Navigation – Plan du site

When corpus analysis refutes common beliefs: the case of interpolation in European Portuguese dialects

Catarina Magro
p. 115-136

Résumés

Quand l’analyse de corpus réfute des idées reçues : le cas de l’interpolation dans des dialectes du portugais européen
Cet article analyse l’interpolation (c’est-à-dire la possibilité d’occurrence d’un proclitique séparé du verbe) comme un trait des dialectes du portugais européen (PE) contemporain, tel qu’il est montré par les données fournies par le Syntax-oriented Corpus of Portuguese Dialects – CORDIAL-SIN. Les objectifs de cet étude sont les suivants : (i) décrire les propriétés des constructions d’interpolation dans les dialectes contemporains du PE ; (ii) comprendre la corrélation entre l’interpolation et les propriétés d’autres constructions avec clitiques dans les mêmes dialectes ; (iii) fournir une perspective diachronique sur l’interpolation non-standard en tenant compte de l’histoire générale de l’interpolation. La recherche sur ce thème, dont les résultats sont présentés en détail dans Magro (2007), montre que, dans beaucoup de dialectes contemporains du PE, l’interpolation actuelle est une construction très productive qui possède des traits particuliers, qui la différencient de l’interpolation du PE ancien. Cette étude montre que, contrairement à ce qu’on croit d’habitude, l’interpolation n’est pas un trait archaïque mais plutôt une innovation récente dans la grammaire du PE. L’analyse présentée adopte la structure de la grammaire définie dans le cadre de la Morphologie Distribuée (Distributed Morphology, Halle & Marantz 1993, 1994), en considérant l’interpolation comme une instance d’une opération de déplacement post-syntaxique qui a lieu au stade final d’une dérivation.

Haut de page

Texte intégral

1. Introduction

1It is well known that EP differs from most Romance languages with respect to clitic placement in that enclisis and proclisis occur both in finite and non-finite domains. In standard EP, the two patterns are in complementary distribution in finite clauses: proclisis is triggered by the presence of an overt complementizer, a negation operator, certain adverbs in preverbal position, a displaced wh-, quantificational or ‘affective’ phrase; enclisis surfaces in the remaining environments. The two alternative patterns are illustrated below, the sentences in (1) displaying proclisis while that in (2) displays enclisis.

2In the proclisis contexts, exemplified in (1), the clitic can be separated from the verb by intervening material. The phenomenon of non-adjacency between the proclitic and the verb (known as interpolation in the Romance philological literature) was a very productive phenomenon in Old Portuguese. During that period, a wide variety of constituents could be interpolated. The loss of generalized interpolation occurs in the 17th century and from then on only the negation operator não (not) can disrupt proclitic-verb sequences (cf. Martins 1994; Fiéis 2003; Namiuti 2008). Nowadays, the proclitic-verb adjacency is mandatory for the majority of speakers of standard EP, although não interpolation is still an available option for some others. The sentences in (3) illustrate the marked contrast between Old and Contemporary interpolation:

3Along with the highly restrictive interpolation of standard EP, illustrated in (3b), there are dialectal varieties that behave more freely with respect to this phenomenon. The dialectal interpolation has been interpreted as the remains of the Old Portuguese system and, therefore, the grammars of these dialects have been regarded as conservative grammars (cf. Martins 1994; Barbosa 1996; Fiéis 2003). In this article I will claim that this is an erroneous idea, which stems from insufficient empirical support.

4I will discuss this issue taking into account a large amount of unreported interpolation data drawn from the Syntax‑oriented Corpus of Portuguese Dialects – CORDIAL-SIN. On the basis of this new evidence I will review the traditional perspective about this topic.

5My analysis of dialectal interpolation accounts for (i) the specific properties of dialectal interpolation (contrasting with both the wide interpolation of Old Portuguese and the restrictive interpolation of standard contemporary EP) and (ii) the connection between the interpolation phenomenon and other clitic related phenomena displayed by dialectal varieties in the very same syntactic contexts where interpolation surfaces: free variation between proclisis and enclisis and clitic duplication flanking an interpolated element or a verbal form. My fundamental claim is that interpolation, clitic duplication, and unexpected enclisis are special instances of a single displacement operation taking place in the Morphology component of grammar, namely the operation of metathesis as formulated by Harris & Halle (2005).

6This article is organized as follows. Section 2 describes the defining properties of the dialectal interpolation construction. Section 3 examines the challenges posed by the reported dialectal data and it evaluates earlier accounts of the interpolation phenomenon in the light of the new data. In Section 4 I present an alternative analysis of dialectal interpolation. Section 5 discusses the status of the dialectal phenomenon in the history of the construction. Section 6 closes the paper making the explanatory power of my approach clear.

2. The interpolation in contemporary EP dialects

2.1 The considered dialectal corpus

  • 1  The CORDIAL-SIN project is supported by national and European funding (PRAXIS XXI/P/PLP/13046/1998 (...)

7The empirical support for this work has been drawn from the Syntax-oriented Corpus of Portuguese Dialects – CORDIAL-SIN. This corpus is being built up since 1999, at the Linguistics Center of University of Lisbon (CLUL), within the scope of a research project aimed at promoting the study of European Portuguese dialect syntax, a fairly recent area of interest in Portuguese linguistic research.1

8CORDIAL-SIN is a corpus of spoken dialectal EP that collects a geographically representative body of excerpts of spontaneous and semi-directed speech, selected from the oral interviews gathered by the Linguistic Variation Team at CLUL in the course of several Dialect Geography projects. The corpus covers 42 locations within the (continental and insular) territory of Portugal and it compiles about 600 000 words.

Map I. Geographical distribution of CORDIAL-SIN locations

9Map I shows the geographical distribution of the CORDIAL-SIN locations. The compilation of CORDIAL-SIN represents a turning point in the history of Portuguese dialectology, because for the first time a corpus of dialectal syntactic data was set up and systematically studied.

10As in the case of other Romance languages, traditional dialectal studies do not focus on syntax; thus, data gathering methods and instruments (like questionnaires) were not designed to elicit and collect syntactic information.

11However, the archives of the CLUL Linguistic Variation Team contain raw syntactic data in those cases were spontaneous speech have been recorded during the fieldwork sessions (mostly to serve ethnographic purposes). These data have been retrieved, transcribed and annotated by the CORDIAL-SIN team. The resulting material constitutes the first corpus of dialectal Portuguese (and the only existing one so far).

12The lack of syntactic data available to researchers explains why some features of Portuguese dialects remained unnoticed until recently, while others have been described and explained without enough empirical support, as in the case of the interpolation phenomenon.

13On the other hand, dialectal variation became a central issue to syntacticists studying linguistic variation within the Principles & Parameters framework, following the proposals by Chomsky (1981) and subsequent work. This led to a need for an accurate and enlarged empirical basis on Portuguese dialects fulfilling the demands of dialect syntax comparative inquiry.

2.2 The dialectal interpolation data

14The collection of data drawn from CORDIAL-SIN provides a general picture of contemporary dialectal interpolation. Besides the negation operator não (that can also be interpolated in standard contemporary EP), there is a wide range of constituents that dialectally may intervene between the proclitic and the verb. Table I lists the interpolated elements of the considered dialectal corpus.

Table I. The interpolated elements of CORDIAL-SIN

pronouns

eu (I)

ele/ela (he/she)

nós/ a gente (we/the people[=we])

eles/elas (they)

esta (this [feminine])

isso (that [neuter])

isto (this [neuter])

adverbs

aqui (here [+close to speaker])

(there [-close to speaker/+close to addressee])

ali (there [-close to speaker/addressee])

(here [+close to speaker])

(there [-close to speaker/addressee])

agora (now)

depois (after)

então (then)

hoje (today)

ontem (yesterday)

ainda (still)

(already)

assim (like this)

prepositional phrases

para lá (to there [=there])

para aí (to there [=there])

a nós (to we [=to us])

negation operator

não

15These interpolated constituents –which belong to different morphosyntactic categories– play diverse syntactic and discursive functions, as it is presented at Table II.

Table II. Categories and functions of the interpolated constituents

morphosyntactic category

syntactic/discursive function

subjects

pronouns

peripheral expletives

topics

modifiers

adverbs

displaced objects

displaced small-clauses predicates

modifiers

prepositional phrases

displaced objects

displaced clitic-doubling phrases

16Examples (4) through (13) illustrate the interpolation of the different kinds of constituents.

2.3 The class of the interpolated elements

  • 2 Up to the 16th century, all kinds of subjects and IP-scrambled constituents could occur between the (...)
  • 3 In this respect, Barbosa (1996) and Fiéis (2001, 2003) propose, respectively, that dialectal interp (...)

17CORDIAL-SIN provides a wide and varied inventory of interpolated elements. Actually, the interpolated constituents vary in morphosyntactic class, syntactic structure, grammatical and discursive function and metrical structure. However, contemporary dialectal interpolation is far from the generalized and highly permissive interpolation of Old Portuguese.2 We need thus to identify the grammatical property shared by all the interpolated elements, that is, the property that makes it possible for these elements (and not for any other) to be interpolated.3

18I claim that these dialectal interpolated elements share a semantic property: all of them are referentially deficient elements with a deictic interpretation (elements bearing a [+dependent] feature, in formal terms).

19Table III lists the dialectal interpolated elements according to their type of deixis. As the table shows, the traditional categories of deixis are evenly covered (and in some cases almost filled) by the dialectal interpolated elements.

Table III. Interpolated elements – types of deixis

personal deixis

eu (I)

ele/ela (he/she)

nós/ a gente (we/the people[=we])

eles/elas (they)

a nós (to we [=to us])

spatial deixis

aqui (here [+close to speaker])

(there [-close to speaker/+close to addressee])

ali (there [-close to speaker/addressee])

(here [+close to speaker])

(there [-close to speaker/addressee])

para lá (to there [=there])

para aí (to there [=there])

esta (this [feminine])

isso (that [neuter])

isto (this [neuter])

temporal deixis

agora (now)

depois (after)

então (then)

hoje (today)

ontem (yesterday)

ainda (still)

(already)

manner deixis

assim (like this)

20With the exception of the negation operator não, all the dialectal interpolated elements are classified as deictics in a straight way. The special case of não will be addressed in section 5.

3. Previous interpolation analyses and new dialectal data

3.1 The syntactic approaches

21The word order found in interpolation structures excludes a purely syntactic treatment of this phenomenon (Martins 1994, 2003, 2005; Fiéis 2001, 2003). There are several arguments against a syntactic approach to dialectal interpolation. Here, I will sketch three of them.

22In the first place, it is problematic to associate the clitic with an invariable structural position. In interpolation constructions within a single dialect, the clitic either (i) precedes elements of the high left periphery, like peripheral expletives (cf. Carrilho 2005), as in (15a), or topics, as in (16a), or (ii) is preceded by elements of the IP domain, like aspectual adverbs, as in (15b)-(16b), or subjects:

23On the other hand, the fact that a clitic may intervene between the elements of a phrase is incompatible with an analysis that derives interpolation by clitic movement across the interpolated element to a higher functional category.

24Finally, constituents with similar syntactic properties are not affected by interpolation in the same way. The contrast between pronominal subjects and regular DP-subjects interpolation, illustrated below, reveals the particular shape of the dialectal construction. This contrast does not hold for Old Portuguese interpolation.

3.2 The prosodic approaches

25Likewise, the metrical structure of interpolated elements and some other aspects concerning clitics prosodization disprove the predictions made by an analysis that accounts for interpolation in prosodic terms (Barbosa 1996). Barbosa (1996) attributes dialectal interpolation to a restructuring process at the prosodic level that enables the clitic to form a unit with a prosodic phrase that exhaustively dominates a monosyllabic prosodic word. Her analysis predicts that interpolated elements are monosyllabic prosodic words and that leftward phonological cliticization is a ruled out option in these dialects’ grammars. None of these predictions is supported by CORDIAL-SIN data. The list in (20) shows that interpolated elements have a variable metrical structure. In (21), the mid central vowel is produced as a glide. This means that this segment doesn’t occur at the prosodic word boundary and concomitantly that the clitic is left cliticized.

4. An alternative analysis

26My analysis of contemporary dialectal interpolation assumes the organization of grammar as envisioned by the theory of Distributed Morphology (Halle & Marantz 1993, 1994 and subsequent work), namely a Morphology component in PF with the ability to change the syntactic output.

27Under my proposal, contemporary dialectal interpolation is the result of a morphological readjustment rule that manipulates a linearized syntactic output. Specifically, I claim that this phenomenon is derived by a metathesis process, in the sense of Harris & Halle (2005), whereby a designated subsequence in a given phonological string is duplicated and the peripheral elements in each of the generated copies are deleted afterwards.

28According to Harris & Halle (2005), metathesis is just a special instance of a more general duplication process in which a designated substring in a base form can be repeated in its entirety (full duplication) or in part (partial duplication) in a derived form.

29The formalism of Harris & Halle (2005) is presented below. The abstract derivation in (26) displays the metathesis device as conceived by the authors.

30In my analysis, the metathesis operation is triggered when the deictic-clitic sequence occurs in a linearized string and is motivated by the [+dependent] feature borne by the two elements involved (note that the clitic is a referentially dependent element itself). My hypothesis is that the grammar of EP dialects that exhibit interpolation includes the following rule:

31Since metathesis and duplication are just different instances of the same operation, we should expect to find clitic duplication data in the same syntactic contexts in which these informants produce interpolation. In fact, CORDIAL-SIN data validate this prediction. Examples (29)-(30) illustrate the interpolation with clitic duplication construction. Note that in these cases the two copies of the clitic are flanking the interpolated element.

32Clitic duplication will thus be the Spell-Out of a partial duplication operation:

5. A different history of interpolation

33From what has been said so far, we are naturally led to formulate the following question: what is the connection between contemporary dialectal interpolation and Old Portuguese interpolation, given the apparent post-syntactic nature of the first one and the pure syntactic status of the last one? I claim that dialectal interpolation doesn’t have its direct origins in Old Portuguese interpolation, corresponding, on the contrary, to a recent innovation in the history of Portuguese.

34Two facts can be invoked to support such a hypothesis: (i) texts from the 19th century start to attest interpolation data, similar to the dialectal interpolation data, after a gap of one hundred and fifty years (texts from the second half of the 17th century and from all the 18th century don’t attest interpolation data of any constituent but não) and (ii) dialectal EP contrasts with dialectal Galician in what concerns the interpolation phenomenon, what suggests the late emergence of dialectal interpolation in the history of Portuguese (happening at a time in which Portuguese and Galician are distinct linguistic systems).

35Under my hypothesis, these morphological displacement operations emerge in Portuguese grammar in the 17th century as the outcome of a reanalysis process of não interpolation. In this reanalysis process the clitic-não-verb sequence is interpreted as the output of a metathesis operation involving the clitic and the functional category (Laka 1990; Martins 1994 and subsequent work). The relevant contexts for the reanalysis to take place are those in which is endowed with a [+dependent] feature, that is, the contexts where the requirement concerning the strong nature of is satisfied by syntactic merger between and C (cf. Costa & Martins 2003, 2004). The metathesis operation then spreads over other contexts where the clitic is preceded by an element bearing a [+dependent] feature as well, that is, a deictic element. Under this alternative view, the deictic interpolation data are the manifestation of a recent and innovating change.

36In this picture, the case of the interpolation of não plays a central role in the history of interpolation in Portuguese. In fact, its course along the time is the key to explain the contrast between: (i) the generalized syntactic interpolation of Old Portuguese, (ii) the highly restricted interpolation of contemporary standard Portuguese and (iii) the less constrained and lexically-oriented interpolation of contemporary dialectal varieties.

6. Other cases of metathesis and duplication in EP dialects

37Such an analysis is descriptively adequate as it thoroughly covers the identified interpolation data (a welcome result in itself), and allows us to account for seemingly disparate phenomena as special cases of a single grammatical device.

38In addition to interpolation, let’s consider two other phenomena displayed by dialectal varieties in the very same syntactic contexts (standard proclisis contexts) where interpolation surfaces: (i) free variation between proclisis and enclisis and (ii) clitic duplication flanking a verbal form or an interpolated element. This cluster of constructions is illustrated in (32)-(36) with data produced by a speaker of the dialect of Lavre (a southern EP variety from Alentejo).

39Besides standard proclisis with adjacency, as in (32), this speaker produces, in typical proclisis contexts, interpolation, as in (33), interpolation with clitic duplication, as in (34), regular enclisis, as in (35), and pre and post verbal clitic duplication, as in (36). All these constructions coexist in the speech of a single individual and they are not occasional performance errors but systematic realizations. To the best of my knowledge, this is an unattested paradigm across the Romance languages.

40I take these coexistent constructions to be closely related phenomena and I propose a formal treatment that captures their relatedness appropriately, which previous interpolation analyses have not been able to do. My fundamental claim is that interpolation, unexpected enclisis and clitic duplication are special instances of a single displacement operation taking place in the Morphology component of grammar.

41Assuming Costa & Martins (2003, 2004) analysis of clitic placement in EP, I propose that the object derived in the syntactic component of the grammar corresponds to that in (37), in which the clitic is left-adjoined to the maximal projection of T (clitics left-adjoin to the highest functional head targeted by verb movement, that, in most cases, corresponds to T in EP) (cf. Kayne 1991).

  • 4 According to Costa & Martins (2003, 2004) proposal, can be licensed in PF, as a last resource, un (...)

42I hypothesize that whenever the visibility requirement of is satisfied in the syntactic component (where is lexicalized under syntactic merger or under -to-C movement),4 the clitic is free to enter metathesis operations with an adjacent element both to its left and to its right. That is, I propose that the grammar of dialects with interpolation includes a metathesis/ duplication rule that operates in both directions.

43When none of these rules applies, standard proclisis is displayed, as in (39). The rule of left metathesis/duplication will derive plain interpolation and interpolation with clitic duplication, as in (40); similarly, the rule of right metathesis/ duplication will derive regular enclisis and clitic duplication in pre and post verbal position, as in (41).

44Finally, I suggest that this kind of operations affects adjacent elements bearing the same feature. In this particular case, it is the fact that the clitic bears a dependent-feature and V-features that makes it possible for the clitic to enter metathesis/ duplication operations with a deictic element that may precede it and with the verb that always follows it.

Haut de page

Bibliographie

Barbosa P. (1996). « Clitic placement in European Portuguese and the position of subjects », in A. Halpern & A. Zwicky (eds), Approaching Second: Second Position Clitics and Related Phenomena. Stanford, California : CSLI Publications, 1-40.

Carrilho E. (2005). Expletive Ele in European Portuguese Dialects. PhD Dissertation, Universidade de Lisboa.

CORDIAL-SIN – Syntax-oriented Corpus of Portuguese Dialects, http://www.clul.ul.pt/sectores/variacao/cordialsin/projecto_cordialsin.php

Costa J. & Martins A. M. (2003). « Clitic placement across grammar components », Going Romance 2003. Nijmegen University, November 2003.

Costa J. & Martins A. M. (2004). « What is a strong functional head ? », Lisbon Workshop on Alternative Views on the Functional Domain. Universidade Nova de Lisboa, July 2004.

Fiéis A. (2001). « Interpolação no português medieval como adjunção a XP », Actas do XVI Encontro Nacional da Associação Portuguesa de Linguística. Lisboa : APL, 197-211.

Fiéis A. (2003). Ordem de Palavras, Transitividade e Inacusatividade. Reflexão Teórica e Análise do Português dos Séculos XIII a XVI. PhD Dissertation, Universidade Nova de Lisboa.

Halle M. & Marantz A. (1993). « Distributed Morphology and the pieces of inflection », in K. Halle & S. J. Keyser (eds), The View from Building 20 : Essays in Linguistics in Honor of Sylvain Bromberger. Cambridge, Massachussetts : MIT Press, 111-176.

Halle M. & Marantz A. (1994). « Some key features of Distributed Morphology », in A. Carnie & H. Harley (eds), MITWPL 21 : Papers on Phonology and Morphology. Cambridge, Massachussetts : MIT Press, 275-288.

Harris J. & Halle M. (2005). « Unexpected plural inflections in Spanish : reduplication and metathesis », Linguistic Inquiry 36, 2 : 195-222.

Laka I. (1990). Negation in Syntax. On the Nature of Functional Categories and Projections. PhD Dissertation, MIT.

Magro C. (2007). Clíticos : Variações sobre o Tema. PhD Dissertation, Universidade de Lisboa.

Martins A. M. (1994). Clíticos na História do Português. PhD Dissertation, Universidade de Lisboa.

Martins A. M. (2003). « From unity to diversity in Romance syntax : Portuguese and Spanish », in K. Braunmuller & G. Ferraresi (eds), Aspects of Multilinguism in European Language History. Amsterdam & Philadelphia : John Benjamins, 201-233.

Martins A. M. (2005). « Clitic placement, VP-ellipsis and scrambling in Romance », in M. Batllori, M-L Hernanz, C. Picallo & F. Roca (eds), Grammaticalization and Parametric Change. Oxford : Oxford University Press, 175-193.

Namiuti C. (2008). Aspectos da História Gramatical do Português: Interpolação, Negação e Mudança. PhD Dissertation, Universidade Estadual de Campinas.

Haut de page

Notes

1  The CORDIAL-SIN project is supported by national and European funding (PRAXIS XXI/P/PLP/13046/1998; POSI/1999/PLP/33275; POCTI/ LIN/46980/2002; PTDC/LIN/71559/2006).

2 Up to the 16th century, all kinds of subjects and IP-scrambled constituents could occur between the proclitic and the verb (Martins 1994, 2003, 2005).

3 In this respect, Barbosa (1996) and Fiéis (2001, 2003) propose, respectively, that dialectal interpolated elements are monosyllabic prosodic words and verbal adjunct heads. CORDIAL-SIN data don’t confirm their proposals.

4 According to Costa & Martins (2003, 2004) proposal, can be licensed in PF, as a last resource, under morphological merger between and the verb. For the morphological merger to take place, and the verb must be adjacent. In these cases, Local Dislocation with inversion applies and the clitic becomes enclitic.

Haut de page

Table des illustrations

URL http://corpus.revues.org/docannexe/image/1859/img-1.png
Fichier image/png, 16k
URL http://corpus.revues.org/docannexe/image/1859/img-2.png
Fichier image/png, 16k
URL http://corpus.revues.org/docannexe/image/1859/img-3.png
Fichier image/png, 12k
Légende Map I. Geographical distribution of CORDIAL-SIN locations
URL http://corpus.revues.org/docannexe/image/1859/img-4.png
Fichier image/png, 29k
URL http://corpus.revues.org/docannexe/image/1859/img-5.png
Fichier image/png, 23k
URL http://corpus.revues.org/docannexe/image/1859/img-6.png
Fichier image/png, 41k
URL http://corpus.revues.org/docannexe/image/1859/img-7.png
Fichier image/png, 37k
URL http://corpus.revues.org/docannexe/image/1859/img-8.png
Fichier image/png, 22k
URL http://corpus.revues.org/docannexe/image/1859/img-9.png
Fichier image/png, 18k
URL http://corpus.revues.org/docannexe/image/1859/img-10.png
Fichier image/png, 7,0k
URL http://corpus.revues.org/docannexe/image/1859/img-11.png
Fichier image/png, 10k
URL http://corpus.revues.org/docannexe/image/1859/img-12.png
Fichier image/png, 15k
URL http://corpus.revues.org/docannexe/image/1859/img-13.png
Fichier image/png, 19k
URL http://corpus.revues.org/docannexe/image/1859/img-14.png
Fichier image/png, 12k
URL http://corpus.revues.org/docannexe/image/1859/img-15.png
Fichier image/png, 19k
URL http://corpus.revues.org/docannexe/image/1859/img-16.png
Fichier image/png, 2,9k
URL http://corpus.revues.org/docannexe/image/1859/img-17.png
Fichier image/png, 15k
URL http://corpus.revues.org/docannexe/image/1859/img-18.png
Fichier image/png, 12k
URL http://corpus.revues.org/docannexe/image/1859/img-19.png
Fichier image/png, 10k
URL http://corpus.revues.org/docannexe/image/1859/img-20.png
Fichier image/png, 6,1k
Haut de page

Pour citer cet article

Référence papier

Catarina Magro, « When corpus analysis refutes common beliefs: the case of interpolation in European Portuguese dialects », Corpus, 9 | 2010, 115-136.

Référence électronique

Catarina Magro, « When corpus analysis refutes common beliefs: the case of interpolation in European Portuguese dialects », Corpus [En ligne], 9 | 2010, mis en ligne le 04 juillet 2011, consulté le 19 août 2017. URL : http://corpus.revues.org/1859

Haut de page

Auteur

Catarina Magro

Centro de Linguística da Universidade de Lisboa

Haut de page

Droits d’auteur

© Tous droits réservés

Haut de page
  • Logo Revues électroniques de l’université de Nice
  • Revues.org