Beyond Piedmontese

  • Cecilia Goria
Part of the Studies in Natural Language and Linguistic Theory book series (SNLT, volume 60)


Optimal Agreement, the twofold model proposed in the previous chapters, has proven advantageous in a number of aspects especially in comparison to extant analyses of SCLs which rely on multiple SCL projections. Optimal Agreement offers a more economical alternative to the view that the morphological specifications for person, including deixis, number and gender require distinct structural positions, increasing the complexity of sentence structure. The same holds for distributional properties of SCLs such as optionality and omission in coordination, including the peculiar distribution of 2sg and 3sg and 3p1 SCLs. As we have seen, concentrating on two varieties of Piedmontese, Optimal Agreement successfully promotes structure-minimality in accounting for the structural position and function of Piedmontese SCLs and the variation related to the morphological and distributional properties of these elements. Nonetheless, comparative considerations are necessary to test and refine Optimal Agreement in relation to different SCLs systems. The comparative discussion that follows is divided into two main sections: one which, relating Optimal Agreement to SCLs across the NIDs, takes into account the claim that SCLs are in T and denote T’s properties, and one which extends beyond Piedmontese the treatment of morphologically distinct SCL systems central to the OT component of Optimal Agreement.


Strong Negation True Imperative Agreement Feature Alignment Constraint Optimal Agreement 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    A point worth making, valid for all accounts of invariable clitics so far, including the considerations just advanced in the text, is that the agreement-less nature of invariable SCLs raises the important question as to why invariable clitics do not occur with non-fiinite verb forms and true imperatives (section 3 below in the text and chapter 6). If, in fact, the incompatibility of true-SCLs with non-finites is linked, as traditionally believed, to the lack of agreement feature of non-finites, nothing in principle prevents invariable clitics from occurring with agreement-less forms. I leave this question open to further study.Google Scholar
  2. 2.
    See Poletto (2000b: ch. 3 and p. 82) for a summary of the CPs that may be activated in interrogatives in the NIDs.In this book interrogatives are dealt with in chapter 6.Google Scholar
  3. 3.
    For discussions on Romance interrogatives see among others the works cited in fn. 1 in chapter 6.Google Scholar
  4. 4.
    As for OCLs, strong negation always precedes OCLs, while weak negation may, in certain varieties, follow some OCLs, namely 1st person, 2nd person and reflexives, but precede 3rd person OCLs, locatives and genitives. See Zanuttini (1997) for more details and Parry (1997a) for a discussion on clitic ordering focussing on North-West Italian dialects.Google Scholar
  5. 5.
    Frascarelli’s (2000:124) view is that Italian non is dependent upon the verb not for syntactic reasons but for phonological reasons: non needs a host word to form a Phonological Phrase. For this reason non always attaches to the verb, and then it excorporates’ leaving the verb behind if further raising is needed for independent reasonsGoogle Scholar
  6. 6.
    Note that, according to Zanuttini’s (1997) distinction, this negative marker in Loreo is weak, as it requires the post-marker mina. Despite Zanuttini’s analysis of weak marker as occupying a position different from that of strong markers, Poletto still takes the word order in (13) and (16) as evidence in favour of her analysis (see Poletto 2000b:33 fn. 21 for a remark by Poletto on this point).Google Scholar
  7. 7.
    One aspect of SCLs not mentioned in the text but dealt with by Poletto (2000b: ch. 6) is the way in which SCLs double overt preverbal subjects. The result of her investigation is that invariable SCLs tolerate no overt preverbal subjects while deictic clitics double all types of preverbal subjects (tonic pronoun, DP, QP, variable). With regards to agreement clitics there is a high degree of variation across the NIDs depending especially on the type of subject. As Poletto claims the issue of doubling with agreement clitics is not related to their position but to their featural content and its relation with verbal inflection. As I do not have a strong argument to propose an alternative explanation, I leave this issue open to further research, referring however the reader to section 5.3.3 in chapter 3 where I express my objections to Poletto’s view about the relation between SCLs and verbal agreement. Furthermore, I find the incompatibility between invariable SCLs and overt subjects not clear. Firstly because, I have found no such a phenomenon in Astigiano, the SCLs of which qualify as invariable clitics according to Poletto’s criteria. Secondly, because even though the invariable position is claimed to be higher than the position of preverbal subjects (cf. (56) section 3.2 chapter 3), nothing in principle within Poletto’s system should block the co-occurrence between these clitics and preverbal subjects. In the light of the discussion in section 2.1.1 in this chapter, my suggestion is that if true invariable clitics are in CP, their incompatibility with preverbal subjects is expected because the intervening subject would prevent their cliticisation to the verb.Google Scholar
  8. 8.
    See Graffi (1996, and references therein) for a distinction between imperativo sintattico’ and imperativo nozionale’. The former refers to the imperative mood, the latter to the notion of command.Google Scholar
  9. 9.
    See Zanuttini’s (1997) own work (and references therein) for an explanation of the fact that 2sg imperative of 2nd conjugation shows up as -i instead of the expected thematic vowel -e. Google Scholar
  10. 10.
    As we know, Piedmontese has no strong negation, thus the authenticity of imperative forms cannot be tested against negation. NIDs with strong negation are expected to behave like standard Italian.Google Scholar
  11. 11.
    Beukema and Coopmans (1989) define imperatives as [—Tense +Agr] and place their agreement features in C.Google Scholar
  12. 12.
    Graffi’s analysis relies on TP, AgrSP, CP as separate functional projections.Google Scholar
  13. 13.
    I do not deal here with questions concerned with the subjects of imperative forms and I assume that, if EPP is universally strong, imperatives have strong EPP, hence a subject. Note that lack of agreement features in T is not a problem for the subject’s Φ features as these are interpretable and need not be checked. The reader is referred to Beukema and Coopmans (1989) for the idea that imperatives have a subject, and to Platzack and Rosengren (1994) for a claim in the opposite direction.Google Scholar
  14. 14.
    Note that true imperatives are commonly regarded as being also deficient in tense (Zanuttini 1997, among many others. Thus, the ban of SCLs with true imperative may be related to deficiency for tense. In fact, if richer Tense specification is required, as in the case of Piedmontese future imperative (Brero and Bertodatti 1993), verbs with indicative morphology replace true imperatives and, as one would expect, are accompanied by SCLs. Nonetheless, this observation raises questions about the use of SCLs with subjunctives and with pseudo-relatives (see chapter 7) which are also defective in tense, in the sense that their tense properties cannot be determined by the time of utterance.Google Scholar
  15. 15.
    See among others Cresti (1994) for the Tuscan variety of San Gimignano; Cuneo (1997a) for the Ligurian variety of Cicagna; Jones (1992, 1993, 1999) for Sardinian; Ledgeway (2000) for the Southern Italian Dialects and a number of other Romance varieties; Menching (2000) for the Romance languages in general; Torrego (1998b) and Rigau (1995) for Spanish and Catalan.Google Scholar
  16. 16.
    My reference to Barron’s work is concerned only with the points relevant to the present discussion. The reader is referred to Barron (2000 and references therein) for more details on pseudo-relatives.Google Scholar
  17. 17.
    Rigau (1995) examines personal infinitive clauses with temporal interpretation in Spanish and Catalan. These are exemplified in i) and ii) respectively: i) Todo el mundo applaudió al acabar el concierto everybody applauded at-the to finish the concert ii) Tothom va aplaudir en acabar el concert everybody applauded in to finish the concert ‘Every body applauded when the concert was finished’ She claims that personal infinitives are marked as [—Tense +Agr] that are the covert counterpart of inflected infinitives. That is, “in contrast with Portuguese and Galician [which exhibit inflected infinitives], Catalan and Spanish have no verbal form to express the poverty of tense and richness of agreement properties at the same time” (Rigau 1995:288). Without going into the details of Rigau’s (1995) proposal, it is clear that a proviso must be made for those varieties which have both personal and inflected infinitives. The availability of both types within one and the same variety, as in Sardinian (Jones 1992, 1993, 1999), shows that lack of morphological means to express the combination [—Tense +Agr] cannot be the only explanation for personal infinitives. This observation weakens Rigau’s (1995) proposal and therefore I will not explore it further.Google Scholar
  18. 18.
    Note that Cicagnino has a type of inflected infinitive that allows SCLs. However, the inflected infinitive found in Cicagnino (Cuneo 1997a) differs from the type considered here in that it occurs only with the verb ’to be’ after a modal verb: i) U dev’êse (dêse) lê SCL must be.3sg him It must be him’ As Cresti (1994) explains these forms are the result of Reconstruction and slittamento ’ (slipping, movement — my translation) of verbal morphology from the modal to the infinitive which, thanks to Reconstruction, has come to form a single VP with the modal. Therefore, the presence of the SCL u in sentence i) above is irrelevant for the present discussion. This type of inflected infinitives is also found in other dialects, such as in the Tuscan variety of San Gimignano (Cresti 1994) and in Altamurano (Loporcaro 1988).Google Scholar
  19. 19.
    Note that even the behaviour of complement clitics does not help in clarifying how inflected infinitives would affect the distribution of SCLs. According to Vincent (1996) inflected infinitives allow proclisis as well as enclisis of complement clitics. While this increases the ambiguity of inflected infinitives with respect to Finiteness (see chapter 7), it is inconclusive with respect to SCLs.Google Scholar
  20. 20.
    System 2 has not been included since it is claimed (Poletto 2000b:13 fn.3) that the SCLs of such a system have unique syntactic properties differing from the SCLs of all the other system.Google Scholar
  21. 21.
    I thank an anonymous reviewer for raising this issue.Google Scholar
  22. 22.
    Note that the optimal candidate for 3p1 is not specified for number. This is consistent with the fact that in Astigiano, if 3sg has a unique form for the masculine, SCL a is used for 3sg feminine as well as 3p1.Google Scholar
  23. 23.
    The highest SCL projection of the Agreement Field is exempt from this condition as it does not encode subject features.Google Scholar
  24. 24.
    I have already mentioned that the notion of non-re-rankable constraints is exploited in Aissen (2001) in order to explain markedness universals. The idea that the ban on certain feature combinations is a consequence of constraint interaction is central to Grimshaw’s (2001) account of the clitic inventory in Romance. Similar to our suggestion, Grimshaw concludes that the interactions between constraints that penalise and constraints that reward a pronoun for encoding input specifications and ’combination constraints’ make sure that certain feature combinations are never selected as the optimal ones. Hence, they are never found. The conceptual similarities between the issues dealt with in our discussion and Aissen’s and Grimshaw’s proposals are striking, despite their relying on different theoretical assumptions.Google Scholar
  25. 25.
    For clarity and lack of space, the irrelevant persons and constraints have been omitted, and the tableau has been split into three separate tableaux.Google Scholar
  26. 26.
    An analogous observation holds for System 5, where the 3sg SCL is fully specified, while the 3p1 SCL is. Google Scholar

Copyright information

© Springer Science+Business Media Dordrecht 2004

Authors and Affiliations

  • Cecilia Goria
    • 1
  1. 1.University of NottinghamUK

Personalised recommendations