Clitic-placement in the History of Portuguese

and the Syntax-Phonology Interface

 

Charlotte Galves, State University of Campinas

 

 

 The complex pattern of clitic-placement in Modern European Portuguese tensed sentences has long been noticed and discussed in the framework of Generative Grammar. In this talk, I'll propose an approach on this topic based on the comparison between the modern and the classical syntax of clitic-placement. I'll show that such a comparison sheds a new light on the way  phonology and syntax  interact to produce the observed patterns. Furthermore, it helps us to understand better the relationship between  clitic-placement and the position of pre-verbal subjects .

 

 

1.          Clitic-placement  in Modern European Portuguese  (EP)

 

Enclisis is obligatory in V1 constructions, and with topics, adjuncts and referential subjects preceding the verb in root affirmative sentences, as illustrated from (1) to (3).

 

(1)        a. Encontrei-o ontem

            b. *O encontrei ontem

"I met him yesterday"

(2)        a.   Ontem, encontrei-o

            b.   *Ontem, o encontrei

" Yesterday, I met him"

(3)        a. O Paulo falou-me

b. *O Paulo me falou

"Paulo spoke to me"

 

Proclisis is obligatory in all the other cases:

 

            a)         in negative clauses

                        (4)        a. O Paulo não me fala

b.* O Paulo não fala-me

“Paulo does not speak to me

            b)         in subordinate clauses

   (5)        a.  Todo mundo  sabe que a viste

               b. *  Todo mundo  sabe que viste-a

               "Everybody  knows that (you) saw her

(6)        a. Se tu me tivesses dito...

            b. *Se tu tivesses-me dito    

          "If  you had said to me.."

c)         in clauses in which the preverbal phrase is a quantifier (7), a WH operator (8), a focalized phrase (9), or an adverb of a certain class (10)

(7)        a. Alguém me chamou

b. *Alguém chamou-me

"Somebody called me"

            (8)        a. Quem me chamou?

                        b. * Quem chamou-me?

                        "Who called me"

            (9)        a. Só ele a entende.

                        b. *Só ele entende-a

                        "Only him understands her"

            (10)      a. eu (sempre, aínda, já) a encontrei no mercado

                        b. * eu (sempre, aínda, já) encontrei-a no mercado

                        "I (always, still, already) met her at the market"

 

2.             Barbosa's analysis

 

Barbosa  (1991, 1996, 2000) argues that all enclitic constructions are V1 constructions. In other terms, according to her, enclisis derives from the application of the  Tobler Mussafia Law which bans unstressed words at the absolute beginning of sentences (cf. also Salvi 1991, and Benincà 1995). This is straightforward for (1) and (2), but (3) needs an auxiliary hypothesis in order to be derivable from this analysis. If (3) is a case of the application of the Tobler-Mussafia law, this means that  pre-verbal subjects in EP do not occupy a position internal to the clause, but are dislocated, like topics. Barbosa argues at length in favor of this hypothesis, in the general framework of the discussion of the position of subjects in null subject languages, in the line of Vallduvi (1992) and others. According to this line of argumentation, the A-position for subjects in NSLs is the post-verbal position and pre-verbal subjects occupy a A' -position. As for pre-verbal referential subjects in EP, Barbosa (2000) argues that this position is of adjunction to IP.

 

3.             The comparison between EP and ClP

 

In this talk, I shall bring evidence that this analysis is problematic for EP. In contrast, I'll show that it fits well with the behavior of Classical Portuguese (henceforth ClP), the language written in Portugal between the 15th and the 18th century. The contrast between the two languages will help us to understand better the phenomenon of enclisis, and its relationship with the position of the verb in the clause.

             First, I'll present  arguments that enclisis is a V1 phenomenon in ClP.   Second, I'll show  that Barbosa's proposal for EP faces empirical problems.

 

3a           Enclisis is a V1 phenomenon in ClP

 

The striking fact about ClP is that, in  contrast with EP,  only  when the verb is in absolute first position is enclisis categorical.   In sentences like (2) and (3), where some phrase - including a non quantified subject - immediately precedes the verb,   we find a variation between enclisis and proclisis, as illustrated below with examples from the Sermons of Padre Antonio Vieira (born in 1608) :

 

(11)      As outras prophecias cumprem-se a seu tempo

the other prophecies  achieve-SE ("are achieved") in their time

(12)      Estes thesouros, pois, que agora estão cerrados, se abrirão a seu tempo

these treasures, therefore, that now are closed, SE-will-open ("will be opened") in its time

 

Moreover, proclisis is much more frequent than enclisis in texts.  Figure 1, from Galves Britto and Paixão de Sousa (2002), shows the  variation between enclisis and proclisis  in V>1 sentences in texts from authors born from 1541 to 1836 (the data are drawn from the Tycho Brahe Corpus cf. http://www.ime.usp.br/~tycho/corpus ).   It is important to note that this variation does not affect the cases of obligatory proclisis defined above for EP, which remain constant along all the history of Portuguese. Therefore, crucially, contexts like the ones defined in (4)-(11) are not taken into consideration in this graph.

In Figure 1, we see that up to 1700, enclisis is a marginal choice.   For all the texts except one,   proclisis appears in at least 80% of the cases, and a great majority displays less than 10% of enclisis.   We find an exception in Padre Antonio Vieira,   who uses enclisis in almost 40% of  the cases.

 

Figure1

This fact was already noted by Martins (1994), who used this fact as an evidence that Vieira was already a speaker of  EP, and therefore that the change from ClP to EP had already taken place at the beginning of the 17th century. But Figure 1 clearly shows that this is an isolated case. Moreover, Vieira himself, in his letters, is as proclitic as his Contemporaries. Therefore it is not Vieira himself who is an exception but Vieira's Sermons. This apparent contradiction in Vieira's use gives us in fact an important key to understand the variation between enclisis and proclisis in ClP. In effect, it turns out that in the totality of the cases of enclisis with pre-verbal subjects, the subject is clearly a contrastive topic, entering in opposition with another term, as exemplified in (13) (cf. Galves 2001). 

 

(13)      Elles conheciam-se, como homens, Christo conhecia-os, como Deus.

They knew-themselves, as men, Christ knew-them, as God

 

The example already given above in (11) is a clear illustration of the same case. (14) gives the whole sentence in which we see that the subject "As outras profecias" is contrasted   with "esta do dia do Juiso".

 

(14)      As outras prophecias cumprem-se a seu tempo, esta do dia do Juiso tem o seu cumprimento antes de tempo; 

The other prophecies  achieve-SE ("are achieved") at their time, this one of the Final Judgement have its achievement  before its time

 

Note that in (13) and (14), the contrast is expressed not only by the  subjects but also by the rest of the sentence:  as men/as god,  at their time/ before its time.

The conclusion we can draw from this is that enclisis in ClP does correspond to a structure in which the pre-verbal phrase is outside the boundaries of the clause. This is stylistically marked, and the normal pattern, when the verb is not in the first position is that proclisis show up. Since the variation we observe affects all kind of pre-verbal phrases, we can infer that there is an internal pre-verbal position which is available both for subject and non-subjects, like in V2 languages.  Like in V2  languages, there is an internal topic position, and an external topic position, which is a marked option.  Figure 2  shows that pre-verbal subjects  follow the general pattern of the variation between enclisis and proclisis evidenced in Figure 1. This is what we expect in V2 languages.

 

Figure 2

Both Figure 1 and Figure 2 evidence a change taking place from the beginning of the 18th century on, leading to the pattern described at the beginning of this talk: in all the cases of variation, enclisis becomes the only option.  An interesting evidence that this change involves a change in the  position of the pre-verbal subjects with enclisis  is found in the following fact,  described by Paixão de Sousa (2002).  In the 12  texts from authors born before 1700,  there are 43 sentences in which the verb is preceded by a subject which is preceded itself by another phrase (XP-Subj-V)[1].  It turns out that 42 of those sentences are proclitic, and only one is enclitic - again in the Sermons, in the context of contrastive topics[2]. This is exactly what the analysis put forth above predicts. Since there are two positions for topics, one external and one internal,  if two topics show up together, we expect proclisis to arise, not enclisis.   In contrast, in the last author considered in Graphs 1 and 2, Ramalho Ortigão, born in 1836, we find 12 cases of  XP-Subj-V-cl.  This is what we predict if in  this author, enclisis is no more a V-1 phenomenon. In this case, the subject is internal to the sentence, and the pre-verbal XP is the topic, an order which is very common in EP, as we shall see below.

 One of the contexts in which we also find a variation between proclisis and enclisis in ClP is when an adverbial clause immediately precedes the verb. As suggested to us by Tony Kroch, this context provides us with an interesting way of testing the effect of the Tobler-Mussafia law. In effect, we expect that the longer the pre-verbal clause is, the more likely it corresponds to an autonomous intonational phrase,  in which case the verb will be the first element of the main clause.  We have therefore measured the length of pre-verbal clauses in terms of the number of words. The table below shows the results we obtained per century (cf. Galves, Britto and Paixão de Sousa 2002) :

 

Sensitivity of clitic-placement to the length of pre-verbal clauses

 

1500 to 1599

 

 

1600 to 1690

 

 

1691 to 1799

 

 

1800

 

 

 

proc

encl

%encl

proc

encl

%encl

proc

encl

%encl

proc

encl

%encl

1-4 w

68

29

29.89

38

35

47.94

12

35

74.46

0

10

100

5-8 w

33

18

35.29

27

28

50.90

16

31

65.95

0

4

100

9+  w

8

9

52.94

10

25

71.42

15

15

50.00

0

0

0

 

109

56

 

75

88

 

43

81

 

0

14

 

 

            Interestingly, we see that both in the sixteen and the seventeen century, the frequency of enclisis significantly increases when the pre-verbal clause has more than eight words. In contrast,  in the 18th century,  which corresponds to the period during which the frequency of enclisis with subjects raises from 0 to 80 %, we observe the reverse tendency.   The raise of enclisis and the decline of sensitivity to the length of preverval clauses put together give us a strong evidence that at this period, enclisis ceases to be governed by the Tobler -Mussafia law.

 

3b      Subjects are in a clause internal A-position in EP

 

The second part of the argument against Barbosa's analysis consists in  bringing synchronic arguments that in EP pre-verbal subjects are  internal to IP.  This can be done along two lines of evidence:

 

1)         EP lacks the properties of the Null Subject Romance Languages which provided evidence that  pre-verbal subjects are not  in A-position in these languages.

2)         There is evidence  that topics and subjects have different properties

 

Costa (1998) argues that  preverbal definite subjects in EP do not exibit A-bar properties. His main point is that pre-verbal subjects do not  produce minimality effects in WH constructions. In contrast with what happens in languages like Spanish (cf. Ordoñez and Treviño, 1995) and Greek (cf. Alexiadou and Agnastopoulou,1996), the well-formedness of sentences like (15) and (16) show  that the pre-verbal subject do not block  A-bar extraction[3]:

 

(15)      esses livros o Paulo leu

           these books Paulo read

(16)      que livros o Paulo leu?

               which books Paulo read

 

Another argument put forth by Costa (1998) relies on certain contrasts between configurations with multiple non-subject topics and configurations with one topic followed by a subject. According to Costa, multiple topicalization constructions are possible in EP, but they are slightly marked and require intonational breaks between the topics, which explains the contrast of acceptability between (17) on the one hand and (18) and (19) on the other hand  (the judgements are his):

 

(17)      Sobre o tempo falei com o Pedro

           about the weather (I) talked with Pedro

(18)      ?/?? Sobre o tempo, com o Pedro, falei

           about the weather with Pedro (I) talked

(19)      ?/?? Com o Pedro, sobre o tempo, falei

with Pedro about the weather (I) talked

 

The interesting  point is that no such break is necessary between a preposed constituent and a subject, in sentences like (20). Furthermore, changing the order between the subject and the PP affects the acceptability of the sentence. (21) is thus much more marked than (22).   In this case, it must be reduplicated by an anaphoric DP in pre-verbal position, which identifies it as a hanging topic. 

 

(20)      Com a Maria, o Pedro falou

           With Maria, Pedro talked

(21)      ??O Paulo com a Maria falou rapidamente

           Paulo with Mary talked quickly

(22)      ?O Paulo com a Maria esse sacana falou rapidamente

           Paulo with Mary  that jerk talked quickly

 

  There is therefore a clear contrast between sentences with preverbal subjects and sentences with multiple topics, which seems to support the claim that subjects are not simply adjoined to the clause like topics, and are in fact in a A-position internal to the clause.

In conclusion, from both synchronic and diachronic points of view, word order considerations strongly suggest that enclitic pre-verbal subjects are not dislocated in EP, but occupy a clause internal A-position. In contrast, enclisis in ClP has been shown to correspond to a marginal construction in which the pre-verbal phrase, be it subject or not, is in  a position external to IP. If this analysis is on the right track, this means that the position of clitics is driven by the Tobler Mussafia Law in ClP, but not in EP.

 

4.     Subject position and clitic-placement in EP

 

 An alternative way to derive enclisis in EP relies on the hypothesis that in contrast to the other Romance languages, root clauses in EP are characterized by the presence of an extra functional category between C and T (see, among others, Cardinaletti and Roberts (1991), Costa and Galves (2001), Martins (1994), Raposo&Uriagereka (1996), Rouveret (1992), Uriagereka (1995a,b)),  with different labels.  These authors argue that such a functional projection is needed in order to account for the two apparently unrelated phenomena considered above: enclisis in affirmative main clauses on the one hand, and on the other hand the relative order between wh-elements (or left-peripheral focused constituents) and left-dislocated constituents, in particular referential subjects. I'll adopt here the implementation of this proposal proposed in Costa and Galves (2001) -cf. also Costa 1996 - with F=Agr, the subject in the Specifier of Agr and the verb in T, as represented in (23)[4]:

 

(23)      [AgrP Subj  Agr [TP V-T(-cl) [….]]]

  

From this point of view,  the Tobler Mussafia law cannot be invoked to explain enclisis. However, I shall now argue that the EP pattern of clitic-placement can be linked to a peculiarity of its prosody. Independently of the existence of an initial stress at the level of the Intonational Phrase,  EP has an initial stress at the level of the Prosodic Word, as it can be seen from the distribution of secondary accents (Andrade and Laks 1992, Vigário 2001). This straightforwardly explains, independently of the position of the verb with respect to the boundaries of the clause, why prosody bans clitics in pre-verbal position[5]. The question is now why EP does not instantiate a purely enclitic pattern. The answer lies in the relationship between syntax and phonology in the general architecture of Grammar. 

 

 

5.    What prevents enclitic pattern in all the cases in EP?

 

From a purely phonological point of view, we would therefore expect a systematically enclitic pattern in EP, independently of the position of the verb in the clause.  But this is obviously not borne out, since as already mentioned above, the contexts of obligatory proclisis remained unchanged during the history of European Portuguese.

One of the main issues raised by the complex pattern of clitic-placement in European Portuguese, and one in which people disagree a lot is the derivation of enclisis itself.  In the line of Kayne (1991, 1994) many authors have adopted a derivation of enclisis by which the clitic  left-adjoins  to the verb in INFL, and then V moves to a higher functional category, stranding the clitic in INFL.   This analysis faces the theoretical problem of excorporation (Cf. Dobrovie Sorin and Galves 2001) and fails to account for the fact that  V+cl behaves much more as a morphological unit than cl+V, as shown by Benincà and Cinque (1994).

From a phonological point of view, Vigário (1999, 2001) brings segmental and accentual evidence that the prosodization of proclitic and enclitic pronouns with respect to their verbal host is different in EP.  She concludes that while proclitics are adjoined to prosodic word, enclitics are incorporated to it. This nicely accounts for the fact that V+cl, but not cl+V, behaves as a morphological unit. If following  Kayne,  we assume that clitic-placement is universally to the left of the verb in T,  an alternative way of accounting for enclisis compatible with this fact is to derive it from a morphological process which applies to incorporate the clitic to the right of the verb in accordance with the prosodic ban on weak syllables at the left edge of  prosodic words (cf. Barbosa 1991 for a similar proposal in the framework of Distributed Morphology). The question is why this process is blocked in all the cases described above of obligatory proclisis.

This brings us back to the structure of the clause represented in (23). The peculiarity of this structure is that the subject position is not in a spec/head agreement relationship with the verb.  I'd like to argue that this is the crucial property  which licenses enclisis. By contrast with the analyses mentioned above, the role attributed to the extra functional category here is not to host the verb, stranding the clitic in T (INFL), but to host a specifier outside the projection of T containing V. This is this externality of the subject which is crucial for enclisis. In other terms, enclisis is excluded when the verb enters in a spec/head agreement relationship. This is easily derivable from the morphological approach suggested above. In effect, it is natural to think that the morphological incorporation of the clitic to the right of the verb has the effect of blocking any further checking operation. The morphological operation which affixes the clitic on the right of the verb encapsulates the word in such a way that checking operations are no longer possible.  This is incompatible to the presence of a specifier for the category containing the verb.

On the basis of this reasoning, we can conclude that all the cases of obligatory proclisis are cases in which T obligatorily has a specifier. This means either that Agr is not projected (as an independent category) or that despite of the fact it is projected, either the movement to Spec/Agr is obligatorily through Spec/T, or some phrase independently occupies Spec/T.

 I'll briefly show that the cases of obligatory proclisis described in (4)-(11) are all amenable to one of these situations.

 

a)                             The obligatory proclitic pattern when C has features (cf. 5, 6 and 8) indicates that in this case Agr is not projected as an independent category.  A natural hypothesis is that when C is active, it selects Tense as its complement, blocking the projection of an extra category between C and T[6].

b)                             The obligatory  proclitic pattern  with quantified and focalized subjects (cf. 7 and 9) derives from the fact  that in spite of being an A-position, Spec/Agr is not a V-related position[7]. Non-referential phrases must be in a V-related position. They therefore raise to Spec/Agr through Spec/T.

c)                             The obligatory  proclitic pattern  with aspectual and focalizing adverbs straightforwardly derives from the fact that they themselves occupy Spec/T.

 

  Sentential negation is a special case, since the functional negative item não is stressed and is able to satisfy the constraint inside the complex T which contains the verb.

 

6.     Concluding remarks

 

-           ClP and EP display two different enclitic systems, only in ClP is enclisis dependent on V1. But in both cases, enclisis is licensed by the fact that the verb is in a functional category which has no specifier.

-           The analysis proposed here supports the parallel derivation of syntax and morphology. If morphology applied after all the sentence is derived, we would not expect a syntactic constraint on enclisis, since the checking operation would have already occurred.

-           This analysis is able to derive all the cases of  obligatory proclisis, in contrast with accounts based on the local effect of a lexical higher functional category, as in Raposo and Uriagereka (2002), which are unable to explain the permanence of the proclitic pattern when some phrase occurs between this lexical category and the verb .

-           Contrary to Barbosa (2000), this analysis accounts for why enclisis is not the unique pattern observed in EP.

-           In contrast with many authors, Duarte e Mattos (1995, 2000)  claim that enclisis is the basic pattern of EP, and proclisis is a derived one, in the sense that it causes more computation, driven by the presence of certain categories. From the point of view of the analysis proposed here,  the basic syntactic pattern is proclisis. It is from a phonological point of view that the basic pattern is enclitic. The distribution of the two patterns depends on an interaction between the requirements of phonology and the constraints on the syntactic computation.

In conclusion, the pattern for EP could be formulated in these terms:

Be enclitic whenever you need  (for phonological reasons) AND you can (for syntactic reasons)

 

 

References:

Barbosa, pilar. 1991 "Clitic-placement in EP", Manuscript, MIT.

Barbosa, pilar. 1996, in Approaching Second.

Barbosa, pilar. 2000. "Clitics: a window into the null subject property" in joão costa (ed.).

Benincà, paola. 1995. “Complement clitics in medieval Portuguese: the Tobler-Mussafia Law”, in A.battye and  i. roberts (eds.), Language Change and Verbal Systems. Oxford University Press.

Cardinaletti, anna and ian roberts 1991

costa, joão. 1996. "Positions for subjects in European Portuguese". in WCCFL XV Proceedings. CSLI: Stanford

costa, joão. 1998. Word Order Variation: a constraint-based approach. HIL/Leiden University.

costa, joão (ed.) 2000.  Portuguese Syntax. Oxford University Press, New York.

costa, joão and charlotte Galves. 2002. "External subjects in two varieties of Portuguese: evidence for a non-unified analysis",  In Claire Beyssade, Reineke Bok-Bennema, Frank Drijkoningen, Paola Monachesi (eds). Proceedings of Going Romance 2000, Benjamins, Amsterdam/Philadelphia.

Dobrovie-sorin, carmen and charlotte galves 2002.

Duarte, inês, and Gabriela Mattos  2000. in joão costa (ed.).

Galves, charlotte. 2000. in joão costa (ed.).

Galves, charlotte. 2001. "Syntax and Style: clitic-placement in Padre Antonio Vieira", to appear in Santa Barbara Portuguese Studies. Vol. VI.

Galves, charlotte, helena britto and maria clara paixão de sousa.2002.  "The change in clitic placement from classical to modern european portuguese: results from the Tycho Brahe Corpus", Ms, Unicamp.

gonçalves vianna, acrísio. 1883 Essais de phonétique et de phonologie de la langue portugaise d'après le dialecte actuel de Lisbonne, in Estudos de Fonética portuguesa, 1973, Lisboa: Imprensa Nacional-Casa da Moeda, pp. 83-152.

haegeman, liliane

Kayne, richard. 1991

Kayne, richard. 1994

Kleinhenz, ursula. 1997 "Domain typology at the phonology/syntax interface", in Interfaces in Linguistic Theory,

martins ana maria.  1994. Clíticos na história do português, unpublished doctoral thesis, University of Lisbon.

Raposo, eduardo and juan uriagereka. 1996 "Indefinite SE", Natural Language and Linguistic Theory, pp.749-810.

Raposo, eduardo and juan uriagereka. 2002 "Clitic Placement in Western Iberian: A Minimalist View", Ms, UMD and UCSB.

Rizzi, luigi. 1997. "The fine structure of the left periphery", in liliane haegeman (ed.) Elements of Grammar: Handbook of Generative Syntax, Dordrecht: Kluwer.

salvi, gianpaolo. 1990. “La sopravvivenza della legge di Wackernagel nei dialettioccidentali della peninsola iberica”, Medioevo Romanzo 15: 177-210.

teyssier, paul. 1987. História da Língua Portuguesa. Lisboa, Livraria Sá da Costa Editora.

vallduvi, eric. 1992. "A  preverbal landing site for quantificational operators", Catalan Working Papers in Linguistics 1992, Barcelona: University Autònoma of Barcelona.

vigário, marina. 2001 The prosodic word in European Portuguese, unpublished doctoral thesis, University of Lisbon.

 

 



[1] It must be noted that V>3 is not a frequent pattern. Before 1700, we find it in 11% of the cases (47 cases for a total of 438 sentences). After 1700, the proportion raises to 16% (40 cases for a total of 252 sentences).

[2] Se/CONJS o/D zelo/N vos/CL come/VB-P a/P vós/PRO ,/, a/D-F vossa/PRO$-F substancia/N converte-se/VB-P+SE em/P zelo/N ;/. e/CONJ se/CONJS vós/PRO comeis/VB-P do/P+D zelo/N ,/, o/D vosso/PRO$ zelo/N converte-se-vos/VB-P+SE+CL em/P substancia/N ./

[3] Barbosa acknowledges this fact. In order to explain it, she claims that while in Spanish there is only one pre-verbal A'position (Spec/IP), EP also uses  Spec/CP  as a landing site for operators.  From this point of view,  two parameters are required to explain the two differences observed between Spanish and Catalan on one side and Portuguese on the other side. 

[4]  In Galves 2000, I propose that this category is Person.

[5] It is beyond the limits of this paper to present a deeper argumentation with respect to this point. From a diachronic point of view,  there are some facts which evidence the existence of a change affecting the pronunciation of words in European Portuguese at some point of the second half of the 17th century.    It is a well-known fact about the Modern Portuguese pronunciation that there is a very strong tendency to reduce non stressed vowels, in pre-tonic as well as in post-tonic syllables. It is generally admitted that Brazilian Portuguese in which vowel reduction productivity is limited to post-tonic syllables is closer to the pronunciation of Classical Portuguese.  In 1883, the Portuguese phoneticist Gonçalves Vianna already complained about the fact that his Contemporaries were no more able to read classical poetry with an adequate rhythm because they failed to pronounce  pre-tonic syllables. The first  reference we find in the literature to the reduction of pre-tonic syllables is the Petite Grammaire portugaise, quoted by Teyssier 1987, which was  published in Paris in 1675, which mentions the variation between the pronunciation [morar] and [murar] for the verb morar ("to live"). The 18th century grammarian Jerônimo Soares Barbosa, born in 1737, whose Grammatica Philosophica da Lingua Portuguesa was first published in 1822 mentions the pronunciation [dif'rente] for the word diferente (op. cit. p. 24). 

The attested emergence of the reduction of pre-tonic vowels can be related with a more general change in the way rhythm is implemented in the language. There is evidence among languages that the languages which have strong vowel reduction processes and complex codas are languages which organize their post-lexical phonology with respect to the prosodic word rather than to the phonological phrase (Cf. Kleinhenz, 1997). If we make reference to a more popular rhythmic typology, we can say that the phonological change which occurred from ClP to EP at  the end of the 17th  century and affected the realization of pre-tonic vowels can be described as a change from a syllable-timed language, with bound feet, to a stress-timed language with unbound feet, and a p-word based rhythm.

 

5 It could be argued that this condition is too strong since it would block the projection of  Focus in embedded clauses.  If we assume the complex C layer proposed by Rizzi (1997),   Foc is part of the C system,  and therefore it is not selected by another  member of this system.

6 This analysis supports Haegeman's claim that the typology of positions for NPs should integrate both the A/A-bar and  V-relatedness  dimensions.