Exploring cognitive and linguistic predictors of speech recognition improvement in adults after cochlear implant activation

Postlingually deafened adults who are provided with a cochlear implant (CI)
face a great challenge: they have to learn to interpret a novel and rather
artificial acoustic signal. Intriguingly, within a couple of weeks only,
dramatic increases in patients* speech understanding performance have been
reported. However, the benefit a patient experiences from a cochlear implant
varies considerably between individuals. For example, individual improvement in
spoken word recognition after six months of CI activation was found to range
between 2% and 66% among CI recipients (Heydebrand et al., 2007). This raises
the question which factors predict speech perception outcome after
implantation. Knowledge about such factors is of clinical relevance as it may
pave the way for individualized training and rehabilitation programs.

Over the last decades, mainly impairment-related factors have been studied as
possible factors accounting for this inter-individual variability. Although age
at onset of hearing loss (e.g., Kaplan et al., 2003), duration of hearing loss
(e.g., Chan et al., 2007; Hamzavi et al., 2003; Holden et al., 2013; Oh et al.,
2003), and to lesser extents etiology (e.g., Blamey et al., 1996; Geier et al.,
1999) and age at implantation (e.g., Holden et al., 2013) have been shown to
explain adults' variability in speech perception skills after implantation,
these factors were found to explain only 10 to 21% of the variance in
CI-outcome (cf., Blamey et al., 1996; Blamey et al., 2013). This suggests that
a large amount of variability between individual CI-recipients remains
unexplained.

Individual cognitive abilities may account for additional variance in
CI-recipients. Particularly in children with CIs, researchers have argued that
individual differences in cognitive abilities that relate to perception,
attention, learning or memory processes may explain the large variability in
CI-outcome (Ingvalson & Wong, 2013; NIH Consens Statement, 1995; Pisoni et al.,
1999). Indeed, research found that measures of behavioral inhibition (Horn et
al., 2005), novel word learning ability (Davidson et al., 2014), short-term
memory (Cleary et al., 2000; Edwards & Anderson, 2014; Pisoni et al., 1999;
Willstedt-Svensson et al., 2004) and working memory (Cleary et al., 2000;
Pisoni & Davis, 2003; Willstedt-Svensson et al., 2004) were associated with
language outcome in children with CIs. However, children are typically
implanted at such a young age that it is not possible to measure cognitive
abilities prior to implantation. Moreover, children's cognitive and language
development is highly intertwined and the relationship between cognitive and
linguistic skills may be reciprocal. On the one hand, cognitive abilities may
facilitate successful adaptation to a CI and hence language development. On the
other hand, access to auditory and linguistic input after CI-activation may
stimulate children's cognitive development by increasing children's attentional
skills (Khan et al., 2005).

Testing adult postlingually deafened CI-recipients, who acquired linguistic and
cognitive skills normally, allows us to approach the influence of specific
cognitive abilities (as assessed prior to implantation) on CI adaptation
success more directly. Although the number of postlingually deafened
CI-recipients is rapidly increasing, only few studies have tried to link
cognitive performance prior to surgery to speech perception outcome after CI
activation. Early studies reported working memory (Lyxell et al., 1998) and
fast visual sequence processing (Gantz et al., 1993; Knutson et al., 1991) to
relate to speech perception performance at nine to eighteen months
post-surgery. More recently, a study found patients* verbal learning ability to
be predictive of their word recognition ability six months after CI activation
(Heydebrand et al., 2007). These research findings are promising as they
suggest cognitive abilities to play a role in CI-outcome. However, studies so
far have only linked cognitive abilities to medium-term outcome. As
CI-recipients receive intensive auditory training mainly during the first weeks
after CI activation, knowledge about the influence of cognitive abilities on
the initial adaptation process may be particularly valuable. For example, if
memory abilities are associated with better hearing progress, treatment may be
adapted for participants with poorer memory skills by extending the
rehabilitation program or by providing memory exercises.

Evidence for the potential role of cognitive abilities in initial CI adaptation
success comes from studies on perceptual learning in speech. In these studies,
participants are exposed to unfamiliar speech input such as accented or
noise-vocoded speech (which is a simulation of the auditory signal of a
cochlear implant) and participants* improvement in speech understanding
performance over exposure is investigated. Speech recognition improvement has
been reported to relate to measures of statistical learning ability (Neger et
al., 2014b), vocabulary knowledge (Janse & Adank, 2012; Neger et al., 2014b),
and selective attention (Janse & Adank, 2012), with more improvement over
exposure for those with better statistical learning, better vocabulary, and
better selective attention. As CI recipients also have to adapt to a novel
listening situation, adaptation to the signal of a CI may be considered a
functional concept of perceptual learning. Further evidence backing up this
assumption comes from the finding that, good CI performers show increased
activity in brain areas (i.e., inferior frontal gyrus and angular gyrus)
(Giraud & Lee, 2007) that have been linked to perceptual learning performance
(Eisner et al., 2010) as well as to executive functioning (Colom et al., 2013),
compared to poor CI performers (Lyness et al, 2013).

Importantly, most studies on predictors of speech perception performance in
CI-users used simple consonant-vowel-consonant (CVC) words. However, in
everyday speech, CI recipients are generally confronted with conversational
speech. In contrast to carefully designed audiological speech material,
conversational speech consists of sentences that vary in speech rate and
include hesitations and sloppily articulated speech etc. That is, audiological
testing material is clearly different from conversational speech and, hence,
from the speech material cochlear implant recipients are exposed to during the
adaptation process. Processing of longer stretches of speech may, thus, be more
ecologically valid to measure patients' progress in speech understanding
performance. Therefore, individuals' cognitive abilities and individuals'
subjective CI-benefit will be compared to speech recognition performance in
three conditions of increasing ecological validity: a standard CVC-recognition
task, a sentence repetition task (clear speech) and a word identification task
in conversational speech.

In sum, this study aims to investigate the role of learning ability, linguistic
knowledge, and general cognitive abilities of processing speed, memory and
attention during the initial adaptation phase after CI-activation. We
hypothesize that these individual cognitive abilities can be used to predict
patients' amount and time course of improvement over the first three months of
CI-use. Moreover, we assume that individuals' speech recognition performance
will be associated with their subjectively reported benefit, particularly in
ecologically valid sentences recognition tasks.

The primary objective of the study is to investigate which cognitive abilities
predict amount and time-course of individual improvement in speech
understanding performance within the first three months after cochlear implant
activation. The primary hypothesis is that part of the variance in patients'
improvement in speech understanding performance during the first weeks after
cochlear implant activation is explained by patients' learning ability,
linguistic knowledge, processing speed, memory and attention. Moreover, we aim
to assess the use of ecologically valid speech materials for audiological
testing in CI-users. The hypothesis is that with increasing ecological validity
speech perception measures are more predictive of the subjective benefit
patients experience with a cochlear implant.

The secondary objective of the study is to investigate participants' changes in
articulatory precision (i.e., one measure of verbal ability in the current
study) in relation to their individual changes in speech perception. Prior to
implantation, CI recipients typically produce imprecise sound contrasts in
their speech due to insufficient auditory feedback (Lane et al., 2007). After
CI activation, auditory feedback is restored but the auditory input is rather
artificial and shifted in frequencies. That is, patients have to retune their
speech productions mechanisms based on this altered auditory feedback (Lane et
al., 2007). Links have been demonstrated between discrimination of speech
sounds and articulatory precision in normal-hearing adults (Perkell et al.,
2004a; Perkell et al., 2004b). Therefore, we hypothesize that individual
improvement in general speech perception abilities is related to individual
improvement in articulatory precision after CI activation.

The investigation is designed as an explorative, observational, longitudinal
study and will take place within one single center. Patient inclusion of this
clinical investigation will be performed among patients that already have been
audiologically and otologically evaluated and have been assessed as suitable
candidates for treatment with a cochlear implant. The group includes
approximately 30 patients. All participants will undergo the regular clinical
practice. However, in addition, participants will be tested at five points in
time. Patients will come for testing of individual cognitive and verbal
abilities one to three weeks prior to surgery and will come for check-ups one
week, three weeks, seven weeks and eleven weeks following cochlear implant
activation. These visits will include a clinical evaluation of patients* speech
understanding performance with the cochlear implant. Moreover, a short test of
processing speed will be administered at the beginning of each visit to measure
patients' alertness and general cognitive processing ability per test session.
Additionally, patients' articulatory precision will be tested during the first
and last follow up moment.
All data is to be registered in digital Case Report Forms.

Project duration
The experiment will run from 01-01-2015 to 30-11-2015. During that time span,
postlingually deafened patients, who are provided with a CI at the Hearing &
Implant department of the Radboud UMC in Nijmegen, who meet the criteria for
inclusion in our study, and who confirm their participation in written consent,
can participate in the project.

Session duration
All testing will take place at the Radboud UMC. Evaluation of participants'
cognitive skills will require 2 hours and will be combined with one of the
preoperative visits. Similarly, testing of patients' speech recognition
performance after cochlear implant activation will be combined with four
follow-up sessions at the Radboud UMC to cause minimal inconvenience for the
participants. The first and last follow-up test moments will take 40-55
minutes, as multiple measures of speech understanding performance as well as a
measure of articulatory performance will be administered. The second and third
follow-up test moments will only take 15 minutes, implying a time investment of
4 hours per participant (spread over 4 months).

Tasks per session
To measure individual abilities, we administer a set of cognitive and
linguistic tests prior to surgery. These tests include:

Cognitive and verbal abilities
To measure individual abilities, we administer a set of cognitive and verbal
tests prior to surgery. These tests include:
Learning ability
• Statistical learning: visual artificial grammar learning task (ability to
derive implicit regularities based on frequency of co-occurrences) (Neger et
al., 2014b). As participants have to be unaware of the aim of the task to be
able to measure implicit learning, participants are informed that the task
measures motor speed.
Memory
• Short term memory: Digit Span Forward Test (Daneman & Carpenter, 1980)
• Verbal short term memory: Non-word repetition task (Gathercole et al., 1994)
• Working memory: Digit Span Backward Test (Daneman & Carpenter, 1980)
Attention
• Attention switching/mental flexibility: Trail-Making Test (Salthouse et al.,
2000)
• Selective attention: Flanker test (Richard Ridderinkhof et al., 1999)
Processing speed
• Motor and mental speed: Digit-Symbol-Substitution task (Wechsler, 2004)
• Scanning speed: Letter comparison task (Salthouse & Babcock, 1991)
Verbal ability
• Vocabulary: Vocabulary knowledge test (Andringa et al., 2012)
• Language proficiency: Cloze test (ability to correctly guess which words
might be missing from incomplete texts) (Neger et al., 2014a)
• Articulatory precision: Word reading task (ability to disambiguate sound
categories). Words are embedded in short carrier sentences. Please note that
the speech of the participant will be recorded in this task.
Subjective cochlear implant benefit
To evaluate participants' subjective benefit in hearing, we collect data on
patients' hearing experience prior to surgery and eleven weeks after
implantation. As patients are standardly asked to fill in this questionnaire
prior to implantation, we only administer the questionnaire eleven weeks after
implant activation:
• Nijmegen Cochlear Implant Questionnaire

Speech understanding
Participants perform three different task to measure speech recognition
performance after activation of the cochlear implant. Participants perform all
tasks in their best-aided, bimodal hearing condition Participants perform all
tasks in their best-aided, bimodal hearing condition (i.e., with the hearing
aid(s) and CI they use in their daily life at that moment) to resemble
patients' everyday listening experience:
• Speech audiometry: Percentage of correctly repeated speech sounds of
auditorily presented consonant-vowel-consonant words at 70 dB SPL. This task is
standardly used to measure patients' hearing ability with their CIs and to
adjust CI-settings. It is part of the general follow-up examinations and will,
therefore, impose no additional load on participants.
• Sentence recognition test: Percentage correctly identified keywords in
sentences. Sentences are presented at a fixed intensity of 70 dB SPL and
participants have to repeat what they understand. Sentences come from audiology
testing materials (Versfeld et al., 2000). During each test moment participants
are presented with three lists of 13 sentences (different lists at each test
moment), each sentence containing three to four keywords.
• Word identification in conversational speech: Accuracy and speed with which
participants are able to correctly identify one out of four possible words in
conversational speech (Koch & Janse, 2014). Speech materials are presented at a
fixed intensity of 70 dB SPL.

Following the risk classification guidelines of the NFU, the risk of
participating in this study is negligible. The implant, surgery and
rehabilitation process are identical to the regular clinical practice and will,
therefore, not contribute to any additional risk. No type of complications is
anticipated other than those that might occur when rehabilitating patients with
a cochlear implant outside the study. Thus, there are no medical risks
associated with participation.
Overall, participation in the study requires visits for the patient with the
same frequency, but with longer duration of some of these visits. This imposes
extra load on patients, however, patients will receive financial compensation
for their time investment (i.e., travel reimbursement per visit and an
once-only payment of 40 ¤ after completion of all follow-ups). Additionally, by
participating in the study, patients have the possibility to gain detailed
insight into their speech understanding improvement in communicative settings,
which is one of the key aspects of rehabilitation. We conclude that the risks
of this exploratory study are acceptable when weighted against the gain in
terms of research results in a field that has received little attention but may
be important for long-term patient outcomes.