Review of Focus Structure in Generative Grammar
Date: Fri, 25 Feb 2005 13:52:24 +0100
From: Judit Gervain
Subject: Focus Structure in Generative Grammar
AUTHOR: Breul, Carsten
TITLE: Focus Structure in Generative Grammar
SUBTITLE: An integrated syntactic, semantic and intonational approach
SERIES: Linguistik Aktuell/Linguistics Today 68
PUBLISHER: John Benjamins
Judit Gervain, SISSA Cognitive Neuroscience Sector, Trieste, Italy
"Linguists seeking the integration of information-structural categories into
syntax have a serious challenge to face and a number of strong arguments
to counter in order to maintain the viability of their program", claims
Polinsky (1999:58). It is precisely this challenge that Breul takes up in
his "Focus Structure in Generative Grammar". The work is an attempt to
integrate information structure into the core of phonological, syntactic and
semantic computations. The main theoretical stance of the book is that
pragmatics/information structure is not just another add-on to the more
basic symbolic computations of language, but part and parcel of it.
Breul takes up Lambrecht's (1994) theory of information structure, and
fleshes out its phonological, syntactic and semantic consequences. The
essence of Lambrecht's proposal, which Breul adapts, is that utterances
have one of the following three information structures: (i) thetic, found
discourse-initially, introducing a new state of affairs (1a); (ii) categorical,
introducing a statement/predication about an entity (1b); or (iii)
identificational, determining what is the entity about which a statement
(1) [Breul 2004:1; capitals indicate primary pitch accent]
a, Q: What happened?
A: My car broke DOWN.
b, Q: What happened to your car?
A: It broke DOWN.
c, Q: I heard that your motorcycle broke down?
A: My CAR broke down.
In the spirit of the general integrative commitment of the book, the basic
assumption is that one of these information structure types, called focus
structures by Breul, is always realized in the phonology, syntax and
semantics of _all_ utterances. Since the analysis is set in a generative
framework, Breul proposes to motivate this primarily in the syntactic
component; the phonological and semantic reflexes are then derived from
the syntactic representation. The main syntactic claim, termed the FocP
Hypothesis states that:
(i) All main clauses instantiate one of the three focus structures.
(ii) The focus structure is formally determined by the presence or absence
of a [+/-foc] property, i.e. feature in the minimalist generative terminology
(Chomsky 1995): the absence of the feature indicates thetic focus structure,
the presence of the [-foc] feature corresponds to categorical focus, and the
presence of the [+foc] feature confers identificational focus.
(iii) The [+/-foc] feature is accommodated into the structure of a main
clause through the Focus Phrase (FocP), a functional projection that
introduces the whole clause; it is to the specifier position of the FocP that
the lexical item to which the [+/-foc] feature is assigned by the conceptual-
intentional system move to check the feature against the Foc head. The
constituent that carries [-foc] is a topic, the one that carries [+foc] is an
identificational focus. Since the [+/-foc] feature is not present in thetic
utterances, these don't have a FocP, and they are directly introduced by an I
(iv) Movement to FocP can be overt or covert, depending on the strength of
the [+/-foc] feature in the given language. When the feature is strong, it
triggers overt movement, as e.g. in English.
One immediate consequence of the FocP Hypothesis is that the C
(omplementizer)P that is traditionally believed to introduce clauses is
replaced by the FocP, so all phenomena accounted for by the CP in previous
theories have to be re-analyzed in terms of the new functional projection,
FocP. Note that this is a non-trivial claim, since some of the most basic
structures, for instance questions, are usually derived via the CP. Moreover,
the CP is crucially involved in different sorts of embeddings. The author,
therefore, has to face the challenge of demonstrating the feasibility of the
FocP Hypothesis in syntax. In order to achieve that, he tackles the most
important of the issues raised by the replacement of CP with FocP, including
fronting, interrogatives, inversion with downward-entailing lexical items
(such as 'rarely', 'hardly' etc.), verb second and multiple fronting in matrix
and in embedded clauses. To give a flavor of the kind of FocP syntax Breul
proposes, the most important of these constructions, namely interrogatives
and embedding will be sketched out here.
To account for interrogatives in the FocP framework, Breul starts out from
the empirical observation that interrogatives and fronted topic/focus
constituents are in complementary distribution. Therefore, argues Breul, it
is reasonable to assume that wh-words (or the interrogative operator in
yes/no questions) occupies the same structural position as topic/focus
constituents, i.e. [Spec,FocP]. Consequently, the wh-words have to carry a
[+/-foc] feature in order to be attracted to FocP. In addition, to ensure the
interrogative interpretation, they also carry a [+int] feature that does not
get deleted in syntax (it's a weak feature that does not trigger movement),
but carries over to the logical form/semantics. Since there is only one FocP
in a clause, and it is the ultimate landing site of movement, it follows that in
every clause, there can only be at most one constituent that carries a [+/-
foc] feature, and thus moves to FocP. Thus the complementary distribution
of wh-words and topic/focus follows straightforwardly. Moreover, it also
falls out readily from the assumptions why only one wh-word can be
fronted in multiple wh-questions (*'Who what wants?'). Note that this
account does away with the usually assumed [wh] feature completely.
Another consequence is that languages in which the whP remains in situ,
and does not move to the left periphery of the clause, e.g. Chinese, can be
readily explained by simply assuming that in these languages the [+/-foc]
feature moves to FocP covertly, i.e. it does not bring phonological material
along. Auxiliary inversion ('do'-support) is accounted for by the
intermediate strength of the [int] feature on the Foc head. This feature is
not fully strong in English, so it cannot attract full verbs, but it has some
intermediate value, strong enough to attract auxiliaries.
As for embedding, previous accounts assume that the left periphery of main
clauses and embedded clauses is by and large similar, both being
introduced by a CP. In Breul's analysis, this parallelism breaks down, since
an utterance has a unique focus structure in a context, independently of
how many, if any, subordinate clauses make it up. In other words, in the
whole of an utterance, there can only be maximally one FocP, and it is
usually at the left periphery of the main clause, since most often that is the
clause that carries the information structural/pragmatic meaning.
Therefore, it is focus-structurally deviant for both the main clause and the
embedded clause to be introduced by FocP, thus embedded clauses are
usually CPs, not FocPs. However, there is no syntactic reason for this to be
so; and indeed, in special cases where it is the embedded clause that
conveys the focus-structural meaning and the main clause is only some
kind of an adsentential element, embedded clauses can be introduced by a
FocP. The author illustrates this general point through the discussion of
several different constructions, such as relative clauses, embedded
interrogatives (indirect questions) etc. Let us only take a quick look at
indirect questions here. In the default case, indirect questions are
pragmatically not real questions, so they are not introduced by a FocP, and
no lexical item carries a [+/-foc] feature in them. However, the whP still
moves to the left periphery of the embedded clause, but there is no [+/-foc]
feature to motivate this. Instead, claims Breul, the [int] feature, which does
not trigger movement in main interrogatives, is strong in embedded ones
and thus motivates the movement of the whP to the embedded CP. This
account, therefore, treats main and embedded interrogatives radically
differently--a rather unwelcome proposal.
As already pointed out above, the [+/-foc] feature of the Foc head is not
deleted during the syntactic derivation, rather it survives until the
phonological and the logical/semantic interfaces, and enters into
computations there as well. In phonology, the feature determines the
assignment of the primary pitch accent of the utterance according to the
Focus Projection Principle, which Breul basically adapts from Selkirk (1995).
One lexical item is marked for focus, i.e. F-marked, which may project to
the phrases that contain it, following the two basic principles of Focus
Projection, namely that (i) the F-marking of a head licenses the F-marking
of the head's projection, and (ii) the F-marking of a complement licenses
the F-marking of its head. F-marking is interpreted as the primary pitch
accent of the utterance. In the three focus structures, the primary pitch
accent as determined by the Focus Projection Principle is placed in the
(i) In a thetic sentence, the whole IP is licensed as intonational focus.
(ii) In a categorical sentence, the complement of the Foc head contains or is
a phrase that is licensed as intonational focus.
(iii) In an identificational sentence, the specifier of the Foc head contains or
is a phrase that is licensed as intonational focus.
It follows from the above, then, that [+foc] featured phrases that move to
[Spec,Foc] have to be licensed as intonational focus, i.e. be F-marked, or
have to contain a phrase that is licensed as such, while [-foc] featured
constituents also move to [Spec,Foc], but cannot be licensed as intonational
focus, nor can they contain such a constituent. A pitch accented constituent,
however, is not necessarily F-marked and [+foc], because focus is not the
only factor to determine pitch accent. Activeness of reference and contrast
also have their own characteristic intonational patterns, which interact with
The semantic contributions of focus-structure are also significant. As
already hinted at earlier, Breul takes truth conditions to be always and
necessarily context/pragmatics-sensitive; no truth value can be computed
for an utterance without taking its focus structure into account. The input
for the semantic analysis is provided by the semantic features, which remain
in situ in their base-generated position even if their corresponding
phonological features move, i.e. when there is overt movement in the
surface form. The semantic interpretation proceeds in a basically
Montegovian model-theoretical, type-semantic fashion, with the additional
Davidsonian assumption that events and eventualities are also entities in
the universe. This additional property of the framework has two immediate
consequences. First, verbs subcategorize for an event argument in addition
to their usual syntactic ones (noun phrases, clauses, prepositional phrases
etc.). Second, the temporal/aspectual functional projections of a clause (e.g.
T(ense)P, VoiceP etc.) denote e-type entities (eventualities), not truth values.
The three different focus structures are semantically distinguished on the
basis of what relevance presuppositions they are consistent with and what
additional information they assert. For a thetic sentence, the relevance
presuppositions are null, and the whole utterance conveys a new assertion.
For categorical utterances, the presupposition contains the topic
expression, e.g. 'it'(='my car') in (1b), and the utterance makes an assertion
about this entity. For identificational utterances, the presupposition
contains the syntactic predicate, e.g. 'broke down' in (1c), and the utterance
makes an assertion about to what entities this syntactic predicate applies.
This focus-structural organization of utterances allows for the precise
definition of what semantic subjects and semantic predicates are. The
semantic subject is defined as the constant in the presupposition, and the
semantic predicate as the assertion made by the utterance.
The semantic effect of the [+/-foc] feature is that it determines the type of
the object that the phrase carrying the feature can denote. Since it is a
common observation that topics are generally referential, Breul argues that
[-foc] featured phrases necessarily denote e-type entities. Consequently,
phrases that are usually believed to be non-referential, e.g. quantified noun
phrases, indefinites etc., can only appear as topics (i.e. have a [-foc] feature
and move to [Spec,Foc]) if they are used exceptionally as e-type denoting,
that is referential, which accords well with the empirical data (4).
(4) [Breul 2004:394-396]
a, I was curious whether the _boys_ would make it to the station. In fact,
MOST boys WALKED to the station, but others didn't get there at all.
b, I was curious whether the _boys_ would make it to the station. #In fact,
EVERY boy WALKED to the station, so none even had to take a taxi.
c, I was curious whether the _kids_ would make it to the station. In fact,
EVERY boy got there in TIME, but most of the girls didn't make it.
The requirement of referentiality holds true for non-nominal topic
expression as well. This might appear as a problem for e.g. verb phrase
fronting (that Breul re-analyzes as PerfP/VoiceP fronting). However, given
the Davidsonian assumption that the temporal/aspectual functional phrases
denote e-type eventualities, the problem disappears.
Focus expressions, that is phrases carrying the [+foc] feature, on the other
hand, denote type shifted, generalized quantifier-like <,e> or
<>,>-type expressions /bear in mind at this point the
Davidsonian modification; the t-types have been replaced by e-types/.
Because in Breul's framework, semantics is supposed to work on semantic
features in their base positions, one very important set of phenomena,
scope ambiguities of different kinds (5), cannot be derived any more in the
usual quantifier-raising fashion, since there is no semantically motivated
raising of any kind at all.
(5) Some boy kissed every girl.
a, There was one boy who kissed all of the girls.
/the existential quantifier takes scope over the universal/
b, For each of the girls, there was at least one boy who kissed her (and this
can be a different boy for each girl).
/the universal quantifier takes scope over the existential/
Rather, what Breul proposes is that the string appearing in (5) is actually the
orthographic representation of two focus-structurally different utterances,
one that has categorical focus corresponding to the reading in (5a), and
another that has identificational focus structure corresponding to (5b). To
put it differently, there is no ambiguity, only the orthographic coincidence
of two discursively different sentences. The semantics nicely follows, since
the noun phrase 'some boy' is the semantic subject in (5a), and combines
with the predicate 'kissed every girl', which already contains the universal
quantifier, and thus can "take scope over it", while the existential noun
phrase is the semantic predicate in (5b), and takes 'kissed every girl' as its
subjects, with the universal "taking wide scope". Moreover, in (5a), 'some
boy' denotes an e-type entity, i.e. refers to a particular boy, whereas in (5b),
it denotes a generalized quantifier, ranging over the set of boys.
Notice that such a semantic analysis requires flexible type assignment, since
depending on the focus structure of an utterance, the same noun phrase
'some boy' can be assigned type e or <,e> / <>,>.
Breul is aware of this issue and makes a few unconvincing attempts to
refute Heim and Kratzer's (1998) arguments against flexible type
Since Breul's work is essentially the syntactic/phonological/semantic
formalization of a pragmatic/information structural phenomenon, the
question of cross-linguistic validity inevitably arises. In this respect, Breul
notes that some languages, e.g. Hungarian, have different functional
projections to accommodate the [-foc] and the [+foc] features, or in more
traditional terms, have different Topic and Focus phrases, and goes on to
give a somewhat more detailed analysis of some German syntactic
Overall, Breul's book is an important attempt at unifying pragmatics, syntax,
semantics and phonology. Such much-needed integrative works are rare,
and thus precious. The proposal described in the book is certainly a feasible
one, since the intuition has been very strong and ubiquitous in the field that
information/focus structure is a Janus-faced phenomenon, relating as
much to pragmatic as to syntax.
However, the way in which this proposal is developed in the book leaves the
reader with a certain feeling of unease and perplexity. One has the feeling
of seeing bits and pieces of a mosaic, but missing the big picture. Many
issues and phenomena are treated in more or less superficial ways, with a
lack of in-depth analysis. While this impression stems, to a large extent,
from problems of organization, exposition and style, some content issues
also arise. In the remainder, I will point out a few of them.
The most general problem is that Breul fails to relate his proposal to
previous work on the syntax of focus, and thus leaves his notion of focus
rather hard to interpret for the syntacticians. (He does give a detailed review
of what the different conceptions of focus mean with respect to information
structure/pragmatics, tough.) This negligence is problematic, since focus
does have an extensive literature in syntax, and ignoring it leads to
overlooking fiercely debated issues, which, nevertheless, pertain to the core
of the matter. One such issue is the debate between proponents of a
feature-based account of focus (e.g. Brody 1995 and subsequent work) and
advocates of an intonational account (e.g. Reinhart 1995, 1996, Szendroi
2003, 2004a,b). Since Breul's position is not totally alien to either side, it
would have been instructive if he had considered the question in some
This also leads to empirical problems in Breul's account. A sentence such as
(6) does not contain identificational focus in Breul's sense (only categorical),
whereas according to the standard account (e.g. E. Kiss 1998, Szendroi
2003), sentences of this kind contain the kind of contrastive focus that even
Breul himself qualifies as identificational in other constructions.
(6) I bought a new CAR (and not a new BIKE).
If sentences like (6) do indeed contain identificational focus, it is expected
that the focused constituent, 'a (new) CAR' moves to the left periphery,
which is exactly the case in so-called discourse-configurational languages
such as Hungarian. In English, however, contrastive focus is most often
conveyed by intonation (nuclear pitch accent on the focused element), and
not by leftward movement (Szendroi 2003). The question arises, then, why
is focus-structurally motivated leftward movement not more common in
English? Breul's answer is that discursive and processing constraints
preclude this configuration from occurring with a higher frequency. This is,
however, a particularly weak argument, since these constraints should be
universal. Nevertheless, the facts show that languages like Hungarian have
no problem whatsoever implementing focus-motivated leftward movement.
A further empirical issue arises in connection with Breul's treatment of
quantifier-raising/scope ambiguity. He argues that these two phenomena
do not exist. Rather the relevant sentences simply allow both an
identificational and a categorical focus structure, this is how their apparent
ambiguity emerges (recall the analysis of example (5)). The empirical
consequence of this account is that the string in (5) should have two
intonational patterns as in (7a-b), moreover that one of them, (7a) should
be associated with the meaning in (5a), while the other, (7b) with (5b).
a, Some boy kissed EVERY GIRL.
b, SOME BOY kissed every girl.
This does not seem to correspond to facts. There has been no claim in the
literature to this effect; neither have the informants I interviewed confirmed
this empirical pattern. Actually, Breul himself quotes counterexamples, and
observes that judgements are very delicate and conflicting in this regard.
It might also be repeated here that, as pointed out before, Breul's treatment
of embedded clauses, and embedded interrogatives in particular, is rather
All in all, Breul's book is an important and resourceful contribution for all
those who are specialized in syntactic, semantic or intonational research on
Brody, Michael (1995) Focus and checking theory. In Kenesei (1995), 31-43.
É. Kiss, Katalin (1998) Identificational focus versus information focus.
Language, 74, 245-273.
Heim, Irene and Angelika Kratzer (1998) Semantics in Generative Grammar.
Kenesei, Istvan, ed. (1995) Levels and Structures. Approaches to Hungarian
5. Szeged: JATE Press.
Lambrecht, Knud (1994) Information structure and sentence form.
Polinksy, Maria (1999) Review of Lambrecht 1994. Language, 75, 567-582.
Reinhart, Tanya (1996) Interface economy - Focus and markedness. In
Wilder et al (1996).
Reinhart, Tanya (1995) Interface Strategies. OTS working papers in
Szendroi, Kriszta (2003) A stress-based approach to the syntax of
Hungarian focus, The Linguistic Review, 20(1), 37-78.
Szendroi, Kriszta (2004a) Introduction: Focus and the interaction between
syntax and pragmatics. Special issue of Lingua, 114(3), 229-254.
Szendroi, Kriszta & Ad Neeleman (2004b) Superman sentences. Linguistic
Inquiry 35(1), 149-159.
Wilder, C., H. M. Gaertner & M. Bierwisch, eds. (1996) The role of economy
principles in Linguistic Theory. Berlin: Akademic Verlag.
| ABOUT THE REVIEWER:
ABOUT THE REVIEWER
Judit Gervain is currently a 3rd year PhD student at the Language, Cognition
and Development Lab, Cognitive Neuroscience Sector, SISSA, Trieste, Italy
under the supervision of Prof. Jacques Mehler. Her first degree is in English
Philology, French Philology and Theoretical Linguistics from the University
of Szeged, Hungary. She wrote her MA theses in English Philology and in
Theoretical Linguistics about focus-raising phenomena in Hungarian. She
has published several papers in this topic. Currently, she is working on
language acquisition. Her precise research topic is the acquisition of the
foundations of syntax in the first year of life. At the same time, she
continues to do research in theoretical linguistics, in the area of left
peripheral phenomena (focus and wh-raising etc.).