LGLex-Lefff syntactic lexicon version 3.3 - 2011/03/31
http://infolingu.univ-mlv.fr/
License: LGPL-LR

The LGLex-Lefff lexicon is a syntatic lexicon of French verbs and nouns 
playing the predicative role converted in the Lefff format (Tolone & 
Sagot, 2011; Tolone, 2011) from the LGLex lexicon.
For more details about the Lefff lexicon and the framework it is based on, 
named Alexina, see (Sagot, 2010).

_____________________________________________
Description of the intensional Lefff format :

Each entry in the intensional lexicon corresponds to a unique meaning of 
the corresponding lemma. It contains the following information:
- an entry identifier generated by concatenating its grammatical category, 
its class (or table) it comes from and the index of the entry in the table;
- a morphological class, which defines the patterns that build its inflected 
forms, using inflection classes from the Lefff;
- a category (or part-of-speech);
- the initial sub-categorization frame;
- additional syntactic information represented by macros;
- the list of possible redistributions;
- an example of sentence with the entry;
- a comment after the "#" symbol.

For example, the intensional entry in the LGLex-Lefff lexicon for the French 
lemma "clouer___V_36SL_28" (to nail) is as follows:

clouer___V_36SL_28	v-er:std	100;Lemma;v;<Suj:cln|sn,Obj:sn,Loc:(avec-sn|et-sn|à-sn|sur-sn)>;cat=v;%actif,%passif,%ppp_employé_comme_adj	Ex.: Max a cloué cette planche(avec+contre+sur)celle-là # BASE CONSTR = N0 V N1 Loc N2 (N0 V N1 et N2 ; N0 V N1 Prép N2 ; N0 V N1hum Loc N2abs ; N0 V N1) [[passif par]] <> ; orig base constr = N0 V N1 Loc N2

It describes a transitive entry with the following information:
- its identifier is V_36SL_28, which correspond to the 28th entry in the 
verb class 36SL;
- its morphological class is "v-er:std", the class of standard first-conjugation 
verbs (ending "-er");
- its semantic predicate can be represented by the "Lemma" as is, i.e., "clouer";
- its category is "verb" (v);
- it has three arguments canonically realized by the syntactic functions Suj 
(subject), Obj (direct object) and Loc (locative argument); each syntactic 
function is associated with a list of possible realizations, but the Loc is 
optional as shown by the brackets;
- it allows for three different redistributions: %active, %passive, and 
%ppp_employé_comme_adj;
- its example is Max a cloué cette planche(avec+contre+sur)celle-là (Max 
has nailed this board(with+by+on)that one).

____________________________________________
Description of the extensional Lefff format:

The compilation process builds one extensional entry for each inflected form 
and each compatible redistribution, by inflecting the lemma according to the 
definition of its morphological class and by applying the formalized 
definitions of these redistributions.
For example, the only inflected forms of "clouer" that are compatible with the 
passive redistribution are the past participle forms. The extensional passive 
entry for "cloués" (nailed) is the following (Kmp is the morphological tag for 
past participle masculine plural forms):

cloués	100	v	[pred="clouer___V_36SL_28__1<Suj:sn,Loc:(avec-sn|et-sn|sur-sn|à-sn),Obl2:(par-sn)>",@passive,@pers,cat=v,@Kmp]	clouer___V_36SL_28__1	PastParticiple	Kmp	%passif

The original direct object (Obj) has been transformed into the passive 
Subject, and an optional Agent (Obl2) realized by a noun phrase preceded 
by a preposition ("par-sn") was added.

_______________________________________________________________________________
Syntactic functions, realizations, redistributions and macros in the LGLex-Lefff lexicon:

For verbs, the Lefff format uses the following syntactic functions:
- Suj for subject: cliticization with the nominative clitic;
- Obj for direct object: cliticization with the accusative clitic, commutable 
with "ceci/cela" (this/that), impacted by passivization when it is possible;
- Objà for indirect object canonically introduced by the preposition "à": 
commutable with "à+non-clitic pronoun" but not with "ici" (here) or "là(-bas)" 
(there), may be cliticizable into the dative clitic or "y";
- Objde for indirect object introduced by the preposition "de": cliticization 
with "en", not commutable with "d'ici" (from here) or "de là" (from there);
- Loc for locative argument: commutable with "ici" (here) or "là(-bas)" (there), 
cliticizable with "y": e.g. "à Paris" in Pierre va à Paris (Peter goes to Paris);
- Dloc for delocative argument: commutable with "d'ici" (from here) or "de là" 
(from there), cliticizable with "en": e.g., "de Paris" in Pierre vient de Paris 
(Peter comes from Paris);
- Att for (subject, object or "à"-object) attribute and pseudo-object: e.g., 
"3 euros" in J'ai acheté ceci 3 euros (I bought this 3 euros);
- Obl and Obl2 for other (non-cliticizable) arguments; Obl2 is used for verbs 
with two oblique arguments, such as "plaider auprès de quelqu'un en faveur de 
quelqu'un d'autre" (To plead in front of somebody for somebody else).

For predicative nouns, that can be headed by a support verb, the same set of 
functions are used.

For verbs and nouns, possible realizations are threefold:
- clitic pronouns: cln for nominative clitic: e.g., "il" in Il donne ce livre 
à Marie (He gives this book to Mary), cla for accusative clitic: e.g., "le" in 
Il le donne à Marie (He gives it to Mary), cld for dative clitic: e.g., "lui" 
in Il lui donne ce livre (He gives her this book), y: e.g., Max y va (Max goes 
there), en: e.g., Max en mange (Max eats it);
- direct phrases: sn for noun phrase: e.g., "La belle dame" in La belle dame 
arrive (The beautiful lady arrives), sa  for adjectival phrase: e.g., "verte" 
in La robe est verte (The dress is green), sinf for infinitive clause: e.g., 
"dire aurevoir" in Pierre est parti dire aurevoir (Peter left to say goodbye), 
scompl for completive clause: e.g., "que Marie est belle" in Pierre dit que 
Marie est belle (Peter says that Marie is beautiful), qcompl for interrogative 
clause: e.g., "combien il gagne" in Pierre dit combien il gagne (Peter said 
how much he earns);
- prepositional phrases: a direct phrase introduced by a preposition (e.g., 
à-sn, de-scompl, pour-sinf).

For verbs, the inventory of possible redistributions is the following:
- %actif, a dummy redistribution that has almost no effect on the initial 
sub-categorization information;
- %passif for the standard passive in "par": e.g., Jean assistait Max depuis 
des années-> Max était assisté par Jean depuis des années (John has assisted 
Max for years -> Max has been assisted by John for years);
- %passif_de for the passive in "de": e.g., Marie aime Pierre -> Pierre est 
aimé de Marie (Mary loves Peter -> Peter is loved by Mary);
- %actif_impersonnel for active impersonal constructions with inverted subject, 
if any: e.g., Un accident est arrivé à Jean -> Il est arrivé un accident à Jean
(An accident happened to John);
- %passif_impersonnel for passive impersonal constructions with inverted 
subject, if any: e.g., Cette nouvelle information clarifie pourquoi Max est 
faché -> Il est clarifié par cette nouvelle information pourquoi Max est faché 
(This new information clarifies why Max is angry -> It is clarified by this 
new information why Max is angry);
- %ppp_employé_comme_adj for past participles used as adjectives: e.g., Marie 
s'assoie -> Marie est assise (Mary is sitting -> Mary is seated).

Predicative nouns only use the %default construction that builds a final 
sub-categorization frame which is identical to the initial one.

For verbs and nouns, macros represents additional syntactic information such as:
- control: for instance, @CtrlSujObj indicates that if it is realized as an 
infinitive clause, the object is controlled by the subject: e.g., "chercher 
du pain" is controlled by "Pierre" in Pierre va chercher du pain (Peter goes 
to get bread);
- mood of the complementizer phrase: @SCompInd indicates that if the subject 
is realized as a completive clause, its mood is indicative: e.g., "qu'il fait 
beau" in Pierre dit qu'il fait beau (Peter says that the weather is good); on 
the contrary, @CompSubj indicates that if the direct object is realized as a 
completive clause, its mood is subjonctive: e.g., "qu'il fasse beau" in Pierre 
veut qu'il fasse beau (Peter wants that the weather be good); the following 
abreviations are used: S for subject, nothing for direct object, A for indirect 
object introduced by the preposition "à", De for indirect object introduced by 
the preposition "de"; by default, the two modes are possible;
- human or non human noun phrase: @ObjàNhum indicates that the indirect object 
introduced by the preposition "à" can denote a person, or an animal 
linguistically assimilated to a person when is realized by a noun phrase: e.g., 
Vercingetorix s’est rendu à l'ennemi (Vercingetorix surrendered to the enemy); 
on the contrary, @ObjàN-hum indicates that the indirect object introduced by 
the preposition "à" can be non human: e.g., Jean s’est rendu à mon opinion 
(John finally accepted my opinion); in this case, @ObjàNhum and @ObjàN-hum 
are not in the same entry because the sense of "se rendre" (to accept/to 
surrender) is different; in other cases, for instance, Max va tomber (Max 
goes down) and Le verre va tomber (The glass will fall), @SujNhum and 
@SujN-hum indicate that the subject can denote a human or a non human for 
the same entry; if no indication is given for an argument, we can consider 
both are always possible.

___________
References:

Tolone, Elsa & Sagot, Benoît (2011). Using Lexicon-Grammar tables for French 
verbs in a large-coverage parser. Edited by Z. Vetulani in Human Language 
Technology, Forth Language and Technology Conference, LTC 2009, Poznán, 
Poland, November 2009, Revised Selected Papers, Lecture Notes in Artificial 
Intelligence (LNAI). Springer Verlag. To appear.

Tolone, Elsa (2011). Analyse syntaxique à l'aide des tables du Lexique-Grammaire 
du français. Thèse de doctorat, LIGM, Université Paris-Est. 326 pp.

Sagot, Benoît (2010). The Lefff, a freely available and large-coverage 
morphological and syntactic lexicon for French. Proceedings of the 7th 
Language Resource and Evaluation Conference (LREC'10), 8 pp. La Valette, Malta.
