposed articles (členna forma), marking definiteness of noun phrases. These are in most situations attached as a suffix or enclitic to the first element of the noun phrase. This is also one of the most typical "Balkan" phenomena in Balkan Slavic, as similar enclitics appear in Albanian and Romanian.
Today standards and most varieties use only the first case row. The "orthographic" m.sg dependent form -a (like šte vzema vlaka 'I will take the train'), standardized in post-WW2 BG, is homographic to the old gen/acc ending, and thus handled as such a morphem. When the context makes the distinct function clear, like in the example diavola sę prestruvaše (mentioned above, from Ned.1856), a secondary tag is employed for the noun ("Nmsoy"). The annotation similarly handles other shortened forms of articles, like kupeco 'the merchant' (PPS 040). These forms are handled as an ending, but thus modified noun (or adjective) carries an UD extension tag "p_nom" (or "p_adj"; for kupeco here, the full tag is "csubj:p_nom"). In this way they can be compared with articles handled as separate tokens.
Beside these, the corpus shows many examples of case marking on the article as already discussed in the §1. These are handled in the same way as adnominal demonstratives handled below:

	
	3.2.1. "simple" BG articles (Pd)
		sg			pl
		m	f	n	f/m	n
	n	tь	ta	to	te	ta
	g	togo
	d	tomu
	a		tu/tǫ
	o		tь
	

3.2.2. CS short demonstratives

Etymologically, articles are based on (short) adnominal demonstratives. The actual distinction between the two is controversial, as we have discussed elsewhere. A common distinction between an article and an adnominal demonstrative is its syntactic position (cf. Mirčev 1978:197). In Church Slavonic, the position is rather free - it can appear as the first element of the noun phrase (togo města 'of that place'; Zogr.107), but also as today - postposed after the first element of the noun phrase (samyi+tъ crstvuǫštii grad 'the ruling city [i.e. Tsarigrad] itself'; Zogr.107). However, other words may appear between it and the preceding word (e.g. vъ nošti že toi 'and in that night'; Zogr.107). The t-stem pronouns may also be used as personal pronouns, i.e. without any further elements in the noun phrase (cf. above §3.1.2). Another difference is that in "simple" BG texts, the articles show only the stem t-. CS texts, on the other hand, also show deictically marked stems s- (proximal) and on- (distal), both as "free" and postposed (e.g. slovo ono prročьskoe 'that prophetic word'; Vuk.1536). Suffixes based on these stems are attested in the dialects especially in Rhodopes (cf. Adamou 2011) and Macedonia (cf. Topolińska 2006), but not in our "simple" BG texts. An example of a distal article is possibly attested in a letter from Basarab IV the Young: vaše+no molenie 'your prayers' (№ 76), but the text is not very clear in general.
The underlying endings are those of pronominal stems (§2.2.5-6): hard for t- and on-, soft for s-. The endings in later sources also reflect phonetic changes discussed above, with ambiguities resolved by secondary tags (e.g. toǫ zastǫplenie 'her protection'; Zogr.107; toǫ: Pp3fsi, Pp3fsg).

	
	3.2.2. CS short demonstratives (Pd)
		sg			dl	pl
		m	f	n
	n	sь/sii	sia	sie	sia/sii	sii/się/sia
	g	sego	seę	sego	seju	sixь
	d	semu	sei	semu	sima	simь
	a	[sego]	sijǫ	[se]		się/[się]/[sia]
	l	semь	[sei]	semь		[sixь]
	i	simь	sejǫ	simь		simi

		sg			dl	pl
		m	f	n	
	n	onъ	ona	ono	ona/oni	oni/ony/ona
	g	onogo	onoę	onogo	onoju	oněxъ
	d	onomu	onoi	onomu	oněma	oněmъ
	a	[onogo]	onǫ	[ono]		ony/[ony]/[ona]	
	l	onomь	[onoi]	onomь		[oněxъ]
	i	oněmь	onojǫ	oněmь		oněmi

		sg			dl	pl
		m	f	n	
	n	tъ	ta	[to]	ta/ti	ti/ty/ty
	g	togo	toę	togo	toju	těxъ
	d	tomu	toi	tomu	těma	těmъ
	a	[togo]	tǫ	[to]		tę/[ta]
	l	tomъ	[toi]	tomъ		[těxъ]
	i	těmъ	tojǫ	těmъ		těmi
	
	
	t-stems in later sources
		sg				pl
		m	f	n
	n	tь	ta	[to]		tě
	g	togo	toe	togo		těxь
	d	tomu	toi	tomu		těmь
	a	[togo]	tu	[to]		[těxь]
	l	tomь	[toi]	tomь		[těxь]
	i	těmь	toju	těmь		těmi
	

3.2.2. extended demonstratives

Another type of demonstrative common in BG/MK varieties is extended by a suffix after the inflected ending. The first part of the demonstrative usually shows a t- or an on-stem with distinctive dependent and dative endings:

	
	znamenie tomuzi deto šte da dodi (Berl.d.)
	'a sign for that, what shall come'
	

Such inflections are considered archaic today. The on-stem variant was considered to be functionally equal to the t-article the damaskini by Miletič (1923:28). Common variants of the extension are -ja, -va and -zi. Their distribution is more or less dialectal. Damaskini prefer -zi in most singular forms (like znamenie tomuzi), while Punčo prefers -ja or -va (e.g. tomuva pravedniku 'to that righteous man'; PPS 017). The -zi is also common in the Legend of Troy (e.g. here). The -ja forms are common in plural of both damaskini in Punčo with exception of Berl.d. (e.g. vъ tezi raboti 'in these things' here).
Lemmatization is always based on the extension: the variants are toja, tova and tozi (or onja, onova and onzi). The extensions are usually not marked for any nominal category like case or gender; in the literature (e.g. Mirčev 1978:183) they are considered "particles" (častici). An exception are the f.sg.dep forms tuju and onuju, occasionally used by Punčo (e.g. onuju ženu 'that woman' mentioned above; PPS 007). Sometimes the extension is truncated (e.g. vъ onui vreme 'in that time'; Berl.d.). Some of these forms may reflect an older stage. The extension -ja was likely marked for the gender earlier; Mirčev (1978:182) mentions forms like tъi, taja and toje already attested in OCS (but not in our corpus). The first of these is used as an adverb in "simple" BG texts (e.g. tьi se běše podkanila 'thus she had decided [to live]'; Tixon.d.), but in sentences like i tъi dumaše 'and thus (an angel) spoke' (Berl.d.) the demonstrative meaning may be denoted too.
All such extended demonstratives, even if used as nominal arguments (e.g. tagged "nsubj", "obj" etc.), are marked with an UD tag extension "ext".

	
	3.2.3. extended demonstratives (Pd)
		sg				pl
		m	f	n	
	n	toja	taja	tova		tezi
		tozi	tazi			tija
		toizi
	g	togova		togova
		togozi		togozi
	d	tomuva		tomuva
		tomuzi		tomuzi
	a		tuju
	o		tьzi

		sg				pl
		m	f	n	
	n	onja	onaja	onova		onezi
		onzi	onazi			onija
	g	onogova		onogova
		onogozi		onogozi
	d	onomuva		onomuva
		onomuzi		onomuzi
	a		onuju
	o		onьzi
	

3.3. relative pronouns

3.3.1. simple "BG" relative/interrogative pronouns
3.3.2. simple "BG" marked relative (etc.) pronouns
3.3.3. CS relative/interrogative pronouns
3.3.4. CS marked relative (etc.) pronouns
3.3.5. pronoun iže

Balkan Slavic varieties show considerable differences in handling of other pronominal categories. This fact can be observed both dialectally and diachronically; some of the differences have been integrated into BG/MK standards. For example, Bulgarian uses unsuffixed stems like koi 'who' as interrogatives, extending them by the suffix -to when used in a relative sense, as in Botev's verse toz, kojto padne v boj za svoboda 'he, who falls in the fight for freedom'. The suffix is already attested in 14th century (Mirčev 1978:187, e.g. here). In Macedonian, the pronoun without a suffix can be used in both interrogative and relative senses, especially when it follows a preposition (e.g. kniga, bez koja nemožam da rabotam 'a book, without which I cannot work'; Lunt 1952:44). The form kojšto, used in modern MK, is not attested in the corpus. Possibly, one sentence in the Legend of Troy may show an archaic form of it:

	
	azь imamь kto+šte naše slъzy utoliti (Vat.slav.2)
	'I have someone to swipe our tears'
	

Beside these, uninflected pronouns like što (< CS čьto 'what') or deto (< kъde+to 'where') can be used instead. In our sources, the options are often free and tend to mix. For example, Punčo uses the relative suffix -to mostly in chapters shared with the damaskini. Elsewhere, he prefers to use koi in both interrogative and relative senses when denoting a person, or simply deto as a general relative pronoun:

	
	ljudie, koito xodatь po volja bžia (PPS 013)
	'people, who walk under God's will'
	
	dovedoxa togova člveka deto beše slepь (PPS 014)
	'they brought the man, who was blind'
	
	ni pomeni koi čini bezakonie (PPS 024)
	'nor take heed of those, who commit lawlessness'
	

3.3.1. "simple" BG relative/interrogative pronouns

Generally, syntactic annotation shows better the difference. In the corpus, all three variants act as markers introducing a subordinate sentence, and thus carry the same UD tag "mark". Morphologically, only koito ("Pr-mpn") and deto ("Pr") are tagged as proper relative pronouns, as these forms do not appear in the other role. On the other hand, koi has a PoS tag of an interrogative pronoun ("Pq-msn"). In other words, the tag "Pr" marks the presence of the -to marker. In the same way, we can also distinguish forms like koga and kogato 'when', kъde and kъdeto 'where', and so on. A similar case is the Punčo's use of kakvo ('what' or n.sg 'what kind of' in st.BG) in the modal meaning 'how' (e.g. here; st.BG kak).
The PoS tag of relative pronouns usually includes information on gender and number, but it is basically inferred from the context or related elements in the main clause (e.g. ljudie: "Nmpny", so koito: "Pr-mpn"). Most sources make no difference between m.sg кой and plural кои, like standard BG. In interrogative clauses, the basic form koi should be handled as unmarked for gender and number (e.g. koi ede tija gozbi 'who eats that food?'; PPS 010; tagged "Pq---n"). Gender-specific forms like koja only rarely appear as a nominal argument in the sentence (koja ti e nevolja 'what kind of worry besets you'; PPS 009; tagged "Pq-fsn"). When they are used adnominally, like in ot koi si ljudie 'from which people are you?' (PPS 014), an adjectival tag is used instead ("Ampnn"), as already discussed above in this section.

	
	3.3.1. "simple" BG relative/interrogative pronouns (Pq)
		sg				pl
		m	f	n		
	n	koi	koja	koe		koi
	g	kogo
	d	komu
	a		koju
	
	kind	kakъv (f kakva, n kakvo, pl kakvi)
	mode	kak
	number	kolko
	place	kъde, gde
	time	koga, kogi
	thing	što
	

3.3.2. "simple" BG marked relative (etc.) pronouns

In the same way as the annotation handles pronouns using -to extension as ones specifically "marked" as relative, also indefinite pronouns are denoted by the prefix ně- and negative by the ni-. Under relative pronouns ("Pr") are classed also some pronouns built by the demonstrative root t-. These commonly show suffixes, which are employed on extended demonstratives, i.e. -ja, -va and -zi. These include ones used adnominally, like takьvja člvekь 'such a man' (PPS 014; tagged "Pr-msn"; always tagged as "Pr", when showing a non-inflected extension), as well as temporal adverbs like togazi 'then' (e.g. here, or Moesian variant togizi here; tagged just "Pr"), famously used for dialectal analysis of damaskini translations by Velčeva (1964). These variants are dialectal, usually appearing consistently in a text.

	
	3.3.2. "simple" BG marked relative (Pr), indefinite (Pi) and negative (Pz) pronouns
	
		Pr		Pi		Pz
		-to/t-		ně-		ni-
		
	n	koito		někoi		nikoi
	g	kogoto		někogo		nikogo
	d	komuto		někomu		nikomu
	
	kind	kakъvto		někakъv		nikakъv
		takъvzi
	mode	kakto		někak		nikak
		taka
	number	kolkoto		několko		nikolko
	place	(kъ)deto	někъde		nikъde
	thing	štoto		nešto		ništo
	time	kogato		někoga		nikoga
		togazi	
		togizi
	

Punčo sometimes does not clearly distinguish endings and extensions, as we could see above with the demonstrative onuju. This confusion is well seen on his use of adjectival pronoun takъv 'such', which sometimes shows even adjectival long-form endings. Hence the pronoun is tagged as an adjective (e.g. takovoe: tagged "Ansny"; PPS 001) in the Sbornik. The extension -va is not old: the form takovъ is already attested in OCS (cf. BAN VII 771). Still, Punčo often assimilates the second vowel to the ending, using extensions -ja in m.sg and -va in plural. In contrast, Berl.d. consistently uses takъv- (or tъkъv-) as a stem, with an extension -zi after the inflected ending (m.sg forms are attested on 174v and 196r, not included in the corpus).

	
	attested forms of takъv 'such'
	PPS
		m.sg	f.sg	n.sg	m/f.pl	n.pl
	n	takьvja	takava	takovoe	takviva	takvava
		takovi	takvava	takvovo	takiva
		takovь			takovo	
	g	takovago		takovago
	d	takvovomu
	a		takovu
			takuvu
	Berl.d.
		m.sg	f.sg	n.sg	pl
	n	tak͛vъzi tak͛vazi tak͛vozi	tak͛vizi
	g	tak͛vazi
	

3.3.3. CS relative/interrogative pronouns

Church Slavonic uses of k- stem for both interrogative and relative meaning. Sg.nom form of the personal 'who'-pronoun was in OCS kъto, besides adjectival kyi 'which'. Analogic levellings of oblique forms of the latter (like koego) seem to have replaced both with the modern form koi seen above already in Middle BG period (Mirčev 1978:186, e.g. here). Similarly, the inst form cěmь was replaced by kyimь (cf. Lunt 2001:63). In the Legend of Troy we probably see the next stage of levellings with forms like m.sg.inst pod koimъ 'under which' (Vat.slav.2), as well as the relative marker -to (e.g. here).
The replacement of root jer in kogda 'where' likely reflects influence of forms like kogo, kotorъ, koliko, etc., or of East Slavic redactions. In vernacular, the root jer was likely elided, as attested by forms like deto or gde. Such forms are attested in our CS sources too (tou něgde 'somewhere there'; Vuk.1536).
Old čьto 'what' is rarely seen in the damaskini (e.g. here), but it was likely shortened to more common što already. Both kъto and što were inflected for case. The same can be said about kyi, kotorъ 'which' (usually long kotory; e.g. here), and even koliko 'how much' (e.g. kolici mnogaždi 'how many times'; Zogr.107).
Limited inflection can also be observed on pronouns denoting place: besides the kъde denoting static location, there were also an ablative (denoting origin: e.g. ot koudu bě 'whence she came'; Vuk.1536) and lative (direction: ne imějaše kamo děnouti se lit. 'he could not place himself anywhere'; Vuk.1536) forms. In the corpus, these are tagged to reflect their use in modern varieties, i.e. mostly as adverbs ("R"), anyway. In CS texts, the analysis is indicated in the secondary tag (e.g. kamo: "R, Pq---a"). Also see §6 for more details.

	
	3.3.3. OCS relative/interrogative pronouns (Pq)
	
		person	thing		kind
					sg	pl
	n	kъto	čьto		kyi	ci
	g	kogo	česo		koego	cěxъ
	d	komu	česomu		koemu	cěmъ
	a	[kogo]	[čьto]		[koego]	[ci]
	l	komь	čemь		komь	[cěxъ]
	i	cěmь	čimь		cěmь	cěmi
	
	kind	kyi, kotorъ
	mode	kako
	number	koliko
	time	kъgda (> kogda)
	
	place
	n	kъde
	g	kǫdu (> kudu)
	a	kamo
	
	
	attested personal forms in the Legend of Troy
		sg			pl
		m	f	n	m/n	f
	n	koi	ko(j)a	koe	koi	koǫ
	g	koego	
	d	komu	koei
	a	koego	koǫ
	l	koemъ	
	i	koimъ	
	
	

3.3.4. CS marked relative (etc.) pronouns

Similarly as in modern varieties (and also outside Balkans), these can be modified by prefixes ně- and ni- for more specific roles. Negative pronouns often show an extension -že, e.g. i ničtože ot nyxь ino slyšaše se 'and nothing else was heard from them' (Vuk.1536). The same extension is used with most relative pronouns. In a similar way, these morphems could be combined with various demonstrative and other stems like sь 'this', inъ 'other', vьsь 'every, all' etc. (e.g. vьsǫdu 'everywhere'; Zogr.107). For the variation between adverbial pronouns denoting place (e.g. tu-tǫdu-tamo) see §6.
To mark identity ('the same'), a demonstrative root t- (§3.2.2) is used with the extension -žde. These are handled as relative pronouns too, e.g. vь tьžde dnь 'on the same day' (BAR 287; tagged "Pr-msn").

	
	3.3.4. OCS marked relative (Pr), indefinite (Pi) and negative (Pz) pronouns
	
		Pr			Pi		Pz
		-že/t-	t-žde		ně-		ni-že
			
	person	iže	tъžde		někъto		nikъtože
	kind	takyi	takyžde		někyi		nikyiže
					někotorъ	nikotorъ
	mode	tako	takožde		někakъ		nikakože
					eliko
	time	togda	togdažde	někogda		nikogdaže
					egda
	thing	to	tožde		něčьto		ničьtože
	
	place	
	n	ideže			někъde		nikъde
		tu	tužde		něgde
	g	[n]ǫduže	
		tudu	tudužde
	a	jamože					nikamo
		tamo	tamožde
	

3.3.5. CS pronoun iže

A characteristic CS pronoun is iže, which is used primarily as a adnominal clause marker. Similarly as later koito, it usually agrees with its head (or referent) in gender and number, while the case is determined by the sense of the clause, e.g. vьsa jaže kь čьsti oustroi 'he duly did everything for honor' (Vuk.1536). However, iže often appears as a general form (like the later deto), referring to nouns of various genders and numbers (e.g. here), as mentioned above. Typical of later CS redactions is also the placing of adnominal clauses in front of the referent, e.g. iže vsa nazirajušteje oko 'only that all-seeing eye' (Vuk.1536). Maybe for this reason some medieval grammarians (cf. Jagić 1895:334) considered iže to be an equivalent to Greek article (različie). N.sg form eže is often found in clauses with infinitive verbs with the meaning 'in order to' (e.g. eda obrěštet se kto eže poučiti me 'could there be found anyone to teach me?'; Kiev.d.), not unlike the particle da, typical of "simple" varieties - and thus likely avoided in CS (cf. da ma nauči in Sv.d.). Morphologically, the pronoun usually shows the same *jь-stem used in long-form and soft pronominal adjectives (§2.2.3. and §2.2.6.):

	
	3.3.5. OCS pronoun iže 'who' (Pr)
		sg				dl		pl
		m	f	n
	n	iže	jaže	eže		jaže/iže	iže/ęže/jaže
	g	egože	eęže	egože		ejuže		ixže
	d	emuže	eiže	emuže		imaže		imže
	a	[n]ьže	jǫže	[eže]				ęže/[ęže]/[jaže]
	l	[n]emže	[eiže]	[n]emže				[ixže]
	i	imže	ejǫže	imže				[n]imiže
	

4. Numerals

Although traditionally considered a separate word category, numerals morphologically differ much from each other. Among cardinal numerals, edin, the word for '1', has forms for all genders, and also a plural form; dva '2' is a distinct masculine form (f/n.sg dve). In our sources, the former may also appear with neuters (e.g. dva kozlišta 'two goat kids'; PPS 026). Typical for Balkan Slavic varieties are - as in nouns - marking of definiteness and count forms, used mostly together with masculine persons. The article usually reflects the gender and number. In some of our sources, the article on numbers may show the form -těxь, its old plural dependent (or gen/acc) form. Such forms are still productive in the dialects of the Northeast (Maslov 1981:181), from whence we have also the following example:

	
	etu sasъ takvizi pakosti, Avva Zosima, paštaxъ v sidim+ na+ dise+těxъ onizi godini
	'thus, I was annoyed by such evils, o Father Zosima, for those seventeen years' (Sv.d.)
	

The count (or, rather, "personal") forms of numbers, showing endings -ama or -ima, appear from the 14th century. Its origin was a generalized dl.dat/inst form dъvěma, used for pairs of a man and woman, later spreading to larger groups (Mirčev 1978:194). Even if the number is higher, the word uses the "dual" (or f.sg? n.pl?) form of the article -ta, e.g. trima+ta ierarxi 'the three hierarchs' (NBKM 1064). It appears already in earlier damaskini, but in 19th century sources it becomes more frequent:

	
	i na utrě izlězoxa i dva+ta (Tixon.d.)
	i na utrina+ta izljazuxa i dvama+ta (NBKM 1064)
	'and the both went out in the morning'
	

The morphological character of these forms is not fully clear; their emergence likely follows the general tendency to mark objects high in the animacy scale. In fact, Punčo uses also a f.dl form dveme for a pair of women (e.g. here). However, the forms sometimes show interaction with derivations denoting groups, being functionally the same (cf. Mirčev 1978:194). In our corpus, such constructions are handled as nouns with their own lemmas; they may, however, be tagged as a numeral modifier ("nummod") in the syntactic annotation. One such common derivation is stotina '(a) hundred'. Punčo uses troica (e.g. here), a word common in CS for groups of three, and the suffix -ina for a group of five:

	
	izbraxu se petina+ta braikja naedno (PPS 022)
	'the five brothers assembled themselves'
	

In Church Slavonic, the inflection of numerals was diverse too (cf. Lunt 2001:153). Numbers 1-4 were inflected like pronominal stems; counted words agreed with them in number and case, e.g. vъ xramě triexъ Stitelei 'in the Temple of Three Apostles' (Rost.1689). On the other hand, 5-10 (also when in constructions involving teen-numbers and units), 100 and 1000 were handled as nouns, while modified elements were in the pl.gen case, e.g. sedmь na dsetь lět 'seventeen [lit. seven-over-ten] years' (Kiev.d.). In later sources, these rules were not always so easy to follow, as the genitive lost its partitive function. Thus, we may see pl.nom after higher numbers (e.g. viděx dsetь junoše 'I saw ten boys'; Kiev.d.) or pl.gen after smaller ones (za *g* lět' 'for 3 years'; ODNB 1/112), and also dl.nom - the modern BG/MK count form (za *i* msca 'for 10 months'; NBKM 326). Old genitives are preserved in designations of hundreds from 400 (BG četiristotin); following numbers 2 and 3 they show plurals in BG (e.g. trista '300'), while MK standardized the dual form dveste for '200'. Punčo's Sbornik represents a compromise between the two standards:

	
	žive adamь dve+ste i tri+desetь godini (PPS 024)
	'Adam lived for 230 years'
	
	noja beše člvekь ot petь stotinь godini (PPS 025)
	'Noah was a five hundred years old man'
	

As it can be seen above, compound numerals are split into multiple tokens in our corpus. The sources themselves - as it can be seen on dve+ste and petь stotinь in the examples above - are not consistent in this respect, although the use of articles (like sidim+na+dise+těxъ) and building of ordinals (e.g. četirnaisie 'the 14th'; NBKM 1064) indicates, that they indeed form a single morphological unit. Teen numbers are split into phrases, in which the unit represents the head, and the 'ten' the dependent. In numbers from 20 and higher, the last element of a numeral phrase not following a preposition or conjunction is the head. Units and lesser powers of ten are then handled as conjuncts. To show it schematically:

	
	analysis of compound numerals
	
	- powers of 10
	požive meleleilь svi dni negovi leta osmь stotinь i devet desetь i petь (PPS 024)
	'Mahalalel lived through all the days of 895 years'
	
	leta		1	0	root
	osmь		2	3	nummod
	stotinь		3	1	nummod
	i		4	6	cc
	devet		5	6	nummod
	desetь		6	3	conj
	i		7	8	cc
	petь		8	6	conj
	
	- teens
	takvizi pakosti paštaxъ [...] v sidim+ na dise+těxъ onizi godini (Sv.d.)
	'I was annoyed by such evils for those seventeen years'
	
	v		1	7	case
	sidim+		2	7	nummod
	na		3	4	case
	dise+		4	2	nummod
	těxъ		5	4	det:p_adj
	onizi		6	7	det:ext
	godini		7	0	root
	

Cardinal numerals are annotated with a PoS tag Ml, the second character denoting the "letter" type. Other types reflect different arithmetic systems used in the texts to represent numbers. Most commonly this means the Cyrillic number system, which has been in use up to the 19th century, when misreadings like naxama (i.e. *#axma* '1641'; NBKM 728) start to appear. The tag might also be used for Greek or Hebrew numbers, if they are ever attested. Roman and digital (Arabic, Indian) numerals are rather rare in the corpus.

	
	Ml	letter		četiri
	Mc	alphabetic	*d*
	Md	digital		4
	Mr	Roman		IV
	

Only proper cardinal numerals are tagged "Ml". Derivations denoting groups using -ica, -ina or -oe suffixes are analyzed as nouns. Ordinal numerals (e.g. treti 'third') are handled as adjectives. Both classes are also lemmatized separately. Word edin is also handled as an adjective, using the hard pronominal stem paradigm (§2.2.5). Its function is in most cases a pronominal one, as a marker of indefiniteness. In the numeric sense, a secondary tag "Ml" can be used, although it is more clearly marked in the syntactic annotation ("nummod"). Other inflected numerals - count forms, dva in modern texts, and others in CS - may use secondary adjectival tags too.

	
	attested forms and derivations of dva 'two' (Ml/A) in PPS
		
	dva		Ml, Amdnn/Anpnn
	dvama		Ml, Amddn
	dve		Ml, Afdnn/Andnn
	dveme		Ml, Afddn
	dvamina		Nfsnn
	dvoe		Ansny
	dvoica		Nfsnn
	

5. Verbs

5.1. present tense
5.1.1. "simple" BG present tense
5.1.2. CS present tense
5.1.3. present tense of irregular verbs

5.2. aorist & imperfect
5.2.1. "simple" BG aorist
5.2.2. "simple" BG imperfect
5.2.3. CS aorists
5.2.4. CS imperfect

5.3. periphrastic tenses
5.3.1. perfects
5.3.2. future tenses

5.4. other moods
5.4.1. conditional
5.4.2. imperative
5.4.3. infinitive

5.5. participles
5.5.1. "simple" BG participles
5.5.2. CS present participles
5.5.3. participial declension

5.6. irregular verbs

From the point of view of dependency grammar, a verb is the root of sentence, its most important part. However, its analysis is not easy in Slavic varieties. One problem is the aspect, which cannot be sufficiently described from the point of view of morphology. Such a situation can be already seen in Church Slavonic. Analysis into basic constituents like stem and ending does not help, as many verbs show two (present and aorist/infinitive) stems; some verbs, like byti 'be', show even more (1sg.prs esmь, 3pl sǫtъ, 2/3sg.aor bě or bystъ). Most of them could be classified into a couple of classes according to auslaut of the aorist/infinitive stem (cf. Lunt 2001:85). Excluding irregular verbs (mostly using old consonant stems), only two sets of thema vowels in present tense constructions appear, i.e. -e- (and -ǫ- in 1sg/3pl.prs) on the one hand, or -i- (and -ę- in 3pl.prs) on the other.

	
	aor/inf stem					prs stem
	-ø-		rešti (< *rek-ti) 'say'		-e-	rekǫ, rečeši, rekǫtъ
	-ě-		sěděti 'sit'			-i-	sěždǫ, sědiši, sědętъ
	-ěj-		smějati 'dare'			-e-	smějǫ, směeši, smějǫtъ
	-i-		xvaliti 'praise'		-i-	xvaljǫ, xvališi, xvalętъ
	-a-		lěgati 'lay down' (iprf.)	-e-	lěgajǫ, lěgaeši
	-nǫ-		legnǫti 'lay down' (prf.)	-e-	legnǫ, legneši, legnǫtъ
	-ova-		kupovati 'buy'			-e-	kupujǫ, kupuješi, kupujǫtъ
	-Ca-		kazati 'show, teach, tell'	-e-	kažǫ, kažeši, kažǫtъ
	-j-		biti 'beat'			-e-	bьjǫ, bieši, bьjǫtъ
	irreg.		byti 'be'				esmь, esi, sǫtъ
			dati 'give'				damь, dasi, dadǫtъ
			iměti 'have'				imamь, imaši, imǫtъ
			vъzęti 'take'				vъzьmǫ, vъzьmeši, vъzьmǫtъ
	

In the following centuries, 2/3sg and 1/2pl.prs thema vowels did not change very much. For this reason, modern grammars (e.g. Lunt 1952:73, Maslov 1981:211, Nicolova 2008:271), and also our dictionary, classifies verbs according to them. Most lexemes, which could be classified as e- or i-verbs in OCS, retain this conjugation well until today. An innovation affected former e-verbs with a- or ova-stems in infinitive, which were taken over to present paradigms, changing the thema vowel in 2/3sg and 1/2pl.prs forms to -a- (e.g. lěgaeši > ljagaš). The same thema vowel can be observed on loanwords showing the sequence -isa- in the stem (e.g. kurtulisam 'be saved'; NBKM 1423). These form a new class, which we designate a-verbs, also called "third" conjugation in modern grammars.

	
	e-verbs		reka, rečeš
			smeja, smeeš
			legna, legneš
			kaža, kažeš
			bija, bieš
	i-verbs		sedja, sediš
			xvalja, xvališ
	a-verbs		ljagam, ljagaš
			kupuvam, kupuvaš
			imam, imaš
	irreg.		sъm, si
			dam, dadeš
			vzema, vzemeš
	

In traditional grammars, an Indo-European verb form shows following features: person (BG lice), number (čislo), tense (vreme), voice or diathesis (zalog), mood (naklonenie), and, in Slavic varieties, aspect (vid). Slavic verbs tend to use synthetic expressions sparingly for them. Voice is marked only on participles, the mood and many tenses only periphrastically. For example, Church Slavonic expressed future tense and some conditionals of perfective verbs using the present tense form. Compare the following CS sentence with its translation in a damaskin:

	
	kto oubo toje po česti izrečetь dějania že i čjudesa (Vuk.1536)
	koi šte da može da iskaže neini+te počesti, raboti i i čjudesa+ta (Berl.d.)
	'who would tell about her works and miracles in detail?'
	

Formally, šte, može and iskaže are all marked for 3sg.prs present tense too, with particles da marking dependency between the auxiliary and main verb. The difference is only that šta became obligatory as an auxiliary verb for future tense/conditional marking - in CS, auxiliaries are used only with imperfective verbs, like gl[agol]ati xotěxъ 'I would have spoken' (Zogr.107). However, as šta is a more explicit future tense marker, the editor deemed it necessary to add može to emphasize the specific modal meaning, lit. 'who will be able to'. In the earlier translation, the meaning is fully modal by the use of rare infinitive clause:

	
	neino čjudo i neinь pos[t] koi ezykь može iskaza (Tixon.d.)
	'which tongue can tell of her miracles and fasting?'
	

Still, both modality denoted by može and future tense denoted by šta are features realized on the syntactic level. From the point of view of the ending, Balkan Slavic verb contrasts with the noun and its case syncretism. Besides person and number, it also preserved old marking of two past tenses, aorist and imperfect. From the Slavic perspective, it is not common. Distinctive aorist and imperfect forms are preserved in Serbo-Croat even better than in most BG/MK varieties (i.e. still with 2/3pl.aor forms -ste and -še), but their use is very limited (Tomić 2006:364). Upper Sorbian also retains the aor/impf forms, but with a different function (Scholze 2007:182). The resultative or l-participles, common in other Slavic varieties in past tense constructions, are neither finite verb forms, nor necessarily marking past. Such examples can be found in our sources too:

	
	da te blgoslovi dša moę dogde ne+smь umrelь (PPS 026)
	'may my soul blesses you before I am dead' 
	

In this sentence, l-participle actually stands on a place of a future or conditional construction in CS (prežde daže ne oumru; Gen 27:4, My-Bible). It is used (especially with the auxiliary based on CS by) in conditionals elsewhere in Slavic too, but also, with other auxiliaries or without them, in sentences marked for evidentiality (ne znael če si dša ta pogubuva 'he did not [seem] to know he was losing his soul'; PPS 017) or narrativity (da skažeme čudesa što se e pravilo 'let us tell about the wonders which were done'; PPS 011), past perfect tense (e.g. beše dalь 'he had given'; PPS 006), and, today, even as adnominal modifiers (Nicolova 2008:435). In other words, the exact function of individual morphological expressions are not apparent from the verbal form, without analysis on the syntactic level.

The verbs are lemmatized according to BG tradition, using the 1sg.prs form, as these are mostly used in relevant dictionaries. Despite the advantage of showing the thema vowel, the 3sg form (as in MK) is used only if the verb is available only in MK sources or commonly used in BG too (like trěbva 'need'). Lemmas specific for CS (e.g. glagolati 'speak') are given in infinitive. Aspectual variants (e.g. legna and ljagam 'lay down') are given as separate lemmas. A PoS-tag of a verb begins with V. Following letter denotes the syntactic type (m/a, i.e. main or auxiliary), which is inferred from the sentence structure and meaning. Important for the rest of attributes is the verbal form, which determines the paradigm used for conjugation. Among these, conditionals particles are always tagged as auxiliaries:

	
	tag	form		examples
	Vmi	indicative	sъm, beše
	Vmm	imperative	bъdi
	Vmn	infinitive	byti
	Vao	conditional	bi
	Vmp	participle	sǫštii, bilь
	

The next attribute tense denotes the specific form expressed by the ending. It is relevant for indicative finite verbs and participles. Participles are tagged as adjectives too (cf. §2). As already mentioned, one of three tenses can be explicitly marked (a/i/p - aorist, imperfect or present) on a token. As indicatives show many homographic forms from this respect, syntax may be consulted to determine the form; if unclear, secondary tags are used (e.g. for gošteva 'treat as a guest' here). Future ("fut"), perfect ("prf") and past perfect ("pprf") tenses are tagged on the auxiliary as UD tag extensions. Person (1/2/3) is relevant for indicative, imperative and conditional verb forms; number (s/p/d) for all but infinitives. The final position denotes aspect. Only two options are included (e/i - perfective or imperfective), but other meanings or biaspectuality may be indicated in the description of the lemma.

	
	pisaxь - PoS-tag: Vmia1si, lemma: piša
	V	verb		property of lemma
	m	main		property of syntax
	i	indicative	property of ending and syntax
	a	aorist		property of ending and syntax
	1	1st person	property of ending
	s	singular	property of ending
	i	imperfective	property of lemma
	

5.1. present tense

5.1.1. "simple" BG present tense
5.1.2. CS present tense
5.1.3. present tense of irregular verbs
Present forms show dialectal variation especially in 1sg and 3pl. In standard Macedonian (Lunt 1952:73) and some Western BG dialects, the ending -m following a thema vowel has been generalized as the 1sg form. From the historical perspective, the innovation can be dated to 14th century, and it was also shared with Serbo-Croat and Central Slovak (Krajčovič 1988:131). The ending can be well observed in Punčo's Sbornik with a high degree of consistency - in about 242 out of 299 1sg.prs main verbs in first 35 chapters. The endings appear in verbs of all stem classes:

	
			1/2sg.prs in OCS	1sg.prs in PPS
	-ø-/-e-		rekǫ, rečeši		rečemь
	-ě-/-i-		sěždǫ, sědiši		sedimь
	-ěj-/-e-	smějǫ, směeši		smeju (011), smeemь (013)
	-i-		xvaljǫ, xvališi		falim
	-nǫ-/-e-	legnǫ, legneši		(dosegnemь 'reach')
	-Ca-/-e-	kažǫ, kažeši		kažemь
	-j-/-e-		bьjǫ, bieši		biemь
	irreg.		esmь, esi		smь
			damь, dasi		dademь (022), damь (ibid.)
			imamь, imaši		imamь
			vъzьmǫ, vъzьmeši	zememь
	

5.1.1 "simple" BG present tense

In the damaskini, on the other hand, the ending -m appears mostly on irregular and a-verbs. In about 13th century (Mirčev 1978:207), sequences -a- and -ova- became productive as suffixes, modifying aspect of originally perfective verbs, e.g. CS kazati 'show, tell (prf.)' > BG 1sg.prs kazvam 'call (iprf.)' (e.g. here; besides prf. BG 1sg.prs kaža). However, sometimes they can be seen on other verb classes too:

	
	zašto ne+mogu da trьpim azь za mnogo toizi smradь (Tixon.d.)
	'because I cannot suffer that stench anymore'
	

The distribution was likely based on lemmas. The i-verb da trьpim 'to suffer' has been changed by the Ljub.d.-editor, using a vowel ending (i.e. trьpę, here), which is plausible from the more Eastern origin of this edition (cf. Demina 1968:57). Another common verb is the descendant of OCS iskati 'want', 1sg.prs ištǫ, where the preferred forms seem to develop across centuries, with present-day form iskam taking over in the latest source:

	
	Tixon.d. (s.242 f.)	Ljub.d.  (s.241 f.)	NBKM 1064 (s.251 f.)
		
	ne+ištu vamь		ne+išta ot vas		ni+išta as ut vas
	ami vy ištemь		ami vy išta		ami vi iskam
	tova ištem		tova ištem		tjax iskam
	

Both 1sg and 3pl were originally affected by denasalization (§10.6) with different reflexes according to both area and orthography. Damaskini are often inconsistent in marking the reflex of the nasal:

	
	ot se+to da sa otdelě, toko štoto želaę azъ da poluča (Berl.d.)
	'may I cast away all of that, just to receive what I desire'
	
	a dvci+te štoto idatъ podirъ crca+ta, i vъrvъtъ podirъ neę (Berl.d.)
	'and the virgins, who follow the Queen, and walk with her'
	

The variation is likely caused by lack of fixed orthography for underlying phonems. When the stem auslaut is hard (like in idatъ), a phonetic jer (for /ă/, cf. §1.1.3) or -a (preferred in present-day BG/MK orthography) appear. Soft-auslaut stems show often -e, -ě or -ę, all possibly representing the sequence /jă/. Distinction of juses typical of Tarnovo school (§10.12) may have had an influence on the choice of vowel too. However, poluča and vъrvъtъ (both typical i-verbs) do not fit into theory. Sometimes also -u (or -ju) appears, a reflex typical of Resava school, e.g. mlju se 'I pray' (Tixon.d.). Punčo's Sbornik oscillates between -atь and -utь as 3pl.prs endings. Rarely, even a shortened ending -u appears:

	
	oni se povrьnuxu da idu na pazarь da si kupatь maslo (PPS 021)
	'they returned to go to the market to by oil'
	

Punčo uses endings -atь and -utь with roughly equal frequencies (106 -atь vs 123 -utь in first 35 chapters). All verbs showing the shortened ending -u are also attested with "full" ones too (e.g. both idatь in PPS 017, idutь in PPS 021).
Slavic varieties with a -m in 1sg.prs forms also show an extended 1pl ending. 1pl.prs ending in the Sbornik can be -me (more common; e.g. ne+znaeme 'we do not know'; PPS 007) or -mo (da pokažemo 'let us show'; PPS 006). A lone -m appears very rarely in fixed CS phrases (e.g. stanemь dobrem stanemь so straxomь 'may we stand up well, stand with fear [of God]'; PPS 013). In such cases, the verb also shows secondary PoS tags (for both stanemь: "Vmip1pe" and "Vmip1se"). In the same way we handle 1sg.prs -m endings appearing on e- and i-verbs in damaskini sources (like here).

	
	5.1.1. "simple" BG present tense (V?ip)
	e-/i-verbs
		sg		pl
	1	-a/-u/-ь	-▼-mь
		-e/-ě
	2	-▼-šь		-▼-te
	3	-▼-ø		-atь/-utь/-ьtь
				-etь/-ětь
				
	a-verbs
		sg		pl
	1	-a-mь		-a-me
	2	-a-šь		-a-te
	3	-a-ø		-atь

	conjugation in PPS
	all regular verbs
		sg		pl
	1	-▼-mь		-▼-me/-▼-mo
	2	-▼-šь		-▼-te
	3	-▼-ø		-atь/-utь
	
	▼ - thema vowel specific for the verb class (a/e/i)
	

5.1.2. CS present tense

The main difference between modern varieties and CS is the use of present forms of perfective verbs, which have a future tense meaning. However, the future meaning could be emphasized by an auxiliary like xotěti 'want' already in OCS (cf. below §5.3).
Within the paradigms, the 2sg ending -ši leaves no trace anywhere else in Slavic - it was possibly a local innovation in the dialect around Salonica, while the 3sg ending -tъ was likely truncated in late OCS period (Mirčev 1978:10). The both endings appear regularily used; vernacular forms are rarely attested (e.g. da ne čjue glas 'to not hear any voice'; ODNB 1/112). 1sg and 3pl forms show variation typical for the respective CS redactions (§10.6, §10.12). The choice of jus in texts of Tarnovo redaction does not seems to follow any clear rules in our sources; soft stems show both -ǫ and -ę.

	
	vьsě usrъdstvuǫ dati (...) da želaemoe mně polučę skrovište (Zogr.107)
	vьsa ousrьd'stvouju dati (...) da želaemoje mně poloučju skrovište (Vuk.1536)
	'I am ready to give away everything (...) to receive the desired treasure'
	

Sources of Resava redaction like Vuk.1536 mark palatality of the ending more consistently, although there are exceptions, when "hard" -u appears on i-verbs (e.g. sьtvoru mnoga dobra 'I will do much good'; NBKM 326).

	
	5.1.2. OCS present tense of thematic verbs (V?ip)
	e-verbs
		sg		pl		dl
	1	-ǫ		-e-mъ		-e-vě
	2	-e-ši		-e-te		-e-ta
	3	-e-tъ		-ǫtъ		-e-te
	
	i-verbs
		sg		pl		dl
	1	-jǫ		-i-mъ		-i-vě
	2	-i-ši		-i-te		-i-ta
	3	-i-tъ		-ętъ		-i-te
	
	present tense forms in later sources (V?ip)
	e-/i-verbs
		sg		pl
	1	-u		-▼-mь
	2	-▼-ši		-▼-te
	3	-▼-tь		-utь/-etь
	

5.1.3. present tense of irregular verbs

Besides these regular (or thematic) verbs, many verbs in OCS showed irregular paradigms, going back to old consonant stems. In modern varieties, these have mostly generalized one stem and adopted endings from other paradigms. Truly irregular is only sъm 'be'. In Punčo's Sbornik, irregular present-tense conjugation is also attested with dam 'give', its derivates, and similar jam 'eat'. 3pl stems seems to have been generalized here (unlike in st. BG), as Punčo prefers 1sg.prs forms like dademь (e.g. PPS 022) or edemь (here). Punčo and damaskini differ mostly in the 3pl: sources from the East prefer local reflexes of the nasal like sь (e.g. here) or dadatь (here).
Present stem variation existed in CS xotěti 'want' (e.g. 2sg xošteši, 3pl xotętъ); the variant with -št- was generalized, as we can see in the shortened variant used later (e.g. ne+štutь here). Thus, šta can be analyzed now as a regular e-verb. Another irregular verb věděti 'know' and its derivatives, though common in CS, do not appear in our sources representing "simple" varieties. More paradigms of these verbs are described below (§5.10).

	
	5.1.3. present tense of irregular verbs (V?ip)
	
	"simple" BG sources
	smь 'be'
		sg		pl
	1	(e)smь		sme
	2	si		ste
	3	e		su (PPS), sa/sь (dam.)
				
	damь 'give'
		sg		pl
	1	damь/dademь	dademo
	2	dadešь		dadete
	3	dade		dadutь (PPS), dadatь (dam.)
	
	
	OCS irregular verbs
	byti 'be'
		sg		pl		dl
	1	esmь		esmъ		esvě
	2	esi		este		esta
	3	estъ		sǫtь		este

	dati 'give'
		sg		pl		dl
	1	damь		damъ		davě
	2	dasi		daste		dasta
	3	dastъ		dadętъ		daste	
	

5.2. aorist & imperfect

5.2.1. "simple" BG aorist
5.2.2. "simple" BG imperfect
5.2.3. CS aorists
5.2.4. CS imperfect

Balkan Slavic varieties use multiple past tenses, among which there still are two synthetic ones - aorist and imperfect. The main difference between the two is aspectual: aorist (BG minalo svъršeno, lit. 'past perfective') denotes singular actions and events, while imperfect (minalo nesvъršeno, lit. 'past imperfective') denotes perpetual or repeating ones. When both are used in a single complex sentence, the imperfect denotes the context or circumstance, and aorist the event:

	
	i doide do epivati čto běše neino otčьstvo
	'and she came to Epibates, what was her fatherland' (Tixon.d.)
	

In such clauses, verbs in aorist commonly are lexically perfective (like doide in the example), while imperfects tend to be used on imperfective verbs (like běše). However, lexical aspect and tense are strictly kept apart - unlike in Upper Sorbian - so perfective verbs can build imperfects and imperfective ones aorists too, although the latter is rather less common (Lunt 1952:90). Resulting combinations allow fine nuances in sense:

	
	aorist
	prf.:	ot srdce moe napisaxь 'I wrote (this book) from my heart' (PPS 001)
	iprf.:	no pisa ruka grešna člvečeska 'but a sinful human hand wrote' (ibid.)
	
	imperfect
	iprf.:	kamili imaše tri xiljadi 'he had three thousand camels' (PPS 006)
	prf.:	i umaljaše se voda posle po sto i pe+desetь dni 'and the water was becoming less for 150 days' (PPS 025)
	

Imperfective verb in aorist (like pisa) denotes an activity in succession of events, like writing - reading - cursing the author (as hinted here). On the other hand, perfective verbs in imperfect tense like umaljaše denotes repeating (iterative) events: the water became a bit less in each of the 150 days.
While functionally distinct, the two tenses show considerable syncretisms. Functionally, both share the function of an "evidential" past tense, denoting witnessed events or situations (cf. below §5.3). Morphologically, 2/3sg forms are identic within tense paradigms, so they can be distinguished only by context. 1sg ending of both tenses was the same (-xъ) already in OCS. The difference was levelled in most stems ending in a vowel after contractions (§10.8). For this reason, 1sg.aor and impf forms are homonymous in a-verbs; the corpus does not give any examples of secondary distinction by stress (cf. Mirčev 1978:212). Elsewhere, the difference was restored later through generalization of -ox as 1sg.aor ending and -ěx (and its reflexes) as the imperfect form. This process shows traces in Punčo's Sbornik, although the forms are often used to translate CS imperfects too - in other words, he does not seem to distinguish them functionally:

	
	vidoxь ti slnce tьmno (PPS 013; vděxъ in Miltenova 1981:102)
	'I saw your dark sun'
	
	tebe videxь člkь pravedenь (PPS 025)
	'I saw you (to be) a righteous man'
	

The -o- was originally a hiatus vowel, added in front of aorist endings in stems ending in a consonant. This is also the only way to distinguish most pl.aor forms from imperfects - at least in standard BG/MK. The endings may also be homographic with other tenses and moods. With the emergence of short infinitive, the difference between it and the 2/3sg.aor form was levelled, both using the same stem. In i-verbs, whose aorist and present stems were identic already in OCS, neither 3sg.prs forms differ:

	
	lesno e bilo da se nameri imane (PPS 011)
	'it was easy to find the wealth'
	
	po sičky světь štešъ nameri neino+to ime (Berl.d.)
	'you shall find her name all over the world'
	
	i nameri čeloveci (NBKM 1064)
	'and he found (some) men'
	

First from the syntax we can determine, that the first sentence goes back to a present tense (because of da; tagged "Vmip3se") and the second - a shortened infinitive (because of lack of da after šta-auxiliary; "Vmn---e"). The aorist in the third sentence (tagged "Vmia3se") is apparent only from the broader context of the narrative. The same form could also represent an imperative. When the narrative for some reason switches between the tenses (like here), the choice may become troublesome - in such cases, the more plausible analysis is given as the primary tag, the other as a secondary one.

5.2.1. "simple" BG aorist

There are also some differences in aorist stems between the Punčo's Sbornik and standard BG/MK. In ě-stems like sedja 'sit', 2/3sg.aor forms likely show Eastern reflex of the jat (§10.4). In Punčo's Sbornik, the ě-stems ending is -e (e.g. tamo sede mlogo vreme 'she sat there for a long time'; PPS 008). In nǫ-stems, the former final nasal (§10.6) is reflected as -a- in the standards. It is also common in the damaskini, besides -ь (preminъ se vreme 'time passed'; Berl.d.). In the Sbornik, nǫ-stems show both -a (e.g. more+to pobegna 'the sea ran away'; PPS 013) and -u (kaiafь pobegnu 'Caiaphas fled'; PPS 014). In these stems, 2/3sg.aor forms are homographic with 1sg.prs forms in the standard and damaskini (cf. also Mirčev 1978:212) and can be resolved only by the context (or secondary tags). This may have motivated adoption of 1sg.prs ending -m with thema vowel in some varieties (like Punčo's).

	
			2/3sg.aor in OCS	2/3sg.aor in PPS	
	-ø-		reče			reče
	-ě-		sědě			sede
	-ěj-		směja			smeja (011), smja (027)
	-i-		xvali			(xrani 'fed')
	-a-		lěga			(bega 'ran')
	-nǫ-		(inf. legnǫ-ti)		legnu (but pobegna 'ran away')
	-ova-		kupova			kupuva
	-Ca-		kaza			kaza
	-j-		bi			(pi 'drink')
	irreg.		bystъ			bistь (011), bi (015)
			(2sg.prs dade-ši)	dade	
			(inf. vъzę-ti)		uze
	

The annotation of verbs in our corpus is based on the shape of ending. Generalization of plural forms of imperfect (-xme, -xte, -xa) seems to have affected most of BG/MK dialects. Old pl.aor endings (-xom, -ste, -še) are occasionally used in our sources too. Punčo (and other sources from the West) uses them somewhat more frequently, sometimes also on place of generalized forms in damaskini (i rekoše onia gospodare 'the lords said'; PPS 010; rekoxa in Tixon.d.). Punčo also uses two 3pl forms: besides the "regular" South Slavic reflex -še also the East Slavic variant -ša (e.g. otpadoša 'they fell down'; PPS 029), which is slightly preferred. Although some peripheral BG varieties still show the endings -ste and -še (Stojkov 2002:166), it is likely they were considered archaic in Punčo's dialect. The prevalence of 3pl.aor -ša also hints at influence of his sources.
The generalized forms are always annotated as imperfects in the corpus (e.g. rekoxa: "Vmii3pe"). Forms ending in -ax following the last stem consonant (e.g. ne+raždax 'I did not give births'; PPS 015) are annotated as aorists ("Vmia1s"). Only -ex or -ěx are handled as imperfects (dogde xoděxъ 'as long as I was walking'; Berl.d.). As both 3pl.aor (-šę) and impf (-xǫ) endings contained a nasal in OCS, their reflexes differ across redactions and dialects (§10.5-6); the annotation does not distinguish the variants.
For aorist forms like bix and bi, see below the section on conditionals (§5.4.1).

	
	5.2.1. "simple" BG aorist (Vmia)
	- based on aorist stems
	
	regular consonant stems
		sg		pl (old)	pl (new)
	1	-o-xь		-o-xomь		[-o-xme]
	2	-ø		-o-ste		[-o-xte]
	3	-ø		-o-še/-o-ša	[-o-xu]/[-o-xa]
	
	regular vowel stems
		sg		pl (old)	pl (new)
	1	-xь		-xomь		[-xme]
	2	-ø		-ste		[-xte]
	3	-ø		-še/-ša		[-xu]/[-xa]
	

5.2.2. "simple" BG imperfect

Imperfect stems reflect those of present tense. This can be observed especially in verbs going back to Ca-stems, showing an "-a preceded by a hard consonant other than v" (Lunt 2001:123), where the aorist stem does not show jotation (e.g. 1sg.prs kaža 'tell', 2/3sg.aor kaza). However, some of such present stems may have been taken over to aorist paradigms in some varieties (e.g. i kaže mu 'and [Philipp] told him'; Berl.d.).
Imperfects of e- and i-verbs can be analyzed as present stems with an extension -ě- replacing the thema vowel. In the damaskini, this extension is commonly reflected with the jat (e.g. xvalěše ba 'he praised God'; Tixon.d.), but also as -e- (posteše se 'he was fasting'; Tixon.d.), as in the standard, or -ja- (mljaše se bci 'she was praying to the Mother of God'; Tixon.d.). After palatal stem auslaut, the extension is reflected as -a- (kato sa grižaše 'as she was trying'; Berl.d.). In Punčo's Sbornik, the reflex is always an -e- (ležeše kato nikoi 'he was lying like nobody'; PPS 006).
Plural endings show various local developments. Punčo sometimes uses ending -xmo for 1pl.impf (e.g. kakvo se radvaxmo 'as we were happy'; PPS 006). In NBKM 728 we can see regular MK reflexes like imaa 'they had' (here) and turivte 'you put' (here; an aorist).
In a-verbs, there is no difference between imperfect, aorist and present stems at all. The ending is always preceded by an -a- (i dumaše 'and she was saying'; Berl.d.).
The forms of the verb 'be' became often homonymous with those of aorist (if the difference was actually made) - the given forms are always tagged as imperfects.

	
	5.2.2. "simple" BG imperfect (Vmii)
	- based on present stems
	
	e-/i-verbs
		sg		pl
	1	-ěxь		-ě-xme
	2	-ě-še		-ě-xte
	3	-ě-še		-ě-xu/-ě-xa	
	
	a-verbs
		sg		pl
	1	[-axь]		-a-xme/a-xmo
	2	-a-še		-a-xte
	3	-a-še		-a-xu/-a-xa
	
	irregular verb sъm 'be'
		sg		pl
	1	běx		běxme
	2	běše		běxte
	3	běše		běxu/běxa
	
	

5.2.3. CS aorists

In Church Slavonic, aorist stems are a productive material, also used in participles and imperfects. The morphology slightly differs from that of modern BG/MK varieties. In OCS, there are three types of aorist paradigms appearing, distributed lexically and already showing levellings. Two of them - root and s-aorist - seem to have been unproductive already at the time in East BG, while their use was limited to stems ending in a consonant (Lunt 2001:102). Distinctive root or s-aorist forms are scarce in our corpus (e.g. rěx ti 'I told you'; Vat.slav.2). Endings with the hiatus vowel -o- are preferred instead (rekox ti 'I told you'; Kiev.d.).
In nǫ-stems, the sequence was omitted in aorist if preceded by a consonant in the infinitive (e.g. ego za rouku vъzdviže '[a soldier] raised him by the arm'; Vuk.1536; OCS inf. vъzdvignǫti). In the damaskini, aorist forms show reflexes of the -nǫ- as today (ta go vdigna 'and he raised him'; Berl.d.), and it can be seen in some of our CS sources too (kosnu se emu sta 'the saint interrupted him'; Kiev.d.; inf. kosnǫti). It is possible, that the ending of the stems, where the sequence followed a vowel (like vъspoměnu 'remembered' here), was generalized for all stems of this type.
The development of nǫ-stems can also be interpreted as generalization of present stems (OCS 1sg.prs legnǫ). A similar process affected irregular verbs with 2/3sg.aor forms like vze or uze 'took' (e.g. here; OCS vъzętъ) and dade 'gave' (here; OCS dastъ).
The 2/3sg.aor ending -tъ was also used by other stems ending in an -ę, -ě, or -i: besides all having front long vowels, the affected stems also show mobile accents. In most sources it appears quite regularily (e.g. načet že smradь isxoditi 'a stench began to come out'; Vuk.1536), but not always (nače plakati 'he began to weep'; Kiev.d.). In the damaskini, the -tъ is regularily omitted. Stems ending in -ę show various reflexes. For OCS -čętъ, we first find forms like nače (e.g. PPS 013, Tixon.d.; the form being also preserved in Serbo-Croat), and later also načena (Berl.d.), as in today's BG standard. The latter is influenced by the 1sg.prs form, possibly being generalized (alongside forms like legna and dade) in the East early in the damaskini era. However, the development did not affect all roots with 2/3sg.aor -tъ.
Finally, later CS sources show a single 2/3dl.aor form (e.g. jako vь sьně zgovorěsta se 'as they made a plan in a dream'; NBKM 326). It is likely that 2dl form had been generalized before the distinctive dual was discarded. The 3dl.aor form is attested in Kiev.d. and Zogr.107 (e.g. here), always with an unexpected -ě.

	
	5.2.3. OCS aorists (Vmia)
	- based on aorist stems

	regular consonant stems
		sg		pl 		dl
	1	-o-xъ		-o-xomъ		-o-vě
	2	-e		-o-ste		-o-sta
	3	-e		-o-šę		-o-ste
	
	regular vowel stems
		sg		pl 		dl
	1	-xъ		-xomъ		-vě
	2	-ø/-tъ		-ste		-sta
	3	-ø/-tъ		-šę		-ste
	
	root aorist: pad- 'fall'
		sg		pl 
	1	padъ		padomъ	
	2	pade		padete
	3	pade		padǫ
	
	s-aorist: vьzę- 'take'
		sg		pl 
	1	vъzęsъ		vъzęsomъ
	2	vъzę		vъzęste
	3	vъzę		vъzęsę
	
	irregular verb sъm 'be'
		sg		pl		dl
	1	byxъ		byxomъ		byxově
	2	bystъ/bě	byste		bysta
	3	bystъ/bě	byšę		byste
	

According to Lunt (2001:103), the 2/3sg.aor ending -tъ appears in following stems: na-čę- 'begin', klę- 'swear', mrě- 'die', nrě- 'sink', pi- 'drink', pě- 'sing', ras-pę- 'crucify', strě- 'spread', vi- 'wrap', žrě- 'consume'. In st.BG, 2/3sg.aor forms show -na in two former ę-stems (na-čena, pro-kъlna). Other stems show only the stem now: ě-stems end in -ja (u-mrja, pja, pro-strja, but also raz-pja) and i- (or, rather, j-) stems in -i (pi, vi).
Root aorists appear in following stems: gręz- 'sink', id- 'go', krad- 'steal', leg- 'lie', lěz- 'crawl', mog- 'can', vъ-nьz- 'pierce', pad- 'fall', sъ-rět- 'meet', sěd- 'sit', vrьg- 'throw', o-xrьm- 'get lame', ěd- 'ride', as well as many nǫ-stems. S-aorists appear on: bljud- 'watch', bod- 'pierce', na-čę- 'begin', cvьt- 'bloom', vrьz- 'tie', nes- 'carry', ras-pę- 'crucify', ved- 'lead'. In some s-aorists, 1sg ending -xъ appears, e.g. rěxъ 'I said' (stem rek-); similarly handled are also stems lęk- 'bend', sěk- 'cut', tek- 'run', vlěk- 'drag', žeg- 'burn'. Čit- 'read', met- 'stir' and tręs- 'shake' show both root and s-aorists.

5.2.4. CS imperfect

Imperfect stems in OCS can be basically analyzed as those of aorist/infinitive with a tense-marking extension -ěa-. The extension is phonotactically truncated in ě- and ěj- stems (e.g. mьně- 'think' + -ěa- > mъněa-) and dejotated after soft consonants (mož- 'can' + -ěa- > možaa-) and a-verbs (xožd- iter. 'go' + -ěa- > xoždaa-). The endings are etymologically based on root aorist with a preceding -x- (Lunt 2001:100). Regular aorist endings seem to have been generalized for dual and plural already in OCS, while the extension -ěa- gets contracted (§10.8), as it is commonly seen in our sources too:

	
	tamo ne+veštestъvnoe i agglskoe prěbyvaaše žitie (Zogr.107)
	tamo ne+veštestv'noje i agglskoe prěbyvȁše žitie (Vuk.1536)
	prěbyvaše tamo bezmlьvnoje žitie (BAR 287)
	'there she dwellt immaterial and angellic (BAR 287: silent) life'
	

Zogr.107 here writes the sequence in OCS fashion, while Vuković still marks the double vowel by diacritics. However, it was usual to omit the second vowel altogether. The contractions like this are commonly attested from the 13th century onwards. There are also other changes affecting the building of imperfects appearing in the period (cf. Mirčev 1978:216). The mentioned generalization of 1sg.prs ending -ěx took place by that time. Also the use of dejotized stems like isxodeše 'it was going out' (CIAI 1161; isxoždaše in NBKM 326) became common.
The development of 1/2pl forms is hard to observe in our corpus due to the lack of examples. Forms like ne+rabotaxmy 'we were not subdued' (Dobr.) appear already in the 13th century (Mirčev 1978:217). In our corpus, they can be seen only in the Legend of Troy, mostly denoting aorists (e.g. obrětoxmy 'we found'; Vat.slav.2). Endings -xme and -xmo are common in the letters of Wallachian princes, e.g. molěxmo 'we were praying' (letter № 89 of Basarab IV). Both endings are used for both aorist and imperfect. A 2pl.impf ending is not found in our corpus, but elsewhere we can see aorist endings. The form drъžaste 'you were holding' (i-verb), is attested in a source from 1498 (Rila 4/5 168v). The dual form tako tvoresta 'thus they were doing' (also an i-verb; NBKM 667) hints that aorist endings were used in dual too.
In vernaculars of the 13th century, the 3pl.impf form was already adopted into the aorist paradigm (Mirčev 1978:217), and 1/2pl aorist endings were being taken over to that of imperfect. The new 1/2pl endings -xmy (later -xme) and -xte likely developed on the model of 3pl -xǫ afterwards, already for both tenses. The CS norm was reflecting these levellings in its dual and 2pl forms. The difference between 3pl.impf -xǫ (or -xu in Resava orthography) and 3pl.aor -šę (or -še) was preserved in the norm. Later CS grammars (Smotrickyi 1648:191v, Bončev 1952:46, Mironova 2010:154) show a mixed set of endings (1pl -xomъ, 2pl -ste, 3pl -xu) with contracted stem extensions (-ę- or -a-) too. New 1/2pl endings, as in §5.2.2., are common first in damaskini and later texts.

	
	5.2.4. OCS imperfect (Vmii)
	- based on aorist stems

	hard consonant/vowel stems
		sg		pl 		dl
	1	-ěa-xъ		-ěa-xomъ	-ěa-xově
	2	-ěa-še		-ěa-šete	-ěa-šeta
	3	-ěa-še		-ěa-xǫ		-ěa-šete
	
	soft consonant stems and a-verbs
		sg		pl 		dl
	1	-aa-xъ		-aa-xomъ	-aa-xově
	2	-aa-še		-aa-šete	-aa-šeta
	3	-aa-še		-aa-xǫ		-aa-šete
	
	imperfect in later sources
		sg		pl 	
	1	-ě-xь		-ě-[xomь]/-ě-xmy (?)
	2	-ě-še		-ě-[ste]
	3	-ě-še		-ě-xu

	irregular verb sъm 'be'
		sg		pl	
	1	běxъ		běxomъ
	2	běaše		běste		
	3	běaše		běaxǫ		
	

5.3. periphrastic tenses

5.3.1. perfects
5.3.2. future tenses

Besides the three tense forms marked on the main verb itself, modern BG/MK varieties show a number of constructions involving participles and conjugated auxiliary verbs. However, the exact function and distribution differs between varieties and standards. Bulgarian grammar tradition (e.g. Nicolova 2008:292-318) distinguishes six periphrastic tenses for the indicative mood:

	
	Perfect (minalo neopredeleno)					az sъm nosil 'I have carried'
	Plusquamperfect (minalo predvaritelno)				az bjax nosil 'I had carried'
	Future (bъdešte)						šte nosja 'I will carry', njama da nosja 'I will not carry'
	Future in the Past (bъdešte v minaloto)				štjax da nosja 'I was going to carry'
	Future Perfect (bъdešte predvaritelno)				az šte bъda nosil 'I will have carried'
	Future Perfect in the Past (bъdešte predvaritelno v minaloto) 	štjax da sъm nosil 'I was going to have carried'
	

As it can be seen, the tenses differ from an interplay of features of two main elements: root of the auxiliary ('be' denoting perfect, 'have'/'want' denoting future) and its tense (present or imperfect). The type of the main verb is selected basically according to the root of auxiliary: present (or optative) form with da particle follows 'have'/'want' auxiliaries, while 'be' are followed by l-particles. This is not the case in some MK dialects, where 'have' auxiliary with past passive (n/t-) participles can build perfects (e.g. ima dojdeno 's/he has come'; Lunt 1952:99, Friedman 1994:108). According to Koneski (1986:200) they are already attested in early 18th century, but not in our corpus, unless we interpret God as the subject in the following sentences:

	
	taka glagoletъ adonai gsdъ Eloi savaotъ, i sega ima dvě vrata otvoreni (Krčovski 1814)
	'thus speaks Adonai, the Lord, Elohim Sabaoth; and now there are (He has?) open two gates'
	

5.3.1. perfects

Under perfect-like constructions we understand those using a finite auxiliary 'be' and an l-participle. In modern BG/MK varieties they mark speaker's distance from the event (Lunt 1952:91). When the auxiliary is in present tense (e.g. nosil e 'he has carried'), the distance is evidential: in contrast to aorist or imperfect, it denotes reported, unwitnessed events. If it is in imperfect (e.g. beše nosil 'he had carried'), the distance is temporal: an event happening before the time referenced in the discourse (i.e. plusquamperfect tense).
The perfect had originally a resultative meaning - denoting a result of past action relevant for the time of utterance (Mirčev 2000:117). This function occurs today less frequently than in OCS. Modern varieties can also distinguish constructions with or without the auxiliary in 3sg (it is always omitted in MK), as well as the use of aorist (nosil) or imperfect (nosel) stems for the l-participle. According to BG grammar tradition, auxiliary omission (in BG) and imperfect-stem l-participles denote a mood mostly called "renarrative" (preizkazno naklonenie), sometimes functionally distinguished from dubitative, admirative and similar categories (cf. Nicolova 2008:319). In other words, perfect construction, stem of the l-participle and auxiliary omission mark different degrees of speaker's distance. This is commonly seen as a Balkan phenomenon (Friedman 1994:108), and a similar grammaticalization of speaker's distance can be seen in Turkish (Mirčev 1978:232).
When looking at sentence roots in our corpora, the number of l-participles among them seems rather restricted, with a notable exception in original Nedělnik and NBKM 728. Their preference for perfect-like constructions cannot be explained by the influence of their East Slavic source (Rostovski 1689), which prefers synthetic forms.

	
					aor		impf		prs		l-ptcp
	PPS (001-036)			2013		949		1010		175
	Zogr.107 (CS Petka)		101		28		95		3
	Tixon.d. (Petka)		99		84		46		13
	Berl.d. (Petka)			134		116		119		6
	Rostovski 1689 (CS Petka)	48		19		8		2
	NBKM 1423 (various)		126		57		47		6
	Nedělnik 1806 (Petka)		12		22		47		100
	NBKM 728 (Petka)		-		4		5		76
	NBKM 1064 (Petka)		101		103		61		35
	

Generally speaking, similar frequencies of perfects as in Punčo's Sbornik (4.22%) and the damaskini (5.37% in Tixon.d.) can be seen in OCS and modern prosaic texts too (Bunina 1970:56). The given statistics include plusquamperfects ('be'.impf + l-ptcp) and conditionals (using the auxiliaries like bi), the use of which does not differ much in function between OCS and modern varieties.
The situation in historical texts shows various handling of perfects. The distribution of forms (presence and root of auxiliary, stem type of the participle) and functions (resultative, distance marking) may also differ. In Punčo's Sbornik, constructions without the auxiliary appear freely with the resultative meaning too, as in MK varieties. Examples can be seen in sentences, which are used in conjunction with ones in present tense:

	
	oči+te mi ot temnicu potьmneli i ne vidimь cru tvoe crsko lice (PPS 022)
	'my eyes have become dark by the dungeon, and I cannot see your royal face, o king'
	

Elsewhere, the context hints, that the same construction could appear simply as an alternative to other past tenses too:

	
	a iosifь ne+štel i ne slušaše noina ta duma (PPS 022)
	'but Joseph did not want and he did not listen to her words'
	

Sonnenhauser (2015:47-51) considers the presence of auxiliary to reflect the "presence" of the narrator and figures in the story. In both previous sentences, Joseph is the one, who does not want and who reports his eyes having become dark. But still, Punčo generally does not use perfect-like constructions very frequently.
The other typical feature of speaker's distance marking, l-participles with imperfective stems, are hard to distinguish due to consequent reflex of jat with e: forms like znael 'knew' (e.g. here) can be traced to both stem types. Other participles formally seem to show imperfect stems, but appear in plusquamperfects, like beše oslepel 'he had become blind' (PPS 026). It is likely the distanced form became widespread first in 19th century. In our sources, such participles also appear late, often replacing finite verbal imperfects:

	
	Trěva tokmo jadęše i drugii plodъ, što sa naxoždaše tamo po onaę pustynę (Ned.1806)
	jadęla samo travǫ i plodove, koito sę namervali vъ onaę gorǫ (Ned.1856)
	'she ate only grass and fruit, which was to find in that forest/desert'
	

It is possible, that some varieties started to use perfect-like constructions as the general past tense form (as mostly elsewhere in Slavic), reinterpreting the stem distinctions to mark context/event difference of synthetic aorist and imperfect forms. Mirčev (1978:232) explains the rise of imperfect l-participles as Turkish influence, which did not encompass Western BG areas.
Finally, there is also an older construction using two auxiliaries with an l-participle (with aorist stem) to denote a renarrative plusquamperfect tense. The construction is older, appearing already in late CS texts (e.g. su bili kupili 'they had [supposedly] bought' in the letter № 210 by Vlad III) and also early damaskini and Punčo's Sbornik. The construction appears in damaskini too, but the second auxiliary seems to be optional. In other words, perfect construction with auxiliary present tense can be used with the same function:

	
	diavolь razumě čto e reklь mal'xь na ženъ+ta si (Tixon.d.)
	'The Devil understood, what Malach had said to his wife'
	
	ja rekoxь če šte da umre a onь ne+ e bilь imalь samrьtь (PPS 017)
	'I said, that he will die, but he did not have death'
	

The tense does not really differ between the examples: both relate to events preceding the reference time of the main verb - as it is in aorist, a plusquamperfect would be plausible. However, the degree of evidentiality differs: in the first case, the Devil is sure about the report, while in the second, he realized that he was deceived. It is likely the addition or omission of auxiliaries did not follow strict paradigms; rather, they were employed as particles to widen the (evidential or temporal) distance. From a diachronic view, the two types of distance were likely interchangeable.
Beside these forms, the second auxiliary is sometimes replaced by a finite, imperfect form (i.e. denoting a plusquamperfect tense, referring to the time of aorist in the main clause), which supports the interpretation of the e as an evidentiality marker:

	
	i prekri go kakvoto e i go beše prekrila maika mu (PPS 040)
	'and he covered (the child), as it had been covered by its mother'
	

The corpus is designed as agnostic towards interpretation of these construction as marking a resultative, evidential or any other (e.g. future in above-mentioned dogde ne+smь umrelь; PPS 026) tense or function. As periphrastic constructions, the individual elements are morphologically annotated as auxiliary finite verbs or participles, e.g. in nosil sъm the main verb as "Vmp--si", the auxiliary as "Vaip1si". Participles using an imperfect stem are tagged distinctively, if there is a visible difference (e.g. nosel: "Vmpi-si"). The perfect-like construction is indicated in the syntactic annotation on the auxiliary ("aux:prf"), or on the main verb (e.g. "root:prf"), if the auxiliary is omitted - in other words, if an l-participle appears as root with no dependent auxiliaries.

5.3.2. future tenses

Future tense in most modern varieties uses an auxiliary based on roots meaning 'want' or 'have', which are uninflected for person and number (BG šte, njama, MK ḱe) followed by the main verb (or perfect-marking auxiliary) in present tense. In our sources, the main verb is usually preceded by a da particle, which is used today only in negative sentences with njama. The auxiliary is also mostly marked for the person:

	
	ili šteme da turime ognь u tvoi+te dvorove (PPS 010)
	'or we will throw fire to your court'
	

Unlike in Church Slavonic, the future auxiliary is obligatory also in phrases involving perfective verbs. In OCS, present forms of these verbs denoted future tense alone. In our sources, auxiliaries also appear in sentences, which otherwise preserve the archaic morphosyntax quite exactly:

	
	azъ že, se, navedu potopъ, vodu na zemlju (Gen 6:17, My-Bible)
	azь xoštu da navedemь potopь vodu na zemlju (PPS 025)
	'I am going to bring floodwaters on the earth'
	

Also note the archaic spelling of the auxiliary - a 1sg.prs form of xotěti (OCS 1sg.prs xoštǫ). The shortened form šta is common already in the Legend of Troy (e.g. ne+štǫ tebe ostaviti 'I will not abandon you'; Vat.slav.2). In OCS, also other auxiliaries were used, especially iměti 'have' and forms of 1sg.prs bǫdǫ, perfective of 'be', and sometimes also načęti 'begin'. Among these, iměti was originally used to denote quasi-future or destiny (Lunt 2001:154), as it can be seen in our sources too:

	
	ašte li sego ne+sьtvoriši, po istině poznavai jako ognjem imate ot b[og]a pogibnouti (BAR 287)
	'if you will not do this, be sure, that you are going to die by fire from God.
	

Modern CS grammars prefer iměti as the future auxiliary without this shade of meaning (e.g. Mironova 2010:139). In our texts, both CS and "simple", a 'have'-auxiliary is preferred in positive future statements, but it appears also in some negative ones, e.g. če ne+šte iovь da poxuli na bga 'that Job would not defame God' (PPS 006). The auxiliary also appears in our CS sources with perfective verbs too, like xoštu povedati 'I will tell' (Kiev.d.). The main verb in CS periphrastic futures is always in infinitive.
Traces of old infinitive in constructions denoting future tense also appear in our sources, e.g. šteme slezna 'we will go down' (PPS 008). The status of this construction as a distinctive tense ("future indefinite", bъdešte neopredeleno) is controversial (cf. Mirčev 1978:225, Tomić 2006:457), and it seems to be abandoned in most BG/MK varieties today. These appear alongside forms with a finite 'have'-auxiliary, possibly as relics of a previous diachronic stage.
The first text showing consistently an auxiliary unmarked for person (i.e. a generalized 3sg.prs form) is that of Krčovski (kako kě da dadete 'how you will give', here). However, due to the lack of sources from MK area, it is hard to specify the date of emergence (cf. Koneski 1986:203). The da in positive statements starts to be omitted only in very late sources (šte čuete 'you will hear'; Ned.1856). In this way, following stages of future tense development can be posited:

	
	'want'.prs + full inf		xoštu povedati		CS
	'want'.prs + short inf		šteme slezna		damaskini/PPS
	'want'.prs + da + prs		šteme da turime		damaskini/PPS
	šte + da + prs			kě da dadete		Krčovski 1814
	šte + prs			šte čuete		Nedělnik 1856
	

Both auxiliary conjugation and da particle are still used in future-in-the-past constructions ('want'.impf + da + prs). Constructions with the main verb in imperfect common in Macedonia (e.g. ḱe dojdeše 's/he was going to come'; Koneski 1986:202; cf. also Mirčev 1978:226) are not attested in our corpus. In modern varieties (cf. Maslov 1981:253), the meaning is rather conditional. In Punčo's Sbornik, this function can be seen in "indefinite" examples, i.e. when the auxiliary is followed by an infinitive form of the main verb. In sentences with an inflected main verb, the construction shows more a past expectation instead:

	
	da ne beše xrstosь xaidukь ne+štexme go tebe preda (PPS 017)
	'if the Christ was not a criminal, we would not bring him to you'
	
	ta se digaxu talase kolьkoto šteše da se potopi gemija+ta (ibid.)
	'and then waves were rising as much as the ship was going to sink'
	

In the first sentence, the speaker's resolution to bring the Christ to the Pilate's court is sufficient to make the idea of not bringing him there merely hypothetic. In the second instance, the threat of sinking was real, even if the ship eventually survived. However, the construction itself is too rare to say, whether a "future-indefinite-in-the-past" tense was a productive category in Punčo's variety.
In CS, of course, future-in-the-past is expressed with the main verb in infinitive:

	
	i xotěše pogubiti Menelauša i Urekšiša ašte ne+bi ix Prějamušь zastǫpilъ (Vat.slav.2)
	'and he would have killed Menelaos and Ulysses, if Priam would have not protected them'
	

Future and perfect constructions can be combined to express what is called future perfect or "futurum exactum", denoting results of actions relevant for a future reference time. They are rare in OCS, using the auxiliary bǫdǫ and an l-participle (Mirčev 2000:119). While not attested in our CS sources, the same construction can be seen in Punčo's Sbornik, used in premises, still without the additional šte auxiliary:

	
	ako ne+bude bgь vilь izelь tova jastie (PPS 010)
	'if god Bel will not have eaten that food'
	

Future-perfect-in-the-past constructions ('want'.impf + da + 'be'.prs + l-ptcp) are not attested in our corpus.

5.4. other moods

5.4.1. conditional
5.4.2. imperative
5.4.3. infinitive

The morphological variety of verbs between CS and modern varieties is a subject to different interpretations. As we could see in the previous section, perfect-like constructions ('be' + l-ptcp) can be used to mark evidentiality. Some future tenses also denote conditional. The function of synthetic infinitives was severely limited throughout the time, in most functions replaced by the construction da + prs by now. The difference between constructions like xotěti + da in the future tense ('will') and optative ('want to') meaning is not always clear.
The design of the corpus marks the non-indicative (also not as "Vmi") only those verb forms, which can be observed on the word level, i.e. using a specific stem or ending. For this reason, only conditional, imperative, and infinitive are relevant for our study - the rest being left for the user to interpret.

5.4.1. conditional

The basic elements of conditional mood do not develop much away from the OCS state of things. Both old and modern varieties use auxiliaries based on the 'be'-stem and l-participles of main verb. Already in OCS, the auxiliary form tends to oscillate between specific conditional variants and aorists. The main difference is in the 1sg form - bimъ or byxъ (or bixь in later texts). Our sources show only the latter variant, which is also common in modern varieties:

	
	ašte ne by xotělь bъ viděti te, ne byx xodilь toliky put (Kiev.d.)
	'if God did not want (me) to see you, I would not walk such (a long) way'
	

The conditional constructions are also often used together with particles davno, dano, or simply da, denoting wishes, e.g. dano bi i ja isьcelelь 'may I be healed too' (PPS 007). The construction can be seen already in the Legend (e.g. here).
The following forms are always marked as "conditionals" on the third position of the tag (Vao). The form by (or bi) is tagged as such also if it represents an indicative aorist, e.g. on bi tolko ne+mltivь 'he was so merciless' (PPS 040). The person of these forms is - as with aorist and imperfect - inferred from the context. As plural forms would in damaskini era be homonymous with aorist forms (§10.3.), they can be distinguished according to the context, i.e. the form of the head verb; both aorist and conditional (as secondary) tag can be used. The conditional relation itself is marked in the syntactic annotation.

	
	5.4.1. "simple" BG conditional auxiliaries (Vao)
		sg		pl 
	1	bixь		[bixme]
	2	bi		[bixte]
	3	bi		[bixa/bixǫ]
	
	5.4.2. early OCS conditional auxiliaries (Vao)
		sg		pl 
	1	bimь		bimъ
	2	bi		biste
	3	bi		bǫ/bišę
	

5.4.2. imperative

As many other verbal functions, also imperative shows both synthetic and analytic means of expression. Both are actually old, going back at least to OCS. The synthetic one reflects PIE optative, which rendered a final diphthong *-ai in Proto-Slavic (Lunt 2001:246), later shortened to -i (e.g. 2sg.imp nesi 'bring!') or monophthongized to -ě (e.g. 1pl otъpaděmъ 'let us fall away'; Mirčev 1978:227). Shortenings, phonetically irregular, appear also later in most thematic verbs, which end now in /-j/, e.g. stoi 'stay!'. Athematic verbs used an ending *-ji (e.g. BG viž 'see!' here < OCS viždь < PS *weiˀd-ji), which is likely a Slavic innovation. Phonetically they developed into a zero ending, and it has been adapted by other stems with a /z/ or /ž/ in the auslaut too (e.g. izlezь nadvorь 'go out'; PPS 014; Mirčev 1978:228). The shortenings themselves are hard to follow on the texts of our corpus. Their orthography does not really make a difference between characters for /i/ and /j/, although some characters used for the former (e.g. ы and ѵ) do not appear for the latter. Distinctive forms are now used only for the second person, although Church Slavonic also had a distinctive form for 1pl:

	
	5.4.3. "simple" BG imperative (Vmm)
		sg		pl
	2	-i/-ь		-te

	5.4.4. OCS imperative (Vmm)
		sg		pl
	1			-mъ
	2	-i/-ь		-te
	

Imperative for other persons (but also for the second) can be expressed by the construction involving particle da and the main verb in present tense, e.g. da uzmešь žito lit. 'may you take grain' (PPS 006). In Church Slavonic, these phrases have more an optative meaning, often coming in prayers and similar fixed phrases (e.g. čitatelja bgь da vrazumit 'may God inspire the reader'; PPS 002). Particles dano (e.g. here) or davno can be used too. Constructions with the auxiliary in the main sequence can be searched for by UD tag ("aux:opt").
Besides these, modern varieties also show other auxiliaries. Today widely used neka is still very rare in our corpus (attested only twice in PPS 059 - neka pie 'let him drink!'). Constructions like idi reči 'go and say' (PPS 038), when the conjunction i is omitted, are handled as synthetic imperatives too: i.e. idi is tagged as an auxiliary ("Vam-2si"). Negative imperatives often appear using the auxiliary ne dei, which can be followed by the construction da + prs (ne+dei da posečešь mene skoro 'do not quickly behead me'; PPS 014) or short infinitive (ne+dei go dovrьša 'do not finish him'; PPS 013). The construction appears already in the Legend with the full infinitive (e.g. here). The 'can'-auxiliary ne moi (cf. Mirčev 1978:229) appears only once, followed by full infinitive - ne+moite mene grešnago proklinjati 'do not curse me, a sinful one' (PPS 001) - likely a deliberate archaism.
The basic endings are -i for singular and -te for plural in both OCS and modern BG/MK varieties. However, there are differences in handling by respective stem classes. In e/i-verbs, the -i is attached directly to the stem; the plural ending uses a thema vowel -e- in most cases (so the ending is actually an -ete). In a-verbs, the thema vowel is used also in singular. The a-verb plural ending is -aite: i.e. the plural marker extends the sg form as a separate suffix. As already mentioned, 2sg forms of i-verbs and 2pl forms of most e-verbs are now homonymous with indicative forms, and can be distinguished only by the context.

	
			1sg.prs		2sg.imp 	2pl.imp
	e-verbs		reka		reči		rečete
			smeja		smei		smeite
			legna		legni		legnete
			kaža		kaži		kažete
			bija		bi		biite
	i-verbs		sedja		sedi		sedete
			xvalja		xvali		xvalete
	a-verbs		ljagam		ljagai		ljagaite
			kupuvam		kupuvai		kupuvaite
			imam		imai		imaite
	irreg.		sъm		(bъdi)		(bъdete)
			dam		dai		daite
			vzema		vzemi		vzemete
	

Two loanwords show only imperative forms - xaide and ela (with pl forms xaidate and elate; e.g. here), both lit. meaning 'let us go'.

5.4.3. infinitive

Development of infinitive is one of the characteristic features of Balkan varieties within the Slavic context. The use of infinitive forms was restricted to contexts following specific auxiliaries only. Elsewhere, the "optative" construction with particle da and the main verb in present tense is used. The main verb then agrees with the subject in person and number; the tense is expressed by the auxiliary. The da-construction could be used in all contexts, where the synthetic infinitive was productive too - already in the damaskini.

	
	ne+mogy světlostь i krasotou oněxь sь drьznoveniemь zrěti (Vuk.1536)
	'being unable to dare watch their shine and beauty'
	
	oti ne+možaše da gleda texnia+t světlostъ očivěstno (Berl.d.)
	'because he could not watch their shine with his eyes'
	

Following the auxiliary moga 'can', the Life of St.Petka in Berl.d. shows in 9 instances 8 da-constructions and a single synthetic infinitives, in a phrase koi može iskaza 'who could tell?'; in another instance, a da-construction is used in the same context too (here). The process of restriction of use of synthetic infinitive progressed later onwards, as it could be seen above on the development of the future tense.
Besides the functional shift, the synthetic infinitive also developed morphologically. Old Church Slavonic actually distinguished two forms based on aorist stem: proper infinitive (with ending -ti) and supine (-tъ), used after words denoting movement (e.g. azь že vy posьlaxъ žętъ 'I sent you to reap'; Jn 4:38, Cod.Assem.). In our sources, only the proper infinitive is used (e.g. izydox poiti kъ iordanu lit. 'I departed to go to Jordan'; Kiev.d.). Sometimes, the infinitive apppears with emphatic eže (cf. §3.3.5) in such contexts (e.g. here). The ending -tъ appears in the Legend of Troy (e.g. here; total 10 times out of 169 infinitives), but in contexts usual for -ti. This seems to have been the productive infinitive ending at that time, having developed from -ti by the loss of final vowel, as the supine was likely lost already before (Mirčev 1978:233). Later, the -t was lost too, leaving the aorist stem bare, like in može iskaza. This form is used now after specific auxiliaries, besides moga 'can' also smeja 'dare', rača 'like to' (Stojkov 2002:266), as well as negative imperatives with ne+dei 'do not' and stiga 'stop' (Mirčev 1978:235), and, at least in our texts, also after future marker šte. The use of most of these forms is limited to Central Balkan and Moesian dialects, which are underlying most texts of our corpus, but it can be found in literature too, e.g. in a title of a Javorov's poem. If the main verb is an i-stem, the form becomes formally ambiguous:

	
	ne+dei ma otfъrli mene sirota (Berl.d.)
	'do not cast my poor self away'
	

Unlike the examples mentioned above in §5.2, otfъrli 'cast away' could be analyzed both as a short infinitive and as imperative, agreeing in person and number with the subject. In such instances, a secondary PoS-tag can be used.
The respective types of infinitives (old -ti, shortened CS/supine -tъ, and later short one with zero ending) are not distinguished in the annotation (all using Vmn).

5.5. participles

5.5.1. "simple" BG participles
5.5.2. CS present participles
5.5.3. participial declension

Participles represent an ambiguous category. Syntactically, they often take roles typical of verbs (roots of sentences and subordinate clauses and their auxiliaries), while their morphology is more like that of adjectives (expressing case, gender, and number). Still, at least in CS, participial morphemes also clearly mark tense (present or past - prs/aor) and even voice (act/pass), which is not indicated on other verbs at the token level. In CS grammar books, they are commonly described as a separate word category. Complexity of their use - especially of l-participles - makes other grammarians focus on periphrastic constructions involving them (like conditional or perfect tense), instead of the category of participles as such.
As mentioned above, participles are tagged both as verbs and as adjectives in the corpus. The role as a part of periphrastic construction is marked syntactically on auxiliaries, or the main verb if they are omitted:

	
	i mnogo puti ot rimljane i ot grьci danokь uzimali (PPS 003)
	'and many times they were taking taxes from the Greeks'
	

In this case, the participle is clearly used as a root of the sentence, but it is used without an auxiliary; thus the UD tag is "root:prf". The PoS-tag is "Vmp--pi", indicating the form of an l-participle, aorist stem, number and aspect. In plurals, the ending is always -i, so an adjectival tag (here would be "A-pnn" plausible) is optional.

5.5.1. "simple" BG participles

Current BG/MK standards use three main types of participles as verbal elements: l-participles used for already discussed periphrastic constructions, past passive (or n/t-) participles for passive constructions, and gerunds, etymologically uninflected forms of old present active participles. An innovation from the historical point of view is the construction of l-participles from imperfect stems, used for renarrative (Mirčev 1978:232) or distanced (Lunt 1952:91f.) constructions. These appear first in later texts, possibly already as a narrative device, as here, for describing miraculous events:

	
	i tečalъ i oustremlęl sę na staę (Ned.1806)
	'and he was running and attacking the saint'
	

At least one of these imperfect-stem participles reflects a finite imperfect verb in the CS source (cf. here). It is possible the difference was based on aspect; earlier attestation is hard to find, also because the stem is the same on a-verbs (both tagged "Vmp-" in the corpus). Both the morphology and functions of l-participles in modern varieties are wider than in CS. Besides narrative, distanced and evidential constructions mentioned above, they can also be used adnominally (e.g. jedno neino dete umrelo 'one dead child of hers'; NBKM 1423), or as heads of participial clauses similar to old ptcp.prs.act (edinь ot naši bratia znajalь nixni ezikь 'one of our brothers, [who] knew their language'; Jan.s.).
Concerning the old ptcp.prs.act itself, the form was lexicalized for some verbs (like sušti bratia 'brothers by blood', lit. 'being brothers'; PPS 007). Elsewhere, it also reflect the text tradition (onova tělo vonešto 'that stinking body'; Tixon.d.; cf. here). These are handled as adjectives in the corpus, with a secondary tag for verbal properties (e.g. vonešto: "Ansnn:Vmpp-sia").
The participle developed in some dialects into an adverbial gerund form. The most common form, standardized in both standards, is based on the MK reflex of the old f.sg/m.pl ending, as -ki or -ḱi. Among our texts, it is attested only in NBKM 728 from the area (kopaiki grobo '[while] digging the grave'; here). Mirčev (1978:242) also mentions gerunds with ending -štem productive in Southeast BG; this is closer to gerund ending -šte, which seems to be attested in OCS (Lunt 2001:159). A possibly related gerund ending -škom rarely appears in the damaskini:

	
	kato čjuxa, i razbraxa tui člci+te, sički+te tičiškom otidoxa (Berl.d.)
	'when the people heard and understood it, everyone went there running'
	

If the -šk- reflects the old ptcp.prs.act marker, it is phonetically irregular. The ending resembles sg.inst, so the form may have been influenced by adverbs built by the old case from nouns (like redom mentioned in §1.2.1). It could also reflect pl.dat (continuing dative absolute?). In CS Kiev.d., a form plačjuštumi 'crying' (here) is attested. Even if -m(i) was based on old sg.inst (or pl.dat), it was likely used as an adverbial extension regardless of gender and number. However, the marker -šk- is also once attested in NBKM 1423, where the participle is inflected for gender and number without the extension (utidi ticescka+ta 'she went away running'; here).
The old ptcp.prs.pass form, using an -m- suffix, is still used adjectivally. However, it received a new meaning, denoting a subjective quality of the modified noun, like the English suffix -able. Such forms are common in our corpus (e.g. deto e silen i ne+vidimi 'who is strong and invisible'; PPS 018), and they are attested in some CS texts (ne+sьtrьpimago smrada onogo trьpěti 'to suffer that insufferable stench'; Vuk.1536) too.
Concerning ptcp.aor.pass forms, these did not change in function much since OCS. Regional differences can be seen in lexical distribution of respective -n- and -t- markers, (e.g. izgnito 'rotten' in Berl.d. vs. izgnieno in NBKM 728). As mentioned, the annotation differs when the participle is used adnominally (doraste sьvrьšena devica 'she became a full-bloomed maiden'; PPS 020; primary tag: "Afsnn") and as a head of a clause (běše isplьnjena dxa stago 'she was filled by Holy Spirit'; Tixon.d.; primary tag: "Vmpa-se").

	
	5.5.1. "simple" BG participles
	
	Vmp-		l-ptcp (aor.stem)	umrelo (dete)		Ansnn:Vmp--se
						znajalь (nixni ezikь)	Vmp--si:Amsnn
	Vmpi		l-ptcp (impf.stem)	tečalъ			Vmpi-si:Amsnn
	
	Vmpp		gerund			kopaiki			Vmpp-pia:Ampnn
						tičiškom		Vmpp-sia:Amsin
						
			"-able" marker		(deto e ne+) vidimi	Amsny:Vmpp-si
						
	Vmpa		ptcp.aor.pass		sьvrьšena (devica)	Afsnn:Vmpa-se
						(běše) isplьnjena	Vmpa-se:Afsnn
	

Ptcp.aor.act m.sg ending -vš (e.g. postradavš 'having suffered'), described by Mirčev (1978:240) as a Russian influence in later literature, is not seen in our corpus. Some old forms corresponding to ptcp.aor.act appear due to text tradition (e.g. omočivi ruku va solilo 'the one who dipped his hand in the bowl'; PPS 017; cf. Mt 26:23), but otherwise they regularily appear in CS texts only.

5.5.2. CS participles

As hinted in the previous section, Church Slavonic has five types of participles. Morphologically, old present participles use a present stem, while past and l-participles use an aorist/infinitive stem. This is extended by a suffix denoting tense and voice, and an adjectival ending, denoting case, gender and person. In OCS, they were regularily used as heads of subordinate clauses; only the main clause tended to have a finite verb as a head. If the main and subordinate clause had the same subject, a nominative form of the participle was used; if not, the participle and its clausal subject were in dative (so-called "dative absolute", BG datelen samostojatelen). It could also appear in adnominal clauses, agreeing with the head in its case, gender and number. An example close to this ideal can be seen in Euthymius' Life of St.Petka:

	
	Ne bo vъzgnǫšaet sę prpdbnaa mti, nas vidęšti sice pririštǫštǫ (Zogr.107)
	'Because this Reverend Mother is not disgusted, when seeing us thus running to her.'
	
	vrěmeni že ne+malu mimošedšu, svoe otxoždenie eže ot+sǫdu razumě (Zogr.107)
	'And after not a small bit of time passed, she understood that she will depart from hence.'
	

The system shows some irregularities already in OCS (Mirčev 1978:286). Dative absolute appears rarely in other texts; there are no traces of it in the Legend of Troy. When it appears, it is often used for the subject of the main clause, and/or only the participle or subject are in dative case. Many clauses of complex sentences also show attempts to retain the use of participles and finite verb forms, while being separated by conjunctions:

	
	viděvъ to aleѯandrъ farižъ. i rče svoimъ otrokamъ: privezěte mi brъǫę katrъgǫ (Vat.slav.2)
	'Having seen it, ("and") Alexander Paris told his boys: bring a fast ship to me'
	
	i vь edin' dnь sědešti ei na větrьnicě i vidě muža vьz' more jazdešta (NBKM 326)
	'and one day, as she sat on the balcony, ("and") she saw a man riding along the sea'
	

It is also not uncommon to see old active participles as heads of sentences. Of course, this can be formally said about any clause separated by the i, but it may appear in both parts of such complex sentences too:

	
	i viděvъ ego prějamoušь mrъtva. i vьzdъxnǫvъ ot srdca velmi. (Vat.slav.2)
	'And having seen him dead, Priam sighed from his heart deeply'
	

Further indications of archaic status of old participles is their limited inflection, like m.sg forms appearing for female subjects (e.g. glje 'speaking', used for Magdona of the Tale here). Elsewhere, the text tradition shows, when an active participle is replacing a finite verb form:

	
	i vъ to vrěmě iměaše crca dete vъ outrobě (ODNB 1/112
	i vь to vrěme imušti crca ego vь črěvě děte (NBKM 326)
	'and in that time the queen had a child in her bosom'
	

In our corpus, the tagset does not functionally distinguish modern (e.g. gerunds, "-able" markers) and archaic functions of participles, which are left - according to general design - for interpretation by the user. Thus the annotation scheme does not differ from that of modern varieties. Old active forms (using -št-, -vš- and related markers) are marked with an additional "a" at the 8th position of the tag, so that they can be distinguished from passive forms in the Search Engine.

	
	5.5.2. CS participles
	
	Vmp-		l-ptcp			znajalь (nixni ezikь)	Vmp--si:Amsnn
	
	Vmpp		ptcp.prs.act		sědešti			Vmpp-sia:Afsnn
			ptcp.prs.pass		(deto e ne+) vidimi	Amsny:Vmpp-si
						
	Vmpa		ptcp.aor.act		viděvъ			Vmpa-sia:Amsnn
			ptcp.aor.pass		(běše) isplьnjena	Vmpa-se:Afsnn
	

5.5.3. participial declension

Individual types of participles show different declensions. The both passive participles are inflected regularily as adjectival hard stems. Both short (§2.2.1) and long (§2.2.2) forms appear. The l-participle is not used adnominally in CS, so it does not appear in non-nominative cases, the ending is short. However, a long form is possible in modern contexts (e.g. ne+izgniloe tělo 'unrotten body'; Ned.1806). In any case, an l-participle may be considered a hard stem too.
Old active participles show more variety. Basically, they show endings similar to mixed pronominal stems (of the vse-type, cf. §2.2.7) and comparatives. However, their m.sg.nom form is different from the rest due to the omission of the suffix marking tense and voice, and they also use different long forms (cf. Lunt 2001:68, Trunte 2022:90). The endings are subject to the same contractions and vowel shifts as other adjectives, so we may commonly find forms like m.sg.gen/acc xvaleštago 'boasting' (Vuk.1536), but also less regular ones, like m.sg.dat crstvoujuštomou 'ruling' (Vuk.1536). Other endings (like vonešto) show analogic levelling - the endings are then tagged according to soft adjectival paradigm, i.e. -i as pl.nom, -o as n.sg.nom, and so on.

	
	5.5.3. declension of OCS active participles
	
	ptcp.prs.act (Vmpp___a)
	- table for e-verbs
	
	m.sg
		sf		lf			corpus examples
	n	-y		-y-i
	g	-ǫšt-a		-ǫšt-aego
	d	-ǫšt-u		-ǫšt-uemu		> crstvouj-ušt-omou
	a	-ǫšt-ь		-ǫšt-ьi
	l	-ǫšt-i		[-ǫšt-iimь]
	i	-ǫšt-emъ	-ǫšt-iimь		(tič-išk-om)
	
	f.sg
		sf		lf
	n	-ǫšt-i		-ǫšt-ia			> im-ušt-i, s-ǫšt-i
	g	-ǫšt-ę		-ǫšt-ęę
	d	[-ǫšt-i]	-ǫšt-ii			> oudroučeva-jušt-ii
	a	-ǫšt-ǫ		-ǫšt-ǫjǫ		(plač-jušt-u-mi)
	l	[-ǫšt-i]	[-ǫšt-ii]
	i	-ǫšt-ǫjǫ	[-ǫšt-ǫjǫ]

	n.sg
		sf		lf
	n	-y		-ǫšt-ee			(von-ešt-o)
	g	-ǫšt-a		-ǫšt-aego
	d	-ǫšt-u		-ǫšt-uemu
	a	-ǫšt-e		-ǫšt-ee
	l	-ǫšt-i		[-ǫšt-iimь]
	i	-ǫšt-emъ	-ǫšt-iimь
	
	m.pl
		sf		lf
	n	-ǫšt-e		-ǫšt-ei			izvěstvou-jušt-e; (s-ušt-i, kopa-ik-i)
	g	-ǫšt-ь		-ǫšt-ьixъ
	d	-ǫšt-emъ	-ǫšt-iimъ
	a	-ǫšt-ę		-ǫšt-ęę
	l	-ǫšt-ixъ	[-ǫšt-iixъ]
	i	-ǫšt-i		-ǫšt-iimi	

	f.pl
		sf		lf
	n	-ǫšt-ę		-ǫšt-ęę
	g	-ǫšt-ь		-ǫšt-ьixъ
	d	-ǫšt-amъ	-ǫšt-iimъ
	a	[-ǫšt-ę]	[-ǫšt-ęę]
	l	-ǫšt-axъ	[-ǫšt-iixъ]
	i	-ǫšt-ami	-ǫšt-iimi	
	
	n.pl
		sf		lf
	n	-ǫšt-a		-ǫšt-aa
	g	-ǫšt-ь		-ǫšt-ьixъ
	d	-ǫšt-emъ	-ǫšt-iimъ
	a	[-ǫšt-a]	-ǫšt-aa
	l	-ǫšt-ixъ	[-ǫšt-iixъ]
	i	-ǫšt-i		-ǫšt-iimi	

	m.dl
		sf		lf
	n	-ǫšt-a		-ǫšt-aa
	g	-ǫšt-u		-ǫšt-uju
	d	-ǫšt-ema	-ǫšt-iima
	
	f.dl
		sf		lf
	n	-ǫšt-i		-ǫšt-ii
	g	-ǫšt-u		-ǫšt-uju
	d	-ǫšt-ama	-ǫšt-iima
	
	n.dl
		sf		lf
	n	-ǫšt-i		-ǫšt-ii
	g	-ǫšt-u		-ǫšt-uju
	d	-ǫšt-ema	-ǫšt-iima	
	
	- for i-verbs
	
		m.sg
		sf		lf
	n	-ę		-ę-i			> glj-e
	g	-ęšt-a		-ęšt-aego		> jazd-ęšt-a, xval-ešt-ago
	etc.
		f.sg
		sf		lf
	n	-ęšt-i		-ęšt-ia			> sěd-ešt-i, vid-ęšt-i; (tic-esck-a)
	etc.
	
	
	ptcp.prs.act (Vmpa___a)
	- the v is deleted in stems ending in a consonant
	
	m.sg
		sf		lf
	n	-v-ъ		-v-ъi			> vidě-v-ъ, omoči-v-i, vьzdъxnǫ-v-ъ
	g	-vš-a		-vš-aego
	d	-vš-u		-vš-uemu
	a	-vš-ь		-vš-ьi
	l	-vš-i		[-vš-iimь]
	i	-vš-emъ		-vš-iimь
	
	f.sg
		sf		lf
	n	-vš-i		-vš-ia
	g	-vš-ę		-vš-ęę
	d	-vš-i		-vš-ii
	a	-vš-ǫ		-vš-ǫjǫ			> poslědova-vš-ǫǫ
	l	-vš-i		[-vš-ii]
	i	-vš-ejǫ		[-vš-ejǫ]

	n.sg
		sf		lf
	n	-v-ъ		-vš-ee
	g	-vš-a		-vš-aego
	d	-vš-u		-vš-uemu		> mimošed-š-u
	a	-vš-e		-vš-ee
	l	-vš-i		[-vš-iimь]
	i	-vš-emъ		-vš-iimь
	
	m.pl
		sf		lf
	n	-vš-e		-vš-ęi
	g	-vš-ь		-vš-ьixъ
	d	-vš-emъ		-vš-iimъ
	a	-vš-ę		-vš-ęę
	l	-vš-ixъ		[-vš-iixъ]
	i	-vš-i		-vš-iimi	

	f.pl
		sf		lf
	n	-vš-ę		-vš-ęę
	g	-vš-ь		-vš-ьixъ
	d	-vš-amъ		-vš-iimъ
	a	[-vš-ę]		[-vš-ęę]
	l	-vš-axъ		[-vš-iixъ]
	i	-vš-ami		-vš-iimi	
	
	n.pl
		sf		lf
	n	-vš-a		-vš-aa
	g	-vš-ь		-vš-ьixъ
	d	-vš-emъ		-vš-iimъ
	a	[-vš-a]		-vš-aa
	l	-vš-ixъ		[-vš-iixъ]
	i	-vš-i		-vš-iimi
	
	m.dl
		sf		lf
	n	-vš-a		-vš-aa
	g	-vš-u		-vš-uju
	d	-vš-ema		-vš-iima
	
	f.dl
		sf		lf
	n	-vš-i		-vš-ii
	g	-vš-u		-vš-uju
	d	-vš-ama		-vš-iima
	
	n.dl
		sf		lf
	n	-vš-i		-vš-ii
	g	-vš-u		-vš-uju
	d	-vš-ema		-vš-iima
	
	

5.6. irregular verbs

In previous sections, some verbs were mentioned to show irregular conjugations. Most of them were archaic already in OCS; at later stages, analogy with other verbs levelled most differences. In modern BG/MK varieties, the most specific conjugation is that of the verb 'be'. For the other verbs we will give full paradigms only for OCS.

	
	sъm 'be'
	dam 'give'
	ida 'go'
	reka 'say'
	ęti 'take'
	
	
5.6.1. sъm 'be'
	- "simple" BG
	
	prs
		sg		pl
	1	(e)smь		sme
	2	si		ste
	3	e		su/sa/sь

	impf
		sg		pl
	1	běxь		běxme
	2	beše		běxte
	3	beše		běxa/běxu

	- OCS
	
	prs
		sg		pl		dl
	1	esmь		esmъ		esvě
	2	esi		este		esta
	3	estъ		sǫtь		este

	aor
		sg		pl		dl
	1	byxъ		byxomъ		byxově
	2	bystъ/bě	byste		bysta
	3	bystъ/bě	byšę		byste

	impf
		sg		pl		dl
	1	běxъ		běxomъ		běaxově
	2	běaše		běste		běašeta
	3	běaše		běaxǫ		běašete
	
	ptcp.prs.act		sy - sušti
	ptcp.aor.act		byvъ - byvši
	l-ptcp			bylъ
	imp			(bǫdi - bǫdete)
	inf			byti
	
	
	
5.6.2. dam 'give'
	- "simple" BG
	
	prs
		sg		pl
	1	damь/dademь	dademo
	2	dadešь		dadete
	3	dade		dadutь/dadatь
	
	- OCS
	
	prs
		sg		pl		dl
	1	damь		damъ		davě
	2	dasi		daste		dasta
	3	dastъ		dadętъ		daste	

	aor
		sg		pl		dl
	1 	daxъ		daxomъ		daxově
	2 	dastъ		daste		dasta
	3 	dastъ		dasę		daste
	
	impf
		sg		pl		dl
	1	daděaxъ		daděaxomъ	daděaxově
	2 	daděaše		daděašete	daděašeta
	3 	daděaše		daděaxǫ		daděašete

	ptcp.prs.act		dady - dadǫšti
	ptcp.prs.pass		dadimъ
	ptcp.aor.act		davъ - davši
	ptcp.aor.pass		danъ
	l-ptcp.			dalъ
	imp			daždь - dadete
	inf			dati


	
5.6.3. ida 'go'
	prs
		sg		pl		dl
	1 	idǫ		idemь		idevě
	2 	ideši		idete		ideta
	3 	idetъ		idǫtъ		idete

	aor
		sg		pl		dl
	1 	idъ		idomь		idově
	2 	ide		idete		ideta
	3 	ide		idǫ		idete

	impf
		sg		pl		dl
	1	iděaxъ		iděaxomъ	iděaxově
	2 	iděaše		iděašete	iděašeta
	3 	iděaše		iděaxǫ		iděašete

	ptcp.prs.act		idy - idǫšti
	ptcp.prs.pass		idomъ
	ptcp.aor.act		šьdъ - šьdši
	l-ptcp			šьlъ
	imp			idi - idite
	inf			iti
	
	
	
5.6.4. reka 'say'
	prs
		sg		pl		dl
	1 	rekǫ		rečemъ		rečevě
	2 	rečeši		rečete		rečeta
	3 	rečetъ		rekǫtъ		rečete

	aor
		sg		pl		dl
	1 	rěxъ		rekoxomъ	rekoxově
	2 	reče		rěste		rekosta
	3 	reče		rěšę		rekoste

	impf
		sg		pl		dl
	1 	rečaaxъ		rečaaxomъ	rečaaxově
	2 	rečaaše		rečaašete	rečaašeta
	3 	rečaaše		rečaaxǫ		rečaašete

	ptcp.prs.act		reky - rekǫšti
	ptcp.prs.pass		rekomъ
	ptcp.aor.act		rekъ - rekъši
	ptcp.aor.pass		rečenь
	l-ptcp			reklь
	imp			reči - rečete
	inf			rešti


	
5.6.5. ęti 'take'
	prs
		sg		pl		dl
	1	imǫ		imemь		imevě
	2	imeši		imete		imeta
	3	imetъ		imǫtъ		imete

	aor
		sg		pl		dl
	1	ęsъ		ęsomъ		ęsově
	2	ętъ		ęste		ęsta
	3	ętъ		ęsę		ęste

	impf
		sg		pl		dl
	1	iměaxъ		iměaxomъ	iměaxově
	2	iměaše		iměašete	iměašeta
	3	iměaše		iměaxǫ		iměašete

	ptcp.prs.act		imy - imǫšti
	ptcp.prs.pass		imomъ
	ptcp.aor.act		emъ/imъ - ęvši
	ptcp.aor.pass		ętъ
	l-ptcp			ęlъ
	imp			imi - imete
	inf			ęti
	
	

6. Adverbs

The part-of-speech category of adverbs is a traditional one, integrated to many dictionaries and grammar books, as well as MultextEast tagsets. However, it is hard to distinguish a Slavic adverb from other words on the basis of formal morphology. Many adverbs go back to old fixed phrases, showing developments specific for each lemma (e.g. pak 'again', shortened from *opaky 'from the other side'; BAN V 14), as well as morphems, unproductive in the current state of language (dnes 'today' < CSl. *dьnь sь '[on] this day'; Holzer 2015:11). An adverb in the corpus is indicated with a PoS-tag R.
Adverbs based on adjectival roots share their structure with adjectives, e.g. in ugodno upisaxь 'I described neatly' (PPS 001), the adverb is marked with an ending -o, resembling n.sg.nom of adjectives. The practice is found everywhere in Slavic; it possibly denotes an elision of the phrase 'as it is neat' (e.g. BG *upisax, kakto e ugodno). Such adverbs most commonly show the only morphological variation reflected on adverbs in the corpus - marking of the comparative degree, which is productive in Church Slavonic:

	
	positive (R)		CS comparative (Rc)	
	jasno 'clearly'		jasněiše
	mnogo 'much'		množae
	prьvo 'first'		prьvěe
	skoro 'quickly'		skorěe
	

The endings of CS comparative are formally the same as adjectival n.sg.nom forms, described in §2.2.4 above. "Simple" BG comparative marker po (as well as superlative nai) are marked as separate tokens, as with adjectives. Positive forms show also other endings: besides the prьvo, common in both CS and modern texts, we can find also forms like original sg.gen isprьva (or is+prьva; PPS 003) or prьvinь (PPS 024). Other adjectival roots form adverbs using short sg.loc suffixes, e.g. dobre 'well' (PPS 010; < CS dobrě) or peši 'on foot' (Jan.s.); the subject being "in the state" of semantic property, denoted by the root. Some adverbs with attested comparatives in CS sometimes sometimes show "locative" forms too, but always with a preposition (e.g. na mnoźě 'much', here; vъ+skorě 'quickly', here; the author uses mnogo only adjectivally, e.g. here). The sg.loc ending is sometimes used with pronominal roots, like CS nyně 'now' (PPS 005; cf. Gr. nyn, Ger. nun), as well as nominal ones like gore 'above' (PPS 013).
Some nominal and pronominal stems also use the ending -u, which seems to go back to u-stem sg.gen/loc, o-stem m/n.sg.dat or dl.gen/loc, e.g. dolu 'down, below' (Sv.d.). It is possible, that -ě (besides -de or -dě) formed a single paradigm together with -u and -mo (< PIE deverbal *-mos described here?), as mentioned above in §3.3.4; this may have distinguished two (static gorě vs. directional/lative goru) or three (static kъde vs. dir./lative kamo vs. dir./ablative kǫdu) cases. If the paradigm was ever productive, it was not stable across Slavic for long. The word kamo is attested in early damaskini in its original meaning (cf. Demina et al. 2012:392), but it is more frequently used to emphasize indeterminate objects:

	
	i ot golěma radostь ne+maše kamo što da stori (Tixon.d.)
	'and he did not know what to do because of the great joy'
	

The demonstrative form tu (< *tu-de ?) was anomalous already in OCS. Today, tu(ka) and tamo are distinguished in BG/MK, but, as elsewhere in Slavic, according to distance (Lunt 2001:79). The abolition of the static vs. directional distinction progressed further in BG/MK, as old ablative forms with -u were replaced by the forms for static location: ot gde (PPS 017), ot tuka (PPS 013), as well as ot tamo (PPS 022) are used. In other roots, one form was usually generalized too, but differently across the dialects. From the CS pair dolě 'below' vs. dolu 'downward', Punčo prefers dole (e.g. here), while BG/MK standards prefer dolu, as in the Milev's verse: dolu bog! 'down with the God!'.

	
		CS				"simple" BG
		
	n	kъde 'where'			kъde, gde, kude, kade 'where'
		tu 'there'			tuka 'here' (prox.)
		
	g	(ot+)kǫdu 'whence'		-
		(ot+)tǫdu 'from there'		-
		
	a	kamo 'from where'		kamo 'what'
		tamo 'to there'			tamo 'there' (dist.)
	

These forms are tagged as adverbs, but with a secondary pronominal tag in CS sources (cf. §3.3.4 for reference). Adverbs based on adjectival roots, as well as those based on nominals using -e or -u suffixes, are tagged as such. The choice was mostly based for each lemma, following dictionaries. Modern varieties also show some adverbs based on old instrumental forms like noštem 'in the night' (PPS 040), which are tagged as nouns in this corpus. However, the adverbial function can be well observed in syntactic annotation (using "advmod" tag).

7. Prepositions

Another traditional category of indeclinable words is actually called "adpositions" in the MultextEast terminology, as similar morphemes appear on various positions in a noun phrase. In most Slavic varieties, only prepositions are productive, placed at its beginning. Church Slavonic had also two postpositions: dělja and radi, both meaning 'for, for the sake of'. The latter appears in later texts too, mostly following the first nominal element of the noun phrase (e.g. skaži mi ljubvě radi bžie 'tell me, for the sake of love of God', Kiev.d.); sometimes it also appears as a preposition (e.g. here). All adpositions use the PoS tag S and the UD tag "case".
In Church Slavonic, the meaning of a preposition often depends on the case of the declinable elements of the noun phrase. Accusative then usually denotes movements in a direction (lative), while locative and instrumental denote static locations.

	
	obače ubo i ne+xotęšti pustynę ostavlь vъ mirъ vьzvrati se (Zogr.107)
	'but even if she did not want to leave the desert, she returned to the (inhabited) world'
	
	vьse bo iže vь mirě čьstno, ne+dostoino ei estъ (Zogr.107)
	'for everything, what is honorable in the (inhabited) world, is not worthy of her'
	

The annotation partly reflects this variation. The PoS tag of the preposition contains information about the expected case of the following noun phrase. While some prepositions always appear with the same case (e.g. k+dat: kъ čьsti 'for honor', Zogr.107; tagged "Sd"), ambiguous ones like v 'in, into' are tagged according to the root verb of the clause. The verb vьzvrati se 'she returned' denotes movement, thus the preposition is tagged "Sa"; čьstno (estъ) '(is) honorable' denotes static location, so "Sl" is used for the expected locative. The two examples are in accord with the CS norm, but many texts are not (e.g. aleѯanьdrь stojaše vь paliopolь 'Alexander stood in Paleopol', NBKM 326; prep. tagged "Sl", cf. vь paleopoli in NBKM 667). Together with UD extensions, this allows analysis of case development on a formal basis - as in modern varieties, a dependent or case-unmarked form follows any preposition.. Prepositions unattested in CS, like podir 'along, after' (e.g. here), are tagged simply with a "S".

	
	Sg		bez 'without', do 'until', iz 'from', kromě (and razvě, svěně) 'except' (all CS), naměsto (CS; or vmesto) 'instead',
			okolo 'around', ot 'from, since', posred 'through the middle', prez 'through', (za)radi 'for', u 'at'
	
	Sg or Sa	za 'for (+acc), during (+gen)'
	
	Sd		k 'to' (CS), protivъ 'against'
	
	Sd or Sl	po 'along (+dat), after (+loc)'
	
	Sa		niz 'downward' (CS), pre 'for', vъrxu 'on', vrъz 'on', vъz 'from, in exchange for' (CS)
	
	Sa or Sl	na 'on', pri 'next to', v 'into, in'
	
	Sa or Si	nad 'over', pred 'in front', pod 'under', zad 'behind'
	
	Sl		o 'about' (CS)
	
	Si		meždu 'between', s 'with'
	
	S		kъm(to) 'towards', podir 'after, along', sled 'after', sproti 'against'
	

8. Conjunctions

Slavic conjunctions are usually placed at the beginning of a sentence, a subordinate clause, or conjunct dependent elements. In Church Slavonic, there are also conjunctions like bo 'because' placed always as a clitic after the first element of the clause:

	
	azъ bo mlstiva ei xoštǫ byti (Vat.slav.2)
	'because I will be merciful to her'
	

Otherwise, the developments were mostly lexical, e.g. CS ašte 'if' was replaced with ako, jako 'as, because' with kato, and so on. As conjunctions show no morphological variety per lemma, they use the PoS tag C with no additional positions. Unlike in the other MultextEast tagsets, the syntactic role (sub- or coordinating) is indicated in the UD tag only.
As many adverbs and prepositions, also many conjunctions are compounds based on fixed phrases. Church Slavonic shows such words especially among those denoting question 'why', like vъskǫjǫ (< vъzъ kǫjǫ lit. 'in exchange for what', cf. Lunt 2001:151; e.g. here) and počьto (lit. 'following what'; here), reflected in modern BG/MK as zašto (lit. 'for what'; here). As long as they are lemmatized, they are handled as single tokens in the corpus, unless the phrase is clearly denoting an oblique argument, e.g. če ne+ma i togiva za što da si kupimo 'because then there will be no (money) to buy (grain) for, too' (PPS 006). The difference between conjunctions and adverbs of manner is not always clear. When they are followed by other adpositions, like in počto radi běgaeši 'why do you run?' (Kiev.d.), they are syntactically tagged as adverbial modifiers ("advmod").
If both elements of such a phrase are conjunctions, like in kato če 'as if' (e.g. here) or, frequently used by Punčo, ta pa 'and then' (e.g. here), they are given as multiple tokens. The first one uses an UD tag of a conjunction ("cc" or "mark"), and the second is handled as a part of a fixed phrase ("fixed", dependency on the preceding conjunction). Complex structures like ašte...to 'if...then' or nito...nito 'neither...nor' are split, with each element analyzed as dependent on the root of the clause. In 'if/when...then'-constructions, the consequent is usually handled as the main clause. Punčo often uses conjunctions a 'and, but' or i 'and' in front of consequent clauses:

	
	i kogi videše nekogo siromaxa a onъ si zatiskaše uši (PPS 036)
	'and when he saw some pauper, ("and") he blocked his ears'
	

Similar constructions can be seen in later CS sources (e.g. here) too, but is it is not usual in OCS, nor in modern standards.
The syntactic role of a conjunction can be fulfilled by words of other PoS categories too. Subordinating clauses are commonly introduced (UD tag "mark") by interrogative or relative pronouns (cf. above §3.3). Rarely, some adpositions like radi appear in the role, e.g. radi da nas spdobitь xs zde prěiti 'so that the Christ makes us pass here' (Kiev.d.). Although CS že and to appear in the role of a conjunction, they are annotated as demonstrative particles ("Qd"). The following lemmas are tagged as conjunctions:

	
	C	a 'and, but', ako 'if', ala (or ali, ama, ami) 'but', ašte 'if' (CS)
		bo (or ibo) 'because' (CS)
		če 'that'
		da 'to'
		i 'and', ili 'or'
		jako 'as, because' (CS), jakože 'as well as' (CS)
		kato 'as'
		nelo 'because', neželi 'than', ni(to, niže in CS) 'nor', no 'but'
		obače 'but', oti 'because'
		pa 'and then'
		počto 'why' (CS), poneže 'because'
		štom 'that'
		ta 'and, thus'
		ubo 'as, for' (CS)
		vъskǫjǫ 'why' (CS)
		zašto(to) 'because', zato(va) 'thus'
	

9. Particles

Traditionally, particles represent a category encompassing most words which do not fit well to the others. They fulfill diverse functions, which lack the concrete semantics of adverbs and prepositions, but also lack purely syntactic relations of conjunctions. Some of them, like ne 'not' or comparative marker po, can modify both verbal and nominal elements. Others, like da 'to', are used as conjunctions, but also have a diverse role in simple sentences, e.g. denoting imperative/optative mood in da otidešь na pazarь 'go to the market' (PPS 006). Church Slavonic že is also commonly used as a conjunction ('and'), but also as a determiner (e.g. toi+že i tavliju sьtvori 'that one created backgammon too'; NBKM 667). A similar situation is with the pronoun to (cf. §3.1.2; Mirčev 1978:270), which can be used to mark consequent clauses.
Particles are marked with a Q in the PoS tag. The second letter denotes the type of the particle; the particles themselves show no morphological variance at the lemma level.

	
	Qc	comparative		nai, po
	
	Qd	demonstrative		ete, eto, se (CS), to, že (CS)
	
	Qg	general			dako, dano, daže, de, dori, ei (CS), gode, makar, sireč, xič
	
	Qq	interrogative		dali, eda (CS), li, nali
	
	Qz	negative		ne
	

Comparative particles are handled as prefixes in standard BG/MK, although marked by a hyphen in the former. They are marked as separate tokens, because they can appear also in front of verbs, e.g. po se šiubelendissa i mažia+t i 'her husband became even more anxious' (NBKM 1423). Historically, marking of degree by these particles is a more recent one. While nai also appears in other Slavic varieties, the degree was marked by a suffix in OCS (cf. §2.2.4, §6). Prefixed markers appear in later CS sources like the Legend of Troy, followed by both positive and comparative forms of the adjectives or adverbs:

	
	tri vily proročicǫ, koǫ běxǫ nai+lěpěšǫę vь morskyixъ otocěxъ (Vat.slav.2)
	'three fairies-prophetesses, who were the most beautiful on the islands of the sea'
	
	koa estь ot nas nai lěpa, toi daite siǫ ablъkǫ (Vat.slav.2)
	'give the apple to the one, who is the most beautiful among us'
	

In a similar manner, the negative particle ne is handled as a separate token, when it modifies an adjective, adverb or a verb follows. One reason is to limit the number of lemmas. The other is the fact, that it appears separately too. In modern BG/MK, verbal ne appears in the beginning of clitic chains, so various elements like reflexive pronouns or mood markers may appear between the ne and the verb. First examples of this process can also be seen already in the Legend:

	
	i dašǫ emu orǫžie acileševo, da imъ ne bi prosilъ acileša (Vat.slav.2)
	'and they gave him the arms of Achilles, so that he would not ask Achilles (himself)'
	

10. Relevant Developments

Differences between Church Slavonic and later Balkan Slavic varieties reflect a thousand years of language drift. This reference grammar does not aim to explain all the changes in the detail. The following shifts and orthographic features are mentioned here for a better orientation in the corpus.

10.1. Loss of weak jers - any ь or ъ placed according to Havlík's law (i.e. odd when counted leftwards from the end of word or a syllable with non-jer vowel) are elided. Affects all Slavic varieties. However, CS and "simple" BG orthographies preserve the writing of final jers. Resava and damaskini prefer ь for these "orthographic jer", while East Slavic, Tarnovo and early BG standard prefer ъ.

10.2. Strong jers > a (in the West/MK) or /ă/ (in the East) - strong jers according to Havlík's law are preserved, but later develop into other vowels. Resava and Tarnovo redactions prefer to represent reflects with jers (usually ъ in prepositions and ь elsewhere; according to Velčeva 1966) too.

10.3. Shift y > i - spreads across South Slavic and Central Slovak in about 11th-13th century (cf. Pauliny 1963:190, Mirčev 1978:131, Holzer 2011:72). The change was not standardized in Resava/Tarnovo redactions, but oscillations between ы and и appear rather frequently.

10.4. Shift ě > e (in the West/MK) or /ja/ (in the East) - The change was not standardized in Resava/Tarnovo redactions. Degree of regularity differs between particular texts.

10.5. Denasalization ę > e (in the West/MK) or /jă/ (in the East) - Resava redaction adopted e as a regular reflex of the front nasal.

10.6. Denasalization ǫ > u (in the West/MK) or /ă/ (in the East) - Resava redaction adopted u as a regular reflex of the back nasal.

10.7. Reduction of unstressed vowels, e.g. e > /i/, o > /u/ (cf. Mirčev 1978:143). Although affecting many East BG dialects, the change was rather avoided in most damaskini and modern standards. It was adopted in non-Cyrillic sources like NBKM 1064 (e.g. here).

10.8. Contraction of groups VjV > V̄. In our corpus, the contraction is apparent on the development of adjectival long forms (dobraago m.sg.gen 'good' > dobrago) and imperfect endings (xoždaaše 2/3sg.impf 'was going' > xoždaše). Resava orthography has mostly adopted contracted forms, especially in adjectives.

10.9. Law of rising sonority. In OCS, only open syllables were allowed, i.e. with vowels or syllabic resonants in auslaut. In principle, it sums up multiple sound changes appearing in Common Slavic before OCS: loss of final plosives and -s (cf. Kortlandt 2002:4,8), monophthongization (e.g. tautosyllabic *ei > *ī > CS i; Holzer 2011:44), liquida-metatheses, and emergence of nasal vowels (e.g. tautosyllabic *em,*im,*en,*in > ę; cf. Holzer 2011:63). With the loss of weak jers (§10.1) closed syllables appear again. However, it remains relevant for CS orthography, which requires vowel markers at the end of each word.

10.10. Jer distinction - The choice of jer (ь, ъ or none) is orthographic in the Balkans, with ь preferred in Serbian (Kratovo, Resava) redactions, and also in most damaskini and by Punčo.

10.11. Writing of sequence /ja/ - Throughout the history of Balkan Slavic literature, the sentence could be written by many letters. Early Cyrillic used the digraph ıа, as well as the jat (ѣ - e.g. in Tarnovo CS) or small jus (ѧ - e.g. in damaskini, East Slavic CS).

10.12. Confusion of nasals. Following the desanalization (§10.5-6), certain dialects in Macedonia and Rhodopes likely did not distinguish between reflexes of the two nasals from OCS. Further levellings between the two were caused by the reduction of unstressed vowels (§10.7). In Tarnovo region, where the front nasal developed into sequence /ja/, the difference was likely reinterpreted as that of palatal correlation: ǫ reflected the middle vowel /ă/, and ę the sequences /ja/ or /jă/, according to stress. However, palatal correlation did not last long, at least in dialects of most scribes/editors of the era (cf. Mirčev 1978:114). This results in irregularities especially in f.sg inflection patterns:

	
	obrětše košutǫ vily proročicǫ felešǫ g[ospo]ždǫ (Vat.slav.2)
	'having found the stag of Lady Pallas the Fey'
	

While the direct object košutǫ 'stag' and the possessor vily 'of fairy' are marked regularily, the rest of elements show a back nasal instead of the expected jā-stem f.sg.gen -ę (cf. prročcę in a similar phrase here). Phonetical levelling coincided with - and likely also catalyzed - the development of analytic marking of old case relations. Legend of Troy is a major source for the studies of this phenomenon since long (e.g. Miklosich 1871).

10.13. Lack of separate graphem for /j/. A feature of both Cyrillic and Glagolitic script confusing generations of scholars in the field (e.g. Lunt 2001:65). An unambiguous graphem appears first in modern standards (BG й, MK j), earlier it can be implied from iotation of vowels.

10.14. Differential object marking. As mentioned in the section on nouns, there is an important difference between the dependent form, commonly used in the damaskini and Punčo's Sbornik, and the CS nominal cases: it is used only on nouns placed high on the animacy scale, like proper names, theonyms or humans. It also appears in singular, and mostly on nouns of masculine or feminine gender. The same could be said about forms appearing in these sources, which reflect other of the old case endings. Possessors are often expressed in the damaskini using synthetic genitive. Both masc. and fem. nouns show them, but already alongside the constructions with the preposition na. Place of the possessor on the animacy scale seems to be relevant too: proper names are more commonly marked than other animates. Possessive genitives in "simple" BG texts also appear in places, different from CS originals; thus they are not to be attributed to text influence only:

	
	nь vьsexvaluju prěbodobnyje raku 'but (he wanted) the most praiseworthy coffin of the Reverend (Petka)' (Vuk.1536)
	epivati, deto ima tamo mošti stěi petki 'Epibates, where the relics of St.Petka are placed' (Tixon.d.)
	

Possessive adjectives denoting feminine (or, rather ā- and jā-stems) possessors, preferred by Punčo (e.g. evini čeda'children of Eve'; PPS 024), are used rarely in the damaskini. It is possible that forms like petki (possibly including the adjective) were lexicalized "possessor forms" before f.sg.gen became obsolete; they may have fulfilled the same role as the poss. adjectives in Punčo's dialect, built from names and other nouns high on animacy scale by derivation. However, already in damaskini these forms compete with constructions using the preposition na, especially when the possessee denotes a kinship relation or action (e.g. na prpodobnaa petka pametь 'remembrance of Reverend Petka'; Tixon.d.). M.sg nouns - again, if high on animacy scale - show quite often old dative endings in such constructions. The dative also often replaces the (possessive) genitive in CS sources. The construction with na with a dependent form also appears:

	
	snu starago cra asěnja 'son of the old king Asen' (Vuk.1536)
	snь starimu blьgarskimu caru asěnju 'son of the old Bulgarian king Asen' (Tixon.d.)
	sinъ na starago crę asěna 'son of the old king Asen' (Berl.d.)
	

Thus, it can be said that case rows in "simple BG" marked primarily the animacy and the degree of syntactic dependency. The old nominative is unmarked in the latter aspect: denoting a subject or predicate. A general dependent form (dep), based usually on the old accusative, is used in direct objects and after most prepositions. A specific one is used for indirect objects (iobj): arguments of polyvalent verbs, which are not direct objects. Forms like petki show, that these constructions used to share the morphological expression with indirect objects earlier - most likely the dative case (and, afterwards, constructions with prep. na). Possessors could be secondarily distinguished from indirect objects: by poss.adjectives in the Sbornik, by using another preposition (like od) or by specific possessor forms (like petki, while using na+nom for f.sg indirect objects) elsewhere. In the Sbornik, the following system of animacy/dependency marking is plausible:

	
			dep	iobj
	človeku		+	+
	človeka		+	-
	človekь		-	-
	
			dep	iobj
	Evi		+	+
	Evu		+	-
	Eva		-	-
	
			dep
	ženu		+
	žena		-
	

Literature

Adamou, Evangelia (2011) Temporal uses of Definite Articles and Demonstratives in Pomak (Slavic, Greece). Lingua 121/5. 879-889. link
BAN: Georgiev, Vladimir I. (ed., 1972-2006) Bălgarski etimologičen rečnik. Tom I-V. Sofia: BAN. Todorov, T.A. (ed. 2002-2010) Bălgarski etimologičen rečnik. Tom VI-VII. Sofia: Marin Drinov.
Bončev, arximandrit Atanasii (1952) Cărkovnoslavjanska gramatika. Sofia: Sinodalno izdatelstvo.
Bratishenko, Elena (2003) Genitive-accusative and possessive adjective in Old East Slavic. Scando-Slavica 49/1. 83-103. link
Bunina, I.K. (1970) Istorija glagolьnyx vremen v bolgarskom jazyke. Moskva: Nauka.
Conev, Benjo (1923) Opis na slavjanskite răkopisi v Sofijskata narodna biblioteka. Tom II. Sofia: Narodnata Biblioteka.
Demina, Evgenija I. et al. (2012) Rečnik na knižovnija bălgarski ezik na narodna osnova ot XVII vek. Sofia: Valentin Trajanov.
Holzer, Georg (2011) Glasovni razvoj hrvatskoga jezika. Zagreb: Institut za hrvatski jezik i jezikoslovlje.
- (2015) Urslavische Wortlautungen II. Ricerche slavistiche 13/59. 5-34.
Ivanova, Ana (1967) Trojanski damaskin. Bъlgarski pametnik ot XVII vek. Sofia: BAN.
Jagić, Vatroslav (1895) Rassuždenija južnoslavjanskoj i russkoj stariny o cerkovno-slavjanskom jazyke. Issledovanija po russkom jazyku, Tom III. Sankt-Peterburg: Otdelenie russkago jazyka i slovesnosti imperatorskoj akademii nauk.
Knoll, Vladislav (2019) Církevní slovanština v pozdním středověku. Praha: Scriptorium.
Koneski, Blaže (1986) Istorija na makedonskiot jazik. Skopje: Kultura.
Krajčovič, Rudolf (1988) Vývin slovenského jazyka a dialektológia. Bratislava: SPN.
Lunt, Horace G. (1952) Grammar of the Macedonian Literary Language. Skopje 1952.
- (2001) Old Church Slavonic Grammar, 7th edition. Berlin: Mouton de Gruyter.
Maslov, Jurij (1981) Grammatika bolgarskogo jazyka. Moskva: Vysšaja škola.
Miklosich, Franz (1871) Trojanska priča - bugarski i latinski. Zagreb: D.Albrecht.
Miletič, Ljubomir (1923) Svištovski damaskin: novobălgarski pametnik ot XVIII věk (= Bălgarski starini 7). Sofia: BAN.
Miltenova, Anisava (1981) Apokrifăt za borbata na arxangel Mixail săs Satanail v dve redakcii. Starobălgarska literatura 9. 98-113.
- (2019) Sbornik na pop Punčo ot 1796 g. Online publication. link
Mirčev, Kiril (1978) Istoričeska gramatika na bălgarskija ezik. Sofia: Nauka i izkustvo.
- (2000) Starobălgarski ezik. Sofia: Faber.
Mironova, Tatjana (2010) Cerkovnoslavjanskij jazyk, izdanie tret'e. Moskva: Izdatel'stvo Moskovskoj Patriarxii.
Mladenov, Maksim Sl. (1963) Členuvani prilagatelni formi na ij v severoiztočnite bălgarski govori. Bălgarski ezik, XIII, kn. 4–5, 404–410.
Mladenova, Olga M. (2007) Definiteness in Bulgarian: Modelling the Processes of Language Change (= Trends in Linguistics. Studies and Monographs 182). Berlin and New York: Mouton de Gruyter.
Nicolova, Ruselina (2008) Bălgarska gramatika: Morfologija. Sofia: Univerzitno izdatelstvo "Sv. Kliment Oxridski".
Pauliny, Eugen (1963) Fonologický vývin slovenčiny. Bratislava: SAV.
Scholze, Leńka (2007) Das grammatische System der obersorbischen Umgangssprache unter besonderer Berücksichtigung des Sprachkontakts. Konstanz: Universität Konstanz (Dissertation).
Smotrickyi, Meletii (1648) Grammatika. Moskva: Pečatnyi dvor.
Sobolev, Andrej N. (1991) K istorii utraty balkanoslavjanskogo sklonenija. Zbornik Matice Srpske za filologiju i lingvistiku, XXXIV/2. Novi Sad: Matica Srpska, 7-41.
Sonnenhauser, Barbara (2015) Pop Punčo als Geschichte(n)erzähler. 3. Münchner Bulgaristik-Tag, 13.11.2015.
Stojkov, Stojko (2002) Bălgarska dialektologija. Sofia: Akad. izd. "Prof. Marin Drinov".
Šaur, Vladimír (1970) Pop Punčov sbornik kak istočnik istoriko-dialektologičeskix issledovanij. Praga: ČSAV.
Tomić, Olga M. (2006) Balkan Sprachbund Morpho-syntactic Features (= Studies in Natural Language and Linguistic Theory 67), Dordrecht: Springer.
Topolińska, Zuzana (2006) Trojniot člen - da ili ne. Južnoslovenski filolog 62. 7-15.
Velčeva, Borjana (1964) Pokazatelni mestoimenija i narečija v novobălgarskite pametnici ot XVII i XVIII v. Izvestija na Instituta za bălgarski ezik, kniga X. Sofia: BAN. 159-233.
- (1966) Norma i tradicija v bălgarskija knižoven ezik ot XVI-XVIII v. Bălgarski ezik, XVI, kn. 2, 111–121.