Show Summary Details

Page of

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, LINGUISTICS (linguistics.oxfordre.com). (c) Oxford University Press USA, 2016. All Rights Reserved. Personal use only; commercial use is strictly prohibited. Please see applicable Privacy Policy and Legal Notice (for details see Privacy Policy).

date: 21 November 2017

Pragmatics of Focus

Summary and Keywords

Generally speaking, ‘focus’ refers to the portion of an utterance which is especially informative or important within the context, and which is marked as such via some linguistic means. It can be difficult to provide a single precise definition, as the term is used somewhat differently for different languages and in different research traditions. Most often, it refers to the linguistic marking of (i) contrast, (ii) question-answering status, (iii) exhaustivity, or (iv) discourse unexpectability. An illustration of each of these possibilities is given below. In English, the focus-marked elements (indicated below with brackets) are realized with additional prosodic prominence in the form of a strong pitch accent (indicated by capital letters).

(i) An [AMERICAN] farmer met a [CANADIAN] farmer…

(ii) Q: Who called last night?

A: [BILL] called last night.

(iii) Only [an ELEPHANT] could have made those tracks.

(iv) I can’t believe it: The Ohioans are fighting [OHIOANS] !

The underlying intuition common to all these instantiations is that a focus represents the minimal information needed to convey an important semantic distinction. Focus can be signaled prosodically (e.g., in the form of a strong pitch accent), syntactically (e.g., by moving focused phrases to a special position in the sentence), or morphologically (e.g., by appending a special affix to focused elements), with different crosslinguistic focus marking strategies often carrying slightly different restrictions on their use.

Example (i) evokes a set of two contrasting alternatives, {‘American farmer,’ ‘Canadian farmer’}, and the meaning ‘farmer’ is common to both members of the set. That is, within this evoked set of alternatives, ‘farmer’ is redundant, and it is the nationality of the farmers which differentiates the two people. Example (ii) exhibits a similar property. One of the standard theories of question semantics represents questions as sets of possible appropriate answers. For (ii), this would be a set of propositions like {‘Bill called last night, ‘Sue called last night,’ etc.}. As with (i), there is an evoked set of meanings whose members share some overlapping semantic material. Within this set, the verb phrase meaning ‘called last night’ is redundant, and it is the identity of the subject that serves to differentiate the true answer. Example (iii) demonstrates a relationship between focus and certain words like only. The sentence means something like ‘of all the animals who might have made these tracks, it must be an elephant.’ As with (i) and (ii), this involves a set of alternatives: the set of possible track makers. That the sentence serves to single out a unique member of this set as being the true track maker makes the subject an elephant a natural focus of the sentence. Finally, in (iv), we see that focus on ‘Ohioans’ is being used to contrast the semantic content of the sentence with some preconception, namely that Ohioans are unlikely fighters of Ohioans. Examples (iii) and (iv) point to more specific uses of focus in different languages. In Hungarian, so-called identificational focus, which is marked syntactically, requires an exhaustive interpretation, as if a silent only were present. And in some Chadic languages, a meaning of “discourse unexpectability,” as in (iv), is required to mark focus via syntactic or morphological means.

Keywords: focus, contrastive focus, prosody, information structure, question under discussion, givenness

1. Overview

Focus refers generally to the linguistic marking of elements within a sentence or utterance, whether via prosodic prominence (as in English), syntactic position, or morphological affixation, that convey information that is prominent or important within the discourse context. Complete, precise definitions are elusive, because the term ‘focus’ has been used to describe several distinct but related phenomena in different languages and within different theoretical frameworks. For English and other related languages, the term is most often used to refer to a marking of contrast. The notion of contrast can itself be construed in various ways. After Rooth (1985, 1992), a contrast between two elements can be thought of in terms of a partial overlap in meaning, where the nonoverlapping material receives a focus feature. A famous example from Rooth, given in (1), illustrates. (Brackets around a word or phrase denote the assignment of a focus feature, often called F,’ which marks the constituent understood to be focused, with capital letters indicating a strong pitch accent.)

(1)

Pragmatics of Focus

Example (1) introduces a set of contrasting NP denotations, {‘American farmer,’ ‘Canadian farmer’}, and the meaning ‘farmer’ is a common semantic component of both members of that set. The nonoverlapping nationalities ‘American’ and ‘Canadian’ minimally differentiate each member from the other, therefore those two adjectives are marked as foci within their NPs, and both receive a strong pitch accent as a result. Refinements have been made over the years to restrict the space of possible alternative sets, in order to avoid the overgeneration of possible focus configurations.

Focus has also been presented as the reflection of a contrast at the level of information structure. Two popular approaches to information structure are the ‘information packaging’ approach of Vallduví (1990), which conceives of information structure as a sentence-level structure reflecting how semantic information is to be incorporated into the hearer’s knowledge store, and the ‘question under discussion’ (QUD) approach of Roberts (1996/2012), which conceives of information structure as a structure over the larger discourse context reflecting what is under discussion. The approaches are different, but overlap with respect to one basic idea: the information conveyed by an utterance can be partitioned by whether that information is part of some salient question or proposition which is expected to be addressed in the current discourse context. This is especially apparent in simple question-answer exchanges, as in (2).

(2)

Pragmatics of Focus

The question in (2) sets up a context where it is expected that the answerer will convey some information of the form ‘x called last night,’ and it is the identity of ‘x’ which serves to differentiate the intended propositional meaning from its alternative possibilities (e.g., ‘Sue called last night,’ ‘Pat called last night’). Rooth’s notion of contrast applies to (2) as well, because the question, whose meaning can be formalized as the set of possible answers (Hamblin, 1973), acts as a set of alternatives, such that {‘Bill called last night,’ ‘Sue called last night,’ ‘Pat called last night’} takes on the same role as {‘American farmer,’ ‘Canadian farmer’} for (1). The whole proposition, ‘Bill called last night,’ contrasts with the other potential answers to the question. On the other side, the traditional information structure approach to focus does not account for (1) so straightforwardly. However, one appeal of the information structure approach is that, unlike formal descriptions like Rooth’s, it aims directly at an intuitive explanation of focus based on language use in context: it seems natural that content which is less predictable given the state of the current discourse would be given more prominence, while content which could conceivably be recovered or figured out from context would be de-emphasized. Variations have also been proposed which center around the notion of givenness (Schwarzschild, 1999). Such analyses boil focus assignment down to a pressure to shift accent from or to de-accent (i.e., prevent the assignment of strong prosodic prominence to) elements which are salient to both interlocutors, or ‘given.’ Such approaches account for cases like (1) and (2), but more recent research (e.g., Wagner, 2006, 2012; Büring, 2016) has revealed that some notion of contrast is nonetheless required to avoid overgeneration.

Contrastive focus is known to associate with certain ‘focus-sensitive’ elements such as only, as in (3), where it is assumed that the lexical meaning of only makes reference to a set of alternatives from which a particular focus structure is derived.

(3)

Pragmatics of Focus

In (3a), only requires a set of alternatives such as {‘I talk to American farmers,’ ‘I talk to Canadian farmers,’ …}, where the use of only implies that the proposition that is conveyed is the only true member of that alternative set. In other words, the focused element picks out an exhaustive set of farmer types that I talk to. Contrast this with (3b), where the focused element picks out an exhaustive list of types of Americans. The set of alternatives necessary to interpret the scope of only also acts to license a particular contrastive focus. Beaver and Clark (2009) adapt Roberts’s QUD-based information structure approach to the semantics of focus-sensitive elements, proposing that the lexical meaning of only requires that the associated proposition be the maximally informative true answer to a (possibly implicit) question under discussion, for example, ‘what kind of farmers do you talk to?’ While many approaches agree that the correlation between focus and the interpretation of only is indirect, there are languages other than English in which an exhaustive interpretation as in (3) is conventionally associated with the use of focus itself, at least in some cases. For example, in Hungarian, there is a special syntactic position at the left edge of the clause for foci, but unlike English, movement to this position has been said by many to semantically entail an exhaustive interpretation, as if a silent only were present (Szabolcsi, 1981; Kiss, 1998; Szendröi, 2003; Kenesei, 2006; Kiss, 2007; Horvath, 2010; see Onea & Beaver, 2011 for a rebuttal).

Zimmermann and Onea (2011) note that there is no one-to-one mapping between formal definitions of focus and the crosslinguistic phenomena the term is used to describe. Hungarian is just one example where focus is thought to conventionally involve additional semantic content beyond the analyses based on English. Zimmermann (2008) examines syntactic and morphological focus marking in West Chadic languages and concludes that in those languages, focus conveys a meaning of ‘discourse unexpectability.’ While the varied crosslinguistic subtleties and differences in theoretical perspectives make a single exact definition of ‘focus’ perhaps untenable, there is a core pragmatic intuition underlying the different conceptions which goes back to early work on the subject by Halliday (1967): the portions of an utterance which are maximally informative or important for interpretation are linguistically marked as such.

2. Formal Accounts

2.1 Rooth’s (1985, 1992) Alternative Semantics

Under the ‘alternative semantics’ approach, focus is proposed as a feature in the syntactic derivation of a sentence which requires a c-commanding silent operator, ~ (often referred to as ‘squiggle’), to define the syntactic level at which the focus feature is interpreted. The focus feature, or F-feature, affects phonological form (PF) by requiring the assignment of prosodic accent within the focus-marked constituent, and the ~ operator affects logical form (LF) by introducing a presupposition requiring an appropriate antecedent in the sentence or discourse whose meaning is a member of the focus semantic value (FSV) of the syntactic constituent to which the ~ operator is adjoined. The FSV of a constituent is a set consisting of all (relevant) semantic formulae obtained by substituting an open variable of the appropriate semantic type in for the focus-marked element. To illustrate, consider example (1) again.

(1)

Pragmatics of Focus

The syntactic structure for a Canadian farmer is claimed to be roughly as follows.

Pragmatics of Focus

The focus feature on Canadian is responsible for its contrastive stress, and requires a c-commanding ~ operator for interpretation. The placement of ~ tells us to calculate the FSV of the NP node: first, we take the ordinary denotation of Canadian farmer—[λ‎x. canadian(x) & farmer(x)]—then, for the predicate ‘Canadian,’ which is contributed by the focused adjective Canadian, we substitute an open property variable P to obtain [λ‎x. P(x) & farmer(x)], which we can represent as a set of meanings {λ‎x. P(x) & farmer(x)}, which is shorthand for the set of NP denotations {‘American farmer,’ ‘Canadian farmer,’ ‘Japanese farmer,’ …}. By hypothesis, the presence of ~ adjoined to [NP CanadianF farmer] introduces a presupposition: there exists elsewhere in the sentence or in the discourse context an antecedent whose denotation is a member of this FSV. We see in (1) that this requirement is satisfied; the structure licensing focus on Canadian refers anaphorically to the antecedent American farmer, whose denotation is a member of {λ‎x. P(x) & farmer(x)}. If we then allow a similar cataphoric reference to Canadian farmer at the American farmer node, then both contrastive foci are licensed.

Alternative semantics can also be used to analyze focus in answers to questions as in example (2) above, reproduced below.

(2)

Pragmatics of Focus

After Hamblin (1973), the semantic denotation of the question Who called last night? can be represented as the set of possible answers, for example, {‘Bill called last night,’ ‘Sue called last night,’ ‘Pat called last night,’ …}. In this sense, the question, in virtue of having a wh-word in subject position, has a meaning identical to the FSV of any one of the possible answers with focus on the subject. Therefore, focus on the subject is always licensed. To focus only the verb, on the other hand, would fail to meet the interpretation requirements, because no antecedent of the form [last-night(P(bill))] can be recovered. A different context can change the acceptability of the pattern, as shown in (4).

(4)

Pragmatics of Focus

Also, alternative semantics offers a natural link between the semantics of certain adverbial elements like only and the focus patterns that correlate with them. Example (3) above illustrates the semantic effect of only.

(3)

Pragmatics of Focus

Rooth (1985, 1992) proposes the lexical meaning of only to be a function that takes a VP denotation and returns true iff that VP denotation is the sole, unique VP denotation in some contextually determined set of VP denotations, C, which is true of the subject. The focus structures in (3a) and (3b) have FSVs which suffice to supply a value for C. For (3a), the resulting interpretation is that of all (relevant) members of the set {‘talk to P farmers’}, only the predicate ‘talk to American farmers’ is true of the speaker. For (3b), the resulting interpretation is that of all (relevant) members of the set {‘talk to American Ps}, only the predicate ‘talk to American farmers’ is true of the speaker. The different focus structures change the assumed value for the contextual variable required to interpret only. This is just one example of a lexical item whose interpretation associates with focus placement. We see similar effects for a variety of lexical items, for example as in (5).

(5)

Pragmatics of Focus

Although these examples instantiate a simple one-to-one mapping between focus marking and the assignment of strong pitch accents, the mapping is more subtle and relies on independent parameters of prosodic phonology (see e.g., Selkirk, 1995; Ladd, 1996). For example, the strongest pitch accent within a phrase or clause is by default placed at the right edge of that phrase or clause in English. Also, function words tend to resist accent. Finally, it is clear that within longer constituents, there must be sufficient peaks and valleys, so that (6a) is not a valid intonation pattern, but rather something like (6b) is produced, where non-right-edge prominences are weaker than the right-edge, ‘nuclear’ accent.

(6)

Pragmatics of Focus

For this and other reasons, it is generally taken for granted that focus involves an abstract phonological representation which only partially determines intonation at the level of phonetics.

2.2 Schwarzschild’s (1999) Givenness

The examples above could alternately be described in terms of conditions on nonaccent rather than conditions on accent. This is the approach taken by Schwarzschild (1999), who, operating within an optimality-theoretic (Prince & Smolensky, 1993) framework for assigning prosodic accent, proposes the following constraints on accent placement (p. 156).

Pragmatics of Focus

F-features are considered to be purely phonological in this framework: The features themselves to do not induce any LF interpretation. Rather, F-feature assignment is constrained so that only elements that are given should lack an F-feature. A constituent is formally classified as given by considering its existential F-closure. The existential F-closure of a constituent is defined as the existential type shift—the proposition obtained by existentially closing a semantic formula after having removed the lambdas binding the variables in that formula—of the denotation of that constituent, with any dominated foci replaced by open variables. For example, the existential F-closure for CANADIAN farmer, with focus on Canadian, is obtained by taking the denotation of the noun phrase, λ‎x. canadian(x) & farmer(x), replacing the focused adjective denotation with an open variable to obtain λ‎x. P(x) & farmer(x), and then existentially type-shifting to get ∃P. ∃x. P(x) & farmer(x).

If the proposition resulting from the existential F-closure of a phrase with a certain focus structure is sufficiently salient in the discourse—that is, assumed to be in the consciousness of the interlocutors (Chafe, 1974)—then the phrase is given. The tree below for the CANADIAN farmer node in (1) illustrates the constraints. The existential F-closure at each node is given in parentheses, with nodes for which the existential F-closure is contextually salient in boldface. Assuming the context of example (1), each of the bold nodes counts as given, and only the adjective node, which is not given, must bear an F-marking. The ‘Givenness’ constraint dictates that the adjective node should bear an F-feature. The ‘AvoidF’ constraint dictates that no other nodes should bear an F-feature. The interface between F-features and the phonological system dictate that some accent must be assigned under the adjective node if it is F-marked, resulting in a stress on Canadian.

Pragmatics of Focus

The intuition underlying this formal analysis is that focus is assigned only to elements which the interlocutors are not already thinking about. In (2), for example, the interlocutors are already thinking about the fact that somebody called last night, and it is only the subject, Bill, which conveys ‘new’ information in the context. This works for (3) and (5) as well; by accommodating the salience of ‘I talk to P farmers’ in (3a), for example, one can infer that the relevant set is one for which all members are of that form. Example (4) also works, in that the notion that Bill did something last night is salient, but the fact that the “something” was ‘calling’ is nonsalient information. Finally, the givenness approach is especially intuitive for examples like the following, from Ladd (1996, p. 175). Here, the salience of the German language suffices to prevent stress on German at the right edge.

(7)

Pragmatics of Focus

Although the givenness approach is more general than the alternative semantic approach, for which examples like (7) are not as straightforward, it has been argued (see Selkirk, 2007; Stevens, 2014) that both givenness and a contrastive notion of focus are needed to account for the interaction between context and prosody in languages like English. One prominent phenomenon argued to involve the marking of both categories is second-occurrence focus, which involves a contrastive focus being de-accented in virtue of its having just been mentioned in the prior context. The following is an example from Rooth (1992, p. 109).

(8)

Pragmatics of Focus

We’d expect the scope of only to be specified by a focus on the second occurrence of rice, yielding the correct interpretation that people who grow rice eat exclusively rice, and yet this is not in line with the placement of the prominences. Rather, there are strong contrastive accents on grow and eat only. This is not entirely unexpected. Firstly, the placement of the verbal accents is licensed by a contrast between growing and eating a crop. Secondly, if the scope of only can be specified pragmatically, a focus on rice may not be needed, because scope over rice is the only sensible scope for only to take in this context. However, von Fintel (1994, p. 48) points out that examples like (9), which show that the second occurrence of rice cannot be reduced to a completely unstressed pronoun, suggest the presence of an “almost undetectable focus,” phonetically quite weak, but still sufficient to block an interpretation where only scopes over eat.

(9)

Pragmatics of Focus

This has led to analyses along the lines of (10), where the status of the second-occurrence focus rice as given information, encoded in a ‘G-feature,’ weakens its relative prominence.

(10)

Pragmatics of Focus

Finally, givenness-based accounts encounter problems in certain contexts where an explicit contrast is required, and a simple de-accenting as in (7) is not possible. This problem is detailed below.

2.3 Contrasting Antecedents

Recent literature has addressed a complexity in the conditions which license focus (see Wagner, 2006, 2012; Stevens, 2013, 2014; Büring, 2016): In some cases, but ostensibly not all, a contrasting antecedent is required to license contrastive accent. Informally, a contrasting antecedent is an antecedent in the sentence or discourse context with a mutually exclusive denotation. Mutual exclusivity as a semantic notion is clear in example (1); under common contextual assumptions, to a reasonable approximation, American farmers and Canadian farmers are nonoverlapping sets. If we change (1) so that the two focused adjectives denote properties which are semantically orthogonal—in other words, which denote sets with considerable overlap—acceptability is significantly degraded.

(11)

Pragmatics of Focus

The following example from Wagner (2012, p. 13) illustrates the requirement further.

(12)

Pragmatics of Focus

The accent pattern in (12a) is said to be felicitous because there is an antecedent in the context, ‘high-end convertible,’ which contrasts with ‘cheap convertible’ (the two denotations cannot describe the same object), and the element with the meaning ‘cheap’ which does not overlap with its antecedent receives the contrastive accent.

On the other hand, example (7) above, a motivating example for givenness-based accounts, is a case where no such antecedent seems necessary: it does not bear on the felicity of (7) whether the prior context evokes a denotation to contrast with ‘read German’ (like ‘write German,’ for example). In this case, the accent on read seems only to serve as a strategy to avoid accenting the discourse-salient direct object German. Example (13) below further illustrates; it is felicitous in any context where ‘Oxford’ is salient in the context of the utterance, and the effect seen in (11) is not found.

(13)

Pragmatics of Focus

Wagner (2012) allows certain relaxations of the contrast requirement in order to avoid a theory that relies on independent focus and givenness features. In general, there is disagreement about whether focus and givenness need to be given separate status, with givenness being called upon to explain cases like (7), (8), and (13), or whether they can derive each other.

3. Pragmatic Explanations

3.1 Information Structure and Questions Under Discussion

Vallduví’s information packaging theory models focus as an aspect of how the propositional information carried by a sentence is stored and accessed. Vallduví assumes a framework along the lines of Heim’s (1988) ‘file change semantics,’ in which entities in discourse are organized into ‘file cards’ (or ‘discourse referents’) and assertions in discourse serve to update the context by writing propositional information onto the file cards. For example, the sentence “Bill called last night” would update the context by locating the file card for the entity ‘Bill’—on which information from prior context would already be written, for example, that Bill is a friend of the speaker—and adding the new information to that card, resulting in a new representation like the following.

Pragmatics of Focus

Under Vallduví’s extension of this framework, the information stored on a file card can take the form of a salient open proposition. To illustrate what this means, consider example (4) from above.

(4)

Pragmatics of Focus

This is most natural in a context where the idea that Bill somehow communicated with the speaker last night is salient in the minds of the interlocutors. For example, we might imagine a context for (4) which includes an utterance like “Bill told me last night that he’s doing quite well.” This introduces into the discourse the assumption that Bill communicated with the speaker in some way, which we can represent as a proposition with an open variable, [last-night(P(Bill))], where the open variable P is some member of {‘come over,’ ‘call,’ ‘text,’ …}. At this point in the discourse, Bill’s file card would look as follows.

Pragmatics of Focus

Information packaging assumes that foci act as instructions to locate and fill in open variables, such that the answer in (4) instructs the hearer to first locate the open proposition [last-night(P(Bill))], which is the proposition obtained by replacing the focus with an open variable of the appropriate semantic type, and then to replace the open variable by the denotation of the focus, ‘called.’ Intuitively, the focus signals that ‘called’ serves to specify new information about some vague proposition that has been made salient by the prior context.

Information packaging is one specific conception of the broader notion of information structure, which refers to a level of structure at which the discourse context constrains linguistic form. Another popular conception of information structure is based on Roberts (1996/2012). Where Vallduví’s theory holds that the information conveyed by an utterance is organized by how it relates to a knowledge store that contains salient open propositions, Roberts’s theory holds that information more generally is organized by how it relates to a question under discussion (QUD). A QUD is formally a question denotation, formalized after Hamblin (1973) as a set of propositions which are possible answers to the question, and is taken to be a formalization of the speaker’s obligations and/or the hearer’s expectations about what kinds of utterances might be produced at a given point in the discourse. QUDs can correspond to overt questions or implicit questions merely suggested by the discourse.

Consider example (4) in a context where the speaker had previously uttered something like “Bill told me last night that he’s doing quite well.” That this prompts the question-answer exchange in (4) suggests a possible underlying motivation for the question: that the hearer wants to know in what manner the speaker was told this information by Bill. To the extent that the speaker can infer or recover this motivation, it is as if two questions are actually posed: the implicit question ‘How did you communicate with Bill last night?’ and the overt question ‘Did Bill come over last night?’ The latter is a subquestion of the former, meaning that a complete answer to the former would entail an answer to the latter (Groenendijk & Stokhof, 1984). Under this framework, such nested QUDs form a stack, such that QUDs must addressed in the reverse order that they were posed, which leads to the response in (4) as illustrated below.

Pragmatics of Focus

This nested nature of these questions can also be represented as a tree with the root node question dominating multiple subquestions (see also Büring, 2003). The list of subquestions dominated by the top-level question can be thought of as a strategy for answering the top-level question. Below is an illustration of such a strategy for (4), considering only two possible subquestions for simplicity’s sake.

Pragmatics of Focus

The answer in (4) anticipates the sister question to “Did Bill come over last night?” and, rather than waiting to be asked that question, simply addresses the root node question directly. That we can tie the answer in (4) to this implicit wh-question gives us a natural understanding of the observed focus pattern. The implicit question provides a set of alternatives, {‘Bill came over last night,’ ‘Bill called last night,’ ‘Bill texted last night,’ …}, such that the verb ‘called’ in (4) conveys the only nonredundant, nonrecoverable information in that context.

Beaver and Clark (2009) extend Roberts’s theory to account for association with focus. Under their account, a certain class of focus-sensitive elements, such as only, have lexical denotations that make reference to a current question under discussion. To illustrate, consider example (3), given again here.

(3)

Pragmatics of Focus

It is easier to interpret these against some implicit QUD, which must match the focus pattern:

(14)

Pragmatics of Focus

In (14a), it is easy to accommodate an implicit question ‘What kind of farmers do you talk to?’ but not possible to accommodate an implicit question ‘What kinds of Americans do you talk to?’ In (14b), the opposite is true, and this is reflected in the focus possibilities. Informally, the meaning of “I only talk to [AMERICAN] farmers” is taken to be, ‘Among all members of the QUD, {‘I talk to American farmers,’ ‘I talk to Canadian farmers,’ ‘I talk to American and Canadian farmers,’ …}, the proposition ‘I talk to American farmers’ is the most informative true proposition.’

The QUD approach is tied closely to Rooth’s alternative semantics. The intuition that foci should be elements whose denotations are not in every possible QUD answer corresponds closely to the status of foci as nonoverlapping denotations in some alternative set. But for the QUD approach, the alternative sets are tightly constrained by the flow of discourse: They must be part of a stack of QUDs introduced by the discourse context. Although this does not explain the purely contrastive cases of focus marking as in “farmer” sentences like in example (1), it takes a step toward explaining focus on a pragmatic level. In general, where formalizations such as alternative semantics seek to accurately describe the distribution and interpretation of focus, information-structural approaches seek to explain where these formalizations come from, the “why” behind the “what.”

3.2 Noisy Communication

In languages such as English and other Germanic languages, where focus is marked prosodically, it seems natural that the focus should receive increased prosodic prominence, rather than the opposite. We could imagine a hypothetical version of English where focus is marked by de-accenting, and everything that is not in focus must receive prominence, as in the following hypothetical example.

(15)

Pragmatics of Focus

We do not find, and do not expect to find, such “anti-focus” languages. The intuitive reason for this is functional: the focus corresponds to the most important material for interpretation, and as such should be emphasized, not de-emphasized. This is especially apparent in question-answer contexts in which everything except the focus may be elided.

(16)

Pragmatics of Focus

Given the restrictions placed by the QUD, the subject “Bill” conveys the only information necessary to interpret the utterance correctly. Schmitz (2008) calls this the i-critical information. Endowing i-critical elements with greater phonological prominence is an optimal way to communicate through a noisy channel (Shannon, 1948). The noisy channel refers to the fact that linguistic utterances, like other kinds of information-carrying signals, are transmitted imperfectly. The speaker’s intended utterance may be distorted, misunderstood, misheard, mis-segmented, or misparsed by the hearer. That speakers of language are conscious of the possibility of noise can play a role in how utterances are selected and interpreted (see also Benz, 2012; Bergen & Goodman, 2014). There may be a number of reasons to answer the question in (16) with a full sentence rather than simply saying “Bill”—for example, to facilitate processing or understanding—but even when a full-sentence answer is preferred, the rules of focus guarantee that the i-critical subject elevates to a higher degree of prominence than the other constituents in the sentence. This higher prominence raises the probability that the i-critical subject will survive any noise by lowering the likelihood of mis-segmentation, mishearing, etc. In general, although the mapping between phonological prominence and phonetic prominence is imperfect (see the discussion of example (6) in 2.1), a higher degree of prosodic prominence on a constituent at the phonological level should tend to ameliorate the effects of a noisy channel on the interpretation of that constituent. The optimal accent pattern under the analysis of Schmitz (2008) is one which minimizes accent overall, but which maintains strong prominence on i-critical material whose meanings cannot be recovered in the event of a noisy transmission. Below is a visual illustration of an optimal linguistic signal, from Stevens (2016).

Pragmatics of Focus

This account can be extended to cases where the nonfocused material is obligatory, such as (14), by considering optimal relative prominence within the DP. Imagine two different ways in which noise could obscure such an utterance.

(17)

Pragmatics of Focus

It is clear that (17b) is wholly uninterpretable: The open slot introduced by noise could be filled with almost anything. On the other hand, the effect of noise in (17a) is less drastic, because the hearer is likely to be able to work out that ‘farmers’ should fill the open slot. However, it would not be grammatical in this case to elide “farmers” altogether.

The role of noise is less obvious in the traditional farmer sentence in (1).

(1)

Pragmatics of Focus

It would not be correct to say that (18) was somehow interpretable out of the blue.

(18)

Pragmatics of Focus

However, note that the felicity of the focus pattern in (1) crucially depends on both slots being filled by the same meaning (in this case, ‘farmer’). It is possible that such parallel contrastive foci instantiate a kind of “risky bet” that at least one of the instances of “farmer” will get through, such that we end up with one of the following.

(19)

Pragmatics of Focus

This is not completely parallel to (17a), because it is still not immediately obvious that the open slot should be filled with ‘farmer,’ but with contrastive stress on either “American” or “Canadian” the hearer might reason pragmatically that “farmer” was intended, since any other element would also require contrastive stress, and thus presumably would have survived the noise.

3.3 Explaining Contrast

It still remains to be explained why some notion of semantic contrast is often required, as shown in (11) and (12):

(11)

Pragmatics of Focus

(12)

Pragmatics of Focus

These parallel focus patterns must set up a true contrast, for example, between cheap convertibles and expensive convertibles, but not between blue convertibles and expensive convertibles or between American farmers and tall farmers. This does not fall out of a noise-based account directly. Focus possibilities must be further constrained.

Vallduví and Vilkuna (1998) introduce a notion of contrast based on quantificational domains. The idea is that contrastive stress, for example on “American” in “American farmer,” generates some ‘membership set,’ which they call M, that restricts the domain of quantification. That pattern in (1) would generate the membership set M = {‘American,’ ‘Canadian,’ …}, which would be used as the domain for future quantifications about farmers, as shown in example (20):

(20)

Pragmatics of Focus

Without the contrastive foci, this sentence would be ambiguous: Either (i) American farmers tend to distrust farmers who are not American, or (ii) American farmers tend to be such that each American farmer distrusts all other farmers. With the foci, the first reading is preferred. Vallduví & Vilkuna’s notion of a membership set can be applied to this case. First, posit a simple semantics for the sentence.

Pragmatics of Focus

Along the lines of Rooth’s semantics for only, we have assumed that there is a contextually specified variable, C, which determines the domain of quantification over which other is interpreted. One natural candidate for C would be a set of functions identifying individuals, {λ‎x. x=x1, λ‎x. x=x2, …}; in this case, we would get the reading under which American farmers tend to distrust farmers who are not themselves. Under the quantificational domain approach to contrastive focus marking, the focus on “American” in (20) generates a quantificational domain in the form of a membership set M = {‘American,’ ‘Canadian,’ ‘Japanese,’ …}, and then M acts as the default value for C, and the truth conditions simplify to the following, which entails that American farmers distrust farmers who do not hail from the USA:

Pragmatics of Focus

The facts in (11) and (12) would then reduce to a requirement that these quantificational domains consist of members that are, in a certain sense, distinct from each other, such that ‘American’ and ‘Canadian’ can easily form a quantificational domain, but not ‘American’ and ‘tall.’ The following illustrates further.

(21)

Pragmatics of Focus

This seems to implicate that American farmers are unintelligent, even though this is not a semantic requirement of the sentence. It could be the case that for each American farmer, that farmer doesn’t trust another farmer if the other farmer is intelligent. Yet, the insulting reading is more prominent because the foci on “American” and “intelligent” set up a quantificational domain that is a superset of {‘American,’ ‘intelligent’}. To the extent that quantification domains are restricted to sets of ‘distinct’ properties, then the implication is some sort of semantic opposition between the two predicates.

Wagner formalizes the notion of semantic opposition as a form of mutual exclusivity, a formalization of the notion that ‘American’ and ‘Canadian’ denote, under certain contextual assumptions, nonoverlapping properties of individuals. Where this accounts nicely for examples like (11) and (12), the example (22), from Katzir (2013), shows that we must relax the notion to further account for the influence of context.

(22)

Pragmatics of Focus

It is not the case that ‘red’ is any inherent semantic opposition with ‘high-end.’ We need a more relaxed way of licensing quantificational domains which allows {‘red,’ ‘high-end’} in this case. It is possible that the opposition is not at the level of inherent lexical semantics, but rather at the level of information structure: it is assumed that there is one specific type of car that each relevant individual collects, such that a collector of red convertibles does not also collect high-end convertibles. In any case, context plays a crucial role in determining what is contrastive for purposes of focus placement.

4. Focus Crosslinguistically

4.1 Hungarian

Hungarian is a frequently cited example of a language with a syntactic position for focus. However, though there is debate over the details (see, e.g., Szabolcsi, 1981; Kiss, 1998; Roberts, 1998; Szendröi, 2003; Kenesei, 2006; Kiss, 2007; Horvath, 2010), it is generally accepted that syntactically marked focus in Hungarian behaves quite differently from prosodically marked focus. The following from Horvath (2010, p. 1359) illustrates.

(23)

Pragmatics of Focus

The focused direct object occurs in a special position in the left periphery, triggering movement of the verb past the particle fel, which would normally precede it. Although only foci end up in this position, not all foci end up in this position. Namely, left-edge focus in Hungarian is said to require an exhaustive interpretation, as if a silent only were adjoined to the focused element. This is argued to be a truth-conditional entailment, as evidenced by (24), which is meant to be a continuation of (23); the answer to the question in (23) can be declared to be false on the grounds that Mary was called in addition to John. Unlike in English, for example, a continuation of the form, ‘Yes, and they called Mary up, too’ is not possible.

(24)

Pragmatics of Focus

For this reason, Kiss (1998) draws a distinction between identificational focus and information focus, whereby the former is interpreted exhaustively and triggers syntactic movement, and the latter corresponds to “new information”, and behaves more like prosodic focus in English. Identificational foci in Hungarian are often translated as clefts, for example, “It was JOHN who called up,” where clefts suggest exhaustivity and are sensitive to focus. Note that the claim of an exhaustivity requirement has been disputed, for example by Onea and Beaver (2011).

4.2 West Chadic

Zimmermann (2008) gives a brief survey of focus marking in some West Chadic languages, beginning with Bole, where a morphological “focus” marker seems to mark a form of identificational focus.

(25)

Pragmatics of Focus

Unlike in the Hungarian example in (23), the focused element ‘millet’ does not uniquely answer the question under discussion, ‘What did Lengi do?’ Zimmermann argues that the function of identificational foci in these languages, which in Bole seems to be independent of the function of an information focus, is to indicate that the information conveyed by the focus-marked element is somehow surprising or improbable in the context.

In contrast to Bole, Gùrùntùm marks information focus morphologically, and like Hungarian, it has a special syntactic position for identificational foci which must also be marked as information foci. In (26), the focus marker picks out ‘a crocodile’ as being the information focus, similarly to prosodic focus in English, and movement to the left edge of the clause indicates an emphatic cleft-like interpretation like that of the answer in (25).

(26)

Pragmatics of Focus

Finally, Hausa can move partial foci to the left edge for emphasis as in (27), where ‘robbers’ has been fronted.

(27)

Pragmatics of Focus

This has a parallel in German, as in the following from Fanselow and Lenertová (2011), where the entire sentence is focused in terms of the question under discussion, but the direct object is fronted for emphasis.

(28)

Pragmatics of Focus

The combination of strong prosodic accent and leftward movement is emphatic, suggesting that rabbits are a particularly noteworthy thing to catch. A similar movement operation is also said to be possible in Yiddish-influenced dialects of American English, a construction which Prince (1981) calls “Yiddish-movement.”

4.3 The Cartographic Approach

Information-structural categories including focus are often incorporated as functional heads and phrases within the left periphery of an abstract phrase structure. Following Rizzi’s (1997) “cartographic” approach to syntax, foci are determined via agreement with a silent functional head “Foc” which forms a focus phrase (FocP) that dominates the core of the clause. This is taken to be a universal and fixed component of the syntactic “spine” of a clause, although whether agreement with “Foc” forces movement into the specifier of FocP must be language-dependent. Rizzi arrives at the following core structure: ForceP determines the force of the whole clause (declarative, interrogative, etc.) and dominates any number of iterated TopPs (topic phrases), and somewhere mixed in with the TopPs there must be a single FocP, where FocP dominates FinP (finiteness phrase), from which point a more traditional phrase structure for the core clause takes over.

Pragmatics of Focus

An illustration from Italian, from Rizzi (1997, p. 295), is given in (29).

(29)

Pragmatics of Focus

The subordinate clause would be analyzed roughly as follows.

Pragmatics of Focus

We see some evidence for left-edge focus along these lines in English, but it is restricted to a particular pragmatic usage. In what Prince (1981) calls “focus-movement,” it is possible to front a narrowly focused constituent in a corrective or emphatic context as in (30):

(30)

Pragmatics of Focus

The cartographic approach aims to provide a universal inventory and hierarchy of functional heads that can explain systematic relationships between information structure and word order across many languages. One potential downside to the approach is that it requires a high number of syntactic primitives, expanding the richness of Universal Grammar at the expense of purely pragmatic explanations of functional patterns in word order variability. Shlonsky (2010) gives an overview of the cartographic approach, citing claims in the literature of between 40 and 400 different features posited in various languages by researchers in this tradition.

Further Reading

Beaver, D. I., & Clark, B. Z. (2009). Sense and sensitivity: How focus determines meaning. Oxford: Blackwell.Find this resource:

Bolinger, D. (1976). Accent is predictable (if you’re a mind-reader). Language, 48, 633–644.Find this resource:

Büring, D. (2003). On d-trees, beans, and B-accents. Linguistics and Philosophy, 26(5), 511–545.Find this resource:

Chomsky, N. (1971). Deep structure, surface structure and semantic interpretation. In D. Steinberg & L. Jakobovits (Eds.), Semantics: An interdisciplinary reader in philosophy (pp. 183–216). Cambridge: Cambridge University Press.Find this resource:

Farkas, D., & Bruce, K. (2010). On reacting to assertions and polar questions. Journal of Semantics, 27, 81–118.Find this resource:

Katz, J., & Selkirk, E. (2011). Contrastive focus vs. discourse-new. Language, 87, 771–816.Find this resource:

Kiss, K. É. (1998). Identificational focus versus information focus. Language, 74, 245–273.Find this resource:

Krifka, M. (2001). For a structured meaning account of questions and answers. In C. Féry & W. Sternefeld (Eds.), Audiatur vox sapientia: A Festschrift for Arnum von Stechow (pp. 287–319). Berlin: Akademie Verlag.Find this resource:

Roberts, C. (1996/2012). Information structure in discourse: Towards an integrated formal theory of pragmatics. Semantics and Pragmatics, 5(6), 1–69.Find this resource:

Rooth, M. (1992). A theory of focus interpretation. Natural Language Semantics, 1(1), 75–116.Find this resource:

Schwarzschild, R. (1999). Givenness, AvoidF and other constraints on the placement of accent. Natural Language Semantics, 7(2), 141–177.Find this resource:

Selkirk, E. (2007). Contrastive focus, givenness and the unmarked status of “discourse-new.” In C. Féry, M. Fanselow, & M. Krifka (Eds.), The notions of information structure (pp. 125–145). Interdisciplinary Studies on Information Structure 6. Potsdam: Universitätsverlag.Find this resource:

Truckenbrodt, H. (2007). The syntax-phonology interface. In P. de Lacy (Ed.), The Cambridge handbook of phonology (pp. 435–456). Cambridge: Cambridge University Press.Find this resource:

Vallduví, E. (1990). The informational component (Unpublished PhD diss.). University of Pennsylvania, Philadelphia.Find this resource:

Zimmermann, M., & Onea, E. (2011). Focus marking and focus interpretation. Lingua, 121, 1651–1670.Find this resource:

References

Beaver, D. I., & Clark, B. Z. (2009). Sense and sensitivity: How focus determines meaning. Oxford: Blackwell.Find this resource:

Benz, A. (2012). Errors in pragmatics. Journal of Logic, Language & Information, 21, 97–116.Find this resource:

Bergen, L., & Goodman, G. (2014). The strategic use of noise in pragmatic reasoning. In P. Bello, M. Guarini, M. McShane, & B. Scassellati (Eds.), Proceedings of the 36th annual Cognitive Science Conference (pp. 182–187). Austin, TX: Cognitive Science Society.Find this resource:

Büring, D. (2003). On d-trees, beans, and B-accents. Linguistics and Philosophy, 26(5), 511–545.Find this resource:

Büring, D. (2016). Unalternative semantics. In S. D’Antonio, M. Moroney, & C. R. Little (Eds.), Proceedings of the 25th Semantics and Linguistic Theory Conference, held at Stanford University, May 15–17, 2015 (pp. 550–575). Washington, DC: Linguistic Society of America.Find this resource:

Chafe, W. (1974). Language and consciousness. Language, 50, 111–133.Find this resource:

Fanselow, G., & Lenertová, D. (2011). Left peripheral focus: Mismatches between syntax and information structure. Natural Language and Linguistic Theory, 29, 169–209.Find this resource:

von Fintel, K. (1994). Restrictions on quantifier domains (Unpublished PhD diss.). University of Massachusetts, Amherst.Find this resource:

Groenendijk, J., & Stokhof, M. (1984). Studies on the semantics of questions and the pragmatics of answers (Unpublished PhD diss.). University of Amsterdam, Amsterdam.Find this resource:

Halliday, M. (1967). Notes on transitivity and theme in English. Journal of Linguistics, 3, 199–244.Find this resource:

Hamblin, C. L. (1973). Questions in Montague English. Foundations of Language, 10, 41–53.Find this resource:

Heim, I. (1988). The semantics of definite and indefinite noun phrases. New York: Garland Publishing.Find this resource:

Horvath, J. (2010). “Discourse-features,” syntactic displacement, and the status of contrast. Lingua, 120, 1346–1369.Find this resource:

Katzir, R. (2013). A note on contrast. Natural Language Semantics, 23, 1–11.Find this resource:

Kenesei, I. (2006). Focus as identification. In V. Molnár & S. Winkler (Eds.), The architecture of focus (pp. 137–168). Berlin: Mouton de Gruyter.Find this resource:

Kiss, K. É. (1998). Identificational focus versus information focus. Language, 74, 245–273.Find this resource:

Kiss, K. É. (2007). Topic and focus: Two structural positions associated with logical functions in the left periphery of the Hungarian sentence. In C. Féry, M. Fanselow, & M. Krifka (Eds.), The notions of information structure (pp. 69–81). Interdisciplinary Studies on Information Structure 6. Potsdam: Universitätsverlag.Find this resource:

Ladd, D. R. (1996). Intonational phonology. New York: Cambridge University Press.Find this resource:

Onea, E., & Beaver, D. (2011). Hungarian focus is not exhausted. In E. Cormany, S. Ito, & D. Lutz (Eds.), Proceedings of the 19th Semantics and Linguistic Theory Conference, held April 3–5, 2009 at The Ohio State University (pp. 342–359). Washington, DC: Linguistic Society of America.Find this resource:

Prince, E. F. (1981). Topicalization, focus-movement, and Yiddish-movement: A pragmatic differentiation. In D. K. Alford et al. (Eds.), Proceedings of the Seventh Annual Meeting of the Berkeley Linguistics Society (pp. 249–264). Berkeley: Berkeley Linguistics Society.Find this resource:

Prince, A., & Smolensky, P. (1993). Optimality theory: Constraint interaction in generative grammar. Rutgers University Center for Cognitive Science Technical Report 2.Find this resource:

Rizzi, L. (1997). The fine structure of the left periphery. In L. Haegeman (Ed.), Elements of grammar (pp. 281–337). Amsterdam: Kluwer.Find this resource:

Roberts, C. (1998). Focus, the flow of information, and Universal Grammar. In P. Culicover and L. McNally (Eds.), The limits of syntax (pp. 109–160). Academic Press.Find this resource:

Roberts, C. (1996/2012). Information structure in discourse: Towards an integrated formal theory of pragmatics. Semantics and Pragmatics, 5(6), 1–69.Find this resource:

Rooth, M. (1985). Association with focus (Unpublished PhD diss.). University of Massachusetts, Amherst.Find this resource:

Rooth, M. (1992). A theory of focus interpretation. Natural Language Semantics, 1(1), 75–116.Find this resource:

Schwarzschild, R. (1999). Givenness, AvoidF and other constraints on the placement of accent. Natural Language Semantics, 7(2), 141–177.Find this resource:

Schmitz, H.-C. (2008). Accentuation and interpretation. New York: Palgrave MacMillan.Find this resource:

Shannon, C. E. (1948). A mathematical theory of communication. Bell System Technical Journal, 27(3), 379–423.Find this resource:

Shlonsky, U. (2010). The cartographic enterprise in syntax. Language and Linguistics Compass, 4(6), 417–429.Find this resource:

Selkirk, E. (1995). Sentence prosody: Intonation, stress, phrasing. In J. Goldsmith (Ed.), Cambridge handbook of phonological theory (pp. 550–569). Cambridge: Blackwell.Find this resource:

Selkirk, E. (2007). Contrastive focus, givenness and the unmarked status of “discourse-new.” In C. Féry, M. Fanselow, & M. Krifka (Eds.), The notions of information structure (pp. 125–145). Interdisciplinary Studies on Information Structure 6. Potsdam: Universitätsverlag.Find this resource:

Stevens, J. S. (2013). Information structure, grammar and strategy in discourse. (Unpublished doctoral dissertation). University of Pennsylvania, Philadelphia.Find this resource:

Stevens, J. S. (2014). Against a unified analysis of givenness and focus. In R. E. Santana-LaBarge (Ed.), Proceedings of the 31st West Coast Conference on Formal Linguistics (pp. 438–446). Somerville, MA: Cascadilla.Find this resource:

Stevens, J. S. (2016). A signaling account of contrastive focus. In N. Bade, P. Berezovskaya, & A. Schöller (Eds.), Proceedings of Sinn und Bedeutung 20 (pp. 694–711).Find this resource:

Szabolcsi, A. (1981). Compositionality in focus. Folia Linguistica, 15, 141–161.Find this resource:

Szendröi, K. (2003). A stress-based approach to the syntax of Hungarian focus. Linguistic Review, 20, 37–78.Find this resource:

Vallduví, E. (1990). The informational component (Unpublished PhD diss.). University of Pennsylvania, Philadelphia.Find this resource:

Vallduví, E., & Vilkuna, M. (1998). On rheme and kontrast. In P. Culicover & L. McNally (Eds.), The limits of syntax (pp. 79–108). Syntax and Semantics 29. New York: Academic Press.Find this resource:

Wagner, M. (2006). Givenness and locality. In M. Gibson and J. Howell (Eds.), Proceedings of the 16th Semantics and Linguistic Theory Conference, held March 22–24, 2006 at the University of Tokyo, 295–312. Washington, DC: Linguistic Society of America.Find this resource:

Wagner, M. (2012). Focus and givenness: A unified approach. In I. Kučerová and A. Neeleman (Eds.), Contrasts and positions in information structure (pp. 102–147). Cambridge: Cambridge University Press.Find this resource:

Zimmermann, M. (2008). Contrastive focus and emphasis. Acta Linguistica Hungarica, 55, 347–360.Find this resource:

Zimmermann, M., & Onea, E. (2011). Focus marking and focus interpretation. Lingua, 121, 1651–1670.Find this resource: