GrammarSoft ApS |
|
||||
|
|
Eckhard Bick
PORTUGUESE SYNTAX 3. Clause level functions Clause level arguments (valency governed) The functional pivot of most clauses is a verbal constituent (V), also called predicator (P). Complex verb chains can consist of both main verbs (MV) and auxiliaries (AUX), linked by a dependency relation, and possibly by an auxiliary subordinator (SUBaux). For the sake of simplicity, we will here stick to single verbs, and treat complex predicators in another chapter. In Portuguese, there are four main types of clause level arguments, the subject (S), objects (O), argument adverbials (A) and complements (C). Objects are subclassified according to pronominal case, argument adverbials and complements as to whether they relate to the subject or - if present - to the direct object. In the examples, complex constituents are "united" by underlines. The different types of arguments in the examples can be distinguished by pronominal substitution: S (subject) demands nominative case when pronominalized (eu, tu). The subject has person and number agreement with its clause's finite verb (or, possibly, leading infinitive). Oacc (direct or accusative object) demands accusative case when pronominalized (o, a, os, as): "Trouxe-o". Both S and Oacc can be pronominalized with "o_que". Opiv (prepositional object) is always a pp [prepositional phrase] and demands prepositional case (also called oblique or prepositive: mim, ti) when the argument of its preposition is substituted by a pronoun: "Gosta de ti." Adverbials can be pp's, too, but prepositional objects can be distinguished from argument adverbials (or adverbial objects, A) by the fact that they can't be replaced by adverbs, and from adjunct adverbials by the fact that they are valency bound (cp chapter 3.2). Odat (pronominal dative object) is the function assigned to the pronominal form 'lhe'. Dative objects typically occur as optional number 2 object in the presence of a number 1 direct object (Oacc): "Lhe (Odat) dá um presente (Oacc)." Lhe alternates with pp-objects introduced by the prepositions 'a' or 'para', which is why such pp's could be regarded as dative objects, too: "Dá um presente a ela (Odat)." The corresponding non-pronominalized construction is, however, (form wise!) indistinguishable from a prepositional object, and will here, for the sake of formal consistency, be tagged as such: "Lhe ajuda" (Odat) - "Ajuda a ele" (Opiv). C (complements or predicatives) can be substituted by either "tal" or "isto", but normally not by personal pronouns: "Parece tal" (Cs) , "O elegeram isto" (Co) . Complements differ from objects in complementing both the clause's main verb and its subject (Cs) or direct object (Co). That's why they are called predicatives - like adnominal modifiers, they predicate something of a noun, while the clause's main verb is reduced to a kind of connecting device (called copula for the Cs-predicative) without much semantic content of its own. For focusing, Cs can be fronted, while Co can't (rico [Cs] não é - *engraçado [Co] não o acho). Adjectives and participles with predicative function have number and gender agreement with their nominal referent, Cs with the subject, Co with the object. A (argument adverbials or adverbial objects) can be substituted by an adverbial pronoun: "Viajará lá." (As), "Pôs a metralhadora lá." (Ao). Like complements (C), some argument adverbials (A) can be distinguished with regard to subject or object connection (As and Ao). Place and direction adverbials, in particular, "feel" very "predicative": "Mora lá" (As or Cs?), "Colocou-o lá" (Ao or Co?), and the same is true of "Está bem" (As or Cs?). Still, in all three cases we will folow the adverb substitution test and settle for the adverbial function tag (A). A very special case are the measuring verbs durar [7 horas], custar [7 coroas] and pesar [7 gramas]. Superficially, the arguments of these verbs seem to ask for direct object function (Oacc), but both the accusative pronoun substitution test and the "o_que"-substitution test fail. Only substitution with "quanto"/"tanto" works, and in the framework of this grammar, we will opt for an A analysis (argument adverbial), adding "quanto" - at least where it doesn't alternate with "o_que" - to the short test list of adverbial interrogative pronouns ("onde", "quando", "como"). Each Portuguese verb has a fixed set of valency patterns. The examples given concern "maximal valency", including both obligatory and optional complements:
(b) Durava muito tempo. (As:np) (c) Nadarmos regularmente seria bom para a nossa saúde. (S:icl) (d) Temia que não o conseguisse. (Oacc:fcl) (e) Quis saber quando voltaria o professor. (Oacc:fcl) (f) Está com febre. (Cs:pp) (g) O perigo a tornou
numa fera. (Co:pp)
<vq> cognitiv S (human) V que-conj Oacc:fcl (finite subclause) <v+interr> cognitiv S (human) V qu-word Oacc:fcl (interrogative subclause)
Clause level adjuncts (not valency governed) Adjuncts, while still being clause level constituents, differ from clause level arguments in that they aren't bound by verbal valency. We will mark adjuncts by a little f (for free) in the function symbol. Two main types will be distinguished here, adjunct (or free) adverbials (fA) and adjunct (or free) predicatives (fC) . Both functions (adverbial and predicative) also occur as arguments, i.e. argument adverbial (A) and argument predicative (or complement - C), respectively. Like their argument counterparts, all free predicatives and some free adverbials (especially place and direction adverbials) can be related not only to the verb, but at the same time to either subject (fCs, fAs) or object (fCo, fAo) . The difference between argument adverbials and adjunct adverbials, or between argument predicatives and adjunct predicatives, can be tested by the predicate isolation test, where "fazer" or "acontecer" is used to substitute for the predicate (the verb plus its arguments). Adjuncts (in italics) can be isolated from the verb, while valency bound arguments (in bold face) cannot. (b) Caíu no chão. -- *O que fez/aconteceu no chão? - Caiu. (A) (c) Trabalha no Rio. -- O que faz no Rio? - Trabalha. (fA) (d) Chegou no país depois da guerra. -- *O que fez/aconteceu no país? - Chegou depois da guerra. / O que fez/aconteceu depois da guerra? - Chegou no país. (A and fA) (e) Se tornou rico. -- *O que fez rico? -Se tornou. (Co) (f) Nadava nua. -- O que fez nua? - Nadava. (fCs) (g) O filhos cresceram
grandes e fortes. - *O que fizeram grandes e fortes? - Cresceram.
(Cs)
(i) Acha-a maravilhosa (Co). -- *Acha-a. (j) Acariciava o cavalo entre as orelhas. (fAo) -- Acariciava o cavalo. (k) No filme "Titanic" (fA), o jovem artista retratou a heroína nua (fCo). -- Retratou-la.
Note that there is a problem in using the predicate isolation test for fCo or fAo constituents (j-k) , since their link to the direct object may be enough to ensure test failure, - with or without verbal valency. fAs and fCs pass the test since they are linked to the subject which is outside the predicate for non-ergative verbs. fAo and fCo don't pass, simply because they lack their Oacc link. Therefore, the adverbial subject adjunct [fAs] in (l) may be detected and distinguished from the two argument adverbials in the same sentence, but not the adverbial object adjunct [fAo] in (m). (l) Veio de Portugal (As) para Brasil (As) num navio inglês (fAs). (m) Mandou a filha de Portugal (Ao) para Brasil (Ao) no barco do rei (fAo). Still, the fAo reading for no barco do rei can be defended on the (less formal) grounds that this constituent is a circumstantial manner adverbial and as such is more loosely linked to the verb than the direction pp's de Portugal and para Brasil, which match the semantics of the "transitive movement" verb mandar. For valency bound object complements, the semantic link between verb and argument is usually causativity: What the Co predicates about the Oacc, is only true by force of the verb, not in any independent way, as becomes clear from the translations of (n-p): (o) Acho a proposta ridícula (Co). [... that it is ridiculous) (p) No fim de semana, pintou a casa de azul (Co). [... such that it was blue] (q) Bebe o chá quente (fCo)! [... while hot) (r) Prefiro a sopa
forte (fCo). [... if strong]
Sometimes both adjunct and argument readings are possible after the same verb, suggesting two different readings: (s2) Ele surpreendeu-a
com um presente.
This sentence integrates the two statements she surprised him and he [object] was with another woman. In (s2), the pp is an adjunct - a free predicative -, and the meaning is he surprised her and he [subject] had a present. Adjunct function is not restricted to adverbials (A) and predicatives (C). Free (i.e. adjunct) objects (fO) are not entirely unthinkable. For instance, the concept offers itself where free dative objects (fOdat) occur with the semantic role of beneficiary with verbs that otherwise do not have a dative object in their valency pattern. Comprar, for instance, is an ordinary monotransitive verb, governing an (obligatory) direct object (Oacc) - still, an optional free dative object can be added: As a formal test for distinguishing between Odat and fOdat, substitution with an Opiv (or fOpiv) can be used. Ordinary valency bound dative objects prefer the preposition a, free dative objects prefer para: Lhe (Odat) deu um diadema. - Deu um diadema a ela (Opiv). Lhe (fOdat) comprou um diadema. - Comprou um diadema para ela (fOpiv). An argument for avoiding the concept of free objects altogether is the fact that the isolation test for adjuncts does not work as convincingly for fO as it does for fA: Lhe deu um diadema. *O que lhe fiz? (Odat) Deu um diadema a ela. *O que fiz a ela? (Opiv) Lhe comprou um diadema. O que lhe fiz (fOdat) Comprou um diadema para ela. O que fiz para ela (fOpiv) Lhe deu/comprou um diadema na loja. O que fiz na loja? (fA) Since bound and free objects of the same type (i.e. Odat and fOdat or Opiv and fOpiv) are not allowed to co-occur in the same clause (uniqueness principle), while adjunct and argument adverbials do co-occur (viajará para Londres [A] para comprar livros [fA]), we shall usually mark the adjunct-argument distinction for adverbials, but not for objects. Even whole statements can be adjuncted. Consider the following sentence: Here, the subclause complementizer (the pronoun o que) is relative not to a noun or np, but to a whole statement, yielding a kind of anaphor effect. In fact, one could split the sentence in two and rewrite it in the following way: Tristemente para ela, morreu o cachorro da velha. Finally, we will describe also vocatives as clause level predicative adjuncts, called vocative adjuncts (fCvoc): Salve-me, meu Deus! Desliga, amor,
que tem gente na linha!
A very special form of constituent is the agent of passive constituent in a passive clause, which in the corresponding active sentence is considered subject. Agents of passive pp's, then, appear to be a kind of "ex-subject"-argument. At first sight, they do not appear to pass the isolation test (for adjuncts): Foi convidado pelos sogros (1) pela primeira vez (2). (1) - O que aconteceu pela primeira vez? - Foi convidado pelos sogros. (fA) (2) *- O que aconteceu pelos sogros? - Foi convidado. (ARGpass?) The question is, however, whether we have applied the isolation test correctly. From a CG or dependency grammar point of view, the clause to test is not the whole sentence, but the participle clause convidado pelos sogros pela primeira vez, which functions as complement of auxiliary (AUX<, cp. chapter 5). Therefore, we should replace only the predicate of the AUX< subclause with a dummy (feito), and not try to include a higher level predicator (foi). The adapted test does, as it should, distinguish between fA (3) on the one hand, and arguments like Co (4) and Ao (5) on the other.
(2') - O que foi pelos sogros? - Foi convidado. (fApass)
Exercise: Identify clause level constituents! 2. Este livro parece interessante. 3. Achei o livro bem interessante. 4. Achei o livro na última hora. 5. Achei o livro na mala. 6. O livro caiu. 7. O livro caiu no chão. 8. Elena nadava. 9. Elena nadava no mar. 10. Elena nadava nua. 11. Chove. 12. A tartaruga nada. 13. Chegou um cliente. 14. A criança bebe leite. 15. Deu-lhe um presente. 16. A sua namorada está grávida. 17. Encontrou o país transformado. 18. O Rio de Janeiro se tornou um palco de desgraças. 19. No domingo 6, 200000 hinduístas demoliram uma mesquita na cidade de Ayodhya, no norte da Índia. 20. Ela andava muito assustada ultimamente. Syntactic function vs. semantic function In a more semantically oriented analysis, clause constituents can be assignet so-called case roles, as first proposed by Fillmore. The most common are:
As can be seen from the examples, a certain semantic role does not necessarily match the same syntactic function in different sentences. Subjects can be both agents, patients, experiencers and instruments, and arguments of passive can be both agents, experiencers and instruments (but not patients), depending on the semantic function of the subject in the active sister-clause. In some cases, interferences between morpho-syntactic (form, inflection) and semantico-syntactic (case role) criteria have lead grammarians to disagree on which function to assign certain constituents: Opiv's tagged as Odat Deu um presente de Natal à namorada (para a namorada). In this sentence the benefactive (BEN) constituent is a prepositional group (pp), and thus looks morphologically like a prepositional object (Opiv). Substitutability with a dative pronoun, however, as well as the benefactive case role itself support a dative object analysis (Odat). Não ama mais a mim.. O homem a quem amava desapareceu na guerra. In both sentences, the patient (PAT) constituent in bold face is a prepositional group (pp), but would by most analysts be regarded not as an Opiv, but as a direct object (Oacc). Substitutability with an accusative pronoun supports this analysis, as does the valency class (monotransitive) of the verb amar. Reflexive Oacc's tagged as S Consider the following, very divergent, examples of the function of the Portuguese reflexive pronoun se : (a) Com a lua subindo no céu, eles (AG) se (PATrefl) banharam num mar de prata. (b) Os dois (AG) se (PATreci) detestam (um ao outro). (c) Hector (PAT) tornou-se (-) um verdadeiro Robin Hood, defensor dos pobres. (d) Trata-se (-) de um livro que li o outro dia. (e) Entre a Dinamarca e a Suécia, se (PASS) constrói uma ponte gigantesca (PAT). (f) Cobram-se (PASS) mensalidades altíssimas (PAT). (g) Celebrou-se (PASS) o fim do ano (PAT) com toda animação. (h) Jamais se (EXP) soube como fugiram do forte (PAT). (i) Está-se (PAT) diante de uma crise econômica mundial (LOC). (j) Compra-se (AG) casas (PAT). (k) Carina (AG) se (BEN) permitiu mais um dia na cama (PAT). Morphologically, se is ambiguous between accusative and dative. Substitution with lhe shows that only in the last example can se be regarded as a dative pronoun, suggesting Odat analysis. So the easy analysis in all other cases would morphologically be accusative case and syntactically Oacc function ... or would it? (a) is the prototypical reflexive case, where se is a patient-object and refers to the same entity in the "real world" as the agent-subject. (b) is similar, with a patient-object, but se is plural and functions reciprocally, as can be shown by adding um ao outro. In (b) and (c) the verbs are so-called pronominal verbs (verbos pronominais) where the reflexive pronoun has no semantic function at all, but is incorporated in the verb as such: tornar-se - to become, tratar-se de - to be about s.th.. Still, syntactically, nothing seems to stand in the way of an Oacc-reading: Alternatively, in order to stress the verbs incorporating the pronoun, we could use a complex predicator with clause form: This way, the object complement (Co) turns into subject complement (Cs): (e') O governo (S-AG) constrói uma ponte (O-PAT). (f') O governo (S-AG) cobra altas mensalidades (O-PAT). (g') O governo (S-AG) celebra o novo ano (O-PAT). Now, though ponte (e), mensalidades (f) and novo ano (g) clearly are subjects (as can be seen, for instance, from the plural agreement between cobram and mensalidades) - they are patient subjects, as in the passive versions of (e'), (f') and (g'): (e'') Uma ponte (S-PAT) é construída. (f'') Altas mensalidades (S-PAT) são cobradas. (g'') O novo ano (S-PAT) é celebrado. Therefore, though syntactically Oacc, se in (e-g) functions semantically more like a passive marker. In some cases, however, neither a reflexive, pronominal verb or passive analysis will work. Consider (h) and (i). Estar, in (i) cannot take direct objects (Oacc) at all, souber, in (h), asks for experiencer - not clausal - subjects (EXP) in active clauses, and neither (i) nor (h) can be replaced by ordinary passives: *Foi sabido que ... *É estado diante de ... Tagging se as subject (S), solves all these problems at once: One could say that, in the evolution of the Portuguese language, se is slowly advancing from reflexive object - via passive marker for patient subjects - towards the semantic space reserved for impersonal pronouns in other languages, like si in Italian, on in French, one in English and man in Danish. Of course, such a process does not happen over night, which is why agreement restrictions are still strong with regard to the "ex-subject" (now Oacc) in such sentences, allowing - in most cases - an alternative, more conservative, analysis of se as Oacc: Performance is, however, gaining fast on competence, - to use Chomskyan terms, and singular se-predicators in connection with plural nouns or np's do occur, forcing an agent subject reading on se: Here, casas cannot be subject for agreement reasons, so se fills the empty space - at the same time allowing casas to become direct object (Oacc) without breaching the uniqueness principle (which forbids two - unco-ordinated - direct objects in the same clause). |