University of Southern Denmark
Portuguese -> Info -> Portuguese Syntax  Visual Interactive Syntax Learning  
Portuguese VISLSentence AnalysisEdutainmentCorporaDictionariesMachine Translation

Printer-friendly version


Eckhard Bick


11. Topic and focus constructions

Running text can be viewed as sentences linked by topic-focus (or thema-rhema) relations. Pragmatically, topic constituents tell us what a statement is about, focus constituents introduce new information, often about the topic. In the examples, topics are underlined, focus constituents are in bold face.

(a) Na beira do rio, dormia um hipopótamo (S).

(b) O hipopótamo estava contente (Cs) e sonhava com uma bela hipopótama.

(c) Hipopótamas eram raras (Cs) na região, mas finalmente tinha encontrado uma.

(d) Era muito bom (Cs) isso.

(e) Ela iria amansar a sua vida e lhe dar filhos fortes (Oacc).

According to Togeby (1993, "Praxt"), focus is the last sentence constituent, that is not definite. Topic material, by contrast, is normally known in advance (from the last sentence, or from extra-lingual context), and will therefore appear in definite form, as names, pronouns or generic terms.

Given the linear nature of language flow, topic constituents are likely to appear left, focus constituents right, as is the case in all examples but (d). In (d), the focus is a subject complement (Cs), as in (b) and (c), and it is located in the usual Cs-position, immediately to the right of the predicator, - so one could say that it's the subject (isso), that is placed in a special way (as opposed to isso era muito bom).

In copula-constructions, subjects are usually definite and topicalized, while the subject complement is indefinite and focused. However, there is no general link between subject and topic, neither for intransitive verbs (a) nor for transitive verbs (f):

(f) A pequena Maria passeava na mata. De repente, um lobo comeu a menina.

Intransitive verbs with agent subjects like dormir or trabalhar, usually place topic subjects to the left (o hipopótamo dormia, 1) and focus subjects to the right (a). Note that the topic subject works fine with only a verb for focus, while the focus subject is not content with a verb as "topic" (2a, 4a-b).

Ergative verbs with patient subjects, like cair or morrer, allow both topic-less focus constituents (2b) and topic subjects (2c) to the right, i.e. after the verb. Focus constituents left of the verb (4), however, without a syntactic topic, sound awkward (though not outright agrammatical) in all cases:

(1a) O pássaro dormia. --- (1b) O pássaro morreu.

(2a) ?Dormia um pássaro. --- (2b) Morreu um pássaro.

(3a) ?Dormia o pássaro. --- (3b) Morreu o pássaro.

(4a) ?Um pássaro dormia. --- (4b) ?Um pássaro morreu.

In order to make possible focus fronting, Portuguese as well as English and Danish uses cleft sentences rather than type (4) sentences:

(4') Era um pássaro que dormia. --- Foi um pássaro que morreu.

In these constructions the focus constituent is moved out of the original clause as a subject complement of a copula verbs (mostly ser), while the original clause seems to become a subordinated subclause, with a relative pronoun as "dummy" for the removed constituent. Syntactically, the finite subclause could either be analysed as an "absolute relative" (a), or as a postnominal adject (DN:fcl) of pássaro (b). The ambiguity seems to enjoy morphological support, since there is a difference in verbal inflection between (a) Fui eu quem o fez and (b) Fui eu que o fiz. In (a), quem governs 3.person singular (fez), suggesting an independent fcl constituent (S:fcl). In (b), que governs 1.person singular (fiz), suggesting closer ties with eu (DN:fcl):



However, neither (a) nor (b) is fully satisfying from a semantic point of view. While the intended meaning is it was a bird that died, analysis (a) literally means something like what died was a bird, while analysis (b) translates as it was a bird which died (rather than another kind of bird). (a) is obviously closer to the intended meaning than (b), but the supposed S:fcl fails the o que/quem-substitution test for "true" absolute relatives:

(O) que morreu foi um pássaro (true absolute relative in S:fcl)

Foi um pássaro (?o) que morreu (different meanings)

Somos nós que/*quem o queremos (agrammatical with quem)

Thus, que in our sentence cannot be replaced by o que like in ordinary subject clauses.

Also, while the relative pronoun analysis works more or less for the focusing of nominal constituents, it seems unsatisfying for pp- or adverbial constituents:

In this example, it is an Opiv constituent that has been focused, stranding gostava without a prepositional filler for its valency slot inside the subclause. Even if we accept que as a complement of gostava, it is hard to see how a nominal pronoun can be placeholder for a pp. And semantically, it doesn't make sense that what he liked most "consists of meat" - which is the literal translation that matches the syntactic analysis given. To even get a semantically viable type (a) analysis, we have to introduce crossing branches or discontinuous constituents into the tree:

With an adverb in focus it becomes even more difficult to treat que as a relative pronoun - we would need to add que to the short and otherwise closed list of adverbial relative pronouns. More natural seems an analysis of que as subordinating conjunction (conj-s):

Still, like in the other focus sentences we have analysed, syntax does not match semantics - that we will get to know his answer is not today in the same way it could be certain or dubious. Also, the supposed Cs hoje is syntactically bound to the focus position left of the supposed subject fcl, while a Cs certo could be moved into the normal Cs position right of the subject:

Que vamos saber a resposta dele é certo.

*Que vamos saber a resposta dele é hoje.

A completely different approach to such cleft sentences is a flat analysis where the original constituent functions (from the non-cleft sentence) are maintained, and one regards e/era/foi/será and que as two parts of a syntactic "focus marker bracket" . In Constraint Grammar, this solution is straight forward:

In this notation, we have a focus marker head (@FOC>), with its arrow pointing towards the focused constituent (de carne, hoje), and a focus marker dependent (@FOC<), with its arrow pointing back at the focus marker head.

In a syntactic tree, this would translate into a disjunct focus marker constituent, which could be attached as dependents to the focused constituent as a whole, in the same way in which we have treated "operator adverbs" (não, ainda, até - cp chapter 6):

Gostava não de carne, mas de peixe. Come até rã.

Ainda hoje vamos saber a resposta. Hoje mesmo acabará.

Since the head of the focus marker constituent is a verb, we might regard its form - tentatively - as vp (verb group). Thus, we get:

That the focus marker in such sentences is, in fact, a hypotactic group with the 3. person form of ser as its head, is made plausible by the fact that the focus marker head verb can be used on its own, too (without que), especially in colloquial language:

Comeu foi o peru inteiro.

Gosta é de briga.

By attaching the focus marker as a dependent to the focused constituent, we get:

Interestingly, que seems to be necessary for closing the focus bracket only if the focused constituent is fronted (a). And que cannot be used without fronting (b,c), - unless the focused material is an np, allowing substitution with o que/quem and S:fcl analyses (a'). (a) Foi o peru inteiro que comeu.

(a') (O que) comeu foi o peru inteiro.

(b) *Que gostava mais, era de carne.

(c) *Que vamos saber a resposta, é hoje.

One could say that que in (a) functions as a syntactic cleft marker, separating the two finite verbs (foi and comeu), somewhat like a comma would, - which is not necessary in (a'), since foi in focus-fronted constructions has a position in the sentence that allows it to function as both focus marker and cleft marker.

In a way, que in cleft sentences is a subordinator that does not head a subclause, but merely chains two parts of the same (cleft) main clause. When introducing predicator vp's, we have assigned a similar "chaining function" (SUBaux) to the auxiliary particle que in verb chains after ter :

Here, too, we had the choice of syntactically showing the subordination - and creating a unique one-purpose constituent (auxiliary complement, AUX<,?que comer carne?) in the process -, or else of placing the subordinator on one syntactic level with the two verbs chained by it, as in the analysis shown here.

Similarly, we could keep the cleft-sentence constituent que at main clause level as a focus subordinator, SUBfoc:

In this analysis, we avoid the awkward stranding of a conjunction as rightmost part of a disjunct dependent, while still marking both focus scope (the head of Dfoc Era) and cleft position (SUBfoc que).

We have seen that the "natural" order of topic and focus is left-to-right, and that a focus constituent can be marked by fronting and/or clefting. Subject topics are usually fronted and thus marked anyway, but non-subject topic constituents can be marked by fronting, too:

(T1) O seu carro, vendeu para o vizinho.

Para o vizinho, vendeu o seu carro.

(T2) A princesa, sonhava com ela cada noite.

*Com a princesa, sonhava com ela cada noite.

(T3) À filha, prometeu-lhe um cavalo.

*À filha, prometeu a ela um cavalo.

Aos inimigos, matou-os todos.

A dinheirama, ganhou-a no jogo do bicho.

(T1) is pure topicalization, the direct object o seu carro or the prepositional object para o vizinho being fronted into the topic-position to the left, otherwise occupied by topic subjects.

(T2) is quite different, since the topicalized constituent, com a princesa, still has a - pronominal - place-holder in its old position (com ela). Also, the topicalized constituent (unlike T1), is "normalized" to nominative case (a princesa, not com a princesa), i.e. normalized to subject case. Mateus et. al. (1989, "Gramática de Língua Portuguesa") call this construction for pending topic ("Deslocação à Esquerda de Tópico Pendente"). Pending topics with other than "nominative" case (T3) are only grammatical with enclitic placeholders (-lhe, but not a ela, -os but not os inimigos).

While (T1) allows a straightforward syntactic analysis - (Oacc P Opiv) or (Opiv P Oacc) -, pending topics do not. In (T2), there is in the topic constituent a clash between form and function ("nominative" np vs. Opiv), and (T3) looks like a violation of the uniqueness principle.

Since we don't want 2 Opiv's anyway, a solution for the "nominative" topic in (T2) could be to view it as a free topic predicative of the place-holder Opiv:

In (T3), the breach of the uniqueness principle can be cushioned somewhat by analysing pending topic objects as free objects (fO) , i.e. as not valency bound. Also, the syntactic distinction between enclitic Odat and non-pronominal Opiv helps defend the uniqueness principle in the dative case:


Word classes (morphological form)

When discussing syntax, one does not usually pay much attention to the smallest building blocks of syntactic form, words. Word boundaries are viewed as stable and self-evident, with blank spaces being all that's needed to determine what a word is. Consider:

(1) galinha-d'angola

(2) dá-me-lo

(3) dar-me-iam

(4) O Porto, Estados Unidos

(5) do que (than), mesmo que (though)

(6) em vez de

(7) de maneira que

(8) fato de banho

From a "blank space perspective", 1-3 would be words, while 4-8 would not. However, semantically (1) and (8) both denote objects that cannot be dissembled. A fato de macaco is not a suit for apes! Also names (4) could be regarded as morphological words, as indicated by the lack of inflection and the capitalisation of nouns and adjectives inside a name string. On the other hand, the "words" in (2) and (3) incorporate pronouns - enclitic ("post-word") and mesoclitic ("in-word"), respectively - that need to be isolated as bearers of syntactic function (objects).

A glance at a list of Portuguese conjunctions reveals a majority of complex words like (5) and (7), that in theory would allow further syntactic analysis, but usually are regarded as smallest units of syntactic form. For complex prepositions (6) , the internal form of such units is typically that of adv+prp, or prp+np+prp, with the last preposition valency governing the complex preposition's argument.

Often, words are not assigned a form category at all, and word class is derived from syntactic function - as when calling como for all of adverb, conjunction and preposition, depending on its being used as adverbial, fcl-subordinator or acl-subordinator ("pp"). Similarly, adjectives are often assigned noun class when functioning as (head of) subject or object (um dinamarquês), and nouns are tagged adjectives, when they appear postnominally (um político pacifísta).

In this approach, Portuguese word classes will be regarded as form categories and defined primarily by morphological criteria, in particular by inflexion category inventory. Thus, nouns, proper nouns, adjectives and (cardinal) numerals feature number and gender, but while both these categories are word form categories for adjectives (i.e., freely inflecting for the individual word), only number is "free" in nouns, whereas gender is a lexeme category to be derived directly from the lexicon, i.e. fixed for the individual word. For numerals, the inverse is true: gender is a word form category (duas), while number is a lexeme category, i.e. fixed - as might be expected for semantic reasons ... In the case of proper nouns, finally, both gender and number are lexeme categories.

Pronouns can be subdivided morphologically by the same criteria: determiner pronouns (pron-det: estas, seus, cuja) inflect like adjectives, while independent pronouns (pron-indp: nada) inflect like proper nouns, i.e. they don't. Personal pronouns are morphologically special in that they also inflect for person and case. Articles should, along this line of reasoning, form part of the determiner pronoun class, but will be distinguished for reasons of grammatical tradition.

Finite verbs are easily defined morphologically due to their characteristic combination of inflection categories (person, number, tense, mode). From this perspective, it is the finite verbs that are the "true" verbs, while the non-finite verbal subcategories could logically be subdivided as "substantival" infinitives (inflecting for number and person, but not gender), "adjectival" participles (word form inflecting for both number and gender) and, finally, "adverbial" gerunds (without gender or number).

For adverbs (with the exception of -mente-adverbs), prepositions, conjunctions and interjections, morphological word class criteria are difficult or impossible to establish - barring closed word lists, and they might therefore be lumped together in one "particle" class. However, as a tribute to grammatical compatibility with other systems of analysis, the traditional categories (and syntactic definitions) have been maintained for these cases.

symbol category examples
n noun
substantiv (nomen)
árvores n(F P)
um oitavo n(<num> M S)
prop proper noun
nome próprio
proprium (egenavn)
Estados=Unidos prop(M P)
Dinamarca prop(F S)
adj adjective
belas adj(F P)
terceiros adj(<num> M P)
num numeral
duas num(F P)
17 num(<cif> M P)
v v-fin finite verb
verbo finito
finit verbum
(bøjet i tid)
fizessem v-fin(IMPF 3P SUBJ)
  v-inf infinitive
fazermos v-inf(1P)
  v-pcp participle
comprados v-pcp(M P) [attributive]
tem comprado v-pcp [verbal]
  v-ger gerund
correndo v-ger
art article
os membros art(<artd> M P) [definite]
uma criança art(<arti> F S) [indefinite]
pron pron-pers personal pronoun
pronome pessoal
personligt pronomen
mim pron-pers(1S PIV)
tu pron-pers(2S NOM)
  pron-det determiner pronoun
pronome determinativo
determinativt pronomen
(adjektivisk pronomen)
estas pron-det(<dem> F P) [demonstrative]
muita pron-det(<quant> F S) [indefinite]
cujos pron-det(<rel> M P) [relative]
quantos pron-det(<interr> M P) [interrogative]
minhas pron-det(<poss 1P> F P) [possessive]
  pron-indp independent pronoun
pronome independente
independent pronomen
(substantivisk pronomen)
isto pron-indp(<dem> M S) [demonstrative]
algo, nada pron-indp(<quant> M S) [indefinite]
os=quais pron-indp(<rel> M P) [relative]
quem pron-indp(<interr> M S) [interrogative]
adv adverb
facilmente, devagar adv [modals]
aqui, adv [pronominals]
muito, imensamente adv [intensifiers]
onde, quando, como adv [relatives or interrogatives]
não, até, adv [operators]
prp presposition
contra prp
em=vez=de prp <c>
in interjection
oi! in
conj conj-s subordinating conjunction
conjunção subordinativa
underordnende konjunktion
que conj-s
embora conj-s
  conj-c coordinating conjunction
conjunção coordenativa
sideordnende konjunktion
e conj-c
ou conj-c
pu punctuation
, pu [komma]


In order to continue using the Java applets, see troubleshooting tips and Download Java.
On Windows use Internet Explorer 11. macOS no longer supports Java applets.
The Chrome extension CheerpJ Applet Runner may work for some use-cases.

Copyright 1996-2022 | Report a Problem / Contact Us | Printable Version