Professional Documents
Culture Documents
17.1 Introduction
The fact that the Romance languages, since their earliest attestations, appear to be, from the
point of view of their syntactic type, much closer to one another than to their documented
common Latin ancestor is often cited as a most striking case of parallel development. As
such, it poses a serious challenge to non-directional theories of syntactic change (cf. the recent
discussion in Roberts 2007: 351-376, and Longobardi, this volume). In accounting for
syntactic change, a diachronic theory must be able to distinguish between three different
factors:
(1) a. interference
(parametric value) from one language to the other. However, for the syntactic domain, it has
often been proposed that (1.c) can also arise more indirectly, as the result of a chain of shifts,
and that consequently it is often difficult to empirically distinguish between (1.b) and (1.c).
That is, apparent cases of chance parallelism may in fact be due to a sequence of changes
brought about by an actually inherited grammatical property. The best known formulation of
the problem is represented by Sapir’s (1921: 172) notion of drift: ‘The momentum of the
more fundamental, the pre-dialectic, drift is often such that languages long disconnected will
pass through the same or strikingly similar phases’. Since Lightfoot (1979), theories of
explanations for syntactic change, on the basis of the fundamental assumption that the
memories’ of any sort. However, it has more recently been proposed that a better
understanding of the functioning of parametric systems can help derive long-term effects on a
principled basis (Roberts 2007: 340-357, Longobardi, this volume, Roberts, this volume).
In this chapter I will tackle the issue of parallel development and the problem of
distinguishing between (1.b) and (1.c) by focusing on the observed sequence of morpho-
syntactic changes affecting the realization of arguments of nominal heads (genitives) from
Latin to Romance. I will present data from a corpus search over Late Latin (3rd-4th cent. CE)
and Old French texts (11th-13th cent. CE), with the addition of some Middle French data (until
1600). The expression of adnominal arguments undergoes a deep restructuring from Latin to
the modern Romance languages: genitives, which are inflectionally encoded in Latin and
occur in a variety of configurations with respect to the head noun, come to be realized, in the
from Latin to Romance. However, neither the substantial similarity of outcomes in the
Western Romance languages, nor the existence of intermediate stages with coexisting
ultimate syntactic cause of the observed similarity in the Western Romance outcomes. I will
also propose a syntactic analysis of Old French inflectional genitives (realized with the cas-
régime absolu), which is able to capture the diachronic link to the Latin genitive constructions
and to elucidate the succession of steps leading to the generalization of the prepositional
change (1.c), thus supporting the idea that a parametric model of syntactic variation combined
with some theoretical assumptions about the dynamics of change can lead us to choose in a
In the following section, I will propose a parametric model for the syntax of genitives,
together with some of its diachronic implications. In section 17.3 I will discuss the Late Latin
situation, and in section 17.4 I will present the Old French data. Section 17.5 summarizes my
conclusions.
centers on the idea that genitive (i.e. the formal realization of adnominal arguments) is a
others. Parametric variation exists with respect to the kind of configuration allowed in a
language, and more than one mechanism can be present. While many languages have
languages (e.g. Classical Latin) the same morphological realization can be given to structures
linguistic variation found in this domain can be reduced to three fundamental strategies of
(2) a. GovG: a non-iterating in situ genitive, realized via a government configuration in the
position of Merge and requiring adjacency to its head noun (cf. Shlonsky 2004);
constraints (cf. analyses like Kayne 1994 and following, or Cinque 1994 for a parallel
The model in (2) is very similar to the one adopted in the parametrization of DP-internal
syntax by Gianollo, Guardiano, and Longobardi (2008), with the addition of the government
configuration in (2.a). Some of the cross-linguistic data on which it is based are presented and
discussed in Longobardi (2001) and Gianollo (2005). The typology in (2) is particularly well
illustrated by Classical Greek, a language that displays all the options above, and that, despite
giving them the same morphological realization, requires different distributional contexts for
(Thuc., II.58.2)
(Thuc., II.39.1)
(Thuc., III.115.6)
(Thuc. II.57.1)
(Thuc. II.57.1)
5
The examples in (3-5) show that: (i) a post-N genitive adjacent to the noun can express the
subjective (3.a) and the objective (3.b) function (but see Shlonsky 2004 on cross-linguistic
and construction-specific restrictions); (ii) genitives can also be pre-N (4.a); two genitives can
co-occur in pre-N position, respecting the thematic hierarchy Agent > Theme (4.b); (iii) in
definite DPs, it is possible to see that there is a further post-N configuration in Classical
Greek, in which the genitive is preceded by the doubling of the matrix DP’s definite article
(5.a). The same phenomenon appears with post-N adjectives (5.b); determiner doubling is in
fact the only possibility to license post-N adjectives (cf. the discussion of determiner
As seen in (3-5), the same morphological form can appear in different syntactic
configurations. This seems to be the case also for prepositional genitives cross-linguistically.
There is an ongoing debate with respect to their nature, motivated especially by the fact that
they are somehow ‘freer’ than their inflectional counterparts (can be iterated, can occur in
orders which do not respect the thematic hierarchy, can be non-adjacent to the head noun), but
at the same time are sensitive to structural constraints, such as possessivization, extraction,
and binding (cf. Giorgi and Longobardi 1991, Longobardi 2001, Androutsopoulou and
Español-Echevarría 2003 for a summary of the debate). It is plausible that this heterogeneity
may arise from a combination of licensing strategies. In 17.4.4 I will propose an analysis of
the Old French prepositional genitive, hinting however to some evidence suggesting that the
sample, the configuration in ModG is always realized post-nominally (with all probability
harmonically with the post-N realization of relative clauses). Thus, in these languages a GN
order can only be analyzed as an AgrG configuration, in one of the two dedicated projections.
The fact that there are two distinct positions is indicated not only by the possibility of
6
multiple occurrences (cf. 4.b), but also by the positioning with respect to adjectives. The two
projections for structural genitive licensing, tagged GenS and GenO, and their position with
(6) [D [GenS [Num [H1 [S-or [M1 H2 [M2 H3 [Arg H4 [GenO [αP [S [O…N]]]]]]]]]]]]
In (6), the lexical layer containing the noun and its arguments is indicated with α. S-or
refer to hierarchically ordered positions for adjectives. The positions tagged H1, H2, etc.,
Bernstein 1993).
The possibility for the noun head or the NP to raise across functional projections brings us
to the next point, i.e. the structural ambiguity of the NG order. A linear string NG in a
language can result in principle from three different configurations: (i) a GovG; (ii) one of the
two positions where AgrG is licensed, in languages where N(P)-raising to higher projections
is present; (iii) a ModG. As we have already mentioned, distinguishing between the three is
types (2.a) and (c), an inflectional form is compatible with all types of licensing. The
interplay with other DP-internal parameters is, instead, revealing. In particular, adjectives
represent a crucial diagnostics for the syntax of genitives. Since the functional projections
hosting adjectives are situated in between the two AgrG positions, the relative order of noun,
adjective and genitive is a crucial cue for the activation of AgrG projections. Also, an order
N-A-G immediately rules out the possibility of analyzing the genitive construction as a
7
I propose that the structural ambiguity of the NG sequence is primarily responsible for the
diachronic lability of post-N genitives. In the next sections, I will also argue that conditions
on structural economy play a role in the historical dynamics affecting these constructions.
The three basic types in (2) are listed in an order that mirrors their degree of syntactic
complexity, defined in terms of the interplay between the number of additional projections
that have to be generated and the movement operations necessary for licensing. Syntactic
complexity plays a crucial role in the process of reanalysis, in that learners’ choices are
guided by principles of structural economy in case of ambiguous input (cf. Roberts and
Roussou 2003, Van Gelderen 2004, and the discussion in Roberts 2007: 127 ff. and 226 ff.).
In the analysis of Late Latin genitive constructions that I will give in the next section, it will
be argued that the choice between two possible parametric interpretations of the NG string is
17.3.1 Corpus
My database for Late Latin (LL) is represented by a corpus of ca 80,000 words. It is,
therefore, much smaller than the database used for Old French (OF), for which a syntactically
parsed electronic corpus is available. The included texts date from the beginning of the 3rd
century CE (Passio Perpetuae et Felicitatis) to the end of the 4th (Peregrinatio Egeriae and
the four gospels in the Vulgata translation). If it is true that a major part of the corpus is
represented by the gospels, which are translations from the Greek, their uniformity with
respect to the native texts observed in the domain of nominal syntax allows for their use as
legitimate representatives of a non-artificial variety of Latin (see Gianollo 2011 for arguments
in support of this position). I will also make reference to the results of a wider study on
8
Classical Latin (CL), presented in Gianollo (2005, 2006). In general, the reader is referred to
these previous works for a fuller account of the Latin situation, which I will only be able to
Prepositional genitives in the Western Romance varieties are attested since the earliest
documents. They share the most fundamental syntactic characteristics (cf. Giorgi and
Longobardi 1991, Androutsopoulou and Español-Echevarría 2003) and the preposition di/de
introducing them can be formally traced back to a common Latin origin, the ablative
preposition dē. However, the genitive function of the prepositional phrase with dē does not
The inflectional genitive realization was the only way of encoding real arguments within
the CL nominal phrase. This situation persists significantly also in the LL texts included in
my survey: the prepositional phrase with dē + ablative is still overwhelmingly found with its
as that of earlier –especially pre-Classical– texts (cf. Molinelli 1996, Rosén 1999: 137-149,
Vincent 1999). The expression of real arguments with dē + ablative is extremely rare, and this
conclusion seems to hold also for later attestations (cf. Bonnet 1890: 607 f. on the few
examples found in Gregory of Tours, 6th cent., where the original ablative value of the
preposition is still clear, and Väänänen 1981 for an overview of documents from the 7th and
8th century).
However, despite the retention of the original inflectional system, LL shows an extremely
sample genitives almost invariantly follow their head noun, with only a few exceptions, which
can be straightforwardly accounted for as idiomatic expressions (cf. Gianollo 2005: 157 ff.).
GENITIVES NG GN
A similar shift is not observable with adjectives, which instead occur in pre- and post-N
ADJECTIVES NA AN
10
Evangelium sec. Ioannem 88 (68.8%) 40 (31.2%)
The major shift in the distribution of genitives, whose ultimate causes are admittedly
unclear (cf. Gianollo 2011), but do not seem to be reducible to concurrent morpho-syntactic
changes, results in a decisive loss in syntactic variation with respect to CL. Moreover, one of
the consequences of this shift is that it is no more possible to simultaneously express two
arguments of the same head noun. In my LL corpus there are no genuine instances of multiple
genitive realization within the same DP. The construction is generally rare also in CL, but the
examples show that it was possible to iterate the genitive structure both in pre- and post-N
position, and that, when both genitives were pre-N, the thematic hierarchy (subjective >
Multiple occurrences of ordered, pre-N genitives in CL were one of the crucial hints towards
As for post-N genitives, the main pieces of evidence were represented by the possibility of
iteration and the possibility for other DP-internal elements, such as adjectives, to occur
between the head noun and the genitive. This excluded the GovG configuration, leading to an
In LL, the fact that genitives start to be consistently realized to the right of the head noun
leads to the generation of an ambiguous input for acquisition: the NG sequence becomes
strongly P-ambiguous, in Roberts’ (2007: 233) terms, in the sense that it may be analyzed by
11
the learner according to different parametric values. As we have seen in section 17.1, three
possible parametric configurations can yield a NG order. In LL, however, the GovG
configuration is ruled out by the absence of the adjacency requirement: as in CL, adjectives
can intervene between the head noun and the genitive, as shown in (7).
In Gianollo (2005), Crisma and Gianollo (2006) it has been proposed that the primary
evidence available in LL is responsible for triggering the reanalysis of the post-N genitive
from type (2.c) to type (2.b). The CL parametric interpretation of the post-N genitive as a
ModG is now ruled out on the basis of (i) empirical evidence, consisting in the absence of
multiple occurrences, and (ii) a principle of structural economy, valuing the AgrG
configuration as less syntactically complex than the ModG one. The learner opts for the
the existence of N-A-G sequences, as seen in (7), and of pre-N adjectives (cf. Table 17.2 and
7.a), the only position available is the lowest projection below ordered adjectives (GenO in 6).
12
According to this proposal, the reanalysis of the genitive construction, due to the interaction
with adjectival syntax, is responsible for the birth of N(P)-raising, i.e. for the assumption on
the part of the learner of a movement operation which takes the noun to land in an
N(P)-raising is shared by all the Romance languages, and it is therefore highly significant
that, if the present reconstruction is correct, it is actually possible to trace back its origin to the
common LL stage. In the following section I will argue that precisely the inheritance of this
pan-Romance feature and of the post-N genitive will trigger the sequence of changes leading
to the system of genitive realization witnessed by Old French, and, ultimately, to the
17.4.1 Corpus
For Old French (OF), and for the partial overview of Middle French (MF) data that I will
present, I have relied on the syntactically parsed electronic corpus MCVF edited by Martineau
(2008). I complemented the evidence offered by the MCVF corpus with another text in
verses, La Vie de Saint Alexis (11th cent., 4.636 words), which represents a particularly
archaic stage of the language. The MCVF texts included in my search are presented in Table
TABLE 17.4 Old and Middle French texts from the MCVF corpus used in the study
13
Le Voyage de Saint Brendan 1120-1125 10,829
Later
TOTAL 784,702
Some of the queries have been performed on the whole material. In other cases I concentrated
14
(9) a. the cas-régime absolu (Foulet 1928): also called Juxtaposition Genitive in the literature,
e.g. Arteaga (1995), Delfitto and Paradisi (2009); inflectional realization in the oblique
la teste [d’Agolant]
la chambre [a la pucele]
Type (9.c), which is marginally retained in Modern French, is restricted to the expression of
[+human] arguments, and is always rare in the texts examined. For the present purposes, I
will disregard this configuration, to concentrate on the relation between the cas-régime absolu
In the most ancient text included in the survey, the Vie de Saint Alexis, the distribution of
the prepositional phrase introduced by de attests its full grammaticalization as the expression
15
of adnominal arguments. However, prepositional genitives occur alongside inflectional
realizations of genitives in the cas-régime absolu. Table 17.5 shows that in the Vie de Saint
Alexis the number of occurrences of prepositional genitives only slightly exceeds that of
inflected ones.
TABLE 17.5 Prepositional vs. inflected genitives in the Vie de Saint Alexis
GENITIVES DE G BAREG
loss of the inflectional realization, and a long period of co-existence of the two constructions
has to be accounted for. As Figure 1 shows, the use of the cas-régime absolu decreases during
the 12th and 13th century, to disappear by the MF period, following the more general fate of
the two-case declension (on which see Schøsler 1973, 1984, Plank 1979, Reenen and Schøsler
2000). 3
FIGURE 17.1 The disappearance of the inflected genitive (column shading indicates the
!"#$%&'&#"()*&(*&+*#,&-*./(&
!"+$#
!"+!#
!"&)#
!")*#
!"'(#
!"'%#
!"%&#
!"!$#
!"!!#
3 1#
>= 59#
.3 0#
,- 3#
78 3#
/= 3<#
>/ #
<#
,- #
0/ <#
>> #
<< #
5; 0#
9- 9 #
7 #
C, B-7 #
@= 0>#
6 >#
0< 50
:= 3<
.3 03<
-5 3<
9:
6 00
A8 50
.3
/0 0
>
.6
2, /0
4 01/
/,
9: /5
90
9- ,?
5<
8,
,
3<
-,
.
6
,-
5.
16
Already in OF an overt morphological distinction between the nominative (cas-sujet) and the
oblique (cas-régime) was consistently visible only in the masculine paradigm of nouns and
determiners, as shown in Table 17.6. Most feminine nouns had a unique form for both cases
and retained only the morphological marking for plural number. Only the few feminine nouns
that did not end in -e muet could mark the cas-sujet singular with an -s, borrowed from the
masculine paradigm; in addition to these, a handful of nouns retained a marked cas-sujet (e.g.
suer/seror ‘sister’, none/nonain ‘noon’, Eve/Evain ‘Eve’). In any case, for feminine nouns the
two-case distinction was restricted to the singular. Nonetheless, also feminine nouns were
allowed in the BareG configuration, although very rarely. This point will, in fact, turn out to
M F M F
sg pl sg pl sg pl sg pl
In this section, I will examine the syntactic behavior of the OF inflectional genitive
construction (type 9.a). In order to do so, I will start from the traditional description of the
cas-régime absolu given by Foulet (1928), and discuss it in light of the data coming from my
17
Foulet (1928: 14-23) offers an analysis of the distribution of the inflectional construction,
Preferences
Requirements
To this points it should be added that, according to the data in the corpus, BareG is never
iterated, i.e. there can be no more than one BareG for DP.
Studies such as e.g. Palm (1977), Herslund (1980), Schøsler (1984), Arteaga (1995),
Delfitto and Paradisi (2009), have already confirmed the accuracy of Foulet’s description with
respect to points (10.a, b, e). I will rather comment on the remaining ones, which are of more
direct import to understand the syntactic configuration in which OF inflected genitives occur.4
Table 17.7 shows the distribution of inflected genitives in two representative OF texts,
according to (i) their order with respect to the head noun; (ii) the nature of the complement
with respect to the proper name/common noun distinction; (iii) the complement’s number. 5
Table 17.7 largely confirms point (10.c), also in its not being an absolute restriction, since
plural complements are not impossible. Point (10.d), on the other hand, does not appear to
qualify as an accurate generalization, since common nouns may even represent the majority of
It has been long noticed (e.g. Togeby 1974, Palm 1977, and more recently Delfitto and
Paradisi 2009) that also generalization (10.f) is incorrect: although there is a predominance of
possessive and subject-like relations, the cas-régime absolu can also have the function of an
b. le servise Jhesucrist
In the corpus examined by Palm (1977), represented by literary texts of the second half of the
12th cent.-first quarter of the 13th cent. and partially overlapping with ours, over a total of
1395 BareG with singular nouns, 62 (4%) can be analyzed as objective genitives. The rate is
comparable to that observed among prepositional complement introduced by a (3%), and not
19
Turning to the definiteness requirement in (10.g), our search over the corpus confirms the fact
that the BareG must always be definite. No examples occur of article-less inflectional genitive
DPs, apart from the cases where proper names, pronouns like celui, or definiteness-inducing
elements such as possessive adjectives or demonstratives are present. The only cases where
the complement can be indefinite are found after the noun mi ‘middle’ (e.g. par mi un val
herbus ‘across a grassy valley’, Roland 994). This construction, however, is better analyzed
parmi), governing, as all other prepositions, the oblique case, which is therefore not a real
BareG anymore (cf. fn. 5; cf. also the fact that after mi the complement can also be [-human]).
Thus, it is possible to conclude that the inflectional genitive always needs a strong element in
D.
The definiteness requirement does not hold for the matrix DP, as is instead implied by
Delfitto and Paradisi’s (2009) treatment of the BareG construction. Although examples in the
corpus are rare in argument position (whereas in predicative position article-less DPs can
more freely occur), they show that DPs with an indefinite determiner can have a BareG
complement, as in (12).
Apparently, therefore, no agreement in definiteness features holds between the matrix and the
genitive DPs. This is a first, important hint toward the fact that a construct-state analysis for
20
these constructions is not correct (since definiteness inheritance is a necessary property of
construct states, cf. Ritter 1991, Siloni 1997, Longobardi 1995, 1996). A second piece of
evidence against a construct-state analysis is represented by the fact that, as shown in (13),
adjectives may intervene between the determiner and the head noun in the matrix DP: the
N(P), therefore, cannot be argued to raise to the projection of D, as is commonly assumed for
Semitic (and marginal Romance) construct states (on the frequent confusion between
Romance construct states and the cas-régime absolu construction cf. Longobardi 1995: 307,
‘to the solemn celebration of St. Michael, (patron) of the danger’ (Roland 125)
This last observation brings us to our last point, (10.h), relative to the position of BareG. As
seen in Table 17.7 above, the post-N order is overwhelmingly dominant (respectively in
95.5% and 96% of the cases). Foulet (1928: 18 f.) states that pre-N cas-régime absolu is
typically found with the name ‘God’. With other nouns, the order is almost always found in
the rhyming part of the verse. For the purposes of this paper, the pre-N inflected genitive will
be considered a relic and will be disregarded in the following discussion. In order to fully
confirm (10.h), also the requisite of immediate adjacency has to be checked against the data in
21
the corpus. Foulet (1928: 17-18) states that a word appears between the head noun and the
inflected genitive only very seldom. The intervening element, judging from Foulet’s
examples, is never part of the DP, but results from independent sentence-level parameters,
governing the position of connectives and verbs, which are most often prosodically weak:
The query on the MCVF corpus and on the Vie de Saint Alexis confirms the adjacency
requirement. In particular, adjectives, the most likely candidates for DP-internal intervention,
never occur between a noun and a BareG.6 Since this aspect, as with LL, turns out to be
crucial for the diachronic interpretation, I will postpone a more thorough discussion of the
data to the following section, where I will propose my analysis for the structural configuration
in which BareG is licensed, and reconstruct its relation to the Late Latin inheritance.
To sum up this section, I propose in (15) below a modified version of (10), where some
Preferences
c. the complement is overwhelmingly singular, but there is no ban against plural DPs
22
d. the complement does not show a distinct preference for being a proper name
Requirements
g. it must always be definite, but there is no obligatory definiteness agreement with the
matrix DP
In this section I will argue that OF inflected genitives are, in fact, a continuation of Latin from
a syntactic point of view. More precisely, I will propose that they represent the result of a
Some of the conditions in (15) suggest that the BareG construction represents the outcome
of the transmission of some fundamental parametric values relative to the nominal domain
from LL to OF. If it is impossible to straightforwardly derive the OF system from the CL one,
the hypothesis that the OF BareG inherits the characteristics of the intermediate LL stage
sheds light on a number of shared properties: (i) the fact that the noun raises to an
intermediate projection in the functional layer, landing in between the ordered adjectival
projections, (ii) the fact that a non-iterated, inflectional genitive, which is able to express any
thematic relation with the head noun, occurs in a post-N position. We may add to this the
observation that the residual pre-N instances of cas-régime absolu appear in a position that is
arguably the same of the residual pre-N genitives in LL, i.e. in the high AgrG projection,
immediately below D.
23
Despite these basic similarities, the OF construction displays an innovative structural
property with respect to the LL post-N genitive realized in the low AgrG projection, namely
strict adjacency with respect to the head noun. We have seen in (7) that this was not the case
in LL, as N-A-G sequences were possible. Therefore, to corroborate the hypothesis that the
cas-régime absolu continues the LL post-N genitive, a principled explanation has to be found
as to why the new requirement (15.h) would arise. In what follows, I will try to do so,
connecting the adjacency requirement to another peculiar feature of BareG, the definiteness
requirement on the complement (15.g), and establishing a structural parallelism with the DeG
configuration.
I ended the last section by discussing the respective positioning of BareG and adjectives. In
Table 17.8 I present an overview of the data obtained from the MCVF corpus, also with
A-N-BareG 3 2 12
N-A-BareG 0 0 2 (1?)
Adjectives can co-occur with the BareG construction (cf. also 13), but only if they precede the
head noun. The only safe example of adjective intervention in a DP hosting a BareG
24
lexicalized expression (dou frere germain le duch de Bretagne ‘of the duke of Brittany’s
The corpus study demonstrates that the requisite of adjacency imposed on such a
realization is also shared, until the Middle French period, by the prepositional genitive. This is
a noteworthy difference with respect to Modern French, where the intervention of an adjective
between the head noun and the prepositional genitive is perfectly grammatical (in 16.a the
The only example of a N-A-DeG sequence in Roland shows, in fact, a still strong ablatival
value of the de-PP, which may therefore not qualify as a proper argument:
The counterexamples found in the corpus are extremely rare in the OF period (4 to 7,
adjunct in some cases). They slightly increase in the MF period, but the great majority of the
instances counted in Table 17.8 (104 out of 165, 63%) are attested in the latest text included
25
Also another fact observable in the corpus differentiates the DeG configuration from the
modern prepositional genitive: there are no instances in which DeG is iterated. In Modern
the same DP, as shown by the example in (18) (taken from Sportiche 1990; multiple orders
are possible):
What is more, there are no instances in the corpus where DeG and BareG co-occur as
realizations of arguments of the same head noun. This hints toward the fact that they may
compete for the same position. The only difference between DeG and BareG observed in the
literature (e.g. Foulet 1928, Palm 1977) is that, especially in the older texts, de is preferably
used with animals, inanimate referents, or (human) kinds, and when the relation is that of an
objective genitive. However, these are only tendencies: the examples in (19) show that, since
the earliest stages of the language, DeG could express any thematic relation –respectively,
b. cunseill d’orguill
counsel of pride
My hypothesis is that, in fact, the DeG and the BareG realizations have a uniform structural
source. They are licensed in a configuration that inherits from Late Latin the post-N position,
its non-iterability, and the possibility of realizing any thematic role, thus the fundamental
syntactic cues. It is further reanalyzed in OF, however, due to changes in the morphological
cues.
(1987), Plank (1992) have shown that the morpho-syntactic development of the English
genitive was conditioned by the increased uniformity of the genitive exponent, i.e. by the
reduction of the declensional system to a unique genitive mark for all declensional classes.
The same change can be observed in OF, apart from lexical exceptions. Moreover, most
feminine nouns occur in the construction completely unmarked for case, and are in fact quite
rare. Adjectives display a strong tendency to lose case marking: there is a unique form for
feminine agreement and for masculine forms ending in -e; adjectives ending in a consonant
show many exceptions. In most instances with masculine nouns, the only unambiguously
case-marked element within the genitive constituent is the element in the Determiner position
(article, demonstrative, possessive adjective; cf. Palm 1977, Reenen and Schøsler 2000).
In this situation, OF case morphology is too weak to represent, in the process of language
projection: thus, the head noun’s extended projection loses its ability to license nominal
Given the disappearance of the morphological cue, a further reanalysis takes place at this
configuration (type 2.a). A GovG configuration cannot be iterated (condition 15.i); it also
27
requires strict adjacency between the head noun and the complement (condition 15.h), and
that is why the N-A-G sequence, probably already underrepresented in the primary data,
given the tendency of OF adjectives to occur prevalently in pre-N position (cf. Boucher 2002,
2004), becomes ungrammatical. The fact that in OF the post-N genitive is ‘frozen’ in place by
the licensing under government does not imply, however, that raising out of the lexical shell
is lost; rather, the whole NP complex reaches a higher position in the functional layer,
In the spirit of the treatment of lexical case by Bayer, Bader, and Meng (2001), I assume
that the genitive complement takes the form of an outer KP, embedding a DP and acting as a
probe for the valuation of formal features of the complement. Both inflectional endings and
lexical features, such as de, cf. Vincent (1999)– can act as exponents of K0. More specifically,
(20) Move
c. possessive adjectives
Merge
d. functional preposition de
The Merge option yields the DeG configuration, while the Move option results in BareG.
This derives the requirement in (15.g) that the constituent in the cas-régime absolu be always
definite, accounting in a principled way for Foulet’s (1928: 20) observation that the
distinct de tous les autres individus’. The semantic properties of the construction, under this
28
account, are only a by-product of the morphosyntactic requirement to fill K0. An overt
indefinite determiner is not able to satisfy such requirement because it is not in D0, but
Lyons 1999).
requirement holding for post-N genitives in Modern German, the so-called ‘genitive rule’ (cf.
Gallmann 1998, Lindauer 1998): according to it, a non-prepositional post-N genitive is only
possible if (i) represented by a proper name, or if (ii) there is at least one inflected form within
the DP (determiner, possessive pronoun or other inflected pronoun, adjective with strong
suffix). Further research is needed to assess whether the diachronic emergence of this
requirement can be linked to a reanalysis of the configuration for post-N genitives along the
The cas-régime absolu and the OF prepositional genitives have the same underlying
structure and coexist, since the earliest documents, and until the two-case declension
eventually disappears. The DeG configuration, being subject to no constraint relative to the
syntactic and semantic nature of its DP complement, is bound to become more frequent. The
The mentioned increase in the occurrence of the N-A-G sequence after the Middle French
period signals that, at some point, the prepositional construction itself undergoes the
reanalysis which brings about the ‘modern’ characteristics of de-genitives, i.e. the possibility
of non-adjacency with the head noun and the possibility of multiple occurrences within the
same DP. In the proposed system, these properties are compatible with an analysis in terms of
a ModG configuration (type 2.c). This further diachronic step, which would attest once more
29
to the lability of the post-N genitive, must however be investigated on a much broader textual
basis.
established, the ultimate source of the prepositional construction can be traced back to the
crucial shift occurring in Late Latin, which unequivocally transmits to the daughter languages
My account, although grounded into different assumptions about the nature of genitive
licensing, is in substantial agreement with the previous formal treatment of the cas-régime
absolu construction proposed by Arteaga (1995): the diachronic processes affecting this form
of genitive is ultimately due to the featural weakening of the functional projection licensing it,
Ultimately, if we assume, as is plausible, that similar conditions may have hold for the
other Western Medieval Romance varieties, the reconstruction proposed here accounts for the
parallel development of prepositional genitives in these languages, whose ancestor only had
traditionally interpreted as an instance of (1.b) turns out to be, instead, at least for OF, an
17.5 Conclusion
In this chapter I have traced the history of genitive realization from Late Latin to Old French.
I have proposed that a bundle of parametric features, crucially comprising the realization of
transmitted from Late Latin to Old French and is responsible for the retention of the genitive
construction displays some innovative structural properties with respect to the originating
30
Latin construction, most notably strict adjacency to the head noun; moreover, it alternates
since the first texts with the prepositional expression of genitives. The existence of a mixed
linked to the loss of the inflectional realization, but also points to some restrictions that
prevented the inflectional genitive from being fully productive. I have argued that the ultimate
explanation for this state of affairs has to be sought in the interplay between morphology and
syntax.
The more general conclusion that can be drawn from my analysis is that post-nominal
genitive is diachronically labile because the space of parametric variation within genitive
syntax and the conspiracy with the values of other parameters allows, in principle, for three
different grammars for NG sequences. The Old French data seem to suggest that choosing
among the three in language acquisition may also be a matter of morphology, as deflexion
post-nominal genitives has to be carefully taken into account. Some parametric values are
more likely to change than others: in some cases their lability can be attributed to the fact that
they are subject to the implicational effect of superordinate parameters (Roberts 2007: 357),
but in other cases, like the one under exam here, the ambiguity of the trigger arising under
Primary sources
Vulgata
Nestle, Eberhard, Erwin Nestle, Barbara and Kurt Aland et al. (19943). Novum Testamentum
31
Passio Perpetuae et Felicitatis
Van Beek, Cornelius (1936). Passio sanctarum Perpetuae et Felicitatis. Nijmegen: Dekker &
Van de Vegt.
Peregrinatio Egeriae
Heraeus, Wilhelm (1908) (1939). Silviae vel potius Aetheriae Peregrinatio ad Loca Sancta,
Heidelberg: Winter.
Storey, Christopher (1968). La Vie de Saint Alexis, Texte du Manuscrit de Hildersheim (L).
Corpus MCVF
Martineau, France (2008). Corpus MCVF : Modéliser le changement: les voies du français. U.
Ottawa, http://www.voies.uottawa.ca/corpus_pg_fr.html
References
Alexiadou, Artemis and Chris Wilder (1998). ‘Adjectival Modification and Multiple
Determiners’, in Artemis Alexiadou and Chris Wilder (eds.), Possessors, Predicates and
32
Anderson, Stephen R. (2002). ‘Syntax and Morphology are Different. Remarks on Jonas’
28, 3-17.
Arteaga, Deborah (1995). ‘On Old French genitive constructions’. In Jon Amastae et al.
Bayer, Josef, Markus Bader and Michael Meng (2001). ‘Morphological underspecification
meets oblique case: Syntactic and processing effects in German’. Lingua, 111, 465-514.
Bernstein, Judy (1993). Topics in the Syntax of Nominal Structure across Romance. Ph.D.
dissertation, CUNY.
Bonnet, Max (1890). (repr. 1968). Le latin de Grégoire de Tours. Hildesheim: Georg Olms.
Boucher, Paul (2002). ‚Determiner Phrases in Old and Modern French’. In Martine Coene
and Yves D’Hulst (eds.), From NP to DP. Volume I. The syntax and semantics of noun
Anna Giacalone Ramat et al. (eds.), Papers from the 7th International Conference on
Cinque, Guglielmo (1994). ‘On the evidence for partial N-movement in the Romance DP’. In
Guglielmo Cinque et al. (eds.), Paths Towards Universal Grammar. Studies in Honor of
33
Crisma, Paola and Chiara Gianollo (2006). ‘Where did Romance N-raising come from? A
parallel study of parameter resetting in Latin and English’. In Jenny Doetjes and Paz
González (eds.), Romance Languages and Linguistic Theory. Selected Papers from ‘Going
Delfitto, Denis and Paola Paradisi (2009). ‘Towards a Diachronic Theory of Genitive
Foulet, Lucien (1928). Petite Syntaxe de l’ancien français. Paris: Champion, 3rd revised
edition.
Gallmann, Peter (1998). ‘Case Underspecification in Morphology, Syntax and the Lexicon’.
In Artemis Alexiadou and Chris Wilder (eds.), Possessors, Predicates and Movement in
Gianollo, Chiara (2005). Constituent structure and parametric resetting in the Latin DP: a
Gianollo, Chiara (2006). ‘Tracing the value of syntactic parameters in ancient languages: the
Latin nominal phrase’. In Karlene Jones-Bley et al. (eds.), Proceedings of the Seventeenth
Gianollo, Chiara (2011). ‘Native syntax and translation effects. Adnominal arguments in the
Greek and Latin New Testament’. In Eirik Welo (ed.), Proceedings of the International
Gianollo, Chiara, Cristina Guardiano and Giuseppe Longobardi (2008). ‘Three fundamental
34
Giorgi, Alessandra and Giuseppe Longobardi (1991). The Syntax of Noun Phrases.
Guardiano, Cristina (2003). Struttura e storia del sintagma nominale nel Greco antico:
Jespersen, Otto (1918). ‘Chapters on English’. Reprinted 1960 in Selected Writings of Otto
Kayne, Richard (1994). The Antisymmetry of Syntax. Cambridge, MA: MIT Press.
Press.
Alexiadou and Chris Wilder (eds.), Possessors, Predicates and Movement in the
Longobardi, Giuseppe (1995). ‘A case of construct state in Romance’. In Roberto Ajello and
Saverio Sani (eds.), Scritti Linguistici e Filologici in onore di Tristano Bolelli. Pisa: Pacini,
293-329.
Longobardi, Giuseppe (1996). The Syntax of N-raising: a minimalist theory. Utrecht: OTS
Longobardi, Giuseppe (2001). ‘The Structure of DPs: Some Principles, Parameters, and
Problems’. In Mark Baltin and Chris Collins (eds.), The Handbook of Contemporary
35
Molinelli, Piera (1996). ‘Casi e preposizioni in latino: lo sviluppo del genitivo e del dativo.
Palm, Lars (1977). La Construction ‘Li filz le rei’ et les constructions concurrentes avec ‘a’ et
‘de’ étudiées dans des oeuvres littéraires de la seconde moitié du 17. siècle et du premier
Plank, Frans (1979). The functional basis of case systems and declension classes: from Latin
Plank, Frans (1992). From cases to adpositions. In Nicola Pantaleo (ed.), Aspects of English
Reenen, Pieter van and Lene Schøsler (2000). ‘Declension in Old and Middle French. Two
opposing tendencies.’ In John C. Smith and Delia Bentley (eds.), Historical linguistics
1995. Vol. 1. General issues and non-Germanic languages. Amsterdam: Benjamins, 327-
344.
Ritter, Elizabeth (1991). ‘Two functional categories in Noun Phrases: Evidence from modern
Hebrew’. In Susan Rothstein, Syntax and Semantics 26. San Diego: Academic Press, 37-
62.
Roberts, Ian and Anna Roussou (2003). Syntactic Change. A Minimalist Approach to
Rosén, Hannah (1999). Latine loqui. Trends and Directions in the Crystallization of Classical
36
Schøsler, Lene (1984). La Déclinaison bicasuelle de l’ancien français. Odense: Odense
University Press.
Shlonsky, Ur (2004). ‘The form of Semitic Noun Phrases’. Lingua, 114.2, 1465-1526.
Siloni, Tal (1997). Noun Phrases and Nominalizations. The syntax of DPs. Dordrecht:
Kluwer.
Sportiche, Dominique (1990). ‘On Movement, Agreement and Case’. Ms., UCLA.
Forlag.
Vincent, Nigel (1999). ‘The evolution of c-structure: prepositions and PPs from Indo-
1
For the treatment of CL post-N adjectives, see Gianollo (2005, 2006), where they are
text is provided on the MCVF website. Some dates, which do not correspond to the scholarly
37
Dictionnaire Étymologique de l’Ancien Français (DEAF). I am grateful to an anonymous
remarkably rich variety of syntactic and semantic contexts, could be interpreted as a sign of a
longer persistency of the construction in the Northern and Eastern dialects, especially Picard.
This would be consistent with the longer preservation, in such varieties, of the two-case
declension, due at least in part to the fact that in the Northern regions final -s disappears later
(cf. discussion in Schøsler 1973: 260 f., Reenen and Schøsler 2000). However, the Middle
French texts included in my search are too few and too scattered in time to allow for a safe
conclusion on this point. I thank Richard Ingham for suggesting this possibility to me.
4
In my system, the [+human] feature does not play a syntactically relevant role and, as such,
remains unexplained. Delfitto and Paradisi (2009) propose that it may be related to the
retrieved by the MCVF search, but which I have set apart in my analysis, in order to comply
with the criteria that I have adopted in my analysis of Latin. They are, respectively, a
construction with a gapped noun, and an instance of en-cliticization. On the other hand, in
order to favor the replicability of the query, I have kept in the total a number of structures
which are tagged as inflected genitives in the corpus, but which may not, in fact, represent
adnominal arguments: these are forms of cas-régime complementing the noun mi ‘middle’,
which however appears in most occurrences to form a complex preposition together with par
38
based on a limited number of examples, which will be hopefully expanded by future research
immediately post-N PPs headed by de, without recognizing their function: thus, for instance,
also partitives and complements of quality and material are included. Due to the large number
of occurrences, I did not perform a case-by-case analysis for this combination; therefore I
offer in the table both the total number of de-PPs and an estimate of the number of argumental
genitive among them (60% of the total). The estimate is based on the case-by-case exam
precise data in the future, especially with respect to the diachronic developments in the
functions of de-PPs.
8
Notice that here no isomorphism is necessarily assumed between syntactic structure and
certain parametric value, representing a surface cue to parameter setting during acquisition.
The impoverishment of such a cue can conspire in creating a situation of strong P-ambiguity,
thus triggering reanalysis (cf. the discussion in Anderson 2002 and Roberts 2007: 136 ff.). I
39