limits the choice of the next word or phrase in such a way that results in a
sufficiently close reproduction of the text (cue-item discriminability). Con-
straints include imagery, theme, rhyme, alliteration, rhythm, and music. At the
beginning of the song, genre-specific constraints provide initial cues. The“sing-
er”starts out with an initial word or phrase, rhythm, or melody, and follows the
various constraints, often implicit and subconscious ones, to produce the next
word, phrase, or line, until the end of the text is reached. Rubin puts particular
emphasis on the local nature of cuing, as opposed to relying on an overall
cognitive template. He quotes interesting examples of how particular details of
a text are accessible even to experts only after a“running start.”For example, an
exorcist on Sri Lanka, when asked to give information on a particular demon,
suggested the following procedure:“I will sing it and you tell me when the demon
you want has his name mentioned. Then I will go slow so that you can put it onto
tape recorder”(Rubin, 1995, p. 190).
Rubin observes that there is a group of texts, which he calls“sacred texts,”
that are supposed to be recalled verbatim (p. 181). When recalling the
Preamble of the American Constitution or Psalm 23, for example, the use of
synonyms, substitute words, or embellishments is not acceptable. The ambi-
tion to memorize some texts as accurately as possible also existed in antiquity.
Rhetoricians developed techniques that enabled them to memorize and deliver
speeches with great accuracy (cf. Quintilian,Institutio oratoria11.2), and
techniques of memorization are known from rabbinical literature, including
intonation, chunking (organization into smaller units), and markers, such as
sequences of letters or numbers (Bowker, 1969, pp. 50–3). In spite of the
intention to transmit“sacred texts”faithfully, however, they are often recited
with typical errors, which go unnoticed by the performer, and sometimes even
by the listeners, probably because they fulfill various other constraints. As
experimental results about confidence in memories as well as the ethnographic
records of Lord and Parry regarding the notion of“verbatim”performance
among professional performers demonstrate, claims to accuracy are not to be
confused with the actual accuracy of the memories and the performance
(Rubin, 1995, pp. 6–7).
Serial recall might seem to be at odds with script theory, atfirst sight. While
the theory of cognitive schemata emphasizes the overall organization of
memories, serial recall relies on local organizing principles. So how do we
remember stories, after all? Do we store in our memory narratives with a
logical structure, so that we can tell what happens with the hero in the end,
even if we do not tell all the details about the middle part of the story? Or do
we store a chain of words and sentences, so that we have to go down the chain
in a linear fashion if we have to retrieve information about some detail?
Obviously, we can tell that any sick person in a healing story will be healed
in the end, without spelling out every detail of the narrative. If we have to recall
the exact words of a character in a story, however, we tend to search in our
Memory and Transmission 75