1. Processing of Syntactic Movement
in Children with Cochlear Implants
Zara Waldman
CUNY - Second Examination
July 2013
2. Content Areas for Today’s Talk
Language
Development in
Children with
Cochlear
Implants
Typical and
Atypical Syntax
Development
The relationship
between
working memory
and syntax
processing
And eventually… how these three areas fit
together and rationale for further exploration of these
topics.
3. Cochlear Implants
• CI are internal, biomedical devices that enable
people with damaged cochleae to experience the
sensation of hearing
• Usually implanted in children at 6-12 months
– In the past, implantation occurred after 12 mos.
– Period of auditory deprivation/altered access has
profound effect on language development
• So, how DO children with CI develop language?
– Sensitive periods
• Experience-independent
• Experience-dependent
4. Language Outcomes in CI
Experience-Independent
• Earlier CI = better standardized language
scores (Connor, Heiber, Arts & Zwolan, 2000; Connor, Craig,
Raudenbush, Heavner & Zwolan, 2006; Nicholas & Geers, 2006, 2007)
– Older CI recipients widen the gap between typical,
hearing (NH) children while younger CI recipients
shorten it (Niparko, Tobey, Thal, Eisenberg, Wang, Quittner &
Fink, 2010; Hay-McCutcheon, Kirk, Henning, Gao & Qi, 2008)
• The “magic” of 1 year of CI use
– CI results in burst of vocabulary, speech (Tomblin,
Barker, Spencer, Zhang & Gantz, 2005)
• Because this “magic” appears to wane by
4-5 years of use (Connor et al., 2006)
• And few CI users reach typical levels of
language, even by age 5 (Svirsky, Teoh & Neuberger,
2004)
• Morphology still not quite right (Nicholas & Geers,
2007)
• When within-group differences were explored,
Duchesne, Sutton & Bergeron (2009) identified
vocabulary and syntax as two major deficits for
children with CI
• And great variability remains
• These studies are framed as an effect of
chronological development, not
necessarily accounting for variations in
language experience
Experience-Dependent
• CI children 4 mos. behind acquiring first
words, 3 mos. behind acquiring word
combinations (Nott, Cowan, Brown & Wigglesworth,
2009)
• Slower reaction time to word
recognition (Grieco-Calub, Saffran & Litovsky, 2009)
– This might be influenced by perception
• Still part of experience!
• Poorer fast mappers, especially after 14
mos. (Houston, Stewart, Moberly, Hollich & Miyamoto, 2012;
Tomblin, Barker & Hubbs, 2007)
– But better if you could hear before your
CI! (Houston et al., 2012)
– And if you had more experience using your
CI! (Tomblin et al., 2007)
• Also found in language scores!
(Nicholas & Geers, 2006; Niparko et al., 2010)
• Reading scores agree with Duchesne et
al. (2009)(Geers et al., 2008)
5. CI Children and Development of
Complex Language (Syntax)
• Despite typical overall language scores from CI subjects…
– Grammar was so disordered further comparison to typical
children wasn’t possible (Caselli, Rinaldi, Varuzza, Giuliani & Burdo, 2012)
– Only 36% of children implanted under 4 years have age-
appropriate syntax (Nikolopoulous, Dyar, Archbold & O’Donoghue, 2004)
• Majority of difficulties centered in verb cases (Szagun, 2000),
article omission (Caselli et al., 2012, Szagun, 2004), and syntactic movement
(Geren, 2010)
– Suggests that the issue starts as a perceptual deficit, but
failure to correct syntax after multiple exposures might
indicate a more involved cause
– But, syntax is better with more maternal input and better
pre-CI hearing (Szagun, 2004b)
6. What is Syntactic Movement?
• One part of the sentence is relocated
– This disrupts subject-verb-object structure
• A trace exists in the gap left behind by the relocated
part
– When the listener reaches the gap, they must reactivate the
filler noun to complete the sentence (Love & Swinney, 1996; Roberts, Marinis, Felser &
Clahsen, 2007; Marinis & Van Der Lely, 2007; Love, 2007)
– This filler is stored in working memory
Subject Question:
[Which dog]1t1 licked Jessie?
Object Question:
[Which girl]1 did the dog lick t1?
7. How can we confirm this is the
underlying process?
• Priming studies
– Evidence of trace reactivation in children as young as
4 years old (Love, 2007)
• Eye-tracking
– Difference in eye movements between wh-questions
and yes/no questions (Sussman & Sedivy, 2003)
– Longer reading times for object relatives and
questions (Traxler, Morris & Seeley, 2002)
– Significant proportion of looking time during the gap
focused on the gap-filler (Dickey, Choy & Thompson, 2007)
• Even true in aphasic participants!
8. Processing Syntactic Movement
• Typical children develop their ability to comprehend
and produce syntactic movement between ages of 2-3
years (Stromswold, 1995)
– But children as young as 15 mos. Have shown recognition of
pictures in response to wh-questions (Seidl, Hollich & Juscyzk, 2003)
– VERY FEW (>2%) children show evidence of NO syntactic
movement abilities (deVilliers, Roeper, Bland-Stewart & Pearson, 2008)
• And there is a hierarchy to syntactic movement
– Object relatives and questions are more difficult than subject
relatives and questions (Friedmann, Belletti & Rizzi, 2008)
• And syntactic comprehension is sensitive to the
preferences/interest level of the child (Hurewitz, Brown-Schmidt, Thorpe, Gleitman &
Trueswell, 2000)
– As well as semantic plausibility (Friedmann et al., 2008)
9. What do we know about
syntactic movement abilities in
children with CI?
Is this deficit unique?
10. Language Disorder & Syntax
Cochlear Implants
• Comprehension and production
of relative clauses and wh-
questions significantly worse
than their NH peers (Friedmann & Sztermann,
2005, 2010, 2011)
– Only 61% of object relatives
were correct and 70% of those
used resumptive pronouns
• Deaf subjects were aided by
explicit noun phrases (Berent, 1996)
– Possibly suggestive of a
“semantic strategy”
• Poor accuracy in repetition of
relatives and wh-questions (Friedmann
& Sztermann, 2011)
Specific Language Impairment
(SLI)
• Comprehension and production of
relative clauses and wh-questions
signifcantly worse than their typical
peers (Deevy & Leonard, 2004; Marinis & van der Lely, 2007;
Friedmann & Novogrodsky, 2006, 2011)
• Trace reactivation shown with
priming only at verb offset (Marinis & van der
Lely, 2007)
– Might be “semantic strategy” (sort
of likely)
– Might be delayed trace (very likely)
• They can process the gap but not
always its relationship to semantic
components in the sentence (Friedmann &
Novogrodsky, 2006)
– As suggested by paraphrasing
11. But Why?
So far, the research community wants us to
blame it on “thematic role confusion”
• But, this might imply total inability to
comprehend/produce syntactic movement
• But, this might impede the usage of resumptive
pronouns
• But, even aphasics can process
syntactic movement
12. When two populations have
“thematic role confusion,” we
should consider a shared cause.
Perhaps, their parsing is in tact,
but something else is not quite
right.
13. What’s something we SHOULD
consider?
• Processing a sentence requires:
Parsing: ability to break down
components of sentences, understand
those components, and reconfigure them
• How we use knowledge of syntax/grammar
[Functional] Working Memory (WM):
ability to store, process, and control
attention while parsing
• How we remember, activate, and integrate all the
sentence components together
14. Working Memory can be defined
by capacity…
• Idea that there is a finite amount of WM
resources in the brain, amount is unique to
every individual (Just & Carpenter, 1992)
– Typical adults with lower reading span capacity
had poorer sentence comprehension (King & Just, 1991; Just & Carpenter,
1992)
• And typical children! (Gaulin & Campbell, 1994)
• Much more of these WM resources are used
to comprehend sentences with syntactic
movement (King & Just, 1991; Just & Carpenter, 1992)
– Even evidenced in eye tracking and ERP (Traxler, Williams, Blozis &
Morris, 2005)
15. Or by interference
• The simultaneous activation of related
concepts while trying to retrieve one specific
concept (Gordon, Hendrick & Levine, 2002)
– Adequate inhibition of this excess activation
enables the brain to retrieve the specific concept
• When subjects had to retain related items in
memory, comprehension of cleft sentences
and relative clauses was poorer (Gordon, Hendrick & Levine, 2002; Fedorenko,
Gibson & Rohde, 2005)
– Poorer comprehension also occurs from syntactic
interference, when verb plausibility is
manipulated (Van Dyke & McElree, 2006)
16. So, is there evidence of working memory
issues in populations with language/syntax
impairment?
And again, is this working memory issue
unique to one population?
17. Working Memory & CI
• CI children have poorer backward/forward
WISC digit spans than their NH peers (Cleary, Pisoni & Geers, 1999;
Cleary, Pisoni & Kirk, 2000; Pisoni & Cleary, 2001; Burkholder & Pisoni, 2003)
• Implicit sequencing is also impaired, as shown
on SIMON task (Pisoni, Conway, Kronenberger, Henning & Anaya, 2008)
– Link to how grammar is learned?
• Better digit span also linked to faster
articulation of sentences (Pisoni & Cleary, 2001; Tobey, Geers, Morchower, Perrin, Skellett,
Brenner & Torretta, 2000)
– Better WM = faster retrieval?
18. Working Memory & SLI
• Reading span capacity and sentence
comprehension linked in SLI population (Leonard, Ellis-Weismer, Francis,
Tomblin & Kail, 2007; Montgomery & Evans, 2009)
– Nonword repetition abilities also linked to sentence
comprehension (Montgomery, 1995)
– Performance is poorer than age matched peers, but
not peers matched for syntax (Montgomery & Evans, 2009)
• Abilities affected more by sentence complexity
than sentence length (Marton & Schwartz, 2003)
• Are not helped by slowed presentation (Montgomery, 2004)
• Do not appear to rehearse or have a particular
pattern/strategy to help them retain information
(Marton & Schwartz, 2003)
19. Do these working memory deficits play
a role in syntactic processing?
It would appear so, but it more
specific research is needed to affirm
this hypothesis.
20. Rationale for Research
• Two groups, but ONE problem!
• There is evidence that “thematic role reversal” is not
the whole story
• We need to confirm if another factor, working
memory, could be responsible
– Or could be the shared, underlying issue between
populations with syntactic deficits
– For children with CI, need more precise research in working
memory (!)
• The opportunity to study both processes (syntax and
working memory) in concert with each other, at the
very moment they are happening
21. Questions & Discussion
With my many thanks to:
Richard Schwartz, PhD
Klara Marton, PhD
Derek Houston, PhD
for their knowledge and guidance in
helping me prepare for this examination.
Editor's Notes
So, as I’m sure you already know, I’m Zara Waldman and my area of focus (and hopefully future research) is in how children with cochlear implants develop and process complex syntax and grammar.
Today’s talk will focus on three major areas that are needed to explore children with implants and syntactic processing skills. First, I will discuss a little background on cochlear implants and the impact the device has on language development. Second, I will talk about syntax development in typical and atypical populations, and the syntax of children with cochlear implants. And lastly, but certainly not least, I will talk about working memory, the role it plays in syntactic processing, and the affect it has on the syntax processing abilities of individuals. And of course, I will find some way to illustrate how these three areas go together and support my chosen area of research.
So here’s the easy part, let’s define what a cochlear implant is and how it permits language acquisition in children. The CI is a biomedical device that is surgically inserted into the non-functional cochleae of a person with severe to profound hearing loss which is coupled with a microphone and sound processor worn on the ear, much like a hearing aid. This sound processor and mic pick up sound from the outside world and transmit it to the internal part, allowing it to send information about the sound on to the brain. It is important to note that sounds are not the same as you and I hear it with natural hearing, rather they are a modified version of sounds. This altered input can affect the child’s perception of sound. Typically, children are identified as having hearing loss at birth, which enables implantation as early as 12 months. However, in the past, children received these implants much later (sometimes as late as 7 years of age) and some children today still do receive implants past 12 months for a variety of reasons. Whether they receive them at 1 year or later, that does not change the fact that most children with CI go through a period of auditory deprivation, where they have no access to sound at all. This period of deprivation is capable of having a profound effect on language development, as it restricts early access to phonology, and subsequently, semantics and syntax. To put it figuratively, the “ground” lost during this period is very difficult for this population to regain, but with early implantation becoming the norm, it has gotten better. But the difficulties children with CI face in language development are often framed around two approaches to sensitive periods. A sensitive period refers to a time where the child is able to acquire certain skills with ease. There is debate in the CI literature as to whether the evidence points to difficulties in language because of experience-independent sensitive periods, or a sensitive period that is based around a chronological window, OR experience-dependent sensitive periods, which is a sensitive period that depends on prior knowledge or skills in order for another ability to be acquired. Children with CI likely feature problems with their language development that is reflective of both, which I will demonstrate on the next slide.
On the left side we have a whole list of research that has supports a chronologically-based window of sensitivity to language. An incredible amount of studies in CI have established that if a child receives their CI earlier, their language abilities are better. This is not true of older CI recipients, as they tend feature wider (and continuously widening) gaps in language scores, as compared to normal hearing, typical children. Older CI users also have a lack of reciprocracy between earlier test scores and later test scores (Hay-McCutcheon, Kirk, Henning, Gao & Qi, 2008). In the first year of CI use, language development is especially promising, as children seem to pick up speed in vocabulary development and language concepts. However, this speed begins to wane at around age 4-5, when language starts to become more complex, and many of these children lack chronologically-appropriate language even by age five. Out of 12 children evaluated, Svirsky, Teoh & Neuberger (2004) found that only 3 met their age criteria on the Reynell and MacArthur assessments. So what was their shortcoming? Well, age 4-5 means the start of more complex language acquisition, namely grammar and syntax. Nicholas & Geers (2007) noted that out of a majority of spontaneous language samples, morphology still fell way below average, especially in the area of grammatical morphemes. When Duchesne, Sutton & Bergeron (2009) profiled 27 Quebecois children they found that while many had typical language otherwise, many had delays in either vocabulary, syntax, or both. It’s interesting that vocabulary and syntax are the two major deficits, because these two areas are somewhat experience-dependent by nature. These areas rely on prior knowledge (prior experience) to build onto. And there is also growing evidence that prior experience plays a role in the language development of CI children, mostly from the area of word learning and mapping. We know that many children with CI are behind in acquiring first words and word combinations, which is likely a factor of how they recognize and map novel words. They are slower to recognize and map new words than their hearing peers, but children with CI who had better residual hearing and more experience using their CI were able to map better. In a reconsideration of their data, Nicholas & Geers (2006) and Niparko et al. (2010) found that residual hearing also impacted language scores. These studies are a change from the earlier studies that focused on timing but neglected variations in the population that could influence CI success. It just shows that while a majority of the CI literature revolves around chronological constraints, it has overlooked possible individual variations that could influence data and results. And an interesting side note comes from the reading research, with reading itself being an obviously experience-dependent skill. Geers et al. (2008) did a small reading assessment and the reader profiles she assembled mirrored the profiles Duchesne et al. (2009) found. Many CI readers had impoverished vocabulary, poor grammar, or both.
Talking about experience-dependent deficits, I introduced both vocabulary and syntax. The last slide explained vocabulary, so here’s the syntax portion. Problems with syntax in the deaf and hard of hearing have been noted as far back as 1974 in a study by Quigley, Wilbur & Montanelli. They can persist even when other language skills and vocabulary are unimpaired. READ SLIDE. Syntactic deficits were centered around perceptual things, such as articles and verb cases, but they also had difficulty with syntactic movement. This suggests the issue might start as a perceptual problem, but then, how does one address the fact that children with CI fail to fix their syntactic mistakes even after repeated exposure? This might be related to a more deep-seated cause, and perhaps the basis for this other deficit in complex, syntactic movement. And interestingly, as if just to attest to the experience-dependent nature of it, better residual hearing and maternal input have an effect on the syntax of children with CI.
Here is just a quick overview of syntactic movement and what makes it so complex in comparison to a simple subject-verb-object sentence.
So by priming children at the gap of the sentence, we can assess if they have reactivated the filler phrase. And evidence of priming has been noted in children as young as four. Using eye-tracking while reading the question, Sussman & sedivy demonstrated that processing syntactic movement is a unique type of linguistic processing. The longer reading time during object relatives and questions in Traxler et al. (2002) confirms the difficulty of object relatives and questions over subject relatives and questions. And lastly, when the question or relative clause is pictured and then read, typical populations spend a significant proportion of time looking at the filler during the gap of the sentence. While not significant, this proportional trend is also true in aphasic participants!
Well let’s look at how this part of syntax develops in typical, normal hearing children. READ SLIDE.
So we know that children with CI have impaired processing for syntactic movement, but how exactly is this manifested? And, is this a deficit that is solely found in children with CI, or is there another population who also struggles with syntactic movement?
Well let’s take a look at children with CI and for comparison’s sake, another population with noted syntactic processing deficits in movement, children with SLI. So that answers our latter question that no, children with CI are not unique in this deficit. As I mentioned before, problems with syntax have been noted in deaf and hard of hearing people since Quigley’s study in 1974. Their comprehension and production of syntactic movement is significantly worse than normal hearing peers, but there are signs that they’ve tried to employ syntactically and semantically appropriate compensatory strategies to help them cope with their deficit in syntactic movement. For one, they rely on resumptive pronouns. Use of resumptive pronouns occurs grammatically in Hebrew, the language this study was conducted in, and essentially is a “reduplication” of the filler using a pronoun…so it would be something along the lines of “who did the grandma kiss him?” This suggests that while children with CI struggle to make sense of movement, they have an idea that there is a filler and that filler needs to be filled by something. Unfortunately, Friedmann & Sztermann did not report on how many of these resumptive pronouns were accurate, but if many of them were, it would suggest they understand a lot more about movement than they let on. Namely, this might suggest that they understand the syntactic construction, but not the components involved. Secondly, Berent in 1996 found that deaf people who were highly proficient in spoken English tended to have an easier time making grammatical judgements about questions and relative clauses when the noun phrase was explicit, suggesting that semantic interpretation may also be used to help comprehend syntactic movement. And their poor accuracy in repetition (80% of subject questions repeated accurately, 55% of object questions repeated accurately) again points to an underlying deficit. Switching gears, let’s see what children with SLI do with syntactic movement. Again, just like the kids with CI, their comprehension and production of movement is significantly worse than their typical peers, but there is evidence of understanding the syntactic construction. A priming study by Marinis & van der Lely (2007) only demonstrated trace reactivation at the offset of verbs in children with SLI, which they claimed to be evidence of a semantic strategy, attachment of the verb to its semantic counterpart. But what’s more likely in their large age range of subjects (ages 10-17), where reaction times may be more volatile, is that this is evidence of delayed reactivation of the trace. In a reading assessment by Friedmann & Novogrodsky (2006), children with SLI were asked to read relative clauses and wh-questions. Their phrasing implied that they understood where the gap was located and parsed accordingly. However, when asked to paraphrase, they often switched or repeated subjects and/or objects. This seems to defeat the idea of a “semantic strategy” as it appears they may not know or remember enough about the semantic parts of the sentence to integrate them into their understanding of the syntactic structure. Delayed reactivation and this issue with paraphrasing, might imply that children with SLI, like children with CI, appear to comprehend the syntactic structure but struggle with the components.
Okay, so now we’re left with two populations with similar deficits. Why? So far, the only suggested theory in the literature is one of “thematic role confusion,” or the idea that children with CI and children with SLI do not understand who does what in the sentence, who acts as the subject and who as the object. If they had truly had a major deficit in role assignment, they might show NO evidence of understanding syntactic movement and they might show more problems in simple sentences and other more complex, yet canonical sentences. BUT, that’s not true. If they had a deficit in this, it might cause them to be unable to use resumptive pronouns to fill a gap. BUT, that appears to be one of their coping strategies. And my hmmm moment…even aphasics who are severely language impaired show evidence of gap-filling and thematic role assignment, albeit delayed. So if their linguistic functions are in tact, maybe we need to start considering another reason for their difficulty.
And the same is true for our kiddos with CI and SLI. Just because they are two separate populations does not necessarily mean the are exclusive. With a similar pattern of difficulties with syntactic movement, I think it behooves the research community to start looking for a shared root of these problems.
So if, as we considered on the previous two slides, that parsing is not solely to blame, then we need to consider another part of sentence processing that may be impaired across the two populations. One possible cause is an underlying cognitive processing deficit, namely in working memory. READ SLIDE. To illuminate this concept, I have my broken plate and glue. Parsing is our ability to break down sentence components, semantics, syntax, pragmatic cues, etc. Parsing gives us all those pieces of the plate. Working memory is our glue. It lets us store and process those sentence components and integrate, or glue them all back together in a manner that enables us to comprehend the sentence’s meaning.
So how exactly does working memory factor into sentence processing? How can it affect our ability to comprehend complex syntax? Two factors can influence working memory, one is capacity and the other is interference. Since the early 1990s, the work of King & Just and Just & Carpenter have established this theory that there are a finite amount of working memory resources in everyone’s brains and this amount is unique to every individual. They have demonstrated variations in this “capacity” (using reading span activities, where you have to repeat sentences while remembering the last word or nonword of each sentence you hear) even in typical populations. When these typical adults were asked to comprehend syntactically simple and complex sentences, King & Just (1991) found that the adults with lower reading span scores had poorer sentence comprehension scores. Gaulin & Campbell (1994) were able to replicate this finding in typical children as well. And these correlations were especially strong in sentences with syntactic movement. This difficulty in comprehension of sentences with syntactic movement was also discovered in eye tracking and event related potentials, and again, also correlated with reading span scores. These studies actually establish a correlation between measures of working memory and measures of language comprehension. They expose a preexisting limitation on working memory resources that is further exhausted as sentence complexity increases.
The second way to look at working memory is through individuals’ ability to control interference, or READ THE SLIDE. While the research in this area is relatively newer than the research on capacity, we know that interference interferes with complex sentence processing because when typical subjects were asked to remember related and unrelated semantic items while comprehending complex sentences, their comprehension was poorer with related items. Much like the idea of priming, it is likely that these activated related items conflicted with the reactivation of fillers for cleft and relative clauses. Gordon et al. (2002) and Fedorenko et al. (2005) establish that semantic content can interfere with parsing ability, but it is key to note that syntactic interference is also possible. Van Dyke & McElree (2006) found that manipulating the plausibility of the verb in relation to the subject also affected their typical subjects’ ability to parse adequately. This research suggests that when interference is not inhibited or suppressed, it can impede language comprehension. So, these two interpretations illustrate the effect of working memory on complex sentence processing. Both limitations in capacity size and ability to inhibit interference influence parsing of complex syntax, even in normal populations. It is most likely that these things work together to exhaust working memory.
So if capacity and interference have such a great effect on syntax comprehension in typical populations, what about special populations? Have we investigated if working memory deficits exists in populations with syntax impairment, such as the ones we already discussed, children with CI and children with SLI? And to mirror our questions about syntax, since the syntax deficits we saw in the CI population were not solely unique to them, what about working memory? Are certain working memory issues unique to just one population?
So let’s look at our first population of interest, children with CI. They’ve demonstrated deficits in backward and forward digit spans and memory for sequences and their ability to learn novel sequences in comparison to their age-matched, normal hearing peers. And these digit span scores have also been linked to speed of articulation, with children with CI who have better digit span able to articulate faster. Unfortunately, the authors of these studies have, in my opinion, stretched a little far in extending the implications of their results. Digit span in all of those studies was interpreted as a measure of working memory, however there is not much functional about remembering sets of numbers. There is no manipulation or application to meaning involved. The same stands for the SIMON task, while it does use sequences and sequence memory, again, there is not much manipulated so it is hard to differentiate this task from a short-term memory task (and the authors themselves have used it as a short-term memory task, a working memory task, AND a sequencing task). And the interpretation of the correlation between faster articulation and digit span as suggestive of faster retrieval has the potential to be intriguing, but there was little to no report on the types of sentences used and this is likely more indicative of phonological memory skills. Although I could go on and on about how these methods are conflicting, they are all we have in terms of research in the area of working memory and CI. Poor scores on digit recall and ability to remember sequences might indicate that their working memory capacity is limited. While these tasks are not ideal, they could be hinting that an underlying problem exists and warrants further investigation.
To answer our second question, are these difficulties in working memory unique to the CI population? The answer is, of course, no. Their buddies with SLI also demonstrate similar deficits in working memory. They have poorer reading span than their typical peers, and this has been correlated to their ability to parse complex sentences, as has nonword repetition (nonword repetition is a measure of phonological WM). Interestingly, when compared to peers matched for syntax, their performance on a reading span task is equivalent. This suggests that their issues with syntax are NOT a function of their parsing ability or grammar knowledge, but rather something else…working memory. These results indicate that children with SLI possibly have fewer working memory resources, a.k.a. limited capacity. And these struggles with sentence parsing are not alleviated by shortened sentences or slowed presentation, rather it is an inherent difficulty with increased complexity. Providing more time to process or less to process does not appear to “free up” more of that working memory capacity. Additionally, when Marton & Schwartz (2003) analyzed their errors in sentence recall, they found no evidence of recency priming (easily remembering the last items) or rehearsal (remembering the first items because they were rehearsed). Their haphazard recall of items in somewhat random order suggests that perhaps children with SLI have not developed a strategy for storing items that will need to be retrieved later on. This makes it difficult for them to either a) cope with lessened working memory capacity or b) ward off interference from other similar items. Working memory difficulties in children with SLI shows the connection between poor working memory and syntax comprehension for children with a language deficit. It has been documented enough that working memory deficits are considered somewhat hallmark of children with SLI and are, at least partially acknowledged for their problems with syntax.
So here’s the final question, do these populations struggle with complex syntax because of their deficits in working memory? Well, it appears to be a solid possibility, but is certainly overlooked by the majority of researchers in syntax processing. Thus, a study that would seek to compare working memory and complex syntax processing in a real-time, functional way, would be really helpful in determining the true cause of these populations’ similar problems with syntax and with working memory.
