Not all lexical access tasks are created equal: Lexical development between three and five

Abstract

Lexical access tasks are designed to measure efficiency of lexical access, but task demands and methods vary greatly. Many lexical access tasks do not account for confounding factors including competence in other linguistic abilities. In this study, preschoolers were given two lexical access tasks. In the single-category naming (SCN) task, children rapidly named animals. In the multi-category naming (MCN) task, children named one item from each of 10 categories. Children were also tested on vocabulary, syntax, and articulation. First, scores on each task increase with age. Second, the two lexical access scores were only moderately correlated, suggesting that different ‘lexical access’ tasks measure different abilities. Finally, we compared scores on each of the lexical access tasks with scores on the other linguistic tasks, demonstrating that different linguistic abilities are involved in each task at different ages. Results suggest that even very similar tasks capture different aspects of lexical access, and that the SCN task is a purer lexical access test.

Keywords

Animal naming language development language testing lexical access mental lexicon preschool language verbal fluency word retrieval

To understand or produce a sentence, a person must not only know the pronunciation, meaning and syntactic behavior of the words that compose the sentence but must also be able to retrieve these words efficiently and accurately (Swinney, 1979). The ability to retrieve words from one’s lexicon is therefore a key component of human language processing and production, of great import to psycholinguists. Lexical access is of central importance to clinicians as well: impairments in lexical access have been noted in many neuropsychological disorders (Tombaugh, Kozak, & Rees, 1999). The vocabularies and lexical access abilities of adults and school-age children have been well studied (Forster, 1976; Levelt, 2001; Marslen-Wilson, 2001; see Caramazza, 1997, for a discussion). Additionally the vocabularies of young children have been well studied (Dickinson, McCabe, Anastasopoulos, Peisner-Feinberg, & Poe, 2003; Funnel, Hughes, & Woodcock, 2006; Messer & Dockrell, 2006; Weizman & Snow, 2001). However, the lexical access abilities of young children remain comparatively understudied (Kambanaros, Grohmann, & Michaelides, 2013). In this article paper we examine lexical access in typically developing preschool-aged children.

In adults and school-aged children, researchers have explored lexical access through a variety of techniques, including analyses of access failures (e.g., slips of the tongue and tip of the tongue), various types of priming-based tasks (e.g., lexical decision tasks, rapid automatized naming, confrontational naming, modified Stroop task), and category prompting tasks in which people are asked to name members of categories (e.g., naming animals). Although all of these techniques are designed to measure the efficiency of lexical access, there are advantages and disadvantages of each. These are summarized in Table 1. In typical adults and school-age children, category prompting tasks have generally been used to gain insight into category structure. For example, Estes (1974) argued that success on these tasks requires the participant to ‘organize output in terms of clusters of meaningfully related words.’ These tasks have often been used with elementary-school-aged children (e.g., Crowe & Prescott, 2003; Koren, Kofman, & Berger, 2005; Snyder & Munakata, 2010), and occasionally with preschool-aged children (e.g., Lucariello, Kyratzis, & Nelson, 1992), and researchers have used connectionist modeling to search for shifts from thematic to taxonomic organization throughout childhood.

Table 1.

Lexical access techniques for children.

Technique	Advantages	Disadvantages
Slip of tongue	Naturalistic: no task demands	Drawing conclusions about normal access from failure
		Limited datasets
		Ascertainment bias (e.g., only funny ones recorded
		Could reflect incorrect entry rather than access failure
Tip of tongue (TOT)	Naturalistic: no task demands	Drawing conclusions about normal access from failure
		Naturally occurring TOT: can’t manipulate factors
		Elicited TOT: effect of elicitation unknown
		High metalinguistic demands
		Could reflect incomplete entry rather than access failure
Priming tasks	Experimental: can manipulate factors independently	Priming effect could reflect strategies rather than automatic process
		Reading required for some priming tasks
Category prompting	Relatively few task demands	Limited ability to manipulate factors
Category prompting	Suitable for young children & people with impairments	Some metalinguistic demands

Two common forms of category prompting tasks are what we will call single-category naming (SCN) tasks and multi-category naming (MCN) tasks. In an SCN task (Thurstone, 1938), a person is given a single category (e.g., ‘animal,’ or the letter ‘s’) and asked to name as many items as he or she can from that category in a set period of time. In an MCN task (Loftus & Freedman, 1972), a person is given a series of categories (e.g., animals, foods, colors) and asked to name one item that belongs to each category. As soon as the person gives a response to the first category (e.g., an animal), he or she is given another category (e.g., foods).

Although both SCN and MCN tasks are used to measure lexical access ability, differences between the two tasks could entail that they measure slightly different abilities. SCN tasks are a traditional measure of verbal fluency used in standardized intelligence tests which require an ability to retrieve items exhaustively from a single category. MCN tasks are less widely used and require an ability to switch between categories quickly. Developments in performance on both tasks could be affected by metamnemonic index: the degree to which a person is aware of his or her own memory and learning processes (e.g., Benjamin & Bjork, 1998); a conceptual restructuring from thematic to taxonomic category structure (e.g., Lucariello et al., 1992; Mandler, 1983); or development of component linguistic skills.

Understanding the linguistic components necessary for success on a lexical access task is essential for fully appreciating a child’s performance on the task and for remediating poor performance. To investigate the differences between SCN and MCN lexical access tests, we used both of these measures to evaluate the lexical access abilities of a single group of preschool children. Furthermore, we tested these children on receptive vocabulary, receptive syntax, and articulation and investigated the relationships between scores on the lexical access tests and scores on the other linguistic tests. Finally, we separately analyzed three-, four-, and five-year-olds’ scores to determine how lexical access ability, and the linguistic factors that may facilitate it, develop over this age span.

There are at least two ways in which lexical access could develop. The first way is quantitative: Older children might name more animals and items than younger children, whether or not the underlying lexical access process develops during the preschool years. The second way is qualitative: The processes used and the factors involved in lexical access could be different at different ages (e.g., vocabulary size could be more important for lexical access at one age than another). In the following, we report evidence of both quantitative and qualitative development. We discuss our findings for children at each age and suggest implications for choosing an appropriate category prompting task.

Method

Participants

Participants were 275 children drawn from a larger twin study. We recruited participants by advertising on internet sites and newsletters affiliated with mother of twins clubs. Although we attempted to recruit children from diverse socioeconomic backgrounds, over half of the children had an annual family income of US$75,000 or more (indicating middle to high socioeconomic status), and two-thirds of the children had at least one parent with a bachelor’s degree. For all of the children, English was the only language spoken in the home, and over 90% of the parents self-identified as non-Hispanic Caucasian.

For this study, we included all three- to five-year-old monolingual English-speaking children who completed all tasks relevant to our analyses (see below) and who had not been diagnosed with any speech or language disorder or received speech/language therapy. The mean age of children, calculated from due date, was 4.61 years (range 3.01–5.90 years), which included 53 three-year-olds (M = 3.51 years, range = 3.01–3.93 years); 139 four-year-olds (M = 4.49 years, range = 4.00–4.92 years); and 82 five-year-olds (M = 5.41 years, range = 4.95–5.90 years). There were approximately equal numbers of females (N = 139) and males (N = 136).

Measures

We used the Parent Administered Language (PAL) test¹ (Stromswold, 2006), designed to assess children’s skills in the areas of spoken and receptive language most frequently assessed in standardized language tests (lexical access, receptive vocabulary, syntax, articulation).

Procedure

Parents received a test packet instructing them to test their child in a quiet room away from distractions and explaining the importance of not coaching their child. The tests were completed in the following order: articulation, SCN, MCN, vocabulary, and syntax. Instructions for administering each test and recording responses directly preceded each test. Tests were designed to minimize judgments by parents in that parents simply recorded children’s responses for us to code.

Single-category naming (SCN) task

Children had 30 seconds to name as many animals as they could. The number of unique animals named was used as the first measure of lexical access ability. Parents were given the following instructions:

For this section you will be timing how many animals your child names in 30 seconds. You will need either a stop watch, a clock with a ‘second hand’ or a watch that lets you see when 30 seconds is up. Write down all the things your child says in 30 seconds (including things that are not animals). Do not give your child any examples besides the one example in the instructions. Start timing right after you say ‘Ready? Go!’ to your child.

and read the following script to their child:

Now we’re going to play a thinking game. Let’s think of animals … A dog is an animal. Now it’s your turn to think of as many other animals as you can. Ready? Go!

If a child named the same animal more than once, the animal was only counted once. As a general rule of thumb, animals were counted as unique, non-repetitions if they would be given their own entry in a standard encyclopedia. Several more specific instructions follow from this rule of thumb. If a child named two pronunciation variants of a single animal (e.g., ‘lamb’ and the diminutive form ‘lamby,’) credit was only given for a single animal. However, if a child named two phonologically distinct words that can be used to refer to a single animal, credit was given for each animal named. Therefore, ‘puma’ and ‘panther’ were counted as distinct animals, as were adult and baby names of the same animal (e.g., ‘deer’ and ‘fawn’); male, female and sex-neutral names of animals (e.g., ‘buck,’ ‘doe,’ ‘deer’); and animals and their superordinate class (e.g., ‘iguana’ and ‘reptile’). In the same spirit, if an adjective and animal name can be compounded to form a distinct lexical item (e.g., ‘bluebird’ or ‘polar bear’), the animal was counted as a unique animal. However, if an adjective simply modified an animal name (e.g., ‘red bird’ or ‘white bear’), these animals were counted as being identical to the animal being modified. For example, if a child named both ‘bear’ and ‘white bear,’ ‘bear’ was only counted once, but if the child said ‘hummingbird and bluebird,’ the child received credit for saying two animals. Additionally, words that referred to imaginary or mythological creatures were counted (e.g., ‘unicorn’ and ‘dragon’), but words that clearly did not refer to animals were not counted (e.g., ‘tree’). Lastly, proper names that clearly referred to an animal were counted (e.g., ‘Barney’ or ‘Fluffy’), regardless of whether the animal category had previously been named. All tests were scored by two research assistants who were blind to the details of the experiment, with a third research assistant acting as a tie breaker for any discrepancies. The intercoder concordance rate for number of animals named in the SCN test was over 95%.

Multi-category naming (MCN) task

Parents were given the following instructions:

For this section you will be timing how many items your child answers in 30 seconds. You will again need either a stop watch, clock with a ‘second hand’ or a watch that lets you see when 30 seconds is up. Begin timing right as you begin saying the first item. If your child does not respond to an item within a few seconds or says ‘I don’t know,’ mark the item ‘No response’ and go to the next item. If you get through all of the items before 30 seconds are up, you can go back to items your child skipped. Mark any items that you don’t get to in 30 seconds as ‘No response.’

and read the following script to their child:

Now we’re going to play a game where I’ll tell you what to name, and you’ll tell me something as fast as you can. For example, if I say, ‘Name something cold,’ you could say ‘ice.’ Are you ready? Let’s play!

Five prompts asked for a member of a superordinate category (a vegetable, a number, a drink, a toy, and a piece of clothing), three asked for an object having a physical characteristic (something round, something red, something big), and two asked for a part of an object (a part of the face, a part of a car). In order to receive credit, a response must have fit into the right type of category. For example, ‘convertible’ would not have been a correct response to ‘name a part of a car.’ In this sense, we were not testing vocabulary or word association but rather the ability to retrieve lexical items from lexical categories at the right level.

Receptive vocabulary

During this forced-choice picture-pointing task, children listened to words and were asked to point to pictures of the corresponding words. Children saw 12 pictures simultaneously, eight of which were targets (nurse, dentist, mittens, sandals, helicopter, kayak, trumpet, saxophone) and four of which were distracter items (canoe, astronaut, doctor, guitar). Parents were given the following instructions:

On the next page, there are twelve pictures labeled with the letters A, B, C, D, E, F, G, H, I and K. You will say the 8 words listed below and ask your child to point to the picture that the word means. Write the letter of the FIRST picture your child points to, even if he or she changes his or her mind and points to another. If your child doesn’t point to a picture, write ‘No Response’ on the blank.

and read the following script to their child:

Now we’re going to play a game where I say a word, and have to find picture for the word. Ready? Let’s play!

The PAL receptive vocabulary test differs from typical forced-choice picture pointing tasks in that all 12 items are present for all trials. Therefore, the probability of correctly identifying any item by chance alone is one in 12. The receptive vocabulary task assessed children’s knowledge of multiple lexical items from a single semantic category (e.g., the test included three musical instruments, of which ‘saxophone’ and ‘trumpet’ were target words). The receptive vocabulary task assessed children’s knowledge of lexical items from what adults would consider to be multiple superordinate categories (musical instrument, clothing, vehicle, and medical occupation), although it is currently unclear whether preschoolers have these categories. The number of items correctly identified was taken as a measure of receptive vocabulary ability, with a maximum possible score of eight. The intercoder concordance rate for the vocabulary test was 100%.

Syntax

This task is a forced-choice, picture–sentence matching test with 12 sentences. Children listened to a sentence while simultaneously viewing two pictures and pointed to the picture that best depicted the propositional content of the sentence. Parents were given the following instructions:

In this task, you will read 12 sentences to your child, one by one. These sentences are listed below. Each sentence corresponds to a page in the ‘Understanding Sentences Picture Booklet.’ For example, sentence 1 corresponds to page 1, sentence 2 corresponds to page 2, etc. Your child’s task is to listen to the sentence you read and then point to one of the two pictures shown on that page. Familiarize your child with the animals: All of the animals used in the picture booklet are shown on the next page of this packet. Introduce your child to the animals on this page. For example, for the first two animals, point to the fox and say, ‘This is a fox.’ Point to the lion and say, ‘and THIS is a lion.’ Then ask, ‘Can you point to the FOX? Now, can you point to the LION?’ If there are any animals that you are not confident your child can identify correctly after being introduced to them, please circle them. On to the task! Now we are ready for the 12 sentences. Read each sentence to your child only once, and ask your child to point to the picture that matches the sentence. Remember, read each sentence to your child only once. … Hint: Some of the sentences are a bit tricky, so read each sentence silently to yourself before reading the sentence aloud to your child.

and read the following script to their child:

‘I am going to read you a sentence, and you have to pick the picture that matches the sentence. Listen very carefully and think before you choose! Are you ready for the first one?’ Circle the picture your child points to. If your child does not respond, write ‘No response’ next to the pair of pictures.

The sentences in this test are semantically reversible insomuch as children cannot select the correct one without having mastered particular aspects of English morphosyntax that previous research has shown preschool-aged children sometimes find difficult (e.g., Slobin, 1966; Wexler & Chien, 1985). Half of the sentences are active and passive sentences (e.g., the pig kissed the sheep and the pig was kissed by the sheep, respectively) and half are sentences with reflexive and non-reflexive pronouns (e.g., the bunny scratched himself and the bunny scratched him, respectively). The number of pictures correctly pointed to was taken as a measure of receptive syntax ability, with a maximum possible score of 12.

Articulation

The PAL articulation test is a standard word repetition task. Children repeated 12 high frequency, mono-morphemic, monosyllabic words, and the number of correctly articulated onsets was taken as a measure of articulation ability. The onsets were selected to be appropriately difficult for each age group (see Sanders, 1972; Vihman, 1996), thereby reducing the likelihood of ceiling or floor effects while minimizing the number of tested words. Parents were given the following instructions:

Ask your child to repeat each of the words below, one-by-one. For each word, if your child says the underlined sound correctly, mark it ‘Correct.’ If your child does not say the underlined sound correctly, mark it ‘Incorrect.’ If incorrect, write the child’s incorrect pronunciation on the line next to the box. Incorrect sounds might be your child omitting a sound (e.g., saying ‘tuck’ for ‘truck’) or substituting another sound (e.g., saying ‘dat for ‘that’). If your child says nothing, write ‘no response’ on the line. Say each word only once.

Validity of measures

The results of validation studies suggest that the PAL tests are valid measures of children’s linguistic abilities. For example, in one study (Stromswold, Sheffield, Truit, & Molnar, 2013), 122 three- to five-year-old monolingual English-speaking children participated in a validation study in which parents administered all PAL tasks and an experimenter administered the Clinical Evaluation of Language Fundamentals – Preschool 2 (CELF) core language tasks (Wiig, Secord, & Semel, 2004) and, because the CELF does not evaluate articulation, the Denver Articulation Screening Exam (DASE, Drumwright, Van Natta, Camp, Frankenburg, & Drexler, 1973). Correlations were significant between the CELF Expressive Vocabulary test and the PAL receptive vocabulary test (r = .55, p < .0001); the CELF Sentence Structure test and the PAL syntax test (r = .46, p < .0001); and the DASE and the PAL articulation test (r = .42, p < .0001), indicating that the PAL tests are valid measures of language in preschoolers. Furthermore, the PAL tests are extremely specific, with lower correlations between scores on PAL subtests (all rs between .24 and .38) than between the CELF subtests (all rs between .68 and .72).

Results

Sex and language

Several studies have indicated that female preschoolers have a larger vocabulary than their male counterparts (e.g., Bornstein, Hahn, & Haynes, 2004), although sex differences on other aspects of language development have not been as widely investigated. However, in our study, males did not perform differently than females on any task at any age or collapsed across ages (all unpaired t-test p’s > .1).

Development of lexical access

Consistent with previous findings (e.g., Korkman, Kirk, & Kemp, 2007), children’s performance on both lexical access tasks improved dramatically between the ages of three and five. On the SCN task, three-year-olds named an average of 3.51 animals (SE = 0.32); four-year-olds named an average of 6.05 animals (SE = 0.20); and five-year-olds named an average of 7.61 animals (SE = 0.25). A one-way ANOVA revealed a significant effect of age group on number of animals named (F(2,273) = 48.16, p < .0001). Pairwise analyses revealed significant differences between all pairs of age groups (all p’s < .0001). Number of animals named was also highly correlated with age as a continuous variable (r = .55, p < .01). On the MCN task, three-year-olds named an average of 4.30 items (SE = 0.44); four-year-olds named an average of 6.62 items (SE = 0.22); and five-year-olds named an average of 7.51 items (SE = 0.23). A one-way ANOVA revealed a significant effect of age group on number of items named (F(2,272) = 26.26, p < .0001). Pairwise analyses revealed significant differences between all pairs of age groups (all p’s < .01). Number of items named was also moderately correlated with age as a continuous variable (r = .46, p < .01).

Relationship between SCN and MCN tasks

Scores on the two lexical access tasks were moderately correlated for each age group: three-year-olds (r = .46, p < .01); four-year-olds (r = .48, p < .01); and five-year-olds (r = .34, p = .01), with none of the coefficients being significantly different from any of the others (all ps > .10).

Although scores on the two lexical access tasks were moderately correlated at each age, performance on one test only predicted between 12% and 23% of the variance on the other test, suggesting that the two tasks tap different skills. In our final set of analyses, we explored the relationship between lexical access development and the development of other linguistic abilities through a series of simple and multiple regression analyses.

Receptive vocabulary, syntax, and articulation scores are presented in Table 2. Note that there were no ceiling or floor effects on any task.

Table 2.

Mean scores by age.

Task	Age (n)						Max. score
	3 (53)		4 (139)		5 (82)
	M	SE	M	SE	M	SE
SCN	3.51	0.32	6.05	0.21	7.61	0.25	—
MCN	4.30	0.44	6.62	0.22	7.51	0.23	10
Vocabulary	4.77	0.26	6.03	0.14	6.42	0.17	8
Articulation	8.89	0.38	10.00	0.20	10.74	0.26	12
Syntax	8.26	0.25	9.92	0.16	10.09	0.18	12

Regression analyses

Correlations between the scores on each of the two lexical access tasks and each of the other linguistic tasks are presented in Table 3. We performed standard multiple regression analyses at each age (i.e., all independent variables were entered in the same step) to determine which linguistic factors – vocabulary, syntax, and/or articulation – were significant independent predictors of scores on each of the lexical access tasks. These results are presented in Table 4. Given that the syntax and articulation tasks differ for each age group, it was not possible to conduct a single multiple regression analysis with age as a variable. As the SCN and MCN tasks were both designed to test lexical access, inclusion of either task in a multiple regression analysis as an independent predictor of the other task would have overshadowed the predictive power of the other linguistic factors. Therefore, we did not include the SCN or MCN task scores as independent variables in the multiple regression analyses.

Table 3.

Correlations between lexical access and other tasks.

Task	Age
	3		4		5
	SCN	MCN	SCN	MCN	SCN	MCN
Vocabulary	.35*	.46*	.35*	.35*	.12	.38*
Syntax	.26	.25*	.41*	.32*	.08	.39*
Articulation	.06	.24	.26*	.39*	.08	.31*

Note: Two-tailed p < .05.

Table 4.

Predictors of lexical access scores by age.

Task		Age
		3		4		5
		SCN	MCN	SCN	MCN	SCN	MCN
Con.	B	.29	–2.92	–1.35	–.88	5.76*	1.33
	95% CI	[–2.93, 2.51]	[–4.23, 4.25]	[–3.74, 1.04]	[–3.29, 1.53]	[2.18, 9.34]	[–1.60, 4.26]
Voc.	B	.38*	.72*	.29*	.34*	.12	.40*
	95% CI	[.05, .70]	[0, .84]	[–.05, .43]	[–.02, .46]	[.02, .58]	[–.02, .58]
Syn.	B	.29	.23	.45*	.12*	.05	.17
	95% CI	[–.10, .58]	[–.32, .58]	[.09, .57]	[–.05, .43]	[–.14, .42]	[–.14, .42]
Art.	B	–.11	.21	.12	.29*	.05	.17
	95% CI	[–.35, .09]	[–.12, .48]	[–.06, .30]	[.09, .45]	[–.19, .29]	[.02, .38]
R ²		.18	.28	.23	.24	.02	.22
F		3.52	6.15	13.28	14.28	.43	7.19

Note: Con. = constant; Voc. = vocabulary; Syn. = syntax; Art. = articulation; CI = confidence interval.

p < .05.

At age three, multiple regression analyses revealed that only receptive vocabulary (β = .31, p = .02) was a significant independent predictor of number of animals named on the SCN task (F(3,49) = 3.52, p = .02, adjusted R² = .13) and only receptive vocabulary (β = .42, p < .01) was a significant independent predictor of number of items named on the MCN task (F(3,49) = 6.15, p < .01, adjusted R² = .23). At age four, receptive vocabulary was a significant independent predictor (β = .12, p = .02), as was syntax (β = .33, p < .0001) of four-year-olds’ SCN task scores (F(3,135) = 13.28, p < .0001, adjusted R² = .21). Receptive vocabulary (β = .23, p < .01), syntax (β = .19, p = .02), and articulation (β = .27, p < .01) were all significant independent predictors of four-year-olds’ MCN task scores (F(3,134) = 14.28, p < .01, adjusted R² = .23). At age five, none of the three linguistic tasks was an independent predictor of number of animals named (F(3,78) = .43, p = .73.) and only receptive vocabulary (β = .30, p < .01) was a significant independent predictor of MCN task scores (F(3,78) = 7.19, p < .01, adjusted R² = .22).

Discussion

Developmental trajectories

The current study is the first to investigate performance on two types of lexical access tasks in the same group of preschool-age children. Results reveal a steep developmental trajectory in lexical access between the ages of three and four. Scores on the SCN and MCN lexical access tasks were not perfectly correlated with each other and differed in their correlations with scores on the other linguistic tasks. Although both tasks assess children’s lexical access ability, regression analyses suggest they may tap different linguistic abilities, which may have different developmental trajectories.

Differences in the lexical access tasks themselves, including the linguistic components they involve, can account for their different developmental trajectories. For the three-year-olds, receptive vocabulary was the only independent predictor of scores on either lexical access task. One explanation we considered was that vocabulary size could play such an integral role in lexical access only when vocabulary is small and thus, a limiting factor. In other words, for the three-year-olds, the lexical access tasks could have essentially been tests of receptive vocabulary. However, if vocabulary was a limiting factor on the SCN task, this would entail that the three-year-olds only knew the names of approximately three or four animals, the mean number of animals named, or as few as 0 animals, the lowest score a three-year-old obtained on the SCN task. In stark contrast, according to the MacArther–Bates Communicative Development Inventories, by 30 months the majority of children can name almost 60 animals (Dale & Fenson, 1996). It is therefore more likely that for three-year-olds, the receptive vocabulary task measured not only vocabulary but also lexical access, which had the effect that scores on the receptive vocabulary task predicted scores on both the SCN and the MCN tasks independently. As at age three the independent predictors did not differ for the lexical access tasks, at this age the tasks are likely equally useful for testing and diagnosis.

For older children, the independent predictors of scores on the two lexical access tasks diverged. At age four, articulation score was an independent predictor of MCN task score but not of SCN task score. By age five, none of the other core linguistic tasks were independent predictors of SCN task score, but receptive vocabulary was still an independent predictor of MCN task score. If the goal is to obtain a measure of lexical access ability that is less tainted by other linguistic demands, the SCN test is probably better than the MCN test.

Differences in retrieval demands

Several differences between the SCN and the MCN tasks stem from the nature of the lexical categories. First, success on the MCN task requires accessing a single lexical item from multiple categories. In contrast, success on the SCN task requires accessing multiple items from a single category. Second, optimal performance on the MCN task necessitates quickly locating the prompted category, retrieving a word from that category, and then quickly switching to the next prompted category. Conversely, performance on the SCN task necessitates retrieving multiple lexical items from a single category.

Third, in the MCN task, some of the prompts asked children to name a member of a superordinate category (e.g., a vegetable); some asked children to name objects with a given physical characteristic (e.g., something round); and some had children name a part of an object (e.g., a part of the face). In the SCN task, the prompt always asked children to name members of a superordinate category (animals).

Fourth, the two lexical access tasks differ in that there are more members of the animal category than there are members of many of the MCN task categories. Previous research on the effect of semantic neighborhood density on lexical access has yielded mixed results, with some studies suggesting that high semantic neighborhood density facilitates lexical access and others suggesting high density inhibits lexical access. These studies are difficult to compare in that they differ in operationalization of semantic neighborhood density, psychometric measures, distance of neighbors, and participants’ lexical access abilities (for varied perspectives, see Bormann, 2011; Mirman & Magnuson, 2008; Pexman, Hargreaves, Siakaluk, Bodner, & Pope, 2008; Yap, Pexman, Wellsby, Hargreaves, & Huff, 2012). Although the effect of semantic neighborhood density is still under debate, it is likely that the size of a category is one factor that affects the ease with which a category can be traversed. Any effect of semantic neighborhood density, whether positive or negative, would distinguish the SCN task from the MCN task in terms of difficulty of retrieval. A possible explanation for why, overall, children performed better on the MCN task than on the SCN task is that the high semantic density of the animal category inhibited lexical access. In theory, the MCN task is more liable than the SCN task to ceiling effects in that, on the MCN task, it is only possible to obtain a score as high as the number of category prompts (in this case, 10), whereas on the SCN task, the score is only limited by the child’s performance. However, ceiling effects were not obtained in our study.

Finally, both the SCN task category prompt (‘animals’) and the highest frequency animal words are higher frequency words than the MCN task category prompts and the highest frequency members of MCN task categories. However, again, children said more words on the MCN than SCN task, suggesting that frequency did not impede performance.

Differences in associated linguistic components

The SCN and MCN tasks also place different demands on children’s receptive language. On all three levels – vocabulary, syntax, and speech perception (the receptive counterpart of articulation) – the receptive demands are likely higher for the MCN task. In terms of receptive vocabulary, to succeed on the MCN task, a child must understand the names of several categories. In contrast, to succeed on the SCN task, a child only must understand the name of one category (animals) – and the word animal is acquired by most children before age two (Dale & Fenson, 1996). In terms of receptive syntax, category prompts were syntactically more complex on the MCN task, which included three different types of prompts: Name a ___, Name something ___, Name a part of a ___. For each question, the child needed to comprehend a new category, some of which included prepositional phrases. On the SCN task, the child only needed to comprehend the syntax of the category prompt once, and the syntax was rather straightforward. Similarly, the speech perception demands were greater for the MCN task, in which the child needed to process 10 prompts. For these reasons, the SCN task is inherently a purer test of lexical access. The comparative difficulty of the MCN task category prompts is a possible explanation for why, overall, receptive vocabulary, syntax, and articulation scores were more correlated with, and better independent predictors of, performance on the MCN task than on the SCN task.

The SCN and MCN tasks also differ in terms of how likely they are to foster phonological priming effects. Previous research suggests that adults phonologically cluster during SCN tasks (e.g., Slowiaczek & Hamburger, 1992) – e.g., following/kangaroo/with/cougar/ – although we are not aware of any research investigating whether adults phonologically cluster during MCN tasks. If children phonologically clustered, as adults do, we might expect articulation scores to correlate more highly with SCN than MCN task scores in that strong phonological awareness could lead both to success on the articulation task and to self-priming on the SCN task. Phonological self-priming should be less likely on the MCN task, where children’s responses are interspersed with adults’ prompts and where the semantic categories are both narrower and less consistent from response to response. The fact that articulation scores were only correlated with SCN task scores at age four and that articulation scores did not independently predict SCN task scores at any age suggests that children do not phonologically self-prime.

In contrast, articulation scores were correlated with MCN task scores at every age and did independently predict MCN task scores at age four. One reason for greater correlation between articulation scores and MCN task scores than SCN task scores could be the additional articulatory demands of the MCN task. First, because children likely knew more lexical items within the SCN task category (animals) than within the MCN task categories, children could avoid words they had trouble pronouncing during the former but not the latter task. Second, MCN task lexical items tend to be more phonologically complex (more syllables, more consonant clusters) than SCN task lexical items. One possibility then is that poor articulation ability inhibited some children from succeeding on the MCN task but not on the SCN task.

Our results suggest that different lexical access tasks – including different category prompting tasks – capture different aspects of lexical access. Exploring the differences between different lexical access tasks is a powerful way of determining the linguistic components necessary for successful lexical access. Furthermore, learning how these components interact at different ages is important for understanding how lexical access develops. Although not perfect, the SCN task appears to provide a purer measure of preschoolers’ lexical access abilities than the MCN task. Given that a deeper appreciation of the linguistic components of lexical access will be necessary for designing better lexical access tests, future research could separately test receptive and expressive language in order to better understand the importance of these components in lexical access.

Footnotes

Funding

This research was supported in part by grants from the the Bamford-Lahey Children’s Foundation, the Charles and Johanna Busch Biomedical Research Fund, and the National Science Foundation (BCS-0446850, DGE-0549115).

Notes

References

Benjamin

A. S.

Bjork

R. A.

(1998). The mismeasure of memory: When retrieval fluency is misleading as a metamnemonic index. Journal of Experimental Psychology: General, 127(1), 55–68.

Bormann

(2011). The role of lexical-semantic neighborhood in object naming: Implications for models of lexical access. Frontiers in Psychology, 127(2).

Bornstein

M. H.

Hahn

C. S.

Haynes

O. M.

(2004). Specific and general language performance across early childhood: Stability and gender considerations. First Language, 24, 267–304.

Caramazza

(1997). How many levels of processing are there in lexical access? Cognitive Neuropsychology, 14, 177–208.

Crowe

Prescott

T. J.

(2003). Continuity and change in the development of category structure: Insights from the semantic fluency task. International Journal of Behavioral Development, 27, 467–479.

Dale

P. S.

Fenson

(1996). Lexical development norms for young children. Behavioral Research Methods, Instruments, & Computers, 28, 125–127.

Dickinson

D. K.

McCabe

Anastasopoulos

Peisner-Feinberg

E. S.

Poe

M. D.

(2003). The comprehensive language approach to early literacy: The interrelationships among vocabulary, phonological sensitivity, and print knowledge among preschoolers. Journal of Educational Psychology, 95, 465–481.

Drumwright

Van Natta

Camp

Frankenburg

Drexler

(1973). The Denver articulation screening exam. Journal of Speech and Hearing Disorders, 38(1), 3–14.

Estes

W. K.

(1974). Foundations of stimulus sampling theory. In Krantz

D. H.

Atkinson

R. C.

Luce

R. D.

Suppes

(Eds.), Contemporary developments in mathematical psychology, volume 1: Learning, memory, and thinking (pp. 163–183). San Francisco, CA: W.H. Freeman & Company.

10.

Forster

K. I.

(1976). Accessing the mental lexicon. In Wales

R. J.

Walker

(Eds.), New approaches to language mechanisms (pp. 257–287). Amsterdam, The Netherlands: North-Holland.

11.

Funnel

Hughes

Woodcock

(2006). Age of acquisition for naming and knowing: A new hypothesis. Quarterly Journal of Experimental Psychology (Hove), 59, 268–295.

12.

Kambanaros

Grohmann

K. K.

Michaelides

(2013). Lexical retrieval for nouns and verbs in typically developing bilectal children. First Language, 33, 182–199.

13.

Koren

Kofman

Berger

(2005). Analysis of word clustering in verbal fluency of school-aged children. Archives of Clinical Neuropsychology, 20, 1087–1104.

14.

Korkman

Kirk

Kemp

S. L.

(2007). NEPSY II: Clinical and interpretative manual. San Antonio, TX: Psychological Corporation.

15.

Levelt

W. J. M.

(2001). Spoken word production: A theory of lexical access. Proceedings of the National Academy of Sciences, 98, 13464–13471.

16.

Loftus

E. F.

Freedman

J. L.

(1972). Effect of category-name frequency on the speed of naming an instance of the category. Journal of Verbal Learning and Verbal Behavior, 11, 343–347.

17.

Lucariello

Kyratzis

Nelson

(1992). Taxonomic knowledge: What kind and when? Child Development, 63, 978–998.

18.

Mandler

J. M.

(1983). Representation. In Flavell

J. H.

Markman

E. M.

(Eds.), Handbook of child psychology: cognitive development (Volume 3; Mussen

P.H.

, Series Ed.) (pp. 420–494). New York, NY: Wiley.

19.

Marslen-Wilson

W. D.

(2001). Access to lexical representation: Cross-linguistic issues. Language and Cognitive Processes, 5–6, 699–708.

20.

Messer

Dockrell

J. E.

(2006). Children’s naming and word finding difficulties: Descriptions and explanations. Journal of Hearing, Speech & Language Research, 49, 309–324.

21.

Mirman

Magnuson

J. S.

(2008). Attractor dynamics and semantic neighborhood density: Processing is slowed by near neighbors and speeded by distant neighbors. Journal of Experimental Psychology: Learning, Memory, and Cognition, 34(1), 65–79.

22.

Pexman

Hargreaves

Siakaluk

Bodner

Pope

(2008). There are many ways to be rich: Effects of three measures of semantic richness on word recognition. Psychonomic Bulletin and Review, 15, 161–167.

23.

Sanders

L. J.

(1972). Procedure guides for evaluation of speech and language disorders in children. Danville, IL: Interstate Printers and Publishers.

24.

Slobin

(1966). Grammatical transformations and sentence comprehension in childhood and adulthood. Journal of Verbal Learning and Verbal Behavior, 5, 219–227.

25.

Slowiaczek

L. M.

Hamburger

(1992). Prelexical facilitation and lexical interference in auditory word recognition. Journal of Experimental Psychology: Learning, Memory and Cognition, 18, 1239–1250.

26.

Snyder

H. R.

Munakata

(2010). Becoming self-directed: Abstract representations support endogenous flexibility in children. Cognition, 116, 155–167.

27.

Stromswold

(2006). Parent administered language test (PAL) (Unpublished assessment). Department of Psychology and Center for Cognitive Science, Rutgers, The State University of New Jersey – New Brunswick, Piscataway.

28.

Stromswold

Sheffield

Truit

Molnar

(2013). Parents can test preschool children’s language: The parent-administered language test (PAL). Manuscript submitted for publication.

29.

Swinney

(1979). Lexical access during sentence comprehension: (Re) consideration of context effects. Journal of Verbal Learning and Verbal Behavior, 18, 645–659.

30.

Thurstone

L. L.

(1938). Primary mental abilities (Psychometric monographs, 1). Chicago, IL: University of Chicago Press.

31.

Tombaugh

T. N.

Kozak

Rees

(1999). Normative data stratified by age and education for two measures of verbal fluency: FAS and animal naming. Archives of Clinical Neuropsychology, 14, 167–177.

32.

Vihman

(1996). Phonological development: The origins of language in the child. Cambridge, MA: Blackwell Publishers.

33.

Weizman

Z. O.

Snow

C. E.

(2001). Lexical input as related to children’s vocabulary acquisition: Effects of sophisticated exposure and support for meaning. Developmental Psychology, 37, 265–279.

34.

Wexler

Chien

Y. C.

(1985). The development of lexical anaphors and pronouns. Papers and Reports on Child Language Development, 24, 138–149.

35.

Wiig

E. H.

Secord

W. A.

Semel

(2004). Clinical evaluation of language fundamentals preschool (2nd ed.). San Antonio, TX: Harcourt Assessment, Inc.

36.

Yap

Pexman

Wellsby

Hargreaves

Huff

(2012). An abundance of riches: Cross-task comparisons of semantic richness effects in visual word recognition. Frontiers in Human Neuroscience, 6, 1–10.