PQJE209998
TECHSET COMPOSITION LTD, SALISBURY, U.K.
12/23/2006
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY
0000, 00 (0), 1 – 24
Focus identification during sentence comprehension:
Evidence from eye movements
Kevin B. Paterson
University of Leicester, Leicester, UK
Simon P. Liversedge
University of Southampton, Southampton, UK
Ruth Filik
University of Glasgow, Glasgow, UK
Barbara J. Juhasz
Wesleyan University, Middletown, CT, USA
Sarah J. White
University of Leicester, Leicester, UK
Keith Rayner
University of Massachusetts, Amherst, MA, USA
Three eye movement experiments investigated focus identification during sentence comprehension.
Participants read dative or double-object sentences (i.e., either the direct or indirect object occurred
first), and a replacive continuation supplied a contrast that was congruous with either the direct or the
indirect object. Experiments 1 and 2 manipulated focus by locating only adjacent to either the direct or
indirect object of dative (Experiment 1) or double-object (Experiment 2) sentences. Reading-time
effects indicated that the surface position of the focus particle influenced processing. In addition,
Experiment 1 reading times were longer when the replacive was incongruous with the constituent
that only adjoined, and particle position modulated a similar effect in Experiment 2. Experiment 3
showed that this effect was absent when only was omitted. We conclude that the surface position
of a focus particle modulates focus identification during on-line sentence comprehension.
Although the assignment of focus in a sentence is
recognized as an important aspect of semantic
interpretation (e.g., Jackendoff, 1972; Krifka,
1992; Rooth, 1992), there have been few empirical
investigations of its computation during language
comprehension. In this paper, we report three
Correspondence should be addressed to Kevin Paterson, School of Psychology, Henry Wellcome Building, Lancaster Road,
University of Leicester, Leicester, LE1 9HN, UK. Email: [email protected]
This research was supported by Grant 12/S19168 from the Biotechnology and Biological Sciences Research Council and by
Grant HD17246 from the National Institute of Health. We are grateful to Tim Slattery for his assistance in collecting the data
reported in Experiment 1. We thank Ekkehard König for giving us useful advice. We also thank three anonymous reviewers for
their detailed and perceptive comments on earlier versions of the paper.
# 0000 The Experimental Psychology Society
http://www.psypress.com/qjep
1
DOI:10.1080/17470210601100563
PATERSON ET AL.
eye movement experiments investigating grammatical constraints on focus identification during
written sentence comprehension.
Focus either marks information that is newly
asserted or indicates that a contrast is to be made
between current information and its alternatives
(Chomsky, 1971; Halliday, 1967; Jackendoff, 1972;
Kiss, 1998; Rochemont & Culicover 1990; Rooth,
1992; Selkirk, 1995). Theoretical approaches to
focus interpretation emphasize the role of prosody
in marking a sentence’s focused syntactic constituent (e.g., Jackendoff, 1972; Kadmon, 2001;
Selkirk, 1995), and although there is considerable
debate concerning precisely how prosody indicates
focus, it generally is agreed that focus correlates
with a peak of prosodic prominence in a sentence.
Thus, in a sentence such as MARY kissed Tom
(with capitals indicating the peak of prosodic prominence), focus is assigned to Mary, and readers
will understand the sentence to assert that it was
Mary who kissed Tom, rather than some other contextually relevant person. Empirical research has
shown that speech comprehension is easier when
focused information is marked using prosody than
when it is not (Birch & Clifton, 1995, 2002; Bock
& Mazzella, 1983; Noteboom & Kruyt, 1987;
Terken & Noteboom, 1987), that listeners perceive
focused information more easily than nonfocused
information (Cutler & Fodor, 1979; Hornby,
1974), and that memory for focused information
is enhanced (Malt, 1985; Singer, 1976). Other
studies have shown that using prosody to assign
focus can influence how ambiguous sentences are
interpreted (Schafer, Carlson, Clifton, & Frazier,
2000).
It also has been widely observed (e.g., Jackendoff,
1972; König, 1991; Rooth, 1992) that function
words termed focus particles (e.g., only, even, just,
also, too) influence focus interpretation. These
signal that a contrast is to be made between the referent of a sentential constituent and its alternatives,
although the precise nature of this contrast
depends on the particle’s lexical characteristics.
Some, such as only and just, specify an exclusive contrast such that a property of the referent does not also
belong to its alternatives, whereas others, such as
even, also, and too, specify an additive contrast in
2
which the referent’s property is shared with its
alternatives. For example, Only Mary kissed John
means that Mary kissed John, and others did not,
whereas Even Mary kissed John means that Mary,
in addition to some others, kissed John.
Jackendoff (1972) was the first to argue that particles such as only “associate with focus”, meaning
that they often, but not always, are interpreted as
specifying a contrast between the referent of a
focused syntactic constituent and its alternatives.
The sentences in (1) and (2) frequently are used
to illustrate how the choice of focus can affect the
interpretation of a sentence containing only:
1. Mary only introduced [Bill]F to Sue.
2. Mary only introduced Bill to [Sue]F.
For these sentences, the focused constituent is
indicated by brackets and the subscript F. The sentences differ only in terms of the distribution of
focus, which is allocated to the direct object
(Bill) in (1) and to the indirect object (Sue) in
(2). If, as Jackendoff (1972) proposed, the particle
usually associates with the focused constituent,
then only should associate with the direct object
in (1) and with the indirect object in
(2). Consequently, (1) should be understood to
mean that Bill alone was introduced to Sue, and
(2) should mean that he was introduced to noone other than Sue. These examples illustrate
that sentences that contain a focus particle but
differ with respect to the distribution of focus
can receive qualitatively different interpretations.
However, it also can be shown, using a sentence
such as (3), that it is not mandatory for the particle
to associate with the focused constituent:
3. [Mary]F only introduced Bill to Sue.
For this sentence, Mary is the focused constituent, but the sentence cannot be interpreted with
only associating with it. Instead, readers are likely
to associate only with either the direct or indirect
object or the entire verb phrase (i.e., introduced
Bill to Sue), despite none of these constituents
being a recipient of focus.
Jackendoff (1972) accounted for these interpretative preferences by proposing that grammatical
constraints govern the range (or scope) over which
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 0000, 00 (0)
FOCUS IDENTIFICATION DURING SENTENCE COMPREHENSION
a focus particle can operate, such that it associates
with a focused constituent within its range.
Jackendoff did not define these constraints but
other researchers, such as Reinhart (1999), have
proposed that only ranges over constituents that it
c-commands in a parse tree. Thus, when only is
included in a sentence, it associates with a focused
constituent within this syntactic domain. For
Sentences 1–3, this stipulates that only associates
with the direct object (i.e., Bill), the indirect
object (i.e., Sue), or the verb phrase (i.e., introduced
Bill to Sue) but cannot associate with the subject
noun phrase (i.e., Mary). We investigated if
readers employ this grammatical knowledge
during normal sentence comprehension.
Jackendoff’s (1972) suggested that prosody
might contribute to focus identification in sentences containing focus particles by disambiguating the choice of focus within a syntactic
domain. He argued that when prosody marks the
focused constituent, the particle associates with
this constituent if it is within its range; but if it
marks a constituent that is outside of the particle’s
range, then the sentence will be perceived as
anomalous. However, and very importantly, explicit prosodic cues are absent in silent reading (but
see Fodor, 2002). Consequently, other factors
may play an important part in focus identification
during written sentence comprehension, particularly the grammatical cues supplied by the
surface position of the focus particle.
Many focus particles, including only, occur
freely in different syntactic positions in a sentence,
and the flexibility in their placement affords a
means of evaluating grammatical influences on
their interpretation. As already observed, when
only is in an adverbial position, as in (3), it can
associate with the direct or indirect object or the
entire verb phrase. Now consider variants of this
sentence (e.g., 4 and 5) in which the particle
adjoins either the direct or the indirect object:
4. Mary introduced Bill to only Sue.
5. Mary introduced only Bill to Sue.
When the particle adjoins the indirect object
(e.g., Sue in 4), it should associate with this constituent, as it alone is within the particle’s syntactic
domain. However, when it precedes the direct
object (e.g., Bill in 5) the sentence has two grammatically permissible analyses. On one analysis, the
particle associates with the direct object
(e.g., Bill), but on the other it adjoins a constituent
containing both postverbal phrases (e.g., Bill to Sue)
and can associate with either the direct or the indirect object. If grammatical constraints influence
focus identification, then (4) means that Bill was
not introduced to anyone other than Sue, By contrast, when only associates with the direct object
in (5), the sentence means that no-one other than
Bill was introduced to Sue, but if it is analysed as
adjoining a constituent containing both postverbal
phrases, then the sentence will be ambiguous
between these two interpretations. We adopted
the working hypothesis that only associates locally,
to the adjacent phrase, during sentence comprehension. This assumption enabled us to investigate
whether the surface position of the particle
informs the allocation of focus. Key questions are
whether and when this information is employed
in normal comprehension.
Relatively few studies have directly investigated
the cognitive processes underlying focus identification. Some researchers have used the focus sensitivity of only to investigate the development of
linguistic constraints on sentence interpretation
(e.g., Crain, Ni, & Conway, 1994; Paterson,
Liversedge, Rowland, & Filik, 2003; Paterson,
Liversedge, White, Filik, & Jaz, 2006; Philip &
Lynch, 1999). For instance, Crain et al. argued
that young children do not employ grammatical
constraints to restrict the range of only, but that
they instead interpret sentences with presubject
only, such as Only the fireman is holding a hose, as
having the same meaning as counterparts with
preverbal only, such as The fireman is only holding
a hose. Paterson et al. (2003) disputed this
account of children’s errors, however, arguing
that children make errors because they have difficulty in computing the alternatives to a focused
referent rather than by failing to restrict the particle’s range. With respect to the current investigations, it must be noted that both groups of
researchers found that adults almost always interpreted sentences with only by associating the
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 0000, 00 (0)
3
PATERSON ET AL.
particle with a constituent within its c-command
domain and took this as evidence for grammatical
restrictions on focus identification in adult sentence comprehension. However, as this evidence
was obtained from tasks in which participants
formed explicit judgements about sentence
meaning, the results may reflect the operation of
processes that are not normally involved in sentence comprehension.
Other research has investigated whether the
choice of focus in a sentence can guide the processing of syntactic ambiguities (e.g., Clifton, Bock, &
Rado, 2000; Filik, Paterson, & Liversedge, 2005;
Liversedge, Paterson, & Clayes, 2002; Ni, Crain,
& Shankweiler, 1996; Paterson, Liversedge, &
Underwood, 1999; Sedivy, 2002). Much of this
research was in response to Ni et al.’s claim that
using only to indicate contrastive focus can guide
the parsing of temporarily ambiguous sentences
such as (6):
6. Only businessmen loaned money at low interest
kept accurate records of their expenses.
For this sentence, the phrase loaned money at
low interest is temporarily ambiguous between a
main clause analysis and a reduced relative clause
analysis, but it is disambiguated at kept in favour
of the latter. Such ambiguities are a cause of difficulty, since perceivers typically assign a main clause
analysis to the ambiguous material and incur a processing cost when it is disambiguated as a reduced
relative clause (e.g., Frazier & Rayner, 1982).
However, Ni et al. (1996) proposed that including
only can eliminate such difficulty. They argued that
on encountering only businessmen in (6), the parser
establishes a contrast between subsets of the head
noun (e.g., two sets of businessmen), and anticipates further information specifying the difference
between these subsets. Such information may be
supplied by the relative clause analysis of the
ambiguity. Thus, according to Ni et al., referential
processing demands associated with contrastive
focus can predispose readers to adopting the relative clause analysis of an ambiguity. Consistent
with this claim, Ni et al. found that only eliminated
reading difficulty for ambiguous sentences like (6).
Sedivy (2002) replicated this finding. However,
4
Clifton et al. (2000) found no effects of only on
the parsing of relative clause ambiguities, and
Paterson et al. (1999) and Liversedge et al.
(2002; see also Filik et al., 2005) found that its
influence depended on which syntactic analyses
were available to the parser and therefore was
obtained only for certain syntactic ambiguities.
Although the question of whether contrastive
focus influences syntactic ambiguity resolution is
not directly relevant to the present investigation,
it is important to note that Ni et al.’s (1996)
account implicitly assumes that focus identification occurs sufficiently rapidly for its referential
consequences to influence initial parsing decisions,
indicating very rapid on-line focus identification
during reading. However, since the data pertaining
to whether only does guide parsing are equivocal
(with some studies showing such effects and
others failing to do so), then even this indirect
evidence is not conclusive.
There is a growing interest in more basic questions concerning the processing of focus identification (e.g., Carlson, 2004; Carlson, Frazier,
Clifton, & Dickey, 2005; Stolterfoht, Friederici,
Alter, & Steube, 2003), although only one published study, by Gennari, Meroni, and Crain
(2004), has directly investigated its computation
during sentence comprehension. This employed
a visual world eye-tracking task in which participants viewed a picture while listening to sentences
that included adverbial only (e.g., The mother only
brought some milk to the boy). The constituent
that associated with only was either ambiguous or
else it was disambiguated by including phonological stress on the direct object (e.g., some milk).
Participants were most likely to visually inspect
the direct object’s referent and a contrasting
item in the picture when stress was included.
Furthermore, they were more likely to judge sentences that included phonological stress as correctly
describing the picture when it depicted a contrast
based on this referent. Thus, the data suggested
that prosody can disambiguate focus. However,
because participants formed explicit judgements
about auditory sentence meaning, the data do not
indicate whether focus identification is mandatory
in written sentence comprehension.
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 0000, 00 (0)
FOCUS IDENTIFICATION DURING SENTENCE COMPREHENSION
Our experiments used measures of eye movements during reading to investigate whether locating only in different surface positions modulates
focus identification, since this methodology provides an indication of the moment-to-moment
processes that occur in normal text comprehension
(Liversedge & Findlay, 2000; Rayner, 1998). The
sentences included a ditransitive verb that permitted dative alternation, as shown in Table 1.
For dative sentences, the direct object (e.g., the
salt) preceded the indirect object (e.g., her mother),
and this constituent order was reversed for doubleobject sentences. Each sentence was continued by
a replacive (but not pepper too/but not father too)
that supplied a congruous contrast for either the
direct or the indirect object. Experiment 1 investigated the effect on processing of locating only in a
position adjoining either the direct or the indirect
object in dative sentences, and Experiment 2
investigated this effect in double-object sentences.
Experiment 3 examined the processing of sentences that omitted the particle.
Our working assumption was that the particle
would associate with its immediately adjacent
constituent. If the sentences are interpreted with
only associating with the indirect object (e.g., her
Q14 Table 1. Examples of sentences used in Experiments 1–3
Experiment
1. Dative sentences
with only
2. Double-object
sentences with
only
3. Dative and
double-object
sentences
without only
Sample sentence
At dinner, Jane passed 1j [only] the salt
to [only] her mother 2j but not [the
pepper/her father] 3j as well because
4j she couldn’t reach 5.
At dinner, Jane passed 1j [only] her
mother [only] the salt 2j but not [the
pepper/her father] 3j as well because
4j she couldn’t reach 5.
At dinner, Jane passed 1j the salt to her
mother 2j but not [the pepper/her
father] 3j as well because 4j she
couldn’t reach 5.
At dinner, Jane passed 1j her mother the
salt 2j but not [the pepper/her father]
3j as well because 4j she couldn’t
reach 5.
Note: Vertical lines delimit regions of analysis, slashes denote
alternatives, and parentheses indicate the alternate positions
of the focus particle.
mother) then readers should have difficulty with
replacives supplying a contrast that is incongruous
with this constituent (e.g., but not the pepper) as
compared with congruous replacives (e.g., but not
her father). Conversely, if only associates with the
direct object (e.g., the salt) then readers should
have difficulty for incongruous replacives (e.g.,
but not her father) but not for congruous replacives
(e.g., but not the pepper). The results also may be
informative about the time course of focus identification. Recall that Ni et al. (1996) assumed that
focus identification occurs very early during processing, sufficiently so that it might affect initial
parsing decisions. The earliest point within the
sentence at which we might anticipate disruption
due to the incongruity of the replacive is in
reading-time measures that reflect early processing
at the replacive region. Alternatively, if such effects
are delayed, then we might detect disruption in
measures reflecting later stages of processing or
in regions downstream from the replacive.
EXPERIMENT 1
Method
Participants
A total of 36 American English speakers from the
University of Massachusetts at Amherst participated for extra course credit or cash payment.
Materials and design
We constructed 32 sentences that permitted dative
alternation, with the direct object preceding the
indirect object, as illustrated in Table 1. The
direct object was always inanimate (e.g., the salt)
and the indirect object was always animate (e.g.,
her mother), to maximize the likelihood of the replacive ultimately being interpreted as supplying a
contrast for the intended constituent. The particle
only adjoined either the direct or indirect object,
and the sentence was continued by a replacive
that supplied a congruous contrast for one of
these constituents. Thus, we employed two independent variables: the position of only and the congruity of the contrast supplied by the replacive. The
dependent variables were measures of reading time.
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 0000, 00 (0)
5
PATERSON ET AL.
Norming data. We conducted three studies to
assess readers’ comprehension of the sentences
used in Experiment 1. The first examined the perceived acceptability of replacives that supplied a
contrast for the sentence’s indirect object (e.g., At
dinner, Jane passed only the salt to her mother but
not her father too). For these sentences, when the
replacive includes a preposition (e.g., to, as in but
not to her father) it is syntactically disambiguated
as supplying a contrast for the indirect object.
The sentences in Experiment 1 did not include
this preposition, to avoid disambiguating the contrast. However, it was possible that readers would
find such sentences to be less grammatically acceptable than those that included the preposition, and
this could affect reading behaviour. We therefore
examined whether omitting the preposition
affected the perceived acceptability of the sentences. A total of 20 participants provided acceptability judgements for sentences with only
adjoining the indirect object and a replacive that
supplied a contrast for this constituent and either
did or did not include a preposition, using a
7-point scale (1 indicating that the sentence was
entirely unacceptable, 7 indicating that it was
entirely acceptable). Acceptability ratings did not
differ (“with preposition” ¼ 4.42, “without preposition” ¼ 4.40), t(31) , 1, indicating that
readers found sentences with and without the
preposition to be equally acceptable.
A second study was conducted to ensure that
when the replacive was congruous it ultimately
was interpreted as contrasting with the particle’s
adjacent constituent and no other constituent.
A total of 22 participants viewed dative sentences
with only adjoining either the direct or the indirect
object and the replacive supplying a congruous
contrast (e.g., At dinner, Jane passed only the salt
to her mother but not the pepper too). An additional
16 filler items had a ditransitive construction, but
with only adjoining the subject noun, and a replacive supplying a congruous contrast for this constituent (e.g., At the auction, only the dentist sold
some equipment to the collector and not the doctor
too). The contrastive noun was underscored
(e.g., but not the pepper as well), and participants
indicated which constituent contrasted with it.
6
In total there were three cases in which participants
did not select the constituent that only adjoined
(i.e., 99.7% correct responses). Thus, when the
contrast was congruous, the replacive almost
always was interpreted as contrasting with the
constituent adjoining the focus particle and was
rarely interpreted as providing a contrast with any
other constituent. Note that these results simply
indicate that supposedly congruous replacives
were indeed congruous with the intended constituent and did not ultimately form a more congruous
contrast with another constituent. The eye-tracking
data would reveal how the replacive is processed
during on-line sentence comprehension.
Finally, we collected sentence completion data
to assess readers’ preferences for assigning focus.
These data would reveal whether readers interpreted the sentences by associating only with its
adjacent constituent. A total of 32 participants
provided completions for dative sentence fragments, with only preceding either the direct or
the indirect object. The fragments were truncated
at the replacive (e.g., At dinner, Jane passed only the
salt to her mother but not . . . ). A total of 16 filler
items also had a ditransitive construction and
were truncated at the replacive, but for these sentences only preceded the subject noun (e.g., At the
auction, only the dentist sold some equipment to the
collector and not . . . ). Sentence completions were
categorized as providing an appropriate contrast
for either the direct or the indirect object, or providing an inappropriate contrast. The majority of
completions did supply an appropriate contrast,
and the mean proportion of these responses is
shown in Table 2.
Table 2. Percentage completions that supplied a contrast for either
the direct object or the indirect object of dative sentence fragments
with only
Focus particle position
Contrasting constituent
Indirect object (her mother)
Direct object (the salt)
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 0000, 00 (0)
Preindirect
Predirect object
(only the salt to object (the salt to
her mother)
only her mother)
39
59
91
6
FOCUS IDENTIFICATION DURING SENTENCE COMPREHENSION
The results indicated that although there was a
preference for associating the particle with its adjacent constituent, this was substantially weaker
when only adjoined the direct rather than the
indirect object (i.e., when it appeared early
rather than late in the sentence), t(31) ¼ 6.34,
p , .001. Thus, it appeared that when only
adjoined the direct object, the sentence could be
interpreted with the particle associating with
either the direct or the indirect object, albeit that
there was a narrow preference for it to associate
with the adjacent phrase (i.e., the direct object).
By comparison, when only adjoined the indirect
object, there was an overwhelming preference for
it to associate with this phrase. We note that this
finding is inconsistent with our working hypothesis that only associates with its immediately
adjacent constituent. However, the finding is consistent with participants assigning focus in line
with grammatical constraints on focus interpretation. Recall that when only adjoins the indirect
object of a dative sentence then it should associate
with this constituent, as it alone is within the
particle’s syntactic domain. The overwhelming
preference for producing a contrast based on the
indirect object when only adjoined this constituent
was consistent with this being the only grammatically permissible analysis. Recall also that when
only precedes the direct object the sentence can
be analysed with the particle attaching to this constituent or to one that includes both postverbal
phrases. Under the former analysis there should be
a preference for forming a contrast based on the
direct object. However, the particle can associate
with either direct or indirect object under the
latter analysis. Thus, the less clear-cut preference
for producing continuations that supplied a contrast
for the direct object when only appears early in the
sentence might reflect the availability of a grammatically permissible analysis that permits the particle
to associate with either postverbal constituent.
The eye-tracking data would reveal if similar preferences occur during on-line comprehension.
Procedure
Eye movements were monitored via a Fourward
Technologies Dual Purkinje Generation 6
Eye-Tracker located at the University of
Massachusetts eye-tracking laboratory. The eyetracker has an angular resolution of 10 minutes of
arc, and a PC displayed materials on a VDU
61 cm from readers’ eyes. The tracker’s output was
sampled to produce a sequence of eye fixations
recorded as x and y character positions, with start
and finish times. Before the start of the experiment,
participants received an explanation of the eyetracking procedure and were instructed to read normally and for comprehension. Participants were
seated at the eye-tracker (a bite-bar was used to
minimize head movements), and a calibration procedure was completed.
Before the start of each trial, a fixation box
appeared in the upper left half of the screen.
Once participants had fixated this box the
experimenter prompted the computer to present
a sentence, with the first character of this sentence
replacing the fixation box. The experimenter
received feedback on the estimated position of participant’s fixation point. If this did not match with
the fixation box then the experimenter recalibrated
the eye-tracker. Participants were permitted to
take breaks as required.
The sentences were divided into four lists. Each
sentence appeared once in a list, and each list
included an equal number of sentences in each
condition. The lists included an additional
45 filler sentences. Each participant viewed the
sentences in one list. Comprehension questions
(e.g., Did Fred have much money?) followed 35%
of the experimental and filler sentences. Half of
the questions had yes, and half had no answers.
Participants responded by pressing a key and
received feedback on their responses. Participants
responded correctly 90% of the time.
Results and discussion
Regions. Sentences were divided into five scoring
regions, as indicated by vertical lines in Table 1.
Region 1 was the locative, subject noun phrase,
and verb. Region 2 was the direct and indirect
object and the focus particle. Region 3 was the
replacive. Region 4 contained an additive focus
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 0000, 00 (0)
7
PATERSON ET AL.
particle and the connective because. Region 5 was
the remainder of the sentence.
Analysis. An automatic procedure pooled short
contiguous fixations. Fixations less than 80 ms
were incorporated into larger adjacent fixations
within one character. Fixations over 1,200 ms
also were deleted. Prior to analysing eye movement
data we eliminated trials where either participants
failed to read the sentence or there had been
tracker loss. This involved removing trials where
zero first-pass reading times were recorded for
Regions 2, 3, or 4, accounting for 4.9% of the data.
We computed several standard reading-time
measures (Rayner, 1998; Rayner, Sereno, Morris,
Schmauder, & Clifton, 1989). First-pass reading
time (or gaze duration for a single word)
summed the duration of fixations made on first
entering a region until exiting it. This measure is
generally taken to be an index of initial processing
effects. We also computed regression path reading
times for Regions 3 and 4. Regression path reading
time is the sum of temporally contiguous fixations
made following the onset of the first fixation
within a region until a saccade transgresses the
right region boundary (Konieczny, Hemforth,
Scheepers, & Strube, 1997; Liversedge, Paterson,
& Pickering, 1998a; Rayner & Duffy, 1986).
This measure includes fixations made to reinspect
earlier portions of text and is usually taken to
reflect an indication of initial processing difficulty
along with (at least some) time spent reinspecting
the sentence in order to recover from such difficulty. Finally, we computed total reading time,
which summed all fixations made in a region
until the participant pressed a button to indicate
that they had completed reading the sentence.
This measure is usually taken to provide a measure
of overall comprehension difficulty associated with
a region of the sentence.
Each region’s data were subjected to two 2
(particle location) 2 (contrast congruity) analyses
of variance (ANOVAs), treating participants (F1)
and sentences (F2) as random variables. We
calculated minimum F0 (min F0 ) from the F1
and F2 ANOVAs (Clark, 1973; Raaijmakers,
Schrijnemakers, & Gremmen, 1999). Table 3
shows mean reading times for Regions 2 – 4, and
Table 4 shows the inferential statistics.
First-pass reading time. Region 2 reading times
were longer when only adjoined the indirect
rather than the direct object (926 vs. 870 ms,
95% confidence interval, CI ¼ 46 ms). This
Table 3. Experiment 1: First-pass, regression path, and total reading times for Regions 2–4 of dative sentences with
only preceding the direct or indirect object and the replacive providing a congruous or incongruous contrast
Particle Position
Region
2
3
4
Measures
First-pass reading time
Regression path reading time
Total reading time
First-pass reading time
Regression path reading time
Total reading time
First-pass reading time
Regression path reading time
Total reading time
Predirect object
(only the salt to her mother)
Congruousa
Incongruousa
Preindirect object
(only the salt to her mother)
Congruousa
Incongruousa
882 (31)
997 (43)
1035 (54)
564 (23)
603 (30)
639 (26)
427 (20)
483 (21)
481 (21)
921 (36)
1067 (41)
1094 (47)
566 (21)
607 (24)
623 (23)
424 (17)
467 (17)
470 (19)
857 (35)
1003 (39)
1023 (43)
576 (21)
599 (30)
642 (29)
438 (23)
520 (31)
507 (25)
Note: Reading times in ms; standard errors in parentheses.
a
Contrast type.
8
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 0000, 00 (0)
931 (35)
1065 (46)
1109 (51)
579 (27)
620 (29)
682 (30)
427 (20)
549 (29)
523 (25)
FOCUS IDENTIFICATION DURING SENTENCE COMPREHENSION
Table 4. Inferential statistics for Experiment 1
F1
Region
First-pass
reading time
2
3
4
Regression path
2
3
4
Total reading
time
2
3
4
Source of variance
Particle location
Contrast congruity
Particle Location Congruity
Particle location
Contrast congruity
Particle Location Congruity
Particle location
Contrast congruity
Particle Location Congruity
Particle location
Contrast congruity
Particle Location Congruity
Particle location
Contrast congruity
Particle Location Congruity
Particle location
Contrast congruity
Particle Location Congruity
Particle location
Contrast congruity
Particle Location Congruity
Particle location
Contrast congruity
Particle Location Congruity
Particle location
Contrast congruity
Particle Location Congruity
Min F 0
F2
Min F 0 value
df
F1 value
MSE
df
F2 value
df
Contrast
1, 35
1, 35
1, 35
4.25
0.13
0.69
115,769
1,991
10,849
1, 31
1, 31
1, 31
5.18
0.30
1.41
1, 66
1, 59
1, 61
2.33
0.09
0.46
Contrast
1, 35
1, 35
1, 35
0.03
1.16
0.00
202
5,453
19
1, 31
1, 31
1, 31
0.07
0.47
0.11
1, 59
1, 53
1, 37
0.02
0.33
0.00
Contrast
1, 35
1, 35
1, 35
0.24
0.31
0.10
1,902
1,863
492
1, 31
1, 31
1, 31
0.51
0.71
0.00
1, 61
1, 59
1, 59
0.16
0.22
0.02
Contrast
1, 35
1, 35
1, 35
7.93
0.00
0.30
156,697
96
543
1, 31
1, 31
1, 31
7.14
0.02
0.04
1, 66
1, 40
1, 55
3.76þ
0.02
0.03
Contrast
1, 35
1, 35
1, 35
0.60
0.15
0.35
5,080
684
2,575
1, 31
1, 31
1, 31
0.79
0.01
0.51
1, 66
1, 35
1, 65
0.34
0.01
0.21
Contrast
1, 35
1, 35
1, 35
0.03
7.00
0.88
317
56,442
6,394
1, 31
1, 31
1, 31
0.08
17.98
0.83
1, 57
1, 58
1, 65
0.02
5.04
0.43
Contrast
1, 35
1, 35
1, 35
9.89
0.05
0.21
187,797
123
6,415
1, 31
1, 31
1, 31
11.01
0.00
1.35
1, 66
1, 61
1, 45
5.21
0.05
0.18
Contrast
1, 35
1, 35
1, 35
1.43
6.51
9.90
200
593
1,017
1, 31
1, 31
1, 31
1.01
3.28þ
1.58
1, 63
1, 57
1, 41
0.59
2.18
1.36
Contrast
1, 35
1, 35
1, 35
0.31
9.04
0.88
317
126,097
6,394
1, 31
1, 31
1, 31
0.02
9.90
0.79
1, 35
1, 68
1, 65
0.02
4.73
0.42
p , .05 þ .1 . p . .05; p , .01; p , .001.
effect was unanticipated but has several possible
explanations. One possibility is that there are
differential costs in computing contrastive focus
when only associates with an indirect rather than a
direct object. That is, the grammatical status of
the constituent may modulate the processing cost.
Alternatively, the effect may have occurred
because of the surface position of the particle,
readers having more difficulty in processing contrastive focus when only was flanked by the direct
and indirect objects than when it preceded them
both. It also was possible that readers prefer the
focused constituent to appear early in a sentence,
as this parallels the preference for given information
to precede new information (e.g., Chafe, 1976).
Finally, it was possible that sentences with only in
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 0000, 00 (0)
9
PATERSON ET AL.
different surface positions differ in their frequency
of usage, and that readers had less difficulty in processing the more frequent form. We return to this
unanticipated effect later. No reliable effects were
obtained at Regions 3 and 4, indicating that no
first-pass disruption was associated with processing
an incongruous replacive.
Regression path reading time. Region 2 reading
times were longer when only preceded the indirect
rather than the direct object (1,066 vs. 1,000 ms,
95% CI ¼ 47 ms), replicating the first-pass effect
for this region. At Region 4 there was a main
effect of contrast congruity with longer regression
path reading times when the contrast was incongruous (535 vs. 475 ms, 95% CI ¼ 37 ms). Thus,
regression path reading times for the postreplacive
region provided the earliest indication that sentence processing was disrupted if the replacive supplied an incongruous contrast.
Total reading time. Region 2 total reading times
were longer when only adjoined the indirect
rather than the direct object (1,102 vs. 1,029 ms,
95% CI ¼ 51 ms), replicating the first-pass effect
for this region. The Region 3 effect of contrast congruity was marginal, with longer reading times
when the replacive was incongruous (662 vs.
631 ms, 95% CI ¼ 27 ms). This effect was reliable
at Region 4, however, where total reading times
were longer when the replacive was incongruous
(515 vs. 476 ms, 95% CI ¼ 30 ms). Thus, consistent with the regression path data for Region 4, total
reading-time effects at Regions 3 and 4 indicated
that readers experienced disruption to processing
when the replacive supplied a contrast that was
incongruous with the focused referent.
To summarize, we obtained two key effects in
Experiment 1. First, we obtained clear evidence
that the inclusion of only influenced focus identification during on-line sentence processing by
causing readers to assign focus to its immediately
adjacent constituent. Consistent with this effect,
we found that readers had difficulty when the replacive was incongruous with the focused constituent,
this effect occurring in regression path and total
reading times for the postreplacive region. The
effect contrasted with the one obtained in the
10
completion data for dative sentences, which indicated that although there was a preference for
associating the focus particle with the adjacent
phrase, this preference was substantially weaker
when the particle appeared early in the sentence
(i.e., adjoining the direct object) than when it
appeared late (i.e., adjoining the indirect object).
We attributed this effect to the different grammatically permissible analyses of the sentence when the
particle was located in these two surface positions.
When it appeared early in the sentence it could
associate with only the adjacent phrase, but when
it appeared late it could associate with either of
the postverbal phrases—hence the stronger preference for completions supplying a contrast for the
adjacent constituent when the particle appeared
late in the sentence. It is noteworthy that a
similar effect did not occur in on-line sentence
comprehension, where readers associated the particle with the immediately adjacent phrase irrespective of its surface location. We return to this point
in the General Discussion. It also should be noted
that reading-time effect was delayed with respect
to the processing of the replacive, occurring in the
postreplacive region rather than at the replacive
itself. This delay in detecting an incongruous
replacive may reflect the operation of inferential
processes that are needed to evaluate the congruity
of the focused referent and the contrast supplied by
the replacive.
A second effect was obtained in first-pass and
total reading times for the region of text containing
the particle, reading times being longer when only
adjoined the indirect rather than direct object. We
attributed this effect to readers having sentenceprocessing difficulty due to either the grammatical
status of the focused constituent or the surface
position of the focus particle.
EXPERIMENT 2
In Experiment 1, we observed congruency effects
that depended on the surface position of the
focus particle in a dative construction. Readers
had more difficulty in processing a replacive
when it was incongruous with the constituent
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 0000, 00 (0)
FOCUS IDENTIFICATION DURING SENTENCE COMPREHENSION
that only adjoined, irrespective of whether this was
a direct or indirect object. This finding was consistent with the claim that readers initially associate the particle with its adjacent constituent. In
Experiment 2, we conducted a further test of this
hypothesis, using double-object sentences, like
those shown in Table 1 with the indirect object
(e.g., her mother) preceding the direct object (e.g.,
the salt) and, as in Experiment 1, with only adjoining either the direct or indirect object. As before,
the replacive provided a contrast that was either
congruous or incongruous with the constituent
adjoining the focus particle.
When only adjoins the direct object in these
sentences, it must associate with this constituent.
However, when it precedes the indirect object
the sentence has two grammatically permissible
analyses. On one analysis, the particle associates
with the indirect object, but on the other analysis
it can associate with either postverbal phrase. In
Experiment 1 the sentence-processing data (but
not the completion data) showed that there is a
strong preference for associating only with its
adjacent constituent during comprehension.
Experiment 2 investigated whether similar effects
would be observed in a syntactic structure other
than a dative construction and to further
examine the time course of focus identification.
Assuming that this is similar across the two structures, then disruption to processing should be
observed in regression path and total reading
times when the replacive is incongruous. The
results from Experiment 1 suggested that this
effect would occur at the postreplacive region.
Experiment 2 also enabled us to investigate the
effect of particle position that we obtained in
Experiment 1. In Experiment 1, the first-pass
reading times for Region 2 were inflated when
the particle adjoined the indirect object. We
offered several possible explanations for this
effect, two of which were: (a) the main effect was
due to the grammatical status of the constituent
adjoining the focus particle (i.e., the indirect
object or the direct object), or (b), it was due to
the surface position of the particle (either early
or late). We tested between these alternatives in
Experiment 2. Although the order of postverbal
phrases was reversed for the Experiment 2 sentences, their grammatical status was unchanged.
It therefore should be possible to determine
whether the effect was due to the order of the constituents or to their grammatical status. If the
effect is due to constituent order, then in line
with the Experiment 1 results, we should observe
shorter reading times when the particle appears
early in the sentence (preceding the indirect
object), than when it appears late (preceding the
direct object). The opposite pattern should occur
if the effect is due to the grammatical status of
the constituent it adjoins.
Method
Participants
A total of 32 native English speakers with normal
or corrected vision from the Universities of
Durham and Derby participated in the experiment,
with the participants from each university spread
evenly across each experimental condition.
Materials and design
The order of direct and indirect objects of the
sentences in Experiment 1 was reversed, and the
items were modified where necessary to create
double-object sentences, as shown in Table 1.
The focus particle preceded the direct or indirect
object, and the replacive continuation supplied a
congruous contrast for one of these constituents.
Thus, we manipulated two independent variables:
the position of only and the congruency of the
replacive. The dependent variables were measures
of reading time.
Norming data. As in Experiment 1, we conducted
a test to ensure that, when the replacive supplied
an appropriate contrast, this was interpreted as
contrasting with the constituent adjoined by the
focus particle and no other sentential constituent.
A total of 22 participants viewed the doubleobject sentences used in Experiment 2, with only
preceding either the direct or the indirect object,
and with the replacive supplying an congruous
contrast (e.g., At dinner, Jane passed only her
mother the salt but not her father too). An additional
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 0000, 00 (0)
11
PATERSON ET AL.
16 filler items had a ditransitive construction, with
only preceding the subject noun phrase and a
replacive supplying a congruous contrast for this
constituent. The contrastive noun was underscored
(e.g., but not the pepper as well), and participants
indicated which constituent contrasted with it.
There was only one trial in which participants
did not select the constituent that only adjoined,
indicating that, when the contrast was congruous,
the replacive was interpreted as supplying a contrast for the constituent that was adjacent to the
focus particle and not interpreted as providing a
contrast with any other sentential constituent.
We collected sentence completions in order to
assess readers’ preferences for assigning focus.
A total of 32 participants provided completions
for double-object sentences, with only preceding
the direct or indirect object. The sentences were
truncated at the replacive (e.g., At dinner, Jane
passed only her mother the salt but not . . . ). A total
of 16 filler sentences with a ditransitive construction, and with only preceding the head noun, also
were truncated at the replacive. The completions
were categorized as providing an appropriate contrast for either the direct object or the indirect
object, or providing an inappropriate contrast.
Most completions supplied an appropriate contrast, and the mean percentage proportion of
these completions is shown in Table 5.
There was a preference for interpreting sentences by associating only with its adjacent constituent, although this preference was weaker
when only preceded the indirect rather than the
direct object (i.e., when it was early rather than
late in the sentence), t(31) ¼ 2.06, p , .05. We
Table 5. Percentage completions that supplied a contrast for either
the direct object or the indirect object of double-object sentence
fragments with only
Focus particle position
Contrasting constituent
Indirect object (her mother)
Direct object (the salt)
12
Predirect object
Preindirect
(her mother
object (only her
only the salt) mother the salt)
13
84
73
24
observed a similar effect in the completion data
from Experiment 1, where we argued that when
only appeared early in the sentence, it could associate with either of the postverbal phrases, but
when it appeared late, it could associate only with
the indirect object. This explanation can account
for the completion data obtained in Experiment
2. Such an effect is consistent with participants
employing grammatical constraints to establish
which constituents can associate with the focus
particle. However, it is inconsistent with the
hypothesis that the particle associates with its
immediately adjacent constituent. The eye-tracking
data would reveal if the same preferences are
observed during normal sentence comprehension.
Procedure
Eye movements were monitored via Fourward
Technologies Dual Purkinje Eye-Trackers located in the Durham (Generation 5) and Derby
(Generation 6) eye-tracking laboratories, using the
same data acquisition and analysis software as
those in Experiment 1. The experimental sentences
were divided into four lists with eight sentences in
each condition, each sentence appearing once in
each list and each list including an additional
49 filler sentences. Each participant viewed one
list. Comprehension questions were presented
following 40% of the experimental sentences, half
having “yes” and half having “no” answers.
Participants responded by pressing a key and
received feedback on their responses, responding
correctly 91% of the time.
Results and discussion
Regions. As in Experiment 1, the sentences were
divided into five scoring regions, as indicated by
vertical lines in Table 1. Region 1 was the locative
phrase, subject noun, and verb. Region 2 was the
direct object, indirect object, and focus particle.
Region 3 was the replacive. Region 4 was the
additive focus particle and the following connective, and Region 5 was the remainder of the
sentence.
Analysis. We employed the same procedure as that
in Experiment 1 to pool contiguous fixations and
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 0000, 00 (0)
FOCUS IDENTIFICATION DURING SENTENCE COMPREHENSION
to delete short and long fixations. Prior to analysing the eye movement data we eliminated trials
where readers failed to read the sentence or there
had been tracker loss. This involved removing
trials where zero first-pass reading times were
recorded for Regions 2, 3, or 4, accounting for
10.4% of the data. Data for each region were subjected to two 2 (particle position) 2 (contrast
congruity) ANOVAs, treating participants (F1)
and sentences (F2) as random variables, and min
F0 was calculated by combining the results of the
ANOVAs based on participants and items
means. Table 6 shows the mean reading-time
data for Regions 2 –4, and Table 7 shows the
inferential statistics.
First-pass reading time. Although we anticipated a
main effect of particle position in Region 2, no
such effect occurred. However, there was a reliable
first-pass effect of particle position at Region 3,
with longer reading times when only adjoined the
direct rather than the indirect object (603 vs.
537 ms), 95% CI ¼ 38 ms. Thus, it appeared
that although the effect of particle position
observed in Region 2 in Experiment 1 also
occurred for double-object sentences in
Experiment 2, it differed in two respects. First,
the effect appeared in a region downstream. The
relative delay in the appearance of this effect may
be due to Region 2 of double-object sentences
containing fewer words and lacking syntactic
markers, as compared with dative counterparts.
The lack of syntactic markers may have created a
temporary syntactic ambiguity, whereby the sentences initially were analysed as having a transitive
construction (e.g., Jane passed her mother . . . in the
car) rather than as a ditransitive. The cost incurred
in resolving this ambiguity may have caused other
aspects of sentence processing, including focus
computation, to be delayed. Second, we found
that readers incurred a cost when the particle
occurred late in the sentence, adjoining the direct
object, despite the constituent order being reversed
for these sentences as compared to those in
Experiment 1. It therefore appears that the
surface position of the particle was responsible
for the reading-time cost observed in both experiments and not the grammatical status of the
constituents.
At Region 4 there was a significant interaction
of particle position and contrast congruity (95%
CI ¼ 32 ms), with longer first-pass reading times
when only adjoined the indirect object and the
replacive was incongruous, but no such difference
when only preceded the direct object. Thus, firstpass times indicated an incongruity effect during
Table 6. Experiment 2: First-pass, regression path, and total reading times for Regions 2–4 of double-object sentences with
only preceding the direct or indirect object and the replacive providing a congruous or incongruous contrast
Particle Position
Region
2
3
4
Measures
First-pass reading time
Regression path reading time
Total reading time
First-pass reading time
Regression path reading time
Total reading time
First-pass reading time
Regression path reading time
Total reading time
Predirect object
(her mother only the salt)
Congruousa
Incongruousa
Preindirect object
(only her mother the salt)
Congruousa
Incongruousa
786 (38)
1,085 (45)
1,240 (71)
585 (27)
686 (33)
763 (38)
412 (22)
472 (29)
489 (27)
760 (44)
1,084 (57)
1,295 (85)
532 (24)
720 (34)
745 (44)
388 (17)
477 (26)
497 (29)
757 (40)
1,047 (44)
1,262 (73)
622 (40)
745 (45)
835 (59)
400 (22)
533 (35)
515 (30)
841 (44)
1,032 (56)
1,378 (110)
542 (27)
705 (42)
811 (55)
453 (32)
676 (58)
593 (27)
Note: Reading times in ms; standard errors in parentheses.
a
Contrast type.
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 0000, 00 (0)
13
PATERSON ET AL.
Table 7. Inferential statistics for Experiment 2
F1
Region
First-pass
reading time
2
3
4
Regression path
2
3
4
Total reading
time
2
3
4
Source of variance
Particle location
Contrast congruity
Particle location contrast
congruity
Particle location
contrast congruity
Particle location contrast
congruity
Particle location
Contrast congruity
Particle location contrast
congruity
Particle location
Contrast congruity
Particle location contrast
congruity
Particle location
Contrast congruity
Particle location contrast
congruity
Particle location
Contrast congruity
Particle location contrast
congruity
Particle location
Contrast congruity
Particle location contrast
congruity
Particle location
Contrast congruity
Particle location contrast
congruity
Particle location
Contrast congruity
Particle location contrast
congruity
df
F1 value
Min F0
F2
MSE
df
F2 value
df
Min F0 value
1, 31
1, 31
1, 31
0.65
1.09
2.92
26,337
20,719
97,626
1, 31
1, 31
1, 31
1.43
0.84
2.77
1, 54
1, 61
1, 62
0.45
0.47
1.42
1, 31
1, 31
1, 31
10.64
1.31
0.64
143,069
17,592
5,705
1, 31
1, 31
1, 31
11.96
0.50
0.14
1, 66
1, 52
1, 44
5.63
0.36
0.11
1, 31
1, 31
1, 31
0.73
2.69
6.19
7,022
23,473
45,040
1, 31
1, 31
1, 31
0.62
1.34
5.00
1, 62
1, 56
1, 64
0.34
0.89
2.77þ
1, 31
1, 31
1, 31
0.03
1.87
0.12
53,057
64,648
1,591
1, 31
1, 31
1, 31
0.06
1.78
0.01
1, 56
1, 62
1, 36
0.02
0.91
0.01
1, 31
1, 31
1, 31
0.01
0.61
1.49
258
15,904
43,479
1, 31
1, 31
1, 31
0.62
0.12
1.34
1, 32
1, 43
1, 62
0.01
0.10
0.71
173,054
544,355
153,015
1, 31
1, 31
1, 31
2.71
8.44
5.34
1, 57
1, 57
1, 66
1.80
5.63
2.66þ
1, 31
1, 31
1, 31
5.38
16.88
5.32
1, 31
1, 31
1, 31
3.82
1.51
0.55
231,680
89,727
29,042
1, 31
1, 31
1, 31
0.77
1.56
1.03
1, 43
1, 62
1, 57
0.64
0.77
0.36
1, 31
1, 31
1, 31
0.55
4.26
0.02
14,040
150,552
320
1, 31
1, 31
1, 31
1.15
2.31
0.08
1, 55
1, 58
1, 46
0.37
1.50
0.02
1, 31
1, 31
1, 31
2.47
6.89
5.71
58,020
119,090
39,687
1, 31
1, 31
1, 31
4.71
5.34
2.06
1, 62
1, 64
1, 51
1.62
3.01þ
1.51
p, .05 þ .1 . p . .05; p , .01; p , .001.
early processing of the postreplacive region when
only adjoined the indirect object, but not when it
adjoined the direct object.
Regression path reading time. Regression path
reading times were longer at Region 4 when the
replacive was incongruous (605 vs. 475 ms, 95%
CI ¼ 61 ms). However, there was also a significant
interaction of particle position and contrast congruity (95% CI ¼ 61 ms). Reading times did not
14
differ when only adjoined the direct object, but
they were longer when it adjoined the indirect
object, and the replacive was incongruous. Thus,
the regression path data at the postreplacive
region showed a similar pattern of data to the
first-pass reading times for this region.
Total reading time. The Region 3 effect of contrast
congruity was not reliable, although total reading
times were numerically longer when the replacive
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 0000, 00 (0)
FOCUS IDENTIFICATION DURING SENTENCE COMPREHENSION
was incongruous (823 vs. 754 ms, 95% CI ¼
57 ms), irrespective of the particle’s surface position.
Total reading times at Region 4 were longer when
the replacive was incongruous (554 vs. 493 ms, 95%
CI ¼ 44 ms), and although the interaction was not
reliable, the numerical trend in the data was similar
to that observed in first-pass and regression path
reading times at this region. The pattern of means
observed in the total time data was probably due
to a residual influence of the differential incongruity
effects that occurred during the first pass when only
appeared at different surface locations as first-pass
fixations also contribute to the total reading time
Q2 for a region.1 Thus, whereas first-pass and
regression path reading times both showed an
early disruption effect for incongruous replacives
when the particle adjoined the indirect object, but
not when it adjoined the direct object, the total
reading times indicated that, ultimately, disruption
was experienced in both cases. It appears that, in
this experiment, the time course of incongruity
detection differed depending on the particle’s
surface position, occurring earliest when it preceded
the indirect object.
The results of Experiments 1 and 2 clearly
demonstrated that changes in the location of the
particle modulates focus effects, such that readers
interpret only as associating with the adjacent
phrase and incur a processing cost when the replacive is incongruous with this analysis. As in
Experiment 1, the effect obtained in the readingtime data differed from that obtained in the
completion data. The completion data revealed a
preference for associating only with the adjacent
constituent when the particle preceded the direct
object but not when it preceded the indirect
object, and we accounted for this effect by arguing
that when the particle appeared early in the sentence (i.e., preceding the indirect object) an analysis
was available in which the particle could associate
with either postverbal constituent. By comparison,
the reading-time data from Experiments 1 and 2
both indicated that only associates with its adjacent
constituent during sentence comprehension irrespective of which phrase it precedes.
There are two possible explanations for this
reading-time effect. First, it was possible that
only marks the focused constituent, in much the
same manner as a peak of prosodic prominence
might mark focus. If this is the case, then lexical
specification of focus by only is directly related to
the specific surface position of the particle.
Alternatively, it could be that the particle modulates an underlying focus pattern, such that
readers have a default preference for assigning
focus to a particular sentential constituent, and
the inclusion of only simply modulates this preference. We therefore conducted Experiment 3 to
investigate whether readers have default focus preferences for the sentences used in Experiments 1
and 2, by examining the processing of replacive
continuations to dative and double-object sentences that do not include only.
EXPERIMENT 3
The sentences used in Experiment 3 did not
include only. This experiment thereby enabled us
to establish whether there is a default preference
for assigning focus to either the direct or the indirect object in dative and double-object sentences. If
such a default depends either on the surface order
of the constituents (i.e., first vs. last constituent) or
their grammatical function (e.g., direct vs. indirect
object) then processing difficulty should occur
when the replacive is incongruous with the
focused constituent.
1
Second-pass reading times for Region 4, the postreplacive region, in Experiment 2 supported this interpretation of the total
reading-time data. Second-pass reading time is the sum of fixations within a region, excluding those made during the first pass,
and provides an indication of the time spent reinspecting a region of text. The second-pass reading times for Region 4 were
longer when the replacive was incongruous than when it was congruous (136 vs. 98 ms, 95% CI ¼ 33 ms), F1(1, 31) ¼ 8.84, p ,
.01, and F2(1, 31) ¼ 5.67, p , .05; min F0 (1, 59) ¼ 3.45, p ¼ .07, with no other reliable effects (Fs , 1). Thus, the numerical
pattern supporting an interaction in the total reading-time data for the postreplacive region appears to be driven primarily by
those fixations that were made during the first pass.
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 0000, 00 (0)
15
PATERSON ET AL.
Method
Participants
A total of 32 native English speakers from the
University of Leicester participated in the experiment for cash payment.
Materials and design
We used dative and double-object versions of sentences used in Experiments 1 and 2 that omitted
the particle only. These were continued with a
replacive supplying a congruous contrast for
either the direct or the indirect object. Thus, we
manipulated two independent variables: the
order of the direct and indirect objects and the
congruity of the replacive. The dependent variables were measures of reading time.
Procedure
Eye movements were monitored via a Fourward
Technologies Dual Purkinje Generation 6 EyeTracker located at the University of Leicester eyetracking laboratory, using the same data acquisition
and analysis software as those used in the previous
experiments. The experimental sentences were
divided into four lists with eight sentences in each
condition, each sentence appearing once in each
list and each list including an additional 94 filler
sentences. Each participant viewed one of the
lists. Comprehension questions (e.g., Did Fred
have much money?) were presented following 35%
of the experimental and filler sentences. Half of
the questions had yes, and half had no answers.
Participants responded by pressing a key and
received feedback on their responses. They
responded correctly 88% of the time.
Results and discussion
Regions. Sentences were divided into scoring
regions, as indicated by vertical lines in Table 1.
Region 1 was the locative, subject noun phrase,
and verb. Region 2 contained the direct and indirect object. Region 3 was the replacive. Region 4
was an additive focus particle and a connective.
Region 5 completed the sentence.
Analysis. The same procedure as that used in
Experiments 1 and 2 was used to pool short contiguous fixations and to delete long fixations. Trials
where zero first-pass reading times were recorded
for Regions 2–4 were eliminated, accounting for
11.6% of the data. Data for each region were subjected to two 2 (sentence construction) 2 (contrast
congruity) ANOVAs, treating participants (F1) and
sentences (F2) as random variables, and min F0 was
calculated from the F1 and F2 results. Table 8
Table 8. Experiment 3: First-pass, regression path, and total reading times for Regions 2–4 for dative and double-object
sentences without only and a replacive providing a congruous contrast for either the direct or indirect object
Sentence type
Region
2
3
4
Measures
First-pass reading time
Regression path reading time
Total reading time
First-pass reading time
Regression path reading time
Total reading time
First-pass reading time
Regression path reading time
Total reading time
Dative
(the salt to her mother)
Direct objecta
Indirect objecta
553 (20)
629 (24)
728 (33)
469 (26)
521 (18)
606 (17)
324 (18)
406 (27)
378 (16)
521 (20)
645 (40)
723 (42)
484 (28)
564 (18)
633 (21)
327 (21)
446 (40)
334 (15)
Double object
(her mother the salt)
Direct objecta
Indirect objecta
453 (27)
574 (29)
678 (31)
477 (27)
483 (17)
644 (21)
343 (22)
470 (36)
419 (16)
Note: Reading times in ms; standard errors in parentheses.
a
Contrast type.
16
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 0000, 00 (0)
472 (18)
565 (29)
678 (36)
482 (29)
570 (18)
666 (19)
350 (17)
449 (25)
356 (18)
FOCUS IDENTIFICATION DURING SENTENCE COMPREHENSION
Table 9. Inferential statistics for Experiment 3
F1
Region
First-pass reading
time
2
3
4
Regression path
2
3
4
Total reading time 2
3
4
Source of variance
Sentence construction
replacive
Sentence construction
Sentence construction
replacive
Sentence construction
Sentence construction
Replacive
Sentence construction
Sentence construction
Replacive
Sentence construction
Sentence construction
Replacive
Sentence construction
Sentence construction
replacive
Sentence construction
Sentence construction
Replacive
Sentence construction
Sentence construction
Replacive
Sentence construction
Sentence construction
Replacive
Sentence construction
replacive
replacive
replacive
replacive
replacive
replacive
replacive
replacive
replacive
df
F1 value
Min F0
F2
MSE
df
F2 value
1, 31 18.93 176,132 1, 31 14.46
1, 31 0.14
1,210 1, 31 0.17
1, 31 1.91
21,294 1, 31 1.43
1, 31 0.26
282 1, 31 0.01
1, 31 0.34
3,372 1, 31 0.21
1, 31 0.08
10,582 1, 31 0.13
1, 31 1.49
13,724 1, 31 1.27
1, 31 0.18
1,164 1, 31 0.22
1, 31 0.08
37 1, 31 0.00
1, 31 30.10 14,583 1, 31 6.66
1, 31 0.02
279 1, 31 0.05
1, 31 0.21
5,129 1, 31 0.65
1, 31 1.79
36,285 1, 31 1.02
1, 31 0.57
6,881 1, 31 0.29
1, 31 1.71
24,927 1, 31 1.73
1, 31 1.39
37,235 1, 31 0.88
1, 31 0.19
3,438 1, 31 0.23
1, 31 1.21
30,399 1, 31 1.68
1, 31 7.02
73,112 1, 31 3.33þ
1, 31 0.01
207 1, 31 0.04
1, 31 0.01
158 1, 31 0.03
1, 31 2.91þ
39,863 1, 31 1.09
1, 31 1.01
18,125 1, 31 1.22
1, 31 0.02
11,369 1, 31 0.23
1, 31 1.63
23,151 1, 31 1.41
1, 31 0.32
2,854 1, 31 0.43
1, 31 0.12
1,178 1, 31 0.19
df
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
Min F0 value
61
61
61
33
59
59
62
61
34
49
53
50
58
56
62
59
61
60
55
46
50
51
61
36
62
61
59
8.20
0.08
0.82
0.01
0.13
0.05
0.69
0.10
0.00
5.45
0.01
0.16
0.65
0.19
0.86
0.54
0.10
0.73
2.26
0.01
0.01
0.79
0.55
0.02
0.76
0.18
0.07
p , .05 þ .1 . p . .05; p , .01; p , .001.
reports mean reading times for Regions 2–4, and
Table 9 shows the inferential statistics.
First-pass reading time. Region 2 reading times
were longer for dative than for double-object sentences (537 vs. 463 ms, 95% CI ¼ 35 ms), possibly
due to datives including an additional word (i.e., a
preposition) at this region. As there were no other
effects, there was no evidence for default focus preferences in first-pass reading times.
Regression path reading time.. Region 2 regression
path reading times showed the same pattern as
first-pass reading times, with longer reading
times for datives (638 vs. 568 ms, 95% CI ¼
46 ms). There were no other significant effects.
Thus, like the first-pass reading times, there was
no evidence for a default focus preference.
Total reading time. Region 2 total reading times
were longer for dative than for double-object sentences (1,102 vs. 1,029 ms, 95% CI ¼ 48 ms),
replicating the effect obtained in first-pass and
regression path reading times at this region.
Region 3 reading times were numerically longer
for double-object sentences than for datives,
although this effect fell short of significance. No
other effects were obtained at Regions 3 or
4. Thus, the total reading-time data provided no
evidence for default focus preferences. Note that
the results contrast to those obtained in
Experiments 1 and 2, where we observed clear
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 0000, 00 (0)
17
PATERSON ET AL.
congruency effects at Region 4 for regression path
and total reading times.
In summary, Experiment 3 produced no
reading-time effects that were consistent with
readers having a default for assigning focus on
the basis of either the surface order or grammatical
function of the direct and indirect objects.
Therefore, it appeared that the effects obtained
at the postreplacive region in Experiments 1 and
2 were due to only marking the focused sentential
constituent and readers incurring a processing cost
when the replacive was incongruous with this
constituent.
GENERAL DISCUSSION
Our experiments produced two key findings. First,
there was compelling evidence that contrastive
focus is computed on-line during the comprehension of sentences containing only. The effects of its
computation were clearly observed in the readingtime data for the postreplacive region in
Experiments 1 and 2, where readers incurred a
cost when the replacive was incongruous with
the constituent that only adjoined. As no such
effects were observed in Experiment 3 for sentences that omitted the focus particle, it was
clear that the presence of the particle caused the
effects. Therefore, we may conclude that only
evoked focus effects by indicating that a contrast
was to be made between the referent of a particular
syntactic constituent and its alternatives. The
effects occurred in early measures of processing
for the postreplacive region rather than the replacive region itself, and we suggested that this delay
in the detection of the incongruency was attributable to the operation of inferential processes to
evaluate the congruency of the supplied contrast.
Despite the incongruency effect being delayed,
our results nevertheless indicate that contrastive
focus is computed during normal sentence comprehension and sufficiently rapidly for its referential consequences to affect sentence processing.
In this respect the results are consistent with Ni
et al.’s (1996) assumption that contrastive focus
is processed automatically during sentence
18
comprehension, although the present results are
not informative about its putative effects on syntactic ambiguity resolution.
The second key finding was that the surface
position of the particle modulated the incongruency effects. If, for the moment, we concentrate
on total reading-time effects at the postreplacive
region, then readers had difficulty in
Experiments 1 and 2 when the replacive contrasted
with the constituent that only adjoined, regardless
of its grammatical status or its location in the sentence. Thus, the data indicated that readers were
sensitive to the surface position of the particle
and that, depending on its location, the particle
was interpreted as associating with a different constituent. The effect is exactly what should happen
if sentences are processed in line with our working
hypothesis that only associates locally, to the
immediately adjacent constituent, during sentence
comprehension. According to this hypothesis,
when the particle adjoins the direct object (e.g.,
the salt) there is a processing cost for replacives
that are incongruous with this constituent (e.g.,
but not her father). Similarly, when it adjoins the
indirect object (e.g., her mother) incongruous replacives (e.g., the pepper versus her father) incur a cost.
As we obtained disruption to processing for incongruous replacives in both cases, we may conclude
that readers identified focus in line with this
hypothesis.
Total reading times are informative about
overall difficulty in processing portions of text,
whereas first-pass and regression path measures
are sensitive to the early processing of text and
therefore were potentially informative about the
time course of effects. Since in Experiment 1 we
obtained a main effect of incongruency that did
not interact with particle position in the regression
path reading times for the postreplacive region, it
appeared that the incongruency effect had the
same time course of influence irrespective of
whether the particle adjoined the direct or indirect
object of dative sentences. By comparison, in
Experiment 2, we obtained an interaction of congruency and particle position in first-pass and
regression path reading times at the postreplacive
region. This indicated that disruption occurred
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 0000, 00 (0)
FOCUS IDENTIFICATION DURING SENTENCE COMPREHENSION
for incongruous replacives when the particle
adjoined the indirect object, but not when it
adjoined the direct object. However, the interaction was observed only in early measures of sentence processing, and the total reading-time data
indicated that ultimately the incongruency was
detected in both cases. Thus, as in Experiment 1,
the Experiment 2 results indicated that readers
identified focus by associating only with its adjacent constituent, but that the resulting incongruency effects were observed earliest in
processing when the particle adjoined the indirect
object.
This is an intriguing effect, since it suggests
that particle position modulated the time course
over which contrastive focus was processed for
double-object sentences, but not for datives. It is
likely that structural differences between the two
constructions were responsible for this effect, the
obvious differences being the reordering of the
postverbal phrases and the omission of syntactic
markers (i.e., the preposition) in double-object
constructions. One possibility is that the lack of
syntactic markers introduced a short-lived ambiguity, whereby when only adjoined the direct
object the double-object sentences were temporarily ambiguous between analyses in which the
particle associated with either the constituent to
its right or the one to its left. Note that this possibility cannot arise in similar dative sentences.
In the present study we followed theoretical
accounts of focus interpretation (e.g., Jackendoff,
1972; Reinhart, 1999) by assuming that only
ranges over constituents within its syntactic
domain, which in English are those to the right
of the focus particle. While this may be the preferred processing strategy, and the total readingtime data from Experiments 1 and 2 suggest that
it is, it is possible that readers can, under certain
circumstances, process the particle as ranging
over constituents on its left. This certainly
appears to be the case when the particle occurs in
a sentence-final position, as in Mary kissed John
only. Readers are likely to interpret this sentence
with only ranging over John and to understand it
to mean that Mary did not kiss anyone other
than John. Within the linguistics literature, such
effects can be explained by rules, such as extraposition, which govern the dislocation of words and
phrases from their normal surface position in a
sentence. However, from a processing perspective,
it appears that on encountering a focus particle, the
language processor often must select between analyses in which the particle associates with a constituent either to its left or to its right.
No such ambiguity was experienced for dative
sentences in Experiment 1, presumably because
these contained a syntactic marker that ruled out
the possibility of an analysis with only associating
with a constituent to its left when the particle
appeared late in the sentence. Thus, focus identification could proceed unambiguously for datives.
The double-object sentences used in Experiment
2 omitted syntactic markers from the clause containing the postverbal phrases, and therefore
when only appeared late in the sentence readers
may have had temporary difficulty in unambiguously associating the particle with a particular constituent, thereby causing a delay in the detection of
the incongruity. If this explanation is correct, it
suggests that grammatical restrictions on the particle’s range are not an absolute determinant of
focus identification but may instead impose a violable constraint on processing. It will be of interest
to investigate the conditions under which these
restrictions may be violated, particularly those in
which the focus particle is interpreted as associating with a constituent to its left.
The effects obtained in reading-time measures
contrast with those in the sentence completion
data. The reading-time data showed a preference
for associating the focus particle with its immediately adjacent constituent. By comparison, the
completion data for both dative and doubleobject sentences showed that when the particle
appeared late in the sentence the completions
almost always supplied a contrast for its immediately adjacent constituent, but when the particle
appeared early in the sentence, although the
majority of completions still supplied a contrast
for the adjacent constituent, the preference for
this analysis was much weaker. Thus, there was a
strong preference for associating the particle with
the adjacent phrase when it only appeared late in
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 0000, 00 (0)
19
PATERSON ET AL.
the sentence, but when it appeared early it could
associate with either of the postverbal phrases.
These data are important for several reasons.
First, they show that the eye-tracking effects
were neither obvious nor a foregone conclusion.
Second, the mismatch between the two sources
of data accentuates the importance of using appropriately sensitive measures, such as eye tracking,
when investigating sentence processing. Since
the sentence completion methodology is off-line,
it does not provide a good measure of the
moment-to-moment processes associated with
language processing. It can, however, provide an
indication of the interpretations that potentially
are available to the reader. In the present case,
the results from the completion task suggest that
participants can access the alternative grammatically permissible focus interpretations for particular sentence constructions in an off-line task.
When the focus particle appeared late in the sentence, the only possible analysis was one with it
associating with the adjacent phrase. However,
when it appeared early in the sentence it could
be interpreted as associating with the adjacent
phrase, but there was another possible analysis in
which it could associate with either of the postverbal phrases. The data therefore showed that
whereas participants favoured an analysis that
associated the focus particle with its adjacent syntactic constituent during on-line sentence processing, participants were able to arrive at a
dispreferred focus identification when the grammatical constraints permitted such an analysis,
and participants had time to compute it.
In addition to the reading-time effects at the
postreplacive region, we also observed an unexpected effect for an earlier region of text in
Experiments 1 and 2. In Experiment 1 there was
an early first-pass effect of particle position, such
that reading times at Region 2 were longer when
the particle adjoined the indirect rather than
direct object. We argued that this effect might
have been due to the surface position of the particle in the sentence (i.e., early versus late).
Alternatively, it might have occurred due to the
syntactic status of the constituent with which the
particle associated. That is, increased processing
20
costs may be associated with focus identification
when a particle associates with an indirect rather
than a direct object.
In Experiment 2, there was a main effect of particle position in first-pass reading times, but with
processing taking longest when the particle
adjoined the direct object. We believe that this
effect is the same influence that caused the firstpass effect at Region 2 in Experiment 1. Given
that the order of the postverbal phrases was
reversed in the Experiment 2 sentences as compared with Experiment 1, it appeared that it
was the surface position of the particle rather
than the syntactic status of the adjoining constituent that dictated the ease of processing
shortly after the particle was first encountered.
Note, however, that the Experiment 2 effect
occurred in Region 3 rather than Region 2 as it
had in Experiment 1. We attributed the delay in
the effect to the double-object constructions in
Experiment 2 having fewer words and lacking
syntactic markers in the region containing the
focus particle, as compared with the sentences in
Experiment 1. The greater density of linguistic
information in the Experiment 2 sentences, and
the lack of syntactic markers, may have impeded
focus computation when the particle adjoined
the direct object, thereby accounting for the
delayed effect.
There are alternative possible explanations for
this effect. One is that there is a preference for
focused constituents to occur early in a sentence,
paralleling the ordering of “given” and “new”
information (e.g., Chafe, 1976). Alternatively, as
Bouma, Hendriks, and Hoeksema (in press) have Q3
shown, sentences with only in different syntactic
positions differ in their frequency of usage, and
our readers may have had less difficulty in processing the more frequent form. Both explanations
might be correct, since if a sentence violates a
preferred sentential construction, it also is likely
to occur less frequently. Whatever the precise
explanation, the effect does appear to arise from
the surface position of the particle and is therefore associated with processes involved in focus
identification. This in turn indicates that such
processes are initiated during the processing
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 0000, 00 (0)
FOCUS IDENTIFICATION DURING SENTENCE COMPREHENSION
of the clause containing the particle, thereby providing further evidence that contrastive focus
is processed on-line during normal sentence
comprehension.
In conclusion, our experiments provided clear
and compelling evidence for the rapid computation of contrastive focus during sentence comprehension, demonstrating that the surface
position of a focus particle supplies important
information concerning which syntactic constituent it associates with. Our data indicate that
during on-line sentence processing the particle
initially associates locally, to its adjacent phrase,
but that other grammatically permissible possibilities can be computed when time is available. The
effects are consistent with theoretical accounts
that specify grammatical restrictions on focus
identification (e.g., Jackendoff, 1972; Reinhart,
1999), although further work is needed to determine the nature and limits of such constraints on
sentence processing.
Original manuscript received 26 September 2006
Accepted revision received 25 October 2006
First published online day month year
REFERENCES
Birch, S., & Clifton, C. (1995). Focus, accent, and argument structure: Effects on language comprehension.
Language and Speech, 38, 365– 391.
Birch, S., & Clifton, C. (2002). Effects of varying focus
and accenting of adjuncts on the comprehension of
utterances. Journal of Memory and Language, 47,
571– 588.
Bock, J. K., & Mazzella, J. R. (1983). Intonational
marking of given and new information: Some consequences for comprehension. Memory & Cognition,
11, 64 – 76.
Q3 Bouma, G., Hendriks, P., & Hoeksema, J. (in press).
Focus particles inside prepositional phrases: A comparison of Dutch, English and German. Journal of
Comparative Germanic Linguistics.
Q4 Carlson, K. (2004). Only in context. Paper presented at
the Architectures and Mechanisms for Language
Processing conference, Aix-en-Provence, France.
Carlson, K., Frazier, L., Clifton, C., & Dickey, M. W.
Q5
(2005). How contrastive is contrastive focus? Paper
presented at the 18th Annual CUNY Sentence
Processing Conference, Tucson, AZ.
Chafe, W. (1976). Giveness, contrastiveness, definiteness, subjects, topics, and point of view. In
C. N. Li (Ed.), Subject and topic (pp. 225– 255).
New York: Academic Press.
Chomsky, N. (1971). Deep structure, surface structure,
and semantic interpretation. In D. Steinberg &
L. Jakobovits (Eds.), Semantics: An interdisciplinary
reader in philosophy, linguistics, and philosophy
(pp. 183– 216). New York: Cambridge University
Press.
Clark, H. H. (1973). The language-as-fixed-effect
fallacy: A critique of language statistics in psychological research. Journal of Verbal Learning and
Verbal Behavior, 12, 335– 359.
Clifton, C., Bock, J., & Rado, J. (2000). Effects of
the focus particle only and intrinsic contrast on
comprehension of reduced relative clauses. In A.
Kennedy, R. Radach, D. Heller, & J. Pynte (Eds.),
Reading as a perceptual process. Amsterdam: Elsevier.
Clifton, C., Kennison, S., & Albrecht, J. (1997). Q6
Reading the words her, his, and him: Implications
for parsing principles based on frequency and
structure. Journal of Memory and Language, 36,
276– 292.
Crain, S., Ni, W., & Conway, L. (1994). Learning,
parsing and modularity. In C. Clifton, L. Frazier, &
K. Rayner (Eds.), Perspectives on sentence processing.
Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Crain, S., & Steedman, M. (1985). On not being led up Q7
the garden path: The use of context by the psychological syntax processor. In D. R. Dowty,
L. Kartunnen, & A. M. Zwicky (Eds.), Natural
language parsing: Psychological, computational, and
theoretical perspectives. Cambridge, UK: Cambridge
University Press.
Cutler, A., & Fodor, J. A. (1979). Semantic focus and
sentence comprehension. Cognition, 7, 49 –59.
Filik, R., Paterson, K. B., & Liversedge, S. P. (2005).
Parsing with focus particles in context: Evidence
from eye movements in reading. Journal of Memory
and Language, 53, 473– 495.
Fodor, J. D. (2002). Prosodic disambiguation in silent
reading. In M. Hirotani (Ed.), Proceedings of NELS
32. Amherst, MA: University of Massachusetts,
GLSA.
Frazier, L., & Rayner, K. (1982). Making and correcting
errors during sentence comprehension: Eye movements in the analysis of structurally ambiguous sentences. Cognitive Psychology, 14, 178– 210.
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 0000, 00 (0)
21
PATERSON ET AL.
Gennari, S. Meroni, L., & Crain, S. (2004). Rapid
relief of stress in dealing with ambiguity. In J. C.
Trueswell & M. K. Tanenhaus (Eds.), Approaches
to studying world-situated language use: Bridging the
language-as-product and language-as-action traditions.
Cambridge, MA: MIT Press.
Halliday, M. A. K. (1967). Notes on transitivity and
theme in English, Part 2. Journal of Linguistics, 3,
199– 244.
Hornby, P. A. (1974). Surface structure and presupposition. Journal of Verbal Leaning and Verbal Behavior,
13, 530– 538.
Jackendoff, R. S. (1972). Semantic interpretation in generative grammar. Cambridge, MA: MIT Press.
Kadmon, N. (2001). Formal pragmatics. London:
Blackwell.
Kiss, K. É. (1998). Identificational focus versus information focus. Language, 74, 245– 273.
Konieczny, L., Hemforth, B., Scheepers, C., & Strube,
G. (1997). The role of lexical heads in parsing.
Language and Cognitive Processes, 12, 307– 348.
König, E. (1991). The meaning of focus particles. London:
Routledge.
Krifka, M. (1992). A compositional semantics for
multiple focus constructions. In J. Jacobs (Ed.),
Informationsstruktur und Grammatik. Opladen,
Germany: Westdeutscher Verlag.
Liversedge, S. P., & Findlay, J. M. (2000). Saccadic eye
movements and cognition. Trends in Cognitive
Science, 4, 6 – 14.
Liversedge, S. P., Paterson, K. B., & Clayes, E. (2002).
The influence of “only” on syntactic processing of
“long” reduced relative clause sentences. Quarterly
Journal of Experimental Psychology, 55A, 225– 241.
Liversedge, S. P., Paterson, K. B., & Pickering, M. J.
(1998a). Eye movements and measures of reading
time. In G. Underwood (Ed.), Eye guidance in
reading and scene perception. Oxford, UK: Elsevier
Science.
Liversedge, S. P., Pickering, M. J., Branigan, H. P., &
Q8
Van Gompel, R. P. G. (1998b). Processing arguments and adjuncts in isolation and in context: The
case of by-phrase ambiguities in passives. Journal of
Experimental Psychology: Learning, Memory and
Cognition, 24, 461– 475.
Malt, B. (1985). The role of discourse structure in
understanding anaphora. Journal of Memory and
Language, 24, 271– 289.
Q9 Murray, W. S., & Liversedge, S. P. (1994). Referential
context effects on syntactic & processing. In
C. Clifton Jr., L. Frazier, & K. Rayner (Eds.),
22
Perspectives on sentence processing. Hillsdale, NJ:
Lawrence Erlbaum Associates, Inc.
Ni, W., Crain, S., & Shankweiler, D. (1996).
Sidestepping garden paths: The contribution of
syntax, semantics and plausibility in resolving
ambiguities. Language and Cognitive Processes, 11,
283– 334.
Noteboom, S. G., & Kruyt, J. G. (1987). Accent, focus
distribution, and perceived distribution of given and
new information: An experiment. Journal of the
Acoustic Society of America, 82, 1512– 1524.
Paterson, K. B., Liversedge, S. P., Rowland, C., &
Filik, R. (2003). Children’s comprehension of
sentences with focus particles. Cognition, 89,
263– 294.
Paterson, K. B., Liversedge, S. P., & Underwood, G.
(1999). The influence of focus operators on syntactic
processing of “short” reduced relative clause sentences. Quarterly Journal of Experimental Psychology,
52A, 717– 737.
Paterson, K. B., Liversedge, S. P., White, D., Filik, R., &
Jaz, K. (2006). Children’s interpretations of ambiguous sentences with “only”. Language Acquisition, 13,
253– 284.
Philip, W., & Lynch, E. (1999). Felicity, relevance, and
acquisition of the grammar of every and only. In
S. C. Howell, S. A. Fish, & T. Keith-Lucas (Eds.),
Proceedings of the 24th Annual Boston University
Conference on Language Development. Somerville,
MA: Cascadilla Press.
Pickering, M. J., Traxler, M. J., & Crocker, M. W.
(2000). Ambiguity resolution in sentence processing: Q10
Evidence against frequency-based accounts. Journal
of Memory and Language, 43, 447– 475.
Raaijmakers, J. G. W., Schrijnemakers, J. M. C., &
Gremmen, F. (1999). How to deal with “the
language-as-fixed-effects-fallacy”: Common misconceptions and alternative solutions. Journal of
Memory and Language, 41, 416– 426.
Rayner, K. (1998). Eye movements in reading and information processing: 20 years of research. Psychological
Bulletin, 124, 372– 422.
Rayner, K., & Duffy, S. A. (1986). Lexical complexity
and fixation times in reading: Effects of word frequency, verb complexity, and lexical ambiguity.
Memory & Cognition, 14, 191– 201.
Rayner, K., Sereno, S. C., Morris, R. K., Schmauder,
A. R., & Clifton, C. (1989). Eye movements
and on-line language comprehension processes
(special issue). Language and Cognitive Processes, 4,
21 – 49.
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 0000, 00 (0)
FOCUS IDENTIFICATION DURING SENTENCE COMPREHENSION
Reinhart, T. (1999). The processing cost of reference-set
computation: Guess patterns in acquisition (OTS
working papers in linguistics). Utrecht, The
Netherlands: Utrecht University.
Rochemont, M. S., & Culicover, P. W. (1990). English
focus and constructions and the theory of grammar.
New York: Cambridge University Press.
Rooth, M. (1992). A theory of focus interpretation.
Natural Language Semantics, 1, 75 – 116.
Schafer, A., Carlson, K., Clifton, C., & Frazier, L.
(2000). Focus and the interpretation of pitch
accent: Disambiguating embedded questions.
Language and Speech, 43, 75 – 105.
Sedivy, J. C. (2002). Invoking discourse-based contrast
sets and resolving syntactic ambiguities. Journal of
Memory and Language, 46, 341– 370.
Selkirk, E. O. (1995). Sentence prosody: Intonation, stress
and phrasing. In J. Godsmith (Ed.), Handbook of phonological theory (pp. 550–569). London: Blackwell.
APPENDIX
Q13Dative Constructions used in Experiment 1
In the following dative constructions used in Experiment 1, only
occurs in one of two surface positions, and slashes denote the
replacive alternatives. Deleting the preposition and reversing
the order of direct and indirect objects created the doubleobject sentences used in Experiment 2.
1. At weekends, John coached [only] football to [only] boys and
not rugby/girls as well because he didn’t have enough time.
2. At dinner, Lucy passed [only] the salt to [only] her mother but
not the pepper/her father as well because she couldn’t reach.
3. At the farm, John fed [only] carrots to [only] horses but not
straw/sheep too as he was told not to.
4. While abroad, Dan sent [only] postcards to [only] his parents
and not letters/his friends too because the airmail was
expensive.
5. At the chip shop, Harry bought [only] chips for [only] Sally
but not pop/Sue as well because he didn’t get paid until the
weekend.
6. When the pie was stolen, Zoe told [only] lies to [only] her
Gran and not the truth/Mummy also because she was very
scared.
7. On 13 July, the king granted [only] land to [only] dukes and
not money/lords as well since he didn’t have enough to go
around.
8. At the park, Tim threw [only] the ball to [only] his friend
and not the Frisbee/his terrier as well because it was
getting late.
Singer, M. (1976). Thematic structure and integration
of thematic information. Journal of Verbal Learning
and Verbal Behavior, 19, 573– 582.
Stolterfoht, B., Friederici, A. D, Alter, K., & Steube,
A. (2003). The difference between the processing of Q11
implicit prosody and focus structure during reading:
Evidence from event-related brain potentials. Poster
presented at the 6th annual CUNY Conference on
Human Sentence Processing, Cambridge, MA.
Terken, J., & Noteboom, S. G. (1987). Opposite
effects of accentuation and deaccentuation on
verification latencies for given and new information,
Language and Cognitive Processes, 2, 145– 163.
Van Gompel, R. P. G., Pickering, M. J., & Traxler, M. J.
(2000). Syntactic ambiguity resolution is not a form of Q12
lexical ambiguity resolution. In A. Kennedy, R.
Radach, D. Heller, & J. Pynte (Eds.), Reading as a
perceptual process. Amsterdam, The Netherlands:
Elsevier.
9. For fathers day, Jane bought [only] beer for [only] Dad and
not whiskey/Grandad too because she couldn’t afford to.
10. At the awards the host presented [only] plaques to [only]
the winners and not trophies/nominees too as this what
the rules said.
11. When he moved, Jack left [only] books to [only] his dad
but not his clothes/his sister as well since it wouldn’t be
appreciated.
12. During the year, the brewery supplied [only] beers to [only]
pubs and not wine/clubs also, as this was part of their policy.
13. At the zoo, Paul fed [only] meat to [only] cats and not
grain/birds as well because the notice said so.
14. At the end of term, David gave [only] cards to [only] his
friends but not presents/teachers too as he didn’t get
enough pocket money.
15. At Christmas, Father carved only [pork] for [only]
Grandma and not turkey/Auntie too because he’d run out.
16. On the beach, Danny collected [only] pebbles for [only]
mother but not shells Auntie/too, as he was tired.
17. After work, the musician trained [only] clarinet to [only]
prodigies but not bassoon/novices as well because he
didn’t have enough time.
18. For charity, grandma knitted [only] scarves for [only]
children but not scarves/adults also as she did not have
enough wool.
19. Before dinner, mother refused [only] sweets to[only] the
children but not drinks/Father as well as she knew this
would be best.
20. For the Ball, Kerry lent [only] her watch to [only] her
friend but not her pearls/her sister too as she did not
trust her enough.
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 0000, 00 (0)
23
PATERSON ET AL.
21. At Christmas, Fred sent [only] chocolates to [only] mother
but not flowers/grandma too because he didn’t have much
money.
22. In the war, the Red Cross provided [only] food for [only]
civilians and not clothing/soldiers too as the charity’s
funds didn’t stretch that far.
23. On death, Lord Smith left [only] his house to [only] his
daughter and not his farm/wife too because he was mean
and unpredictable.
24. At the wine store, Rebecca sold [only] wine to [only]
adults but not kids/beer too as these were her instructions.
25. On Tuesday evenings, Dan taught [only] Judo to [only]
kids but not karate/adults too because he didn’t enjoy it
as much.
26. On sports day, the teacher awarded [only] a shield to [only]
winners and not medals/losers as well because he was a fair
judge.
24
27. At the pub, Joe bought [only] a drink for [only] Jane but
not crisps/Carole too, because he didn’t want to break
into his ten pound note.
28. After a holiday in France, Eddy brought [only] cheese for
[only] mum but not wine/Gran too as he was allowed a
limited amount of luggage.
29. At playtime, Helen told [only] secrets to [only] girls
and not lies/boys as well because she preferred it this way.
30. Before the lecture, Sam lent [only] a pen to [only] Jill
and not paper/Susie as well because he didn’t have
enough.
31. On Halloween, Emily offered [only] sweets to [only] children but not adults/crisps as well as she had a limited
number of treats.
32. At the show, Susan presented [only] flowers to [only] the
singer but not chocolates/the conductor too since this
was standard theatre practice.
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 0000, 00 (0)
PQJE209998
Queries
Keith Rayner
Q1
Q2
Running head ok as abbreviated?
Superscript 2 changed to 1, as there is no other footnote. OK?
Q3
Q4
Bouma, Hendriks, & Hoeksema (in press). Still in press?
Carlson, K. (2004). Please give month of conference.
Q5
Carlson, Frazier, Clifton, & Dickey (2005). Please give month of conference.
Q6
Q7
Clifton, Kennison, & Albrecht (1997). Text citation?
Crain & Steedman (1985). Text citation?
Q8
Q9
Liversedge, Pickering, Branigan, & Van Gompel (1998b). Text citation?
Murray & Liversedge (1994). Text citation?
Q10
Pickering, Traxler, & Crocker (2000). Text citation?
Q11
Q12
Stolterfoht, Friederici, Alter, & Steube (2003). Please give month.
Van Gompel, Pickering, & Traxler (2000). Text citation?
Q13
Q14
Appendix title OK as inserted? [each appendix must have a title (APA5: 3:90, p. 205)]
This wasn’t really a table; as a list it should have been inserted into the text, but in columns it can
be called a table for convenience as it is referred to so often throughout.