My research program examines the relationship between memory and language processes in young and older adults, specifically the interactions between memory retrieval and language perception, comprehension, and production. My work is theoretically driven by interactive activation theories of language production, such as Node Structure Theory and the Transmission Deficit Hypothesis (e.g., MacKay, 1987; MacKay & Burke, 1990), which represent words in a hierarchical, connectionist network of multi-level systems, including a semantic system containing the meanings of words, a phonological system representing syllables and speech sounds, and an orthographic system representing the spellings of words. A fundamental assumption of the theory is that language and memory are inextricably related, such that failures in retrieving information from memory can be tied to deficits in the activation levels of a word’s phonology and/or orthography. Using priming techniques to implicitly present phonological and orthographic aspects of words, I have tested the hypothesis that strengthening the connections between words and their phonology/orthography will facilitate word retrieval and reduce retrieval failures. I have focused the majority of my research on two specific retrieval failures, tip-of-the-tongue (TOT) states and spelling errors. In addition to their cognitive importance, these phenomena are especially relevant to the functioning of older adults, who experience increased TOT states and errors in spelling. My long-term goals are twofold: (1) to more fully develop connectionist theories in order to account for fundamental language and memory phenomena, and (2) to understand why certain aspects of language and memory functioning are impaired in old age while others are spared.

The overarching goal of my research on TOT states is to offer insight into the organization of the lexicon and the nature of syntactic and phonological processes involved in retrieving and speaking words, processes that are essential to normal language production. TOT states are naturally occurring retrieval failures that are characterized by a temporary inability to recall a known word. Within the Transmission Deficit Hypothesis, a TOT state occurs when the connections to a word’s phonological nodes are weakened, causing a reduction in the transmission of priming to these nodes, which prevents word retrieval from occurring. Previous research has demonstrated that phonological cues and primes (words sharing phonological aspects with the TOT target word) can reduce TOT experiences as well as increase retrieval of the target word. My research has investigated specific aspects of phonology and their role in resolution of TOT states. Abrams, White, and Eitel (2003) found that only primes containing the target's first syllable facilitated TOT resolution, relative to primes containing the target's first letter, first phoneme, middle syllable, or last syllable. These results indicate that activation of a word’s initial syllable is critical for eliciting retrieval of the entire word, such that the initial syllable must be activated before subsequent syllables become available. These findings are important not only for understanding the mechanism underlying TOT errors, but also for understanding how the syllables of a word are produced in the correct sequence, a fundamental problem of serial order. Abrams and Rodriguez (2005) provided more information about serial order mechanisms, specifically the mechanisms underlying syntax. Phonological primes containing the target’s first syllable matched or differed from the target in part of speech. The results showed that different part-of-speech primes facilitated resolution of TOT states, whereas same part-of-speech primes had no effect, relative to phonologically unrelated control words. This result explains why phonologically related words do not always help word retrieval: Syntactic class activates words that are candidates for production so that phonologically similar words within the same syntactic class compete with each other, whereas phonologically similar words in different syntactic classes facilitate word retrieval.

With respect to older adults, TOT states increase with aging and are therefore especially frustrating for older adults, who cite TOT experiences as one of their most frequent complaints. The Transmission Deficit Hypothesis states that this increase in TOT states is caused by an additional weakening of connections throughout the lexical network that occurs as a result of the normal aging process. My research has addressed whether young and older adults equivalently experience the benefits of phonological priming on resolution of TOT states. White and Abrams (2002) documented an interesting dissociation within older adults: Young-old adults (aged 60-72) showed facilitation from first-syllable primes equivalent to that of college students, whereas old-old adults (aged 73-83) were unable to increase word retrieval after reading first-syllable primes. These results demonstrate significant differences between young-old and old-old adults in TOT resolution such that old-old adults need additional remediation beyond first-syllable phonology to compensate for their weakened phonological connections. Abrams, Trunk, and Merrill (2007) examined the role of syntactic class in young-old and old-old adults' resolution of TOT states. An interesting pattern emerged where old-old adults did not demonstrate priming from different part-of-speech primes but showed inhibition from same part-of-speech primes: Their resolution of TOT states was lower after reading same part-of-speech primes relative to unrelated words, unlike young and young-old adults. These data indicate an increased susceptibility in old-old adults for phonologically related words in the same syntactic class to become competitors, reducing resolution of TOT states. Although there has been considerable TOT research in recent years, there is very little research on the TOT resolution process and even fewer studies that mimic the automatic retrieval processes that help resolve naturally-occurring TOTs. My future TOT research will continue to identify specific phonological and syntactic components that influence resolution of TOT states.

The general purpose of my spelling research is to understand the factors that influence the ability to retrieve and produce correct spellings from long-term memory, such as word frequency, recency of exposure to the correct spelling, spelling ability, and aging. Spelling is a language-based skill that utilizes complex cognitive processes, similar to those involved in reading, so investigating the causes of spelling errors can be used to understand the nature of these declines, relative to other language production processes. Within Node Structure Theory and the Transmission Deficit Hypothesis, errors in retrieving correct spelling from memory are similar to TOT states in that they are caused by weakened connections, but to orthographic nodes instead of phonological nodes. I have conducted several priming studies that have examined whether recent presentation of a word’s orthography facilitates later spelling retrieval. Abrams, Trunk, Hadley, and Margolin (in preparation) found that reading or writing a correctly spelled, difficult-to-spell word increased the ability to later retrieve the correct spelling of that exact word. Using auditorily presented primes, Abrams, Trunk, and White (2008) showed that producing (via typing) the correct spelling of primes with a low-frequency spelling (e.g., the “ege” in sacrilege) increased the spelling retrieval of different words containing that spelling (e.g., privilege). Together, these studies demonstrate that both perception and production of orthographic nodes lead to a relatively long-term change in connection strength that enhances future spelling retrievals, an effect analogous to that observed in phonological priming studies of TOT resolution.

Only a few recent studies have examined older adults’ ability to retrieve correct spelling from memory (Abrams & Stanley, 2004; Abrams, Trunk, Hadley, & Margolin, in preparation; MacKay & Abrams, 1998; MacKay, Abrams, & Pedroza, 1999). Similar to increases in TOT states, the results of these studies consistently demonstrated an age-related decline in the ability to retrieve correct spellings, both for auditorily presented words and for briefly visually presented words, although Margolin and Abrams (2007) found age differences only among poor spellers. This decline in written spelling retrieval often occurs despite an intact ability to recognize words as correctly spelled, suggesting that the age-related impairment in spelling occurs during the orthographic retrieval process. Whether old-old adults show declines in priming of spelling retrieval like they did for priming of TOT resolution is a question for future research. Another future research question is whether presenting misspelled words as primes has a negative impact on spelling retrieval in young and older adults. Seeing a misspelling activates its orthographic nodes, and this recent activation may allow the “incorrect” orthographic nodes to receive more node priming than the correct spelling at the time of spelling retrieval. Older adults generally have weaker orthographic representations by virtue of the aging process, especially old-old adults (e.g., Abrams & Stanley, 2004; MacKay & Abrams, 1998), so the detrimental influence of exposure to misspellings on later spelling retrieval may be exacerbated in older adults, relative to young adults.

On the comprehension side of language, I have started a research program investigating the ability to detect misspellings during reading and the effects of perceiving misspellings on comprehension and memory. Understanding how perceiving misspellings can impact reading comprehension and memory retrieval is important because people frequently encounter misspellings during reading in various everyday contexts (e.g., emails, newspapers, advertisements), so the opportunities for perceiving misspellings are widespread. Theoretically, this research addresses older adults’ simultaneous use of top-down semantic processes and bottom-up visual processes, an issue relevant to Node Structure Theory via its architecture, which allows for the examination of top-down and bottom-up processes, e.g., production versus perception (Taylor & Burke, 2002). I received a R03 grant from the National Institute on Aging to investigate whether perception of misspellings remains stable in old age in naturalistic reading, and if so, whether this stability in processing is differentially affected by a context in which the cognition is personally relevant. Abrams, Farrell, and Margolin (submitted) examined older adults’ ability to detect misspellings in paragraphs about health-related topics when simultaneously attempting to read for recall. The results showed that declines in misspelling detection become more pronounced with increasing age (older adults on their 80s had poorer misspelling detection than older adults in their 60s). However, Abrams, Farrell, and Margolin (in preparation) found that the presence of misspellings did not generally disrupt older adults’ recall of what they had read. In fact, the accuracy of misspelling detection was a strong predictor of recall, showing that readers who accurately detected misspellings had greater recall. One interpretation of these results is that embedding misspellings in meaningful, cohesive paragraphs helps older adults to more effectively manage their limited resources for processing.

Together, my research has both theoretical and practical significance for the fields of cognitive psychology and cognitive aging. Theoretically, my studies will provide a better understanding of the cognitive processes underlying memory retrieval and its effects on our comprehension and production of language. Given the paucity of existing theories that incorporate language, memory, and aging, more precisely defining the relationship between retrieval failures, their relationship to other aspects of language production, and the changes that occur throughout the normal aging process will make a strong contribution to developments in both cognitive and aging theories. I plan to continue testing theories of language production and to use my research to refine these theories by detailing the relations between the phonological, orthographic, and semantic systems. In addition, my research will document whether these language processes decline or remain stable as part of the normal aging process. Practically, identification of the conditions that exacerbate or reduce retrieval failures from memory can help older adults to function more effectively in everyday life by encouraging language use. Experiencing more TOT states and spelling errors may cause older adults to withdraw from language production, both spoken and written, which will increase their cognitive declines by further weakening the connections to phonology and orthography that cause TOT states and spelling errors. I look forward to continuing the research programs described here as well as developing new directions that offer innovation and significance to research on language and memory.