Tokens are delimited by a) whitespace, b) apostrophe and c) hyphen as follows:
Whitespace is the default token delimiter
Apostrophe is a delimiter for proclitic variants of prepositions/case markers b’, f’, m’, s’ and t’
Hyphens are a delimiter for the definite article il- and its assimilated forms which include prepositions/case markers with fused definite definite article; these are - for the moment - treated as single tokens.
A comma is treated as a token separator in numbers, e.g. 1,500 is three tokens: 1 - , - 500.
Words with spaces or multi-word tokens do not occur in MUDT.
Morphology
Tags
For the native tagset (XPOS), see the table below. The tagging decisions are made based on three criteria in the following hierarchy:
Semantics (e.g. words designating beings, things and concepts = NOUN; words designating properties = ADJ, words designating actions, processes and states = VERB)
Morphology (e.g. words that can take feminine suffix -a or plural suffix -i- = ADJ, VERBs that can appear in both prefixal and suffixal conjugations = VERB, VERBS that are conjugated with only one affix set = VERB_PSEU)
Syntax (e.g. GEN and LIL could very well fit under PREP, but their syntactic role is quite different from PREP; same goes for FOC which are normally classified under ADV, but their syntactic behavior is quite different; HEMM is normally classified as a VERB_PSEU, but it behaves idiosyncratically)
ID
TAG
Description
ID
TAG
Description
1
ADJ
adjective
25
PART_ACT
active participle
2
ADV
adverb
26
PART_PASS
passive participle
3
COMP
complementizer
27
PREP
preposition
4
CONJ_CORD
coordinating conjunction
28
PREP_DEF
preposition with article
5
CONJ_SUB
subordinating conjunction
29
PREP_PRON
preposition with pronoun
6
DEF
article
30
PROG
progressive particle
7
FOC
focus particle
31
PRON_DEM
demonstrative pronoun
8
FUT
future particle
32
PRON_DEM_DEF
demonstrative pronoun with article
9
GEN
genitive particle
33
PRON_INDEF
indefinite pronoun
10
GEN_DEF
genitive particle with article
34
PRON_INT
interrogative pronoun
11
GEN_PRON
genitive particle with pronoun
35
PRON_PERS
personal pronoun
12
HEMM
existential verb
36
PRON_PERS_NEG
personal pronoun with negative suffix
13
INT
interjection
37
PRON_REC
reciprocal pronoun
14
KIEN
the copular/auxiliary verb kien
38
PRON_REF
reflexive pronoun
15
LIL
oblique particle
39
QUAN
quantifier
16
LIL_DEF
oblique particle with article
40
VERB
verb
17
LIL_PRON
oblique particle with pronoun
41
VERB_PSEU
pseudoverb
18
NEG
verbal negator
42
X_ABV
abbreviation
19
NOUN
noun
43
X_BOR
bordel
20
NOUN_PROP
proper noun
44
X_DIG
digits
21
NUM_CRD
cardinal numeral
45
X_ENG
English words
22
NUM_FRC
fractions
46
X_FOR
other foreign words
23
NUM_ORD
ordinal numeral
47
X_PUN
punctuation
24
NUM_WHD
number one
Some forms can appear in two different syntactic roles - Maltin can be both and ADJ and a NOUN, qabel can function both as a PREP and an ADV. In such cases, they are tagged based on their syntactic role.
MUDT uses all 17 UPOS. The XPOS > UPOS matching is straightforward (see table below), except where UD guidelines regarding the matching of UPOS and relation mandate otherwise.
UPOS
XPOS
Comments
ADJ
ADJ
ADV
ADV, FOC
INTJ
INT
NOUN
NOUN, X_ABV
PROPN
NOUN_PROP
VERB
HEMM, PART_ACT, PART_PASS, VERB, VERB_PSEU
ADP
PREP, PREP_DEF, GEN, GEN_DEF, LIL, LIL_DEF
AUX
KIEN, FUT, PROG, PRON_PERS_NEG
see the note below the table
CCONJ
CONJ_CORD
DET
DEF, QUAN
NUM
NUM_CRD, NUM_FRC, NUM_ORD, NUM_WHD
including higher-order numbers normally declined as nouns such as miljun
with the exception of mathematical symbols and currency symbols
X
X_BOR
The following are considered UPOS AUX in MUDT:
XPOS KIEN (the verb kien) in all its forms
XPOS VERB_PSEU għad and its negative form għadx
XPOS VERB_PSEU tantx (this one only provisionally so pending further study)
XPOS PRON_PERS hux, hix etc. which function as interrogative pronouns
XPOS PRON_PERS_NEG and XPOS PRON_PERS when preceded by XPOS NEG
XPOS VERB ġa used as the passive auxiliary
XPOS FUT (se, ħa, għad) and XPOS PROG (qed and its variants)
Modal verbs are not considered AUX in MUDT
UPOS PART is only used for ma (XPOS NEG)
The distinction between PRON and DET is primarily semantic; consequently, only XPOS DEF and XPOS QUAN are considered DET in the UPOS sense.
Features
Morphological features are not employed in MUDT.
PRON types are encoded in the native tagset (see above).
Syntax
General
Five types of clauses are defined in MUDT:
Verbal clauses (those with XPOS VERB or VERB_PSEU as the root)
Copular clauses
Non-copular verbless clauses
Existential clauses (those with XPOS HEMM as the root)
Non-expletive subjectless clauses (all the rest)
There are four types of copular clauses
No copula (Type i)
Personal pronoun (XPOS PRON_PERS) as the copula (Type ii)
XPOS PART_ACT qiegħed and its forms as the copula (Type iii)
XPOS KIEN as the copula (Type iv)
The relation cop is used for XPOS PRON_PERS, XPOS PART_ACT and XPOS KIEN in copular clauses of Types ii-iv. The negated form of personal pronouns (XPOS PRON_PERS_NEG) when functioning as a negative copula is provisionally tagged aux:neg throughout.
Core and oblique arguments
Core arguments in MUDT are defined based on verbal valency, largely as conceptualized by VALLEX 3.0, especially its notion of obligatoriness
A nominal subject (nsubj) is primarily defined as the obligatory nominal dependent of verbal, copular and existential clauses that
agrees with the predicate in gender, number and person (most verbal and all type (iv) copular clauses)
agrees with the predicate in gender and number (type (ii) and type (iii) copular clauses)
is the only mandatory nominal dependent of the predicate for type (i) copular
clauses and main existential clauses
An object (obj or iobj) is primarily defined as
a verbal dependent that bears the XPOS LIL/LIL_DEF case markers
a verbal dependent not bearing the case markers cited above (or any preposition) that is replaceable or can be co-referential with the respective member of either of the verbal clitic sets
Additionally, MUDT recognizes another type of core argument, the non-canonical object, which is for the purposes of UD designated obl:arg. This relation is used for the following verbal dependents:
obligatory arguments realized by means of prepositional phrases, regardless of how many and what other arguments the verb has
obligatory arguments realized by means of prepositional phrases that alternate with noun phrases (including clitics) normally annotated as obj
obligatory arguments indicating the outcome, effect or origin, regardless of whether they are realized as noun phrases or prepositional phrases and regardless of how many and what other arguments the verb has
All non-obligatory prepositional phrases with adverbial function are attached as obl
Relations overview
The following relation subtypes are used in MUDT:
advmod:neg for verbal negators ma and lanqas
aux:neg for verbal and copular negators with XPOS PRON_PERS_NEG
aux:part for verbal particles
aux:pass for the passive auxiliary ġa
case:det for prepositions and case markers with fused definite article
cop:expl for XPOS KIEN in non-copular verbless clauses
flat:name for personal and place names
nmod:poss for possessive constructions, whether the status constructus or those mediated by XPOS GEN
nsubj:pass for the nominal subject of passive verbal clauses
