Date of Degree


Document Type


Degree Name





Marilyn Shatz

Committee Members

Helen Cairns

Annick Mansfield

Joseph Glick

Gilbert Voyat

Subject Categories



This study was devised to investigate the developmental syntagmaticparadigmatic word association shift. In syntagmatic associations the stimulus and associative response are of different grammatical form classes and appear to be grammatically continuous, as response may follow stimulus in an utterance (e.g. cat-meows). These are the predominant responses of children before the ages of six to eight. Older children and adults shift to making paradigmatic associations in which stimulus and response are from the same form class and may be substituted for one another in an utterance (e.g. cat-dog). This shift was explained in terms of underlying symbolic mediational processes and the epistemic subject, the child’s interpretation of the unstated task demands of the word association test.

The central hypotheses were:

1. The syntagmatic-paradigmatic shift occurs because younger children use verbally-evoked images to generate their word associations and so produce syntagmatic responses. Older children, on the other hand, rely less on verbally-evoked images and more on knowledge of unstated task demands to generate their word associations. Older children can accomplish this because they have acquired explicit or potentially articulated knowledge of hierarchic semantic organization (abbreviated HSO) of words (e.g. cats and dogs are animals), which younger children have only tacit, unarticulated knowledge of. On the basis of other studies (e.g. Mansfield, 1977; Steinberg & Anderson, 1975) it was assumed that all age groups have comparable implicit knowledge of hierarchic semantic organization and that therefore this would not play a significant role in the shift to paradigmatic responding.

2. The ability to change deliberately the mode of representation (i.e. imaginal or linguistic) through which word association responses are mediated depends on the acquisition of two kinds of metaknowledge: a) metalinguistic knowledge and b) explicit knowledge of imagery.

To test these hypotheses, twenty kindergarteners, twenty second graders and ten college students were presented with 48 words, consisting of high and low imagery nouns, verbs and adjectives under five different word association instruction conditions: 1) the standard "firstword"; 2) paradigmatic, in which Ss were trained to give an association which could substitute for the stimulus in a sentence; 3) syntagmatic in which Ss were trained to give an association which together with the stimulus would complete a phrase; 4) imagery, in which the subject was asked to obtain a "picture" of the stimulus word in her head before responding and 5) time delay, which required that the subject wait three seconds before responding. The purpose of this last condition was to determine whether the delay in responding was responsible for a possible increased frequency of syntagmatic associations in the imagery condition. One half the kindergarteners and second graders also received: 1) two convergent measures of metalinguistic knowledge, which tap the child's understanding of words as arbitrary, interchangeable units; 2) two convergent measures of explicit knowledge of imagery, which tap the child's understanding of the distinction between (a) images and words and (b) images and objects; 3) the hierarchic semantic organization and 4) word-word relationship tasks. The latter two tasks respectively tap the child's tacit and explicit understanding of the hierarchic semantic organization of words.

A control group consisted of the remaining ten kindergarten and second grade children who received only the five word association tests. The purpose of this control group was to determine if administration of the metalinguistic and other cognitive tasks influenced subsequent word association responding.

The results showed that there appears to be a developmental progression in the way children approach the word association task. First, very young children seem to produce responses based only on the phonological form of the word; i.e., they produce clang responses, which rhyme with the stimulus, or negated responses, generated through an "negation" rule which consists of prefixing "not" or "un" to the phonological form of the stimulus word to produce a response. Later, children begin to consider the meaning of the stimulus word. At the start, this may be confined to considering the imagery which the stimulus word evokes. Evidence indicates that stimulus imagery mediates the generation of syntagmatic associations. Gradually, as tacitly known hierarchic semantic organization becomes more filled-in,* and the child gains explicit access to it, it is possible for her to produce paradigmatic responses. But in order to do so spontaneously, she must also understand that words are arbitrary, manipulable units. On the other hand, if she is explicitly directed to produce paradigmatic responses, she will be able to do so, as long as she has explicit access to a filled-in hierarchic semantic structure. Metalinguistic knowledge does not play a role in this case since the explicit directions replace its function as an aid to the spontaneous production of paradigmatic responses.

* As determined by absolute number of correct responses.


Digital reproduction from the UMI microform.

Included in

Psychology Commons