The mental representation of the number’s verbal structure
Read the full dissertation (in Hebrew) or the article based on Nadin’s dissertation
How do we represent multi-digit verbal numbers? My MA thesis addressed the question of whether we represent the syntactic structure of the verbal numbers, and examined what are the characteristics of this representation are. Recent studies have shown the existence of syntactic number representation, but there are still gaps regarding the specific characteristics of that mental representation.
We conducted three different experiments using the syntactic chunking paradigm – serial recall of a sequence of number words. The experiments consisted of several almost-identical conditions, with the only difference between the conditions being the word order within each sequence. Each item was a sequence of a number of words; it could be a syntactic sequence (e.g., four hundred and fifty-six) or a non-syntactic sequence (six fifty-four hundred).
The participants remembered syntactic sequences better than non-syntactic ones, a finding that indicates the existence of the syntactic mechanism that allowed them to merge the words of a syntactic sequence into a chunk in working memory, thus improving the level of memorization. The performance was the best in the sequences with two syntactic segments (e.g., one thousand seven hundred and twenty / forty eight), with lower performance in sequences of one syntactic segment (forty-eight thousand seven hundred and twenty). These finding informs about the syntactic representation — specifically, that the participants represented cross-triplet syntactic structures: although single-chunk cross-triplet representation created load on working memory, they still automatically represented the largest syntactic structure that the stimulus allowed.
In two follow-up experiments, we examined: (a) whether the better memorization of syntactic sequences was due to strategies of memory and prior knowledge (e.g. knowledge of categorical order); and (b) whether the syntactic mechanism is efficient and flexible enough to represent different structures that change frequently. Experiment 2 showed a syntactic chucking effect even when we provided cues (e.g., using intonation) as for how to group the words into chunks. This shows that the chunking effect stems from the ability to compress the information by representing the syntactic structure of the number, and refutes the possibility that syntax was not represented, but only provided hints as to how to chunk words. Experiment 3 showed a syntactic chunking effect even when the stimuli had various and constantly-changing syntactic structures, which were not included in the previous experiments. This shows the flexibility of the syntactic mechanism, and its ability to represent irregular and variable structures in a short time.
Overall, my MA thesis shows that adults can represent the syntactic structure of numbers, and they use the syntactic mechanism automatically, involuntarily and are able to represent a wide variety of variable structures without explicit instruction. The syntactic mechanism seems to be inherent in the processing of multi-digit verbal numbers, and is robust enough to represent long syntactic structures (6 digits) using a cross-triplet representation. The creation of this syntactic representation is automatic, and the findings cannot be explained as an effect of external strategies.