1 Introduction 1.1 Notational conventions 1.2 Features 2 ACE texts 3 Queries 3.1 Topicalised questions 3.2 Questions 4 Commands 5 Specifications 5.1 Sentence coordination 5.2 Topicalised sentences 5.3 Composite sentences 6 Verb Phrases 6.1 Verb Phrase Coordination 6.2 Subcategorization of verbs 7 Noun Phrases 7.1 Specifiers 7.2 Variables 7.3 Possessive Noun Phrases 7.4 Arithmetic Terms 8 Prepositional Phrases 9 Relative Clauses 10 Adjective Phrases / Adverb Phrases 11 Function Words 12 Content Words 12.1 Adverbs 12.2 Intransitive Adjectives 12.3 Transitive Adjectives 12.4 Trunk Nouns for Anonymous Pronouns 12.5 Countable Nouns 12.6 Mass Nouns 12.7 Measurement Nouns 12.8 Proper Names 12.9 Ditransitive Verbs 12.10 Transitive Verbs 12.11 Intransitive Verbs
We use some notational conventions in our abstract grammar. The feature values of a constituent are attached to it in square brackets. Required values take a plus sign (+), prohibited values are marked with a minus sign (–). Features which are not discriminatory in the respective rule are not made explicit. Alternatives are separated by a vertical bar (|) and optional elements are put into parentheses.
Gaps left behind by moved constituents are indicated by a slash (/). A minus sign after the slash (/–) indicates that the respective category is not allowed to contain a gap. Feature values that are not indicated explicitly in the abstract grammar are not restricted, or they are passed on by feature percolation.
For many rules we provide an example. Parts of the example that are not covered by the respective rule, but are added to complete a sentence or to make the example more illustrative, are put into square brackets.
We have made explicit use of a limited set of features in the abstract grammar presented in
this report. Nominal constituents are marked as nominative (+NOM
) or oblique (-NOM
), and
they can be either pronominal (+PRO
) or not (-PRO
).
Verb forms can be either finite (-INF
)
or infinite (+INF
). The only infinite verb form that exists in ACE is the bare infinitive. If a
verbal constituent has the copula as its head, it is marked +COPULA
, and if it must not have
a copula head -COPULA
respectively. Other verbs are categorized as intransitive (+ITR
),
transitive (+TR
), or ditransitive (+DITR
).
Furthermore, verbs can be active (-PASSIVE
) or passive (+PASSIVE
).
Noun phrases can either be negative (+NEG
) or non-negative (-NEG
) depending their determiner.
Number, quantification and the WH value are indicated by non-boolean features. A noun
phrase can be definite (+DEF
), existentially quantified (+EXISTS
) or universally quantified
(+FORALL
). Likewise, a non-universally quantified noun phrase, i.e. one that is either definite
or existentially quantified, is marked as -FORALL
. Many constituents can have a WH value. Its
range includes interrogative (+Q
), relative (+R
), either (+WH
), neither (-WH
), non-interrogative
(-Q
) and non-relative (-R
). The number of a constituent is indicated as singular (+SG
), mass
(+MASS
) or plural (+PL
) and the respective complementary values.
Adjectives can be positive (+POS
), comparative (+COMP
), or superlative (+SUP
).
If a sentence coordination is subordinated (+THAT
) then each sentence starts with 'that', which
is not the case for top-level sentence coordinations (-THAT
).
An ACE text can either be a specification, a query or a command. I.e. ACE can be used either in the declarative, the interrogative, or the imperative mode. These different modes of ACE texts cannot be mixed. The declarative mode can be used to assert rules and facts to a knowledge base. Queries can be used to ask for specific information from a knowledge base or try to prove a theorem over it. The imperative mode can be regarded as a special type of declarative mode that can be used to give instructions.
1
ACEText →
Specification | Query | Command
A query can be preceded by one or more declarative sentences.
19
Query →
PropositionOrSentenceCoord Query | TopicalisedQuestion '?'
A topicalised question can start with an existential question topic (23, 24), optionally continued by such that followed by an ordinary declarative sentence coordination.
20
TopicalisedQuestion →
ExistentialQuestionTopic ( such that SentenceCoord[-THAT
] )
Example: Is there a card such that the code of the card is valid [?]
Likewise, a topicalised question can start with an universal topic followed by another topicalised question (21). Note that this is a recursive rule. In example 21, the universal topic is for every card.
21 TopicalisedQuestion → UniversalTopic TopicalisedQuestion Example: For every card does a customer own it [?]
Last, a topicalised question need not have a topic at all but can just be an ordinary question (22).
22
TopicalisedQuestion →
Question
An existential question topic can either consist of an interrogative NP followed by the existential
global question quantor is there (23), or it starts with the existential global question
quantor followed by an NP (24). Like in its declarative counterpart (10), this NP must not
stand in an oblique case (+NOM
) nor must it be unversially quantified (-FORALL
), relative or
interrogative (-WH
). In this way we prevent sentences like
* Is there every card?
* Is there which card?
Furthermore, an NP containing an interrogative element must not be coordinated, preventing sentences like
* A card and what are valid?
* Who and who enters a card?
23 ExistentialQuestionTopic → NP[+NOM
,+Q
] ExistentialGlobalQuestionQuantor Example: Which code is there [?]
24 ExistentialQuestionTopic → ExistentialGlobalQuestionQuantor NPCoord[+NOM
,-FORALL
,-WH
] Example: Is there a card [such that its code is valid ?] Example: Are there a code and a card [?]
There are two types of questions: Yes/no-questions (26–27) and WH-questions (28–34).
25
Question →
YesNoQuestion | WhQuestion
26a
YesNoQuestion →
SentenceQuestion SentenceCoord[+THAT
]
Example: Is it true that John waits [?]
Example: Is it false that John waits [?]
Example: Is it not true that John waits [?]
26
YesNoQuestion →
Aux SentenceWithoutVPCoord[-WH
]/Aux
Example: Does a customer enter a code [?]
Example: Can a customer enter a code [?]
Example: Must a customer enter a code [?]
Example: Is a code entered by a customer [?]
By 'copula' we always denote the forms of be. Note that the VP consists of a gap plus 'valid' in the example of 27. The gap has been left behind by the copula which has been moved to the front of the sentence.
27
YesNoQuestion →
Copula/- SentenceWithoutVPCoord[-WH
]/Copula
Example: Is a card valid [?]
WH-questions can take various forms (28–34). The only WH-question without inversion occurs when one asks for the subject of the sentence (28). In WH-questions that ask for an objekt or an adjunct, the interrogative constituent is either followed by an auxiliary (29, 31, 33) or by the copula (30, 32, 34). In these questions, the interrogative constituent can be an NP (29, 30), a prepositional phrase (31, 32) or an adverb coordination (33, 34). The verbal phrases of these questions contain a gap for the constituent that was moved to the front of the sentence (i.e. the constituent that is asked for). The verb of these verbal phrases stands in the infinitive form. In copula sentences (30, 32, 34), the copula has also been moved to the front and is therefore missing within the VP of that sentence. Interrogative nominal phrases are never coordinations (hence NP instead of NPCoord).
28
WhQuestion →
Sentence[+WH
]
Example: Which customer enters a card [?]
Example: Who enters a card [?]
28b WhQuestion → ExistentialGlobalQuantor NPCoord[+NOM
,-FORALL
,-DEF
,+WH
] Example: There is who [?] Example: There is a man who knows who [?]
29 WhQuestion → NP[+Q
,-NOM
] Aux SentenceWithoutVPCoord/Aux,NP Example: Which code does a customer enter [?] Example: What does a customer enter [?]
30 WhQuestion → NP[+Q
,-NOM
] Copula SentenceWithoutVPCoord/Copula,NP Example: What is a customer interested in [?]
31
WhQuestion →
PP[+Q
] Aux SentenceWithoutVPCoord/Aux,PP
Example: Into what does a customer enter a card [?]
32
WhQuestion →
PP[+Q
] Copula SentenceWithoutVPCoord/Copula,PP
Example: In what is a customer interested [?]
33
WhQuestion →
AdverbCoord[+Q
] Aux SentenceWithoutVPCoord/Aux,AdverbCoord
Example: How does a customer enter a card [?]
Example: Where is a card entered by a customer [?]
34
WhQuestion →
AdverbCoord[+Q
] Copula SentenceWithoutVPCoord/Copula,AdverbCoord
Example: When is a card valid [?]
Commands end with an exclamation mark.
120
Command →
ImperativeSentence '!'
The addressee of the command has to be made explicit at the beginning of the sentence (separated by a comma).
121 ImperativeSentence → NPCoord[+NOM
,-WH
,-NEG
] ',' VP[-WH
,+INF
] Example: John, go to the house [!]
Specifications consist of a sentence coordination followed by a period and optionally one or more subsequent specifications.
2 Specification → PropositionOrSentenceCoord ( Specification ) Example: A man waits. Example: A man waits and a dog barks. Example: A man waits. He eats.
3
PropositionOrSentenceCoord →
PropositionSentence | SentenceCoord[-THAT
]
Sentences can be coordinated by and and or. And refers to the logical conjunction, while or denotes the logical disjunction. The logical conjunction has a higher precedence than the disjunction. Both connectors are right-associative. The expression
A or B and C or D
is therefore ordered like
A ∨ ((B ∧ C) ∨ D)
To enable more combinations, we have introduced comma-and and comma-or. These expressions reverse the order of precedence. To achieve the order
A ∨ (B ∧ (C ∨ D))
we can write
A, or B, and C or D
A sentence coordination in general consists of a sentence coordination of a lower level (thus ensuring right-associativity) optionally followed by the respective connector and a sentence coordination of the same level (4).
4 SentenceCoord → SentenceCoord_1 ( CommaOr SentenceCoord ) SentenceCoord_1 → SentenceCoord_2 ( CommaAnd SentenceCoord_1 ) SentenceCoord_2 → SentenceCoord_3 ( Or SentenceCoord_2 ) SentenceCoord_3[-THAT
] → TopicalisedSentence ( And SentenceCoord_3[-THAT
] ) SentenceCoord_3[+THAT
] → _that_ TopicalisedSentence ( And SentenceCoord_3[+THAT
] ) SentenceCoord_3[+THAT
] → _that_ PropositionVariable ( And SentenceCoord_3[+THAT
] ) Example: A customer enters a card and a clerk enters a code and for every code that the customer enters the card is valid [.] Example: [A man believes] _that_ a dog barks and _that_ a cat eats a cake [.] Example: [A man believes] that a dog barks and that a cat eats a cake [.] Example: [Proposition P: A card is valid. John believes] that P [.]
The function words "Proposition" and "Fact" can be used to define macros. Such macros have a variable which can be used to refer to the macro later in the text.
111 PropositionSentence → 'Proposition' Variable ':' SentenceCoord[-THAT
] PropositionSentence → 'Fact' Variable ':' SentenceCoord[-THAT
] Example: Proposition MyProp12: A dog barks [. It is possible that MyProp12.] Example: Fact X: A dog barks [. John knows that X.]
To avoid scope ambiguities ACE uses the 'principle of surface order' which makes the scope of a quantifier uniquely predictable from the quantifier's position in the sentence. This interpretation principle says that the relative scope of a quantifier corresponds to its surface position. The scope opens at the textual position of the quantified noun phrase and extends to the end of the sentence. If sentences are coordinated, the scope of a quantifier extends only to the end of the sentence conjunct/disjunct containing the quantifier. The sample sentence commonly used to illustrate scope amguities
Every man loves a woman.
is therefore unambiguously interpreted in ACE as
∀ x: (man(x) ⇒ ∃ y: (woman(y) ∧ loves(x,y)))
We may, however, want to express the reverse interpretation
∃ y: (woman(y) ∧ ∀ x: (man(x) ⇒ loves(x,y)))
To achieve the latter interpretation, we have introduced two ACE constructions that allow the user to topicalise a quantifier (i.e. move it to the front of the sentence) and give it a wider scope. The global quantifiers there is/there are (10) and for every/for each (11, 12) expand their scope over the whole sentence. In ACE, the above interpretation can thus be expressed as
There is a woman such that every man loves her.
or
There is a woman that every man loves.
The two readings of the sentence
A man loves every woman.
which are
∃ x: (man(x) ∧ ∀ y: (woman(y) ⇒ loves(x,y)))
∀ y: (woman(y) ⇒ ∃ x: (man(x) ∧ loves(x,y)))
can be made explicit in ACE as
There is a man who loves every woman.
which is equivalent to 'a man loves every woman' and 'for every woman a man loves her'. A topicalised sentence can start with an existential topic (7) or a universal topic (8). It needs, however, not be topicalised at all but can just be an ordinary composite sentence (9).
7
TopicalisedSentence →
ExistentialTopic ( such that SentenceCoord[-THAT
] )
Example: There is a card such that the code of the card is valid [.]
8
TopicalisedSentence →
UniversalTopic SentenceCoord[-THAT
]
Example: For every code there is a card such that the code belongs to it [.]
9
TopicalisedSentence →
CompositeSentence
9a
TopicalisedSentence →
ArithmeticalSentence
Topics consist of a global quantifier enhanced by a noun phrase coordination (i.e. a noun phrase
or a coordination of noun phrases) and a VP coordination (a verb phrase or a coodination of verb
phrases) or by an N' respectively. An existential topic (10) takes the existential global quantifier
there is/there are and an NP coordination that has to be existentally quantified (+EXISTS
)
and must not be in an oblique case (+NOM
).
With these restrictions, we prohibit sentences like
* There is he.
* There is him.
* There is the man.
* There is every customer.
We also prohibit the following
* There are John and Mary.
* There is a man. There are the man and a woman.
10 ExistentialTopic → ExistentialGlobalQuantor NPCoord[+NOM
,+EXISTS
] Example: There is a card [which is valid.] Example: There are a card and a code [such that the code is the code of the card.] Example: There is a man. There is the man's dog. Example: There is John's dog. Example: There are a dog and a cat. Example: There are a man and less than 3 dogs. Example: There are a man and more than 3 dogs.
A universal global quantifier is only followed by an N' and not by a complete NP (* for every a code). This N' must be in the nominative case and can be either in singular (for every card) or plural (for all cards). Since every cannot be used with mass nouns (* every money), we use every in combination with countable nouns (either singular and plural), and all in combination with mass nouns and plural nouns.
11
UniversalTopic →
UniversalGlobalQuantor N'[+NOM
]
Example: For every card [there is a code.]
Example: For all money there is a bank.
Example: For everybody X [a man sees X.]
Example: For all cards [there is a code.]
A universal topic can also start with a distributive global quantifier for each of. The distributive global quantifier must be followed by a complete plural NP coordination in nominative case (12).
12 UniversalTopic → DistributiveGlobalQuantor NPCoord[+PL
,+NOM
] Example: For each of the customers [a clerk enters a code.] Example: For each of a customer and a clerk [a code is valid.] Example: Not for each of the customers [a clerk enters a code.] Example: Not for each of a customer and a clerk [a code is valid.]
13
CompositeSentence →
SentenceInit SentenceCoord[+THAT
]
Example: It is false that John waits [.]
Example: It is possible that John waits [.]
Example: It is not necessary that John waits [.]
A conditional sentence consists of an antecedent, introduced by if, and a consequent, introduced by then. Both the antecedent and the consequent must be complete sentence coordinations.
16 CompositeSentence → if SentenceCoord[-THAT
] then SentenceCoord[-THAT
] Example: If a card is valid then its own code is valid and a customer enters the card [.]
15
CompositeSentence →
Sentence[-WH
]
17
ArithmeticalSentence →
Term '=' Term | Term '\=' Term | Term '<' Term | Term '>' Term | Term '=<' Term | Term '>=' Term
Example: 3+X
> 8
Example: L = {"a","bc","def"}
A simple declarative sentence consists of an NPCoord which is its subject. It must be in nominative case and must not be relative or interrogative. The VPCoord too must not contain any interrogative or relative arguments or modifiers, and its form must be finite.
18 Sentence → NPCoord[+NOM
] VPCoord[-INF
] Example: A customer enters a green card into a slot [.]
18b SentenceWithoutVPCoord → NPCoord[+NOM
] VP[-INF
]
A verb phrase coordination (35) does automatically take the feature value -COPULA
.
And, of course, both coordinated verb phrases have to agree in number and verb form
(either finite or infinite).
35
VPCoord[-COPULA
] →
VPCoord_1 ( CommaOr VPCoord )
VPCoord_1 →
VPCoord_2 ( CommaAnd VPCoord_1 )
VPCoord_2 →
VPCoord_3 ( Or VPCoord_2 )
VPCoord_3 →
VP ( And VPCoord_3 )
Usually, in an declarative context, a verb phrase consists of a V' only (37). However, if the verb phrase is negated, it is preceded by an auxiliary (or the copula respectively) and the negation not (38, 39). Note that the copula has been moved out of the VP in (39). In all cases, the verb form has to be infinite.
38 VP[-INF
] → Auxiliary V'[+INF
,-COPULA
] Example: [A customer] does wait [.] Example: [A customer] is seen by a clerk [.] Example: [A customer] does not enter a card into a slot [.] Example: [A customer] is not seen by a clerk [.] Example: [A customer] can not enter a card [.] Example: [A customer] cannot enter a card [.] Example: [A customer] can't enter a card [.] Example: [A customer] must enter a card [.] Example: [A customer] has to enter a card [.] Example: [Some customers] have to enter a card [.] Example: [A customer] does not have to enter a card [.] Example: [Some customers] do not have to enter a card [.] Example: [Who] does not enter a card into a slot [?]
39 VP[-INF
,+C
opula] → Auxiliary V'[+C
opula]/Copula Example: is not in the garden Example: is not a card Example: [A card] is not valid [.] Example: London can be sunny. Example: London must be sunny. Example: London can't be sunny. Example: London does not have to be sunny. Example: Christmases do not have to be sunny.
37
VP →
V'[-PASSIVE
]
Auxiliaries are split into two parts: Aux and AuxRest. The first part is the one that occurs at the sentence beginning in the case of questions.
130 Auxiliary → Aux AuxRest Auxiliary/Aux → AuxRest Example: is not Example: does not have to
Verb phrase modifiers consist of adverbs and/or prepositional phrases. Adverbs are coordinated (with "and") whereas prepositional phrases are simply concatenated.
131 VModifiers → ( AdverbCoord VModifiersX | PP VModifiers ) VModifiersX → ( PP VModifiers ) Example: in a house in the living-room Example: quickly and thoroughly during one day
In ACE, verbs are subcategorised into intransitive verbs, transitive verbs, ditransitive verbs and the copula. Intransitive verbs take no complement and do not have a passive form.
43
ComplV →
V[+ITR
]
Example: [A customer] waits [.]
Ditransitive verbs take two complements. Rule 48 comes in two versions. Rule 48b is the version with a NP and a PP. Rule 48a is the version with two NPs.
48b ComplV → V[+DITR
] NPCoord[-R
,-NOM
] PP[-R
] Example: [A man] gives a book to a girl [.]
84c ComplV → V[+DITR
,+PASSIVE
] PP[-R
] by NPCoord[-R
,-NOM
] Example: [A book] is sent to Mary by John [.] Example: [A direct object] is sent an indirect object by a subject [.] Example: [It] is sent to her by John [.] Example: [It] is sent to Mary by him [.]
84d ComplV → V[+DITR
,+PASSIVE
] NPCoord[-R
,-NOM
] by NPCoord[-R
,-NOM
] Example: [Mary] is sent a book by John [.] Example: [An indirect object] is sent a direct object by a subject [.] Example: [Mary] is sent it by John [.] Example: [Mary] is sent him by John [.] Example: [Mary] is sent a book by her [.]
Transitive verbs take one complement.
45
ComplV →
V[+TR
] NPCoord
Transitive verbs can also be in passive. In this case the participle form of the verb is used and the verb is surrounded by 'is' and 'by'. The only difference with the active form of the verb is that the arguments are inverted in the predicate condition.
45a ComplV → V[+TR
,+PASSIVE
] by NPCoord Example: [John] is seen by Mary [.] Example: [John] is not seen by Mary [.] Example: [John] is quickly seen by Mary [.] Example: [Some men] are seen by Mary [.] Example: [Are some men] seen by Mary [?] Example: * There is a man who Mary is seen by. Example: [A book] can be written by a man [.] Example: [Are some men] seen by Mary [?] Example: [Can a book] be written by a man [?]
48a ComplV → V[+DITR
] NPCoord[-R
,-NOM
] NPCoord[-R
,-NOM
] Example: [A man] gives a girl a book [.]
The copula always takes a complement. Note that the constraints on complements of copula verbs are different from the constraints on the complements of non-copula verbs.
50
ComplV →
Copula CopulaCompl
The copula can take adjective phrase coordinations (is taller than his brother), noun phrase coordinations that are non-relative and non-pronominal (is a customer), prepositional phrases (is in the garden), and query pronouns (is where?) as its complement (52).
52 CopulaCompl → APCoord | NPCoord[-PRO
,-NOM
,-R
] | PP[-R
] | AdverbCoord[+Q
] Example: [A card is] valid [.] Example: [John is] a customer [.] Example: [A card is] in a slot [.] Example: [A code is] 1234 [.] Example: [A text is] "abcdef" [.] Example: [What is] when [?]
A verb (and its complements) can optionally be modified by an adverb coordination preceding the verb and/or by verbal modifiers following the verb and its complements (41). Such verbal modifiers (adjuncts) can be adverb coordinations, or prepositional phrases. A modifier is either a PP or an adverb. Adverb(s) can precede or follow the verb, while PPs always follow the verb.
41
V' →
( AdverbCoord ) ComplV VModifiers
Transitive verbs can take a sentential complement, marked by that.
41b V' → (AdverbCoord) V[+TR
] VModifiers SentenceCoord[+THAT
] Example: [John] knows that Mary sleeps [.] Example: [John] knows wisely in the park that Mary sleeps [.] Example: [John] does not know that Mary sleeps [.]
A noun phrase coordination can consist of the intermediate category that we call 'unmarked NP coordination'
either preceded (53) or not by the distributive marker each of (54). A distributive noun phrase
coordination is always singular (+SG
) as a whole, but its distributive marker can only precede a
non-negative unmarked NP coordination that is plural and non-universally quantified (53). Noun phrases
like * each of no customer, * each of a customer or * each of every customer do not make sense.
We use the distributive marker each of, as well as its global counterpart for each of (12), to
resolve plural ambiguities. Unmarked NP coordinations that are not preceded by a distributive marker have no
restrictive feature values (54).
53 NPCoord[+SG
]/- → DistributiveMarker UnmarkedNPCoord[-NOM
,-FORALL
,+PL
] Example: each of some customers Example: each of them
54 NPCoord/- → UnmarkedNPCoord Example: some customers Example: he Example: they Example: a card that is valid
A noun phrase coordination can be a gap if it has been moved to the front of the sentence in a question.
56
NPCoord/NPCoord →
[]
If an unmarked NP coordination is an actual coordination of noun phrases, then it is automatically
assigned plural number (+PL
). Interrogative, relative, negative or universally quantified NPs
cannot be coordinated in ACE; hence the -WH
, -NEG
and -FORALL
restriction in the LHS and RHS of
rule. Simple (not coordinated) noun phrases do not underlie such restrictions.
57 UnmarkedNPCoord[+PL
,-FORALL
,-NEG
,-WH
] → NP[-FORALL
,-NEG
,-WH
] and UnmarkedNPCoord[-FORALL
,-NEG
,-WH
] UnmarkedNPCoord → NP Example: a card and a code and a customer Example: [John sees] Mary and more than 2 men [.] Example: [John sees] Mary and less than 2 men [.] Example: * [John sees] Mary and every man [.]
58
NothingButNP →
N'[-SG
] | PropernameNP
59a
NP[-PRO
] →
no N but NothingButNP
Example: Nothing but men [wait .]
Example: [A man enters] nothing but cards [.]
Example: [John eats] nothing but raw and expensive meat [.]
Example: [John likes] nothing but Mary [.]
In general, a noun phrase consists of a specifier followed by an N' (59).
59
NP[-PRO
] →
Specifier N'
Example: a card X
Example: some water
Example: the card that is valid
Example: the card of John
Example: every card
Example: whose card
Example: which card
Example: [a customer] the card of who [is valid enters a code.]
Example: the card of which customer
Example: the card of who
Example: which card of which customer
Example: 4 kg of apples
Example: at most 3 l of milk
Example: 2l of water [flows.]
Example: 2l of water [flow.]
Pronouns (62), proper names and variables (63) are also self-contained noun phrases. Proper names always start with a capital letter. They can be used anaphorically, and they are in the lexicon. Variables also start with a capital letter, but they are not in the lexicon. Variables can be used anaphorically when used in an apposition or as a self-contained NP. They must, however, first be introduced in an apposition.
62
NP[+PRO
] →
Pronoun ( RelativeClauseCoord )
If an NP is a term, is a proper name or is a variable. This is the last alternative for the NP, in order not to overload capitalized function words with an interpretation as a ProperName or Variable. Users are advised not to define and use proper names and variables such as 'It', 'Everybody', 'There' etc. Note that variables and proper names must be capitalized.
63 NP[-PRO
,-WH
] → PropernameNP ( RelativeClauseCoord ) | VariableNP ( RelativeClauseCoord ) | Term
69a
PropernameNP →
Propername
Nouns can be optionally modified by a preceding adjective coordination and/or, following the noun,
an apposition coordination, an of-PP
or a relative clause coordination (64).
64 N' → ( AdjectiveCoord ) N ( VariableApposition ) ( of NPCoord | RelativeClauseCoord ) Example: a red and blue card X of a customer Example: a red and blue card X which is valid
In ACE, specifiers can be determiners, possessive NP coordinations or number phrases. In a measurement noun phrase, however, the specifier consists of a number phrase (2, at least 3) followed by a measurement noun (kg, l, min), and the preposition of. Recursive Saxon Genitive is not supported.
66 Specifier → Determiner | PositiveNumberPhrase | PossessiveNP | PositiveNumberPhrase MeasurementNoun of Example: the Example: every Example: some Example: the clerk's Example: at least 3 Example: 5 Example: 3l of Example: * a clerk's customer's
A positive number phrase consists of a positive number which is optionally preceded by a generalised quantor.
67 PositiveNumberPhrase → GeneralisedQuantor PositiveNumber Example: 3 [men wait.] Example: at most 6 [men wait.] Example: not less than 4 [men wait.] Example: * at least 0 [men wait.] Example: at most 1 [man waits.] Example: * at most 1 [men wait.] Example: [John has] at most 5 [kg of apples.] Example: [John has] 0 [kg of apples.]
Variables in apposition stand immediately after the nouns to which they are attached (see 64). Variables can be arbitrary sequences of letters and numbers, starting with an uppercase letter.
69 VariableApposition → Variable Example: [A man] X [waits.] Example: [A man] M23 [waits.]
69b VariableNP → Variable Example: X Example: Obj54
112 PropositionVariable → Variable Example: [It is false that] X [.] Example: [A man believes that] Prop23 [.]
The specifier of a noun phrase can be a possessive noun phrase. Possessive noun phrases are either noun phrases (a rich man X) that are followed by a Saxon Genitive marker (i.e. so-called genitive noun phrases), or possessive pronouns (his, his own).
76
PossessiveNP →
whose | PossessivePronoun | GenitiveNP
77 GenitiveNP → GenitiveSpecifier GenitiveN' SaxonGenitiveMarker Example: the man 's [dog barks.] Example: a man 's [dog barks.] Example: 5 men 's [dog barks.] Example: everybody 's [dog barks.] Example: [There is] the man 's [card.] Example: [There is] a man 's [card.] Example: [Mary likes] a man 's [fast car.]
78 GenitiveNP → PropernameNP SaxonGenitiveMarker GenitiveNP → VariableNP SaxonGenitiveMarker Example: John 's [card is valid.] Example: [There is a man X.] X 's [card is valid.] Example: Jimmy's [dog barks.] (If declared, 'Jimmy' is interpreted as a variable, otherwise as a proper name) Example: [Mary likes] John 's [fast car.]
Note that GenitiveNP functions as an existentially qualifying determiner.
82 GenitiveSpecifier → Determiner | PositiveNumberPhrase Example: A [man's dog barks.] Example: Some [men's dog barks.] Example: 5 [men's dog barks.]
75 GenitiveN' → (AdjectiveCoord) N (VariableApposition) Example: [A] rich man X ['s dog barks.] Example: [2] men X and Y ['s dog barks.]
Arithmetic terms are treated as normal NPs.
140 Term → Term1 Term1 → Term2 ( Term1_Tail ) Term1_Tail → '+' Term1 | '-' Term1 | '&' Term1 Example: 1234 Example: "abc" & "def" Example: 5*X - 4 - 5
141 Term2 → Term3 ( Term2_Tail ) Term2_Tail → '*' Term2 | '/' Term2 Example: 1234 Example: X * Y * 3 * 5 Example: X / 4
142 Term3 → AtomicTerm | '(' Term ')' | '[' TermList ']' | '{' TermList '}' Example: 1234 Example: ("abc" & "def") Example: [1,2,3] Example: {"a",5,X}
143 TermList → ( Term | Term ',' TermList ) Example: 1,2,3
144 AtomicTerm → String | Integer | Real | PropernameNP | VariableNP Example: "abc" Example: 12 Example: 3.14 Example: John Example: X
There are two types of prepositional phrases in ACE: ordinary prepositional phrases (84) and of-PP
s (64).
Ordinary prepositional phrases consist of a preposition followed by an NP coordination (84). Note that the
prepositional phrase as a whole is always non-pronominal (-PRO
). This is relevant for a prepositional phrase
containing a pronoun as its NP which is the complement of a transitive phrasal verb. Rule 45 states
that only non-pronominal complements can stand after the phrasal particle and prohibits sentences like
A customer looks-up a code.
* A customer looks-up it.
If the complement is a prepositional phrase, it can stand in the end position even if its NP is a pronoun:
A boy looks-up to a famous actor.
A boy looks-up to him.
This means that the pronominal feature is not passed on from an NP to its dominating PP and prepositional phrases are always non-pronominal.
When a prepositional phrase has been moved to the front of a question, it leaves a gap behind (85). Note that, in an implementation, rule (85) can get into an infinite loop, if it is parsed in combination with rule (41) and its gap feature is not properly restricted. An infinite number of emtpy verb modifiers (adjuncts) would be generated in this case.
84
PP[-PRO
] →
Preposition NPCoord
Example: into a slot
Example: to a clerk
Example: with a card and a code
85
PP/PP →
[].
Rule (64) defines at which position relatives clauses occur within an N'. Relative clauses can be coordinated (88). Relative clauses follow the same coordination rules as clauses, with appropriate hierarchy for or, and with and without comma.
88 RelativeClauseCoord → RelativeClauseCoord_1 ( CommaOr RelativeClauseCoord ) RelativeClauseCoord_1 → RelativeClauseCoord_2 ( CommaAnd RelativeClauseCoord_1 ) RelativeClauseCoord_2 → RelativeClauseCoord_3 ( Or RelativeClauseCoord_2 ) RelativeClauseCoord_3 → RelativeClause ( And RelativeClauseCoord_3 ) Example: [a card] that is green and that is valid
A relative clause can either be introduced by the complementizer that (89, 91) or by a relative NP coordination or PP (90, 93). The constituent to which a relative clause refers can either be its subject (89–91), or it can be a complement or adjunct (93). If it is a complement or an adjunct, like in rule 93, then it has been moved from the VP of the relative clause and has left a corresponding gap, namely an NP coordination in an oblique case or a PP, which cannot be interrogative or relative. Furthermore, it is followed by the subject of the relative clause (a non-relative NP coordination in nominative case).
89
RelativeClause →
that VP[-INF
]
Example: [a customer] that enters a card
91 RelativeClause → that NPCoord[-R
,+NOM
] VP[-INF
]/NPCoord[-WH
,-NOM
] Example: [A card] that a customer enters
90 RelativeClause → NPCoord[+R
,+NOM
] VP[-INF
] RelativeClause → NPCoord[+R
,-NOM
] NPCoord[-R
,+NOM
] VPCoord[-INF
]/NPCoord[-WH
,-NOM
] Example: [a table] which is green Example: [a customer] who enters a card Example: [a customer] the card of who is valid Example: [a customer] the card of which is valid Example: [a customer] whose card is valid Example: [A card] which a customer enters Example: [A man] who a customer sees Example: whose brother he sees Example: the code of which someone knows
93 RelativeClause → PP[+R
] NPCoord[-R
,+NOM
] VP[-INF
]/PP[-WH
] Example: at which he looks Example: at each of which he looks
Adjective phrase coordinations (94a, 94b) have a predicative usage only, i.e. they function as the complement of the copula only (50, 52). Attributive usage, modifying a noun (64), is only possible with bare adjective coordinations (101).
Only adjective phrases that do not contain a noun phrase are allowed to be coordinated with "and".
94a APCoord → APCoordX | APItr | APTr Example: more important and rich Example: valid Example: more important than Mary Example: as important as Bill Example: fond-of Mary Example: more fond-of Mary than Bill Example: as fond-of Mary as of Sue
94b APCoordX → APItr ( and APCoordX ) Example: more important and rich
An adjective phrase (AP) can either be an intransitive adjective (99) or a transitive adjective followed by an noun phrase (coordination) (100). Transitive adjective phrases receive the feature values which indicate whether they contain a negative or universally quantified noun phrase as their argument.
100 APTr → TransitiveAdjective NPCoord Example: [A man is] fond-of Mary [.] Example: [A man is] interested-in an account [.]
Note that transitive adjectives must be followed by a noun phrase (coordination) which is linguistically not correct, since instead, they should be followed by a PP. We require the preposition of the PP to be attached to the adjective.
100a
APTr →
TransitiveAdjective[+COMP
] NPCoord than NPCoord
Example: [A man is] more fond-of Mary than Bill [.]
100b
APTr →
TransitiveAdjective[+COMP
] NPCoord than Preposition NPCoord
Example: [A man is] more fond-of Mary than of Bill [.]
100c
APTr →
as TransitiveAdjective[+POS
] NPCoord as NPCoord
Example: [A man is] as fond-of Mary as Bill [.]
100d
APTr →
as TransitiveAdjective[+POS
] NPCoord as Preposition NPCoord
Example: [A man is] as fond-of Mary as of Bill [.]
99 APItr[-FORALL
,-NEG
] → IntransitiveAdjective Example [A] green [card] Example [A card is] green [.]
96
APItr →
IntransitiveAdjective[+COMP
] than NPCoord
Example: [a customer A is] more important than a customer B [.]
96b
APItr →
as IntransitiveAdjective[+POS
] as NPCoord
Example: [a customer A is] as important as a customer B [.]
In attributive position (64), only non-comparative intransitive adjectives or coordinations of them (101) can occur.
101
AdjectiveCoord →
IntransitiveAdjective ( and AdjectiveCoord )
Multiple adverbs must be conjoined by 'and' (not concatenated like PPs).
An adverb coordination can either be a coordination of one or more
non-interrogative (-Q
) adverbs, a single interrogative adverb (102),
or a gap left behind when an adjunct was moved to the front of a question (104).
Note that non-interrogative adverbs are not allowed in copula constructions.
102 AdverbCoord[-Q
] → Adverb ( and AdverbCoord[-Q
] ) AdverbCoord[+Q
] → WhAdverb
104
AdverbCoord/AdverbCoord →
[]
Function words are predefined and cannot be changed by the user.
150
Preposition →
aboard | about | above | across | after | against | along | alongside |
amid | among | amongst | around | as | at | before | behind | below |
beneath | beside | between | beyond | by | despite | down | during | for |
from | in | inside | into | like | near | off | on | onto | out | outside |
over | past | through | throughout | till | to | toward | towards | under |
until | up | upon | via | with | within | without
151
Aux →
( do | does | are | is | must | can )
152
AuxRest →
( not | not provably | not provably not | has to | have to | not have to |
be | have to be | has to be | not have to be | not be )
Only the following combinations of "Aux (...) AuxRest" are possible (the content of Aux is marked by square brackets):
[does] [do] [is] [are] [does] not [do] not [is] not [are] not [does] not provably [do] not provably [is] not provably [are] not provably [does] not provably not [do] not provably not [is] not provably not [are] not provably not [must] [] has to [] have to [does] have to [do] have to [does] not have to [do] not have to [must] be [] has to be [] have to be [does] have to be [do] have to be [does] not have to be [do] not have to be [can] [can] be [can] not [can] not be
Is/are are used for copula and for passives. Rule 153 handles copula, whereas rule 151 is used in case of passives.
153 Copula → is | are Copula/Copula → []
154 CommaAnd → , and CommaOr → , or And → and Or → or
155 Determiner[+DEF
] → the Determiner[+EXISTS
,+SG
] → a | an Determiner[+EXISTS
,-SG
] → some Determiner[+FORALL
,+NEG
] → no Determiner[+FORALL
,+SG
] → every | each Determiner[+FORALL
,+SG
,+NEG
] → not every | not each Determiner[+FORALL
,-SG
] → all Determiner[+FORALL
,-SG
,+NEG
] → not all Determiner[+EXISTS
,-MASS
,+Q
] → which | whose
156
DistributiveMarker →
each of
157
DistributiveGlobalQuantor →
for each of
158 UniversalGlobalQuantor → for | not for Example: for [every] Example: for [all] Example: for [no] Example: not for [every]
159 ExistentialGlobalQuantor → there is | there are ExistentialGlobalQuestionQuantor → is there | are there
160 GeneralisedQuantor → ( at most | at least | more than | less than | not at most | not at least | not more than | not less than ) Example: [John sees] at most [5 dogs.]
161 SentenceInit → it is false | it is not true | it is possible | it is not possible | it is necessary | it is not necessary | it is not provable | it is true | it is not false Example: it is false [that ...] Example: it is not true [that ...] Example: it is possible [that ...]
162 SentenceQuestion → is it true | is it not true | is it false | is it not false Example: Is it true [that ... ?] Example: Is is not false [that ... ?] Example: Is it false [that ... ?] Example: Is it not true [that ... ?]
163
PossessivePronoun →
its | his | her | his/her | her/his | their | its own | his own | her own |
his/her own | her/his own | their own
164
Pronoun →
it | he | she | he/she | she/he | they | him | her | him/her | her/him |
them | itself | himself | herself | himself/herself | herself/himself |
themselves | what | who | which
165 SaxonGenitiveMarker → 's | ' Example: [John]'s [car] Example: [Thomas]' [dog]
167
WhAdverb →
how | where | when
Content words are defined in the lexicon. The users can add their own words.
200
Adverb[+POS
] →
PositiveAdverb
Example: fast
Example: frequently
201 Adverb[+COMP
] → ComparativeAdverb Adverb[+COMP
] → more PositiveAdverb Example: faster Example: more frequently
202 Adverb[+SUP
] → SuperlativeAdverb Adverb[+SUP
] → most PositiveAdverb Example: fastest Example: most frequently
203
IntransitiveAdjective[+POS
] →
PositiveIntransitiveAdjective
Example: rich
Example: expensive
204 IntransitiveAdjective[+COMP
] → ComparativeIntransitiveAdjective IntransitiveAdjective[+COMP
] → more PositiveIntransitiveAdjective Example: richer Example: more expensive
205 IntransitiveAdjective[+SUP
] → SuperlativeIntransitiveAdjective IntransitiveAdjective[+SUP
] → most SuperlativeIntransitiveAdjective Example: richest Example: most expensive
206
TransitiveAdjective[+POS
] →
PositiveTransitiveAdjective
Example: fond-of
Example: pessimistic-about
207 TransitiveAdjective[+COMP
] → ComparativeTransitiveAdjective TransitiveAdjective[+COMP
] → more PositiveTransitiveAdjective Example: fonder-of Example: more pessimistic-about
208 TransitiveAdjective[+SUP
] → SuperlativeTransitiveAdjective TransitiveAdjective[+SUP
] → most PositiveTransitiveAdjective Example: fondest-of Example: most pessimistic-about
Anonymous pronouns like 'something', 'everyone', and 'nobody' are processed as 'a -thing', 'every -one', and 'no -body', respectively. Rule 209 treats the trunks '-thing', '-body', and '-one' as nouns.
209
N →
-thing | -body | -one
210
N[+SG
] →
SingularCountableNoun
Example: mouse
Example: credit-card
211
N[+PL
] →
PluralCountableNoun
Example: mice
Example: credit-cards
212
N[+MASS
] →
MassNoun
Example: water
Example: freedom
213
MeasurementNoun[+SG
] →
SingularMeasurementNoun
Example: kg
Example: man-hour
214
MeasurementNoun[+PL
] →
PluralMeasurementNoun
Example: kg
Example: man-hours
216
Propername[+SG
] →
SingularPropername
Example: John
Example: Switzerland
217
Propername[+PL
] →
PluralPropername
Example: USA
218 V[+DITR
,-PASSIVE
,-INF
,-PL
] → FiniteSingularDitransitiveVerb Example: gives
219 V[+DITR
,-PASSIVE
,-INF
,+PL
] → InfinitePluralDitransitiveVerb Example: give
220 V[+DITR
,-PASSIVE
,+INF
] → InfinitePluralDitransitiveVerb Example: give
221 V[+DITR
,+PASSIVE
] → PastParticipleOfDitransitiveVerb Example: given
222 V[+TR
,-PASSIVE
,-INF
,-PL
] → FiniteSingularTransitiveVerb Example: sees
223 V[+TR
,-PASSIVE
,-INF
,+PL
] → InfinitePluralTransitiveVerb Example: see
224 V[+TR
,-PASSIVE
,+INF
] → InfinitePluralTransitiveVerb Example: see
225 V[+TR
,+PASSIVE
] → PastParticipleOfTransitiveVerb Example: seen
226 V[+ITR
,-PASSIVE
,-INF
,-PL
] → FiniteSingularIntransitiveVerb Example: waits
227 V[+ITR
,-PASSIVE
,-INF
,+PL
] → InfinitePluralIntransitiveVerb Example: wait
228 V[+ITR
,-PASSIVE
,+INF
] → InfinitePluralIntransitiveVerb Example: wait