home issue tracker

This page still pertains to UD version 1.


Note: nmod, neg, and punct appear in two places.

Core dependents of clausal predicates
Nominal dep Predicate dep
nsubj csubj
nsubjpass csubjpass
obj ccomp xcomp
Non-core dependents of clausal predicates
Nominal dep Predicate dep Modifier word
obl advcl advmod
Special clausal dependents
Nominal dep Auxiliary Other
vocative aux mark
discourse auxpass punct
expl cop
Noun dependents
Nominal dep Predicate dep Modifier word
nummod acl amod
appos   det
nmod   neg
Compounding and unanalyzed
compound fixed goeswith
flat foreign
conj cc punct
Case-marking, prepositions, possessive
Loose joining relations
list parataxis orphan
dislocated reparandum
Sentence head Unspecified dependency
root dep

acl: clausal modifier of noun

Head of an adjectival or relative clause, usually marking the verb or nominal predicate in that clause, when the clause is marked by a relative converter (the Coptic language specific POS tag CREL, which is given the label mark, regardless of which form is used, such as ⲉⲧ, ⲉⲧⲉ, ⲉⲧⲉⲣⲉ or ⲉ; for circumstantial conversion see advcl). The arrow points from the modified element (usually a noun) to the predicate of the relative clause.


ⲛⲉⲧⲛ ⲏⲣⲡ ⲉⲧ ⲗⲟⲙⲥ \n your wine which is foul
acl(ⲏⲣⲡ, ⲗⲟⲙⲥ)

edit acl

advcl: adverbial clause modifier

An adverbial subordinate clause (e.g. a clause answering the question why? How? Where? When? or a conditional, etc.), usually introduced by a subordinating conjunction or auxiliary (ⲉⲡⲉⲓⲇⲏ, ⲉⲣϣⲁⲛ), or a circumstantial converter (ⲉ/ⲉⲣⲉ).


ⲉⲛⲉ/CONJ ⲟⲩ/DET ⲙⲟⲛⲁⲭⲟⲥ/NOUN ⲛⲁⲙⲉ/ADV ⲡⲉ/AUX ⲛⲉ/AUX ϥ/PRON ⲛⲁ/AUX ⲣ/VERB ϣⲟⲙⲧ/NUM ⲙ/ADP ⲫⲟⲛⲟⲥ/NOUN \n ...if he were truly a monk, would he have committed three murders?


Rarely, we may also see a subordinate clause governed by a preposition, in which case the preposition is governed by the head of the clause and labeled mark, not case, even if there is also a second conjunction with a mark label.


ⲙⲛ/VERB ϣⲏⲣⲉ/NOUN ... ⲉⲧⲃⲉ/ADP ϫⲉ/CONJ ⲁ/AUX ⲩ/PRON ⲡⲱⲧ/VERB \n There are no children ... because (lit. for that) they have fled.

advcl(ⲙⲛ, ⲡⲱⲧ)
mark(ⲡⲱⲧ, ⲉⲧⲃⲉ)
mark(ⲡⲱⲧ, ϫⲉ)

This analysis keeps a parallel structure with a similar clause without the preposition (e.g. only with ϫⲉ to mean ‘because’).

edit advcl

advmod: adverbial modifier

An adverbial modification, usually modifying a verb or a noun. This can be an adverb like ⲉⲙⲁⲧⲉ ‘very much’, ⲙⲙⲁⲩ ‘there’, a sentence particle (modifying the main predicate) like ⲅⲁⲣ ‘after all’, or a directional adverbial such as ⲉⲃⲟⲗ ‘out’, as well as Greek adverbs in -ⲱⲥ.


ⲙⲛ/VERB ⲗⲁⲁⲩ/NOUN ⲛϩⲏⲧ/ADP ⲟⲩ/PRON ϩⲟⲗⲱⲥ/ADV ⲙⲙⲁⲩ/ADV \n There isn't a thing inside them at all there.
advmod(ⲙⲛ, ϩⲟⲗⲱⲥ)
advmod(ⲙⲛ, ⲙⲙⲁⲩ)

Occasionally nouns will be used as adverbial modifiers depicting manner or time, as in the following:

ⲛ/AUX ⲧⲉⲧⲛ/PRON ⲥⲟⲣ/VERB ϥ/PRON ⲉⲃⲟⲗ/ADV ⲙⲉⲣⲟⲥ/NOUN ⲙⲉⲣⲟⲥ/NOUN \n ...as you divide it out, limb for limb...
advmod(ⲥⲟⲣ, ⲉⲃⲟⲗ)
advmod(ⲥⲟⲣ, ⲙⲉⲣⲟⲥ-6)
conj(ⲙⲉⲣⲟⲥ-6, ⲙⲉⲣⲟⲥ-7)

Inflected modifiers (Scriptorium tag IMOD, cf. Layton 2011: 118-123) are also seen as adverbial. For example, ϩⲱⲱ⸗ is used together with an object pronoun to mean ‘also X’ or ‘X for X’s part. Because of its basic modifier semantics, meaning ‘also’, the combination is seen as adverbial, so that the function of the phrase is again advmod. Note that ϩⲱⲱ is not a preposition, and the analysis treats it similarly to a possessed noun, so that the pronoun is seen as a determiner det:

ϩⲱⲱ/ADV ⲕ/PRON ⲕ/PRON ⲟⲩⲱϣⲧ/VERB \n you also worship/for your part you worship
advmod(ⲟⲩⲱϣⲧ, ϩⲱⲱ)

The same applies to other IMODs: the inflected modifier ⲧⲏⲣ⸗ ‘all of X’ is also seen as advmod, i.e. as syntactically more similar to ‘completely’ than a determiner ‘all’. Like all inflected modifiers, the pronoun is seen as a determiner in this case, similar to a possessive. In the following example, we can think of the meaning as ‘their entirety(-wise)’, or ‘by way of their entirety’.


ⲛ/DET ⲕⲟⲟⲩⲉ/NOUN ⲧⲏⲣ/ADV ⲟⲩ/PRON \n all the others (lit. the others in their entirety)
advmod(ⲕⲟⲟⲩⲉ, ⲧⲏⲣ)

edit advmod

amod: adjectival modifier

This function is reserved to the small closed class of Egyptian adjectives which may follow a noun without mediating ⲛ. The label is only given if the construction is actually noun+adjective: if the preposition ⲛ appears, then the label should be nmod.


ⲟⲩ/DET ϣⲏⲣⲉ/NOUN ϣⲏⲙ/NOUN \n a little boy
amod(ϣⲏⲣⲉ, ϣⲏⲙ)
det(ϣⲏⲣⲉ, ⲟⲩ)

Note that such adjectives are still tagged with the POS tag N, following the Scriptorium POS tagging guidelines.

edit amod

appos: appositional modifier

Marks appositions in all free apposition constructions, as well as the special construction with ⲛϭⲓ ‘namely’. Appositions are preferred to be marked from left to right, including in sentences with nominal subject after the verb (apposition from pronoun to noun). In cases where both a nominal and pronominal realization are given to the same argument with the same function, the tightly bound argument of the verb takes precedence in receiving subject or object marking, with subsequent realizations connected as appositions. Typical cases include normal apposition (two consecutive nominal expressions with same reference and grammatical function) and the ⲛϭⲓ construction.

ⲙⲛ/ADP ⲡⲉⲕ/DET ⲉⲓⲱⲧ/NOUN ⲡ/DET ⲇⲓⲁⲃⲟⲗⲟⲥ/NOUN \n with your father, the devil

appos(ⲉⲓϣⲧ, ⲇⲓⲁⲃⲟⲗⲟⲥ)
det(ⲇⲓⲁⲃⲟⲗⲟⲥ, ⲡ)
det(ⲉⲓⲱⲧ, ⲡⲉⲕ)
ⲁ ϥ ⲥⲱⲧⲙ ⲛϭⲓ ⲡ ⲇⲓⲁⲃⲟⲗⲟⲥ \n He heard, namely the devil.

appos(ϥ, ⲇⲓⲁⲃⲟⲗⲟⲥ)
det(ⲇⲓⲁⲃⲟⲗⲟⲥ, ⲡ)
case(ⲇⲓⲁⲃⲟⲗⲟⲥ, ⲛϭⲓ)

In the first example, your father and the devil, refer to the same thing, and have the same function (both relate to with: with your father, i.e. with the devil). In the second example, an apposition to the pronoun ϥ ‘he’ is mediated by ⲛϭⲓ ‘namely, that is’, which is considered to be a case marker, like a preposition (but with nominative case). As usual, the apposition goes from left to right, to the lexical head ‘devil’ (he = devil).

Unusually, if a nominal subject or object is mentioned before an auxiliary and is then referred to by a pronoun in the verbal complex (e.g. ⲡⲣⲱⲙⲉ ⲁϥⲥⲱⲧⲙ ‘the man, he heard’), the pronoun is treated as the subject or object, and the preceding noun is labeled dislocated.

edit appos

aux: auxiliary

The relation between a lexical verb and a conjugation base marks the base as aux to the verb, as shown below. This applies to all tripartite conjugation bases, but NOT to converters, which receive the mark label.

ⲁ/AUX ⲩ/PRON ⲥⲟⲧⲡ/VERB ⲥ/PRON \n They chose her (or: she was chosen)
nsubj(ⲥⲟⲧⲡ, ⲩ)
aux(ⲥⲟⲧⲡ, ⲁ) 
dobj(ⲥⲟⲧⲡ, ⲥ)

Some other elements that are marked as aux include the future auxiliary ⲛⲁ (tagged FUT) and the potential verb ϣ ‘be able to’, which is marked as an auxiliary to the lexical verb that follows it. This should not be confused with the impersonal verb ϣϣⲉ ‘it is appropriate’, which is treated as a main verb governing an infinitive.

edit aux

auxpass: passive auxiliary

This document is a placeholder for the language-specific documentation for auxpass.

edit auxpass

case: case marking

Used for all prepositions, including the marker of accusative case ⲛ and all other prepositions, which are understood as ‘oblique’ cases. Nouns govern their prepositions as in all other Universal Dependency guidelines, and not the other way around.

ⲡ/DET ⲣⲏ/NOUN ϩⲛ/ADP ⲧ/DET ⲡⲉ/NOUN \n the sun in the sky

det(ⲣⲏ, ⲡ)
nmod(ⲣⲏ, ⲡⲉ)
det(ⲡⲉ, ⲧ)
case(ⲡⲉ, ϩⲛ) 

The hermeneutic particle ⲛϭⲓ, roughly ‘namely’, is also considered a case marker, assigning nominative case (it is only compatible with subject appositions, never objects or obliques, cf. Grossman 2014); see dislocated for more guidelines on ⲛϭⲓ.

edit case

cc: coordinating conjunction

The label for coordinating conjunctions. These are usually ⲁⲩⲱ ‘and’ between clauses and ⲙⲛ ‘and, with’ between phrases, but could also be ϩⲓ in the sense ‘X upon Y’ or ⲏ ‘or’. If the sense is not coordinating (e.g. ⲙⲛ to mean ‘with’), cc should not be used, but nmod as with a regular preposition.

Coordination is marked left to right, with the first coordinate dominating all subsequent coordinations and receiving the incoming grammatical relations. The reasoning is that to retrieve the function of any coordinate, we can check its parent’s function, even without knowing how many coordinates there are (X and Y and Z and …). For example:

ϩⲛ/ADP ⲙ/DET ⲙⲛⲧⲁⲡⲓⲥⲧⲟⲥ/NOUN ,/PUNCT ⲙⲛ/ADP ⲙⲛⲧⲁⲕⲁⲑⲁⲣⲧⲟⲥ/NOUN ⲛⲓⲙ/PRON \n In faithlessness and every impurity

case(ⲙⲛⲧⲁⲡⲓⲥⲧⲟⲥ, ϩⲛ)
det(ⲙⲛⲧⲁⲡⲓⲥⲧⲟⲥ, ⲙ)
cc(ⲙⲛⲧⲁⲡⲓⲥⲧⲟⲥ, ⲙⲛ)
conj(ⲙⲛⲧⲁⲡⲓⲥⲧⲟⲥ, ⲙⲛⲧⲁⲕⲁⲑⲁⲣⲧⲟⲥ)
det(ⲙⲛⲧⲁⲕⲁⲑⲁⲣⲧⲟⲥ, ⲛⲓⲙ)

Note that the function of ⲙⲛⲧⲁⲡⲓⲥⲧⲟⲥ ‘faithlessness’ is nmod. To recover the function of ⲙⲛⲧⲁⲕⲁⲑⲁⲣⲧⲟⲥ ‘impurity’, we can look at ‘faithlessness’, its immediate parent and establish that ‘impurity’ is also nmod. Also note that the word ⲙⲛ ‘and’ is ambiguous with the meaning of comitative ‘with’ (e.g. go somewhere with someone). When used in the latter way, it is not labeled cc + conj but rather nmod + case, as with all other prepositions.

Exceptionally, clause initial ‘and’ or ‘but’ is connected to the root of the clause, pointing backwards, as in this example:

ⲏ/CONJ ϯ/PRON ⲥⲟⲟⲩⲛ/VERB ⲁⲛ/ADV \n Or don't I know? 

cc(ⲥⲟⲟⲩⲛ, ⲏ)
nsubj(ⲥⲟⲟⲩⲛ, ϯ)
neg(ⲥⲟⲟⲩⲛ, ⲁⲛ)

In this case, the word ⲏ ‘or’ cannot be attached to a preceding word, so it is pointed to from the following conjunct ‘know’ with the usual function, cc.

edit cc

ccomp: clausal complement

Marks a complement clause, e.g. an object clause to a verb of saying (said that: …., saw that: …). The dependency goes from the main clause predicate to the subordinate clause predicate. Markers like ϫⲉ are governed by the mark function (see mark).

ⲏ/CONJ ϯ/PRON ⲥⲟⲟⲩⲛ/VERB ⲁⲛ/ADV ⲙⲡⲁⲧ/AUX ⲓ/PRON ϫⲟⲟⲩ/VERB ⲥⲟⲩ/PRON ϫⲉ/CONJ ⲕ/PRON ⲛⲁ/AUX ⲡⲁϩ/VERB ⲟⲩ/PRON \n Don't I know, before I have sent them, that you will tear them up? 

cc(ⲥⲟⲟⲩⲛ, ⲏ)
nsubj(ⲥⲟⲟⲩⲛ, ϯ)
advmod(ⲥⲟⲟⲩⲛ, ⲁⲛ)
ccomp(ⲥⲟⲟⲩⲛ, ⲡⲁϩ)
mark(ⲡⲁϩ, ϫⲉ)
nsubj(ⲡⲁϩ, ⲕ)
aux(ⲡⲁϩ, ⲛⲁ)
obj(ⲡⲁϩ, ⲟⲩ)

Note how the verb “know” (ⲥⲟⲟⲩⲛ) is the source of ccomp, and the subordinate clause main verb, “tear” (ⲡⲁϩ), is the target.

edit ccomp

compound: compound

Used to connect compound noun heads to their modifier. For example, the compound ‘accomplice’ is comprised of ‘friend’ and ‘doing’ (a ‘doing-friend’), which is a type of ‘friend’ (not a type of ‘doing’). Therefore ‘friend’ is the head, and ‘doing’ is attached to it via the function compound:

ⲡⲉ/ART ϣⲃⲣ/NOUN ⲣϩⲱⲃ/NOUN \n accomplice! (lit. 'the friend-doing' or 'the doing-friend')

det(ϣⲃⲣ, ⲡⲉ)
compound(ϣⲃⲣ, ⲣϩⲱⲃ)

Note: This label is only used for cases in which tokenization has left parts of a compound as separate units. Generally speaking, Coptic Scriptorium guidelines specify that compound constituents are only annotated at the morpheme level, and do not constitute independent normalized units which are assigned a part of speech. As a result, this label should almost never be needed in corpora following Scriptorium segmentation practices; but for exceptional cases or corpora not following these practices, the compound label is the alternative.

edit compound

conj: conjunct

This function marks the coordinate (not the coordinating word ‘and’ etc.) and attaches it to the first member of the chain of coordinates. To retrieve the function of the daughter ofconj, only its parent needs to be examined, cf. the guidelines for cc. The same example applies:

ϩⲛ/ADP ⲙ/DET ⲙⲛⲧⲁⲡⲓⲥⲧⲟⲥ/NOUN ,/PUNCT ⲙⲛ/ADP ⲙⲛⲧⲁⲕⲁⲑⲁⲣⲧⲟⲥ/NOUN ⲛⲓⲙ/PRON \n In faithlessness and every impurity

case(ⲙⲛⲧⲁⲡⲓⲥⲧⲟⲥ, ϩⲛ)
det(ⲙⲛⲧⲁⲡⲓⲥⲧⲟⲥ, ⲙ)
cc(ⲙⲛⲧⲁⲡⲓⲥⲧⲟⲥ, ⲙⲛ)
conj(ⲙⲛⲧⲁⲡⲓⲥⲧⲟⲥ, ⲙⲛⲧⲁⲕⲁⲑⲁⲣⲧⲟⲥ)
det(ⲙⲛⲧⲁⲕⲁⲑⲁⲣⲧⲟⲥ, ⲛⲓⲙ)

The first coordinate points right to the second (conj goes from the first member in the coordination to all other members, left to right).

edit conj

cop: copula

This label marks the copula forms ⲡⲉ, ⲧⲉ, ⲛⲉ. Note that the copula is not the root of a copula predication, but rather a dependent of the lexical predicate, usually a noun preceding the copula, but sometimes a verb (especially with preterit conversion, followed by the ‘optional’ copula). Both cases are illustrated below.

ⲛⲉⲓ/DET ⲣⲱⲙⲉ/NOUN ϩⲉⲛ/DET ⲓⲟⲩⲇⲁⲓ/NOUN ⲛⲉ/AUX \n These men are Jews. 

cop(ⲓⲟⲩⲇⲁⲓ, ⲛⲉ)
det(ⲓⲟⲩⲇⲁⲓ, ϩⲉⲛ)
nsubj(ⲓⲟⲩⲇⲁⲓ, ⲣⲱⲙⲉ)
det(ⲣⲱⲙⲉ, ⲛⲉⲓ)
ϩⲁⲙⲟⲓ/CONJ ⲟⲛ/ADV ⲛⲉ/AUX ϣⲁ/AUX ⲧⲉⲧⲛ/PRON ϭⲱ/VERB ⲛ/ADP ⲧⲉⲓ/DET ϩⲉ/NOUN ⲡⲉ/AUX \n Oh, would that you would stop in this way! 

discourse(ϭⲱ, ϩⲁⲙⲟⲓ)
det(ϩⲉ, ⲧⲉⲓ)
case(ϩⲉ, ⲛ)

Also note that in nominal sentences, the subject is nsubj to the predicate, not to the copula. This is also true when the subject follows the copula, as shown below.

ⲛⲧⲟⲕ/PRON ⲡⲉ/AUX ⲧⲉⲩ/DET ϩⲉⲗⲡⲓⲥ/NOUN \n Their hope is you (lit. 'it's you, their hope')

cop(ⲛⲧⲟⲕ, ⲡⲉ)
nsubj(ⲛⲧⲟⲕ, ϩⲉⲗⲡⲓⲥ)
det(ϩⲉⲗⲡⲓⲥ, ⲧⲉⲩ)

edit cop

csubj: clausal subject

Used to mark the head of a clausal subject. The dependency goes from the predicate that governs the subject, to the local root (i.e. the predicate) of the subject clause. This construction is fairly rare, and the subject clause is often an infinitive clause, as shown below.

ⲛⲉ/AUX ⲡ/DET ⲉⲧ/SCONJ ⲉϣϣⲉ/VERB ⲡⲉ/AUX ⲉ/ADP ⲁⲛⲉⲭⲉ/VERB ⲙⲙⲟ/ADP ⲕ/PRON \n it would be fitting to tolerate you

csubj(ⲡ, ⲁⲛⲉⲭⲉ)
aux(ⲁⲛⲉⲭⲉ, ⲉ)
cop(ⲡ, ⲡⲉ)
acl(ⲡ, ⲉϣϣⲉ)
mark(ⲉϣϣⲉ, ⲉⲧ)
obj(ⲁⲛⲉⲭⲉ, ⲕ)
case(ⲕ, ⲙⲙⲟ)

Note that this is just like a nominal subject (nsubj): what would be fitting? To tolerate you. To tolerate you is fitting.

The small class of mostly Greek origin impersonal verbs also takes subject clauses, such as ⲉⲝⲉⲥⲧⲓ ‘it is appropriate’, which points from main verb to subordinate verb:

ⲟⲩⲕ ⲉⲝⲉⲥⲧⲓ/VERB ⲉ ⲁⲁ/VERB ϥ \n it is not appropriate to do it

csubj(ⲉⲝⲉⲥⲧⲓ, ⲁⲁ)

edit csubj

csubjpass: clausal passive subject

The relation csubjpass is not used in Coptic, refer to the documentation on nsubj for (pseudo-)passivization.

edit csubjpass

dep: unspecified dependency

For all cases not covered by these guidelines, the label dep may be used to denote the unusual dependency. Ideally, these cases should be re-examined and integrated into the guidelines at a later date.

edit dep

det: determiner

Function label for determiners, including definite and indefinite articles governed by their noun, but also the postposed ⲛⲓⲙ ‘any’ and the determiner ⲕⲉ ‘other’, which is unique in being allowed to stand for an article.

ⲣⲱⲙⲉ/NOUN ⲛⲓⲙ/PRON \n any man

det(ⲣⲱⲙⲉ, ⲛⲓⲙ)
ⲕⲉ/DET ⲣⲱⲙⲉ/NOUN \n another man

det(ⲣⲱⲙⲉ, ⲕⲉ)

Double determiners

If a noun has both a normal article and ⲕⲉ, both are marked as det:

ϩⲉⲛ/DET ⲕⲉ/DET ϩⲛⲁⲁⲩ/NOUN \n some other matters

det(ϩⲛⲁⲁⲩ, ⲕⲉ)
det(ϩⲛⲁⲁⲩ, ϩⲉⲛ)

Possessive determiners

Both pre-posed possessives and the old suffix pronouns are treated as determiners:

ⲡⲉⲥ/DET ⲏⲓ/NOUN \n her house

det(ⲏⲓ, ⲡⲉⲥ)

Note that the suffix possessors are pure pronouns, and are therefore tagged as PRON (but the relation is still det):

ⲣⲛⲧ/NOUN ⲕ/PRON \n your name

det(ⲣⲛⲧ, ⲕ)

edit det

discourse: discourse element

Function label for discourse particles, such as negative/affirmative answers (yes, …) or interjections (oh! Hah! etc.). The discourse particle is connected to the root of its clause using the discourse function.

ϩⲁⲙⲟⲓ/CONJ ⲟⲛ/ADV ⲛⲉ/AUX ϣⲁ/AUX ⲧⲉⲧⲛ/PRON ϭⲱ/VERB ⲛ/ADP ⲧⲉⲓ/DET ϩⲉ/NOUN ⲡⲉ/AUX \n Oh, would that you would stop in this way! 

discourse(ϭⲱ, ϩⲁⲙⲟⲓ)
advmod(ϭⲱ, ⲟⲛ)
mark(ϭⲱ, ⲛⲉ)
aux(ϭⲱ, ϣⲁ)
nsubj(ϭⲱ, ⲧⲉⲧⲛ)
cop(ϭⲱ, ⲡⲉ)
nmod(ϭⲱ, ϩⲉ)
det(ϩⲉ, ⲧⲉⲓ)
case(ϩⲉ, ⲛ)

edit discourse

dislocated: dislocated elements

This label is used for arguments or ‘hanging topics’ that are preposed before the verbal complex, and which are referred to again as pronouns governed by the verbal complex.

ⲡ/DET ϣⲓⲡⲉ/NOUN ⲙ/ADP ⲡⲉⲓ/DET ⲙⲁ/NOUN ,/PUNCT ⲡ/DET ⲉⲟⲟⲩ/NOUN ⲙ/ADP ⲡⲉⲓ/DET ⲙⲁ/NOUN ⲁ/AUX ⲓ/PRON ⲡⲁⲣⲁⲓⲧⲉⲓ/VERB ⲙⲙⲟ/ADP ⲩ/PRON \n The shame of this place, the glory of this place, I have forsaken them.

det(ϣⲓⲡⲉ, ⲡ)
nmod(ϣⲓⲡⲉ, ⲙⲁ-5)
det(ⲙⲁ-5, ⲡⲉⲓ)
case(ⲙⲁ-5, ⲙ)
conj(ⲙⲁ-5, ⲉⲟⲟⲩ)
det(ⲉⲟⲟⲩ, ⲡ)
nmod(ⲉⲟⲟⲩ, ⲙⲁ-11)
det(ⲙⲁ-11, ⲡⲉⲓ)
case(ⲙⲁ-11, ⲙ)
appos(ⲩ, ϣⲓⲡⲉ)
case(ⲩ, ⲙⲙⲟ)
dobj(ⲡⲁⲣⲁⲓⲧⲉⲓ, ⲩ)
aux(ⲡⲁⲣⲁⲓⲧⲉⲓ, ⲁ) 
nsubj(ⲡⲁⲣⲁⲓⲧⲉⲓ, ⲓ)

Note how the lexical arguments ‘the shame … the glory’ act as dislocated and are pre-posed, to be referred to again as pronouns by the pronoun object ‘them’, which acts as dobj (forsaken what? them. What is them? shame + glory). The dislocation is linked to the word governing the ‘duplicate pronoun’, usually the verb (here ⲡⲁⲣⲁⲓⲧⲉⲓ ‘forsake’, which also governs the object pronoun).

edit dislocated

expl: expletive

This document is a placeholder for the language-specific documentation for expl.

edit expl

fixed: multi-word expression

Multi-word expressions are sequences of tokens which form a fixed expression, for which internal grammatical relations are not represented. For Coptic, this often corresponds to multi-token complex prepositions, often with a frozen adverbial modifier such as ⲉⲃⲟⲗ ‘out’. For example in the following example, the complex sequence ⲉⲃⲟⲗ ϩⲛ ‘out of’, has individual tokens which literally mean ‘out in’.

ⲁ/AUX ⲩ/PRON ⲛⲧ/VERB ⲟⲩ/PRON ⲉⲃⲟⲗ/ADV ϩⲙ/ADP ⲡⲉⲕ/DET ⲏⲓ/NOUN \n they took them out of your house

aux(ⲛⲧ, ⲁ)
nsubj(ⲛⲧ, ⲩ)
dobj(ⲛⲧ, ⲟⲩ)
nmod(ⲛⲧ, ⲏⲓ)
case(ⲏⲓ, ⲉⲃⲟⲗ)
mwe(ⲉⲃⲟⲗ, ϩⲙ)
det(ⲏⲓ, ⲡⲉⲕ)

While an interpretation connecting ⲉⲃⲟⲗ to the verb to mean ‘take out’ is possible, that would leave the sense of ‘in’ to mean ‘of’ unexplained. Rather, the combination ⲉⲃⲟⲗ ϩⲛ ‘out of’ is lexicalized as a multiword expression, or fixed. By convention, mwe’s point in a chain from left to right, whereas the first token in the chain carries the external function of the expression – in this case a preposition (the label case).

edit fixed

flat: name

This label is used to connect parts of multi-word names, pointing from left to right in a chain. The most typical case is titles such as ⲁⲡⲁ, but the guideline applies to all complex names.

ⲁⲡⲁ/NOUN ⲡⲁⲡⲛⲟⲩⲧⲉ/PROPN  \n Apa Papnoute

name(ⲁⲡⲁ, ⲡⲁⲡⲛⲟⲩⲧⲉ)

edit flat

foreign: foreign words

This document is a placeholder for the language-specific documentation for foreign.

edit foreign

goeswith: goes with

This document is a placeholder for the language-specific documentation for goeswith.

edit goeswith

iobj: indirect object

Only used to mark the possessor in the possessive existential construction:

ⲟⲩⲛⲧⲁ/VERB ⲩ/PRON ⲓⲏⲥⲟⲩⲥ/PROPN \n They have Jesus

iobj(ⲟⲩⲛⲧⲁ, ⲩ)
nsubj(ⲟⲩⲛⲧⲁ, ⲓⲏⲥⲟⲩⲥ)

The above example can be interpreted as ‘exists to-them Jesus’, etymologically ⲟⲩⲛⲧⲁ-ⲩ < ‘exists in their hand.’

edit iobj

list: list

This document is a placeholder for the language-specific documentation for list.

edit list

mark: marker

Marker indicating clause status. For subordinate clauses, this is the subordinating conjunction or particle, such as ϫⲉ introducing the object clause of direct or indirect speech, a relative pronoun such as ⲉⲧ, or an adverbial subordination, such as the circumstantial converter ⲉ/ⲉⲣⲉ. In Coptic, some main clauses also have marker elements determining clause status, which are morphologically and paradigmatically comparable to subordinators: the focalizing and preterit converters. As a result, all converters (POS tags CCIRC, CREL, CPRET and CFOC) are treated as mark, modifying their clause’s predicate, but also non-converter conjunctions with similar functions (e.g. ⲉⲛⲉ ‘if’) as shown in the examples below.

ⲉⲛⲉ/CONJ ⲛⲧⲕ/PRON ⲟⲩ/DET ⲉⲛⲧⲏϭ/NOUN \n If you were a weed

mark(ⲉⲛⲧⲏϭ, ⲉⲛⲉ)
nsubj(ⲉⲛⲧⲏϭ, ⲛⲧⲕ)
det(ⲉⲛⲧⲏϭ, ⲟⲩ)
ⲉ/SCONJ ⲕ/PRON ⲧⲣⲩⲫⲁ/VERB \n While you live luxuriously 

mark(ⲧⲣⲩⲫⲁ, ⲉ)
nsubj(ⲧⲣⲩⲫⲁ, ⲕ)
ⲉ/PART ⲓ/PPERS ϣⲁϫⲉ/VERB ⲉⲣⲟ/PRON \n It’s YOU I’m talking to! 

mark(ϣⲁϫⲉ, ⲉ)
nsubj(ϣⲁϫⲉ, ⲓ)
nmod(ϣⲁⲭⲉ, ⲉⲣⲟ)

edit mark

neg: negation modifier

The label for negations such as ⲁⲛ, ⲛ, ⲧⲙ etc. which receive the POS tag NEG. The attachment is to the negated element, often the predicate or verb. Copula sentence negation is attached to the predicate, not to the copula. In circum-negation (ⲛ…ⲁⲛ), both elements are attached to the same element with the neg label.

ⲟⲩ ϩⲱⲃ ⲉ ⲛⲁⲛⲟⲩ ϥ ⲁⲛ ⲡⲉ \n it is not a good deed

det(ϩⲱⲃ, ⲟⲩ)
acl(ϩⲱⲃ, ⲛⲁⲛⲟⲩ)
cop(ϩⲱⲃ, ⲡⲉ)
neg(ϩⲱⲃ, ⲁⲛ)
mark(ⲛⲁⲛⲟⲩ, ⲉ)
nsubj(ⲛⲁⲛⲟⲩ, ϥ)

edit neg

nmod: nominal modifier

A nominal modifier. This is the label given to prepositional objects and other types of nominal dependents which are non-core arguments and not verbal modifiers (i.e. neither subject nor object nor oblique). Note that in keeping with Universal Dependencies for other languages, the nmod noun attaches directly to the lexeme it modifies (usually a noun), while the preposition is seen as a case dependent of the modifying noun.

ⲡ/DET ⲣⲏ/NOUN ϩⲛ/ADP ⲧ/DET ⲡⲉ/NOUN  \n the sun in the sky

det(ⲣⲏ, ⲡ)
nmod(ⲣⲏ, ⲡⲉ)
case(ⲡⲉ, ϩⲛ)
det(ⲡⲉ, ⲧ)

Note that prepositional arguments of verbs are marked as obl, not nmod, including prepositional objects of verbs of perception:

ⲁⲣ/PRON ⲥⲱⲧⲙ/VERB ⲉⲣⲟ/ADP ⲓ/PRON \n You have heard me

nsubj(ⲥⲱⲧⲙ, ⲁⲣ)
obl(ⲥⲱⲧⲙ, ⲓ)
case(ⲉⲣⲟ, ⲓ)

edit nmod

nsubj: nominal subject

Designates the subject of a verb. The relation is from the lexical verb to the subject pronoun or noun, not from the auxiliary, as shown below. The annotation guidelines do not distinguish passive subjects, since the two forms of passive-like expressions in Coptic may be seen as syntactically indistinct from active:

Actional passive: formed by a third person plural with non-plural reference, but if there is no singular ‘by’ phrase, it is formally indistinguishable from the active equivalent:

Ⲁ/AUX ⲩ/PRON ⲥⲟⲧⲡ/VERB ⲥ/PRON  \n she was chosen (or: they chose her)

aux(ⲥⲟⲧⲡ, ⲁ)
nsubj(ⲥⲟⲧⲡ, ⲩ)
obj(ⲥⲟⲧⲡ, ⲥ)

Stative passive: formed by the morphological stative form (pos=VSTAT) with a transitive verb, however syntactically same as statal reading with intransitive verb:

ϥ/PRON ⲕⲏⲧ/VERB \n it is built (transitive ⲕⲱⲧ “build”)

nsubj(ⲕⲏⲧ, ϥ)
ϥ/PRON ϩⲟⲗϭ/VERB \n it is sweet (intransitive ϩⲗⲟϭ)

nsubj(ϩⲟⲗϭ, ϥ)

Note also that the existential predicates (Scriptorium pos=EXIST) take nsubj for the existing entity, even when they are used in the possessive construction:

ⲙⲛ/VERB ⲙⲛⲧⲗⲏⲥⲧⲏⲥ/NOUN ϣⲟⲟⲡ/VERB \n there is no robbery

aux(ϣⲟⲟⲡ, ⲙⲛ)
nsubj(ϣⲟⲟⲡ, ⲙⲛⲧⲗⲏⲥⲧⲏⲥ)

If the possessor is indicated in the EXIST construction, the possessed is still annotated as nsubj, and the possessor is annotated as the indirect object, see iobj (i.e. the construction is interpreted as ‘there exists X to Y’).

ⲟⲩⲛⲧⲁ/VERB ⲩ/PRON ⲓⲏⲥⲟⲩⲥ/PROPN \n they have Jesus (literally: “exists to them Jesus”)

iobj(ⲟⲩⲛⲧⲁ, ⲩ)
nsubj(ⲟⲩⲛⲧⲁ, ⲓⲏⲥⲟⲩⲥ)

In pure existence predication, the existing entity is the subject, and the EXIST predicate is the local root:

ⲙⲛ/VERB ϭⲉ/PART ⲛⲟⲩⲧⲉ/NOUN ⲛⲃⲗⲁ/ADP ⲕ/PRON \n There is no God but you (“not existing” is the main predicate, “God” is the subject)

nsubj(ⲙⲛ, ⲛⲟⲩⲧⲉ) 
det(ⲛⲟⲩⲧⲉ, ϭⲉ)
obl(ⲙⲛ, ⲕ)
case(ⲕ, ⲛⲃⲗⲁ)

The same analysis applies to the presentative particle ⲉⲓⲥ (‘behold’, ‘voila’):

ⲉⲓⲥ/PART ⲟⲩ/DET ⲥⲃⲱ/N ⲛ/ADP ⲃⲣⲣⲉ/N \n Behold, a new teaching

nsubj(ⲉⲓⲥ, ⲥⲃⲱ)
det(ⲥⲃⲱ, ⲟⲩ)
nmod(ⲃⲣⲣⲉ, ⲥⲃⲱ)
case(ⲃⲣⲣⲉ, ⲛ)

edit nsubj

nsubjpass: passive nominal subject

The relation nsubjpass is not used in Coptic. For both transitive stative subjects and third person plural pseudo-passivization, use nsubj.

edit nsubjpass

nummod: numeric modifier

This label is used for number modifiers, counting the modified nouns. Note that for numbers taking a preposition ⲛ, the preposition is attached via the case label to the counted noun, but the article (and any further prepositions preceding the article) are attached to the nominal head as well, as shown below.

ϩⲙ/ADP ⲡⲉ/DET ϩⲙⲉ/NUM ⲛ/ADP ϩⲟⲟⲩ/NOUN \n in the forty days

case(ϩⲟⲟⲩ, ϩⲙ)
det(ϩⲟⲟⲩ, ⲡⲉ)
nummod(ϩⲟⲟⲩ, ϩⲙⲉ)
case(ϩⲟⲟⲩ, ⲛ)

For the number ⲥⲛⲁⲩ ‘two’, which follows the counted noun, the noun is still the head, i.e. the dependency arrow points forward from the noun to the number.

edit nummod

obj: direct object

Designates the direct object of the verb. This can be one of the following:

  1. A nominal object standing directly after a verb, OR

  2. A pronominal object directly after a verb:

Ⲁ/AUX ⲕ/PRON ⲡⲁϩ/VERB ⲟⲩ/PRON \n You have torn them. 

aux(ⲡⲁϩ, Ⲁ)
nsubj(ⲡⲁϩ, ⲕ)
obj(ⲡⲁϩ, ⲟⲩ)
  1. A nominal or pronominal object, mediated by the accusative marker (ⲛ/ⲙ/ⲙⲙⲟ⸗), usually in the durative patterns according to Jernstedt’s Law. The marker itself is linked to the noun with the case function.
Ⲁ/AUX ⲓ/PRON ϥⲓ/VERB ⲛ/ADP ⲛⲉⲕ/DET ⲛⲟⲩⲧⲉ/NOUN \n I have carried your gods.

aux(ϥⲓ, ⲁ)
nsubj(ϥⲓ, ⲓ)
obj(ϥⲓ, ⲛⲟⲩⲧⲉ)
case(ⲛⲟⲩⲧⲉ, ⲛ)
det(ⲛⲟⲩⲛⲧⲉ, ⲛⲉⲕ)

edit obj

obl: oblique argument or adjunct

The obl relation is used for oblique nominal arguments and adjuncts of verbs.

The token marked obl is a noun (or head of a noun phrase) functioning as a non-core (oblique) argument or adjunct. The obl token can provide locational, temporal or other adverbial information, and may be an adjunct as well.


ⲡ/DET ⲇⲓⲁⲃⲟⲗⲟⲥ/NOUN ⲟⲩⲏϩ/VERB ⲛϩⲏⲧ/ADP ⲕ/PRON \n The Devil dwells inside you

ⲡⲉϫⲁ/VERB ϥ/PRON ⲛⲁ/ADP ⲥ/PRON \n He said to her


Note that obl includes prepositional objects of verbs of perception, such as the object of ‘hear’ or ‘see’, which is prepositional in Coptic:

ⲁⲣ/PRON ⲥⲱⲧⲙ/VERB ⲉⲣⲟ/ADP ⲓ/PRON \n You have heard me

nsubj(ⲥⲱⲧⲙ, ⲁⲣ)
obl(ⲥⲱⲧⲙ, ⲓ)
case(ⲓ, ⲉⲣⲟ)

Prepositional objects are always obl and never marked as obj, unless they are marked by the object marker ⲛ in one of the durative tenses, according to Jernstedt’s Law.

edit obl

orphan: remnant in ellipsis

This (relatively rarely used) function is required when ellipsis of a head word results in two words which do not refer to the same thing in the world (cf. dislocated) to be realized as double dependents of the same head. English examples for this relations are sentences such as:

Mary ate the cake, but John the cookies.

In this case, the absence of a second ‘ate’ forces us to consider two conflicting subjects and objects for the first ‘ate’. The solution is to connect the second member in each set to the first one, using the remnant relation. Coptic examples work using the same logic:

ⲛ/ADV ⲥⲉ/PRON ⲣⲭⲣⲉⲓⲁ/VERB ⲁⲛ/ADV ⲛϭⲓ/PTC ⲛ/ART ⲉⲧ/SCONJ ⲧⲏⲕ/VERB ⲙ/ADP ⲡ/DET ⲥⲁⲓⲛ/NOUN ⲁⲗⲗⲁ/CONJ ⲛ/DET ⲉⲧ/SCONJ ϣⲟⲟⲡ/VERB ⲕⲁⲕⲱⲥ/ADV \n The healthy do not need the doctor, but those who are unwell (need the doctor). 

neg(ⲣⲭⲣⲉⲓⲁ, ⲛ-1)
nsubj(ⲣⲭⲣⲉⲓⲁ, ⲥⲉ)
neg(ⲣⲭⲣⲉⲓⲁ, ⲁⲛ)
dislocated(ⲣⲭⲣⲉⲓⲁ, ⲛ-6)
case(ⲛ-6, ⲛϭⲓ)
cc(ⲣⲭⲣⲉⲓⲁ, ⲁⲗⲗⲁ)
acl(ⲛ-6, ⲧⲏⲕ)
mark(ⲧⲏⲕ, ⲉⲧ-7)
nmod(ⲧⲏⲕ, ⲥⲁⲓⲛ)
case(ⲥⲁⲓⲛ, ⲙ)
det(ⲥⲁⲓⲛ, ⲡ)
remnant(ⲥⲉ, ⲛ-13)
acl(ⲛ-13, ϣⲟⲟⲡ)
mark(ϣⲟⲟⲡ, ⲉⲧ-14)
advmod(ϣⲟⲟⲡ, ⲕⲁⲕⲱⲥ)

While both subjects are related to need, only the first instance of the subject may be realized on the overt verb need. The second subject is thought of as belonging to an omitted coordinated verb. For this reason, the conjunction but is seen as having the function cc to the first verb.

edit orphan

parataxis: parataxis

This label is used to link two main clauses that are listed together as one sentence (either by accident, or because they are not quite independent sentences). It is also used for parenthetical clauses in the middle of other clauses. The dependency goes from the root of the first clause, or in parenthetical cases, the non-parenthetical clause, to the other one:

ⲛ/ADP ⲧⲉ/DET ⲩⲛⲟⲩ/NOUN ⲇⲉ/PART ⲁ/AUX ⲩ/PRON ⲕⲁ/VERB ⲛⲉⲩ/DET ϣⲛⲏⲩ/NOUN ⲁ/AUX ⲩ/PRON ⲟⲩⲁϩ/VERB ⲟⲩ/PRON ⲛⲥⲱ/ADP ϥ/PRON \n Immediately they left their nets and followed him.

mark(ⲩⲛⲟⲩ,  ⲛ)
det(ⲩⲛⲟⲩ, ⲧⲉ)
advmod(ⲕⲁ, ⲇⲉ) 
aux(ⲕⲁ, ⲁ-5)
nsubj(ⲕⲁ, ⲩ-6)
dobj(ⲕⲁ, ϣⲛⲏⲩ)
det(ϣⲛⲏⲩ, ⲛⲉⲩ)
parataxis(ⲕⲁ, ⲟⲩⲁϩ)
aux(ⲟⲩⲁϩ, ⲁ-10)
nsubj(ⲟⲩⲁϩ, ⲩ-11)
dobj(ⲟⲩⲁϩ, ⲟⲩ)
nmod(ⲟⲩⲁϩ, ϥ)
case(ⲛⲥⲱ, ϥ)
nmod(ⲕⲁ, ⲩⲛⲟⲩ)

If both of these clauses are seen as one sentence, there is no other relation to call the connection between the first clause and the second one. Note that this is distinct from two coordinated clauses, e.g. with ⲁⲩⲱ ‘and’, for which cc and conj should be used.

edit parataxis

punct: punctuation

The function for punctuation. It is seen as a root function, i.e. punctuation does not depend on any other word in the sentence. If a single tree form is desired, it is also possible to attach punctuation to the local root of the graph.

ⲁⲛⲟⲕ/PRON ./PUNCT \n I.


edit punct

reparandum: overridden disfluency

This document is a placeholder for the language-specific documentation for reparandum.

edit reparandum

root: root

The root of the utterance, that word, which depends on no other word. Usually this is the predicate, a verb if available, otherwise the nominal predicate of a nominal sentence. In fragments, such as a plain nominal phrase or a single interjection, the local root (i.e. the noun head, or the interjection) is the root.

edit root

vocative: vocative

Used to mark direct forms of address, usually introduced by a definite article: ⲡⲉϣⲃⲣⲣϩⲱⲃ! ‘oh you accomplice!’. If the vocative forms the entire utterance, it is labeled vocative and functions as the root of the utterance. If there is a further proposition in the sentence (usually an imperative), then the vocative is attached to the root of the predication:

ⲡ/DET ⲛⲟⲩⲧⲉ/NOUN ⲁⲣⲓ/VERB ⲧ/DET ⲁⲅⲁⲡⲏ/NOUN \n Oh God, have mercy! 

det(ⲛⲟⲩⲧⲉ, ⲡ)
vocative(ⲁⲣⲓ, ⲛⲟⲩⲧⲉ)
dobj(ⲁⲣⲓ, ⲁⲅⲁⲡⲏ)
det(ⲁⲅⲁⲡⲏ, ⲧ)

edit vocative

xcomp: open clausal complement

This label is used to mark dependent clauses that do not contain their own subject, most often infinitive object clauses.

Use with infinitives

ⲧⲉⲧⲛ/PRON ⲕⲱ/VERB ⲙⲙⲟ/ADP ⲟⲩ/PRON ⲁⲛ/ADV ⲉ/ADP ⲥⲣϥⲉ/VERB ⲉ/ADP ⲁⲁ/VERB ϥ/PRON \n you do not let them engage in making it

nsubj(ⲕⲱ, ⲧⲉⲧⲛ)
obj(ⲕⲱ, ⲟⲩ)
case(ⲟⲩ, ⲙⲙⲟ)
advmod(ⲕⲱ, ⲁⲛ)
xcomp(ⲕⲱ, ⲥⲣϥⲉ)
mark(ⲥⲣϥⲉ, ⲉ-6)
xcomp(ⲥⲣϥⲉ, ⲁⲁ)
mark(ⲁⲁ, ⲉ-8)
det(ⲁⲁ, ϥ)

Note how in the example above, the subject of both infinitives marked by ⲉ is the object of the main verb ⲕⲱ, which is external to the infinitive clauses. The infinitive clauses marked by xcomp are inheriting their subject argument from the main clause argument.

Use with causative ⲧⲣⲉ

Additionally, the subordinate infinitive of the causative construction with ⲧⲣⲉ is also analyzed as xcomp, although the etymological subject of the auxiliary ⲧⲣⲉ is attached to the lexical infinitive as a subject. This facilitates syntactic recognition of the construction next to semantic argument structure extraction:

ⲙⲡⲉ/AUX ⲕ/PRON ⲟⲩⲱϣ/VERB ⲁⲛ/ADV ⲡⲉ/AUX ⲉ/ADP ⲧⲣⲉ/VERB ⲛ/DET ϣⲏⲣⲉ/NOUN ⲙ/ADP ⲡ/DET ⲛⲟⲩⲧⲉ/NOUN ϫⲓ/VERB ⲟⲩ/DET ⲡⲣⲟⲥⲫⲟⲣⲁ/NOUN \n you did not wish to make the sons of God carry an offering

aux(ⲟⲩⲱϣ, ⲙⲡⲉ)
nsubj(ⲟⲩⲱϣ, ⲕ)
advmod(ⲟⲩⲱϣ, ⲁⲛ)
cop(ⲟⲩϣⲱ, ⲡⲉ)
xcomp(ⲟⲩⲱϣ, ⲧⲣⲉ)
aux(ⲧⲣⲉ, ⲉ)
det(ϣⲏⲣⲉ, ⲛ)
xcomp(ⲧⲣⲉ, ϫⲓ)
nsubj(ϫⲓ, ϣⲏⲣⲉ)
nmod(ϣⲏⲣⲉ, ⲛⲟⲩⲧⲉ)
case(ⲛⲟⲩⲧⲉ, ⲙ)
det(ⲛⲟⲩⲧⲉ, ⲡ)
det(ⲡⲣⲟⲥⲫⲟⲣⲁ, ⲟⲩ)
obj(ϫⲓ, ⲡⲣⲟⲥⲫⲟⲣⲁ)

Note that in the example above, the first xcomp is the normal infinitive case, with no explicit subject, but the second xcomp illustrates the causative construction: ‘sons’ are both the object of ‘making’ and subject of ‘carrying’.

Predicate of ϣⲱⲡⲉ ‘become’

Predicates of ϣⲱⲡⲉ ‘become’ (and stative ϣⲟⲟⲡ) are not considered objects, but are linked as xcomp to the verb. The preposition ⲛ is still marked as case in this construction.

ⲧⲉⲧⲛ ϣⲱⲡⲉ ⲛ ϫⲣⲟⲡ \n you become an obstacle

xcomp(ϣⲱⲡⲉ, ϫⲣⲟⲡ)
case(ϫⲣⲟⲡ, ⲛ)

This analysis effectively treats predicates of verbs of ‘becoming’ as a small clause: become(x,be(y)), with x serving as the overt subject of become, but also the external subject of the ‘be’ predication, much like a governed infinitive.

edit xcomp