Homework #2 :: Due by 10:00am on 2/20/2022
1) Read the article
2) On the following pages of this document, find a summary of the research paper written by a student in the class 2 years ago. 5 sentences ( which has already been highlighted ) in this summary contain inaccurate information.
Your task is to
1) compare the contents of the paper and the summary,
2) identify which part of the highlighted information is inaccurate/wrong and correct the inaccuracy. Provide your answer as a numbered list at the end of this answer sheet. ** In your , you MUST cite a part of the original paper with APA 7th style that contains the CORRECT information. A failure to do so will result in a deduction of points.
Summary of “Ferguson, B., Graf, E., & Waxman, S. R. (2014). Infants use known verbs to learn novel nouns: Evidence from 15- and 19-month-olds. Cognition, 131, 139-146.”
Fluent English speakers make inferences guided by a verb’s semantic requirements, also called its selectional restrictions. For instance, “He ate the carambola” licences the inference that carambola must be edible. Ferguson, Graf, and Waxman set out to determine if infants less than two years old could label a novel noun as animate or inanimate based on its later use with a known verb. Previous studies have shown that infants violate these selectional restrictions in their own productions but show understanding by 24 months. However, studies on younger infants that could address their categorization of novel nouns based on verb’s linguistic requirements were lacking.
Ferguson, Graf, and Waxman studied 59 infants (30 19-month olds and 29 15-month olds) who had been exposed to a language other than English a maximum of 25% of the time. A list of known words was generated based on a publicly available database of words used in picture books for young children. There were three phases of each trial. The phase consisted of 6 seconds of a pair of objects displayed side-by-side, with no audio. The Dialogue phase showed an abstract screensaver for 9 seconds, during which a native speaker of American English used infant-directed speech to label one of the images. The final Test phase presented the same pair of objects in the same positions for another 6 seconds while the speaker prompted the infant to find the previously mentioned object. Each pair of objects consisted of an animal and an artefact, with randomized left-right positioning. There were two types of trials: familiar and unfamiliar. In the familiar trials, at least 72% of the 15-month old infants supposedly knew the names of the target objects presented. In the unfamiliar trials, unknown objects (rare animals and abstract sculptures) were presented, with nonce names. The verbs used in this study were known by 66% of the 15-month olds. During the trials, infants sat on their caregiver’s lap while the caregiver wore opaque glasses. An eye-tracking system was calibrated to each infant, then infants were exposed to 12 trails. The first six trials were identical familiar trials across all infants. The final six trials were unfamiliar trials. Infants were randomly assigned to the Informative or Neutral condition. The only difference between these conditions occurred during the Dialogue phase. In the Neutral condition, the sentence was expected to bias infants’ case to an animate referent (e.g. The wog is right here) while it was not the case in the Informative condition (e.g. The wog is eating). All labels in the familiar trials were neutral.
Ferguson, Graf, and Waxman predicted that infants would prefer artefacts in general. Familiar trials should prompt infants to match the name with its appropriate image to support the design of the experiment. However, they also predicted that infants would look at the artefact significantly less in the Informative condition than in the Neutral condition of the unfamiliar trials if they were able to utilize the verb’s linguistic information.
In the familiar trials, infants preferred the animal image over the artefact object. When prompted with the animal’s name, both age groups looked at the animal image significantly more than the artefact, though this was less pronounced in the 15-month olds. They also looked at the artefact more than the animal when the former was named. In the unfamiliar trials, 15-month olds did not show any statistically significant difference in preference for the animal image between the Neutral and Informative conditions. However, 19-month olds looked at the animal image significantly more in the Neutral condition than the Informative condition, despite an overall preference for the artefact image. No confounding variables were identified.
19-month-olds were able to identify novel nouns based on their knowledge of the used verbs. However,15-month olds were shown to use this knowledge of selectional restrictions only for the nouns used in the familiar trials. They did not use the same knowledge for the novel nouns introduced in the unfamiliar trials in the experiment. These findings help identify infants’ comprehension of the selectional restrictions of their first verbs. During this active period of development, infants begin to use their knowledge of verbs to increase their knowledge of nouns.
There are two possible (not necessarily mutually exclusive) explanations for the difference between 15-month olds and 19-month olds seen in this study. First, infants may have different representations of familiar verbs at these ages; perhaps 15-month olds have not yet incorporated selectional information in their representations of verbs. Future research would have to specify what types of information are in infants’ verb representations and identify how these representations change between 15 and 19 months of age. Alternatively, the limitations that 15-month olds display may not be in their representation of the verbs but rather their ability to process the sentences and objects in time to display their understanding. They may struggle to encode a novel noun while accessing their knowledge of the verb; there is a processing lag even in the 19-month olds. The processing burden would have to be specified more clearly and reduced to test this possibility.