Jorge Vega Vilanova
5. Old Catalan PPA
In order to better understand how PPA gets lost from Old to Modern Catalan, I gathered Old Catalan data from the 11th to the 16th centuries.4 For each century, the first 100 pages of two or three prose texts were analyzed (excepted for the 11th and 12th centuries: the written records are too scarce). All sentences with past participles in verbal function, e.g., in compound verb tenses, were excerpted, excluding passives, which have always auxiliary BE and obligatory agreement, and masculine singular objects, indistinguishable from the default form of agreement. 1,091 sentences were found, distributed along the centuries as 3 Franco (1994) connects PPA to the categorial status of clitics: Old Catalan clitics, being XP, can enter into a Spec-Head relation with the participle, triggering PPA. When they reduce to the category X°, they act as agreement markers and cannot trigger PPA anymore. The grammaticalization path of clitics has been assumed to be an important factor in the explanation of the emergence of CLD as well (Fontana 1993, but also Vega Vilanova et al., forthcoming).
4 The following texts were used: Llibre de meravelles (1288) by Ramon Llull; Crònica (1299) by Bernat Desclot; Contes i faules (1392) by Francesc Eiximenis; Lo somni (1399) by Bernat Metge; La fi del comte d’Urgell (1433), anonym; Curial e Güelfa (1468), anonym; Col·loquis de la insigne Ciutat de Tortosa (1557) by Cristòfor Despuig; Epistolaris d’Hipòlita Roís de Liori i d’Estefania de Requesens (16th century).
SPECIFICITY AND PAST PARTICIPLE AGREEMENT IN CATALAN: A DIACHRONIC APPROACH
shown in Table 1. As can be seen, the rates of PPA until the 15th century are very high. The 16th century seems to be a point of inflection: almost half of the tokens lack agreement.
Auxiliary HAVE
[+Agreement] Auxiliary HAVE [−Agreement]
11th/12th 12 2
13th 294 16 (~5.1%)
14th 297 58 (~16.3%)
15th 196 18 (~8.4%)
16th 107 91 (~46.0%)
Table 1. General rates of PPA in Old Catalan 5.1 Specificity and PPA
Since PPA seems to be related to the same functional position where aspectual information is encoded, and aspect and specificity are themselves interconnected, I first examined how specificity affects the realization of PPA in Old Catalan (see footnote 2 for the problematic of defining specificity in diachronic data).
Lack of agreement is found in Old Catalan in all kinds of contexts since the very first documents. However, it is especially frequent in the following cases:
• With [−Def] objects:
(11) car oït he moltes coses que . . .
since hear-pp-def have-1sg many thing-f.pl rel
“Since I heard many things that . . .” (14th century)
• In relative clauses, with objects in form of operators, which cannot be considered sensu stricto definite DPs:
(12) ladronices que havia fait
theft-m.pl rel have-impf.3sg do-pp.def
“robberies that he had done” (12th century)
• With inherent accusatives (length and time measures);
• In unaccusative verbs when they are used with the auxiliary HAVE:
(13) Han seguit guerres injustes
have-3pl follow-pp.def war-f.pl unfair-f.pl
“Unfair wars happened afterwards” (14th century)
JORGE VEGA VILANOVA
The tendency to lack agreement increases in the 16th century. Table 2 sums up the results and shows that until the 15th century only around 15% of the indefinite objects lack agreement. In the 16th century, it is almost 70%. Table 3 shows that in all periods less than half of the non-agreeing participles have definite objects, although definite objects are much more frequent in the corpus.
Total [−Def] +PPA −PPA
13th 72 62 10 (~13.9%)
14th 132 107 25 (~18.9%)
15th 44 39 5 (~11.3%)
16th 26 8 18 (~69.2%)
Table 2. Rates of PPA with indefinite objects in Old Catalan Total [−Agr] +Def −Def Relative
clauses Unacc. with HAVE
13th 16 4 (~25.0%) 9 1 2
14th 58 21 (~36.2%) 26 7 4
15th 18 9 (~50.0%) 5 3 1
16th 89 25 (~28.1%) 45 15 6
Table 3. Types of DOs with non-agreeing past participles in Old Catalan
In a nutshell, indefinite DPs can be assumed to be the first context where PPA disappears.
Definite DPs, especially when placed to the left of the verb, still trigger agreement.
5.2 DOM and PPA
Since the preposition-like marker of DOM objects already assigns case to the DP (e.g., Jaeggli 1986), it is expected that DOM blocks the agreement relation between participle and DO. DOM is not yet attested in the oldest Catalan texts. However, when it first appears, DOM and PPA do not usually appear simultaneously in the same clause (14), although there are occasional exceptions (15).
(14) (a) he aja perdonat a tots aquells qui . . .
and have-subj.1sg forgive-pp.def dom all-m.pl that-m.pl rel
“and I had forgiven all those that . . .” (14th century)
(b) los geògrafos que han descrit a Espanya
the geographer-m.pl rel have-3pl describe-pp.def dom Spain-f.sg
“The geographers that described Spain . . .” (16th century)
SPECIFICITY AND PAST PARTICIPLE AGREEMENT IN CATALAN: A DIACHRONIC APPROACH
(15) que havie spolsada una vegada that have-past.3sg expel-pp.f.sg once
a la dita na Grahullana
dom the mentioned-f.sg na Grahullana-f.sg
“that he had once expelled na Grahullana already mentioned”
(Farreny Sistac 2004, 344; 16th century)
In fact, the same pattern is found in Modern Catalan. PPA with 3rd person clitic objects is commonly accepted by all speakers, but when the clitic is doubling a full object DP with DOM (16a), PPA is rejected. In this case, there are two possible repair strategies:
either the full DP must be dislocated (16b) or the participle must show default agreement (16c). In sum, both the synchronic and diachronic data suggest that DOM and PPA stand in a complementary distribution.
(16) (a) *Les ha vistes a elles.
cl.acc.3.f.pl have-3sg see-pp.f.pl dom they-f.pl
“He/she has seen them.”
(b) Les ha vistes, a elles.
cl.acc.3.f.pl have-3sg see-pp.f.pl dom they-f.pl
(c) Les ha vist a elles.
cl.acc.3.f.pl have-3sg see-pp.def dom they-f.pl 5.3 CLD and PPA
Evidence for a correlation between CLD and PPA is elusive in Old Catalan. Only a few instances of accusative CLD are attested until the 15th century, but these do not show up in the relevant contexts. However, clauses where PPA co-appears with CLD are not attested at all. This indirectly confirms the idea that languages allow either CLD or PPA.
5.4 Interim Summary
In this section, I have shown that the loss of PPA in Catalan does not occur randomly, but rather follows a clear path. The 16th century is a turning point: PPA decreases, especially if the DO is indefinite. This leads us to look at other related phenomena like CLD and DOM, which emerge during the exact same period. The presented data suggest that the latter are gradually substituting PPA. All these phenomena show alternation during the 16th century but do not coexist within the same clause. As for the trigger of this change, specificity has been shown to have an effect on PPA and can plausibly be identified as the main trigger for PPA. Hence, the following five stages can be claimed to depict the language change with respect to PPA in Catalan:
JORGE VEGA VILANOVA
• Obligatory PPA (12th to 15th centuries)
• PPA connected to specificity (16th century)
• PPA depending on syntactic placement of the DO (Modern Italian and French)
• Optional PPA (spoken French, Modern Catalan)
• Loss of PPA altogether (Spanish, Portuguese, and Romanian).