|
Summer 99, Final |
Part 1: Multiple Choice. Circle the letter corresponding to the correct answer. Only one answer is correct for each question. (1 point each)1. In light of McKoon & Ratcliff's (1980) results, if experimental subjects are
presented with the story, "The hunter fired the shotgun. The duck fell into the
water. The dog retrieved the duck. The hunter rewarded the dog." we would expect the
word "hunter" to prime (i.e. facilitate) recognition of "duck" _____
"hunter" would prime "dog". 2. If Rumelhart and McClelland (1982) had found that letters are just as difficult to
recognize in unpronounceable non-words that share letter combinations with real words
(e.g., "SLNT") as they are in unpronounceable non-words that don't share letter
combinations with real words (e.g., "XLQJ"), then _____ would have been
falsified. 3. MacWhinney, Bates, and Kliegl's (1984) animacy strategy suggests that the agent of
the pseudo-sentence _____ is _____. 4. In McClelland & Rumelhart's (1981) Interactive Activation Model, a node's
activation level at time t+1 is determined by _____ at time t. 5. Patients with surface dyslexia have no problem with regular words (e.g.,
"like" and "cane") but are unable to read irregular words (e.g.,
"said" and "lose"). This suggests that they have lost the _____ route
proposed by _____ models of reading (e.g., Coltheart, 1978). 6. Because McCloskey and Glucksberg (1979) found that the presence of negative
sentences with high feature overlap (e.g., "A bat is a bird.") does not
attenuate the semantic distance (or typicality) effect, _____ model of semantic memory was
disconfirmed. 7.The goal of _____ is to formulate a small set of rules that can generate all the
grammatical sentences of a language, and no non-sentences. 8. If Slobin (1966) had found that semantically constrained (i.e., nonreversible)
passive sentences (e.g., "The bagel was eaten by the boy.") take longer to
understand than their active equivalents ("The boy ate the bagel."), then _____
would have been supported. 9. Which of the following models explicitly assumes that the structures and processes
underlying language comprehension are independent of the structures and processes that
mediate other cognitive abilities? 10. Seidenberg & McClelland's (1989) parallel distributed processing model of word
recognition and naming predicts that orthographic-phonological regularity will have _____
effect on naming times for words with small neighborhoods as those observed with words
that have large neighborhoods. Part 2: Definitions. In just 1 or 2 sentences, give an operational definition for each of the following concepts. Your definition may come from an experiment you are familiar with or you may make up your own definition (as long as it accurately defines the concept and is operational). (2 points each)Grading Criteria:
11. The Word Superiority Effect To demonstrate the Word Superiority Effect, we could ask people to look at a fixation point on a computer screen then name, as quickly as possible, letters that appear in its place. The Word Superiority Effect can then be defined as the difference in letter naming times (the time that elapses between presenting a letter and detecting the naming response) between letters presented alone (e.g., "O") and letters presented in the context of a word (e.g., "DOG"). 12. Reversible versus Nonreversible Sentences To determine if sentences are reversible or nonreversible I would present them to a group of participants and ask them to judge whether they still make sense when the agent and object are reversed (e.g., "The boy ate the bagel." --> "The bagel ate the boy."). Those for which a majority responded "yes" could then be classified as reversible while those for which a majority responded "no" could be classified as nonreversible. 13. Psychological Distance Between Two Events in a Story To measure the psychological distance between two events in a story I would present the story to a group of participants along with several other stories then ask them to make speeded true/false judgements by pushing a TRUE or FALSE button as quickly as possible in response to test sentences presented on a computer screen. I would measure the reaction time for the second of the two events under two conditions, when it is preceeded by a the first event and when it is preceeded by an event from another story, and use the difference between these two times as my measure of the psychological distance between the two events. 14. The Neighborhood of a Word The neighborhood of any word can be defined as all the words in the latest edition of Webster's Dictonary that that differ from that word by just one or two letters and have the same number of syllables. Thus, the neighborhood of the word "bike" would include words like "hike," "like" and "pike." 15. The Codability of Colors I would present colored squares to participants on a computer monitor and ask them to generate a name for each color as quickly as possible. The average naming latency (as measured by a computer) could be used to measure codability. Part 3: Short Essay. Answer each of the following questions in half a page or less. Be sure to use operational definitions and/or examples were they are appropriate. (5 points each16.Describe the difference between a sequential and a cascaded process and give an example of each from theoretical models in the psychology of language. Grading Criteria:
Example Answer: Both sequential and cascased processing involve two or more levels or stages of processing in which stage n begins before stage n+1. The difference is that in a sequential process, stage n must also be completed before stage n+1 can begin while in a cascaded process stage n continues after stage n+1 begins. A clear example of a sequential process in the psychology of language is Forster's (1981) model of lexical access and sentence understanding. In this model lexical processing must be completed before syntactic processing can begin, and syntactic processing must be completed befroe semantic processing begins. An example of a cascaded process is PANDEMONIUM in which feature processing begins prior to letter processing, but then continues in parallel with it. 17. Design an experiment to test the claim that words are easier to recognise when thay are presented in a meaningful context. Be sure to describe your independent and dependent variables, using operational definitions and/or examples where they are appropriate. What pattern of results would you expect if the claim is true? What pattern of results would you expect if the claim is false? Grading Criteria:
Example Answer: To test this claim I would ask two groups of participants to name out loud as quickly as possible a list of 20 words. For one group, each word would be presented alone following a fixation point which would be presented for 3 seconds (e.g., "+" followed by "CAT" ). For the other group, each word would be presented in the context of a meaningful sentence. The rest of the sentence would be presented first with the target word replaced by a blank. After 3 seconds, the target word would replace the blank in the sentence (e.g., "The dog chased the ___." followed by "The dog chased the CAT."). The independent variable in this experiment would be the presentation context (alone versus in a sentence context). The dependent variable would be the time that elapses between the presentation of a word and when a voice key detects that the participant has begun to pronounce its name. If the claim is true, I would expect words to be named faster when they are presented in sentence contexts. If the claim is false, I would expect to find no difference between words presented alone and those presented in a sentence context. 18. Compare McClelland & Rumelhart's (1981) Interactive Activation Model to Seidenberg & McClelland (1989) model of word recognition and naming. Include in your comparison at least one way in which their representational assumptions are the same, one way in which their representational assumptions are different,one way in which their processing assumptions are the same, and one way in which their processing assumptions are different Grading Criteria:
Example Answer: McClelland & Rumelhart's (1981) Interactive Activation model takes as input a representation of the visual features that make up a word and provides as output the identity of the word itself. Spreading activation and inhibition are used to accolish this goal. It includes three levels of representation: visual features, letters, and words. All information in this model is represented locally. As a result, the processing that it performs is symbolic. Seidenberg & McClelland's (1989) model of word recognition and naming takes a representation of the visual features that make up a word as input but provides as output a set of features specifying the sound of the word. Like the earlier model, it uses spreading activation and inhibition to accomplish its goal. Also like the earlier model it uses three levels of representation, but those levels are orthographic features, a hiden layer, and phonological features. This model uses distributed representations. As a result, the processing it carries out is subsymbolic. The most obvious similarities between these two models are their use of a connected network of nodes to represent information about words along with speading activation and inhibition to processes that information. The most important differences are the use of local representations paired with symbolic processes in McClelland & Rumelhart compared to distributed representations paired with subsymbolic processes by Seidenberg & McClelland. 19. Previous research has shown that the more active the meaning of a word is the longer it takes to name the color of the word. For example, if participants are shown the word "cat" printed in blue ink and required to respond by saying "blue" response times will be longer if the preceeding test word is "dog" ( a close semantic associate which is likely to increase the activation level of the concept cat) then if the preceeding word is "pen". Design an experiment which uses this Stroop-like effect to test the claim that that a reader's attention will always be drawn to the state or event which is causally last in a narrative. Be sure to describe your independent and dependent variable, using operational definitions and/or examples where they are appropriate. What pattern of results would you expect if the claim is true? What pattern of results would you expect if the claim is false? Grading Criteria:
Example Answer: For this experiment, I would use the materials developed by Fletcher, Hummel & Marsolek (1990). These materials consit of a set of narratives that each occur in two versions. In one version (the Antecedent Version) a critical sentence (e.g., "Kate's sister came home and told her that the oven was broken.") is followed by a causal antecedent ("She had tried to use the oven earlier and discovered that it would not heat up."). In the other version (the Consequence Version), the same critical sentence is followed by a causal consequence ("Since she had the batter all ready, Kate decided to borrow the neighbor's oven."). If a reader's attention is always drawn to the information that is causally last, the critical sentence should remain the focus of attention after reading the casual antecedent, but not after reading the causal consequence. To test this claim, I would present stories of each type to a group of participants. They would be allowed to read at their own speed, one sentence at a time, using a button press to request each new sentence. After reading either the antecedent or consequence setnence, they would see a probe word from the critical sentence (e.g., "TOLD") instead of the next sentence in the narrative. This word would be presented in either red, green, blue, yellow, orange, or purple on a white background. All other text would be presented in black on a white background. Participants would required to respond to this probe word by naming its color as quickly as possible. The time that elapses between the presentation of the word, and the detection of a spoken response by a voice key attached to a microphone would be the dependent variable in this study. The two versions of the stories (Antecedent versus Consequence) would be the independent variable. If the claim is ture, I would expect naming times to be slower in the Antecedent versions than in the Consequence Versions because attention is still focused on the sentence containg the probe word. If the claim is false, I would expect to find no difference between the Antecedent and Consequence Versions of the narratives. |
|