Short answer
Paraphrasing indeed negates the spacing effect observed in learning, because it causes mass repetition to be as effective as spaced repetition. The confusion is that paraphrasing does not degrade spaced learning, instead it improves the effectiveness of learning through mass repetition up until the level of spaced repetition.
Background
Basically the question is why
The spacing effect fails to occur with paraphrased repetitions
Krug et al. (1990) define the spacing effect as
...[T]he phenomenon in which material encountered on two different occasions with a lapse of time between the encounters is remembered better than material studied for an equal amount of time on one occasion.
Your cited text from Dempster, 1998) reads:
Thus, this account helps to explain failures to obtain the spacing effect with paraphrased repetitions, that is, repetitions having a changed surface structure (Dellarosa & Bourne, 1985; Glover & Corkill, 1987).
Paraphrasing means
[Restating] a text, passage, or work giving the meaning in another form
Now why does this fail to induce the spacing effect? The author cites two papers. The first from Dellarosa & Bourne (1985) states
On the basis of previous research, it was assumed that memory for surface structure of sentences decays rapidly, and hence can contribute to initial identification of repetitions only at short spacings. Because this manipulation
should hinder recognition of repetitions as repetitions, it was expected to induce full processing of massed repetitions, and thus facilitate recall of these items.
Why is this important? The authors believe that processing is key to the learning of verbal information. They claim that mass repetition, defined as
[M]aterial studied for an equal amount of time [as spaced repetition] on one occasion
Spaced repetition is believed to be more effective, because the information upon repeated exposures is all fully processed, because it is not recognized by the brain anymore as a full repeat. In contrast, mass repetition leads to incomplete processing during repeated exposures, and hence is not processed fully anymore, because the brain recognizes it as an exact repeat of previously handles material Dellarosa & Bourne (1985). The surface structure of sentences is defined as
[The] representation of a string of words or morphemes as they occur in a sentence, together with labels and brackets that represent syntactic structure.
Changing the surface structure disrupts the brain from recognizing it as an exact repeat. Dellarosa & Bourne (1985) hypothesized that changing the surface structure might prevent the brain from recognizing the same material as exact repeats of information in mass repetition, and thus that learning would become equally effective in mass repetitions. And indeed, their experiments showed that
When sentences were repeated verbatim [] or by the same speaker [], the typical spacing effect was obtained. However, when the surface structure or speaker changed at time of repetition, massed repetitions were recalled nearly as well, or as well as their spaced counterparts.
Subsequently, Glover & Corkill (1987), the second paper Dempster cites, extended these findings to written text, showing the same effect, namely that changing surface structure of texts negated the spacing effect.
Your quote from Dempster, 1998)
The spacing effect fails to occur with paraphrased repetitions
says the same thing, albeit in a condensed and indeed less clear form. It took me quite some detective work to solve this obscure sentence indeed!! Very nice question.
References
- Dempster, American Psychologist (1998); 43(8): 627-34
- Glover & Corkill, J Educational Psychol (1987); 79(2): 198-19
- Krug et al. J Educational Psychol (1990); 82(2): 366-71
