PhiN 47/2009: 1

Elisabeth Stark (Zürich)

Romance restrictive relative clauses between macrovariation and universal structures*

Romance restrictive relative clauses show considerable variation as to the 'relative pronouns' used in actual speech. Besides the complex normative systems of French, Italian and Spanish, "substandard" varieties of these languages present constructions based on relative particles with or without resumptive pronouns. After a short description and typological presentation of relative clauses in general and in the three Romance languages in question, both in standard and "substandard" varieties, we will discuss traditional and variationist descriptions of "substandard" relative clauses. They usually identify (at least) two different systems, one typologically rather inconsistent system of relative elements in the standard and one consistent system of particle-based "substandard" relative clauses, which would amount to say that Romance speakers use two different systems, maybe according to extralinguistic factors like education or situation. Empirical data show, however, that the ascription of non-normative relative constructions to "substandard" varieties in Romance fail to describe in a realistic way what is going on in actual usage: the general "degree of accessibility" of the antecedent and the internal syntactic complexity of the relative clause provoke in some rare cases the cross-linguistically well motivated and widespread use of relative particles with or without resumptive pronouns rather than (standard) relative pronouns. This variation is most probably due to medial factors and occurs almost exclusively in phonic, i.e. not planned and revisable communication.

1. Introduction

1.1 Ever since the seminal book of Christian Lehmann (Lehmann 1984), the overall structure of relative clauses together with their history and typology seem, at least for Indo-European languages, to be a well-established and well-understood fact in descriptive linguistics. Additionally, universal tendencies influencing their morphosyntactic make-up have already been described in the important typological article by Keenan / Comrie (1977). And detailed syntactic descriptions like the one in Cinque (31991) for Italian are available in the theoretical syntactic literature at least since Chomsky (1977). However, when it comes to the description and classification of the different relative constructions attested in single languages or language families, there still remain some controversial points. How is the relative construction in (1) to be analysed and classified when comparing it to Modern Standard French?

(1) La santé que j'ai jouie jusqu'ici très vigoureuse (Montaigne, Essais1)
'the very strong health which I enjoyed up to now'

The element introducing the relative clause here is que 'that', identical to the French conjunction que and invariable as to gender, number or case marking. The verb governing this relative element is jouir de 'to enjoy', subcategorising normally an oblique object introduced by the preposition de, which leaves us with a syntactic gap in the relative clause or better which leaves the syntactic relation between the relative element and its verbal governor unmarked.2 Such structures are far from being archaisms, they are, on the contrary, widely attested in Modern French – yet they usually are ascribed to so-called "substandard"3 varieties (cf. Schafroth 1993):4

(2) un carambolage qu'on vous a déjà beaucoup parlé (Blanche-Benveniste 1990: 323)5
'a collision which you have already been told much about'

PhiN 47/2009: 2

Just like in example (1), the finite verb of the relative clause parler de 'to speak about' subcategorises an oblique complement introduced by the preposition de, which would lead to dont as the 'correct' relative element, instead of simple que in example (2), leaving unexpressed the mark of the verbal government.

This paper will be mainly concerned with arguing for an adequate interpretation of apparent variational facts in the structure of Romance relative clauses, irrespective of whether they are labelled "standard" or "substandard" or "colloquial" in the literature,6 in order to identify the triggering factors of the respective constructions and to uncover the underlying syntactic and pragmatic regularities of their formation and distribution in actual speech. It will almost exclusively focus on restrictive relative clauses, i.e. adjuncts inside NPs, not on appositive relative clauses, constituents usually considered to be outside the modified NP and which do not serve to restrict the reference of the individuals in question.7 It is organised as follows: sections 1.2. and 1.3. present a short description and typological presentation of relative clauses in general (1.2) and in three Romance languages in particular (French, Italian and Spanish), both in standard and "substandard" varieties, with special focus on the differing systems of relative clause initial elements ("pronouns" vs. "particles", section 1.3). Section 2 discusses the facts presented in 1 and arrives at the preliminary conclusion that we seemingly deal with a typologically rather inconsistent system of relative elements in the standard varieties as opposed to the consistent system of particle-based "substandard" relative clauses. Section 3 tests this variationist hypothesis against empirical background (quantity and, mainly, quality of "substandard" relative constructions in a few actual corpora of spoken French, Italian and Spanish) and gives some counterevidence. Section 4 finally presents an alternative and more realistic description of Romance "substandard" relative constructions.

1.2 We will start from Lehmann's (1984: 47) definition of a relative clause:

Als Relativsätze sollten wir […] alle untergeordneten Sätze und satzartigen Konstruktionen betrachten, die ein Nominal modifizieren, welches in ihnen selbst eine semantische Rolle hat.

That means that any relative clause or functionally similar construction is the expression of three different operations: subordination, modification of the (nominal) antecedent and indication of the syntactic function and semantic role the wh-moved relative element has inside the relative clause ("Leerstellenbildung", following Lehmann's terminology). In the world's languages, we find essentially two elements explicitly introducing relative clauses, normally combining a pronominal and a complementiser function:8 relative pronouns and relative particles. We will speak of a relative pronoun when at least two of the three constitutive functions are assumed by the same clause-initial element, and of a relative particle when only one function is expressed by it, the others being expressed by other elements inside the relative clause (the so-called resumptive pronouns) or not being expressed at all. The French examples in (3) illustrate all three possibilities, of which only the last one is "correct" from the point of view of normative grammar:

(3) A. le livre que je t'ai parlé (relative particle que without any resumptive pronoun)
B. le livre que je t'en ai parlé (relative particle que + resumptive 'pronoun' en)
C. le livre dont je t'ai parlé (relative 'pronoun' dont – but see the discussion below)
'the book about which I have talked to you'

Typological research of the last decades has shown quite clearcut correlations in the distribution of these different structures with other syntactic properties of the respective languages. Highly inflectional languages with a relatively free word order tend to have relative pronouns (type C), but also make use of type B (e.g. Bulgarian, Modern Greek and also Classical and Vulgar Latin, cf. Catoira 2008), whereas languages which almost completely lack inflectional morphology and have a relatively fixed word order tend to favor relative particles without resumptive pronouns, i.e. are normally of type A (e.g. English, Scandinavian languages, Yiddish, cf. Prince 1990; Kurzowá 1981: 96ff.). The Romance languages, which, compared to Latin, lost a good part of their (nominal) inflectional morphology and tend to have a relatively fixed word order (French more than, e.g., Spanish or Rumanian), seem nevertheless to belong more to the first than the second group: they are considered to have relative pronouns (type C), but types B and even A, typical of the second group of languages, are also widely attested – with the restriction that those latter types seem to belong to "substandard" or "popular" varieties (cf. Downing 1978: 385f., Kurzowá 1981: 80ff.). The question is thus: How do we deal with this seemingly "typological inconsistency"? Do Romance languages constitute a third, a mixed type? Do we have to abstract from the attested variation and concentrate only on standard relative clauses, which would then lead to the assumption of counterexamples to the above mentioned typological classification (especially for French, with relative pronouns despite its reduced nominal morphology and rather fixed word order)? Or do we have to abstract from the highly normalised standards and consider only the attested "substandard" relative constructions as the "real" ones, consistent with the typological findings?

PhiN 47/2009: 3

In what follows, we will try to show that neither of these rather radical steps has to be taken if two conditions are met: first, we have to do away with some traditional, yet erroneous denominations and analyses of relative constructions in Romance, and second, we have to question the sometimes misleading variationist classifications of the different types of relative clauses. Before doing so, we will present a short overview of the system of 'relative pronouns' in French, Italian and Spanish.

1.3 The following table gives an overlook over the most important relative 'pronouns' in Modern Standard French, Italian and (European) Spanish:9

French Italian Spanish
il quale
el cual
(4) la femme qui parle
(4') la donna che parla
(4'') la mujer que habla
(5) il n'y a rien qui m'intéresse
(5') non c'è niente che mi interessa
(5'') no hay nada que me interese
(6) la femme que je vois
(6') la donna che vedo
(6'') la mujer que veo
(7) la femme dont la fille est morte
(7') la donna la cui figlia è morta
(7'') la mujer cuya hija está muerta
(8) la femme dont / de qui/ de laquelle j'ai parlé
(8') la donna di cui / della quale ho parlato
(8'') la mujer de quien / de la que hablé / de la cual hablé
(9) il n'y a rien de quoi je veux parler
(9') non c'è niente di cui voglio parlare
(9'') no hay nada de lo que / de lo cual quiera hablar

Table 1: 'Relative pronouns' in some Romance Standard languages10

PhiN 47/2009: 4

This table presents a rather complex system, with variable features relevant for the single elements. Traditionally so-called 'relative pronouns' like Fr. qui/que, It. che or Sp. que do not carry any gender or number information and seem not to be sensitive to semantic features like [animate] or [abstract] in subject and (direct) object function (cf. examples (4)–(6)), with only French making a morphological distinction between the two functions). Possessive relations ('genitive', cf. examples (7)) are marked with the invariable Fr. dont and It. cui, while Sp. cuyo can be marked for gender and number of the possessed object. Simplifying a bit, especially for Fr. dont,11 which is also possible in these last contexts, indirect and prepositional objects (cf. examples (8) and (9)) finally exhibit the greatest variation and allow in many cases the composite element lequel/il quale/el cual (French and Italian yet not for [abstract] referents, something where French needs a special element after prepositions, i.e. quoi) and Fr. qui and Sp. quien after prepositions only for [human] referents).

As for "substandard" Romance relative clauses, the literature refers to three main types (cf. examples (10) to (17) below, and Cinque 31991: 511ff., Blanche-Benveniste 1990: 330, Schafroth 1993: 24ff., Fiorentino 1999: 23):12

Type (A): only qui/que / che/ que:
(4) la femme qui parle
(4') la donna che parla
(4'') la mujer que habla
(5) il n'y a rien qui m'intéresse
(5') non c'è niente che mi interessa
(5'') no hay nada que me interese
(6) la femme que je vois
(6') la donna che vedo
(6'') la mujer que veo
(10) la femme que j'ai parlé
(10') la donna che ho parlato
(10'') ?la mujer que hablé
(11) il n'y a rien que je veux parler
(11') non c'è niente che voglio parlare
(11''') no hay nada que quiera hablar

Type (B): que + resumptive pronoun:
(12) la femme que j'en ai parlé
(12') la donna che ne ho parlato
(14) el hombre que no sabes cuando lo viste (Fernández Soriano, foll. Brucart 1999, 406)13
(13) il n'y a rien que j'en veux parler
(13') non c'è niente che ne voglio parlare

Type (C): relative pronoun and resumptive pronoun ("pleonastic"):
(15) la femme dont j'en ai parlé
(15') la donna di cui ne ho parlato
(17) recuerdo las murallas de la ciudad […] las cuales las tiraron para hacer casas de pisos
(16) il n'y a rien dont j'en veux parler
(16') non c'è niente di cui ne voglio parlare

PhiN 47/2009: 5

Type C is without any doubt a case of hypercorrectism (cf. Stempel 1964: 224, Godard 1989: 57, Gadet 1995, Bellini 1998: 41, 44, Fiorentino 1999: 16) and scarcely attested, and will therefore not be taken into consideration in what follows. The other two types, however, are attested phenomena, at least in corpora of spontaneous speech and most frequently in prepositional complements. For both French and Italian, it has further been repeatedly noted in the literature that relative elements like dont, cui or lequel/il quale, despite being always enumerated as relative pronouns on a par with qui/que and che in normative and descriptive grammars, appear very rarely in these same corpora (cf. Deulofeu 1981: 140, Blanche-Benveniste 1990: 329, Bernini 1991: 183, Alfonzetti 2002; Spanish el cual and cuyo show a similar distribution).

The common characteristics of type A and B is the overall presence of (a variant of) che (qui/que or que) as the clause-initial element, while the syntactic function of the relativised element either remains unexpressed (type A) or gets expressed by a resumptive element (type B). The only function che/qui/que or que assume in these constructions is the one of subordination – a function that they assume in other cases as well, being the unmarked Romance subjunctions or complementisers par excellence.14

2. Discussion: How to deal with the variation?

As we just saw in the concise analyses of the "substandard" picture, types A and B represent at first sight a considerable "simplification" of the rather complex paradigms of relative elements in table 1. Qui/que, che and que assume the function of subordination, leaving the indication of any semantic or morphological features of their antecedent (modification) or of the syntactic function of the relativised element inside the relative clause ("Leerstellenbildung") either unexpressed or expressed by another element (resumptive pronouns). This leads directly to the conclusion that they are relative particles (cf. Blanche-Benveniste 1990: 326ff., 330) rather than relative pronouns, at last homomorphous (or identical, cf. for Italian, Cinque 31991: 463–466, Fiorentino 1999 and Berruto 92000: 134) to the unmarked complementisers of the respective languages. Following this argumentation, we arrive at a typologically quite different system of relative clauses in the "substandard" varieties of French, Italian and Spanish, when compared with the respective standards, where the existence of a complex paradigm of relative pronouns is assumed (cf. table 1 above).

Now, we have already seen in the short typological overview in section 1.2 that the assumption of Romance belonging to the first group of languages having mainly relative pronouns does not fit very well with their actual characteristics, such as a considerably reduced (nominal) inflectional morphology and a rather fixed word order, at least clearly attested in French. And in fact, when we liberate ourselves from traditional grammatical descriptions of Romance relative clauses, we discover the existence of relative particles also in the paradigm of standard relative clauses (cf. table 1): Fr. qui/que does not carry any information about the semantic features [animate] or [abstract] of its antecedent and is not inflected for gender, number or case (cf. footnote 8 and footnote 13), just like It. che or Sp. que for subjects and direct objects. The same holds true for French dont, except that this element is marked as a pronoun replacing prepositional phrases introduced by de – but in manifold syntactic functions such as the genitive (la femme dont la fille est morte) or oblique objects (la femme dont j'ai parlé).15 Elements marked for gender and number or semantic features of their referents and the syntactic function the relativised element holds in the relative clause – and thereby behaving like pronouns – only appear after prepositions in Romance standard restrictive relative clauses (Fr. qui, quoi, lequel / It. il quale / Sp. el cual). In this perspective, Sp. cuyo, inflected for gender and number and syntactic function ('genitive'), and uninflected It. cui and Sp. que, which can be combined with prepositions, contrary to Fr. qui/que and It. che, are also to be considered as relative pronouns.

PhiN 47/2009: 6

This results then in a – rather unsatisfying – picture of a typologically 'mixed' system for the respective standards in the three Romance languages discussed here,16 and of a rather consistent system for "substandard varieties" based on the complementiser and/or relative particle que/che/que:

La grande différence entre les deux usages [standard and substandard, ES] vient de ce que les usages non normatifs [= substandard varieties, ES] étendent l'emploi de la particule à des points du système où l'usage normatif [= standard, ES] emploie des groupes formés de préposition + pronom (Blanche-Benveniste 1990: 330).

This would equally amount to say that we deal with two different grammars, one for the standard and another one for substandard Romance varieties.17 Besides several principled problems with such a decision (cf. the discussion in Berruto 1985 for Italian, Deulofeu 1981: 166f. for French), concerning e.g. issues of language acquisition, we would have to show at least, in this latter case, the conditions of use the two different systems have and how speakers learn to respect them. That is, in case we assume two (or even more) different varieties as being different subsystems of a language (cf. the critical discussion of this assumption in Dufter/Stark 2002 for any but the diatopic dimension in French), stable correlations between linguistic and extralinguistic factors have to be attested in authentic linguistic material, correlations which permit to consider the different relative clause constructions as functional equivalents of each other.

3. Facts: What about linguistic reality?

3.1 Several studies on relative clauses in Romance report the relative sparsity of the "substandard" constructions.18 In French, not more than 2% of all relative clauses in huge corpora of authentic spoken French occur as one of the three types described in section 1.3. above (cf. Schafroth 1995: 367f., Gadet 1995: 141). These types seem to be equally rare in Italian (cf. Bernini 1991: 185, Schafroth 1993: 278f., Fiorentino 1999: 169, Alfonzetti 2002: 161, 167). Our own study (cf. Stark 2004) has shown that in the LABLITA corpus (cf. Cresti 2000; 7h 11' 54'', 59.845 words), only 8 out of 616 (= 1.3%) relative clauses with che as the relativised element are of the "substandard types". And finally, different studies on Spanish "substandard relative clauses" reveal the same tendency: e.g. in 18 hours of informal conversations in Carácas (cf. D'Introno 1992), only 4.8% of all relative clauses are of the "substandard" types; only 7.4% of over 1000 relative clauses in a corpus of Chilean Spanish (cf. Silva-Corvalán 1996: 386), and only 2.2% of all attested relative clauses in a corpus of Sevilla (cf. Butler 1992).

3.2 The attested Italian occurrences of the substandard relative clauses seem all to be very well motivated by linguistic facts (cf. Stark 2004). That means, structural reasons rather than situational factors lead to their occurrence, something which has already been discussed by Ariel 1999.

Type B, i.e. a relative particle plus a resumptive pronoun in the original syntactic position of the relativised element, occurs in constructions with considerable 'structural distance' or a lot of lexical material between the antecedent and the relativised element (cf. also Suñer 1998: 335). This seems to be related to processing difficulties in spontaneous and phonic, i.e. not written, speech: the more lexical material we find between them, the more likely are 'simple' clause initial and quite adjacent complementisers like che and resumptive pronouns inside the relative clause, not only in Romance languages:19

PhiN 47/2009: 7

(18a) […] è quella che [...] da Malaga / ci vuole quasi / più d' un ora / pe' andà' su //
'it is this one which, coming from Malaga, you need almost more than one hour to go to' (cf. Cresti 2000, vol. 2: 192)

(18b) *STE: que tu es obligé d' en trouver un autre //
'and then, you know, afterwards, its really annoying, when you find yourself at a place where you can't go, which [where] you have to look for another one'
(cf. Cresti/Moneglia 2005)

The Italian standard version of this relative construction would have been something like (18a'):

(18a') è quella per andare sulla quale da Malaga ci vuole quasi più di una ora…

where we see clearly the increased distance between the antecedent and the relativised element.

Furthermore, resumptive pronouns tend to occur preferably in appositive relative clauses (cf. Blanche-Benveniste 1990: 333 for French, Bernini 1991: 179 and Alfonzetti 2002: 69–96, 161f. for Italian, Brucart 1999: 405ff. for Spanish), which we excluded initially from the discussion here. Since appositive relative clauses are considered to be rather independent from their antecedent and as not belonging to the NP constituent (cf. Jones 1996: 500), resumptive pronouns in these syntactic contexts behave quite similar to 'normal' anaphoric elements referring back to their antecedent in the text:

(19a) allora / senti / per l' acquisto della prima casa / ci sono vari documenti da presentare [...] innanzitutto / la domanda // da [/] da riempire e presentare all' INPDAP [...] che / naturalmente / quella è uguale //
'so please listen: for the acquisition of the first house, you have to present several documents, first of all the application to fill in and present to INPDAP, which, of course, this one is the same for everybody'
(cf. Cresti, 2000, vol. 2: 155)

A structurally similar example is the French one in (19.b):

(19b) *ROG : […] par exemple / nous avons eu des belles choses // […] qu' on appelait les grands bureaux de Lens // que grâce à notre maire de [/] de Lens […] il y a pu il y a pu l' avoir pour le franc symbolique […]
'for example, we had nice things, which we called the big offices of Lens, which, thank to our lordmayor from Lens, he could have there, he could have them for just a symbolic franc'
(cf. Cresti/Moneglia 2005)

It is important to underline here that the structural factors triggering type B or resumptive pronouns in relative clauses are cross-linguistically more or less identical and stable phenomena (cf. Ariel 1999), so that the occurrence of type B in a language is not necessarily and even very improbably the result of language change or contact (this against Catoira 2008) – we seem to deal here with almost universal encoding strategies.

Type A seems to appear in prepositional adjuncts with a clear semantics of the antecedent noun, making superfluous any precise indication of the syntactic and semantic function of the relativised element inside the relative clause:

(20a) io / l' unica posizione che posso stà'/ è qui sotto // perché / se sto sopra...
'I, the only position I can stand is the one below, because when I am above'
(cf. Cresti 2000, vol. 2: 304)

(20b) Nos dimos una vuelta por el retiro, un tarde otoñal, creo que ese parque tiene algo que me engancha, creo que es el único sitio que puedo estar libremente con mi amiga [...]
'We had a walk in the Retiro, one evening in autumn, I believe that this park has something which attracts me a lot, I believe that it is the only place which [where] I can stay freely with my girl-friend [...]'
(cf. ; published by Godofredo, 19:56, november 24, 2008, UN DÍA CUALQUIERA)

PhiN 47/2009: 8

(20c) *ROG: il y a beaucoup de choses qui nous appartiennent / mais // que le mineur il n' a pas accès hein
'there are a lot of things which belong to us but which [to which] the miners are not admitted'
(cf. Cresti/Moneglia 2005)

And finally the "Accessibility Hierarchy" (cf. Keenan / Comrie 1977 and 1979, cf. also Fiorentino 1999: 38ff., 167–170, 177ff., Alfonzetti 2002: 161–168 for Italian, Tarallo 1989: 259f. for Brazilian Portuguese) comes into play whenever we try to explain the distributional properties of type A and type B:

SU > DO > IO > OBL > GEN > OCOMP (Keenan / Comrie 1977: 66)

The lower a syntactic function in this hierarchy (examples (20) can be considered a case of "other complements"), the more difficult its accessibility, whereas Subjects (SU) and Direct Objects (DO) can easily be processed. Usually, the probability for resumptive pronouns, i.e. type B, increases with positions at the right end of the scale (cf. Keenan / Comrie 1977: 88f.). But there are also other types of relative constructions imaginable and de facto attested with the different syntactic functions on the scale. For example, as SU and DO are highly accessible, they do not need much coding and can easily be referred to by simple relative particles (see below). The overall generalisation is that it is quite widespread in the languages of the world to find one strategy for SU and DO and another one for IO, OBL etc. (cf. also above, section 2, our findings for the 'mixed' standard systems opposing SU and DO to prepositional constituents of different types). Thus, the Accessibility Hierarchy, being an implicational hierarchy, permits predictions concerning the distribution of relative clause types and explains at the same time a good part of the attested "substandard constructions" in the corpora.

This is also consistent with the finding that another relatively frequent case of type A in Romance is the absence of any indication of the syntactic function of the relativised element with oblique complements (OBL) of certain verbs like parler de / parlare di / hablar de, avoir besoin de / aver bisogno di, se souvenir de / ridordarsi di (cf. Blanche-Benveniste 1990: 330f., Bellini 1998: 56ff., Fiorentino 1999: 37ff.):

(21a) le fotografie / che io / parlavo / con / Gioacchino Calabrò / erano / &fo [/]
'the photographs which I was talking [about] with Gioacchino Calabrò were photographs…'
(cf. Cresti 2000, vol. 2: 379)

(21b) una empresa como el Real Madrid, con todos sus trabajadores, la maquinaria que hablaba ayer eh – Butragueño.
'a company like Real Madrid, with all its workers, the machinery [about] which Butragueño talked yesterday, eh'
(cf. Corpus del español; España Oral: ADEB007; )

(21c) *EST: ma jupe noire là // que Marie-Laure m' a dit / ouais / elle est sympa / avec les poches tout en bas
'my black skirt [about] which Marie-Laure said [talked] to me / it is nice / with the pockets below'
(cf. Cresti/Moneglia 2005)

3.3 Finally, there is no clear-cut correlation in empirical studies between the occurrence of type A and type B and specific extralinguistic factors, justifying, e.g., their classification as sociolinguistic indicators or markers of certain sociolects or styles (cf. Gadet 1995: 153 for French, Alfonzetti 2002: 162 for Italian,20 Stark 2004 for Italian, Cortés Rodríguez 1990: 442, Suñer 1998 for Spanish, Blanche-Benveniste 1990: 324f. for French, Spanish and Portuguese, contrary to the assumptions in Koch / Oesterreicher 1990: 153, 192f.). The 8 attested "substandard" constructions in the Italian LABLITA-corpus mentioned above (cf. Stark 2004) are in fact equally distributed over situations of informal and very formal speech, so that we cannot attribute them to a dimension of "communicative proximity" vs. "communicative distance" (cf. Koch / Oesterreicher 1990, especially 192 or 225, assuming "substandard" relative clause types in Italian and Spanish partially as typical for "communicative proximity").

PhiN 47/2009: 9

All this has to be taken into account when looking for an adequate description and interpretation of the variety observable in Romance (and other languages') relative constructions.

4. Conclusion and outlook

The typological and morphosyntactic analysis of Romance relative constructions, presented in section 1 and discussed in section 2, revealed a considerable inconsistency for the respective standard systems, which are a 'mix' of relative particles for subjects and directs objects and relative pronouns for different kinds of prepositional complements. This finding is consistent, though, with Keenan / Comrie's (1977, 1979) "Accessibility Hierarchy", predicting different relativising strategies available for different points of the scale, opposing frequently SU/DO to other complements (cf. also the findings in Cristofaro / Giacalone Ramat 2007: 12ff. for Galician and Italian). However, it is inconsistent with typological generalisations (cf. Kurzowá 1981), which consider relative pronouns to occur usually in highly inflectional languages with a relatively free word order, like Latin, but much less so in the Romance languages (especially French), with little (nominal) inflectional morphology and relatively fixed word order. There, we would expect relative particles with different "repair strategies" in order to indicate the syntactic function of the relativised element in the relative clause ("Leerstellenbildung"), either leaving a gap (type A) or indicating it by a resumptive pronoun (type B). This is exactly the system "substandard" relative clause types show in Romance, cf. section 1.3. above – which leads at first sight to the conclusion that the respective standards and not the "substandard varieties" represent the deviation from the normal morphosyntactic coding processes in Romance, being artificial and heterogeneous. And in fact, this has been repeatedly stated in the literature (cf. Lehmann 1984: 392, Blanche-Benveniste 1990, Fiorentino 1999: 34ff., Schafroth 1993: 5, 17f.), also beyond the limits of Romance linguistics (cf. Fleischer 2005 for German and German dialects):21

En résumé, en étudiant le standard, on croit étudier une langue et l'on étudie dans beaucoup de secteurs un bricolage de la langue [...] On en arrive même à parler de système du français standard, à le constituer en référence pour les autres études, alors que tout porte à y voir un bricolage fonctionnel d'un ensemble de normes de réalisation sociolinguistiques. (Deulofeu 1981: 165)

This represents, by the way, a major and yet rarely discussed obstacle to typological studies based on the comparison of standard grammars of different languages, leading at least sometimes to superficial and erroneous conclusions – as they might deal with inexistent results of normalisation processes rather than with functionally motivated structures.22

However, if we take into consideration the fact that the seemingly more consistent "substandard" systems with their different relative clause types are scarcely attested and do not correlate in a systematic way with extralinguistic factors such as social class, educational status, style, situational (in)formality etc., we better might think of giving up the whole opposition between standard and "substandard" or non-standard relative constructions. Rather than with two, we are in fact dealing with one system of relativising strategies in Romance languages, depending on universal tendencies: it is the "general degree of accessibility" (cf. Ariel 1999: 256), i.e. the mental accessibility of the antecedent, together with the ease of processing its syntactic role inside the relative clause, which provoke "substandard" type B and A respectively:23

PhiN 47/2009: 10

Relative clauses where the head is highly accessible when the relativized position is processed take gaps, whereas relative clauses which maintain a relatively low degree of accessibility of the head when the relativized position is processed take resumptive pronouns. (Ariel 1999: 217)

In this perspective, we are dealing only with one – functionally rather consistent – system in Romance relative clauses: speakers choose constructions according to processing difficulties and following "default conventions":

Default conventions specify what choice speakers will consider most automatically, what they would tend to choose unless strong indications against it exist. (Ariel 1999: 256f.)

This system is based on three degrees of accessibility: relative particles without resumptive pronouns (type A) are used when the accessibility of the antecedent or the syntactic role of the relativised element in the relative clause is high,24 e.g. for SU and DOs (also in the standard) or for oblique complements with highly accessible referents due to (verb) semantics or context (cf. examples (20) above); relative pronouns encode "medium accessibility" (indirect objects, prepositional objects, genitives in standard relative constructions), and relative particles plus resumptive pronouns mark a relatively low accessibility due to structural distance between relative pronoun and its antecedent or complex syntactic structures inside the relative clause (cf. examples (18) above).

What about, then, the huge discussion about the different "substandard" relative clause types and their classification as being socially or stylistically marked? This is due, we think, to the hearer's, not the speaker's impression and acceptability judgements when hearing such a construction and comparing it with normative grammar s(he) has been exposed to during the whole education process.25 Speakers recur to type A and B in those rare cases where these two types correspond better than others to their processing difficulties, i.e. to either very accessible and easy or very complex imbedded structures, which would create a rather long distance between the antecedent and the relativised element or quite complicated wh-movements ("pied-piping structures" etc). They do this, as the above mentioned empirical studies have shown, independently of their education, style or (in)formality of the situation.

Yet, one parameter often mentioned in sociolinguistic and/or variationist work is not to be forgotten, and that is the medium: "substandard" and "standard" relative clause types, especially the ones with relative pronouns, seem to be related to either phonic, i.e. spontaneous and spoken, and graphic, i.e. planned and written forms of communication.

Ciò ci riporta alla stretta connessione tra struttura non-standard della relativa e meccanismi psicolinguistici del parlato, cioè la micropianificazione, dimostrando la priorità da accordare alla dimensione diamesica [i.e. phonic vs. graphic, ES] nel rendere conto della distribuzione del fenomeno nelle diverse varietà dell'italiano contemporaneo. (Alfonzetti 2002: 31)

With type A and Type B being very well motivated by structural simplicity or complexity, they occur in spoken, not in written texts, where normative pressure and time for planning the message and revising and avoiding 'errors' in complex structures are both active.26 And as written texts tend to be normative models much more than spoken ones, typical structures of the latter tend to be criticised or downgraded by the hearers – even if they do not belong to any special variety of the (Romance) languages.27

PhiN 47/2009: 11


Albrecht, Jörn (1986): "'Substandard' und 'Subnorm'. Die nicht-exemplarischen Ausprägungen der 'Historischen Sprache' aus varietätenlinguistischer Sicht", in: Günter Holtus / Edgar Radtke (eds.): Sprachlicher Substandard, vol. 1, Tübingen: Niemeyer, 65–88.

Albrecht, Jörn (1990): "'Substandard' und 'Subnorm'. Die nicht-exemplarischen Ausprägungen der 'Historischen Sprache' aus varietätenlinguistischer Sicht (Fortsetzung)", in: Günter Holtus / Edgar Radtke (eds.): Sprachlicher Substandard, vol. 3: Standard, Substandard und Varietätenlinguistik, Tübingen: Niemeyer, 44–127.

Alfonzetti, Giovanna (2002): La relativa non-standard – Italiano popolare o italiano parlato?, Palermo: Centro studi filologici e linguistici siciliani.

Ariel, Mira (1999): "Cognitive Universals and Linguistic Conventions: The Case of Resumptive Pronouns", in: Studies in Language 23/2, 217–269.

Auger, Julie (1993): "On the History of Relative Clauses in French and Some of Its Dialects", in: Henning Andersen (ed.): Historical Linguistics 1993. Selected Papers from the 11th International Conference on Historical Linguistics, Los Angeles, 16–20 August 1993, Amsterdam/Philadelphia: Benjamins, 19–32.

Bellini, Simona (1998): I pronomi relativi nell'italiano parlato: un'indagine basata sui corpora LIP, Berlin: Freie Universität Berlin (= Romanistische Linguistik. Berliner Staatsexamens- und Magisterarbeiten, 8).

Bernini, Giuliano (1989): "Tipologia delle frasi relative italiane e romanze", in: Favio Foresti / Elena Rizzi / Paola Benedini (eds.): L'italiano tra le lingue romanze. Atti del XX congresso internazionale di studi, Bologna, 25–27 settembre 1986, Rom: Bulzoni, 85–98.

Bernini, Giuliano (1991): "Frasi relative nel parlato colloquiale", in: Cristina Lavinio / Alberto A. Sobrero (eds.): La lingua degli studenti universitari, Florenz: Nuova Italia Editrice, Scandicci, 165–187.

Berruto, Gaetano (1985): "Per una caratterizzazione del parlato: l'italiano parlato ha un'altra grammatica?", in: Günter Holtus / Edgar Radtke (eds): Gesprochenes Italienisch in Geschichte und Gegenwart. Tübingen: Narr, 120–153.

Berruto, Gaetano (92000): Sociolinguistica dell'italiano contemporaneo, Rom: Carocci.

Blanche-Benveniste, Claire (1990): "Usages normatifs et non normatifs dans les relatives en français, en espagnol et en portugais", in: Johannes Bechert / Giuliano Bernini / Claude Buridant (eds.): Toward a Typology of European Languages, Berlin – New York: Mouton de Gruyter, 317–335.

Brucart, José María (1999): "La estructura del sintama nominal: las oraciones de relativo", in: Ignacio Bosque / Violeta Demonte (eds.): Gramática descriptiva de la lengua española, vol. 1, Madrid: Espasa, 397–522.

Butler, Christopher S. (1992): "A Corpus-Based Approach to the Relative Clauses in the Spoken Spanish of Madrid", in: Hispanic Linguistics 5:1–2, 1–42.

Catoira, Loreto (2008): "Whose Syntax Is It Anyway? Indo-European and Semitic Links in Spanish Relative Clauses with Pronominal Resumption", in: PhiN 45, 1–11 (

Chomsky, Noam (1977): "On Wh-Movement", in Peter W. Culicover / Thomas Wasow /Adrian Akmajian (eds.), Formal Syntax, New York: Academic Press, 71–132.

PhiN 47/2009: 12

Cinque, Guglielmo (31991): "La frase relativa", in: Lorenzo Renzi / Giampaolo Salvi / Anna Cardinaletti (eds.): Grande grammatica italiana di consultazione, vol. 1: La frase. I sintagmi nominale e preposizionale, Bologna: Il Mulino, 457–517.

Cortés Rodríguez, Luis (1990): "Usos anómalos del relativo en español hablado", in: Revista española de lingüística 20/2, 431–446.

Cresti, Emanuela (2000): Corpus di italiano parlato, 2 vols., CD-Rom, Florence: Accademia della Crusca.

Cresti, Emanuela / Moneglia, Massimo (eds.) (2005): C-ORAL-ROM: Integrated Reference Corpora for Spoken Romance Languages, Amsterdam/Philadelphia: Benjamins.

Cristofaro, Sonia / Giacalone Ramat, Anna (2007): "Relativization strategies in the languages of Europe", in: Paolo Ramat / Elisa Roma (eds.): Europe and Mediterranean as linguistic areas: convergences from a historical and typological perspective, Amsterdam/Philadelphia: Benjamins, 1–38.

Davis, Mark (2002): Corpus del Español, Brigham Young University. [URL:].

Deulofeu, José (1981): "Perspective linguistique et sociolinguistique dans l'étude des relatives en français", in: Recherches sur le français parlé 3, 135–193.

Downing, Bruce T. (1978): "Some Universals of Relative Clause Structure", in: Joseph H. Greenberg (ed.): Universals of Human Language, vol. 4: Syntax, Stanford: Standford University Press, 375–418.

Dufter, Andreas / Stark, Elisabeth (2002): "La variété des variétés: combien de dimensions pour la description? Quelques réflexions à partir du français", in Romanistisches Jahrbuch 53, 81–108.

Fiorentino, Giuliana (1999): Relativa debole. Sintassi, uso, storia in italiano, Milano: Francoangeli.

Fleischer, Jürg (2005): "Relativsätze in den Dialekten des Deutschen: Vergleich und Typologie", in: Linguistik online 24/3.

Foulet, Lucien (1928): "La difficulté du relatif en français moderne", in: Revue de philologie française et de littérature 40, 100–124, 161–181.

Gadet, Françoise (1995): "Les relatives non standard en français parlé: le système et l'usage", in: Hanne L. Andersen / Gynver Skytte (eds.): La subordination dans les langues romanes. Actes du colloque international, Copenhague, 5.5.–7.5. 1994, Kopenhagen: Munksgaard, 141–162.

Giacalone Ramat, Anna (1982): "Explorations on Syntactic Change (Relative Clause Formation Strategies)", in: Anders Ahlqvist (ed.): Papers from the 5th International Conference on Historical Linguistics, Amsterdam/Philadelphia: Benjamins, 283–292.

Godard, Danièle (1989): "Français standard et non-standard: les relatives", in: LINX 20, 51–88.

Hirschbühler, Paul / Rivero, María-Luisa (1982): "Aspects of the Evolution of Relatives in Romance", in: Anders Ahlqvist (ed.): Papers from the 5th International Conference on Historical Linguistics, Amsterdam/Philadelphia: Benjamins, 132–152.

D'Introno, Francesco (1992): "Subordinadas relativas en el español de Carácas", in: Elizabeth Luna Traill (ed.): Scripta philologica in honorem Juan M. Lope Blanch a los 40 años de docencia en la Unam y a los 65 años de vida, vol 2: Lingüística Española e Iberoamericana, México: Universidad Nacional Autónoma de México, 541–551.

PhiN 47/2009: 13

Jones, Michael Allan (1996): Foundations of French Syntax, Cambridge: Cambridge University Press.

Keenan, Edward L. / Comrie, Bernard (1977): "Noun Phrase Accessibility and Universal Grammar", in: Linguistic Inquiry 8/1, 63–99.

Keenan, Edward L. / Comrie, Bernard (1979): "Noun Phrase Accessibility Revisited", in: Language 55/3, 649–664.

Koch, Peter / Oesterreicher, Wulf (1990): Gesprochene Sprache in der Romania: Französisch, Italienisch, Spanisch, Tübingen: Niemeyer.

Kurzowá, Helena (1981): Der Relativsatz in den indoeuropäischen Sprachen, Hamburg: Buske.

Lehmann, Christian (1979): "Der Relativsatz vom Indogermanischen bis zum Italienischen. Eine Etüde in diachroner syntaktischer Typologie", in: Die Sprache 25, 1–25.

Lehmann, Christian (1984): Der Relativsatz: Typologie seiner Strukturen; Theorie seiner Funktionen; Kompendium seiner Grammatik, Tübingen: Narr.

Prince, Ellen F. (1990): "Syntax and Discourse: A Look at Resumptive Pronouns", in: Kira Hall et al. (eds.): Proceedings of the Sixteenth Annual Meeting of the Berkeley Linguistics Society, February 16–19, 1990. General Session and Parasession on the Legacy of Grice, Berkeley/California: Berkeley Linguistics Society, 482–497.

Riegel, Martin / Pellat, Jean-Christophe / Rioul, René (1994): Grammaire méthodique du français, Paris: Presses Universitaires de France.

Rovere, Giovanni (1979): "Un testo di italiano popolare del primo Ottocento", in: Vox romanica 38, 74–84.

Schafroth, Elmar (1993): Zur Entstehung und vergleichenden Typologie der Relativpronomina in den romanischen Sprachen. Mit besonderer Berücksichtigung des Substandards, Tübingen: Niemeyer.

Schafroth, Elmar (1995): "À propos d'une typologie panromane des relatifs 'non normatifs'", in: Catherine Bougy / Pierre Boissel / Bernard Garnier (eds.): Mélanges René Lepelley. Recueil d'études en hommage au Professeur René Lepelley, Caen: Musée de Normandie, 363–374.

Silva-Corvalán, Carmen (1996): "Resumptive Pronouns: A Discourse Explanation", in: Claudia Parodi / Carlos Quicoli / Mario Saltarelli / María Luisa Zubizarreta (eds.): Aspects of Romance Linguistics, Washington D.C.: Georgetown University Press, 383–395.

Stark, Elisabeth (2004): "L'italiano parlato – varietà storica o variazione universale?", in: Federico Albano Leoni / Francesco Cutugno / Massimo Pettorino / Renata Savy (eds.): Il parlato italiano. Atti del convegno nazionale, Napoli, 13–15 febbraio 2003 (CD-ROM), Neapel: M. D'Auria Editore, CIRASS, E09.

Stempel, Wolf-Dieter (1964): Untersuchungen zur Satzverknüpfung im Altfranzösischen, Braunschweig: Westermann.

Suñer, Margarita (1998): "Resumptive Restrictive Relatives: a Crosslinguistic Perspective", in: Language 74/2, 335–364.

PhiN 47/2009: 14

Tarallo, Fernando (1989): "Inside and Outside Relative Clauses: Pronominal Redundancy in Portuguese", in: Ralph W. Fasold / Deborah Schiffrin (eds.): Language Change and Variation, Amsterdam/Philadelphia: Benjamins, 255–274.

Tejada Caller, Paloma (1991): "Diachronic Considerations on English Relative Clauses: Semantic and Structural Evolution", in: Revista Canaria de Estudios Ingleses 22/23, 145–164.


* This article owes much to the great collaboration of Charlotte Meisner and Natascha Pomino, Zurich, and to the important critical remarks of an anonymous reviewer. All remaining shortcomings, are, of course, mine.

1 Cf. Deulofeu (1981: 184).

2 The Modern standard equivalent of example (1) would be la santé très vigoureuse dont j’ai joui jusqu’ici, with the relative particle dont, invariable as to gender and number marking, but indicating the status of a complement introduced by the preposition de.

3 We will not go very much into the discussion of the concept of "substandard" (cf. Albrecht 1986, 1990) here, we will simply use it as designating the whole array of non-standard varieties of a language, along the social, stylistic, ‘diasituational’ and maybe even medial (‘phonic’ or ‘graphic’) dimension of linguistic variation.

4 Compare also the following quote: "C’est dans l’emploi du relatif que la langue populaire se sépare le plus complètement de la langue cultivée. L’écart est même si marqué qu’il peut servir à définir l’une par rapport à l’autre. La langue populaire, en France et en ce moment, est essentiellement une langue qui a simplifié le système des relatifs [...] Il n’est pas sans intérêt de rechercher pourquoi trente millions de Français au bas mot sont incapables de se servir du relatif en se conformant aux règles de la grammaire." (Foulet 1928: 100)

5 This example stems from a news speaker on television.

6 For an illustrating overview of the multiple and sometimes contradictory classifications of Italian "substandard" relative clauses cf. Stark (2004: table 1).

7 Cf. Jones (1996: 498ff.).

8 The third strategy being a gap as in English, cf. This is the man ø I met last week, which is not available in Romance and will therefore not be discussed in what follows.

9 It is based on the descriptions found in Cinque 31991 for Italian, Riegel/Pellat/Rioul 1994 and Jones 1996 for French and Brucart 1999 for Spanish. In what follows, we will leave out a discussion of spatial adverbs like , dove, donde 'where', sometimes also appearing as the clause initial element in relative clauses.

10 English more or less literal translations of examples (4) to (9): (4): ‘the woman who speaks’; (5): ‘there is nothing which I am interested in’; (6) ‘the woman whom I see’; (7) ‘the woman whose daughter is dead’; (8) ‘the woman about whom I have spoken’; (9) ‘there is nothing about which I would like to speak’.

11Cf. Jones (1996: 508-512).

12 English more or less literal translations of examples (10) to (17): (10) = (8); (11) = (9); (12) = (8); (13) = (9); (14) ‘the man of whom you do not know when you saw him’; (15) = (8); (16) = (9); (17) ‘I remember the city wallPL of that city, which they broke down in order to build apartment houses’.

PhiN 47/2009: 15

13 The morphosyntactic variation in Spanish relative clauses is a bit different from the other two Romance languages, so that we do not always find exactly comparable examples, but the main generalisations apply to it as well. Especially the phenomenon of obligatory clitic-doubling for indirect and animate direct objects leads to a higher frequency of resumptive pronouns in Spanish relative clauses and to a much higher acceptability of these constructions, cf. Silva-Corvalán (1996: 387), Fiorentino (1999: 42ff.), Brucart (1999: 408).

14 As for French qui in subject function, there is widespread agreement on the fact that it represents an allomorph of que in front of finite verbs; there are several arguments to assume that it is not a relative pronoun like the homophonous qui-element appearing after prepositions in relative clauses: i) contrary to the latter one, it is not sensitive to semantic features of its referent like [± human], [± animate] or [± abstract]; ii) the [-i] can be dropped in casual speech, contrary to the qui of the relative / interrogative pronoun: Je connais la femme qu’est venue vs. *Je ne connais pas la femme de qu’on a parlé; iii) it appears even in contexts as an allomorph for que where it introduces an underlying object, not a relative clause: [CP Spec C [IP tu crois [CP que [IP quel étudiant est arrivé]]]] - [CP quel étudiant crois [IP tu _ [CP que [IP _ est arrivé]]]] for Quel étudiant crois-tu qui / *que est arrivé? (cf. Jones 1996: 507).

15 Cf. with a similar argumentation Godard (1989: 61ff.).

16 Due to, of course, historical events in the respective normalisation processes, especially during the Renaissance period. Cf. Giacalone Ramat (1982: 286), describing the inherited forms from Latin, e.g. It. che, cui, Fr. qui/que etc., and the creation of new inflected forms of the il quale-type, resulting in an innovative, but inconsistent Romance system; see also Lehmann (1979: 19f.) for Italian, Blanche-Benveniste (1990) for French and Fleischer (2005) for German and other languages.

17 Cf. Berruto (1985). See also Jones (1996: 517), with two different rules for French, one which allows the derivation of relative clauses in standard French, assuming wh-movement and wh-NP-deletion, the other one without assuming these operations; cf. also Giacalone Ramat (1982: 291ff.) and Fiorentino (1999: 29ff.) for Italian.

18 Surprisingly, this holds also true for older stages of the Romance languages, with some interlinguistic variation, and an only slight increase of type A and the same triggering conditions for type B as described below, cf. Schafroth (1993: 81-164).

19 Cf. furthermore Prince (1990: 483ff.), Bernini (1989: 92–96), Bellini (1998: 54ff.) for Italian, Gadet (1995: 151f.) for French, Brucart (1999: 405ff.) for Spanish etc.

20 Cf. the following quote: "Oltre a non essere un tratto esclusivo o fortemente caratterizzante dell’italiano popolare, la relativa non-standard non è neppure confinata ai registri più informali dei parlanti colti, cioè all’italiano colloquiale e informale trascurato." (Alfonzetti 2002: 53).

21 "Wenn man die Standardsprache mit den hier untersuchten Varietäten vergleicht, so weist sie ein völlig atypisches Bild auf: Die Standardsprache kennt fast ausschließlich flektierte Relativpronomen; sie kennt beispielsweise praktisch gar keine relativsatzeinleitenden Partikeln oder gespaltene Pronominaladverbien, und die auftretenden Relativpronomen werden nicht mit Partikeln kombiniert. Das System, das die Standardsprache aufweist, findet sich in keinem der untersuchten Dialekte wieder; es ist darum wohl nicht auf eine natürliche Entwicklung, sondern auf Standardisierung unter dem Einfluss anderer europäischer Standardsprachen zurückzuführen. Das System der neuhochdeutschen Standardsprache hat keine direkt vergleichbare dialektale Basis und ist in diesem Sinn unnatürlich." (Fleischer 2005). Cf. also Albrecht (1990: 101), who speaks of "erschwerte Sprache" concerning the overall morphosyntactic characteristics of standards.

22 "First of all, the standard language is not necessarily the variety that most clearly reflects the evolution of a certain structure." (Tejada Caller 1991: 153).

23 "[…] we can see that an RP [= resumptive pronoun, type B, E.S.] is invariably preferred when the degree of accessibility associated with the […] antecedent is relatively low at the point when the (NPrel) anaphor is processed. Zeros are preferred when the degree of accessibility associated with the antecedent is extremely high, and rich verbal agreement (mainly personal agreement) may form an intermediate accessibility marker." (Ariel 1999: 226).

24 With minor interlinguistic differences due to certain morphosyntactic characteristics of the single Romance languages like pro-drop, obligatory clitic-doubling with certain syntactic functions etc., cf. Hirschbühler / Rivero (1982), Ariel (1999: 253), Fiorentino (1999: 177ff).

25 Cf. Deulofeu (1981: 154ff.), Godard (1989: 51).

26 "[…] if resumptive pronouns are indeed the result of a last resort strategy which can be attributed to processing difficulties, such elements are not likely to be frequently found in written texts, where the possibility of correcting one’s writing should in fact eliminate most such constructions." (Auger 1993: 26f.) Here, we can point out that we did not find any "substandard" relative clause construction in a letter of a not very much educated Italian soldier of the nineteenth century (cf. Rovere 1979).

27 Cf. Kurzowa (1981: 89) for a typological survey showing this tendency in many languages.