Predicting foreign-accent adaptation in older adults

24 Pages·2012·0.26 MB·English

This article was downloaded by: [Max Planck Institut Fur Psycholinguistik]
On: 28 August 2012, At: 00:37
Publisher: Psychology Press
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office:
Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
The Quarterly Journal of Experimental
Psychology
Publication details, including instructions for authors and subscription
information:
http://www.tandfonline.com/loi/pqje20
Predicting foreign-accent adaptation in older
adults
Esther Janse a b & Patti Adank c
a Max Planck Institute for Psycholinguistics, Nijmegen, The Netherlands
b Centre for Language Studies, Radboud University Nijmegen, Nijmegen,
The Netherlands
c School of Psychological Sciences, University of Manchester, Manchester,
UK
Accepted author version posted online: 23 Jan 2012. Version of record first
published: 25 Apr 2012
To cite this article: Esther Janse & Patti Adank (2012): Predicting foreign-accent adaptation in older adults,
The Quarterly Journal of Experimental Psychology, 65:8, 1563-1585
To link to this article: http://dx.doi.org/10.1080/17470218.2012.658822
PLEASE SCROLL DOWN FOR ARTICLE
Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions
This article may be used for research, teaching, and private study purposes. Any substantial
or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or
distribution in any form to anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation that the
contents will be complete or accurate or up to date. The accuracy of any instructions, formulae,
and drug doses should be independently verified with primary sources. The publisher shall not
be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or
howsoever caused arising directly or indirectly in connection with or arising out of the use of this
material.
THEQUARTERLYJOURNALOFEXPERIMENTALPSYCHOLOGY
2012,65(8),1563–1585
Predicting foreign-accent adaptation in older adults
Esther Janse1,2 and Patti Adank3
1MaxPlanckInstituteforPsycholinguistics,Nijmegen,TheNetherlands
2CentreforLanguageStudies,RadboudUniversityNijmegen,Nijmegen,TheNetherlands
3SchoolofPsychologicalSciences,UniversityofManchester,Manchester,UK
2
1
20 Weinvestigated comprehensionof and adaptation tospeech in anunfamiliar accent inolder adults.
st Participants performed a speeded sentence verification task for accented sentences: one group upon
u
g auditory-only presentation, and the other group upon audiovisual presentation. Our questions were
u
A whetheraudiovisualpresentationwouldfacilitateadaptationtothenovelaccent,andwhichcognitive
28 andlinguisticmeasureswouldpredictadaptation.Participantswerethereforetestedonarangeofback-
7 groundtests:hearingacuity,auditoryverbalshort-termmemory,workingmemory,attention-switching
3
0: control, selective attention, and vocabulary knowledge. Both auditory-only and audiovisual groups
0
at showedimprovedaccuracyanddecreasingresponsetimesoverthecourseoftheexperiment,effectively
] showingaccentadaptation.Eventhoughthetotalamountofimprovementwassimilarfortheauditory-
k
sti onlyandaudiovisualgroups,initialrateofadaptationwasfasterintheaudiovisualgroup.Hearingsen-
ui sitivityandshort-termandworkingmemorymeasureswereassociatedwithefficientprocessingofthe
g
n novel accent. Analysis of the relationship between accent comprehension and the background tests
holi revealed furthermore that selective attention and vocabulary size predicted the amount of adaptation
c
y overthecourseoftheexperiment.Theseresultssuggestthatvocabularyknowledgeandattentionalabil-
s
P itiesfacilitatetheattention-shiftingstrategiesproposedtoberequiredforperceptuallearning.
r
u
F
ut Keywords: Speech perception; Perceptual adaptation; Audiovisual information; Ageing; Individual
stit differences.
n
I
k
c
n
a
x Pl Much of everyday speech comprehension occurs semanticverificationtask(i.e.,reportingwhethera
Ma under listening conditions that are less than ideal, sentence such as “dogs have four ears” is true or
y [ due to background noise, regional accents, or false)spokeninaregionalaccenttheyarenotfam-
d b speech rate differences, to name a few common iliar with, they show slower response times and
de everyday variations in the speech signal. Listeners higher error scores than when they listen to a
a
o are generally able to successfully comprehend known accent or the standard variant of their
nl
w speech under such adverse listening conditions. language (Adank, Evans, Stuart-Smith, & Scott,
o
D Nevertheless,comprehensionisoftenmoreeffortful 2009).
andlessefficientthanunderlessadverseconditions. This ability to effectively comprehend speech
For instance, when listeners are performing a under challenging listening conditions deteriorates
CorrespondenceshouldbeaddressedtoEstherJanse,CenterforLanguageStudiesRadboudUniversityNijmegen&MaxPlanck
InstituteforPsycholinguisticsNijmegen,POBox310,6500AHNijmegen,TheNetherlands.E-mail:[email protected]
#2012TheExperimentalPsychologySociety 1563
http://www.psypress.com/qjep http://dx.doi.org/10.1080/17470218.2012.658822
JANSEANDADANK
as humans age (Dubno, Dirks, & Morgan, 1984), younger adults, older adults failed to transfer this
mainly due to a high-frequency hearing loss learning to a different compression rate and did
(Gates, Cooper, Kannel, & Miller, 1990). notshowadditionalbenefitwithadditionalpractice
Independent of elevated hearing thresholds, there withtime-compressedsentences.Inanotherstudy,
areindications thatauditoryprocessing—temporal Adank and Janse (2010) compared younger and
processinginparticular—maybeimpairedinolder older adults on their ability to perceptually adapt
adults (e.g., Fitzgibbons & Gordon-Salant, 1998; to speech in an unfamiliar accent, presented in
Grose, Hall, & Buss, 2006; Humes, Kewley-Port, background noise. They equated task difficulty
Fogerty, & Kinney, 2010, and cf. Gordon-Salant, acrosslistenersthroughtheuseofanadaptivestair-
Frisina, Popper, & Fay, 2010, for an overview of caseprocedureinwhichtaskperformancewaskept
2 auditoryageing).Additionally,age-relateddeclines constant at 50% (Kalikow, Stevens, & Elliott,
1
0 in cognitive function (e.g., working memory, 1977; Plomp & Mimpen, 1979). Adaptation to
2
st mental flexibility; Salthouse, Atkinson, & Berish, the accented speech was assessed by evaluating
u
g 2003)maycompromiseeaseofspeechcomprehen- the change in signal-to-noise ratio at which the
u
A sion (cf. Wingfield & Stine-Morrow, 2000). A sentences were presented to the listeners. A
8
2 recent study illustrated this deterioration in the decrease in signal-to-noise ratio over the course of
7
3 ability to process distorted speech by showing that the experiment (which means that participants
0:
0 older adults were more negatively affected by an couldgraduallytoleratemorenoise)signalledadap-
at unfamiliar accent than younger listeners, relative tation. The results showed that the age groups
]
k to their performance on standard Dutch (Adank ended up with a similar amount of adaptation to
sti
ui & Janse, 2010). This is problematic because in the speech in the unfamiliar accent, but that the
g
n societies that become increasingly multicultural, age groups differed in their adaptation curve:
oli such as the UK, older adults may have to interact Older adults reached plateau performance earlier
h
yc more often with people coming from different whereas the younger listeners continued to
s
P language backgrounds, including health care improve with added exposure.
r
u providers. The ability to adapt in older adults may be
F
ut Nevertheless, even thoughsuchchallenginglis- affectedbyage-relateddeclinesincognitiveabilities
stit tening conditions can pose problems for listeners, (cf.Kennedy,Rodrigue,Head,Gunning-Dixon,&
n
I there is compelling evidence that listeners are able Raz, 2009). In this study, we therefore focused on
k
c to quickly perceptually adapt to distortions of the individualdifferencesintheabilitytoadapttodis-
n
a
Pl speech signal, including noise-vocoded speech torted speech. We investigated which individual
x (Shannon, Zeng, Kamath, Wygonski, & Ekelid, abilities predict adaptation in order to study the
a
M 1995), time-compressed speech (Dupoux & mechanisms underlying perceptual learning.
[
y Green, 1997), synthetic speech (Greenspan, Older adults may show more variability in speech
b
d Nusbaum, & Pisoni, 1988), and foreign (Bradlow performance, and in their auditory, linguistic, and
e
d & Bent, 2008; Clarke & Garrett, 2004) and cognitive abilities, than younger adults. Service
a
o
nl regional (Clopper & Bradlow, 2008; Maye, Aslin, and Craik (1993) found that a repetition task (as
w
& Tanenhaus, 2008) accented speech. This ability an index of phonological memory) predicted
o
D
to quickly adapt appears to be preserved in older foreign vocabulary learning in their sample of
adults (Golomb, Peelle, & Wingfield, 2007; olderadults,butthecorrelationwasnonsignificant
Peelle & Wingfield, 2005), although it should be in their sample of younger adults. We therefore
noted that there are differences between age tested adaptation to a novel accent in a group of
groups. For instance, older adults were found to older adults only, in order to gain more insight
adapt to time-compressed speech at a rate and into individual perceptual plasticity for effective
magnitude comparable to younger adults, when communication.
equated for starting accuracy with younger adults Inrecentyears,therehasbeenagrowingunder-
(Peelle & Wingfield, 2005). However, unlike standing of the role of cognitive factors in the
1564
THEQUARTERLYJOURNALOFEXPERIMENTALPSYCHOLOGY,2012,65(8)
ACCENTADAPTATIONINOLDERADULTS
declineoftheabilitytocomprehendspeechinchal- that memory measures relate to word learning in
lenging listening situation situations (Akeroyd, either first or second language acquisition (Atkins
2008; Pichora-Fuller, & Singh, 2006). Akeroyd’s & Baddeley, 1998; Baddeley, Gathercole, &
literature survey (2008) shows that measures of Papagno, 1998; Gathercole, Hitch, Service, &
working memory (reading span in particular) were Martin, 1997; Gupta & MacWhinney, 1997;
most often found to relate to speech comprehen- Papagno, Valentine, & Baddeley, 1991; Papagno
sion ability in challenging conditions. Adank and & Vallar, 1992; Service, 1992; Service &
Janse (2010) investigated the mechanisms under- Kohonen, 1995). Gupta (2003) found relations
lyingthelearningprocessbyrelatingolderlisteners’ between memory and word learning and argued
adaptation to and comprehension of accented that sequence memory is directly involved in tem-
2 speech to their hearing acuity and to measures of porarily maintaining the serial order of the
1
0 cognitive function. They found that individual sequence of sublexical units constituting the novel
2
st hearingacuityandameasureofexecutivefunction word form and in connecting the phonological
u
g (the Trail-Making Test, Reitan, 1958, whichtests and semantic representations (and cf. Vaughan,
u
A attention-switching control) predicted how well Storzbach, & Furukawa, 2006). Further, there is
8
2 olderadultscouldunderstandsentencesinanunfa- morerecentevidencefromthevisualprocessinglit-
37 miliar accent. Nevertheless, they did not find a erature that working memory predicts learning
0:
0 reliable predictor for perceptual adaptation to the (Kennedyetal.,2009).Kennedyetal.usedafrag-
at accented speech. They tested the relationship mentedpicturesidentificationtaskwhereobservers
]
k between perceptual learning (i.e., an increase in viewlinedrawingsofcommonobjectsindescend-
sti
ui the amount of noise participants could tolerate in ing order of fragmentation. With more practice,
ng order to arrive at a fixed identification accuracy observers are able to identify objects more quickly
oli level) and hearing acuity (average pure-tone or at a greater level of fragmentation. Working
h
yc threshold in the better ear), mental flexibility (as memory measures (both verbal and nonverbal)
s
P measured in the Trail Making Test), and proces- were found to predict learning in this fragmented
ur sing speed (as indexed by performance in the pictures identification task, via their effect on a
F
ut Digit-Symbol Substitution Test, which is part of fluid intelligence measure. Efficient working
stit the Wechsler Adult Intelligence Test (2004). memory may facilitate discarding false hypotheses
In Inthepresentstudy,wefirstaimedtodetermine that preclude correct identification of degraded
k
c whetherwecouldidentifypredictorsaccountingfor stimuli (cf. e.g., Engle, Kane, & Tuholski, 1999;
n
a
Pl individual differences in accent adaptation ability, Kane, Conway, Hambrick, & Engle, 2007, on
x using a task that allowed for more fine-grained the relation between working memory and inhibi-
a
M temporal analysis of the adaptation process and torycontrol).Inlisteningtoaforeignaccent,listen-
y [ that allowed participants to listen to the speech ersalsohavetolearn toreject “false friends”:word
b
d signal in quiet conditions. Note that Adank and forms that immediately map onto a standard-
e
d Janse (2010) used a task in which improvement Dutch word, but actually mean something else.
a
o
nl was assessed every 15 sentences, and stimuli were Wealsoinvestigatedwhetherhearinglossanda
w
presented in background noise. Also, we now linguistic measure (vocabulary knowledge) were
o
D specifically included memory measures to investi- predictiveofadaptationtothenovelaccent.Poorer
gate their relation with adaptation: Both auditory hearing(dueeithertohearinglossortopooreraudi-
verbal short-term memory and working memory toryprocessing)maycompromiseperceptuallearn-
measures were included. Being able to keep the ing, as a poorer input signal provides a more
accented sentence in memory may be helpful to fragmented picture of the novel accent, which
learn in what (systematic) ways the accented should make it more difficult to adapt to. Note,
speech deviates from the standard pronunciation however, that Gordon-Salant, Yeni-Komshian,
and to use these regularities in processing sub- andFitzgibbons(2010) found no effect ofhearing
sequentsentences.Therearenumerousindications thresholds on adaptation to Spanish-accented
1565
THEQUARTERLYJOURNALOFEXPERIMENTALPSYCHOLOGY,2012,65(8)
JANSEANDADANK
speech.Linguisticabilitymayverywellbeageneral speech (such as lip opening; Sumby & Pollack,
predictor of speech processing performance, but 1954). This would be less important in the case
may also relate to perceptual adaptation in our of listening to acoustically intact speech in quiet
study, given the linguistic nature of ourdistortion. conditions. Nevertheless, audiovisual information
A number of studies have shown that language- may still help in understanding the speech
relatedvariables,suchasvocabularyandgrammati- because the visible speech provides information
calknowledge,arerobustpredictorsofproficiencyin about speech segments that is redundant as well
a second language (Sparks, Patton, Ganschow, & as complementary to the auditory signal (Jesse &
Humbach, 2009; Van Gelderen et al., 2003). Massaro, 2010) and thus facilitates perception.
Thus,linguisticability,orindividuallanguageapti- Furthermore, audiovisual, rather than auditory-
2 tude(Carroll&Sapon,1959/2000),operationalized only, presentation of the speech may help at a
1
0 inourstudyasvocabularyknowledge,wasexpected moregenerallevel:Thetime-lockedco-modulation
2
st tobepredictiveofnovelaccentadaptation. of speech movements (including movements sig-
gu Second, we evaluated the influence of presen- nalling prominence, such as head nodding and
u
A tation modality of the speech material on adap- movement of eyebrows) may help the listener to
8
2 tation, to address the issue of ecologically valid keep attending the speaker and thus to adapt to
37 communication settings. Typically, both auditory the speaker’sunfamiliarwayof talking.
0:
0 andvisualinformationisavailablewhenconversing Inthepresentstudy,wepresentedtwogroupsof
at with others in everyday settings. Successful com- olderlistenerswithaccentedsentencesinauditory-
k] prehension of speech in everyday settings requires only(A) and auditory–visual(AV) modalities. We
sti
ui the integration of both auditory and visual infor- tested accent processing speed and monitored
g
n mation. The auditory signal may be less available adaptationinbothgroupsusingaspeededsemantic
oli as humans age, due to age-related hearing loss verification test (i.e., judging whether a statement
h
yc that particularly affects sensitivity to high-fre- such as “rats have teeth” is true or false). We
Ps quency sounds. This could partly be compensated related each individual’s general performance, as
r
u for by the availability of visual information in well as individual perceptual adaptation, to their
F
ut audiovisual presentation (Grant & Walden, 1996; auditory, cognitiveand linguisticabilities.
stit MacLeod & Summerfield, 1987; Walden, Iftheavailablevisualinformationimprovespro-
n
I Prosek, Montgomery, Scherr, & Jones, 1977), cessingandadaptation,thentheAVgroupshould
k
c thoughoneshouldnotethatageingalsonegatively showfasterandmoreeffectiveprocessingthanthe
n
a
Pl affectsvisualprocessingandthuslip-readingability Agroupandshouldalsoshowafasterrateofadap-
x (Cienkowski & Carney, 2002; Sommers, Tye- tation. Furthermore, we investigated whether the
a
M Murray, & Spehar, 2005; Tye-Murray, Sommers, audiovisual modality was more beneficial for some
[
y & Spehar, 2007). participants than others: The addition of visual
b
d However, it is unclear whether older adults can information might be more beneficial for those
e
d make use of available visual information about the with poorer hearing and/or for those with poorer
a
nlo speaker when adapting to the speaker’s accent. A selective-attention abilities.
w
recent study of younger listeners found that adap-
o
D
tation to noise-vocoded words was improved
Method
when visual information was present compared
with when only auditory information was present Participants
(Kawaseetal.,2009).Notethatvisualinformation Twogroupsofolderadults(N=66)volunteeredto
maybeparticularlybeneficialinthecaseofacousti- participateintheexperiment.Theyhadresponded
cally degraded speech, because it provides infor- toaninformationletterwithacallforparticipants.
mation that is complementary to the degraded Theyreceivedthisinformationlettereitherbecause
acoustics: Information about place of articulation they had participated in a language experiment at
of consonants may be particularly salient in visual the Max Planck Institute for Psycholinguistics
1566
THEQUARTERLYJOURNALOFEXPERIMENTALPSYCHOLOGY,2012,65(8)
ACCENTADAPTATIONINOLDERADULTS
before, or because they had signed up for partici- limited to avoid too much repetition of content
pation after reading a short advertisement in a words.Thesameheldforthe80sentencesselected
local newspaper (for the Nijmegen area). The forpresentation inthe novel accent.
adults were randomly allocated either to the audi- The novel accent, as used previously (Adank,
tory-only group (N=33, of which 14 were male, Hagoort, & Bekkering, 2010; Adank & Janse,
average age=74 years, SD=6.0 years, range= 2010), was created by instructing the speaker to
65–89 years), or to the auditory–visual group read sentences with an adapted orthography. The
(N=33, of which 11 were male, average age=73 orthography was systematically altered to achieve
years, SD=4.6 years, range=64–82 years). All the following changes in all 15 Dutch vowels as
were native monolingual speakers of Dutch, with listed in Table 1. All sentences are listed in the
2 no history of oral or written language impairment Appendix.Onlyvowelsbearingprimaryorsecond-
1
0 or of a neurological or psychiatric disease (as self- arystresswereincludedintheorthographyconver-
2
st reported on a questionnaire). All participants gave sion. An example of a sentence in standard Dutch
u
g written informed consent and were paid for their andaconverted versionaregivenbelow,including
u
A participation (€8per hour). a broad phonetic transcription using the
8
2 Allparticipantswereaskedtobringtheirglasses International Phonetic Alphabet (International
7
3 to the lab, if they had any. In the experimental Phonetic Association, IPA,1999):
0:
at 0 bwoeoatrhf,otrhweyatccohuinldg tthheencodmecpidueterwshcirceheng,laasnsdesfotor “Rattenhebbentanden”(StandardDutch)/rɑtəhɛbətɑndə/
k] doing any paper-and-pencil task or for filling out After conversion: “Raten heben taanden”
uisti thequestionnaire.Nofurthertestwasadministered /raːtəheːbətaːndə/
g
holin of their vision or visual processing speed. Ofirusrt,retaosoanvsoifdorauscinongfaounnodvelbaectwceenetnwsepreeatkwerofoanldd:
yc Stimulus material accent, thus making sure that the listeners
s
P Thestimulussetconsistedof98sentencesrecorded
r
u in standard Dutch and in an unfamiliar (novel)
F
ut accent. These 98 sentences were selected from a Table1. Vowelconversionsforobtainingthenovelaccent
stit largersetof100sentencepairs(thusfrom200sen-
n
I tences).Sentencepairsweretworelatedsentences: Orthography Phonetic(IPA)
k
nc one being a true statement (e.g., Bevers bouwen a→aa /α/→/a:/
Pla dammen in de rivier; English: “Beavers build dams aa→a /a:/→/α/
x in the river”), and the other false (e.g., Bevers e→ee /ε/→/e:/
Ma groeien in een moestuin; English: “Beavers grow in ee→e /e:/→/ε/
by [ a vegetable patch”). Half of the 98 sentences were iie→→iei //ii//→→//II//
ed true statements, and the other half were false. o→oo /ɔ/→/o:/
ad Importantly, 18 sentences were selected for oo→o /o:/→/ɔ/
nlo presentation in standard Dutch only (half of them uu→u /y:/→/y/
w true,halfofthemfalse):Thesewereusedaspractice u→uu /y/→/y:/
Do oe→u /ε/→/e:/
trials to familiarize participants with the task of
eu→u /ø/→/y/
sentence verification. The other 80 sentences were au→oe /ɔu/→/u/
presented in the novel accent in the test phase. In ei→e /εί/→/e:/
other words, there was no overlap in sentence ui→uu /œy/→/y/
content between the standard-Dutch sentences
Note:Theleftcolumnshowsthealterationoftheorthography
presentedduringthepracticepartandtheaccented
from standard Dutch, and the right column shows the
sentences presented in the test phase. Within the
intended change in pronunciation of the vowel in broad
18 sentences selected as standard-Dutch practice phonetic transcription, using the International Phonetic
trials, the number of true–false sentence pairs was Alphabet(IPA;InternationalPhoneticAssociation,1999).
1567
THEQUARTERLYJOURNALOFEXPERIMENTALPSYCHOLOGY,2012,65(8)
JANSEANDADANK
adaptedtotheaccentandnot(also)tothespeaker’s speaker’smeanspeechrateinthenovel-accentcon-
voice. Second, we wanted to be sure listeners were dition was 4.5syllables per second (SD=0.8).
all equally unfamiliar with the accented speech The video recordings were made with a Canon
(Adank et al., 2009; Floccia, Goslin, Girard, & HV30 camera, and audio recordings were made
Konopczynski,2006). using a Sony ECM-MS907 microphone attached
Note that the perception of vowels is not as to the video camera. The recordings were edited
affected by (age-related) high-frequency hearing using Adobe Premiere CS4. The recordings were
loss as consonant perception would be (but cf. saved as individual files.
Vongpaisal & Pichora-Fuller, 2007, on age-
related decline in periodicity coding affecting Auditory, cognitive, and linguistic backgroundtests
2 vowel perception). Therefore, the way in which Hearing thresholds. Hearing thresholds were
1
0 the novel accent deviates from standard Dutch is assessed with a portable Maico ST 25 screening
2
st relatively salient to the older participant sample of audiometer in a sound-attenuating booth. Only
gu the present study. air-conduction thresholds were established.1 Five
u
A Full-colour-motionvideorecordingsweremade participants had hearing aids in one or two ears,
8
2 in a quiet room of a 37-year-old female native which they were asked not to wear during the
37 speaker of Dutch. Recordings were made of two experiment (nor during the audiometry test
00: versions of the sentences in the Appendix: a stan- obviously). Given the high-frequency hearing loss
at dard-Dutch version and a novel-accent version. associated with ageing, a pure-tone average
]
k Thespeakersatinfrontofadarkbluebackground threshold was computed over 1, 2, and 4 kHz
uisti whilst reciting the sentences and was filmed (insteadofover0.5,1,and2kHz)foreachpartici-
ng from her upper torso up: The speaker faced the pant’sbetter ear.This pure-toneaverage threshold
oli camera so that the front of her face was clearly (PTA) was 27.2 dB HL (SD=11.8): The higher
ych visible. the participants’ PTA, the poorer their hearing
s
P The speaker was instructed to keep a neutral acuity. Pure-tone average threshold in the better
ur facial expression with respect to the truth value of ear did not differ significantly between the audi-
F
ut the sentences. Four repetitions of each stimulus tory-only and audiovisual participant groups, t
stit were recorded, and the authors(both trained pho- (64)=1.2, p..1.
n
I neticians) made sure to select the version that
k
c sounded most novel-accent consistent and to not Cognitive measures
n
Pla select a version with a facial expression or head Attention-switchingcontrol.Participantsweretested
x movement (such as nodding) that might be ontheTrail-MakingTest(TMT;Reitan,1958),a
a
M biasing towards either truth value. All 80 selected paper-and-pencil task, as a measure of executive
[
y accented sentences were inspected for accent con- function, or more particularly, attention-switching
b
d sistency (i.e., did the speaker pronounce the sen- control. Attention-switching control, as a form of
de tences in the Appendix as instructed?). There cognitiveflexibility,wasrelatedtogeneralperform-
a
o
nl were only 3 sentences containing one inconsistent anceontheaccentedmaterialsinourpreviousstudy
w vowel(i.e.,vowelsbeingpronouncedasinthestan- (Adank & Janse, 2010). Mean time to complete
o
D dard-Dutch word). The speaker was thus highly Part A of the TMT, in which participants have
consistent in her production of the accent (if we to connect 25 digits (in ascending order), was
take each sentence to contain minimally three 54.6 s (SD=16.7). Mean time to complete Part
stressed vowels, 99%, i.e., 237 out of 240, of the B, in which participants have to connect letters
vowels were accent consistent). Also, collapsed and digits (alternating between the two dimen-
over the 80 selected novel-accent sentences, the sions: 1–A–2–B–3–C, etc.), was 104.0 s (SD=
1Notethattheremaybesuprathresholdaspectsofhearing(suchastemporalprocessing,cf.Gordon-Salant,Frisina,Popper,&
Fay,2010)thatimpactonspeechprocessingability.Thosewerenotmeasuredinthisstudy.
1568
THEQUARTERLYJOURNALOFEXPERIMENTALPSYCHOLOGY,2012,65(8)
ACCENTADAPTATIONINOLDERADULTS
35.5). We took ratio scores of the two subparts incongruent condition (86%, SD=23.3), while
(TMT–B/TMT–A), rather than the difference accuracy in the congruent and neutral conditions
score, as a measure of attention-switching control was97% (SD=8.2).
(Arbuthnott & Frank, 2000) to take general Mean response times (collapsed only over
slowing into account (Verhaeghen & De correct responses) in the three conditions were
Meersman, 1998). Mean ratio was 1.95 (SD= 612ms(fromvisualpresentationonset)inthecon-
0.48): The higher the ratio, the poorer a partici- gruentcondition(SD=195),758msintheincon-
pant’s attention-switching control. Mean score on gruent condition (SD=283), and 600 ms (SD=
this attention-switching control measure did not 192)intheneutralcondition.Arepeatedmeasures
differ between the auditory-only and audiovisual analysisofvariance(ANOVA)ofeachparticipant’s
2 participant groups, t(64)=1.0, p..1. mean response time (RT) in all three conditions
01 showed a significant effect of condition on RTs,
2
st Selective attention. In this computerized variant of F(2, 64)=51.1, p,.001. Pairwise comparisons
gu the classic flanker task (Eriksen & Eriksen, (Bonferroni-corrected for multiple comparisons)
u
A 1974), participants responded to visual stimuli by showedthattheincongruentconditionwassignifi-
28 clicking either the “z” or the “/” key on the key- cantly slower than the neutral and congruent con-
37 board. A row of five white symbols is shown (in ditions (p,.001) and that the difference between
0:
0 Arial 80 font size), against a black background. theneutralandcongruentconditionwasmarginally
at Themiddlesymbolintherow(thetarget)isaleft- significant (p=.064). Individual performance on
]
uistik wisaflradn-koerdriognhtewitharedr-spidoeinbtiyntgwaorrcoownhgreuaedn.tTohreintacrogne-t tflhainsketraskintwerafserednecteermefinfeecdt: bEyacchomppaurttiincigpantht’es
g
n gruent arrows (same or opposite direction), or by mean logRT in the neutral condition was sub-
oli neutral lines (e.g., for a.target:.....as tracted from each participant’s mean logRT in the
h
yc thecongruentcondition;,,.,,astheincon- incongruent condition and was then divided by
Ps gruentcondition;and --.--asthe neutral con- the mean logRT in the neutral condition. Mean
ur dition).Thetaskoftheparticipantwastoindicate flanker effect was 0.04 (SD=0.02): The higher a
F
ut the direction of the central (middle) target symbol participant’sflankereffect,thepoorertheirselective
stit by pressing the “z” key for leftward pointing and attention. Mean score on this measure did not
In the“/”keyforrightwardpointing,andtomaximize differ between the auditory-only and audiovisual
ck speedandaccuracy.Eachtrialstartedwithabeep(a participant groups (t,1,ns).
n
a
Pl 400-Hzpuretone,presentedat50dBSPL)anda
x fixation cross that remained on the screen for 250 Auditory short-term memory. An auditory nonword
a
M ms. Following this fixation cross, the symbol repetitiontaskwasusedasanindexofverbal/pho-
[
y string was presented for 1,500 ms. After these nological short-term memory (Botting & Conti-
b
d 1,500ms,thestringwasremoved,andparticipants Ramsden, 2001; Gathercole & Baddeley, 1996;
e
d could nolonger respond.Intertrialtime was1,000 Thorn & Gathercole, 1999). Others have referred
a
o
nl ms. There were six different stimuli (two pointing to this task as indexing phonological storage (e.g.,
w
directions for the target, times three different Gathercole, 2006), phonological buffer capacity
o
D flanker conditions). These six different stimuli (Bates et al., 2011), or phonological working
were each presented 12 times in the test part memory (Gathercole, Willis, Baddeley, & Emslie,
(order of trial presentation was randomized for 1994; McGettigan et al., 2011). The task has
each participant) to make 72 trials. Before the test beenwidelyusedinresearchondevelopmentaldys-
part started, 6 practice trials were presented (i.e., lexia(e.g.,Ramus&Szenkovits,2008)andspecific
the six different stimuli). languageimpairment(Botting&Conti-Ramsden,
Mean accuracy of the responses (pooled over 2001).
participants) was 93% correct (SD=12). The task consisted of the presentation of 50
Accuracy was lowest and most variable in the nonwords, all of which were phonotactically
1569
THEQUARTERLYJOURNALOFEXPERIMENTALPSYCHOLOGY,2012,65(8)
JANSEANDADANK
legal in Dutch (de Jong & van der Leij, 1999). 2), the participant was prompted to recall the
The speaker was a professional speech therapist digitsinthereverseorder(e.g.,263).Thepartici-
and female native speaker of Dutch who spoke pant was first presented with two 3-digit trials to
at a consistently slow and clear speaking rate. become familiarized with the task. They would
The nonword items were presented over head- then be tested on 2- up to 8-digit sequences (two
phones at a fixed mean intensity level of 80 dB trials for each sequence length, making up 14
SPL. Participants were seated in a sound-attenu- trials in total). Individual performance on this
ating booth. Each nonword was presented only task was determined by computing the proportion
once, after which participants were asked to of correctly recalled digit sequences (out of 14 test
repeat the nonword. Intertrial time was three trials): The higher the proportion, the better
2 seconds. Nonwords of different syllable lengths working memory. Mean proportion correct in this
01 (two to five syllables long) were presented inter- task was .36 (SD=.13): The higher the score,
2
st mixed, but the order in which they were pre- the better working memory this participant has.
u
g sented was kept constant for all participants. Mean working memory score did not differ
u
A Responses were recorded to allow for offline between the auditory-only and audiovisual partici-
28 scoring. Transcription and scoring was done by pant groups,t(64)=1.4, p..1.
7
3 a native speaker of Dutch. If all syllables of a par-
0:
0 ticular item were reproduced correctly, a score of
at 1 was obtained. If not all syllables were repeated Linguistic measure
k] correctly, a proportion correct was computed Vocabularytest.Thevocabularytestwasareceptive
sti multiple choice test. Participants were asked to fill
ui (number of correctly repeated syllables divided
g in a document in a text-editing program on the
n by total syllable number for that target
oli nonword). Maximum score for the entire task computer (Courier font size 15 was used). The
h
yc was thus in principle 50 (for 50 correctly repeated test was based on a selection of items from
Ps nonword items). Mean score for this task was Hazenberg and Hulstijn (1996), a test originally
ur 31.0 (SD=6.7). Higher scores reflected better developed for speakers of Dutch as a second
F
ut auditory verbal short-term memory.2 Mean language. Target words were presented in a
nstit scores on this task of the auditory-only and nfoeruteraaclhcatarrrigeert)s—enftoenrceexa(mdifpfleer,enfotrctahrreietrarsgeenttewnocreds
I audiovisual participant groups did not differ, t
ck (64),1, ns. mentaliteit (“mentality”), the carrier phrase would
an be: Wat een vreemde mentaliteit! (“What a strange
x Pl mentality!”). Participants had five alternatives to
Ma Workingmemory.Adigitspantask(withbackward choose from, the last one always being: “I really
y [ recall,asubpartoftheWechslerAdultIntelligence don’t know”. New items were added to make the
d b Test, 2004) was used to measure individual test suitable for native speakers. They were con-
de working memory capacity. In the computerized structed according to the same principles as those
a
o variant of this task used here, a series of digits employed by Hazenberg and Hulstijn: Words
wnl would flash up in the centre of the computer should not be too domain-specific, and it should
o
D screen. Each digit was presented during one bepossibletousethewordinasimplecarriersen-
secondandwithonesecondinbetweenconsecutive tence. Care was taken not to introduce any sys-
digits. Digits were presented in a large white font tematicity in the length of the alternatives and the
(Arial, font size 100) against a black background. way meanings were described. The test consisted
After presentation of the digit sequence (e.g., 3 6 of 60 items. Individual score was defined as
2Sincesuprathresholdauditoryprocessingabilities(suchastemporalandfrequencyresolution)werenotassessedinthecurrent
study, we do not know to what extent such auditory processing abilities might be confounded with our auditory short-term
memorymeasure.Thestrengthofthecorrelationbetweenauditoryshort-termmemoryperformanceandaveragehearingthreshold
isaddressedbelow.
1570
THEQUARTERLYJOURNALOFEXPERIMENTALPSYCHOLOGY,2012,65(8)
ACCENTADAPTATIONINOLDERADULTS
Table2. Intercorrelationsbetweenpredictors
Age Hearingloss Switching Selectiveattention AuditorySTM WM
Age
Hearingloss .41***
Attentionswitching(Switching) .11 –.01
Selectiveattention –.03 .10 .35**
AuditorySTM –.32** –.65*** –.15 –.16
Workingmemory –.11 –.30* –.18 .11 .30*
Vocabularyknowledge .11 –.03 –.32** –.03 .27* .35**
2 Note:Pearsoncorrelationcoefficients.STM=short-termmemory.WM=workingmemory.
1 *p,.05.**p,.01.***p,.001.
0
2
st
u
g proportionofcorrectitems(outof60).Meanscore task, and the auditory verbal short-term memory
u
A was.87(SD=.09):Thehigherthescore,thebetter task(in this order).
28 a participant’s vocabulary knowledge. Mean voca- Alllistenersweretestedindividuallyinasound-
7
3 bulary scores of the two participant groups (audi- attenuating booth and received written and oral
00: tory-only and audiovisual group) did not differ, t instructions. Responses were made using a button
at (64),1, ns. box.Participantswereinstructedtousetheirdomi-
k] nant hand (index finger) for “true” responses (the
nguisti Intercorrelations between predictors gfirnegeenr-lfaobrel“lefadlseb”urtetsopno)nsaensd(retdh-eliarbelolethderbutitnodne)x.
oli Table 2 lists all intercorrelations between predic- The stimuli were presented binaurally, over
ch tors. The highest correlation was found between Sennheiser HD 280-13 headphones, at a fixed
y
s hearing loss and auditory short-term memory:
P output level of 80 dB SPL for all participants.
r Those with more hearing loss had poorer
u Stimulus presentation and response recording
F nonword repetition (r=–.65, p,.001). The two
ut were performed using Presentation software
nstit aPtatretnictiiopnantsmweaistuhrepsoowreerreseleccotrirveelataetdtentaisonwaelsllo: (Neurobehavioral Systems, Albany, CA).
I Response times were measured from the onset of
k generally had poorer attention-switching control
c the video. Participants in the AV group were pre-
n (r=.35, p,.01). Further, as expected, the two
Pla memory measures were correlated: Participants sented with the video and the audio signal, and
x the participants in the A group only heard the
a with better auditory short-term memory had
M audio signal, while the computer monitor showed
[ better working memory skills (r=.30, p,.05).
y a black screen. Participants in the AV group were
b Vocabulary knowledge was related to three of our
d instructed to watch the screen while listening to
e four cognitive measures.
d the sentences.
a
o
nl Eachtrialproceededasfollows.First,thestimu-
w Experimental procedure lus sentence was presented. Second, the program
o
D Thepoolofolderadultsparticipatedintwostudies waited for three seconds before playing the next
(theotheronenotreportedhere)andinthebattery stimulus, allowing the participant to respond. If
of background tests described above. To minimize the participant did not respond within three
fatigue, we spread testing over two sessions, each seconds, the trial was recorded as no response.
of which consisted of a speech perception study Participants were asked to respond as quickly and
and a number of background tests. The sessions accurately as possible and they were told that they
were approximately one month apart. The study didnothavetowaituntilthesentencewasfinished,
reported here was done in the first session, as well allowing for negative RTs, as RT was calculated
asthepure-toneaudiometry,theselective-attention from the offset ofthe sound file.
1571
THEQUARTERLYJOURNALOFEXPERIMENTALPSYCHOLOGY,2012,65(8)