Chapter 4 Why “Partitive Articles” Do Not Exist in (Old) Spanish

In: Disentangling Bare Nouns and Nominals Introduced by a Partitive Article
David Paul Gerards University of Leipzig

Search for other papers by David Paul Gerards in
Current site
Google Scholar
Elisabeth Stark University of Zurich

Search for other papers by Elisabeth Stark in
Current site
Google Scholar
Open Access


In this paper a formal morphosyntactic analysis of so-called “partitive articles” available in some Modern Gallo- and Italo-Romance varieties is proposed. The idea that they are, in the singular, mass classifiers located in Div° (cf. Borer 2005) signaling non-individuation is formulated. The de-element contained in “partitive articles”, morphologically complex items, is in complementary distribution to explicit “dividers” such as unambiguous overt plural morphemes. This leads to the prediction that “partitive articles” are not available in Romance varieties with (unambiguous) overt nominal plural marking, a prediction tested against Old Spanish data (technical prose from the 13th to 16th century), a language with unambiguous sigmatic nominal plural morphology and for which the availability of mass classifiers of the Modern Gallo- or Italo-Romance type has been repeatedly claimed. Refuting this claim, the paper shows that what seem to be attestations of “partitive articles” are bare partitives, that is, QPs with zero Q° containing prepositional phrases. The latter, in turn, can contain definites with representative object interpretations licensed by the so-called kind-oriented mode of talk. Such definites share some properties of short weak definites but will be shown to differ from the latter. The prediction of our formal analysis is thus borne out for (historical) Ibero-Romance data.

1 Introduction

The aim of this paper is twofold. Firstly, we present a formal analysis of so-called indefinite “partitive articles” available in some Modern Gallo- and Italoromance varieties. We put forward the idea that these are, sensu lato, a sort of nominal classifier and indicate non-individuation or mass in the singular, a value often considered to be the unmarked, default interpretation of nominals (cf. Borer 2005, 93). In a Pan-Romance perspective, “partitive articles” are in complementary distribution with unambiguous, agglutinative plural markers (like -s in Spanish amigo-s, ‘friend-s’), and lead to a mass reading of the respective nominal. We will argue that their existence is due to a (partial) diachronic loss of unambiguous number markers, that is, vocabulary items (vi) to express interpretable φ-features (number) on nouns and general agree requirements inside nominals (Stark 2008b; Mathieu 2009). We will follow Borer (2005, 93) in assuming identity for elements in complementary distribution across languages and argue that the de-element in Romance “partitive articles” realizes the same functional head as agglutinative plural morphemes. This leads to a strong generalization, namely that in Romance varieties or older stages of Romance languages with unambiguous (usually sigmatic) nominal plural marking, indefinite “partitive articles” in the singular, that is, mass determiners or classifiers, should not be available (cf. also Mathieu 2009, for Old French).1 Secondly, this generalization will be tested against 275 Old Spanish occurrences of con del/de la/de los/de las ‘with of.the’ from the 13th–16th century stemming from the Corpus Diacrónico del Español (Corde) and the Corpus del Español (CdE). Contrary to some authors who have claimed that Modern Gallo- or Italo-Romance-like indefinite “partitive articles” are sporadically attested in this variety, too (e.g., Lapesa 1964; Cano 1992; Eberenz 2008; cf. also Crispim 1996; Mattos e Silva 2008, for Old Portuguese), we will show that Old Spanish did not possess indefinite mass determiners (“partitive articles” in the singular). Instead, what seem to be “partitive articles” are to be analyzed as pps governed by a zero Q° and containing a definite dp with a representative object interpretation licensed by the so-called kind-oriented mode of talk (Krifka et al. 1995, 83–88). Such pps are restricted mainly to the 13th century. Old Spanish, which has agglutinative sigmatic nominal plural marking, like Modern Spanish, is thus no counterexample to our generalization.

In Section 2, we will present the theoretical background (Section 2.1) as well as our analysis of Modern Gallo- and Italo-Romance “partitive articles” as indefinite mass classifiers (Section 2.2). Section 3 will be dedicated to Old Spanish del-constituents. It presents the data, that is, the results of an exhaustive string query in two (Old) Spanish corpora (Section 3.1), followed by an analysis of these Old Spanish data as pps containing definites with representative object interpretations (Krifka et al. 1995), a special class of weakly referential dps merged in D and denoting prototypical instantiations of the respective kind (Section 3.2). This makes Old Spanish del-constituents semantically and syntactically different from Modern Gallo- and Italo-Romance “partitive articles”, that is, from indefinite mass classifiers. Additionally, in Section 3.3, we will address a possible alternative analysis of the Old Spanish data in terms of short weak definites in the sense of Carlson and Sussman (2005) and Carlson et al. (2006). We will show that short weak definites and definites with representative object interpretations, despite sharing some properties, are clearly different from each other and that only an analysis in terms of representative object interpretations can adequately capture the Old Spanish data. Section 4 summarizes the main findings of the paper.

2 Why “Partitive Articles” in Romance?

2.1 Theoretical and Typological Background

Some Romance languages are known for a typologically highly marked element, namely an indefinite determiner encoding mass in the singular (Herslund 1998; Stark 2008a, 2008b, 2016; Cardinaletti and Giusti 2016a, 2018).2 Despite their being “DP-argument languages” in Chierchia’s (1998) parametrization, that is, languages without nominal classifiers in the strict typological sense, Romance languages and varieties feature a systematic opposition between mass and count interpretations (= reference to portions vs. atomic units, individuals, see below) of the nominal predicate, for example ‘bread’ in (1) and (2). This opposition is encoded in their system of nominal determination in two different ways, either by marking the count vs. mass reading by an adequate numeral or quantifier, and having most often zero as the non-marked default case (= mass), or by systematically marking also the mass reading, namely by means of a “partitive article”:






J’ achète







‘I buy bread.’









J’ achète











‘I buy a very tasty bread.’

Even though, etymologically, the element du/del3 in (1) goes back to a composition of the Latin preposition de ‘from, of’ and the definite article (resulting in ‘of the’ when translated literally; a possible reading of the homonymous expressions combining a preposition de or di and a definite article in Modern Romance), there is no doubt that the indefinite determiners under investigation here are no longer compound pps (cf. e.g., the extraction facts discussed in Ihsane 2013, 236; see also Cardinaletti and Giusti 2016a, 2018). Their semantics is also clearly indefinite. Like the indefinite count determiner in (2), stemming from the Latin numeral unus (cf. Givón 1981 and, e.g., the detailed and comparative discussion of the different grammaticalization steps in Mulder und Carlier 2011), the mass determiners (“partitive articles”, pa in the glosses), too, are excluded with textually or situationally given discourse referents, which inherit their mass- vs. count interpretation from their antecedent.4

While in the last decades, in-depth research has been done on the history (e.g., Carlier 2007; Carlier and Lamiroy 2014), syntax and semantics of Romance determiners, especially for French and standard Italian, we want to rather focus on an explanation of their cross-linguistic distribution, as not every Romance variety features all of them. In fact, indefinite mass determiners (“partitive articles”) are only found with a certain regularity in French (cf. Dobrovie-Sorin, this volume) and Northern Italian varieties (cf. Giusti, this volume), in some Occitan varieties and in Francoprovençal (cf. Schurr, this volume, and Stark and Gerards, this volume), whereas they do not exist in Modern Ibero-Romance varieties (see Section 3), central and southern Italian varieties or Romanian (see Bossong 2016; also Giusti, this volume). These latter languages and varieties admit bare plurals and bare singulars in argument position, contrary especially to French, and to a lesser extent to non-central and non-southern Italian varieties (Stark 2008a, 2008b, and 2016). We would like to put forward the hypothesis that the (non-)existence of “partitive articles” and their complementary distribution with bare plural arguments is causally linked to the (non-)availability of a dedicated nominal plural morpheme (cf. Delfitto and Schroten 1991; Mathieu 2009; Carlier and Lamiroy 2014 for a similar descriptive generalization), that is, to morphosyntactic properties of nominal declension in Romance.

Ever since Borer’s (2005) seminal work on (among other things) the count-mass distinction, plural morphemes can be considered signals of “nominal classification” in a broad sense in non-classifier languages (cf. e.g., Cowper and Hall 2012).

Rather, all nouns, in all languages, are mass, and are in need of being portioned out, in some sense, before they can interact with the ‘count’ system. This portioning-out function, accomplished in languages like Chinese through the projection of classifiers, is accomplished in languages like English, by the plural inflection, as well as by the indefinite article.

Borer 2005, 93

Borer (2005, 111, 114, for a preliminary conclusion) later discusses the exact location of the indefinite article, which, contrary to plural morphemes, is to be seen as an “individuator” (cf. Wiese 2012, 72), that is, both as a “divider” and as a “counter” (= some kind of portmanteau-morpheme). It therefore ends up being located higher in the internal structure of nominals.


Figure 4.1

Internal structure of a count nominal following Borer (2005)

Plurals in Modern Romance (and generally Indoeuropean) varieties are thus classifying plurals in the sense that they unambiguously denote sets of sets (cf. already Link 1983; Borer 2005, 127), contrary, for instance, to Latin, where plurals could also denote intensive manifestations of the noun’s denotation or collectives (Stark 2008b). Overt quantifiers and numerals, including the indefinite article deriving from the numeral ‘one’, explicitly assign a specific quantity to the expression. Based on these assumptions, Borer (2005, 109) proposes the structure represented in Figure 4.1 for English count nominals, which we slightly modified for the lexical material (nP).

2.2 Our Analysis

Applying the analysis of indefinite (plural) nominals in Borer (2005) and Mathieu (2009, 2014), for instance, to Spanish, we can analyze the Spanish plural -s as the overt exponent of Div°, in parallel to English (cf. Pomino 2016, 111).

Following general assumptions in Distributed Morphology (Halle and Marantz 1993), we hold it that roots are not specified for grammatical categories, and follow Borer (2005, 93) in that they, that is, nouns, are not specified for mass or count readings either (cf. also Pelletier 2012). Roots combine in syntax with functional heads (n°, v°, a°) in order to form nominal, verbal, or adjectival constituents. These heads may contain lexical properties like animacy, often linked to gender or noun class. n° is, more concretely, the “locus of gender negotiation”; the gender feature of n° is valued under agree with the lexical root (cf. Lowenstamm 2007, 2012, for French; but cf. Stark 2016, for the assumption of a defective n° in French; cf. also Picallo 2008; Alexiadou 2015). nP then merges with a functional projection DivP, hosting the Num feature, expressed by number morphology, if available5 (cf. ClassP in Picallo 2008, 57; NumP in Mathieu 2009; note that Alexiadou (2015) calls the higher functional projection ClassP, i.e., what is called NumP in Picallo’s work and #P in Figure 4.1). Merge of #° brings the counting or quantifying head, carrying a probe for gender and number, for agreeing quantifiers and numerals (e.g.,, ‘much’, ‘many’, in Spanish).

However, applying Borer’s analysis for English to French brings some complication to the picture. The main difference lies in the availability of a mass classifier (cf. examples under (1)), absent in English or Spanish, correlating with the absence of number marking on nouns (cf. e.g., Pomino 2012, 2016). French looks like the complete opposite of English and Spanish, in that a plural exponent for Div° is not available, but rather an exponent for “not portioning out”, namely de. French seems to possess an exponent for non-individuation, in complementary distribution with exponents for individuation, as a result of the interaction of the elements available in #° and Div° (cf. examples under (1) against examples under (2); cf. Borer 2005, 128). A nominal such as [vɛ̃], ⟨vin(s)⟩, ‘wine(s)’ itself is not specified (in the spoken, that is, naturally acquired registers) for number. For plural, this is only achieved by some determiners, numerals or quantifiers rather high in the structure, and by the opposition between un (= count) vs. du (= mass) in the singular:

(3) I drink—Je bois—Bebo

Table 4.1 shows possible continuations of the sequence ‘I drink’ in three languages. In grey, we see the complete underspecification of French nominal roots (and nPs) for number (3a), in parallel to languages like, for instance, Chinese. In contrast to Chinese, quantifying does not automatically lead to disambiguation or individuation (3e), as quantifiers like beaucoup (‘much’ / ‘many’) or peu (‘little’/ ‘few’) are also underspecified for count or mass and obligatorily combined with de, compatible with singular as well as with plural nominals (also under the scope of negation, where de shows up even with singular count nominals, see below). This observation and the sequence of beaucoup—de—[vɛ̃] leads to the assumption that, at least in French, Div° is always projected and overtly realized, also in mass nominals. The detailed adaptation of Figure 4.1 for French goes as represented in Figure 4.2 below: we assume that French roots come with a gender (for details, see Stark 2016, 138–139), but that there is no agree or probing operation between the root and n° (we take as morphological evidence the absence of word class or gender markers in French). Therefore, the root remains in situ, and the result is a highly defective nominal, actually only a property-denoting expression that can usually not occupy an argument position, not even under the scope of negation (*Je ne bois pas vin, intended: ‘I do not drink wine’; correct: Je ne bois pas de vin; *Je n’ ai pas acheté voitures, intended: ‘I did not buy cars’; correct: Je n’ ai pas acheté de voitures). The combination of this root with Div°, carrying no number feature in French and incapable of successful probing (cf. e.g., Mathieu 2009, 147 f., where the probe on his Num° probes unsuccessfully for number, the Modern French nominal not carrying a number feature), does not change things much. This is due to the absence of plural markers and the general assumption that functional heads without any morphological or semantic effect should not be assumed to exist (cf. Heycock and Zamparelli 2005). Above, however, we saw that there exists an element in French, contrary, for instance, to Spanish or English, which always realizes Div°, namely de. This de-element alone (having its own functional projection de-P in Ihsane 2008, 163, cf. also Shlonsky 2014) cannot in itself be analyzed as an explicit mass or non-individuation marker, as it is even found, in some colloquial varieties of French, after numerals (cf. Kayne 1977 citing Bauche 1951, J’ ai deux de bonnets, ‘I have two caps’; cf. Ihsane 2013, 4 f., deux vins or deux bonnets would then be the result of a normative deletion rule).

Table 4.1

Plural marking and mass-count specification of arguments in English, French and Spanish

Plural marking and mass-count specification





Unspecified for number




(some) wine

du [vɛ̃]




a (very good) wine

un [vɛ̃] (très bon)

un vino (muy rico)


Plural > count

(some very good) wine-s

des [vɛ̃] (très bons)

vino-s (muy ricos)


High quantity—unspecified for number/mass-count

beaucoup de [vɛ̃]


High quantity—mass

much wine


mucho vino


High quantity—count

many wine-s

plusieurs [vɛ̃]

mucho-s vino-s


Figure 4.2

Deriving indefinite nominals in French with internal AGREE operations

Now, and as shown in Figure 4.3 below, in the absence of explicit quantification (or individuation) via suitable elements in #°, an obligatory insertion of le/la/les ‘the’ takes place in this position (cf. Ihsane 2008, 163; Mathieu 2009, 148, for a similar idea; see also Borer 2005, 164). This happens, after probing with the root, in order to minimally realize gender and already valued number features of French indefinite nominals in the sense of Greenberg’s (1978) noun markers (note that this is the only locus of gender- and number-marking in indefinite nominals in Modern French). With le/la/les ‘the’ inserted in #° as a default—that is, as a mere number (and gender) morpheme and not as a generic or non-specific definite article (≠ D°; contra Gross 1967; Milner 1978; Kupferman 1979, 1994; Zamparelli 2008)—there is no explicit counting or quantification, potentially leading to individuation (cf. Borer 2005, 128). Subsequently, a post-syntactic operation called lowering (Embick and Noyer 2001)6 takes place in indefinite nominals,7 moving (the exponent of) a hierarchically higher head2 to a lower head1 being part of its complement. This yields de + le = du/de la and de + les = des.


Figure 4.3

Indefinite nominals in French (without probes) with lowering of #° into Div°

As represented in Figure 4.4 above, French quantified indefinite nominals are derived by the same structure, and depending on the element inserted in #°, semantic number is specified (deux, ‘two’, plusieurs, ‘several’, etc.), and sometimes even morphological number and gender (e.g., in agreeing quantifiers such as différent-ef-spl, ‘different’). This may, as a consequence, either yield a count, that is, individuated interpretation, or an interpretation for which individuation is left underspecified (beaucoup, ‘much/many’).


Figure 4.4

Indefinite nominals in French

This analysis accounts for the fact that, at least for French, the interaction described in Borer (2005) between the lexical element (NP) and #P is not enough to yield an unambiguous mass reading of the whole nominal, contrary to, for example, Chinese, English or Spanish. Note that Borer rejects DivP for bare mass nominals as well as #P for bare plurals (Borer 2005, 130), structures not available in French in argument position.8

The proposed analysis for French allows us to account for the underspecification of French bare nominals for number and of some French quantified nominals for individuation (e.g., with beaucoup), for the obligatory insertion of at least de with argument nominals, irrespectively of their mass or count interpretation, and for the observed correlation with a highly defective nominal morphology. It accounts also for the much discussed en-pronominalization facts (cf. Ihsane 2013; Shlonsky 2014); en pronominalizes DivP:


















‘I drink two wines.—I drink two.’




















‘I drink much wine/many wines.—I drink much/many.’















‘I drink wine.—I drink (sm).’
























‘I don’t drink wine(s).—I don’t drink (any).’

Standard Italian features a “partitive article”, too, albeit optionally and with slightly different scope properties (Zamparelli 2008; Cardinaletti and Giusti 2016a, 2016b; Giusti, this volume). As represented in Table 4.2 below, Italian nouns are marked overtly for plural, but the inventory of plural markers is highly syncretic. Except for, all plural endings can also be interpreted as or (;

Table 4.2

Italian noun classes

Italian noun classes

libr-olibr-i (m.) ‘book(s)’

man-oman-i (f.) ‘hand(s)’

bracci-obracci-a (m.- f.) ‘arm(s)’

cas-acas-e (f.) ‘home(s)’

poet-apoet-i (m.) ‘poet(s)’

can-ecan-i (m.) ‘dog(s)’

nott-enott-i (f.) ‘night(s)’

That means that vino in Italian is not a morphological component of vin-i or a constituent of a plural expression in DivP (it is not parallel to vino—vino-s in Spanish or wine—wine-s in English); it cannot incorporate into Div° in order to be pluralized. Rather, we assume that it is already specified for number on the level of nP, carrying a valued number feature (cf. Eichler 2012, 358 f.; Alexiadou 2004, 27; Alexiadou 2015). Like in Spanish (see below), Italian nominal roots are combined with overt exponents of noun classes and incorporate, after a successful agree operation for gender (and noun class) between a probe in n° and valued features of the root (which thus becomes mobile), into n°. However, nP is not the locus of “portioning out”. This can be done in two ways: assuming a number probe on Div°, agree with the expression in n° either triggers movement, and the whole expression is incorporated into Div° (see Figure 4.5), like in Spanish. Or, like in French, no movement takes place and the nominal, stuck in n°, needs a default element in Div°, namely, di as the first part of the Italian “partitive article” (see Figure 4.6).


Figure 4.5

Deriving indefinite nominals without “partitive articles” in Italian with internal agree operations


Figure 4.6

Deriving indefinite nominals with “partitive articles” in Italian with internal agree operations

It seems as if, at least in Standard Italian, no semantic or other substantial difference exists between the bare nominal and the nominal with a “partitive article” in the singular (indefinite reading; cf. Cardinaletti and Giusti 2016a), but this issue has to be further explored. Different Italian varieties, dialects as well as regional varieties, show different patterns of indefinite determination with much less optionality (cf. Cardinaletti and Giusti 2018; Giusti, this volume), so that the standard Italian pattern might also reflect some inconsistency due to standardization and koineization processes (cf. Stark 2007).

Applying the analysis to Spanish yields, of course, a somehow redundant derivation at first sight. This is shown in Figure 4.7 below.

Contrary to Borer (2005), we assume DivP to always be present in indefinite nominals, at least in Romance. In Spanish, due to transparent plural morphology (cf. (3)), the absence of plural (marking) or higher quantifiers or numerals results in a default mass reading.9

Again, we assume Spanish roots to be combined with overt exponents of noun classes and to incorporate, after a successful agree operation for gender between a probe in n° and valued features of the root (which thus becomes mobile), into n° (see Figure 4.8 for deriving Sp. gato-s). Then, they are combined with Div° coming with a gender probe and a valued number feature,10 realized by -s if plural, and incorporate into Div°. Further merge of #°, carrying a gender and number probe for agreeing quantifiers and numerals, may subsequently lead to the combination of the expression in Div° with a numeral (dos, ‘two’) or a quantifier, yielding an (unspecific) quantification with forms inflecting for gender and number (, ‘much’ or much-oM-sPL/-aF-sPL ‘many’). Note that the mass or count interpretation of the complex element vin-o in n°, vino in Figure 4.7 (third line), results from the form being clearly singular and not being combined with an explicit “counter” in #°: un vino would automatically yield a count interpretation (‘one special sort of wine’ or ‘a glass of wine’).


Figure 4.7

Indefinite nominals in Spanish


Figure 4.8

Deriving indefinite nominals in Spanish with internal agree operations

Now, if it is true that for Spanish, just like for English, the interaction of nP with #P is enough to yield a count vs. mass interpretation in bare nominals or quantified nominals, the comparative approach we choose here allows us to account for the typologically complementary distribution in Romance of “partitive articles” always including some variant of de and unambiguous plural marking, in parallel to Borer’s (2005) observation starting from Chierchia (1998) of the widespread complementary distribution of classifiers and plural morphemes in the languages of the world.

If our analyses are on the right track, they make a strong prediction about the cross-linguistic distribution of the indefinite Romance mass classifier (i.e., the traditionally wrongly labeled “partitive article”) containing de: this element is not to be expected in Romance languages or varieties with overt and unambiguous morphological exponents of plural in Div°, as is the case in Ibero-Romance. In the next section, we will discuss apparent counterevidence to this prediction in the history of Spanish.

3 The Case of Old Spanish

Different authors (e.g., Lapesa 1964; Cano 1992; Eberenz 2008) have claimed that Old Spanish featured (optional) “partitive articles” in the sense of indefinite mass classifiers comparable to those of French and Italian discussed in Section 2. Based on data obtained by an exhaustive string search in the Corpus Diacrónico del Español (Corde) and the Corpus del Español (CdE),11 we will show that this claim originates from an incorrect analysis of the data: all Old Spanish del-constituents are superset-denoting pps headed by a zero Q° (which, besides the superset pp, also governs a subset-denoting zero dp, cf. Cardinaletti and Giusti 2006).12 In other words, they are bare partitives in the sense of Hoeksema (1996, 15 f.) and Kornfilt (1996) and never feature “partitive articles”.13 Yet, besides regular strong definite dps, Old Spanish bare partitives can also contain weakly referential definite dps with a so-called representative object interpretation licensed by what is known as the kind-oriented mode of talk (Krifka et al. 1995). Such weakly referential definite dps, responsible for the wrong “partitive article” analysis of the Old Spanish data in older literature, are not to be confounded with short weak definites in the sense of Carlson and Sussman (2005) and Carlson et al. (2006). Although, due to reasons of space, we will focus on Old Spanish data, all analyses presented hold true of Old Portuguese as well. This can be evidenced by a survey of data from the Corpus do Português (cf. also Gerards 2020).14 Neither Old Spanish nor Old Portuguese are, thus, counterexamples to the prediction resulting from Section 2.2.

3.1 The Data

The aim of this section is to determine the morphosyntactic and semantic status of seemingly Modern Gallo- and Italo-Romance-like del-constituents in Old Spanish. To this aim, we performed an exploratory corpus search, for practical reasons restricted to one specific context, namely, del-constituents governed by the preposition con ‘with’.15 As represented in Table 4.3, we obtained 275 occurrences from 34 texts,16 out of which most are from the 13th-, 14th-, and 15th-century; six occurrences are from the first quarter of the 16th century, 1 is from the 17th century, and 1 from the 20th century.17

Table 4.3

Chronologic distribution of the 275 del-constituents governed by con from Corde and CdE

Chronological distribution


N texts

del (disc. new)



62.5 % (172/275)



18.9 % (52/275)



15.6 % (43/275)



2.2 % (6/275)



0.4 % (1/275)



0.4 % (1/275)



100 % (275/275)

Del-constituents are thus a phenomenon characteristic of the 13th century (cf. also Gerards 2020 for more details).

As represented in Table 4.4, out of the 275 del-constituents governed by con, 78 (= 28.4 %) could readily be identified as bare partitive pps governed by a zero Q°, as they contain a discourse-given definite superset dp referring back to a specific entity introduced cotextually by indefinite or bare nominals or given by accommodation of the type the car … the wheels. For the remaining 197 (= 71.6 %) occurrences, in contrast, an analysis as textual or situational definites à la Russell (1905), Christophersen (1939), Hawkins (1978), or Heim (1982) is difficult to construct:


destiénpren=la […]






‘Dissolve it with strong vinegar.’ (Gerardus Falconarius, 13th century)

The occurrence of vinagre fuerte ‘strong vinegar’ in (5) is the first mention of the substance in a recipe text, which, furthermore, comes without an initial list of ingredients. The reason why it is highly implausible to assume that del vinagre fuerte in (5) denotes an unspecified subset portion of a situationally unique or familiar superset portion of vinegar is the enormous diversity of lexemes contained in the 197 occurrences of type (5) without any immediate cotextual givenness (see Section 3.3). Such del-constituents, the only ones of interest in the remainder of this paper, are only attested in the data until the first quarter of the 16th century.

Table 4.4

Distribution of discourse-given and discourse-new con-governed del-constituents from Corde and CdE

Distribution of del-constituents



Total N

Total texts

78/275 (28.4 %)

197/275 (71.6 %)

275 (100 %)


At first sight, a plausible analysis of discourse-new del-constituents governed by con seems to be one in terms of Modern Gallo- or Italo-Romance “partitive articles” seen in Section 2.2: indefiniteness, mass interpretation. A closer investigation, however, shows that such an analysis is incorrect: both intratextually and intertextually discourse-new del-constituents like in (5) alternate under identical discourse-pragmatic conditions and in the same syntactic functions with bare nominals (6), but also with definitely marked nominals (7) (cf. also Eberenz 2008):18







‘Dissolve it with strong vinegar.’ (Modo de Meleçinar las Aves, 16th-century copy of (5))








‘Dissolve it with strong vinegar.’ (Dancus Rex. Esc. V.II.19, 13th century)

Like in example (5) (and like the bare nominal in (6)), the definite nominal el vinagre fuerte in (7) is discourse-new: it is the first mention of the substance, and the use of the definite article is not licensed by the availability of a situationally unique or familiar discourse referent. This strongly suggests that discourse-new del-constituents (5)—unlike “partitive articles” (see Section 2.2)—also contain a definite article and that they are, thus, pps with a zero Q°.19 Crucially, this is the morphosyntax of bare partitives (see introduction to Section 3).

Besides the chronological one, two more restrictions apply to discourse-new del-constituents governed by con (5): first, 195/197 (= 99.0 %) discourse-new del-constituents denote concrete referents.20 Second, in the data analyzed, almost all (187/197 = 94.9 %) such constituents are from technical prose, more precisely medical, veterinary, or culinary treatises.21 This bias is not due to overrepresentation of some ingredient nouns in technical prose, a potential objection raised by an anonymous reviewer: an exemplary string search for ⟨miel⟩ ‘honey’ in a relevant non-technical prose subportion of Corde (register: narrative prose until 1599) yielded 247 attestations, none of which was a discourse-new del-constituent and 93 of which were bare nouns in argument position.

Importantly, this ⟨miel⟩-string search also shows that unembedded discourse-new definites of type (7) display the same genre bias as discourse-new del-constituents: besides the 93 bare nominals, the 247 attestations of miel also contain 91 tokens of definite la miel ‘the honey’. Crucially, all of them instantiate one of Hawkins’ (1978) eight different usage types of definite articles, that is, are regularly unique and/or familiar strong definites (≠ (7)).22

Returning to the question addressed in this paper, that is, whether Old Spanish featured (optional) “partitive articles” in the sense of indefinite mass determiners as available in Modern French and Modern Italian, and summarizing the preceding observations based on data from Corde and CdE, we can state the following:

  1. Old Spanish featured optional discourse-new del-constituents for which co(n)textual uniqueness or familiarity is difficult to construct (5). Such del-constituents are, at first sight, reminiscent of “partitive articles”.

  2. In Old Spanish, and under identical discourse-pragmatic conditions, there is intratextual and intertextual variation between such discourse-new del-constituents and discourse-new nominals with the definite article only (7). This strongly suggests that discourse-new del-constituents in Old Spanish, differently from “partitive articles” (see Section 2.2), also contain a definite article and are, hence, pps with a zero Q°. This is the morphosyntax of bare partitives, not of “partitive articles”. The overt dp embedded in the pp of discourse-new del-constituents is superset-denoting.

  3. Both discourse-new del-constituents and discourse-new definite nominals display a strong genre-bias in our data, being almost exclusively attested in technical prose.

In view of (a)–(c), we need an adequate semantic analysis of the definite article in (5) and (7), as well as an explanation of the genre bias observed.

3.2 Definite Articles in Need of Explanation: Representative Object Interpretations in the Kind-Oriented Mode of Talk

We propose that discourse-new definites such as el vinagre fuerte in (7) and the same type of definite nominal contained in discourse-new del-constituents (5) are to be analyzed as definites with representative object interpretations (ROI) made available by what Krifka et al. (1995, 85–88) have dubbed kind-oriented mode of talk (KoM).23 The label representative object interpretation (ROI) designates in our understanding a weakly referential use of definite nominals, often complements of V° or P°.24 Such definites do not meet uniqueness or familiarity requirements at the object level and come with a “generic flavor” even though they occur with verbal predicates not licensing kind-denoting arguments. In this vein, the grizzly in (8) and el gorila ‘the gorilla’ in Spanish (9) are not generic in the sense of kind denotation, as the whole sentence does not make a straightforward predication about a property of ursus arctos horribilis or gorilla, respectively. Rather, in the roi-reading of interest here, the definites denote arbitrary yet prototypical instantiations of kinds, which were filmed or encountered. As will be shown below, such instantiations are inherently non-specific, the definite itself being semantically number-neutral:

(8) In Alaska, we filmed the grizzly. (Krifka et al. 1995, 78)



















‘Yesterday we had our first encounter with the gorilla.’ (Leonetti 1999, 873)

roi-definites as in (8) and (9) are under-researched and, therefore, not well understood. Yet, what we do know is that they are attested in many Indo-European and at least some non-Indo-European languages and that, truth-conditionally, they are equivalent to indefinites (Oosterhof 2006, 67).25

As pointed out above, roi-definites refer to (an) arbitrary prototypical exemplar(s) of the respective kind, as noted by Krifka et al. (1995) and confirmed, among others, by Mueller-Reichau (2013):26

[…] the object in the situation described is only relevant as a representative of the whole kind [and] a property can be projected from the object to the kind.

Krifka et al. 1995, 79

[an] object term […] function[s] as a kind term [and] reference to the kind is realised indirectly via reference to a representative of the kind.

Mueller-Reichau 2013, 93

The fact that roi-definites are about representative prototypical exemplars of kinds and not about kinds ‘on the whole’ can only mean that such definites are, in principle, subject to the [± specific]-distinction. In other words, they introduce variables bound by an existential quantifier (Oosterhof 2008, 55, 159–161). Yet, we claim that roi-definites, despite being subject to this distinction in principle, are inherently non-specific (see for the same claim, Koss Torkildsen 2002, 83 and, for closely related observations, Kupisch and Koops 2007). This claim is—we believe—in line with Krifka et al.’s (1995) and Mueller-Reichau’s (2013) observations. Support for inherent non-specificity of roi-definites comes from their hitherto unnoticed semantic number neutrality. Both versions of (10), which are about one or two gorillas, respectively, are semantically felicitous continuations of (9):27






de repente



espalda(/s) plateada(/s)






appear.pst.3sg/ 3pl

an/ two


‘When crossing the forest, there suddenly appeared an/two impressive silverback(/s).’

Finally, note that from a syntactic point of view, the only plausible locus where the definite article in roi-definites is located is D°—as this is where referentiality (be it weak or strong) is established.

We believe that discourse-new Old Spanish definites (7) and discourse-new del-constituents (5) are—or, in the case of (5), contain—roi-definites. The genre bias of our Old Spanish data (see Section 3.1) is perfectly compatible with Krifka et al.’s (1995) observations, whose hypothesis is that roi-definites are pragmatically conditioned phenomena sensitive to a varietal bias. Yet, two possible objections need to be addressed.28

First, rois in modern article languages—the exclusive empirical basis of the scarce literature on such definites—are generally count nouns (cf. (8)–(9)). This is not the case of the Old Spanish data (cf. (5), (7)). However, in the case of Old Spanish, we are dealing with a less grammaticalized article system than that of modern languages. Such article systems are known to display greater freedom of article use and/or non-use (cf. e.g., Carlier and Lamiroy 2014). In this vein, for Old French, a language closely related to Old Spanish, it has explicitly been argued that definite articles with mass nouns could be used with non-unique and non-familiar referents in order to signal discourse prominence (Epstein 2001; cf. also Epstein 1993, 1994).29 Our claim that the Old Spanish data involve roi-definites, thus, does not seem far-fetched at all: after all, in recipes, ingredients are clearly central discourse referents (for statistical support of the discursive importance of roi-definites in Old Spanish, see Gerards 2020). Drawing on Epstein (2001) also leads to the prediction that roi-definites become rarer once the definite article of a given language continues to grammaticalize. Again, this is confirmed by the Old Spanish data, in which discourse-new del-constituents are only attested until the first quarter of the 16th century (see Section 3.1).

Second, roi-definites in modern article languages, besides being count nouns, are generally morphologically singular only. Once more, this is not true of the Old Spanish data. Again, the diachronic argument sketched in the preceding paragraph is a reasonable counterargument against this objection. However, it is no longer the only one: experimental investigations on rois in Modern Dutch (Oosterhof 2006, 2008, 159–161), despite confirming that singular count nouns clearly get such readings most easily, reveal that, at least for some speakers, roi-readings seem to not be completely ruled out with bare plurals. From our point of view, this suggests that it is reasonable to propose that Old Spanish plural definites, too, are amenable to rois.

Summing up, an analysis of the Old Spanish data, that is, of both unembedded discourse-new definites (7) and the definites contained in discourse-new del-constituents (5), in terms of roi-definites is perfectly plausible: both are most typical of the 13th century, that is, of a less grammaticalized article system, instantiate a pragmatically-conditioned genre-biased usage type of the definite article, and vary intertextually and intratextually with bare nominals, as they are truth-conditionally equivalent to indefinites. Clearly though, having the same truth-conditions as indefinites does not mean that Old Spanish del-constituents featuring roi-definites are semantically (let alone syntactically) identical to Modern Gallo- and Italo-Romance indefinite mass classifiers (“partitive articles”). Old Spanish del-constituents with roi-definites but not “partitive articles” involve superset reference to prototypical instantiations of kinds. Only Old Spanish del-constituents with roi-definites but not “partitive articles” (see Section 2.2) involve (weakly) referential definite articles located in D°. Furthermore, Modern Romance “partitive articles”, differently from Old Spanish del-constituents, do not involve prepositions and, differently from roi-definites, are not number neutral. In short, Old Spanish del-constituents featuring roi-definites are bare partitives, and do not involve “partitive articles”.

In the following section, we will address a possible alternative analysis of the Old Spanish data in terms of short weak definites. We will show that such an analysis is clearly inferior to one in terms of roi-definites.

3.3 The Old Spanish Data Are Not Short Weak Definites30

Another class of weak referentials besides roi-definites are so-called short weak definites (Carlson and Sussman 2005; Carlson et al. 2006; Aguilar-Guevara and Zwarts 2013).31 Short weak definites are definites that share a number of properties with roi-definites, among others semantic number neutrality, non-uniqueness, non-familiarity, and some fuzzy “generic flavor”. In this vein, in (11), Lola could have taken several potentially different, previously unfamiliar trains:

(11) Lola took the train from Amsterdam to Nijmegen. (Aguilar-Guevara and Zwarts 2013, 34)

Inherent semantic number neutrality, non-uniqueness, and non-familiarity seem to approximate short weak definites to roi-definites. Upon closer scrutiny, however, both turn out to be different classes of weak referentials. In the remainder of this section, this will be shown by means of three morphosyntactic, lexical, and semantic properties of short weak definites that are found neither in roi-definites nor in our Old Spanish data: defectiveness with regard to introducing discourse referents, strong lexical restrictions, and restriction to one morphological number only.32

It is commonplace that short weak definites (swds) are bad at introducing discourse referents (Carlson et al. 2006; Aguilar-Guevara and Zwarts 2013; Aguilar-Guevara 2014, among many others). Pronominal resumption of potential short weak definites, at least with stage-level predicates, obligatorily triggers a strong, “regular” definite reading of the nominal (sd), that is, a regularly unique and/or familiar interpretation:33

(12) Lola listened to the radioi until she fell asleep. She turned iti off when she woke up in the middle of the night. (? swd/sd fine) (Aguilar-Guevara and Zwarts 2013, 35)

roi-definites, in turn, can perfectly function as antecedents of pronominal expressions also with stage-level predicates, without the roi-reading being lost:

(13) In Alaska, we filmed the grizzlyi. Often, we would even be able to observe iti /themi34 interact with itsi/theiri young.

In the Old Spanish data, 101/197 (= 51.3 %) discourse-new del-constituents serve as antecedents for either clitic or strong pronominal expressions with stage-level predicates:35













por tal que










‘Take dill and cook it with water and put it [the water] in front of them so that they drink of it [of the water].’ (Moamín, ca. 1250)

Summing up, the Old Spanish data clearly align with roi-definites and not with short weak definites with regard to the introduction of discourse referents.

Short weak definites are also subject to strong lexical restrictions (Carlson et al. 2006; Aguilar-Guevara and Zwarts 2013; Aguilar-Guevara 2014; Schwarz 2014). This holds true with regard to the noun itself, even for near synonyms (15a–b), the governing verb (15c–d), and the governing preposition (15e–f):

(15) a. I went to the hospital. (swd fine)

b. I went to the clinic. (no swd)

c. Sally checked the calendar. (swd fine)

d. Sally tore the calendar. (no swd)

e. Kenneth is at the store. (swd fine)

f. Kenneth is behind the store. (no swd)

([a], [b] Aguilar-Guevara 2014, 153; [c]–[f] Carlson and Sussman 2005, 76)

The 197 discourse-new Old Spanish del-constituents in the data contain 37 different governing verbs and 42 different nouns. Altogether, 85 different verb + con + noun combinations are attested in the data. Again, the Old Spanish data thus clearly align with roi-definites and not with short weak definites.

Finally, short weak definites are morphologically defective. With a given nominal lexeme, they can either be morphologically singular ((16a) vs. (16b)) or morphologically plural ((16c) vs. (16d)), but never both (Aguilar-Guevara and Zwarts 2011, 181; Espinal and Cyrino 2017b, 130):

(16) a. Sally listened to the radio. (swd fine)

b. Sally listened to the radios. (sd only)

c. Lola went to the mountain. (sd only)

d. Lola went to the mountains. (swd fine)

(Aguilar-Guevara and Zwarts 2011, 181; [a–b] adapted)

rois, in turn—though preferably singular—are, according to previous research (see Section 3.2), not categorically ruled out in the plural and, more importantly, have, for a given nominal lexeme, never been argued to be restricted to only one morphological number. Again, the type of Old Spanish del-constituents studied in this paper align with roi-definites rather than with short weak definites. As proof of this, consider the data in (17) and (18), taken from Gerards (2020), which not only analyzes del-constituents governed by con ‘with’ but also in other syntactic functions:








‘Take cow suet.’ (Moamín, ca. 1250)














‘Take bird suets that suit (lit.: suits) them.’ (Moamín, ca. 1250)

Both (17) and (18)—two examples from the same text—feature discourse-new del-constituents containing the lexical head noun seuo ‘suet’. Yet, seuo is morphologically singular in (17), while it is plural in (18). In addition, note how (18) may be further proof of our claim that the Old Spanish del-constituents under study in this paper—differently from “partitive articles” (see Section 2.2)—feature an empty Q° (see note 14): in (18), the del-constituent is the subject of the modifying relative clause que les conuiene ‘which suit (lit. suits) them’, where conuiene is 3rd person singular. Yet, neither seuos ‘suets’ nor its PP-complement de las aues ‘of the birds’ is morphologically singular. Hence, it is reasonable to assume that 3rd person singular conuiene agrees with an empty Q-head specified for singular and which embeds the entire complex del-constituent.

Summing up, the Old Spanish data discussed are (i) perfectly able to introduce discourse referents and to function as antecedents of coreferential expressions even with stage-level predicates, (ii) lack strong lexical restrictions, and (iii) are not restricted to one morphological number for a given nominal lexeme. They thus clearly align with roi-definites, and not with short weak definites.

4 Conclusion

We have shown in this contribution that the typologically marked existence of an indefinite mass determiner (“partitive article”) in many Gallo-Romance and Italo-Romance varieties and also in standard French (and to a lesser extent in standard Italian) can be correlated empirically with the (non-)availability of word class markers and agglutinative plural morphemes in Romance languages. A formal analysis of the internal structure of indefinite nominals (based on Borer 2005, plus some minimalist assumptions on agree and Distributed Morphology mechanisms) has shown that de is the minimal expression of Div°, in complementary distribution with overt and unambiguous plural morphemes. This analysis makes the prediction that languages like Ibero-Romance varieties with such a plural-s should not possess indefinite mass determiners (“partitive articles”). The prediction, contrary to older claims (Lapesa 1964; Cano 1992; Eberenz 2008), is borne out: the discourse-new Old Spanish del-constituents discussed in this paper, frequent only in the 13th century, turned out to be bare partitives, that is, qps with a zero Q° that contain pps. These pps, in turn, contain weakly referential definite superset dps with representative object interpretations licensed by the kind-oriented mode of talk. In Modern Spanish, these del-constituents are no longer available.36 The reason for this, we believe, is that in the 13th century, the system of nominal determination was much less grammaticalized than in Modern Spanish. Definites in Old Spanish were able to signal discourse prominence more easily than in Modern Spanish, even in the absence of uniqueness and/or familiarity of a discourse referent.


We would like to express our gratitude to two anonymous reviewers and the audience of Going Romance 30 at Goethe-University Frankfurt for many helpful comments on previous versions of this paper. Additionally, we would like to thank the Swiss National Science Foundation (SNSF) and the URPP Language and Space (University of Zurich) for financial support. Usual disclaimers apply.


We only claim complementary distribution of “partitive articles” and unambiguous plural markers in Romance (complementary across languages and varieties) and the grammaticalization of the former due to the loss of the latter (cf. Schurr, this volume, on additional grammaticalization facts). There is no universal generalization intended—except for the claim that every language needs some classification device to encode the conceptual distinction between mass and count.


The equivalent plural form (e.g., Fr. des, It. dei) is different from the singular in its syntactic distribution, semantic function and dialectal distribution (cf. Ihsane 2008; Zamparelli 2008; Garzonio and Poletto 2014; Cardinaletti and Giusti 2016a, 2018; Stark 2016, 132). In what follows, we will focus on the singular but include the plural in our morphosyntactic analyses.


In what follows, we will use the masculine singular form as a representative of the entire paradigm.


Note that there are some Italian varieties where the indefinite mass interpretation typical of du/del-nominals also holds for nominals with the definite article not meeting the standard criteria for definiteness (Kupisch and Koops 2007; Cardinaletti and Giusti 2016b, 2018; Leonetti 2019; Giusti, this volume).


See Pomino (2016, 122–127) for the proposal to locate liaison [-z] in French plural nominals originally in Div°, claiming however a phrasal clitic status for [-z], which may also be realized rather high in the structure, under D° [lezami] (les amis ‘the friends’).


Embick and Noyer (1999, 2001) introduce this operation to account for English verbal inflection in the past: with syntax giving abundant evidence that there is no v°-to-T° movement in English (Pollock 1989), temporal features of T° have to be lowered to v° in order to explain forms like she laugh-ed: [TP T° … [vPv° …]]—[TP … [vP … [v° v° + T°] …]].


For definite nominals and the English determiner the, Borer (2005, 164, example (8a)) speculates about a possible merge in Div° and subsequent movement towards #°, which could account for the absence of de in definite French nominals. Note that *l’ un vin is ungrammatical in French (contrary to the one wine in English), which might be evidence for an initial insertion of le/la/les in Div° in definite nominals (which inherit their mass-count distinction from their antecedent, cf. Borer 2005, 166; recall that there is no plural morpheme on nP available in French to occupy Div°: les vins, [levɛ̃]). We leave the question of an exact derivation of Romance definite nominals and a potential relation to case (cf. e.g., Giusti 2015) open for future research.


Furthermore, Borer (2005, 164) seems to allow for projected, but not overtly realized heads: “[…] that plural or mass interpretations could emerge, in principle, without #P, or alternatively, with ⟨e⟩# but without any range assigned to it […].”


Cf. also Borer (2005, 107, note 18): “As to the possibility that languages project distinct functional structures, I assume, pending evidence suggesting otherwise, that the inventory and interpretation of functional structure is identical across all languages, and that to the extent that the output differs, it must be due to the mode of range assignment selected in a particular structure by a particular language.”


This is different from Mathieu (2009), who does not take gender probing into account.


Corde ( contains data from the beginning of documentation until 1974 (236,709,914 tokens; 34,155 texts) from Spain, Latin America, and the Philippines. The Corpus del Español ( covers Peninsular and Latin American varieties. It comprises 101,311,682 tokens from 1200–1999 (13,926 texts). Cf. Schurr, this volume, for another corpus study of Old Spanish.


Throughout the rest of this paper, the neutral label del-constituent comprises both singular masculine del + N and feminine de la + N, as well as the plural forms de los + N and de las + N.


Note that this use of bare partitive is different from that of Chierchia (1998) and Le Bruyn (2010), who use the label bare partitive for referring to “partitive articles”, that is, D-elements, as discussed in Section 2. Disagreeing with Kupferman (1994), we take as evidence for bare partitives to be headed by a zero Q° two facts. First, with bare partitives in subject function, the verbal predicate clearly agrees with zero Q° (cf. Seržant 2012, for the same observation on partitive genitives in Ancient Greek). This is illustrated by data from Palatian ((i); see also Old Spanish (18) below), a Rhine Franconian variety of German in which bare partitives are particularly frequent (cf. Glaser 1993, for issues of general frequency, but not for agreement facts; Strobel and Glaser, this volume, on partitive markers in some Germanic varieties; cf. also Martin, Carvalho and Alexiadou, this volume, on (dis-)agreement facts between subject and predicate):
















‘There is [still] (a portion) of the beans on the plate.’

(lit.: ‘Of the beans [still] lies on the plate.’)

Second, the pronominal expression sie ‘it’ in Standard German (ii) is clearly coreferential with zero Q°, not with the overt DP der Milch ‘the milk’:
























‘I took of the milk and then drank it.’


Corpus do Português ( covers European and Brazilian varieties. It comprises 45,606,959 tokens from 1200–1999 (55,493 texts).


Restriction to one context was necessary as Corde is not lemmatized and data collection, therefore, was extremely complicated and time-consuming. According to the literature (cf. e.g., Lapesa 1964, 79; Sánchez Lancis 2009), del-constituents are attested as direct objects, as objects of P°, and as subjects of unaccusative constructions, that is, as internal arguments. Gerards 2020 shows that what will be said in this section about del-constituents governed by con ‘with’ holds across the board for del-constituents in direct object and subject function and explains their syntactic distribution in terms of case theory.


The initial number was considerably higher. However, CdE proved to be unreliable due to scanning mistakes in the process of corpus compilation (e.g., pages with two columns scanned as if they contained one). In order to warrant a scrupulous analysis, all data from this latter corpus had to be checked manually. This procedure led to the exclusion of many false positives.


These proportions would be even more biased towards the 13th century if we counted as belonging to the 13th century those attestations from later centuries that are (often verbatim) copies of 13th-century ones (among which the only datum from the 20th century). Note furthermore that both Corde and CdE contain up to three times more tokens for the 15th and up to six times more tokens for the 16th century than for the 13th and 14th century. Obviously, this, too, means that the proportions reflected in Table 4.3 would be even more biased towards the 13th century if the corpora were to contain an equal number of tokens per century.


For reasons elaborated on below, such uses of the definite article are not mentioned in grammars of Old Spanish (cf. e.g., Ortiz Ciscomani 2009).


The use of the definite article in (7) is to be distinguished from that of some Modern Italian varieties in which the definite article is the default morphosyntactic means to encode indefinite mass interpretations (cf. Cardinaletti and Giusti 2016b, 2018, and note 4; Giusti, this volume). In this context, note that Kupisch and Koops (2007, 194, their note 5) sustain that such uses in Modern Italian varieties are the result of a grammaticalization process of definites with representative object interpretations in the kind-oriented mode of talk (see Section 3.2).


From our point of view, the only two attestations with abstract nouns are doubtful as to whether they really instantiate bare partitives.


The percentage could be argued to even be higher, as 6 out of the 10 occurrences from genres other than technical prose are found in clearly instructive passages. This adds further robustness to the genre bias (for the reason of the bias, see Section 3.2).


This is, we believe, the reason why uses of the definite article such as the one in (7) are not mentioned in grammars of Old Spanish: these are mostly based on literary and poetic text genres. For a discussion of the problematic text selection underlying grammars of Old Spanish, see, for instance, Kabatek (2005) and references therein.


As opposed to the default object-oriented mode of talk (Krifka et al. 1995, 87). The first scholar to (briefly) note the existence of this class of dps was, to the best of our knowledge, Bally (41965 [1932], 89–90).


For the discussion of rois in the context of weak referentiality, see Pires de Oliveira (2013, 28–29). For an overview of many different types of weak referentiality, see the contributions in Aguilar-Guevara, Le Bruyn, and Zwarts (2014).


For roi-definites in Indo-European languages, see the references in this section. For Arabic, see Jaber (2014).


See also Leonetti (1999, 872 f.), Oosterhof (2008), and Pelletier (2010, 6).


For further evidence, see also the discussion of example (13) in Section 3.3.


We thank Anna Kocher and an anonymous reviewer of a previous version of this paper for these observations.


Possibly, the pragmatic notion of prominence can be formalized in terms of salience, which some have claimed to be the only universal meaning of definite articles (see, most recently, Von Heusinger 2013).


This section is the fruit of numerous informal discussions over the past three years, in the course of which we were repeatedly challenged to position ourselves with regard to whether short weak definites and roi-definites are one and the same class of nominals. We particularly thank an anonymous reviewer of a previous version of this paper for his/her insightful comments.


The term weak definite is originally due to Poesio (1994), who was, however, mostly concerned with what is now known as long weak definites. For a recent comprehensive typology of weak definites, see Espinal and Cyrino (2017a).


These properties of short weak definites, among others, are why Carlson et al. (2006, 2013) and Schwarz (2014) analyze short weak definites as a special type of incorporation. For different accounts, see Aguilar-Guevara and Zwarts (2013), Beyssade (2013), Corblin (2013), Aguilar-Guevara (2014), and Zwarts (2014).


As a reviewer notes, pronominal resumption of swds is fine with individual or kind-level predicates (Lola listened to the radioi. Iti is her favorite medium for listening to music).


Note how the grammaticality of both a singular and a plural pronominal expression resuming a singular count roi-definite in (13) adds further support to the semantic number neutrality of roi-definites (see Section 3.2).


In order to warrant comparable results, we only took into consideration the two clauses immediately following the one containing the del-constituent. Note that the actual number of del-constituents introducing discourse referents would have even been higher had we taken into account the 14 cases of resumption of del-constituents by means of a lexical dp. However, we decided not to include such data in our count as, so far, the defectiveness of short weak definites with regard to introducing discourse referents has exclusively been discussed and tested for pronominal expressions.


Whereas ROI-definites not embedded in del-constituents are still available, at least with singular count nouns (9).


  • Aguilar-Guevara, Ana. 2014. Weak Definites: Semantics, Lexicon and Pragmatics. Dissertation Series 360. Utrecht: Landelijke Onderzoekschool Taalkunde.

  • Aguilar-Guevara, Ana, Bert Le Bruyn, and Joost Zwarts, eds. 2014. Weak Referentiality, Amsterdam, Philadelphia: John Benjamins.

  • Aguilar-Guevara, Ana, and Joost Zwarts. 2011. “Weak definites and Reference to Kinds.” In Proceedings of SALT 20, edited by Nan Li and David Lutz, 179–196. Cornell University, Ithaca, NY: CLC Publications.

  • Aguilar-Guevara, Ana, and Joost Zwarts. 2013. “Weak Definites Refer to Kinds.” Recherches Linguistiques de Vincennes 1, no. 42: 33–60.

  • Alexiadou, Artemis. 2004. “Inflection Class, Gender and DP Internal Structure.” In Exploration in Nominal Inflection, edited by Gereon Müller et al., 21–50. Berlin: de Gruyter.

  • Alexiadou, Artemis. 2015. “Deriving Fake Mass Nouns.” In Proceedings of NELS 45, edited by Thuy Bui, and Deniz Özyıldız, 11–24, Amherst, MA: GLSA.

  • Bally, Charles.41965 [1932]. Linguistique générale et linguistique française. Bern: Francke.

  • Bauche, Henri. 1951. Le langage populaire. Paris: Payot.

  • Beyssade, Claire. 2013. “Back to Uniqueness Presupposition.” Recherches Linguistiques de Vincennes 42: 127–137. doi: 10.4000/rlv.2190.

  • Borer, Hagit. 2005. In Name Only. Structuring Sense, vol. 1. Oxford: Oxford University Press.

  • Bossong, Georg. 2016. “Classifications.” In The Oxford Guide to the Romance Languages, edited by Adam Ledgeway and Martin Maiden, 63–72. Oxford: Oxford University Press.

  • Cano, Rafael. 1992. “La sintaxis española en la época del Descubrimiento.” In Estudios filológicos en homenaje a Eugenio de Bustos Tovar, edited by José A. Bartol Hernández, Juan F. García Santos and Javier de Santiago Guervós, 183–197. Salamanca: Ediciones Universidad de Salamanca.

  • Cardinaletti, Anna, and Giuliana Giusti. 2006. “The Syntax of Quantified Phrases and Quantitative Clitics.” In Blackwell Companion to Syntax, edited by Martin Everaert and Henk Van Riemsdijk, vol. 5, 23–93. Oxford: Blackwell. doi: 10.1002/9780470996591.ch71.

  • Cardinaletti, Anna, and Giuliana Giusti. 2016a. “The Syntax of the Italian Determiner dei.” Lingua 181: 58–80.

  • Cardinaletti, Anna, and Giuliana Giusti. 2016b. “Indefinite Determiners across Italo-Romance Varieties: Optionality and Variation.” Talk given at the International Workshop “Partitivity and Language Contact”, Zurich, 25–26 November 2016.

  • Cardinaletti, Anna, and Giuliana Giusti. 2018. “Indefinite Determiners: Variation and Optionality in Italo-Romance.” In Advances in Italian Dialectology, edited by Diego Pescarini and Roberta D’Alessandro, 135–161. Language and Linguistics EBooks Online. doi: 10.1163/9789004354395_008.

  • Carlier, Anne. 2007. “From Preposition to Article. The Grammaticalization of the French Partitive.” Studies in Language 31, no.1: 1–49. doi: 10.1075/sl.31.1.02car.

  • Carlier, Anne, and Béatrice Lamiroy. 2014. “The Grammaticalization of the Prepositional Partitive in Romance.” In Partitive Cases and Related Categories, edited by Silvia Luraghi and Tuomas Huumo, 477–520. Berlin: de Gruyter.

  • Carlson, Gregory, Natalie Klein, Whitney Gegg-Harrison, and Michael Tanenhaus. 2013. “Weak Definites as a Form of Definiteness.” Recherches Linguistiques de Vincennes 1, no. 42: 11–32.

  • Carlson, Gregory, and Rachel Sussman. 2005. “Seemingly Indefinite Definites.” In Linguistic Evidence: Empirical, Theoretical, and Computational Perspectives, edited by Stephan Kepsar and Marga Reis, 71–86. Berlin: de Gruyter. doi: 10.1515/9783110197549.71.

  • Carlson, Gregory, Rachel Sussman, Natalie Klein, and Michael Tanenhaus. 2006. “Weak Definite Noun Phrases.” In Proceedings of NELS 36, edited by Christopher Davis, Amy R. Deal and Youri Zabbal, 179–196. Amherst, MA: GLSA.

  • Chierchia, Gennaro. 1998. “Reference to Kinds Across Languages.” Natural Language Semantics 6, no. 4: 339–405. doi:

  • Christophersen, Paul. 1939. The Articles. A Study of their Theory and Use in English. Copenhagen: Munksgaard.

  • Corblin, Francis. 2013. “Weak Definites as Bound Relational Definites”. Recherche Linguistiques de Vincennes 42: 91–122. doi: 10.4000/rlv.2184.

  • Cowper, Elizabeth, and Daniel C. Hall. 2012. “Aspects of Individuation.” In Count and Mass across Languages, edited by Diane Massam, 27–53. Oxford: Oxford University Press. doi: 10.1093/acprof:oso/9780199654277.003.0003.

  • Crispim, Maria de Lourdes. 1996. “Artigos definidos e demonstrativos num corpus do séc. XIII.” In Actas do XII Encontro Nacional da Associação Portuguesa de Linguística. Vol. II: Linguística Histórica. História da Língua, edited by Ivo Castro, 79–90. Lisboa: Associação Portuguesa de Linguística.

  • Delfitto, Denis, and Jan Schroten. 1991. “Bare Plurals and the Number Affix in DP.” Probus 3, no. 2: 155–185.

  • Dobrovie-Sorin, Carmen. This volume. “Negation, des-Indefinites in French and Bare Nouns across Languages.”

  • Eberenz, Rolf. 2008. “Ninguno quiere del agua turbia beber: sobre construcciones partitivas y su representación en algunos géneros textuales del español preclásico.” In Sintaxis histórica del español y cambio lingüístico. Nuevas perspectivas desde las Tradiciones Discursivas, edited by Johannes Kabatek, 151–172. Madrid, Frankfurt a. M.: Iberoamericana, Vervuert.

  • Eichler, Nadine. 2012. Code-Switching bei bilingual aufwachsenden Kindern. Tübingen: Narr.

  • Embick, David, and Rolf Noyer. 1999. “Locality in Post-Syntactic Operations.” MIT Working Papers in Linguistics 34: 265–317.

  • Embick, David, and Rolf Noyer. 2001. “Movement after Syntax.” Linguistic Inquiry 32, no. 4: 555–595.

  • Epstein, Richard. 1993. “The Definite Article: Early Stages of Development.” In Historical Linguistics 1991, edited by Jaap Van Marle, 111–134. Amsterdam, Philadelphia: John Benjamins.

  • Epstein, Richard. 1994. “The Development of the Definite Article in French.” In Perspectives on Grammaticalization, edited by William Pagliuca, 63–80. Amsterdam, Philadelphia: John Benjamins.

  • Epstein, Richard. 2001. “The Meaning of Definite Articles in Cross-Linguistic Perspective.” In Cognition in Language Use, Selected Papers from the 7th International Pragmatics Conference, edited by Enikő Németh, 174–189. Antwerp: International Pragmatics Association.

  • Espinal, M. Teresa, and Sonia Cyrino. 2017a. “The Definite Article in Romance Expletives and Long Weak Definites.” Glossa 2, no. 1: 1–26.

  • Espinal, M. Teresa, and Sonia Cyrino. 2017b. “On Weak Definites and their Contribution to Event Kinds.” In Boundaries, Phases and Interfaces: Case Studies in Honor of Violeta Demonte, edited by Olga Fernández-Soriano, Elena Castroviejo-Miróa and Isabel Pérez-Jiménez, 130–150. Amsterdam, Philadelphia: John Benjamins.

  • Garzonio, Jacopo, and Cecilia Poletto. 2014. “On Partitives in the Northern Italian Dialects.” Talk given at the International Exploratory Workshop “Partitivity in Romance and Beyond”, Zurich, 11–13 December 2014.

  • Gerards, David Paul. 2020. Bare Partitives in Old Spanish and Old Portuguese. PhD Diss. University of Zurich.

  • Giusti, Giuliana. 2015. Nominal Syntax at the Interfaces: A Comparative Analysis of Languages with Articles. Newcastle upon Tyne: Cambridge Scholars Publishing.

  • Giusti, Giuliana. This volume. “A Protocol for Indefinite Determiners in Italian and Italo-Romance.”

  • Givón, Talmy. 1981. “On the Development of the Numeral ‘One’ as an Indefinite Marker.” Folia Linguistica Historica 2, no. 1: 35–53.

  • Glaser, Elvira. 1993. “Syntaktische Strategien zum Ausdruck von Indefinitheit und Partitivität im Deutschen (Standardsprache und Dialekt).” In Dialektsyntax, edited by Werner Abraham and Josef Bayer, 99–116. Wiesbaden: VS Verlag für Sozialwissenschaften.

  • Greenberg, Joseph H. 1978. “How does a Language Acquire Gender Markers?” In Universals of Human Language, Word Structure, edited by Joseph H. Greenberg, Charles A. Ferguson and Edith A. Moravcsik, vol. 3, 47–82. Stanford, CA: Stanford University Press.

  • Gross, Maurice. 1967. “Sur une règle de cacophonie.” Langages 7: 105–119.

  • Halle, Morris, and Alec Marantz. 1993. “Distributed Morphology and the Pieces of Inflection.” In The View from Building 20, edited by Kenneth Hale and Samuel J. Keyser, 111–176. Cambridge, MA: MIT Press.

  • Hawkins, John A. 1978. Definiteness and Indefiniteness: A Study in Reference and Grammaticality Prediction. London: Croom Helm.

  • Heim, Irene R. 1982. “The Semantics of Definite and Indefinite Noun Phrases.” PhD Diss. University of Massachusetts Amherst.

  • Herslund, Michael. 1998. “Le français, langue à classificateurs?” In La ligne claire. De la linguistique à la grammaire. Mélanges offerts à Marc Wilmet à l’ occasion de son 60e anniversaire, edited by Annick Englebert et al., 65–73. Louvain-La-Neuve, Paris: Duculot. doi: 10.3917/dbu.pierr.1998.01.0065.

  • Heycock, Caroline, and Roberto Zamparelli. 2005. “Friends and Colleagues: Coordination, Plurality, and the Structure of DP.” Natural Language Semantics 13, no. 3: 201–270.

  • Hoeksema, Jacob. 1996. “Introduction.” In Partitives: Studies on the Syntax and Semantics of Partitive and Related Constructions, edited by Jacob Hoeksema, 1–24. Berlin: de Gruyter.

  • Ihsane, Tabea. 2008. The Layered DP in French. Form and Meaning of French Indefinites. Amsterdam, Philadelphia: John Benjamins.

  • Ihsane, Tabea. 2013. “EN Pronominalisation in French and the Structure of Nominal Expressions.” Syntax 16, no. 3: 217–249.

  • Jaber, Aziz. 2014. On Genericity and Definiteness in Modern Standard Arabic. Ph.D. dissertation. University of North Carolina at Chapel Hill.

  • Kabatek, Johannes. 2005. “Las tradiciones discursivas del español medieval: historia de textos e historia de la lengua.” Iberoromania 62: 28–43.

  • Kayne, Richard. 1977. Syntaxe du français. Le cycle transformationnel. Paris: Editions du Seuil.

  • Kornfilt, Jaklin. 1996. “Naked Partitive Phrases in Turkish.” In Partitives, edited by Jakob Hoeksema, 107–142. Berlin, New York: de Gruyter.

  • Krifka, Manfred, Gregory Carlson, Gennaro Chierchia, Godehard Link, Francis. J. Pelletier, and Alice Ter Meulen. 1995. “Genericity: An Introduction.” In The Generic Book, edited by Gregory Carlson and Francis J. Pelletier, 1–124. Chicago: University of Chicago Press.

  • Kupferman, Lucien. 1979. “L’ article partitif existe-t-il?” Le français moderne 47: 1–16.

  • Kupferman, Lucien. 1994. “Du: un autre indéfini?” Faits de Langue 4: 195–203.

  • Kupisch, Tanja, and Christian Koops. 2007. “The Definite Article in Non-Specific Direct Object Noun Phrases. Comparing French and Italian.” In Nominal Determination. Typology, Context Constraints and Historical Emergence, edited by Elisabeth Stark, Elisabeth Leiss and Werner Abraham, 189–213. Amsterdam, Philadelphia: John Benjamins.

  • Lapesa, Rafael. 1964 [2000]. “Los casos latinos: restos sintácticos y sustitutos en español.” In Estudios de morfosintaxis histórica del español, edited by Rafael Cano Aguilar and M. Teresa Echenique Elizondo, vol. 1, 73–122. Madrid: Gredos.

  • Le Bruyn, Bert. 2010. “Indefinite Articles and Beyond.” PhD Diss. University of Utrecht.

  • Leonetti, Manuel. 1999. “El artículo.” In Gramática descriptiva de la lengua española. 1: Sintaxis básica de las clases de palabras, edited by Ignacio Bosque and Violeta Demonte, 787–883. Madrid: Espasa Calpe.

  • Leonetti, Manuel. 2019. “On Weak Readings of Definite DPs.” In Proceedings of the IX Nereus International Workshop “Morphosyntactic and Semantic Aspects of the DP in Romance and Beyond”, edited by Natascha Pomino, Arbeitspapier 131, 1–26. Konstanz, Universität Konstanz: Fachbereich Sprachwissenschaft.

  • Link, Godehard. 1983. “The Logical Analysis of Plural and Mass Terms: A Lattice Theoretical Approach.” In Meaning, Use and Interpretation of Language, edited by Rainer Bäuerle, Christoph Schwarze and Arnim Von Stechow, 302–323. Berlin: de Gruyter.

  • Lowenstamm, Jean. 2007. “On Little n, √, and Types of Nouns.” In The Sounds of Silence: Empty Elements in Syntax and Phonology, edited by Jutta Hartmann, Veronika Hegedus and Henk Van Riemsdjik, 105–144. Amsterdam: Elsevier.

  • Lowenstamm, Jean. 2012. “Feminine and Gender, or why the Feminine Profile of French Nouns has Nothing to do with Gender.” In Linguistic Inspirations. Edmund Gussmann in Memoriam, edited by Eugeniusz Cyran, Henryk Kardela and Bogdan Szymanek, 371–406. Lublin: Wydawnictwo Katolicki Uniwersytet Lubelski.

  • Martin, Fabienne, Janayna Carvalho and Artemis Alexiadou. This volume. “Predicates of Personal Taste and Pancake Sentences in Brazilian Portuguese and French.”

  • Mathieu, Eric. 2009. “From Local Blocking to Cyclic AGREE. The Role and Meaning of Determiners in the History of French.” In Determiners. Universals and Variation, edited by Jila Ghomeshi, Ileana Paula and Martina Wiltschko, 124–157. Amsterdam, Philadelphia: John Benjamins. doi: 10.1075/la.147.04mat.

  • Mathieu, Eric. 2014. “Many a Plural.” In Weak Referentiality, edited by Ana Aguilar-Guevara, Bert Le Bruyn and Joost Zwarts, 151–181. Amsterdam, Philadelphia: John Benjamins.

  • Mattos e Silva, Rosa V. 2008. O português arcaico. Uma aproximação. vol. 2. Lisboa: Imprensa Nacional—Casa da Moeda.

  • Milner, Jean-Claude. 1978. De la syntaxe à l’ interprétation. Paris: Editions de Seuil.

  • Mueller-Reichau, Olav. 2013. Sorting the World. On the Relevance of the Kind/Object-Distinction to Referential Semantics. Berlin: de Gruyter.

  • Mulder, Walter de, and Anne Carlier. 2011. “The Grammaticalization of Definite Articles.” In The Oxford Handbook of Grammaticalization, edited by Bernd Heine and Heiko Narrog, 522–534. Oxford: Oxford University Press. doi: 10.1093/oxfordhb/9780199586783.013.0042.

  • Oosterhof, Albert. 2006. “Dutch Bare Plurals, Faded Partitives and Subject-Object Asymmetry.” Belgian Journal of Linguistics 19: 59–91.

  • Oosterhof, Albert. 2008. The Semantics of Genericity in Dutch and Related Languages. Amsterdam, Philadelphia: John Benjamins.

  • Ortiz Ciscomani, Rosa M. 2009. “La creación y generalización del artículo definido.” In Sintaxis histórica de la lengua española. Segunda parte: la frase nominal, edited by Concepción Company, vol. 1, 271–386. México D.F.: Fondo de Cultura Económica/UNAM.

  • Pelletier, Francis J. 2012. “Lexical Nouns are Both +MASS and +COUNT, but they are either +MASS nor +COUNT.” In Count and Mass across Languages, edited by Diane Massam, 9–26. Oxford: Oxford University Press.

  • Pelletier, Francis J., ed. 2010. “Generics: A Philosophical Introduction.” In Kinds, Things, and Stuff. Oxford: Oxford University Press.

  • Picallo, Carme. 2008. “Gender and Number in Romance.” Lingue e linguaggio 7, no. 1: 47–66.

  • Pires de Oliveira, Roberta. 2013. “Weak (In)definitess [sic] and Referentiality.” ABRALIN 12, no. 1: 11–37.

  • Poesio, Massimo. 1994. “Weak Definites.” In Proceedings of SALT IV, edited by Mandy Harvey and Lynn Santelmann, 282–292. Ithaca, NY: Cornell University. doi: 10.3765/salt.v0i0.2465.

  • Pollock, Jean-Yves. 1989. “Verb Movement, Universal Grammar, and the Structure of IP.” Linguistic Inquiry 20, no. 3: 365–424.

  • Pomino, Natascha. 2012. “Partial or Complete Lack of Plural Agreement. The Role of Morphology.” In Inflection and word-formation in Romance languages, edited by Sascha Gaglia and Marc-Olivier Hinzelin, 201–229. Amsterdam, Philadelphia: John Benjamins.

  • Pomino, Natascha. 2016. “On the Clitic Status of the Romance Plural Marker.” In Proceedings of the VII Nereus International Workshop “Clitic Doubling and other Issues of the Syntax/Semantic Interface in Romance DPs”, edited by Susann Fischer and Mario Navarro, Arbeitspapier 128, 105–130. Konstanz: Universität Konstanz.

  • Russell, Bertrand. 1905. “On Denoting.” Mind 14: 479–493.

  • Sánchez Lancis, Carlos. 2009. “La evolución del partitivo indefinido en español: ¿un proceso de gramaticalización preposicional?” Diachronic Linguistics, edited by Joan Rafel Cufí, 407–448. Girona: Documenta Universitaria.

  • Schurr, Hagay. This volume. “Bound To Be? Bare and Partitive-Marked Noun Phrases in Romance Languages and the Emergence of Prominence-Conditioned Patterns.”

  • Schwarz, Florian. 2014. “How Weak and How Definite are Weak Definites?” In Weak Referentiality, edited by Ana Aguilar-Guevara, Bert Le Bruyn and Joost Zwarts, 213–235. Amsterdam, Philadelphia: John Benjamins.

  • Seržant, Ilja A. 2012. “Morphosyntactic Properties of the Partitive Genitive in the Subject Position in Ancient Greek.” Indogermanische Forschungen 117: 187–204. doi: 10.1515/indo.2012.117.2012.187.

  • Shlonsky, Ur. 2014. “One Note on Quantitative Ne-Cliticization.” Talk given at the International Exploratory Workshop “Partitivity in Romance and Beyond”, Zurich, 11–13 December 2014.

  • Stark, Elisabeth. 2007. “Gender, Number, and Indefinite Articles. About the ‘Typological Inconsistency’ of Italian.” In Nominal Determination. Typology, Context Constraints, and Historical Emergence, edited by Elisabeth Stark, Werner Abraham and Elisabeth Leiss, 49–71. Amsterdam, Philadelphia: John Benjamins.

  • Stark, Elisabeth. 2008a. “Typological Correlations in Nominal Determination in Romance.” In Essays on Nominal Determination. From Morphology to Discourse Management, edited by Henrik Høeg Müller and Alex Klinge, 45–61. Amsterdam, Philadelphia: John Benjamins.

  • Stark, Elisabeth. 2008b. “The Role of the Plural System in Romance.” In The Paradox of Grammatical Change. Perspectives from Romance, edited by Ulrich Detges and Richard Waltereit, 57–84. Amsterdam, Philadelphia: John Benjamins.

  • Stark, Elisabeth. 2016. “Nominal morphology and Semantics—Where’s Gender (and ‘Partitive Articles’) in Gallo-Romance?” In Proceedings of the VII Nereus International Workshop “Clitic Doubling and other issues of the syntax/semantic interface in Romance DPs”, edited by Susann Fischer and Mario Navarro, Arbeitspapier 128, 131–149. Konstanz: Universität Konstanz Fachbereich Sprachwissenschaft.

  • Stark, Elisabeth, and David Paul Gerards. This volume. ‘ “Partitive Articles’ in Aosta Valley Francoprovençal–Old Questions and New Data.”

  • Strobel, Thomas, and Elvira Glaser. This volume. “The Rise and Fall of Partitive Markers in Some Germanic Varieties.”

  • Von Heusinger, Klaus. 2013. “The Salience Theory of Definiteness.” In Perspectives on Linguistic Pragmatics, edited by Alessandro Capone, Francesco Lo Piparo, and Marco Carapezza, 349–374. Berlin: Springer. doi: 10.1007/978-3-319-01014-4_14.

  • Von Koss Torkildsen, Janne. 2002. “Representativity—Weak Interpretations of Definite Plural Noun Phrases.” PhD Diss. University of Oslo.

  • Wiese, Heike. 2012. “Collectives in the Intersection of Mass and Count Nouns: A Cross-linguistic Account.” In Count and Mass across Languages, edited by Diane Massam, 54–74. Oxford: Oxford University Press.

  • Zamparelli, Roberto. 2008. “Dei ex machina. A Note on Plural/Mass Indefinite Determiners.” Studia Linguistica 62, no. 3: 301–327. doi: 10.1111/j.1467–9582.2008.00149.x.

  • Zwarts, Joost. 2014. “Functional Frames in the Interpretation of Weak Definites.” In Weak Referentiality, edited by Ana Aguilar-Guevara, Bert Le Bruyn and Joost Zwarts, 265–285. Amsterdam, Philadelphia: John Benjamins.

  • Collapse
  • Expand