Chapter 3 Familiar word recognition
3.1 Lexical processing dynamics
Mature listeners recognize spoken words by continuously evaluating incoming speech for possible word matches. The first part of a word activates multiple candidate words in parallel. These candidates compete as more of the speech signal enters the system, and the best-fitting word is the favored interpretation. For example, the onset “bee” might activate phonologically compatible candidates like bee, beam, beetle, beak, beaker, beginning, and so on, but an additional “m” would narrow the candidates to just beam. Semantic relationships also influence lexical processing, and cascading phonological-semantic effects—for instance, where castle activates the phonologically similar candy which in turn activates the semantically related sweet—have been demonstrated (Marslen-Wilson & Zwitserlood, 1989). Both low-level phonetic cues and high-level grammatical, semantic and pragmatic information can influence this process, but this continuous processing of multiple competing candidates is the essential dynamic underlying word recognition in adults (Magnuson, Mirman, & Myers, 2013).
What about young children who know considerably fewer words? Eyetracking studies with toddlers have suggested a developmental continuity between toddlers and adult listeners. Children recognize words incrementally (Swingley, Pinto, & Fernald, 1999), match truncated words to their intended referents (Fernald, Swingley, & Pinto, 2001), and use information from neighboring words in a sentence to facilitate word recognition. This information can be high-level grammatical or semantic cues. Lew-Williams and Fernald (2007) found that Spanish-acquiring preschoolers can use grammatical gender on determiners (el or la) to anticipate the word named in a two-object word recognition task. Borovsky, Elman, and Fernald (2012) showed that children can use semantic information from an agent and a verb (e.g., the dog chased) to anticipate a plausible noun (the cat). The information can also be low-level phonetic variation: We found that toddlers look earlier to a named image when the coarticulatory formant cues on word the predicted the noun of the sentence, compared to tokens with neutral coarticulation (Mahr, McMillan, Saffran, Ellis Weismer, & Edwards, 2015).
There is some evidence for lexical competition where children are sensitive to phonological and semantic similarities among words. Ellis Weismer, Haebig, Edwards, Saffran, and Venker (2016) showed that toddlers (14–29 months old) look less reliably to a named image when the onscreen competitor was a semantically related word or perceptually similar image. Huang and Snedeker (2011) presented evidence of cascading semantic-phonological activation in five-year-olds such that for a target word like log, the children looked more to an indirect phonological competitor like key (competing through its activation of lock) than they looked to an unrelated image like carrot.
Priming studies also reveal that children are sensitive to phonological similarities among words. Mani and Plunkett (2010) demonstrated cross-modal phonological priming effects in 18-month-olds. In this study, a picture of prime word (e.g., cat or teeth) was presented in silence; then two images (cup and shoe) were presented, one of which was named (cup). Children on average looked more to the target word (cup) when it was primed by an image of a phonological neighbor (cat), and the children performed at chance when the prime was not related to the named word. Mani, Durrant, and Floccia (2012) found a similar result for cascading phonological-semantic priming with 24-month-olds: Children looked more to a target (e.g., shoe) compared to a distractor (door) when primed by an image (clock, assumed to activate sock which primed shoe).1
Altvater-Mackensen and Mani (2013) demonstrated phonological-semantic priming even when the prime is a mispronunciation. German-learning two-year-olds heard a prime word, and 200 ms later two images appeared onscreen (a cow, Kuh and a fork, Gabel), one of which was labeled (“Kuh”). The prime word was a semantically related word (“Schaf”, sheep), an onset-mispronunciation of the related word (“Faf” or “Taf”), or an unrelated prime (“Buch”, book). Children looked to the target about equally well in the two prime conditions: approximately .62 proportion looks to target with a normal prime versus approximately .60 for a mispronounced prime. In contrast, they looked less to the target in the unrelated prime condition (approximately .55). Thus, the children in this study showed cascading activation where the mispronunciation activates the mispronounced word which in term activates a semantically related word.
Chow, Aimola Davies, and Plunkett (2017) performed a very similar study to the one I present in the next chapters. They used the Visual World Paradigm with English-learning 24- and 30-month-olds. Children saw a 2 × 2 grid of images which included a phonological (cohort) competitor and a semantic competitor, and they heard a prompt to view one of the images (e.g., Look at the bee). On filler trials, the target word and an unrelated image appear onscreen alongside the competitors. On test trials, the display had two unrelated images (sandwich, dress), a phonological competitor (bus), and a semantic competitor (cat) and children were prompted to look at an offscreen, unpictured target (Look at the bee). They found a temporary early advantage for the phonological competitor, so that the probability of looking to the phonological competitor was greater than the other competitors. This early advantage was followed by a late, more stable advantage for the semantic competitor. Moreover, they found that increased receptive vocabulary predicted more looks to the phonological competitor and fewer looks to the semantic competitor. (The looks to the semantic competitor were decreased because of the early advantage of the phonological competitor.) Their results support a kind of cascading activation in which phonological information comes online before semantic information.
The above studies involved young children of different ages tested under different procedures, sometimes in different dialects and languages. Averaging these results together, so to speak, the studies suggest that early word recognition demonstrates some hallmarks of adult behavior: Continuous processing of words, integration of information from different levels of representation, and the influence of similar, unspoken words on the recognition of a word. Nevertheless, we only have a fragmented view of how familiar word recognition develops within children.
One open question is how lexical competition develops in young listeners. For example, how and when do phonologically or semantically similar words exert their influence on word recognition? Chow et al. (2017) provide a promising first step, in which two-year-olds looked to the phonological and semantic relatives of a named word. (Yet I am skeptical of any word recognition study where a target word is absent and absent for many trials.) As a guiding hypothesis, we can think of word learning as a gradual process where familiarity with a word moves from shallow receptive knowledge to deeper expressive knowledge. In adult listeners, words compete and they inhibit one another, so that a word is truly “learned” (integrated into the lexicon) when it can influence the processing of other words (a line of reasoning reviewed by Kapnoula, Packard, Gupta, & McMurray, 2015). Increasing sensitivity to similar sounding or similar meaning words over time would reveal that children more thoroughly learn familiar words with age.
3.2 Individual differences in word recognition
We have a rough understanding of the development of word recognition, and these gaps in knowledge matter because young children differ in their word recognition abilities. These differences are usually measured using accuracy (a probability of recognizing a word) or efficiency (a reaction time or some measure of how quickly accuracy changes over time). These differences are consequential too, as word recognition differences correlate with other language measures concurrently and prospectively.
Many studies highlight the predictive power of word recognition ability. Marchman and Fernald (2008) found that vocabulary size and lexical processing efficiency at age 2 jointly predicted working memory scores and expressive language scores at age 8. Fernald and Marchman (2012) found that late talkers who looked more quickly to a named word at 18 months showed larger gains in vocabulary by 30 months compared to late talkers who looked more slowly at 18 months. Weisleder and Fernald (2013) found that lexical processing and language input at 19 months predicted vocabulary size at 25 months and that lexical processing mediated the effect of language input—the basic idea being that rich language input builds up word recognition ability which in turn supports word learning. Lany (2017) found a direct link between lexical processing and word learning: 18-month-olds and 30-month-olds who were faster at recognizing familiar words were also more accurate at recognizing novel words in a word-learning task. Thus, children who are better at recognizing words learn more words over time and perform better at word-learning tasks.
Word recognition performance predicts future language outcomes, so we conclude that individual differences in word recognition are important. But we do not know how word recognition develops within children, so we have no context for evaluating these individual differences. Are these differences in lexical processing persistent over development? Is word recognition a skill where most children catch up and converge on a mature range of performance by a certain age?
3.3 The current study
In the previous two sections, I outlined two gaps in knowledge. The first is that we do not have a clear understanding of how the mechanisms underlying word recognition change in early childhood. We know that children show plenty of adult-like features of word recognition, but each of these findings is an isolated fact. What we need is a coherent set of facts that shows how specific features of word recognition change with age. The second gap is that although we know that individual differences in word recognition are predictive of later outcomes, we do not have a developmental picture of these individual differences.
In this study, I tackle these two lines of research: The development of lexical competition effects and individual differences in familiar word recognition. I report the results of a longitudinal study of word recognition in preschoolers at age 3, age 4, and age 5. The study is described in detail in Chapter 4. Briefly stated, this experiment tested word recognition by presenting prompts like “find the horse” and recording children’s looks to an array of four images. The array of images included the target, a phonological competitor, a semantic competitor, and an unrelated image. In Chapter 5, I analyze the development patterns of familiar word recognition (looks to the target) and how individual differences change over time. I hypothesized that children would show stable individual differences over time, but the range and magnitude of these differences would get smaller as children grew older. I also examine which word recognition measures correlate with future vocabulary size to test how word recognition behavior predicts later outcomes. In Chapter 6, I study how the phonological and semantic competitors influence word recognition, and I test the prediction that children will become more sensitive to competitors as they grow older. Finally, in Chapter 7, I link these two lines of research together and describe both sets of results in terms of lexical processing dynamics, and Chapter 8 reviews the results of my pre-analysis research hypotheses.
References
Marslen-Wilson, W. D., & Zwitserlood, P. (1989). Accessing spoken words: The importance of word onsets. Journal of Experimental Psychology: Human Perception and Performance, 15(3), 576–585. doi:10.1037/0096-1523.15.3.576
Magnuson, J. S., Mirman, D., & Myers, E. (2013). Spoken word recognition. In D. Reisberg (Ed.), The Oxford handbook of cognitive psychology. Oxford University Press. doi:10.1093/oxfordhb/9780195376746.013.0027
Swingley, D., Pinto, J. P., & Fernald, A. (1999). Continuous processing in word recognition at 24 months. Cognition, 71(2), 73–108. doi:10.1016/S0010-0277(99)00021-9
Fernald, A., Swingley, D., & Pinto, J. P. (2001). When half a word is enough: Infants can recognize spoken words using partial phonetic information. Child Development, 72(4), 1003–15. doi:10.1111/1467-8624.00331
Lew-Williams, C., & Fernald, A. (2007). Young children learning Spanish make rapid use of grammatical gender in spoken word recognition. Psychological Science, 18(3), 193–8. doi:10.1111/j.1467-9280.2007.01871.x
Borovsky, A., Elman, J. L., & Fernald, A. (2012). Knowing a lot for one’s age: Vocabulary skill and not age is associated with anticipatory incremental sentence interpretation in children and adults. Journal of Experimental Child Psychology, 112(4), 417–436. doi:10.1016/j.jecp.2012.01.005
Mahr, T., McMillan, B. T. M., Saffran, J. R., Ellis Weismer, S., & Edwards, J. (2015). Anticipatory coarticulation facilitates word recognition in toddlers. Cognition, 142, 345–350. doi:10.1016/j.cognition.2015.05.009
Ellis Weismer, S., Haebig, E., Edwards, J. R., Saffran, J. R., & Venker, C. E. (2016). Lexical processing in toddlers with ASD: Does weak central coherence play a role? Journal of Autism and Developmental Disorders, 46(12), 3755–3769. doi:10.1007/s10803-016-2926-y
Huang, Y. T., & Snedeker, J. (2011). Cascading activation across levels of representation in children’s lexical processing. Journal of Child Language, 38(03), 644–661. doi:10.1017/S0305000910000206
Mani, N., & Plunkett, K. (2010). In the infant’s mind’s ear: Evidence for implicit naming in 18-month-olds. Psychological Science, 21(7), 908–913. doi:10.1177/0956797610373371
Mani, N., Durrant, S., & Floccia, C. (2012). Activation of phonological and semantic codes in toddlers. Journal of Memory and Language, 66(4), 612–622. doi:10.1016/j.jml.2012.03.003
Altvater-Mackensen, N., & Mani, N. (2013). The impact of mispronunciations on toddler word recognition: Evidence for cascaded activation of semantically related words from mispronunciations of familiar words. Infancy, 18(6), 1030–1052. doi:10.1111/infa.12022
Chow, J., Aimola Davies, A. M., & Plunkett, K. (2017). Spoken-word recognition in 2-year-olds: The tug of war between phonological and semantic activation. Journal of Memory and Language, 93, 104–134. doi:10.1016/j.jml.2016.08.004
Kapnoula, E. C., Packard, S., Gupta, P., & McMurray, B. (2015). Immediate lexical integration of novel word forms. Cognition, 134, 85–99. doi:10.1016/j.cognition.2014.09.007
Marchman, V. A., & Fernald, A. (2008). Speed of word recognition and vocabulary knowledge in infancy predict cognitive and language outcomes in later childhood. Developmental Science, 11(3), F9–16. doi:10.1111/j.1467-7687.2008.00671.x
Fernald, A., & Marchman, V. A. (2012). Individual differences in lexical processing at 18 months predict vocabulary growth in typically developing and late-talking toddlers. Child Development, 83(1), 203–222. doi:10.1111/j.1467-8624.2011.01692.x
Weisleder, A., & Fernald, A. (2013). Talking to children matters: Early language experience strengthens processing and builds vocabulary. Psychological Science, 24(11), 2143–52. doi:10.1177/0956797613488145
Lany, J. (2017). Lexical-processing efficiency leverages novel word learning in infants and toddlers. Developmental Science. doi:10.1111/desc.12569
Arias-Trejo and Plunkett (2009) is commonly cited as evidence of semantic priming effects. Toddlers heard sentences like “I saw a cat… dog”. During the word dog, two images (dog and door) are presented. The idea is that cat should prime looks to its semantic neighbor dog. The unnatural stimulus order (a sentence followed by an isolated single word) and a condition effect where 18-month-olds outperformed 21-month-olds make me skeptical that semantic priming is the most plausible explanation of those results.↩