RomClitics

We love clitics.

General remarks

The status of clitics (syntactic word vs. bound morpheme) is problematic. Most Romance clitics have the properties of bound affixes but for primarily orthographic considerations, the Delph-In Romance grammars realize clitics in syntax (exceptions: post-verbal clitics in Spanish and French).

Issues to be addressed:

• determining pre- vs. post-verbal position
• allomorphy and ordering constraints in clitic clusters
• clitic climbing (i.e., raising of complement clitics, realization on a higher verb)
• clitic doubling
• the syntactic deficiency of clitics (restrictions on coordination, modification, distribution)
• subcategorization issues (complements that must be / must not be cliticized)

Spanish

Portuguese

French

• preverbal clitics appear in a fixed order, with 5 slots

  { m(e) | t(e) | s(e) | nous | vous }	{ l(a) | l(e) | les }	{ lui | leur }	{ y }	{ en }

  • maximum cluster length = 4 (at most one clitic per slot, and slots 1 and 3 cannot both be filled)
  • can be extended to the left by a subject clitic and the negative scope marker

    { j(e) | tu | il/elle | nous | vous | ils/elles }	n(e)

• post-verbal clitics

  • non-negated imperatives

    { la | l(e) | les }	{ lui | leur | m(oi) | t(oi) | nous | vous }	{ y }	{ en }

  • post-verbal clitics (pronominal subject inversion) [not implemented]

  • { je | tu | (t-)il/(t-)elle | nous | vous | (t-)ils/(t-)elles }

• Preverbal clitics are separated from verb and each other by a space or apostrophe; Postverbal clitics are separated from verb and each other by a hyphen or apostrophe.

• obligatory allomorphic effects throughout cluster and between clitic and verb:

  • elision (reflected in orthography): deletion of e, a, oi

  • liaison: [t]-insertion (sometimes reflected in orthography), [z]-insertion (non-standard)

• Clitic climbing is found systematically in compound past tenses (auxiliaries être and avoir) and under some conditions with causative faire and laisser.

• Clitic doubling in standard French is limited to the following cases [none implemented]:

  • obligatory les in combination with direct object tous (all):

    • Je vais [tous (*les) massacrer, (*les) massacrer tous] (I'm going to slaughter them all)

  • en in combination with direct object quantifiers (especially numerals)

    • Je vais (*en) massacrer deux (I will slaughter two (of them))

  • NP subject in combination with clitic inversion:

    • A quelle heure Paul va-t-il arriver ? (At what time will Paul arrive?)

• widespread left and right dislocation with clitic realization [not implemented]

Implementation

Miller and Sag (1997) established that French pronominal clitics are bound affixes (pre-verbal subject clitics are possibly less bound). This approach is not uniformly adopted in the LKB grammar, in order to reflect conventional French orthography. Pre-verbal clitics are analyzed as lexemes and inserted in syntax by "compounding" rules (which must apply lower than ordinary syntactic rules). This is made possible by the following hierarchy under sign:

gram-cat := sign.
lexical := sign.
  word := lexical & gram-cat.
  lexeme := lexical.
phrasal := gram-cat.
  phrase := phrasal.
compound := phrasal & lexical.

Post-verbal clitics are realized morphologically using inflectional lexical rules.

Although vowel elision is correctly analyzed, the apostrophe is currently thrown out (not generated). The hyphen is treated as a normal character. Neither one of these approaches is adequate.

Clitic climbing is analyzed using Abeillé, Godard, and Sag's argument raising analysis: temporal and causative auxiliaries systematically inherit all complements of the participle/infinitive (which is not allowed to project a VP). Currently untreated: downstairs cliticization with causative faire and "clitic trapping".

Currently no analysis for any cases of clitic doubling.

Modern Greek