Dative alternation in Brazilian Portuguese : typology and constraints

67 Dative alternation in Brazilian Portuguese: typology and constraints CHRISTINA ABREU GOMES UFRJ (Federal University of Rio de Janeiro) Abstract: The aim of this paper is to contribute to the discussion on dative alternation in natural languages. Dative alternation in BP shows four structural possibilities: VP[V SN1 PP]; VP [V PP SN1]; VP [V SN1 SN2]; VP [V SN2 SN1]. These configurations involve two independent but related processes: variable position of the complement within the VP and variable use of the preposition in the dative complement. These processes are independent in the sense that the use of the preposition is possible with both positions of the indirect complement. They are related in the sense that adjacency of the complement to the verb tends to favor absence of the preposition. We posit that the four observed configurations can be treated as a variable process in the sense of Labov (1972) and investigate the variable position of the complement within the VP and the variable use of preposition. For complement position, we examine competing structural constraints such as size, structural complexity, and certain semantic properties (material or perceptual transfer between complements) of both the accusative and dative complements as well as discourse level constraints such as given/new information in the dative complement. Diachronic data show that the synchronic configurations are the effect of a process of change that involves the decline of the IO DO order. Variable use of the preposition is constrained by adjacency and semantic properties of the verb. Dative alternation in BP exhibits characteristics related to increase of abstractness following the pattern proposed by Heine (1981). 1. Introduction The phenomenon known as Dative Alternation or Dative Shift constitutes a complex challenge in the search for linguistic universals. Languages exhibit two alternative syntactic configurations: the oblique construction [NP V NP PP] and/or the double object construction [NP V NP NP]. The conditions under which alternation occurs and, in particular, the structural, morphological, and semantic factors that constrain the use of the two orders are topics that have been explored in detail within different theoretical frameworks, ranging from generative to functionalist, and using data from several languages (English: Erteschik-Shir 1979, Barss & Lasnik 1986, Larson 1988, Jackendoff 1999, Groefsema 2000; Language Design 5 (2003, 67-78) 68 Christina Abreu Gomes Mandarin/Cantonese: Liejiong & Peyraube 1997; Greek: Kakouriotis 1995; Dolakha Newari: Genetti 1997). The aim of this paper is to discuss the linguistic factors that condition dative alternation in Brazilian Portuguese (BP). In BP, the indirect complement of a verb with both an indirect and a direct object (henceforth called ‘ditransitive’ verbs) can be preceded by a preposition or the preposition may be absent. In this text the traditional labels - dative alternation, dative shift, double object construction - are used even though the data in BP do not conform entirely to the classical patterns exhibited by the Germanic languages and other languages that have "true" dative shift. Spoken vernacular Brazilian Portuguese shows four structural possibilities: V NP1 PP; V PP NP1; V NP1 NP2; V NP2 NP1, as illustrated in (1) - (4) bellow: (1) Maria deu o livro a/para Pedro Mary gave the book to Peter (2) Maria deu o livro Ø Pedro Mary gave the book Ø Peter (3) Maria deu a/para Pedro o livro Mary gave to Peter the book (4) Maria deu Ø Pedro o livro Mary gave Ø Peter the book These configurations, not all of which are present in the standard language, involve two independent but related processes - variable position of the complement within the VP and variable use of the preposition in the dative complement. The two processes, which we examine in detail below, are governed by different constraints. Our results are based on quantitative studies (1) of data collected in two samples of spoken Portuguese in the city of Rio de Janeiro: the “Censo” Project, carried out in 1980, and the "Trend" sample of 2000. The purpose of the latter sample was to furnish data for the study of short-term real-time change in progress (2). Examples (1) - (4) are sufficient to show that Dative Alternation in spoken vernacular Brazilian Portuguese is different from other Romance languages, at least in their standard forms. This directly Dative alternation in Brazilian Portuguese... 69 contradicts the claim of Larson (1988:379), according to which Dative Shift is unavailable in the Romance languages in general. In Larson's terms, the preposition that introduces indirect complements (a 'to' in Portuguese) would assign Oblique Case in Romance languages, but Structural Case in English and other Germanic Languages. However, even though standard French and Spanish apparently both conform to Larson's generalization (see Kayne 1981, Larson 1988, Demonte 1995), vernacular Brazilian Portuguese does not - it exhibits an alternation between the oblique and double object constructions. Furthermore, I can see no principled basis why to "give Jim the book" in English should be analyzed differently from "dar Pedro o livro" in Portuguese, especially with regard to syntactic properties such as case assignment. Larson's claim is simply an unmotivated over generalization that contradicts the empirical facts of Brazilian Portuguese. 2. Synchronic study As shown in examples (1)-(4), the processes are independent in the sense that the use of the preposition is possible with both positions of the indirect complement. Nonetheless, they are related in the sense that adjacency of the complement to the verb tends to favor absence of the preposition. Quantitative analysis of synchronic data from contemporary Brazilian Portuguese reveals that deletion of the preposition is rare. As far as order of the complements is concerned, the structure V DO IO is the most frequent configuration for ditransitive verbs. It is important to note that since anaphoric direct objects (DO) tend to be null (Duarte 1986) in BP, adjacency between an indirect object (IO) and a ditransitive verb may be a consequence of the absence of a direct object. 2.1 Variable use of the preposition Examples (5) to (10) illustrate all of the relevant contexts for the variable use of the preposition that introduces the indirect object of a ditransitive verb. (5) E pede um comprovante ao presidente do morro né? Pede Ø o seu Aurino, pede um comprovante Ø ele (Censo sample - 1980) And (you) ask the President of the Slum for a receipt. (you) ask (0) Mr. Aurino, (you) ask for a receipt (from) him. 70 Christina Abreu Gomes (6) ai dei a ele o jogo (Censo sample - 1980) Then I gave him the game (7) Aí Jesus Cristo deu pra ele uma inteligência (Censo sample - 1980) Then Jesus Christ gave him intelligence (8) o garoto escreveu coisas lindas para o pai (Censo sample - 1980) The boy wrote beautiful things to his father (9) ensinar Ø o povo regras básicas de saneamento (Censo sample - 1980) To teach people basic sanitation norms (10) Ela levava a gente sempre Ø o quadro pra fazer as coisas (Censo sample - 1980) She always went with us to the blackboard to do things Examples (5) to (10) show that there are, in fact, three variants: structures with the preposition a ‘to’, with the preposition para ‘to’ or ‘for’, and with the null preposition. Dative alternation occurs only with verbs with a. In general, the action denoted by the verb involves a transfer, whether material or not (3). A comparison between the data collected in the two samples mentioned above reveals that para is currently replacing a in the spoken language. The preposition a is the formal variant, most frequently used in formal written and formal spoken language. The relative frequency of para in the Censo sample is 49%; in the Trend sample this figure increases to 85%. The examples also show that the three variants can occur in both positions of the IO; more specifically, the preposition may be absent, whether or not IO is adjacent to the verb and adjacency of IO to the verb does not necessarily imply deletion of the preposition. The synchronic data show that there is no context in which any of the four variants occur categorically. In other words, there is no context in which the use of only one or another of the variants under study is permitted. Pairs (5a) - (6), (5c) - (9), (7) - (8) reveal that the given/new dimension of the information contained in the noun phrase does not categorically determine which variants are used. Example (10) shows that the variable use of the preposition affects not only human IO’s in the goal or beneficiary role, but also complements indicating place. Thus, deletion of the preposition can occur in BP even when the prepositional phrase is not the recipient (4). Dative alternation in Brazilian Portuguese... 71 Examples (11)-(12) below exhibit sentences with light verbs (dar ‘give’ + NP) and a human indirect complement. These verbs are semantically empty and form a compound with the following NP. The data in (11) were collected in Censo sample. They show the categorical use of preposition a. But the data in (12), collected more recently in the Trend sample, show the spread of variation to a context that exhibited categorical presence of the preposition a in the past. They are submitted to the same process observed in examples (5) to (10) with full meaning verbs. (11) a. eles não dão muita ênfase a isso (Censo sample - 1980) They don't emphasize that very much b eles não dão atenção ao caso (Censo sample - 1980) they don't pay attention to that (12) a. pra dar continuação Ø o trabalho (Trend sample - 2000) in order to continue the work b. não deu muita inclinação para a parte intelectual (Trend sample - 2000) he didn't have much of a leaning toward intellectual things c. que eu dou muito valor a uma reunião (Trend sample - 2000) I think meetings are important As can be seen, the synchronic data provide evidence that oblique and double object constructions of ditransitive verbs with the preposition a in BP can be treated as variable alternants, constrained by structural and social factors. Since there is no categorical context for any given syntactic structure, they can be labeled as alternants from a sociolinguistic point of view and the syntactic and semantic constraints can be regarded as favoring or disfavoring each variant. Quantitative analysis shows that adjacency favors absence of the preposition, while non-adjacency favors its presence. Adjacency is also the context in which the para variant has the high frequency of occurrence. This distribution reinforces the assumption that the use of para is a strategy to avoid deletion, the stigmatized variant. The particle a is more frequently used when it is distant from the verb. The use of preposition para is also more frequent with human objects and full content verbs. Formal styles and higher educational levels also favor the use of a (see Gomes (1999). 72 Christina Abreu Gomes 2.2 The order of the complements Assuming that all of the observed configurations could be treated as a single variable process in the sense of Labov (1972), we investigated the variable position of the complement within the VP. We examined competing structural constraints on word order such as size(5), structural complexity, and certain semantic properties (material or perceptual transfer between complements) of both the accusative and dative complements as well as discourse level constraints such as given/new information in the dative complement. The data collected in Censo sample exhibit overwhelming occurrence of direct order (89%). Diachronic data show that the synchronic configurations are the outcome of a process of change that involves decline in the use of movement of the dative complement. Table 1 below displays the frequencies found in written documents from Brazil beginning in the 16th century. Table 1. Frequency of V OD OI order XVI XVII XVIII XIX Apl/Total 19/40 17/28 11/16 29/39 Frequency 47% 60% 68% 74% Examples (13) to (18) display a set of distinct contexts: (13) não tem um cara lá na Itália querendo dar um presente Ø o Papa? (Censo sample - 1980) Isn’t there a guy in Italy wanting to give a present (to) the Pope? (14) ensinar Ø o povo regras básicas de saneamento To teach people basic sanitation norms (15)Eu não sabia que você ia pedir o dinheiro a ele”( Censo sample) I didn't know that you were going to ask him for money (16) Aí dei a ele o jogo (Censo sample) Then I gave him the game (17) eu falaria com o João para me dar um emprego melhor Ø os meus filhos (Censo sample) I would tell John to give me a better job (to) my children Dative alternation in Brazilian Portuguese... 73 (18) Que davam condições de vida legal para o pessoal That they provided enough for a good life for them The pairs in (13) - (14) and (15) - (16) show that variable order is possible with both new and given information. Examples (14)-(17) show both possibilities for complement position within the VP, independently of the size of the IO. Finally, (14)-(18), in comparison with (13)-(17), are examples of different levels of phrase complexity of DO. Quantitative analysis (see Gomes et alli, 2002) reveals that direct order [NP V NP PP/ NP] is favored when the dative complement (with or without the preposition) is bigger than DO (more than 5 syllables), is more complex, and carries new information. Verbs of material transfer (dar o livro a Pedro ‘give the book to Peter’) also tend to favor direct order while verbal or perceptual transfer verbs (dizer a verdade a alguém ‘tell the truth to someone’) favor indirect order (6). 3. Further remarks There is currently a debate in the literature as to the structural nature of vernacular Brazilian Portuguese. One line of thought attributes typical variable features such as variable subject/verb concord to a postulated creolized stage of the language in Brazil (cf. Silva Neto (1951), Guy (1981, 1989) Baxter & Lucchesi (1997) Baxter, Lucchesi & Guimarães (1997), Lucchesi (2000)) , while another view attributes these same features to the operation of a drift within the language brought from Portugal (cf. Revah (1963), Naro & Scherre (1993, 2000)). In this context, we reflect on whether the situation described in this article should be ascribed to a possible Creole origin of Popular Brazilian Portuguese or whether is it might be the consequence of an internal drift. In other words, should dative alternation in BP be viewed as the result of contact of European Portuguese with African Languages which differ from Romance Languages in relation to the marking of dative constructions (7) or might it simply reflect internal grammatical possibilities? Evidence from Archaic Portuguese (11th to 15th centuries) shows that some of the variants involved in dative alternation (presence x lack of preposition and different word order) have been present for centuries in the grammar of Portuguese. Elsewhere, we have documented the lack of preposition and the presence of the preposition para in texts from the archaic period (8), as exemplified below: 74 Christina Abreu Gomes (19) "e daqui avante em cada mês ele va requerer [ ] todolos oficiães em presença de seus escriuães que lhe deçlarem se reçeberão mais algũa cousa ou he fecta alguma despesa da que aviam."(p. 19) and from now on each month he will require all officers, in the presence of his notaries, that they declare if they received anything else or if there was any expenditure from the ??? that they had Livro dos Conselhos de El-rei D. Duarte (Livro da Cartuxa), Lisboa, Editorial Estampa, 1982 (texts written between 1423 and 1438) The data were not submitted to quantitative analysis, but they permit an important conclusion about Portuguese grammar: the structural configurations found today are the reflex of structural alternants of the past. We have shown, based on both synchronic quantitative evidence and diachronic data, that dative alternation is within the structural scope of a Romance language. The alternation is better described as a sociolinguistic variable. The situation observed at the end of the 20th century can be viewed as the result of a tendency toward the crystallization of V OD OI order. We also observe the spread of use of the preposition para with increased abstractness in accord with the abstractness cycle proposed by Heine et al., 1981): PERSON > OBJECT> SPACE> TIME> PROCESS > QUALITY. The cycle initially proposed to explain unidirectionality in processes of grammaticalization can also be related to increase of abstractness in meaning. The a ~para ~ Ø alternation is a process of change in which a is being replaced by para in the coding of semantic relations in a continuum from [+concrete] to [+abstract]. Besides occurring with NP’s marked as [+animate], the use of para spreads to contexts in which a was earlier the only possibility or was preferred such as NP’s marked as [-ANIMATE, +ABSTRACT].. Notes (1) see Gomes (1999), (2000) and (2001). (2) The Censo da Variação Lingüística Research Project collected data on the urban dialect spoken in the city of Rio de Janeiro, Brazil, from 1982 to 1984 (cf. Oliveira e Silva, 1996). The sample consists of 64 speakers and is stratified for age, gender, and level of formal education. The sample recorded in 2000 contains data from 32 speakers. In both samples speakers are distributed according to different levels of the social parameters mentioned: a)- age: 7 - 14; 15 - 25; 26 - 49; 50 - 70; b) level of formal education: elementary school, junior high, high Dative alternation in Brazilian Portuguese... 75 school. The sociolinguistic interviews were collected by trained field workers using standard sociolinguistic methodology. The Trend sample is stratified in the same way as the Censo sample. (3) - material transfer: dar ‘to give’ (prototypical) - dar o livro a alguém ‘to give the book to someone’. Examples: emprestar ‘to loan’, ‘to borrow’, enviar ‘to send’. - verbal or perceptual transfer: dizer ‘to say, speak, tell, talk’ (prototypical) - dizer umas palavras a alguém ‘to say some words to someone’. Examples: ensinar ‘to teach’, falar ‘to say’, perguntar ‘to ask’, responder ‘to answer’. (4) Mazurkewich 1984 proposed that two contexts -- one morphological, and the other semantic -- for the double object construction in English. The morphological context distinguishes native from Latin origin verbs and postulates that only native verbs permit the double object construction : a. Peter told/*reported his boss the news; b. The professor showed/*demonstrated us his new methods; c. Paul got/*obtained his girl friend a ticket. There is an exception for Latin verbs with two syllables such offer and promise. The semantic restriction postulates that the entity to which the theme is transferred, that is, the recipient, should be capable of ownership. Places cannot exercise ownership: a.*I sent a letter to England/ * Peter opened the old lady a door. But it we replace England by Helen we get: b. I sent Helen a letter. In the a sentences, the IO indicates a change of position and not of possessor while Helen implies both. MAZURKEWICH,I. (1984) The acquisition of the dative alternation by second language learners and linguistic theory, Language Learning 34, p.91-109. Acho que todos estes detalhes do inglês são irrelevantes para o teu artigo e podem cansar o leitor. (5) Size was measured by number of syllables (6) A possible explanation is that verbal or perceptual transfer verbs tend to appear with sentential objects, which are always in final position in the VP. (7) According to Gonçalves (1996:47-48), in Bantu languages (Tsunga and Yao (Y)) there is no grammatical mark, whether or not the beneficiary NP that corresponds to an oblique construction in Portuguese occupies the position adjacent to the verb: