Part I Theoretical Background
3. Cognitive psychology of proper names
3.3 Retrieval difficulties of proper names: Some responding approaches
3.3.2 Node structure theory
Node structure theory was developed to provide a general model of the production and perception of language (MacKay, 1981, 1982, 1987). This model was used to explain TOT phenomenon of proper names by Burke et al. (1991). According to this theory, the memory system consists of a network of processing units called nodes. The nodes are organized hierarchically and represent different semantic and phonological knowledge.
The nodes of the semantic system represents concepts and are connected to the phonological nodes which are types but not tokens. Two fundamental processes in this model are priming and activation. Burke and his colleagues postulated that type-based and hierarchical organization of phonological nodes and the differences between priming and activation are important for explaining the TOT-phenomenon of proper names.
Activation is necessary to retrieve information as a node represents and proceeds hierarchically and sequentially in production. It is either a fully prohibiting or a fully facilitating process and does not spread. Activation requires a special class of activating mechanism known as sequence nodes that connect every node in a domain.
This domain is a set of nodes that all share the same syntactic function (e.g. noun) or order of occurrences in words and sentences (e.g. initial consonant group). On the contrary, priming prepares a node for possible activation and an activated node primes all connected nodes. It spreads both at higher and lower levels (Burke et al., 1991).
During word retrieval, the translation of thought into speech begins with the activation of a propositional node. Taking the production of the word frisbee for example (example taken from Burke et al., 1991), when one intends to say the word, the prepositional node such as they were throwing a frisbee is activated at the level of the semantic system. Activating this prepositional node primes the lexical node for frisbee, so that it becomes activated as the most primed node in its domain. Priming
spreads from the lexical node to connected phonological nodes. The most primed phonological nodes in their domains will be activated and prime further muscle movement nodes in the articulation system. This procedure is illustrated in Figure 3 (Burke et al., 1991).
Figure 3: Nodes representing frisbee in the semantic, phonological, and muscle movement systems (taken from Burke et al., 1991, p. 544).
Transmission deficit hypothesis
In node structure theory, TOT phenomenon was considered as a deficit in the transmission of priming across critical connections that are necessary for target word production (e.g. Cohen & Faulkner, 1986). There are three factors affecting transmission deficit: frequency of use, recency of use and aging (Burke et al., 1991).
According to MacKay (1981, 1982), connections between nodes become stronger with use. When a node has been frequently activated, the rate and amount of priming transmitted across its connections increases. Frequency of use may explain why phonological speech errors and word substitution occur more often for low frequency
words than high frequency words (Stemberger & MacWhinney, 1986; as cited in Burke et al., 1991, p. 545). Aging has been suggested to weaken the connections within the entire network nodes and reduce the rate and amount of priming transmitted across connections (MacKay & Burke, 1990). Compared with frequency and aging factors that determine the strength of connections, recency of use defines the decay of the strength of connections over time. Connections to phonological nodes that have not been used recently can become weak so that the transmitted priming will not be sufficient enough to enable the activation (Burke et al., 1991).
According to the node structure theory, Burke et al. (1991) predicted that TOT states will involve low frequency words more often than high frequency words. Moreover, TOT states will appear more for older subjects than younger subjects. Finally, recent use will influence the frequency of TOT states where words not recently used words induce more TOT states. Therefore, names of acquaintance who have been not contacted for more than three months (and much longer for older subjects) will evoke more TOT states. Those predictions have been confirmed in the study of Burke et al. (1991). They found a prevalence of proper name TOTs for all age groups and an age-related increase in proper name TOTs. This finding is consistent with the transmission deficit hypothesis. They clarified the effects of proper names on TOTs with an example in which the semantic representation underlying a proper name (e.g. Baker) and a common noun (e.g. baker) is compared. The occupation noun baker is connected to a large number of nodes representing semantic information about bakers, such as bake bread, get up early etc. whereas the proper name Baker is connected to semantic information only via lexical nodes for John Baker or other known individuals sharing the family name Baker. John Baker is connected to a large number of nodes representing semantic information of this person. However, the family proper name Baker does not have a set of connections representing information about this name, such as Baker is an old English name.
When one attempts to produce baker, the large number of connections linking the visual concept node for the person to the occupation noun will provide priming that leads to the activation of baker (occupation noun) and its connected phonological nodes. In contrast, when one attempts to greet an acquaintance named John Baker, no corresponding priming occurs at the lexical node for Baker (family proper name). There is only one connection linking the visual concept node to Baker via John Baker.
Because Baker and its connected phonological nodes receive no convergent connections, they become vulnerable to transmission deficits and TOTs, even though the person is familiar to the speaker (Burke et al., 1991) (see Figure 4).
The transmission deficit hypothesis also predicts that the presentation of alternative words (blockers, persistent alternatives) will not increase the probability of TOTs because TOTs are caused by transmission deficits but not by interference. This prediction has been confirmed by a study of Cross and Burke (2004) in which prior production of a related character name did not affect TOTs.
Recent studies of James (2004) as well as Burke et al. (2004) have provided further evidence for transmission deficit hypothesis in which recall errors occurred more often for people’s names and the phonological priming of proper names reduced the number of TOTs. It has been suggested that the single connections required for proper names, compared with the multiple connections existing to aid in learning and retrieval of common nouns, are more susceptible to transmission deficits (James, 2004). Even when the proper names and common nouns are the same words and only the label (syntactic class) is changed, the retrieval of proper names is still more difficult than common nouns. Aged-related difficulties in learning new names may be explained with this approach as well as where production of homophones of proper names strengthens phonological connections and an increase in the transmission of excitation of nodes and can facilitate the retrieval of proper names (Burke et al., 2004).
Figure 4: Examples of representations in the semantic, phonological and visual concept systems for a person with the proper name Baker a), and with the occupation baker b) (taken from Burke et al., 1991, p. 571).