nmod:unmarked: unmarked nominal modifier
This relation is a subtype of the nmod relation that applies
when a non-possessive modifier within a nominal takes the form of a
nominal lacking a preposition (a.k.a. a noun phrase). It is
“unmarked” in that, unlike most nmods, it has no adposition or
genitive marking.
nmod:unmarked merges two older subtypes, nmod:npmod and nmod:tmod.
It has a clause-level countepart: obl:unmarked.
Examples include:
(i) temporal modifiers
For lunch , any day this week is fine !
nmod:unmarked(day, week)
(ii) rate expressions with meaning equivalent to “per”
IBM earned $ 5 a share
nmod:unmarked($, share)
(iii) elementary properties
I want that color kitten
nmod:unmarked(kitten, color)
a pizza the size of the sun
nmod:unmarked(pizza, size)
(iv) emphatic reflexive pronouns
Einstein himself was in attendance .
nmod:unmarked(Einstein, himself)
(v) category descriptor or quantifier following a personal pronoun
It 's a significant burden for us students .
nmod:unmarked(us, students)
See you all in the morning .
nmod:unmarked(you, all)
(vi) “a couple” as pre-head quantifier
You can have a couple/NOUN cookies .
det(couple, a)
nmod:unmarked(cookies, couple)
(vii) scientific scale postmodifiers
7/NUM degrees/NOUN Fahrenheit/PROPN
nummod(degrees, 7)
nmod:unmarked(degrees, Fahrenheit)
Locations
For discussion of multiword place names like Lake Mead and Route 66, see nmod:desc.
Identifiers of specific locations are often complex, incorporating multiple levels of granularity. For some such descriptions, the locational parts are juxtaposed—perhaps with punctuation like commas or dashes, but without prepositions or other clear grammatical indication of headedness. Common examples include cities, where the city name is conventionally paired with a state/province/country name (Berkeley, California, USA); street addresses, where a house number is related to a street name, and possibly to a subunit; and even textual references (Act IV, Scene 1 in a play). The order of granularity is big-endian for some of these naming patterns and little-endian for others.
In the absence of morphosyntactic cues, we rely on semantics to determine the dependency relations:
the most specific locator serves as the head of all the others in the expression, provided that the
expression is short enough that it would tend to occur as a nominal in a wider sentence.
Dependents attach as nmod:unmarked:
Berkeley/PROPN , California/PROPN , USA/PROPN
nmod:unmarked(Berkeley, California)
nmod:unmarked(Berkeley, USA)
The city name heads both the state and country names, though any pair of these elements might occur without the third.
110/NUM Sproul/PROPN Hall/PROPN
nmod:unmarked(110, Hall)
compound(Hall, Sproul)
Here, Sproul Hall names a building and 110 is the room or office number within the building. The room number is more specific, and is therefore the head. The same applies to apartment numbers, even if presented after the house number and street. Because the house number and street are clearly grouped together, the house number heads the street name:
221B/PROPN Baker/PROPN St./PROPN , Apt./PROPN E/PROPN
nmod:unmarked(221B, St.)
compound(St., Baker)
flat(Apt., E)
nmod:unmarked(Apt., 221B)
Where the name of an institutional branch or campus juxtaposes the name of the institution followed by the name of a place, the institution is the head (but prepositional phrases are structured as usual):
University/PROPN of/ADP Wisconsin/PROPN – Madison/PROPN
nmod(University, Wisconsin)
nmod:unmarked(University, Madison)
If the expression follows an extragrammatical format like a full postal address, the major parts are connected via list:
110/NUM Sproul/PROPN Hall/PROPN , Berkeley/PROPN , California/PROPN , USA/PROPN 94704/NUM
list(110, Berkeley)
list(110, 94704)
Harry/PROPN Potter/PROPN , 12/NUM Grimmauld/PROPN Place/PROPN , London/PROPN
flat(Harry, Potter)
list(Harry, 12)
list(Harry, London)
Dates
Locations in time follow the same principles as spatial locations:
8/NUM[NumType=Card] October/PROPN 1963/NUM[NumType=Card]
nmod:unmarked(8, October)
nmod:unmarked(8, 1963)
October/PROPN 8/NUM[NumType=Card] , 1963/NUM[NumType=Card]
nmod:unmarked(8, October)
nmod:unmarked(8, 1963)
October/PROPN 8th/NUM[NumType=Ord] , 1963/NUM[NumType=Card]
nmod:unmarked(8th, October)
nmod:unmarked(8th, 1963)
BC, AD, and similar eras modify the year:
October/PROPN 1963/NUM[NumType=Card] AD/NOUN
nmod:unmarked(October, 1963)
nmod:unmarked(1963, AD)
Where two labels apply to the same temporal unit, appos may be suitable:
It starts at/ADP 8:00/NUM[NumType=Card] a.m./NOUN EST/PROPN on/ADP Monday/PROPN , October/PROPN 8/NUM[NumType=Card] , 1963/NUM[NumType=Card] .
obl(starts, 8:00)
case(8:00, at)
nmod:unmarked(8:00, a.m.)
nmod:unmarked(8:00, EST)
nmod(8:00, Monday)
case(Monday, on)
appos(Monday, 8)
nmod:unmarked(8, October)
nmod:unmarked(8, 1963)
For date+time collocations in machine-readable formats, list is recommended:
01/01/1970/NUM 12:30:00/NUM PM/NOUN
list(01/01/1970, 12:30:00)
nmod:unmarked(12:30:00, PM)
History: Prior to release 2.15, case (i) (temporal modifiers) had a separate subtype called nmod:tmod, and nmod:npmod was used for the non-temporal ones.
nmod:unmarked in other languages: [cop] [en] [gd] [he]