Eckhard Bick


PORTUGUESE SYNTAX


6. Group forms and group level constituent function

Groups (or phrases) are here defined as syntactic constituents that are not clauses, and consist of more than one word. In order not to be clauses, none of the group node's immediate constituents must be a predicator or a complementizer (subordinator). Every group features 1 head (H) and one or more dependents (D), which may be either modifiers (Dmod) or arguments (Darg). In this text, we will denote all kinds of group level dependents with the umbrella term adjects.

As on the clause level, arguments are valency bound. Modifiers are on the group level what adjuncts are on the clause level - they are "free" constituents without valency slots. The word class inventory of a group's head and dependents defines the group's form category. Accordingly, 4 main types of groups can be distinguished:

group type typical heads typical adjects (modifier or argument dependents)
np noun phrase noun
proper noun
"substantival" pronoun		 adjective
article
"adjectival" pronoun
ap adpositional phrase adjective
adverb
"adjectival" pronoun		 adverb
pp prepositional phrase preposition noun
"substantival" pronoun
icl
fcl
vp verb phrase verb
(main verb [semantically] or 1. auxiliary [dependency])		 verb
(auxiliaries [semantically]
or 2./following verbs [dependency])
preposition or the conjunction que

Groups can be classified in yet another way, as hypotactic (endocentric) , katatactic (exocentric) and paratactic. Hypotactic groups (np- or ap-type groups) can semantically be substituted by their head, which is not true of katatactic groups (pp-type). Some grammarians even base their definition of head on a group being hypotactic according to this substitution rule - thus, a pp would not have a head at all, since none of its constituents can stand for the whole group. Verb groups (vp) are the most special of all: they are hypotactic in the sense that the main verb can semantically replace the whole vp, but in dependency and valency terms, it is the main verb (or a subordinated AUX<:icl), that is an argument of the auxiliary, not the other way around. A paratactic group consists of two co-ordinated constituents, usually of the same form type, that share a common function label. Parataxis will be discussed in detail in chapter 9 (Co-ordination).
symbol category examples
np np
propp
pronp		 noun phrase
sintagma nominal
nominalsyntagme		 Era um homem como outro qualquer. (np)
A velha avó dormia na rede. (np)
Vou fazê-lo eu mesmo. (pronp)
O seu nome era Mário Moreno dos Santos. (propp)
ap adjp
advp
detp		 adpositional phrase
sintagma adposicional
adpositionssyntagme		 As árvores no jardim eram muito velhas. (adjp)
Foi um presidente um pouco iconoclasta. (adjp)
Nesta saia, parece mais jovem do que as amigas. (adjp)
Costuma falar muito devagar. (advp)
Ainda hoje vivem de caça e pesca. (advp)
Era muito mais vinho do que imaginava. (detp)
vp verb phrase
sintagma verbal
verbalsyntagme		 Ele continua mexendo nas tarefas dos outros.
Vem de lhes propor um acordo.
Temos que lhe dar mais dinheiro.
pp prepositional phrase
sintagma preposicional
præpositionssyntagme		 Abriu a janela da sala
Gostou do que viu.
Pedro da Silva
Mudamos para São Paulo.

6.1.

Noun phrases (np)

The prototypical np is headed by a noun, and allows a choice of adnominal adjects (group-level dependents), typically pre- and post-modifiers. Heavy (i.e. long or complex) modifiers and adnominal arguments appear only to the right of the head.

Typical premodifiers are:

articles: um dia, o Manoel

determiner pronouns (adjectival pronouns): todos estes livros

numerals: 7 nanos
 

 Typical postmodifiers are: adjectives: uma casa grande

prepositional phrases (pp) o jatinho do presidente

relative subclauses: o computador que comprou
 

 Atypical position often entails a change in meaning. Thus, adjectives in premodifier position become more "subjective", less "measuring" than in postmodifier position.

Compare: um grande homem (great) - um homem grande (big)

Only certain adjectives tend to occur in premodifier position. In some cases, grammarians even disagree on the word class of a prenominal modifier, because it looks like an adjective (and inflects like one), but functions like a (determiner) pronoun or numeral. Consider the following cases:

a primeira noite (numeral or adjective)

a última unção (adjective or numeral)

umas/algumas/várias propostas (adjective or pronoun)

a mesma cor, outra cor, diferentes cores (adjective or pronoun)

Determiner pronouns in postmodifier position are rare, comprising possessives and - with a change in meaning - todo. próprio and mesmo occur postnominally with independent "substantival" pronouns, that do not allow premodifiers:

responsabilidade sua (cp. a sua mãe - *a mãe sua)

a casa toda ([whole], cp. toda casa [every])

ele mesmo ([himself], cp. a mesma cor [same])

ela própria (cp. o próprio presidente, o próprio Cardoso)

"Adjectival" modifiers need not be individual words, but can become complex forms themselves (adjective phrases), as discussed in chapter 6.2:

Some nouns (titles, professions, family membership terms etc.) may have noun or proper noun (name) modifiers, and complex names may themselves be described as np's with a proper noun head and a chain of one or more proper noun modifiers to the right, yielding "left leaning" (read: head to the left) analyses for name expressions.

Name modifiers can be "isolated" from their head noun by other (adjectival) modifiers:

Not all names are simple chains of proper nouns - some integrate recognizable adjectives, numerals or prepositional phrases that one would want to mark as such:

In the last example, "name-hood" first arises on the np-level, since neither of the two constituents of the modifier np qualifies as a proper noun (prop).

One way of marking what's part of a name and what isn't, in terms of constituents, is by means of constituent bracketing. Thus, instead of marking a and Grande in a Grande São Paulo as "sister"-modifiers on the same level, one would first bracket Grande onto são Paulo, forming a complex head for the article a:

Noun phrases are much more likely to have modifiers than arguments, the latter occurring especially in connection with deverbal noun heads, i.e. nouns that have been derived from verbs. Arguments are here "borrowed" from the valency pattern of the concerning verb. proposta, for instance, can govern an argument replacing the direct object of the verb propor, and viagem borrows its valency from the argument adverbials of the verb viajar:

Deverbal nouns can borrow from their parent verb not only ordinary arguments, but also clause level adjunct adverbials, denoting time, space, manner, or an agent of a passive, turning fA-labels into DNmod-labels.

The prepositions that attach postnominal argument pp's to their head noun, cannot normally be exchanged, and have to be memorized individually in connection with the valency bearing noun: abertura para

afinidade a

confiança em

cumplicidade com

discussão sobre

escolha entre

lei contra

respeito por

semelhança com

temor de
 

 Finally, one might consider cardinal numbers after certain "counting nouns" as arguments: capítulo 7

páginas 8-12

século vinte

no dia 5 de julho 1998
 

Ordinal numbers, however, inflect like adjectives, and we will treat them as modifiers, even in postmodifier position. rei Alfonso III (terceiro)
  A special kind of adnominal adjects are appositions (DNapp) and adject predicatives (DNc) . Both are isolated from the rest of the np by a comma, and thus more loosely bound than ordinary modifier adjects. The difference between the two is largely semantic, but appositions are usually proper nouns or definite np's, while adject predicatives are adjectives, participles or indefinite np's.

Apposition adjects are identifiers and help define or denote the referent of the np in question, while adject predicatives are descriptive and act much like adjunct predicatives (fC) on the clause level. As a matter of fact, ambiguity as to whether a non-argument predicative attaches at clause or group level is not at all rare. Thus, predicatives left of the subject (a), or comma-separated predicatives to the right of the predicator (c) are clearly fC, while the same predicative "feels" more like a DN-constituent if found directly to the right of the subject (b).

(a) Contente com a vida, o hipopótamo dormia na água.

(b) O hipopótamo, contente com a vida, dormia na água.

(c) O hipopótamo dormia na água, contente com a vida.

One argument in favour of the DNc analysis for (b) is the fact that the constituent can be replaced by a DN relative clause (which cannot be moved to other positions in the sentence): O hipopótamo, que era contente com a vida, ... Note that it is the comma-separation between np-head and the DN dependent, that makes a DNc. For que-clauses, the distinction between DN (without comma) and DNc (with comma) equals the semantic distintion between restrictive (necessary) and parenthetical (non-necessary) relative clauses.

Finally, all np's can be modified by certain operator adverbs denoting negation (nem) , set inclusion (também, só) or avaliation (até) :

até ele, nem Pedro, só isso, dinheiro demais

However, in the presence of other modifiers, it becomes clear, that these "operators" don't mingle with other modifiers and it does not seem entirely satisfying to treat them as constituent-sisters of ordinary modifiers. Rather, they modify the whole np, as could be expressed by adding one bracket or tree level to our syntactic analysis, with the "inner np" as the complex head of a kind of "meta-np" (np') :

Further discussion:

One of the functions of the operator adverbs mentioned above is that of focus marker (cp. chapter 11 on focus-constructions). Focus marker dependents (Dfoc) put their head into focus, and they can be attached to heads of any form (x), - words, groups, clauses and compound units, generating a meta-constituent of the same form (x'). By using the Dfoc tag in these cases, we do not have to uphold the awkward distinction between DN, DA, DP and so on, for what appears only one type of function (Dfoc).

As a positive side effect, most independent ("substantival") pronoun groups (até ele, ele mesmo) become "meta-words", - which is more in line with one's view on independent pronouns as "unmodifiable" (cp. 6.4).

In order to avoid conflict with VISL's general definition of words and groups, we will, however, refrain from using terms like pron, n, adj etc. for focused individual words. Thus, ordinary group form categories (or, at most, np?, ap? etc.) will be used in these cases, too.

6.2.

Adpositional phrases (ap)

The term adpositional phrase will here be used to lump together adjective phrases (adjp) , adverb phrases (advp) and (adjectival) determiner pronoun phrases (detp) , all of which allow the same prototypical type of modifier - intensifier/quantifier adverbs, - and not much else. Thus, all three types of ap's can be premodified by the adverb muito .

Only very few postmodifiers can be attached to adjectives or adverbs, but the few are intensifiers/quantifiers:

In the last example, one might argue that mais and ainda form one disjunct constituent, as when both appear left of the ap-head, with an advp - not an adverb - modifying an adjective head:

With the (pre)modifying advp split into two parts of a disjunct constituent, we get the following analysis:

Though "adverbial" in function, the quantifier modifiers in an ap need not be adverbs proper, or even adverbial phrases (advp) , they can instead be borrowed from other form categories:

Determiner phrases are very rare, and restricted to quantifier modifiers, but adjp's and advp's do allow a few other - non-quantifying - modifiers:

manner adverbs: academicamente verboso

time operators: já morto

Like np's, ap's allow certain logical, set or modal operators as premodifiers:

Most modifier variation, however, is seen with "adjectival participles" , i.e. participles used as modifiers in noun phrases, not as part of a verb chain, because participles - even as modifiers - retain their parent verb's affinity for adjuncts and even arguments (the function of which could be attached to the DA-tag in small letters).

Due to the rich clause-like structure in participle ap's, one obvious alternative analysis is that of non-finite clause (icl) in stead of ap (cp. chapter 7.2.3.1):

Participle-based ap's are not the only ones to feature arguments. Adjectives can have a valency, too, the argument being a prepositional phrase (pp) introduced by a specific preposition dictated by the adjective's valency pattern.

Adverbs with a valency pattern are rare:

In the last example inclusive translates as including. There is another reading, equivalent to até (even), where inclusive is an operator adverb and functions as a (focus) dependent rather than as a head (cp. further discussion in chapter 6.1):

Some adverbs form "complex prepositions" (antes de, depois de) or "complex conjunctions" (antes que, depois que), that could be analysed as adverbs heading ap's with pp- or fcl-arguments, respectively. Another case are the comparative adverbs mais, menos, tão, tanto that valency-govern comparandum arguments (DAcom) in constructions like menos formosa do que uma hipopótama (cp. chapter 9).

6.3.

Prepositional phrases (pp)

A prepositional phrase is not hypotactic (or endocentric), like np's and ap's. Rather, it is katatactic (or exocentric), as none of its constituents can syntactically stand for the whole group. However, valency-wise it is the preposition that links the group to a head on the next syntactic level. Thus, it is a specific preposition that is governed and "asked for" when a verb, noun or adjective allows pp-arguments. Therefore, in dependency grammar, the preposition counts as head (H) of the pp, with the rest of the pp rolling as the preposition's [dependent] argument (DParg or, simply, DP) .

The argument position can be filled by almost any type of word class, group or clause, but most typically by np's and those word classes that qualify as np-heads, including infinitives and infinitive clauses.

passeava com a mãe (np)

discutiram sobre você (pron)

preparou-se para a palestra que ia dar o outro dia (np)

gostava de ler na cama (icl)

andava com medo de magoá-la (icl)

However, also adverbs (até hoje, para aqui) and finite clauses (sem que o soubesse) do occur as DP's.

pp's in general do not allow ordinary modifiers like np's and ap's, but only the kind of "operator adverbs" already mentioned in the last chapter, and - in a few cases - premodifying intensifiers. Both only occur as preadjects, and a simple analysis would treat them as "sisters" of the preposition's postadject argument.

Like with np's and ap's, one could argue against this "flat" analysis that ainda and nem are not modifiers of a preposition at all, but rather focus markers for the whole pp, introducing an additional bracket/level and making the kernel pp the complex head of a new form, a kind of "meta-pp" (pp'):

Likewise, intensifiers could be analysed as modifying the pp as a whole, creating an ap in the process:

By opting for analyses involving Dfoc's and DA's, the "real" pp is turned into a kind of complex head within a larger group, and cannot itself contain any dependents but the DParg constituent.

Functionally, pp's can be prepositional objects (Opiv), argument adverbials (A), adjunct adverbials (fA), or - on the group level - arguments or modifiers in np's (DNarg, DNmod) or ap's (DAarg, DAmod).

Não bate no pobre cachorro! (Opiv)

Mandaram-no para a Índia. (Ao)

Sem mais nada, venderam a velha casa, sede da família durante séculos. (fA)

Lhe deu uma coroa com oitenta jóias. (DNmod)

Era grande de corpo e coração. (DAmod)

Era legendária a sua capacidade de tornar em ouro o que tocava. (DNarg)

Escolheu uma senha quase impossível de lembrar. (DAarg)
 

 In a few cases, pp's can appear as predicatives (complements) , either on clause level (Cs, fC) or group level (DNc). Com setenta anos, não queria mais trabalhar. (fC)

Está com febre. (Cs)

Mário Goncalves, de Pernambuco, mantém que até tocou num extraterrestre. (DNc)
 

 With the exception of adjunct adverbials (fA) and adjunct predicatives (fC), pp's are almost always located to the right of their valency head. Like conjunctions, they add - and subordinate - new material with the linear flow of language, making syntactic tree structures "heavy" on the right hand side.

With regard to subordinating function, the prepositions com and sem are a special case. They can create a kind of clausal nexus without verbs, conjunctions or relatives. Consider:

(foi surpreendido) com o rostro na caixa pública

(foram fotografados) com todo mundo já seminu

sem ela para ajudar (não conseguiu nada mais)
 

 In these constructions (bold face), the preposition subordinates a clausal nexus, where a kind of predicative (na caixa pública, já seminu, para ajudar) is predicated of a nominal unit (np, noun, proper noun, independent pronoun) - o rostro, todo mundo, ela. The question is, do we use (a) a real clausal analysis and treat the nominal element as subject (S) and the predicative as subject complement (Cs) or adverbial (As), or do we (b) opt for a group analysis, with the predicative as a group-level DNc?

A group analysis seems more conservative, since it doesn't assign the preposition any unusual function and is structurally close to the even more conservative reading where the DNc tag becomes an ordinary attributive DNmod. Also, clause functions like S and Cs usually presuppose some verbal valency (a copula verb, for instance), -which just isn't there. On the other hand, clause functions like S and Cs are exactly what is needed to build a predicating nexus without a predicator.

As a matter of fact, treating prepositions as more than "pp-headers", as in (a), is not altogether uncommon. Particularly, few grammarians would treat prepositions in verb chains as heads of a pp, introducing tailor-made function categories like infinitive marker instead. Another, maybe more consistent view, is to regard these cases as subordinators, too, yielding the category of verb chain subordinator (SUBaux, as discussed in chapter 5).

There is a certain confusion as to the status of prepositions as a form or a function category, and one could argue that even the prototypical function of a preposition (as "pp-header" governing an np) is really one of subordinator ("SUBpp"). This "functional" view of the category preposition explains why many grammars treat conjunctions (que) or or relative adverbs (como) as "prepositions" if they head a comparandum with an np body: pior que isso, bela como a tia.

6.4.

Pronoun phrases

Pronoun phrases can be divided into determiner phrases (detp) and independent pronoun phrases (pronp). Detp's are a subclass of ap's and the few determiner pronouns - possessives and quantifiers - that allow dependents, take modifiers of the intensifier/quantifier type, like adjp's and advp's:

Foram tão poucos os comunistas no país que nem conseguiram lançar um partido.

Agora, a casa era inteiramente sua.
 

 Interestingly, detp's are usually formed with clause level function (Cs, S), not in the prototypical place of a determiner pronoun (prenominal DN).

Pronp's are treated as a subclass of np's, and have the same functional register as other np's, but they are heavily restricted as to their choice of modifiers, allowing only "operator class adverbs", and - for personal pronouns - mesmo and próprio:

ela mesma

até você

nem isso
 

 Using np terminology, we get the following type of analysis:

Note that "real" attributive adnominals for personal pronouns in Portuguese are circumvented by adding a de which turns the semantic head of the constructions into the argument in a syntactic pp-dependent, as in: pobre de mim. The respective analysis (a) of the whole group as an ap ("poor of me-type") is, however, awkward on the clause level. Therefore, in order to preserve DA-status for poor ("poor me"), an np-analysis (b) with a complex pp-head (de mim) might be preferable:

Further discussion:

One could be tempted to argue that the pronouns in these examples aren't really modified, but focused (as discussed in chapter 11 and 6.1), with the resulting constituent being a meta-word rather than a group:

We will not here pursue this line of thought any further, since the concept of multi-word "meta-words" is in conflict with the VISL-system's general definition of words and groups.

Article groups, finally, are happily non-existent in Portuguese.

symbol category examples
H - D head <-> dependent
núcleo <-> dependente
hoved <-> dependent		 uma grande árvore
sem dinheiro
devagar demais
D DN
DNmod
DNarg
>N, N<		 adnominal adject
adjeto adnominal
adnominaladjekt
(H: noun or pronoun)		 o (1) seu (2) grande (3) carro novo (4) (modifiers)
a (1) mulher do amigo (2) (modifiers)
um tanto (modifier)
cacique Jerônimo (modifier)
Manoel Neto (1) da Silva (2) (modifiers)
a proposta de lhe ajudar (argument)
combinaram a venda da casa. (argument)
predisposição para diabete (argument)
  DNapp
APP		 (adnominal) apposition
aposição (do substantivo)
[epíteto de identidade]
(nominal-) apposition		 O grande cacique, Jerônimo, conhecia o seu país como mais ninguém.
  DNc
N<PRED		 predicative adject 
adjeto predicativo
[epíteto predicativo]
preædikativadjekt		 Jerônimo, um grande cacique, temia ninguém.
com a mão na bolsa
  DA
DAmod
DAarg
>A, A<		 adverbial adject
adjeto adverbial
adverbialadjekt
(H: adjective, adverb or determiner)		 muito devagar (modifier)
devagar demais (modifier)
rico em ouro (argument)
receoso de lhe ter ofendido (argument)
  DAcom
KOMP<		 argument of comparative
complemento comparativo
komparativkomplement		 é mais bonito do que um hipopótamo.
  DP
DParg
DPmod
P<, >P		 argument of preposition
argumento de preposição
præpositionsargument
[styrelse]		 sem dinheiro nenhum (argument)
quase sem dinheiro (modifier)