Files
gf-core/lib/src/api/Constructors.gf

1572 lines
53 KiB
Plaintext

--1 Constructors: the Resource Syntax API
incomplete resource Constructors = open Grammar in {
flags optimize=noexpand ;
-- This module gives access to the syntactic constructions of the
-- GF Resource Grammar library. Its main principle is simple:
-- to construct an object of type $C$, use the function $mkC$.
--
-- For example, an object of type $S$ corresponding to the string
--
-- $John loves Mary$
--
-- is written
--
-- $mkS (mkCl (mkNP (mkPN "John")) (mkV2 "love") (mkNP (mkPN "Mary")))$
--
-- This module defines the syntactic constructors, which take trees as arguments.
-- Lexical constructors, which take strings as arguments, are defined in the
-- $Paradigms$ modules separately for each language.
--
-- The recommended usage of this module is via the wrapper module $Syntax$,
-- which also contains the $Structural$ (structural words).
-- Together with $Paradigms$, $Syntax$ gives everything that is needed
-- to implement the concrete syntax for a language.
--2 Principles of organization
-- To make the library easier to grasp and navigate, we have followed
-- a set of principles when organizing it:
-- + Each category $C$ has an overloaded constructor $mkC$, with value type $C$.
-- + With $mkC$, it is possible to construct any tree of type $C$, except
-- atomic ones, i.e. those that take no arguments, and
-- those whose argument types are exactly the same as in some other instance
-- + To achieve completeness, the library therefore also has
-- for each atomic tree of type $C$, a constant suffixed $C$, and,
-- for other missing constructions, some operation suffixed $C$.
-- These constructors are listed immediately after the $mkC$ group.
-- + Those atomic constructors that are given in $Structural$ are not repeated here.
-- + In addition to the minimally complete set of constructions, many $mkC$ groups
-- include some frequently needed special cases, with two possible logics:
-- default value (to decrease the number of arguments), and
-- direct arguments of an intervening constructor (to flatten the terms).
-- + If such a special case is applied to some category in some rule, it is
-- also applied to all other rules in which the category appears.
-- + The constructors in a group are listed, roughly,
-- *from the most common to the most general*. This does not of course specify
-- a total order.
-- + Optional argument types are marked in parentheses. Although parentheses make no
-- difference in the way the GF compiler treats the types, their presence indicates
-- to the reader that the corresponding arguments can be left out; internally, the
-- library has an overload case for each such combination.
-- + Each constructor case is equipped with an example that is built by that
-- case but could not be built with any other one.
--
--
--2 Texts, phrases, and utterances
--3 Text: texts
-- A text is a list of phrases separated by punctuation marks.
-- The default punctuation mark is the full stop, and the default
-- continuation of a text is empty.
oper
mkText = overload { --%
mkText : Phr -> (Punct) -> (Text) -> Text -- John walks? Yes.
= \phr,punct,text -> case punct of { --%
PFullStop => TFullStop phr text ; --%
PExclMark => TExclMark phr text ; --%
PQuestMark => TQuestMark phr text --%
} ; --%
mkText : Phr -> Text -> Text -- 1. But John walks. Yes! --%
= \x,t -> TFullStop x t ; --%
mkText : Phr -> Punct -> Text --%
= \phr,punct -> case punct of { --%
PFullStop => TFullStop phr TEmpty ; --%
PExclMark => TExclMark phr TEmpty ; --%
PQuestMark => TQuestMark phr TEmpty --%
} ; --%
mkText : Phr -> Text -- 1. But John walks. --%
= \x -> TFullStop x TEmpty ; --%
-- A text can also be directly built from utterances, which in turn can
-- be directly built from sentences, present-tense clauses, questions, or
-- positive imperatives.
mkText : Utt -> Text -- Yes.
= \u -> TFullStop (PhrUtt NoPConj u NoVoc) TEmpty ; --%
mkText : S -> Text -- John walked.
= \s -> TFullStop (PhrUtt NoPConj (UttS s) NoVoc) TEmpty ; --%
mkText : Cl -> Text -- John walks.
= \c -> TFullStop (PhrUtt NoPConj (UttS (TUseCl TPres ASimul PPos c)) NoVoc) TEmpty ; --%
mkText : QS -> Text -- Did John walk?
= \q -> TQuestMark (PhrUtt NoPConj (UttQS q) NoVoc) TEmpty ; --%
mkText : (Pol) -> Imp -> Text -- Walk!
= \p,i -> TExclMark (PhrUtt NoPConj (UttImpSg p i) NoVoc) TEmpty; --%
mkText : Imp -> Text -- Walk! --%
= \i -> TExclMark (PhrUtt NoPConj (UttImpSg PPos i) NoVoc) TEmpty; --%
-- Finally, two texts can be combined into a text.
mkText : Text -> Text -> Text -- Where? When? Here. Now!
= \t,u -> {s = t.s ++ u.s ; lock_Text = <>} ; --%
} ; --%
-- A text can also be empty.
emptyText : Text -- (empty text)
= TEmpty ; --%
--3 Punct: punctuation marks
-- There are three punctuation marks that can separate phrases in a text.
fullStopPunct : Punct -- .
= PFullStop ; --%
questMarkPunct : Punct -- ?
= PQuestMark ; --%
exclMarkPunct : Punct -- !
= PExclMark ; --%
-- Internally, they are handled with a parameter type. --%
param Punct = PFullStop | PExclMark | PQuestMark ; --%
oper --%
--3 Phr: phrases in a text
-- Phrases are built from utterances by adding a phrasal conjunction
-- and a vocative, both of which are by default empty.
mkPhr = overload { --%
mkPhr : (PConj) -> Utt -> (Voc) -> Phr -- but come here John
= PhrUtt ; --%
mkPhr : Utt -> Voc -> Phr -- come here John --%
= \u,v -> PhrUtt NoPConj u v ; --%
mkPhr : PConj -> Utt -> Phr -- but come here --%
= \u,v -> PhrUtt u v NoVoc ; --%
mkPhr : Utt -> Phr -- come here --%
= \u -> PhrUtt NoPConj u NoVoc ; --%
-- A phrase can also be directly built by a sentence, a present-tense
-- clause, a question, or a positive singular imperative.
mkPhr : S -> Phr -- I go home
= \s -> PhrUtt NoPConj (UttS s) NoVoc ; --%
mkPhr : Cl -> Phr -- I go home
= \s -> PhrUtt NoPConj (UttS (TUseCl TPres ASimul PPos s)) NoVoc ; --%
mkPhr : QS -> Phr -- I go home
= \s -> PhrUtt NoPConj (UttQS s) NoVoc ; --%
mkPhr : Imp -> Phr -- I go home
= \s -> PhrUtt NoPConj (UttImpSg PPos s) NoVoc --%
} ; --%
--3 PConj, phrasal conjunctions
-- Any conjunction can be used as a phrasal conjunction.
-- More phrasal conjunctions are defined in $Structural$.
mkPConj : Conj -> PConj -- and
= PConjConj ; --%
noPConj : PConj --%
= NoPConj ; --%
--3 Voc, vocatives
-- Any noun phrase can be turned into a vocative.
-- More vocatives are defined in $Structural$.
mkVoc : NP -> Voc -- John
= VocNP ; --%
noVoc : Voc --%
= NoVoc ; --%
--3 Utt, utterances
-- Utterances are formed from sentences, clauses, questions, and imperatives.
mkUtt = overload {
mkUtt : S -> Utt -- John walked
= UttS ; --%
mkUtt : Cl -> Utt -- John walks
= \c -> UttS (TUseCl TPres ASimul PPos c) ; --%
mkUtt : QS -> Utt -- did John walk
= UttQS ; --%
mkUtt : QCl -> Utt -- does John walk
= \c -> UttQS (TUseQCl TPres ASimul PPos c) ; --%
mkUtt : (ImpForm) -> (Pol) -> Imp -> Utt -- don't love yourselves
= mkUttImp ; --%
mkUtt : ImpForm -> Imp -> Utt -- love yourselves --%
= \f -> mkUttImp f PPos ; --%
mkUtt : Pol -> Imp -> Utt -- don't love yourself --%
= UttImpSg ; --%
mkUtt : Imp -> Utt -- love yourself --%
= UttImpSg PPos ; --%
-- Utterances can also be formed from interrogative phrases and
-- interrogative adverbials, noun phrases, adverbs, and verb phrases.
mkUtt : IP -> Utt -- who
= UttIP ; --%
mkUtt : IAdv -> Utt -- why
= UttIAdv ; --%
mkUtt : NP -> Utt -- this man
= UttNP ; --%
mkUtt : Adv -> Utt -- here
= UttAdv ; --%
mkUtt : VP -> Utt -- to sleep
= UttVP ; --%
mkUtt : CN -> Utt -- beer
= UttCN ; --%
mkUtt : AP -> Utt -- good
= UttAP ; --%
mkUtt : Card -> Utt -- five
= UttCard ; --%
} ; --%
-- The plural first-person imperative is a special construction.
lets_Utt : VP -> Utt -- let's walk
= ImpPl1 ; --%
--2 Auxiliary parameters for phrases and sentences
--3 Pol, polarity
-- Polarity is a parameter that sets a clause to positive or negative
-- form. Since positive is the default, it need never be given explicitly.
positivePol : Pol -- John walks [default]
= PPos ; --%
negativePol : Pol -- John doesn't walk
= PNeg ; --%
--3 Ant, anteriority
-- Anteriority is a parameter that presents an event as simultaneous or
-- anterior to some other reference time.
-- Since simultaneous is the default, it need never be given explicitly.
simultaneousAnt : Ant -- John walks [default]
= ASimul ; --%
anteriorAnt : Ant -- John has walked --# notpresent
= AAnter ; --# notpresent --%
--3 Tense, tense
-- Tense is a parameter that relates the time of an event
-- to the time of speaking about it.
-- Since present is the default, it need never be given explicitly.
presentTense : Tense -- John walks [default]
= TPres ; --%
pastTense : Tense -- John walked --# notpresent
= TPast ; --# notpresent --%
futureTense : Tense -- John will walk --# notpresent
= TFut ; --# notpresent --%
conditionalTense : Tense -- John would walk --# notpresent
= TCond ; --# notpresent --%
--3 ImpForm, imperative form
-- Imperative form is a parameter that sets the form of imperative
-- by reference to the person or persons addressed.
-- Since singular is the default, it need never be given explicitly.
singularImpForm : ImpForm -- help yourself [default]
= IFSg ; --%
pluralImpForm : ImpForm -- help yourselves
= IFPl ; --%
politeImpForm : ImpForm -- help yourself [polite singular]
= IFPol ; --%
-- This is how imperatives are implemented internally. --%
param ImpForm = IFSg | IFPl | IFPol ; --%
oper --%
mkUttImp : ImpForm -> Pol -> Imp -> Utt --%
= \f,p,i -> case f of { --%
IFSg => UttImpSg p i ; --%
IFPl => UttImpPl p i ; --%
IFPol => UttImpPol p i --%
} ; --%
--2 Sentences and clauses
--3 S, sentences
-- A sentence has a fixed tense, anteriority and polarity.
mkS = overload { --%
mkS : Cl -> S --%
= TUseCl TPres ASimul PPos ; --%
mkS : Tense -> Cl -> S --%
= \t -> TUseCl t ASimul PPos ; --%
mkS : Ant -> Cl -> S --%
= \a -> TUseCl TPres a PPos ; --%
mkS : Pol -> Cl -> S --%
= \p -> TUseCl TPres ASimul p ; --%
mkS : Tense -> Ant -> Cl -> S --%
= \t,a -> TUseCl t a PPos ; --%
mkS : Tense -> Pol -> Cl -> S --%
= \t,p -> TUseCl t ASimul p ; --%
mkS : Ant -> Pol -> Cl -> S --%
= \a,p -> TUseCl TPres a p ; --%
mkS : (Tense) -> (Ant) -> (Pol) -> Cl -> S -- John wouldn't have walked
= \t,a -> TUseCl t a ; --%
-- Sentences can be combined with conjunctions. This can apply to a pair
-- of sentences, but also to a list of more than two.
mkS : Conj -> S -> S -> S -- John walks and I run
= \c,x,y -> ConjS c (BaseS x y) ; --%
mkS : Conj -> ListS -> S -- John walks, I run and you sleep
= \c,xy -> ConjS c xy ; --%
-- A sentence can be prefixed by an adverb.
mkS : Adv -> S -> S -- today, John walks
= AdvS ; --%
} ;
--3 Cl, clauses
-- A clause has a variable tense, anteriority and polarity.
-- A clause can be built from a subject noun phrase
-- with a verb, adjective, or noun, and appropriate arguments.
mkCl = overload {
mkCl : NP -> V -> Cl -- John walks
= \s,v -> PredVP s (UseV v); --%
mkCl : NP -> V2 -> NP -> Cl -- John loves her
= \s,v,o -> PredVP s (ComplV2 v o); --%
mkCl : NP -> V3 -> NP -> NP -> Cl -- John sends it to her
= \s,v,o,i -> PredVP s (ComplV3 v o i); --%
mkCl : NP -> VV -> VP -> Cl -- John wants to walk
= \s,v,vp -> PredVP s (ComplVV v vp) ; --%
mkCl : NP -> VS -> S -> Cl -- John says that she walks
= \s,v,p -> PredVP s (ComplVS v p) ; --%
mkCl : NP -> VQ -> QS -> Cl -- John wonders who walks
= \s,v,q -> PredVP s (ComplVQ v q) ; --%
mkCl : NP -> VA -> AP -> Cl -- John becomes old
= \s,v,q -> PredVP s (ComplVA v q) ; --%
mkCl : NP -> V2A -> NP -> AP -> Cl -- John paints it red
= \s,v,n,q -> PredVP s (ComplV2A v n q) ; --%
mkCl : NP -> V2S -> NP -> S -> Cl -- John tells her that we walk
= \s,v,n,q -> PredVP s (ComplSlash (SlashV2S v q) n) ; --%
mkCl : NP -> V2Q -> NP -> QS -> Cl -- John asks her who walks
= \s,v,n,q -> PredVP s (ComplSlash (SlashV2Q v q) n) ; --%
mkCl : NP -> V2V -> NP -> VP -> Cl -- John forces her to walk
= \s,v,n,q -> PredVP s (ComplSlash (SlashV2V v q) n) ; --%
mkCl : NP -> A -> Cl -- John is old
= \x,y -> PredVP x (UseComp (CompAP (PositA y))) ; --%
mkCl : NP -> A -> NP -> Cl -- John is older than her
= \x,y,z -> PredVP x (UseComp (CompAP (ComparA y z))) ; --%
mkCl : NP -> A2 -> NP -> Cl -- John is married to her
= \x,y,z -> PredVP x (UseComp (CompAP (ComplA2 y z))) ; --%
mkCl : NP -> AP -> Cl -- John is very old
= \x,y -> PredVP x (UseComp (CompAP y)) ; --%
mkCl : NP -> NP -> Cl -- John is the man
= \x,y -> PredVP x (UseComp (CompNP y)) ; --%
mkCl : NP -> N -> Cl -- John is a man
= \x,y -> PredVP x (UseComp (CompNP (DetArtSg IndefArt (UseN y)))) ; --%
mkCl : NP -> CN -> Cl -- John is an old man
= \x,y -> PredVP x (UseComp (CompNP (DetArtSg IndefArt y))) ; --%
mkCl : NP -> Adv -> Cl -- John is here
= \x,y -> PredVP x (UseComp (CompAdv y)) ; --%
-- As the general rule, a clause can be built from a subject noun phrase and
-- a verb phrase.
mkCl : NP -> VP -> Cl -- John always walks here
= PredVP ; --%
-- Existentials are a special form of clauses.
mkCl : N -> Cl -- there is a house
= \y -> ExistNP (DetArtSg IndefArt (UseN y)) ; --%
mkCl : CN -> Cl -- there is an old house
= \y -> ExistNP (DetArtSg IndefArt y) ; --%
mkCl : NP -> Cl -- there are five houses
= ExistNP ; --%
-- There are also special forms in which a noun phrase or an adverb is
-- emphasized.
mkCl : NP -> RS -> Cl -- it is John who walks
= CleftNP ; --%
mkCl : Adv -> S -> Cl -- it is here he walks
= CleftAdv ; --%
-- Subjectless verb phrases are used for impersonal actions.
mkCl : V -> Cl -- it rains
= \v -> ImpersCl (UseV v) ; --%
mkCl : VP -> Cl -- it is raining
= ImpersCl ; --%
} ;
-- Generic clauses are those with an impersonal subject.
genericCl : VP -> Cl -- one walks
= GenericCl ; --%
--2 Verb phrases and imperatives
--3 VP, verb phrases
-- A verb phrase is formed from a verb with appropriate arguments.
mkVP = overload {
mkVP : V -> VP -- sleep
= UseV ; --%
mkVP : V2 -> NP -> VP -- love it
= ComplV2 ; --%
mkVP : V3 -> NP -> NP -> VP -- send a message to her
= ComplV3 ; --%
mkVP : VV -> VP -> VP -- want to run
= ComplVV ; --%
mkVP : VS -> S -> VP -- know that she runs
= ComplVS ; --%
mkVP : VQ -> QS -> VP -- wonder if she runs
= ComplVQ ; --%
mkVP : VA -> AP -> VP -- become red
= ComplVA ; --%
mkVP : V2A -> NP -> AP -> VP -- paint it red
= ComplV2A ; --%
mkVP : V2S -> NP -> S -> VP -- tell her that we walk
= \v,n,q -> (ComplSlash (SlashV2S v q) n) ; --%
mkVP : V2Q -> NP -> QS -> VP -- ask her who walks
= \v,n,q -> (ComplSlash (SlashV2Q v q) n) ; --%
mkVP : V2V -> NP -> VP -> VP -- force her to walk
= \v,n,q -> (ComplSlash (SlashV2V v q) n) ; --%
-- The verb can also be a copula ("be"), and the relevant argument is
-- then the complement adjective or noun phrase.
mkVP : A -> VP -- be warm
= \a -> UseComp (CompAP (PositA a)) ; --%
mkVP : A -> NP -> VP -- be older than her
= \y,z -> (UseComp (CompAP (ComparA y z))) ; --%
mkVP : A2 -> NP -> VP -- be married to her
= \y,z -> (UseComp (CompAP (ComplA2 y z))) ; --%
mkVP : AP -> VP -- be warm
= \a -> UseComp (CompAP a) ; --%
mkVP : N -> VP -- be a man
= \y -> (UseComp (CompNP (DetArtSg IndefArt (UseN y)))) ; --%
mkVP : CN -> VP -- be an old man
= \y -> (UseComp (CompNP (DetArtSg IndefArt y))) ; --%
mkVP : NP -> VP -- be this man
= \a -> UseComp (CompNP a) ; --%
mkVP : Adv -> VP -- be here
= \a -> UseComp (CompAdv a) ; --%
-- A verb phrase can be modified with a postverbal or a preverbal adverb.
mkVP : VP -> Adv -> VP -- sleep here
= AdvVP ; --%
mkVP : AdV -> VP -> VP -- always sleep
= AdVVP ; --%
-- Objectless verb phrases can be taken to verb phrases in two ways.
mkVP : VPSlash -> NP -> VP -- paint it black
= ComplSlash ; --%
mkVP : VPSlash -> VP -- paint itself black
= ReflVP
} ; --%
-- Two-place verbs can be used reflexively.
reflexiveVP : V2 -> VP -- love itself
= \v -> ReflVP (SlashV2a v) ; --%
-- Two-place verbs can also be used in the passive, with or without an agent.
passiveVP = overload { --%
passiveVP : V2 -> VP -- be loved
= PassV2 ; --%
passiveVP : V2 -> NP -> VP -- be loved by her
= \v,np -> (AdvVP (PassV2 v) (PrepNP by8agent_Prep np)) ; --%
} ; --%
-- A verb phrase can be turned into the progressive form.
progressiveVP : VP -> VP -- be sleeping
= ProgrVP ; --%
--3 Imp, imperatives
-- Imperatives are formed from verbs and their arguments; as the general
-- rule, from verb phrases.
mkImp = overload { --%
mkImp : VP -> Imp -- go
= ImpVP ; --%
mkImp : V -> Imp -- take it
= \v -> ImpVP (UseV v) ; --%
mkImp : V2 -> NP -> Imp -- come here now
= \v,np -> ImpVP (ComplV2 v np) ; --%
} ; --%
--2 Noun phrases and determiners
--3 NP, noun phrases
-- A noun phrases can be built from a determiner and a common noun ($CN$) .
-- For determiners, the special cases of quantifiers, numerals, integers,
-- and possessive pronouns are provided. For common nouns, the
-- special case of a simple common noun ($N$) is always provided.
mkNP = overload {
mkNP : Quant -> N -> NP -- this man
= \q,n -> DetCN (DetQuant q NumSg) (UseN n) ; --%
mkNP : Quant -> CN -> NP -- this old man
= \q,n -> DetCN (DetQuant q NumSg) n ; --%
mkNP : Quant -> Num -> CN -> NP -- these five old men
= \q,nu,n -> DetCN (DetQuant q nu) n ; --%
mkNP : Quant -> Num -> N -> NP -- these five men
= \q,nu,n -> DetCN (DetQuant q nu) (UseN n) ; --%
mkNP : Det -> CN -> NP -- the first old man
= DetCN ; --%
mkNP : Det -> N -> NP -- the first man
= \d,n -> DetCN d (UseN n) ; --%
mkNP : Numeral -> CN -> NP -- fifty old men
= \d,n -> DetCN (DetArtCard IndefArt (NumNumeral d)) n ; --%
mkNP : Numeral -> N -> NP -- fifty men
= \d,n -> DetCN (DetArtCard IndefArt (NumNumeral d)) (UseN n) ; --%
mkNP : Digits -> CN -> NP -- 51 old men
= \d,n -> DetCN (DetArtCard IndefArt (NumDigits d)) n ; --%
mkNP : Digits -> N -> NP -- 51 men
= \d,n -> DetCN (DetArtCard IndefArt (NumDigits d)) (UseN n) ; --%
mkNP : Digit -> CN -> NP ---- obsol --%
= \d,n -> DetCN (DetArtCard IndefArt (NumNumeral (num (pot2as3 (pot1as2 (pot0as1 (pot0 d))))))) n ; --%
mkNP : Digit -> N -> NP ---- obsol --%
= \d,n -> DetCN (DetArtCard IndefArt (NumNumeral (num (pot2as3 (pot1as2 (pot0as1 (pot0 d))))))) (UseN n) ; --%
mkNP : Card -> CN -> NP -- forty-five old men
= \d,n -> DetCN (DetArtCard IndefArt d) n ; --%
mkNP : Card -> N -> NP -- forty-five men
= \d,n -> DetCN (DetArtCard IndefArt d) (UseN n) ; --%
mkNP : Pron -> CN -> NP -- my old man
= \p,n -> DetCN (DetQuant (PossPron p) NumSg) n ; --%
mkNP : Pron -> N -> NP -- my man
= \p,n -> DetCN (DetQuant (PossPron p) NumSg) (UseN n) ; --%
-- Proper names and pronouns can be used as noun phrases.
mkNP : PN -> NP -- John
= UsePN ; --%
mkNP : Pron -> NP -- he
= UsePron ; --%
-- Determiners alone can form noun phrases.
mkNP : Quant -> NP -- this
= \q -> DetNP (DetQuant q sgNum) ; --%
mkNP : Quant -> Num -> NP -- these five
= \q,n -> DetNP (DetQuant q n) ; --%
mkNP : Det -> NP -- these five best
= DetNP ; --%
-- Determinesless mass noun phrases.
mkNP : CN -> NP -- old beer
= MassNP ; --%
mkNP : N -> NP -- beer
= \n -> MassNP (UseN n) ; --%
-- A noun phrase once formed can be prefixed by a predeterminer and
-- suffixed by a past participle or an adverb.
mkNP : Predet -> NP -> NP -- only the man
= PredetNP ; --%
mkNP : NP -> V2 -> NP -- the man found
= PPartNP ; --%
mkNP : NP -> Adv -> NP -- Paris today
= AdvNP ; --%
mkNP : NP -> RS -> NP -- John, who lives in Paris
= RelNP ; --%
-- A conjunction can be formed both from two noun phrases and a longer
-- list of them.
mkNP : Conj -> NP -> NP -> NP
= \c,x,y -> ConjNP c (BaseNP x y) ; --%
mkNP : Conj -> ListNP -> NP
= \c,xy -> ConjNP c xy ; --%
-- backward compat --%
mkNP : QuantSg -> CN -> NP --%
= \q,n -> DetCN (DetQuant q NumSg) n ; --%
mkNP : QuantPl -> CN -> NP --%
= \q,n -> DetCN (DetQuant q NumPl) n ; --%
} ; --%
--3 Det, determiners
-- A determiner is either a singular or a plural one.
-- Quantifiers that have both singular and plural forms are by default used as
-- singular determiners. If a numeral is added, the plural form is chosen.
-- A determiner also has an optional ordinal.
mkDet = overload { --%
mkDet : Quant -> Det -- this
= \q -> DetQuant q NumSg ; --%
mkDet : Quant -> Card -> Det -- these five
= \d,nu -> (DetQuant d (NumCard nu)) ; --%
mkDet : Quant -> Ord -> Det -- the best
= \q,o -> DetQuantOrd q NumSg o ; --%
mkDet : Quant -> Num -> Ord -> Det -- these five best
= DetQuantOrd ; --%
mkDet : Quant -> Num -> Det -- these five
= DetQuant ; --%
-- Numerals, their special cases integers and digits, and possessive pronouns can be
-- used as determiners.
mkDet : Card -> Det -- forty
= DetArtCard IndefArt ; --%
mkDet : Digits -> Det -- 51
= \d -> DetArtCard IndefArt (NumDigits d) ; --%
mkDet : Numeral -> Det -- five
= \d -> DetArtCard IndefArt (NumNumeral d) ; --%
mkDet : Pron -> Det -- my
= \p -> DetQuant (PossPron p) NumSg ; --%
mkDet : Pron -> Num -> Det -- my five
= \p -> DetQuant (PossPron p) ; --%
} ; --%
the_Det : Det -- the (house)
= theSg_Det ; --%
a_Det : Det -- a (house)
= aSg_Det ; --%
theSg_Det : Det -- the (houses)
= DetQuant DefArt NumSg ; --%
thePl_Det : Det -- the (houses)
= DetQuant DefArt NumPl ; --%
aSg_Det : Det -- a (house)
= DetQuant IndefArt NumSg ; --%
aPl_Det : Det -- (houses)
= DetQuant IndefArt NumPl ; --%
--3 Quant, quantifiers
-- There are definite and indefinite articles.
mkQuant = overload { --%
mkQuant : Pron -> Quant -- my
= PossPron ; --%
} ; --%
the_Quant : Quant -- the
= DefArt ; --%
a_Quant : Quant -- a
= IndefArt ; --%
--3 Num, cardinal numerals
-- Numerals can be formed from number words ($Numeral$), their special case digits,
-- and from symbolic integers.
mkNum = overload { --%
mkNum : Str -> Num -- thirty-five (given by "35")
= \s -> NumCard (str2card s) ; --%
mkNum : Numeral -> Num -- twenty
= \d -> NumCard (NumNumeral d) ; --%
mkNum : Digits -> Num -- 21
= \d -> NumCard (NumDigits d) ; --%
mkNum : Digit -> Num -- five
= \d -> NumCard (NumNumeral (num (pot2as3 (pot1as2 (pot0as1 (pot0 d)))))) ; --%
mkNum : Card -> Num -- almost ten
= NumCard ; --%
-- A numeral can be modified by an adnumeral.
mkNum : AdN -> Card -> Num -- almost ten
= \a,c -> NumCard (AdNum a c)
} ; --%
-- Dummy numbers are sometimes to select the grammatical number of a determiner.
singularNum : Num -- singular
= NumSg ; --%
pluralNum : Num -- plural
= NumPl ; --%
-- Cardinals are the non-dummy numerals.
mkCard = overload { --%
mkCard : Str -> Card -- thirty-five (given as "35")
= str2card ; --%
mkCard : Numeral -> Card -- twenty
= NumNumeral ; --%
mkCard : Digits -> Card -- 51
= NumDigits ; --%
mkCard : AdN -> Card -> Card -- almost fifty
= AdNum ; --%
} ; --%
--3 Ord, ordinal numerals
-- Just like cardinals, ordinals can be formed from number words ($Numeral$)
-- and from symbolic integers.
mkOrd = overload { --%
mkOrd : Numeral -> Ord -- twentieth
= OrdNumeral ; --%
mkOrd : Digits -> Ord -- 51st
= OrdDigits ; --%
mkOrd : Digit -> Ord -- fifth
= \d -> OrdNumeral (num (pot2as3 (pot1as2 (pot0as1 (pot0 d))))) ; --%
-- Also adjectives in the superlative form can appear on ordinal positions.
mkOrd : A -> Ord -- largest
= OrdSuperl ; --%
} ; --%
--3 AdN, adnumerals
-- Comparison adverbs can be used as adnumerals.
mkAdN : CAdv -> AdN -- more than
= AdnCAdv ; --%
--3 Numeral, number words
-- Numerals can be extracted from strings at compile time.
mkNumeral = overload { --%
mkNumeral : Str -> Numeral -- thirty-five (given by "35")
= str2numeral ; --%
} ; --%
-- Some "round" numbers are here given as shorthands.
n1_Numeral : Numeral
= num (pot2as3 (pot1as2 (pot0as1 pot01))) ; --%
n2_Numeral : Numeral
= num (pot2as3 (pot1as2 (pot0as1 (pot0 n2)))) ; --%
n3_Numeral : Numeral
= num (pot2as3 (pot1as2 (pot0as1 (pot0 n3)))) ; --%
n4_Numeral : Numeral
= num (pot2as3 (pot1as2 (pot0as1 (pot0 n4)))) ; --%
n5_Numeral : Numeral
= num (pot2as3 (pot1as2 (pot0as1 (pot0 n5)))) ; --%
n6_Numeral : Numeral
= num (pot2as3 (pot1as2 (pot0as1 (pot0 n6)))) ; --%
n7_Numeral : Numeral
= num (pot2as3 (pot1as2 (pot0as1 (pot0 n7)))) ; --%
n8_Numeral : Numeral
= num (pot2as3 (pot1as2 (pot0as1 (pot0 n8)))) ; --%
n9_Numeral : Numeral
= num (pot2as3 (pot1as2 (pot0as1 (pot0 n9)))) ; --%
n10_Numeral : Numeral
= num (pot2as3 (pot1as2 pot110)) ; --%
n20_Numeral : Numeral
= num (pot2as3 (pot1as2 (pot1 n2))) ; --%
n100_Numeral : Numeral
= num (pot2as3 (pot2 pot01)) ; --%
n1000_Numeral : Numeral
= num (pot3 (pot1as2 (pot0as1 pot01))) ; --%
-- See $Numeral$ for the full set of constructors, and use the category
-- $Digits$ for other numbers from one million.
--3 Digits, numerals as sequences of digits
mkDigits = overload { --%
mkDigits : Str -> Digits
= str2digits ; --%
mkDigits : Dig -> Digits
= IDig ; --%
mkDigits : Dig -> Digits -> Digits
= IIDig ; --%
} ; --%
n1_Digits : Digits
= IDig D_1 ; --%
n2_Digits : Digits
= IDig D_2 ; --%
n3_Digits : Digits
= IDig D_3 ; --%
n4_Digits : Digits
= IDig D_4 ; --%
n5_Digits : Digits
= IDig D_5 ; --%
n6_Digits : Digits
= IDig D_6 ; --%
n7_Digits : Digits
= IDig D_7 ; --%
n8_Digits : Digits
= IDig D_8 ; --%
n9_Digits : Digits
= IDig D_9 ; --%
n10_Digits : Digits
= IIDig D_1 (IDig D_0) ; --%
n20_Digits : Digits
= IIDig D_2 (IDig D_0) ; --%
n100_Digits : Digits
= IIDig D_1 (IIDig D_0 (IDig D_0)) ; --%
n1000_Digits : Digits
= IIDig D_1 (IIDig D_0 (IIDig D_0 (IDig D_0))) ; --%
--3 Dig, single digits
n0_Dig : Dig
= D_0 ; --%
n1_Dig : Dig
= D_1 ; --%
n2_Dig : Dig
= D_2 ; --%
n3_Dig : Dig
= D_3 ; --%
n4_Dig : Dig
= D_4 ; --%
n5_Dig : Dig
= D_5 ; --%
n6_Dig : Dig
= D_6 ; --%
n7_Dig : Dig
= D_7 ; --%
n8_Dig : Dig
= D_8 ; --%
n9_Dig : Dig
= D_9 ; --%
--2 Nouns
--3 CN, common noun phrases
mkCN = overload { --%
-- The simplest way of forming common noun phrases is from atomic nouns $N$.
mkCN : N -> CN -- house
= UseN ; --%
-- Common noun phrases can be formed from relational nouns by providing arguments.
mkCN : N2 -> NP -> CN -- mother of John
= ComplN2 ; --%
mkCN : N3 -> NP -> NP -> CN -- distance from this city to Paris
= \f,x -> ComplN2 (ComplN3 f x) ; --%
-- Relational nouns can also be used without their arguments.
mkCN : N2 -> CN -- mother
= UseN2 ; --%
mkCN : N3 -> CN -- distance
= \n -> UseN2 (Use2N3 n) ; --%
-- A common noun phrase can be modified by an adjectival phrase. We give special
-- cases of this, where one or both of the arguments are atomic.
mkCN : A -> N -> CN -- big house
= \x,y -> AdjCN (PositA x) (UseN y); --%
mkCN : A -> CN -> CN -- big blue house
= \x,y -> AdjCN (PositA x) y; --%
mkCN : AP -> N -> CN -- very big house
= \x,y -> AdjCN x (UseN y) ; --%
mkCN : AP -> CN -> CN -- very big blue house
= AdjCN ; --%
mkCN : CN -> AP -> CN -- very big blue house --%
= \x,y -> AdjCN y x ; --%
mkCN : N -> AP -> CN -- very big house --%
= \x,y -> AdjCN y (UseN x) ; --%
-- A common noun phrase can be modified by a relative clause or an adverb.
mkCN : N -> RS -> CN -- house that John owns
= \x,y -> RelCN (UseN x) y ; --%
mkCN : CN -> RS -> CN -- big house that John loves
= RelCN ; --%
mkCN : N -> Adv -> CN -- house on the hill
= \x,y -> AdvCN (UseN x) y ; --%
mkCN : CN -> Adv -> CN -- big house on the hill
= AdvCN ; --%
-- For some nouns it makes sense to modify them by sentences,
-- questions, or infinitives. But syntactically this is possible for
-- all nouns.
mkCN : CN -> S -> CN -- rule that John walks
= \cn,s -> SentCN cn (EmbedS s) ; --%
mkCN : CN -> QS -> CN -- question if John walks
= \cn,s -> SentCN cn (EmbedQS s) ; --%
mkCN : CN -> VP -> CN -- reason to walk
= \cn,s -> SentCN cn (EmbedVP s) ; --%
-- A noun can be used in apposition to a noun phrase, especially a proper name.
mkCN : N -> NP -> CN -- king John
= \x,y -> ApposCN (UseN x) y ; --%
mkCN : CN -> NP -> CN -- old king John
= ApposCN ; --%
} ; --%
--2 Adjectives and adverbs
--3 AP, adjectival phrases
mkAP = overload { --%
-- Adjectival phrases can be formed from atomic adjectives by using the positive form or
-- the comparative with a complement
mkAP : A -> AP -- warm
= PositA ; --%
mkAP : A -> NP -> AP -- warmer than Paris
= ComparA ; --%
-- Relational adjectives can be used with a complement or a reflexive
mkAP : A2 -> NP -> AP -- married to her
= ComplA2 ; --%
mkAP : A2 -> AP -- married
= UseA2 ; --%
-- Some adjectival phrases can take as complements sentences,
-- questions, or infinitives. Syntactically this is possible for
-- all adjectives.
mkAP : AP -> S -> AP -- probable that John walks
= \ap,s -> SentAP ap (EmbedS s) ; --%
mkAP : AP -> QS -> AP -- uncertain if John walks
= \ap,s -> SentAP ap (EmbedQS s) ; --%
mkAP : AP -> VP -> AP -- ready to go
= \ap,s -> SentAP ap (EmbedVP s) ; --%
-- An adjectival phrase can be modified by an adadjective.
mkAP : AdA -> A -> AP -- very old
=\x,y -> AdAP x (PositA y) ; --%
mkAP : AdA -> AP -> AP -- very very old
= AdAP ; --%
-- Conjunction can be formed from two or more adjectival phrases.
mkAP : Conj -> AP -> AP -> AP -- old and big
= \c,x,y -> ConjAP c (BaseAP x y) ; --%
mkAP : Conj -> ListAP -> AP -- old, big and warm
= \c,xy -> ConjAP c xy ; --%
-- Two more constructions.
mkAP : Ord -> AP -- oldest
= AdjOrd ; --%
mkAP : CAdv -> AP -> NP -> AP -- as old as John
= CAdvAP ; --%
} ; --%
reflAP : A2 -> AP -- married to himself
= ReflA2 ; --%
comparAP : A -> AP -- warmer
= UseComparA ; --%
--3 Adv, adverbial phrases
mkAdv = overload { --%
-- Adverbs can be formed from adjectives.
mkAdv : A -> Adv -- warmly
= PositAdvAdj ; --%
-- Prepositional phrases are treated as adverbs.
mkAdv : Prep -> NP -> Adv -- in the house
= PrepNP ; --%
-- Subordinate sentences are treated as adverbs.
mkAdv : Subj -> S -> Adv -- when John walks
= SubjS ; --%
-- An adjectival adverb can be compared to a noun phrase or a sentence.
mkAdv : CAdv -> A -> NP -> Adv -- more warmly than John
= ComparAdvAdj ; --%
mkAdv : CAdv -> A -> S -> Adv -- more warmly than he runs
= ComparAdvAdjS ; --%
-- Adverbs can be modified by adadjectives.
mkAdv : AdA -> Adv -> Adv -- very warmly
= AdAdv ; --%
-- Conjunction can be formed from two or more adverbial phrases.
mkAdv : Conj -> Adv -> Adv -> Adv -- here and now
= \c,x,y -> ConjAdv c (BaseAdv x y) ; --%
mkAdv : Conj -> ListAdv -> Adv -- with John, here and now
= \c,xy -> ConjAdv c xy ; --%
} ; --%
--2 Questions and relatives
--3 QS, question sentences
mkQS = overload { --%
-- Just like a sentence $S$ is built from a clause $Cl$,
-- a question sentence $QS$ is built from
-- a question clause $QCl$ by fixing tense, anteriority and polarity.
-- Any of these arguments can be omitted, which results in the
-- default (present, simultaneous, and positive, respectively).
mkQS : QCl -> QS --%
= TUseQCl TPres ASimul PPos ; --%
mkQS : Tense -> QCl -> QS --%
= \t -> TUseQCl t ASimul PPos ; --%
mkQS : Ant -> QCl -> QS --%
= \a -> TUseQCl TPres a PPos ; --%
mkQS : Pol -> QCl -> QS --%
= \p -> TUseQCl TPres ASimul p ; --%
mkQS : Tense -> Ant -> QCl -> QS --%
= \t,a -> TUseQCl t a PPos ; --%
mkQS : Tense -> Pol -> QCl -> QS --%
= \t,p -> TUseQCl t ASimul p ; --%
mkQS : Ant -> Pol -> QCl -> QS --%
= \a,p -> TUseQCl TPres a p ; --%
mkQS : (Tense) -> (Ant) -> (Pol) -> QCl -> QS -- who wouldn't have walked
= TUseQCl ; --%
-- Since 'yes-no' question clauses can be built from clauses (see below),
-- we give a shortcut
-- for building a question sentence directly from a clause, using the defaults
-- present, simultaneous, and positive.
mkQS : Cl -> QS
= \x -> TUseQCl TPres ASimul PPos (QuestCl x) ; --%
} ; --%
--3 QCl, question clauses
mkQCl = overload { --%
-- 'Yes-no' question clauses are built from 'declarative' clauses.
mkQCl : Cl -> QCl -- does John walk
= QuestCl ; --%
-- 'Wh' questions are built from interrogative pronouns in subject
-- or object position. The former uses a verb phrase; we don't give
-- shortcuts for verb-argument sequences as we do for clauses.
-- The latter uses the 'slash' category of objectless clauses
-- (see below); we give the common special case with a two-place verb.
mkQCl : IP -> VP -> QCl -- who walks
= QuestVP ; --%
mkQCl : IP -> V -> QCl -- who walks
= \s,v -> QuestVP s (UseV v); --%
mkQCl : IP -> V2 -> NP -> QCl -- who loves her
= \s,v,o -> QuestVP s (ComplV2 v o); --%
mkQCl : IP -> V3 -> NP -> NP -> QCl -- who sends it to her
= \s,v,o,i -> QuestVP s (ComplV3 v o i); --%
mkQCl : IP -> VV -> VP -> QCl -- who wants to walk
= \s,v,vp -> QuestVP s (ComplVV v vp) ; --%
mkQCl : IP -> VS -> S -> QCl -- who says that she walks
= \s,v,p -> QuestVP s (ComplVS v p) ; --%
mkQCl : IP -> VQ -> QS -> QCl -- who wonders who walks
= \s,v,q -> QuestVP s (ComplVQ v q) ; --%
mkQCl : IP -> VA -> AP -> QCl -- who becomes old
= \s,v,q -> QuestVP s (ComplVA v q) ; --%
mkQCl : IP -> V2A -> NP -> AP -> QCl -- who paints it red
= \s,v,n,q -> QuestVP s (ComplV2A v n q) ; --%
mkQCl : IP -> V2S -> NP -> S -> QCl -- who tells her that we walk
= \s,v,n,q -> QuestVP s (ComplSlash (SlashV2S v q) n) ; --%
mkQCl : IP -> V2Q -> NP -> QS -> QCl -- who asks her who walks
= \s,v,n,q -> QuestVP s (ComplSlash (SlashV2Q v q) n) ; --%
mkQCl : IP -> V2V -> NP -> VP -> QCl -- who forces her to walk
= \s,v,n,q -> QuestVP s (ComplSlash (SlashV2V v q) n) ; --%
mkQCl : IP -> A -> QCl -- who is old
= \x,y -> QuestVP x (UseComp (CompAP (PositA y))) ; --%
mkQCl : IP -> A -> NP -> QCl -- who is older than her
= \x,y,z -> QuestVP x (UseComp (CompAP (ComparA y z))) ; --%
mkQCl : IP -> A2 -> NP -> QCl -- who is married to her
= \x,y,z -> QuestVP x (UseComp (CompAP (ComplA2 y z))) ; --%
mkQCl : IP -> AP -> QCl -- who is very old
= \x,y -> QuestVP x (UseComp (CompAP y)) ; --%
mkQCl : IP -> NP -> QCl -- who is the man
= \x,y -> QuestVP x (UseComp (CompNP y)) ; --%
mkQCl : IP -> N -> QCl -- who is a man
= \x,y -> QuestVP x (UseComp (CompNP (DetArtSg IndefArt (UseN y)))) ; --%
mkQCl : IP -> CN -> QCl -- who is an old man
= \x,y -> QuestVP x (UseComp (CompNP (DetArtSg IndefArt y))) ; --%
mkQCl : IP -> Adv -> QCl -- who is here
= \x,y -> QuestVP x (UseComp (CompAdv y)) ; --%
mkQCl : IP -> NP -> V2 -> QCl -- who does John love
= \ip,np,v -> QuestSlash ip (SlashVP np (SlashV2a v)) ; --%
mkQCl : IP -> ClSlash -> QCl -- who does John today
= QuestSlash ; --%
-- Adverbial 'wh' questions are built with interrogative adverbials, with the
-- special case of prepositional phrases with interrogative pronouns.
mkQCl : IAdv -> Cl -> QCl -- why does John walk
= QuestIAdv ; --%
mkQCl : Prep -> IP -> Cl -> QCl -- with whom does John walk
= \p,ip -> QuestIAdv (PrepIP p ip) ; --%
-- An interrogative adverbial can serve as the complement of a copula.
mkQCl : IAdv -> NP -> QCl -- where is John
= \a -> QuestIComp (CompIAdv a) ; --%
-- Asking about a known subject.
mkQCl : IComp -> NP -> QCl -- who is this man
= \a -> QuestIComp a ; --%
-- Existentials are a special construction.
mkQCl : IP -> QCl -- which houses are there
= ExistIP ; --%
} ; --%
--3 IP, interrogative pronouns
mkIP = overload { --%
-- Interrogative pronouns
-- can be formed much like noun phrases, by using interrogative quantifiers.
mkIP : IDet -> CN -> IP -- which five big cities
= IdetCN ; --%
mkIP : IDet -> N -> IP -- which five cities
= \i,n -> IdetCN i (UseN n) ; --%
mkIP : IQuant -> CN -> IP -- which big cities
= \i,n -> IdetCN (IdetQuant i NumSg) n ; --%
mkIP : IQuant -> Num -> CN -> IP -- which five cities
= \i,nu,n -> IdetCN (IdetQuant i nu) n ; --%
mkIP : IQuant -> N -> IP -- which city
= \i,n -> IdetCN (IdetQuant i NumSg) (UseN n) ; --%
-- An interrogative pronoun can be modified by an adverb.
mkIP : IP -> Adv -> IP -- who in Paris
= AdvIP ; --%
} ; --%
what_IP : IP -- what (singular)
= whatSg_IP ; --%
who_IP : IP -- who (singular)
= whoSg_IP ; --%
-- More interrogative pronouns and determiners can be found in $Structural$.
--3 IAdv, interrogative adverbs.
-- In addition to the interrogative adverbs defined in the $Structural$ lexicon, they
-- can be formed as prepositional phrases from interrogative pronouns.
mkIAdv = overload { --%
mkIAdv : Prep -> IP -> IAdv -- in which city
= PrepIP ; --%
mkIAdv : IAdv -> Adv -> IAdv -- where in Paris
= AdvIAdv ; --%
} ; --%
-- More interrogative adverbs are given in $Structural$.
--3 IDet, interrogative determiners
mkIDet = overload { --%
mkIDet : IQuant -> Num -> IDet -- which (songs)
= \i,nu -> IdetQuant i nu ; --%
mkIDet : IQuant -> IDet
= \i -> IdetQuant i NumSg ; --%
} ; --%
which_IDet : IDet
= whichSg_IDet ; --%
whichSg_IDet : IDet --%
= IdetQuant which_IQuant NumSg ; --%
whichPl_IDet : IDet
= IdetQuant which_IQuant NumPl ; --%
--3 RS, relative sentences
-- Just like a sentence $S$ is built from a clause $Cl$,
-- a relative sentence $RS$ is built from
-- a relative clause $RCl$ by fixing the tense, anteriority and polarity.
-- Any of these arguments
-- can be omitted, which results in the default (present, simultaneous,
-- and positive, respectively).
mkRS = overload { --%
mkRS : RCl -> RS --%
= TUseRCl TPres ASimul PPos ; --%
mkRS : Tense -> RCl -> RS --%
= \t -> TUseRCl t ASimul PPos ; --%
mkRS : Ant -> RCl -> RS --%
= \a -> TUseRCl TPres a PPos ; --%
mkRS : Pol -> RCl -> RS --%
= \p -> TUseRCl TPres ASimul p ; --%
mkRS : Tense -> Ant -> RCl -> RS --%
= \t,a -> TUseRCl t a PPos ; --%
mkRS : Tense -> Pol -> RCl -> RS --%
= \t,p -> TUseRCl t ASimul p ; --%
mkRS : Ant -> Pol -> RCl -> RS --%
= \a,p -> TUseRCl TPres a p ; --%
mkRS : (Tense) -> (Ant) -> (Pol) -> RCl -> RS -- that wouldn't have walked
= TUseRCl ; --%
mkRS : Conj -> RS -> RS -> RS -- who walks and whose mother runsx
= \c,x,y -> ConjRS c (BaseRS x y) ; --%
mkRS : Conj -> ListRS -> RS -- who walks, whom I see and who sleeps
= \c,xy -> ConjRS c xy ; --%
} ; --%
--3 RCl, relative clauses
mkRCl = overload { --%
-- Relative clauses are built from relative pronouns in subject or object position.
-- The former uses a verb phrase; we don't give
-- shortcuts for verb-argument sequences as we do for clauses.
-- The latter uses the 'slash' category of objectless clauses (see below);
-- we give the common special case with a two-place verb.
mkRCl : RP -> VP -> RCl -- that loves John
= RelVP ; --%
mkRCl : RP -> ClSlash -> RCl -- whom John loves today
= RelSlash ; --%
mkRCl : RP -> NP -> V2 -> RCl -- whom John loves
= \rp,np,v2 -> RelSlash rp (SlashVP np (SlashV2a v2)) ; --%
-- There is a simple 'such that' construction for forming relative
-- clauses from clauses.
mkRCl : Cl -> RCl -- such that John loves her
= RelCl ; --%
} ; --%
--3 RP, relative pronouns
-- There is an atomic relative pronoun
which_RP : RP -- which
= IdRP ; --%
-- A relative pronoun can be made into a kind of a prepositional phrase.
mkRP : Prep -> NP -> RP -> RP -- all the houses in which
= FunRP ; --%
--3 ClSlash, objectless sentences
mkClSlash = overload { --%
-- Objectless sentences are used in questions and relative clauses.
-- The most common way of constructing them is by using a two-place verb
-- with a subject but without an object.
mkClSlash : NP -> VPSlash -> ClSlash -- (whom) he sees here
= \np,vps -> SlashVP np vps ; --%
mkClSlash : NP -> V2 -> ClSlash -- (whom) he sees
= \np,v2 -> SlashVP np (SlashV2a v2) ; --%
-- The two-place verb can be separated from the subject by a verb-complement verb.
mkClSlash : NP -> VV -> V2 -> ClSlash -- (whom) he wants to see
= \np,vv,v2 -> SlashVP np (SlashVV vv (SlashV2a v2)) ; --%
-- The missing object can also be the noun phrase in a prepositional phrase.
mkClSlash : Cl -> Prep -> ClSlash -- (with whom) he walks
= SlashPrep ; --%
-- An objectless sentence can be modified by an adverb.
mkClSlash : ClSlash -> Adv -> ClSlash -- (whom) he sees tomorrow
= AdvSlash ; --%
} ; --%
--3 VPSlash, verb phrases missing an object
mkVPSlash = overload { --%
-- This is the deep level of many-argument predication, permitting extraction.
mkVPSlash : V2 -> VPSlash -- (whom) (John) loves
= SlashV2a ; --%
mkVPSlash : V3 -> NP -> VPSlash -- (whom) (John) gives an apple
= Slash2V3 ; --%
mkVPSlash : V2A -> AP -> VPSlash -- (whom) (John) paints red
= SlashV2A ; --%
mkVPSlash : V2Q -> QS -> VPSlash -- (whom) (John) asks who sleeps
= SlashV2Q ; --%
mkVPSlash : V2S -> S -> VPSlash -- (whom) (John) tells that we sleep
= SlashV2S ; --%
mkVPSlash : V2V -> VP -> VPSlash -- (whom) (John) forces to sleep
= SlashV2V ; --%
} ; --%
--2 Lists for coordination
-- The rules in this section are very uniform: a list can be built from two or more
-- expressions of the same category.
--3 ListS, sentence lists
mkListS = overload { --%
mkListS : S -> S -> ListS
= BaseS ; --%
mkListS : S -> ListS -> ListS
= ConsS ; --%
} ; --%
--3 ListAdv, adverb lists
mkListAdv = overload { --%
mkListAdv : Adv -> Adv -> ListAdv
= BaseAdv ; --%
mkListAdv : Adv -> ListAdv -> ListAdv
= ConsAdv ; --%
} ; --%
--3 ListAP, adjectival phrase lists
mkListAP = overload { --%
mkListAP : AP -> AP -> ListAP
= BaseAP ; --%
mkListAP : AP -> ListAP -> ListAP
= ConsAP ; --%
} ; --%
--3 ListNP, noun phrase lists
mkListNP = overload { --%
mkListNP : NP -> NP -> ListNP
= BaseNP ; --%
mkListNP : NP -> ListNP -> ListNP
= ConsNP ; --%
} ; --%
--3 ListRS, relative clause lists
mkListRS = overload { --%
mkListRS : RS -> RS -> ListRS
= BaseRS ; --%
mkListRS : RS -> ListRS -> ListRS
= ConsRS ; --%
} ; --%
--.
the_Art : Art = DefArt ; -- the
a_Art : Art = IndefArt ; -- a
---- obsol
mkQuantSg : Quant -> QuantSg = SgQuant ;
mkQuantPl : Quant -> QuantPl = PlQuant ;
this_QuantSg : QuantSg = mkQuantSg this_Quant ;
that_QuantSg : QuantSg = mkQuantSg that_Quant ;
these_QuantPl : QuantPl = mkQuantPl this_Quant ;
those_QuantPl : QuantPl = mkQuantPl that_Quant ;
sgNum : Num = NumSg ;
plNum : Num = NumPl ;
------------ for backward compatibility
QuantSg : Type = Quant ** {isSg : {}} ;
QuantPl : Type = Quant ** {isPl : {}} ;
SgQuant : Quant -> QuantSg = \q -> q ** {isSg = <>} ;
PlQuant : Quant -> QuantPl = \q -> q ** {isPl = <>} ;
-- Pre-4 constants defined
DetSg : Quant -> Ord -> Det = \q -> DetQuantOrd q NumSg ;
DetPl : Quant -> Num -> Ord -> Det = DetQuantOrd ;
ComplV2 : V2 -> NP -> VP = \v,np -> ComplSlash (SlashV2a v) np ;
ComplV2A : V2A -> NP -> AP -> VP = \v,np,ap -> ComplSlash (SlashV2A v ap) np ;
ComplV3 : V3 -> NP -> NP -> VP = \v,o,d -> ComplSlash (Slash3V3 v o) d ;
that_NP : NP = DetNP (DetQuant that_Quant sgNum) ;
this_NP : NP = DetNP (DetQuant this_Quant sgNum) ;
those_NP : NP = DetNP (DetQuant that_Quant plNum) ;
these_NP : NP = DetNP (DetQuant this_Quant plNum) ;
that_Det : Det = (DetQuant that_Quant sgNum) ;
this_Det : Det = (DetQuant this_Quant sgNum) ;
those_Det : Det = (DetQuant that_Quant plNum) ;
these_Det : Det = (DetQuant this_Quant plNum) ;
-- new things
-- export needed, since not in Cat
ListAdv : Type = Grammar.ListAdv ;
ListAP : Type = Grammar.ListAP ;
ListNP : Type = Grammar.ListNP ;
ListS : Type = Grammar.ListS ;
-- bw to 4
Art : Type = Quant ;
the_Art : Art = DefArt ; -- the
a_Art : Art = IndefArt ; -- a
DetArtSg : Art -> CN -> NP = \a -> DetCN (DetQuant a sgNum) ;
DetArtPl : Art -> CN -> NP = \a -> DetCN (DetQuant a plNum) ;
DetArtOrd : Quant -> Num -> Ord -> Det = DetQuantOrd ;
DetArtCard : Art -> Card -> Det = \a,c -> DetQuant a (NumCard c) ;
TUseCl : Tense -> Ant -> Pol -> Cl -> S = \t,a -> UseCl (TTAnt t a) ;
TUseQCl : Tense -> Ant -> Pol -> QCl -> QS = \t,a -> UseQCl (TTAnt t a) ;
TUseRCl : Tense -> Ant -> Pol -> RCl -> RS = \t,a -> UseRCl (TTAnt t a) ;
-- numerals from strings
oper
str2ord : Str -> Ord = \s -> case Predef.lessInt (Predef.length s) 7 of {
Predef.PTrue => OrdNumeral (str2numeral s) ;
Predef.PFalse => OrdDigits (str2digits s)
} ;
str2card : Str -> Card = \s -> case Predef.lessInt (Predef.length s) 7 of {
Predef.PTrue => NumNumeral (str2numeral s) ;
Predef.PFalse => NumDigits (str2digits s)
} ;
str2numeral : Str -> Numeral = (\s -> case s of {
m@(? + _) + "000" => num (pot3 (s2s1000 m)) ;
m@(? + _) + "00" + n@? => num (pot3plus (s2s1000 m) (s2s1000 n)) ;
m@(? + _) + "0" + n@(? + ?) => num (pot3plus (s2s1000 m) (s2s1000 n)) ;
m@(? + _) + n@(? + ? + ?) => num (pot3plus (s2s1000 m) (s2s1000 n)) ;
_ => num (pot2as3 (s2s1000 s))
})
where {
s2d : Str -> Digit = \s -> case s of {
"2" => n2 ;
"3" => n3 ;
"4" => n4 ;
"5" => n5 ;
"6" => n6 ;
"7" => n7 ;
"8" => n8 ;
"9" => n9 ;
_ => Predef.error ("not a valid digit" ++ s)
} ;
s2s10 : Str -> Sub10 = \s -> case s of {
"1" => pot01 ;
#idigit => pot0 (s2d s) ;
_ => Predef.error ("not a valid digit" ++ s)
} ;
s2s100 : Str -> Sub100 = \s -> case s of {
"10" => pot110 ;
"11" => pot111 ;
"1" + d@#digit => pot1to19 (s2d d) ;
d@#idigit + "0" => pot1 (s2d d) ;
d@#idigit + n@? => pot1plus (s2d d) (s2s10 n) ;
_ => pot0as1 (s2s10 s)
} ;
s2s1000 : Str -> Sub1000 = \s -> case s of {
d@? + "00" => pot2 (s2s10 d) ;
d@? + "0" + n@? => pot2plus (s2s10 d) (s2s100 n) ;
d@? + n@(? + ?) => pot2plus (s2s10 d) (s2s100 n) ;
_ => pot1as2 (s2s100 s)
} ;
} ;
idigit : pattern Str = #("1" | "2" | "3" | "4" | "5" | "6" | "7" | "8" | "9") ;
digit : pattern Str = #("0" | #idigit) ;
--- it would be nice to have foldr on strings...
str2digits : Str -> Digits = (\s -> case s of {
d0@? => IDig (s2d d0) ;
d1@? + d0@? => IIDig (s2d d1) (IDig (s2d d0)) ;
d2@? + d1@? + d0@? => IIDig (s2d d2) (IIDig (s2d d1) (IDig (s2d d0))) ;
d3@? + d2@? + d1@? + d0@? =>
IIDig (s2d d3) (IIDig (s2d d2) (IIDig (s2d d1) (IDig (s2d d0)))) ;
d4@? + d3@? + d2@? + d1@? + d0@? =>
IIDig (s2d d4) (IIDig (s2d d3) (IIDig (s2d d2) (IIDig (s2d d1) (IDig (s2d d0))))) ;
d5@? + d4@? + d3@? + d2@? + d1@? + d0@? =>
IIDig (s2d d5) (IIDig (s2d d4) (IIDig (s2d d3) (IIDig (s2d d2)
(IIDig (s2d d1) (IDig (s2d d0)))))) ;
d6@? + d5@? + d4@? + d3@? + d2@? + d1@? + d0@? =>
IIDig (s2d d6) (IIDig (s2d d5) (IIDig (s2d d4) (IIDig (s2d d3)
(IIDig (s2d d2) (IIDig (s2d d1) (IDig (s2d d0))))))) ;
d7@? + d6@? + d5@? + d4@? + d3@? + d2@? + d1@? + d0@? =>
IIDig (s2d d7) (IIDig (s2d d6) (IIDig (s2d d5) (IIDig (s2d d4) (IIDig (s2d d3)
(IIDig (s2d d2) (IIDig (s2d d1) (IDig (s2d d0)))))))) ;
_ => Predef.error ("cannot deal with so many digits:" ++ s)
}) where {
s2d : Str -> Dig = \s -> case s of {
"0" => D_0 ;
"1" => D_1 ;
"2" => D_2 ;
"3" => D_3 ;
"4" => D_4 ;
"5" => D_5 ;
"6" => D_6 ;
"7" => D_7 ;
"8" => D_8 ;
"9" => D_9 ;
_ => Predef.error ("not a valid digit" ++ s)
} ;
} ;
}