Elsevier

Brain and Language

Volume 89, Issue 1, April 2004, Pages 192-202
Brain and Language

Verbal fluency output in children aged 7–16 as a function of the production criterion: Qualitative analysis of clustering, switching processes, and semantic network exploitation

https://doi.org/10.1016/S0093-934X(03)00367-5Get rights and content

Abstract

Developmental changes in children’s verbal fluency were explored in this study. One hundred and forty children aged from 7 to 16 completed four verbal fluency tasks, each with a different the production criterion (letter, sound, semantic, and free). The age differences were analyzed both in terms of number of words produced, and clustering, switching, and semantic network exploration. Analysis of the number of words produced showed a larger difference between the 7–8- and the 9–10-year-olds in semantic than in letter fluency, but this difference gradually disappeared with increasing age for semantic fluency while remaining constant for letter fluency. In letter fluency production, age modified both the number of switches and clusters formed whereas in semantic fluency tasks, only cluster size changed with age. Concerning the semantic network exploration indicators derived from the supermarket fluency task, the number of categories sampled increased from 11 to 12 years, but efficient semantic exploitation occurred only after the age of 13–14 years. These results are discussed in terms of the development of strategic retrieval components and categorical knowledge.

Introduction

Verbal fluency tasks are widely used in cognitive psychology and neuropsychological assessment of strategic search and retrieval processes from the lexicon and semantic memory. Consequently, the verbal fluency test provides useful information about the development of both word retrieval strategies and lexico-semantic networks during childhood. The most common tasks used are letter fluencies, where subjects must generate as many words as possible beginning with a specified letter such as F, A, or S in a limited time (Benton & Hamsher, 1976; Raskin, Sliwinski, & Borod, 1992; Troyer et al., 1997, Troyer et al., 1998a, Troyer et al., 1998b; Troyer, 2000), and semantic fluencies, where words must belong to a specified semantic category like “Fruit” or “Supermarket” (Beatty, Monson, & Goodkin, 1989; Bousfield, 1953; Gruenewald & Lockhead, 1980; Randolph, Braun, Goldberg, & Chase, 1993; Tröster et al., 1995; Testa et al., 1998; N’Kaoua, Lespinet, Barsse, Rougier, & Claverie, 2001; Troyer, 2000; Troyer et al., 1997, Troyer et al., 1998a, Troyer et al., 1998b). As in adult subjects, childhood studies have also reported that the number of words produced by children aged 5–11 is greater in semantic than in letter fluency tasks, thus suggesting that the semantic fluency task is easier (Kremin & Dellotas, 1996; Riva, Nichelli, & Devoti, 2000). In view of the word extent to which each criterion refers, letter fluencies are classically considered as more strategic search based (or more dependent on executive functions such as cognitive flexibility and shifting) than semantic ones. Indeed, retrieval by letter appears to require the exploration of more subsets of words than the retrieval of exemplars of a given semantic category (Miceli, Caltagirone, Gainotti, Massulla, & Silveri, 1981; Monsch et al., 1992; Ober, Dronkers, Koss, DElid, & Friedland, 1986; Riva et al., 2000; Rosen, 1980; Troyer et al., 1997).

Regarding age differences in childhood, Riva et al. reported a clearly age-related improvement in the semantic fluency test, whereas age-related differences were only observed between the 5- and 11-year-olds in the letter fluency test. The improvement in semantic fluency performance with increasing age was explained by the authors as resulting from lexico-semantic memory development. In contrast, the belated age difference for letter tasks was explained by the fact that the strategic search abilities (cognitive flexibility and shifting) required by this task depend on the maturation of the frontal lobe which, according to some authors (Golden, 1981; Passler, Isaae, & Hynd, 1985), occurs at about 12 years of age. This corroborates findings from behavioral studies stating that the major improvement in cognitive strategies, working memory, processing and, attentional capacities is observed between 10 and 12 years (for review, Bjorklund & Douglas, 1997; Cowan, 1997; Guttentag, 1997). Moreover, the study by Kremin and Dellotas (1996) supports the assumption that the letter fluency test is particularly strategic research-based. Obtaining similar results to those by Riva et al. in children aged 5–8, they conducted correlational analyses revealing that letter fluency performance was not specially correlated with a test assessing the segmentational spelling knowledge of the words reported. This led them to propose that letter fluency performance in older children appears relatively more dependent on strategic retrieval processes than spelling knowledge.

Therefore, the age-related differences in the letter fluency test in children aged from 6 to 12 years and over may be salient when subjects are 10–12-years old, whereas they are observed earlier in the semantic fluency test (before 10–12 years). The letter fluency test may thus prove to be a good indicator of the development of the strategic component during young adulthood. This assertion would be more convincing if the specific strategic processes involved in verbal fluency were shown to change with age to a greater extent in letter fluency than semantic fluency.

Strategic retrieval process analysis highlights two kinds of process, i.e., clustering and switching (Abwender, Swan, Bowerman, & Connolly, 2001; Beatty et al., 1989; Bousfield, 1953; Gruenewald & Lockhead, 1980; Raskin et al., 1992; Tröster et al., 1995; Troyer, 2000; Troyer et al., 1997, Troyer et al., 1998a, Troyer et al., 1998b). Both reflect to the extent to which the subject’s productions are characterized by semantic (e.g., “reptile,” “mammal”) and phonemic (e.g., “frog,” “grog”) clusters. A switching process involves the search and retrieval of different semantic or phonemic subcategories and a clustering process involves the search and retrieval of exemplars from the same subcategory. Therefore, irrespective of the type of fluency (letter vs. semantic), a scoring system was established to assess phonemic or semantic clustering and phonemic or semantic switching processes during fluency productions with three main indicators: the number of switches, the number of clusters, and mean cluster size (for review, Abwender et al., 2001). The number of both switches and clusters associated with frontal capacities reflect the necessary cognitive flexibility for new subcategory search. The mean cluster size associated with temporal lobe functions is thought to measure the extent of the knowledge network for a given subcategory (Abwender et al., 2001; Robert et al., 1997; Troyer, 2000; Troyer et al., 1997, Troyer et al., 1998a, Troyer et al., 1998b). These assumptions are supported by several neuropsychological studies showing a decrease in the number of switches in patients with frontal-lobe lesions or Parkinson’s disease (Randolph et al., 1993; Testa et al., 1998; Troyer et al., 1998a, Troyer et al., 1998b), a decrease in the number of clusters in schizophrenic persons (Robert et al., 1997), and smaller cluster sizes in patients with temporal-lobe lesions or Alzheimer’s disease (N’Kaoua et al., 2001; Testa et al., 1998; Troyer et al., 1998a, Troyer et al., 1998b). Thus, both the number of switches and the number of clusters may provide indicators of strategic retrieval processes while cluster size reflect the extent of lexico-semantic knowledge.

Consequently, in addition to examining letter and semantic fluency performance differences, a study of switch and cluster differences may shed light on the development of both word retrieval strategies and lexico-semantic networks during childhood. Although to our knowledge no childhood study has really used this scoring, it is likely that the developmental differences in both switch and cluster numbers are particularly salient in the letter fluency test, as the latter is retrieval more strategic than its semantic counterpart.

Even if strategic retrieval processes explained the delayed development in the letter fluency task observed around 10–12 years of age, it remains to be explained why considerable early differences are observed in semantic fluency tests, as reported by Riva et al. (2000). These differences suggest that developmental factors other than the strategic retrieval component may contribute to semantic fluency performances such as learning categorical knowledge. Numerous studies have established age-related differences in categorization capacities during childhood, particularly in taxonomy-based concept formation abilities. This raises the question of the stage during development when taxonomic-based concept formation starts to contribute to the organization of children’s knowledge. Some authors propose that it is available early (around 3–4-years old), perhaps even in late infancy (Baron-Cohen, 1995; Gergely, Nadasky, Csibra, & Biro, 1995; Deak & Bauer, 1996; Leslie, 1994; Smith, 1979), while others suggest that it intervenes much later (not before the stage of concrete operations) (Carey, 1985; Inhelder & Piaget, 1964; Slaughter, Jaakkola, & Carey, 1999). Blewitt, 1989, Blewitt, 1994 provided a theoretical account of hierarchical knowledge development that helps bridge the gap between these two extremes. She argued that hierarchical knowledge emerges gradually as the child acquires deeper levels of knowledge pertaining to hierarchical relations. At the earliest level, children are able to name the same object with labels or generic words at multiple levels of specificity (e.g., at this stage, “coke” can be indiscriminately classified by children as “soda,” or “drink,” or as “edible goods”). At an intermediate level, children are able to establish a connection between hierarchically related categories, but this connection may be one of overlap rather than inclusion (e.g., at this stage, “coke” is a kind of “soda,” as well as a “drink”). At the highest level, knowledge acquired by children is hierarchical and based on the inclusion relationships that constitute the taxonomic knowledge network (e.g., at this stage, “coke” is a kind of “soda,” a “soda” is a “drink”). The shift from the intermediate to the highest level may occur from 6–7- to 10–12-years old, and is probably effective at roughly 11 years (Blewitt, 1994; Carson & Abrahamson, 1976; Lane & Hodkin, 1985; Markman, 1978). Moreover, the most common labels (i.e., basic level) (e.g., “drink” is more familiar than “beverage”) generally comprise the majority of children’s earliest words, but the acquisition of more specific subordinate names (e.g., “soda”) and more general superordinate (e.g.,“edible goods”) names occurs later and is more gradual (Anglin, 1977; Mervis, 1983; Rosch, Mervis, Gray, Johnson, & Boyes-Braem, 1976). Taken together, these findings suggest that the number of labels produced and the number of categories sampled in a semantic fluency task increase from 6 to 12 years of age. Consequently, an analysis of semantic networks may be particularly relevant in the main, early differences observed in semantic fluency. Another analytical method has been specially designed to measure semantic network organization (Beatty et al., 1989; Tröster et al., 1995). It is applied in the “supermarket” fluency task, which involves generating as many items as possible that can be purchased in a supermarket (Dementia Rating Scale, Mattis, 1988). The method is based on a categorical semantic system (“drinks”, “vegetable,” and “meat/fish”) with 3 hierarchical levels of analysis: a superior level of categories (10 categories defined by the authors, e.g., “drinks”), an intermediate level with generic names of categories and subcategories (the labels) (“drinks,” “beverages,” and “liquor”) and a lower level of exemplars (e.g. “wine”). The method measures the number of categories sampled, as well as the number of labels and exemplars that are produced. Moreover, the words per category sampled ratio provides information about the semantic network area while the category shifts per category sampled ratio is thought to reflect categorical search efficiency. Therefore, it becomes possible to investigate the role of categorical knowledge development with regard to the large, early difference in semantic fluencies.

From the above data, children’s performances in the letter fluency task could be expected to be largely explained by strategic retrieval component development, leading to a real age difference starting at 10–12 years of age, whereas those obtained with semantic fluency are mainly dependent on the development and acquisition of categorical knowledge, shown by the large, early differences. Thus, the aim of this study was to investigate the age-related differences in word production performances during late childhood (7–16 years of age) as a function of the kind of fluency test (letter or semantic criterion). We therefore conducted quantitative and qualitative analyses, evaluating the number of words produced as well as clustering, switching, and semantic network exploration as a function of age.

Section snippets

Participants

Participants included five groups (Table 1): 26 seven–eight-year-olds (2nd grade), 25 nine–ten-year-olds (4th grade), 29 eleven–twelve-year-olds (6th grade), 29 thirteen–fourteen-year-olds (8th grade), and 31 fifteen–sixteen-year-olds (from 10th grade). All participants were selected in schools affiliated with the “Institut Universitaire de Formation des Maîtres” in Limoges (i.e., Limoges university teacher training institute). The children and adolescents were recruited during an information

Results

The quantitative scoring results are presented first, followed by the switch and cluster scores, then the semantic network scores.

Discussion

To investigate the assumption that age-related differences in letter fluency performance reflected strategic retrieval component development whereas those obtained in semantic fluency were related to categorical knowledge development, we conducted quantitative and qualitative analyses of the number of words produced by the 7–16-year-olds children, their switching and clustering processes, and indicators of semantic network exploration.

Quantitative analysis revealed a larger difference between

References (48)

  • A.K Troyer et al.

    Clustering and switching on verbal fluency: The effects of focal frontal- temporal-lobe lesions

    Neuropsychologia

    (1998)
  • D.A Abwender et al.

    Qualitative analysis of verbal fluency output: Review and comparison of several scoring methods

    Assessment

    (2001)
  • J.M Anglin

    Word, object, and conceptual development

    (1977)
  • S Baron-Cohen

    Mindblindeness: An essay on autism and theory of mind

    (1995)
  • W.W Beatty et al.

    Access to semantic memory in Parkinson’s disease and multiple sclerosis

    Journal of Geriatric Psychiatriy and Neurology

    (1989)
  • A Benton et al.

    Multilingual Aphasia Examination

    (1976)
  • D.F Bjorklund et al.

    The development of memory strategies

  • P Blewitt

    Understanding categorical hierarchies: The earliest levels of skill

    Child Development

    (1994)
  • P Blewitt

    Categorical hierarchies: Levels of knowledge and skill

    Genetic Epistemologist

    (1989)
  • W.A Bousfield

    The occurrence of clustering in the recall of randomly arranged associates

    The Journal of General Psychology

    (1953)
  • D Cardebat et al.

    Evocation lexicale formelle et sémantique chez des sujets normaux: Performances et dynamiques de production en fonction du sexe, de l’âge et du niveau d’étude

    Acta of Neurology of Belgia

    (1990)
  • S Carey

    Conceptual change in childhood

    (1985)
  • M.T Carson et al.

    Some members are more equal than others: The effect of semantic typicality on class-inclusion performance

    Child Development

    (1976)
  • N Cowan

    The development of working memory

  • Cited by (129)

    • A Battery of Easily Accessible, Simple Tools for the Assessment of Concussion in Children

      2021, Journal of Pediatrics
      Citation Excerpt :

      We were surprised by this, given its utility in formal neuropsychological assessments and published lists of norms for children wherein performance is influenced not only by age but by the number of years of school completed,44-47 although the utility of brief cognitive tests in general for this population has been repeatedly disappointing. The strategy by which children produce a list of animals appears to be independent of the language spoken48,49 and to be on a par with that of adults by age 11-12 years.48,50 The strategy is to list animals by the environment in which they are experienced,47 a process thought to involve both the temporal (clusters of animals from a particular environment) and frontal (selection of a new animal environment) lobes.48

    • The developmental trajectories of L2 lexical-semantic networks

      2023, Humanities and Social Sciences Communications
    View all citing articles on Scopus
    View full text