French (Indo-European, Romance)

1. Verbs: synchronic description

A number of French verbs show quirky gaps in their paradigms. Boyé (2000) correlates the gaps with the distribution of stem alternants in the paradigm. In his analysis, the verb paradigm has 16 stems, arranged in the morphological inheritence hierarchy given below:

finite forms
non-finite forms
PRS SBJV
PRS		 IMPRF SBJV
IMPRF		 PRET FUT COND IMPV
1SG pr1 su6 imp
ps4
fut
INF inf
2SG pr3 ju2 GER ger
3SG PST PTCP pps
1PL pr4 su4 ju4
2PL pr5 ju5
3PL pr6 su6

The black lines indicate direct inheritence, e.g. by default the Pr4 stem is identical to the Imp stem, and the Pr5 stem is identical to the Pr4 stem, and so on. The red arrows indicate non-hierarchical inheritence: where a stem is lexically specified, inheritence exceptionally travels along these paths. Defectiveness is construed as the lexical specification of a gap, and so is transmitted along these paths (as well as via the default hierarchy). Boyé isolates 8 verbs (from among various which have been cited as defective in normative sources) as being synchronically valid, relatively common examples of morphogical defectiveness.

Traire 'milk' is missing its subjunctive imperfect and preterite, which can be modelled as Ps4=Ø (absoudre 'absolve' follows the same pattern):

finite forms
non-finite forms
PRS SBJV
PRS		 IMPRF SBJV
IMPRF		 PRET FUT COND IMPV
1SG trais traie trayais ------ ------- trairai trairais INF traire
2SG trais traies trayais ------ ------- trairas trairais trais GER trayant
3SG trait traie trayait ------ ------- traira trairai PST PTCP trait
1PL trayons trayions trayions ------ ------- trairons trairions trayons
2PL trayez trayiez trayiez ------ ------- trairez trairiez trayez
3PL traient traient trayaient ------ ------- trairont trairaient

Paître 'graze' has the same gap as traire, plus it is missing its past participle, modelled as Pps=Ø. Note that the absence of the past participle has a profound effect, as the periphrastic TAMs formed from it (which account for about half of the total verb paradigm) are likewise missing.1

finite forms
non-finite forms
PRS SBJV
PRS		 IMPRF SBJV
IMPRF		 PRET FUT COND IMPV
1SG pais paisse paissais ------ ------- paîtrai paîtrais INF paître
2SG pais paisses paissais ------ ------- paîtras paîtrais pais GER paissant
3SG paît paisse paissait ------ ------- paîtra paîtrai PST PTCP ------
1PL paissons paissions paissions ------ ------- paîtrons paîtrions ????
2PL paissez paissiez paissiez ------ ------- paîtrez paîtriez paissez
3PL paissent paissent paissaient ------ ------- paîtront paîtraient

Gésir 'lie' (in restricted senses, applied to the ailing or deceased) is missing the bulk of its paradigm, which can be modelled as Pps=Ø, Fut=Ø and Ju2=Ø.

finite forms
non-finite forms
PRS SBJV
PRS		 IMPRF SBJV
IMPRF		 PRET FUT COND IMPV
1SG gis ------- gisais ------- ------ ------- ------- INF gésir
2SG gis ------- gisais ------- ------ ------- ------- ------- GER gisant
3SG gît ------- gisait ------- ------ ------- ------- PST PTCP ------
1PL gisons ------- gisions ------- ------ ------- ------- -------
2PL gisez ------- gisiez ------- ------ ------- ------- -------
3PL gisent ------- gisaient ------- ------ ------- -------

The gaps in pouvoir 'be able' result from Ju2=Ø. In contrast to gésir, the gap does not spread to the subjuntive (as predicted by the inheritence hierarchy), as it is blocked by lexical specification of the stem puiss- at Su4.2

finite forms
non-finite forms
PRS SBJV
PRS		 IMPRF SBJV
IMPRF		 PRET FUT COND IMPV
1SG peux puisse pouvais pusse pus pourrais pourrais INF pouvoir
2SG peux puisses pouvais pusses pus pourras pourrais ------- GER pouvant
3SG peut puisse pouvait pût put pourra pourrait PST PTCP pu
1PL pouvons puissions pouvions pussions pûmes pourrons pourrions -------
2PL pouvez puissiez pouviez pussiez pûtes pourrez pourriez -------
3PL peuvent puissent pouvaient pussent purent pourront pourraient

The interaction of gaps and blocking is even more strikingly illustrated by frire 'fry', braire 'bray' and clore 'close'. These all involve Imp=Ø. Since this is the mother node of the entire hierachy, inheritence of this specification would mean that there were no forms whatsoever. However, Boyé propose that the various stems that make up the paradigm are specified lower down in the hierarchy:

finite forms
non-finite forms
PRS SBJV
PRS		 IMPRF SBJV
IMPRF		 PRET FUT COND IMPV
1SG fris ------- ------- ------- ------- frirai frirais INF frire
2SG fris ------- ------- ------- ------- friras frirais fris GER -------
3SG frit ------- ------- ------- ------- frira frirait PST PTCP frit
1PL ------- ------- ------- ------- ------- frirons fririons -------
2PL ------- ------- ------- ------- ------- frirez fririez -------
3PL ------- ------- ------- ------- ------- friront friraient
(Imp=Ø; stem specified at Pr3, Inf and Pps)

finite forms
non-finite forms
PRS SBJV
PRS		 IMPRF SBJV
IMPRF		 PRET FUT COND IMPV
1SG brais braie ------- ------- ------- brairai brairais INF braire
2SG brais braies ------- ------- ------- brairas brairais brais GER -------
3SG brait braie ------- ------- ------- braira brairait PST PTCP brait
1PL ------- ------- ------- ------- ------- brairons brairions -------
2PL ------- ------- ------- ------- ------- brairez brairiez -------
3PL braient braient ------- ------- ------- brairont brairaient
(Imp=Ø; stem specified at Pr6, Inf and Pps)

finite forms
non-finite forms
PRS SBJV
PRS		 IMPRF SBJV
IMPRF		 PRET FUT COND IMPV
1SG clos close ------- ------- ------- clorai clorais INF clore
2SG clos closes ------- ------- ------- cloras clorais clos GER closant
3SG clôt close ------- ------- ------- clora clorait PST PTCP clos
1PL ------- closions ------- ------- ------- clorons clorions -------
2PL ------- closiez ------- ------- ------- clorez cloriez -------
3PL closent closent ------- ------- ------- cloront cloraient
(Imp=Ø; stem specified at Ger, Su4 and Pr6)

2. Verbs: diachronic account

Gilliéron (1919) offers an account of the development of defectiveness in clore 'close', frire 'fry' and choir 'fall'.

Clore originally was a completely obsolete verb, driven out of the language by phonological competition with clouer 'nail', with which it was homophonous for much of the paradigm,3 and with which there was also some semantic overlap ('nail (shut)' ≈ 'close'), and replaced by fermer 'close'. It was artificially reintroduced by grammarians of the French Academy, meant to replace fermer in some contexts.

The development of frire also involves competition, but of a different type. Gilliéron argues that there was phonological competition with forms of frir, a syncopated variant of ferir 'strike', which iteslf survives only as the vestigial férir (infinitive and past participle forms only). Here there was no question of semantic overlap, but Gilliéron suggests there was something more subtle going on, involving transitivity. Frire was both transitive and intransitive (with the patient as subject, e.g. 'the fish is frying'), while f(e)rir was transitive only. In those parts of the finite paradigm where the two verbs were homophonous (present indicative singular), frire lost its transtive uses, so that the two verbs were now in complementary distribution with respect to transitivity. Elsewhere in the paradigm, the stem of frire was fris-, and thus not homophonous with those of f(e)rir. These forms developed the specialized sense of 'fry hair (with a curling iron)' > 'curl', which ultimately split off as a separate lexeme, filling in the empty portions of its paradigm. The transitive of frire was now formed as a regular periphrastic causative, faire frire 'make fry'. All that was left of original frire were a few fragments, largely restricted to 3rd person subjects.The paradigm as seen above in section 1a, which is essentially that sanctioned by the Academy, is an artificial extension of these fragments.

The story Gilliéron gives for choir is similar to that he gives for clore: the word was obsolete, and the subject of a revival attempt by the Academy. The story of its death is somewhat different from that given for clore, but more importantly, its revival does not seem to have been successful. Thus Boyé treats it as a 'vestigial' verb and not a defective one. That is, the extent forms (infinitive and past participle) should not be regarded as components of an inflectional paradigm, but simply as unanalyzed words.

Thus, Gilliéron's take on defectiveness is sceptical, treating it as the by-product of ill-advised attempts at language engineering. He writes (p. 33) 'Par l’usage les enfants apprennent l’irrégularité des verbes, la défectivité s’apprend à l’école: c’est là qu’on apprend ce qui n’existe pas.' In the case of clore and choir, the actual shape of the proposed paradigms is said to have been limited by what their creators thought they could get away with in terms of plausability (thus, forms more or less readily derivable from the infinitive are present).4 Still, the account he gives of frire do involve some natural developments. While the putative paradigm, including a 1st and 2nd person singular present, may be a fiction, his account does derive a fragmentary paradigm of frire.

3. Adjectives

The following is based on Bonami and Boyé (2006). French adjectives may have distinct forms for masculine and feminine. When used prenominally, the form of the masculine is sensitive to the phonology of the following noun. If the following noun begins in a consonant, the form is the same as when the adjective is used phrase-finally. But when the following noun begins in a vowel, a distinct form (the 'liaison form') is used. Typically, the liaison form is identical to the masculine or to the feminine. A few adjectives have a unique form, which can be derived from the feminine singular through of a final s or devoicing of a final d. However, some adjectives appear to lack a liason form.

liaison form = M.SG liaison form = F.SG liaison form is unique liason form is absent
M.SG, preconsonantal court (/kur/) rendez-vous
'brief meeting' 		petit (/pəti/) bateau
'small boat' 		grand (/grã/) bateau
'small boat' 		chaud (/ʃo/) rendez-vous
'lively (lit. 'hot') meeting'
F.SG courte (/kurt/) entrevue
'brief discussion' 		petite (/pətit/) auto
'small car'		 grande (/grãd/) auto
'small car'		 chaude (/ʃod/) entrevue
'lively discussion'
M.SG, prevocalic
(liaison form) 		court (/kur/) entretien
'brief discussion' 		petit (/pətit/) avion
'small plane'		 grand (/grãt/) avion
'small plane'		 *chaud entretien
'lively discussion'

Post-nominal position is still acceptable (entretien chaud), and is indeed the preferred position. Bonami and Boyé (2006) treat this as a lexically-specified gap: just as some adjectives may have lexically-specified liaison forms, others may have gaps in this position. It is unclear what the actual inventory of such adjectives is, as judgements vary, but it is something on the order of a dozen, including (at least according to the intuitions of the authors) blanc 'white', blond 'blonde', brun 'brown', chaud 'hot', franc 'frank', froid 'cold', sot 'stupid'.

Notes

1 These being the perfect, pluperfect, anterior past, future past, past subjunctive, future subjunctive and past conditional.

2 Where there is a conflict between normal default inheritence and non-hierarchical inheritence, the former prevails.

2 Evidence for the competition between clore and clouer also comes from dialects in which clouer has become deformed (e.g. as clouter or clouler) to differentiate it from clore (which was subsequently lost).

4 As he colorfully puts it, describing the proposed paradigm of choir (p. 51):

L'Académie, selon le pli qu'elle a pris, qui appert dans clore et frire, et qui consiste à conjuger tant qu'elle ne s'achoppe pas a quelque monstrueux produit, qui mout, tant qu'une pierre n'arrête pas subitement l'engrenage du moulin, dit:

je chois, tu chois, il choit, -------, -------, -------, participe passé chu

References

Bonami, Olivier and Gilles Boyé. 2006. Construire le paradigme d’un adjectif. Recherches linguistiques de Vincennes, 34, L' adjectif. [Available online at http://rlv.revues.org/document1370.html.]

Boyé, Gilles. 2000. Problèmes de morpho-phonologie verbale en français, en espagnol et en italien. PhD. thesis, University of Paris VII.

Gilliéron, Jules. 1919. La faillite de l'étymologie phonétique: Étude sur la défectivité des verbes. Neuveville: Beerstecher.