Gernsbacher's Structure Building Framework (1990) provides an experimentally grounded and theoretically rich account of how people arrive at global representations of discourse meaning from the evidence presented in sequentially ordered clauses. Gernsbacher's model explains both what is present and what is absent in the interpretation arrived at, with special attention paid to shifts from substructure to substructure when the language comprehender constructs a global interpretation of input. Gaps in Gernsbacher's handling of some aspects of sentence complexity, such as heavy nominals and thematic contrasts, and in her discussion of some of the competing accounts of sentence comprehending, do not undermine the value of this important contribution to the field of psycholinguistics.
2. In brief, language comprehenders, like experiencers in general, typically move as quickly as possible to integrate perceptual input with the knowledge base they already possess. Along the way, they cast off sensorily exact (eidetic, verbatim) representations, contextually implausible perceptual or interpretive ambiguities or unclarities, and details of the input that do not appear to be relevant to the interpretive outcome as it has been constructed so far.
3. To the extent that the input is referentially, temporally, locationally, and causally coherent, the developing construal will contain much information that is not triggered directly by items in the input but rather incorporated into the construal by the processes of inference inherent in the structure-building process itself. Indeed, the process typically constructs understandings that the language comprehender may not be able to distinguish between: an understanding of the input proper and an understanding of the state of affairs the comprehender infers from input.
4. Gernsbacher delves deeply into the specifics of the Structure Building process. Her account should help to persuade many who are skeptical about the value of psycholinguistic experimentation over intuition as a gauge of what is going on in language (witness my own cautious appeals to intuition in what follows).
5. There are many details of Gernsbacher's argument that deserve exposition and critical commentary. I hope that the book will receive wide discussion. I will, however, limit my comments to just four of these details.
6. Gernsbacher claims that the Advantage of First Mention, whereby first-mentioned sentence participants are remembered better than later-mentioned participants, is independent of whether the first- mentioned participant is semantic agent or patient, sole grammatical subject, first-occurring nominal, or even an argument of the main verb of the sentence. Given the centrality of the mapping versus shifting routines she describes later, and the consequent importance of first- mentioned items as triggers for the suppression or enhancement of the conceptual content of construals, Gernsbacher does well to focus on the relative accessibility of various sentence constituents as a gauge of their relative status as shifters, suppressors, and enhancers. There are, however, other sentence-structural variables that Gernsbacher leaves untested.
a. Is the Advantage of First Mention independent of whether the first-mentioned participant is the head of its constituent? Specifically, will it be Lisa or Tina, or neither, who exhibit the Advantage in the following sentence?
Lisa's sister Tina beat her in the state tennis tournament.
My very faint intuitions tell me that Lisa will NOT exhibit the Advantage in this instance, but if I'm wrong,
b. Is the Advantage independent of whether the first-mentioned person is a semantic argument of the sentence, as it fails to be in the following sentence?
Lisa's sister Tina beat Pam in the state tennis tournament.
In this case my intuitions again tell me that Lisa will NOT exhibit the Advantage, and if I'm right this would help to explain why the nominals in preposed prepositional phrases like the following:
Because of/According to/Compared with/Except for Tina, ...
exhibit the Advantage less than other first-mentioned participants. This in turn raises the question:
c. How strong would the Advantage be for other first-occurring nominals, such as those in the following sentences:
While visiting Tina, Lisa received the good news. During a visit to New York, Lisa hit a few with John McEnroe. According to Tina's sister, Lisa won the contest. Despite Tina's warning, Lisa took only one spare racquet.
Clearly, a great deal more can be done to nail down the variables that determine the relative accessibility of the concepts labeled by sentence constituents.
7. In discussing the communicative function of first mention, Gernsbacher misses an opportunity to resolve the paradox between first mention as a marker of givenness versus first mention as a marker of importance. Specifically, her own Structure Building Framework makes much of the distinction between "mapping" input onto an existing structure versus "shifting" input into a new structure. It would seem that the often hazy distinction between given/new versus important/unimportant can be replaced entirely in Gernsbacher's model: the first-mentioned item may be given or new, important or unimportant, but it is always the foundation upon which subsequent information will be mapped. The judgment that something mentioned first is "new" or "important" can then be explained as an effect of the extent to which that foundational first mention triggers a shift to a new structure.
8. In discussing the question of whether the mechanisms of language comprehension are automatic or under conscious control, Gernsbacher completely omits reference to Jerry Fodor's claims regarding the automaticity of language comprehension (Fodor, 1983). Perhaps she resents giving the devil his due, or hesitates to pick a fight with the whole "modularity" crew, but much of the evidence Fodor cites in support of his position would support Gernsbacher's just as well or better.
9. A number of terminological ambiguities mar Gernsbacher's work. Two of the most grievous surround the terms "referent" and "theme".
a. Gernsbacher introduces unnecessary confusion by using the terms "anaphor/cataphor" and "referent" as if the latter meant BOTH "coreferential discourse constituent" AND "concept labeled by the anaphor, cataphor, or discourse constituent". This usage detracts from the coherence of her argument that construals are conceptual abstractions away from the linguistic constituents of sentences. It also conflicts with the wide use of the term "referent" to refer to external objects rather than internal concepts.
b. Gernsbacher uses the phrase "convey a theme" to mean something like "cohere to a common gist", as when she says that the following two sentences "convey the same theme":
He sent a letter about it to Galileo. A letter about it was sent to Galileo.
While the following sentence "does not convey the same theme":
Galileo sent him a letter about it.
This usage also detracts from Gernsbacher's argument that discourses do not "convey" but rather trigger the construction of meaning. It also conflicts greatly with the use of the term "theme" in many other linguistic contexts (as witness her own reference on page 12). And extensions of the term "theme" lead to such egregiously misleading phrases as "thematic content" in place of "gist".
10. This book makes an important contribution to our growing ability to make sense of how people make sense of linguistic input. The painstaking experimental methodology exemplified here deserves wide discussion and careful emulation.
Fodor, J.A. (1983) The modularity of mind: an essay on faculty psychology. A Bradford Book, MIT Press.
Gernsbacher, M.A. (1992) Precis of: Language Comprehension as Structure Building. PSYCOLOQUY 3(69) language-comprehension.1.gernsbacher.
Gernsbacher, M.A. (1990) Language Comprehension as Structure Building. Hillsdale NJ: Lawrence Erlbaum.