The trajectory of (in)definiteness, number, and gender in Old French 1. Context. In the diachrony of French, we observe a change in the determiner (D) system such that in earlier stages of French, D was not obligatory, while in Modern French D is obligatory. Based on parsed corpus data from two OF texts — Brendan (circa 1106-1121) and Marie de France (MdeF) (circa 1154-1189) — a quantitative analysis of the distribution of D confirms the six factors listed below. While factors (i-v) have been observed in previous qualitative research [1-6], factor (vi) has not been previously noticed. (i) nominal predicates favour determinerless nouns (i.e. bare Ns); (ii) object position favours bare Ns; (iii) non-count Ns (abstract and mass) favours bare Ns; (iv) indefiniteness contexts favour bare Ns; (v) plural marking favours bare Ns; (vi) feminine gender favours bare Ns, but only in the earlier text (Brendan). 2. Findings. A quantitative analysis confirms the activity of factors (i-vi), and also permits their relative weight to be measured. The latter allows one to analze the interplay between these factors, and to track the stable and changing properties of the grammar. Stable factors: Across the two texts, the relative weight of the factors relating to argument (nominal predicate vs. nominal argument) and grammatical function (object vs. subject) are stable, i.e. they do not change over time. We take this to indicate that these are stable properties of the grammar, and moreover that they reflect properties of Universal Grammar. Changing factors: We observe an increase in range of the factor group semantic class, which is partly due to a surprising increase of bare non-count nouns (mass and abstract) in MdeF. As for definiteness and number, because there are no definiteness nor number contrasts with abstract and mass Ns, the crucial class to track is count Ns. Within the count N domain, definite and indefinite count nouns pattern differently. While the change appears nearly completed with definites, it is in progress with indefinites: (a) (b) (c)

there is a steady decrease of bare indefinite count nouns from Brendan to MdeF. number is significant in both texts with indefinites, but significant with definites only in MdeF gender is a significant factor only with Brendan’s indefinites

3. Analysis. The change in significance and relative weight of semantic class, number, and gender reflects a reorganization triggered by the reanalysis of un from a cardinality numeral (‘one’) to an indefiniteness marker. In Brendan, the use of un as an indefinite marker yields the paradigm in (1), which is organized around a case contrast (Kihm 2012), with -s coding [SUBJECT] case. This has a knock-on effect: objects are more likely to be bare Ns; feminine singular Ns are more likely to be bare Ns; plural Ns are more likely to be bare Ns. Moreover, as shown in (1) and (2), in Brendan, the featural organization of the (in)definite paradigms differ. While the indefinite paradigm has a [NUM>GEN>CASE] ranking, the definite paradigm has a [CASE>NUM>GEN] ranking. Both paradigms are re-organized in MdeF. Specifically, the introduction of the plural indefinite determiner des, restricted to object position, forces the restructuring in (3), where a singular/plural contrast is coded in each cell of the indefinite paradigm. The decrease of bare indefinites in MdeF reflects the fact that the re-organized paradigm restricts bare indefinite plurals to fewer contexts. At later stages, the paradigm is regularized by spreading indefinite plural des to subject position (triggered by the loss of subject case-marking -s), and subsequent spread of des to feminine plural. As for the increase of bare non-count nouns in MdeF, this reflects the introduction of des (with count Ns) which indirectly contributes to an increase in the proportion noun-count bare Ns.

(1)

BRENDAN INDEFINITE PARADIGM: [NUMBER > GENDER > CASE] SINGULAR

PLURAL

MASCULINE

(2)

(3)

FEMININE

SUBJECT

OBJECT

un-s

un

un-e

∅

BRENDAN DEFINITE PARADIGM: [CASE>NUMBER>GENDER] SUBJECT u KASE u NUM SINGULAR

PLURAL

MASCULINE

MASCULINE

FEMININE

l-i

l-e

l-a

l-e-s

MARIE-DE-FRANCE (IN)DEFINITE PARADIGM: [GENDER > CASE > NUMBER] MASCULINE SUBJECT

FEMININE OBJECT

SINGULAR

PLURAL

SINGULAR

PLURAL

SINGULAR

PLURAL

un-s

∅

un l-e

d-e-s l-e-s

un-e l-a

∅ l-e-s

INDEFINITE DEFINITE

l-i

This morphological re-organization has syntactic consequnces. As shown in (4), in Brendan, overt D is definite (l-) with number conditioned by definiteness: overt singular-marking with indefinites, overt plural-marking with definites. As shown in (5), in MdF, overt D can be definite (l-) or indefinite (d-PL), with the latter restricted to plural contexts. This sheds light on why number (qua NUM) is significant in both texts with indefinites, but significant with definites (qua D) only in MdeF, as d-PL introduces a number contrast in D. (4) a.

Brendan INDEFINITE

[Kase [D -sSU.MS.SG INDEF: ∅

b.

DEFINITE

-iSU.MS

MdF

[Kase [D [Num -sSU.MS.SG INDEF: ∅ SG: un INDEF: d-PL PL: -es -iSU.MS DEF: lSG: ∅ PL: -es

(5) a.

INDEFINITE

b.

DEFINITE

DEF:

l-

[Num SG: -un PL: ∅ SG: ∅ PL: -es

[Gender [ NP ]]]]] FEM: -eSG.INDEF MASC: ∅ SG.INDEF FEM: -aSG.DEF MASC: -eSG.DEF [Gender [ NP ]]]]] FEM: -eSG.INDEF MASC: ∅ SG.INDEF FEM: -aSG.INDEF MASC: -eSG.DEF

References. [1] Boucher, P. 2005. Definite Reference in Old & Modern French: The Rise & Fall of DP, in Batllori et al. (eds), Grammaticalization & Parametric Variation. OUP. [2] Carlier, A. 2013. Grammaticalization in Progress in Old French: Indefinite Articles, in D. L. Arteaga (ed.), Research on Old French: The State of the Art. Dordrecht: Springer. [3] Carlier, A. 2007. From preposition to article: The grammaticalization of the French partitive, Studies in Language 31, 1-49. [4] Foulet 1928/1974. Petite syntaxe de l'ancien français. Paris: H. Champion. [5] Mathieu, E. 2009. From local blocking to cyclic Agree: the role and meaning of determiners in the history of French, in Ghomeshi et al. (eds.), Determiners: variation and universals, John Benjamins. [6] Moignet, G. 1976. Grammaire de l’ancien français. Paris : Librairie Klincksieck. [7] Kihm, A. 2012. “Word & Paradigm” perspective and default syncretism, paper presented at Defaults in Morphological Theory, UKentucky.