Grammatical Subjects in home sign: Abstract linguistic structure in adult primary gesture systems without linguistic input

Coppola and Newport. 10.1073/pnas.0509306102.

Supporting Materials and Methods

Files in this Data Supplement:

Supporting Materials and Methods

Study 1: Coding and Analysis of Responses. For each item, the first response produced by the participant that contained a complete target clause was analyzed. Within a response, clauses were defined by using prosodic criteria standardly used in sign language research. These criteria included pauses, eye gaze, holds, head position, head nods, and the rhythm of gesture production. Intercoder reliability for clause boundaries was 0.90 across participants. The target clause was defined as the clause containing gestures referring to both the primary argument and the action or event depicted in an item. For example, in the Agent-Patient item "Man kisses woman," the target clause must contain a gesture referring to the man (the Agent and the primary argument) and a gesture referring to the action "kiss." Responses that did not include any clauses containing both of these elements could not be coded for the devices under examination and were therefore excluded from the analyses, resulting in 50 items for home signer 1, 54 items for home signer 2, and 61 items for home signer 3 (of 66 possible items). Of home signer 1’s target clauses, 47/50 consisted of the entire response; that is, no clause was selected from the response as a whole. The proportions of target clauses that were the entire response for home signers 2 and 3 were 34/54 and 30/61, respectively.

Within the target clause, noun phrases (expressions referring to the people and objects of the target event) were coded for word order position. In this coding, the order of the argument noun phrases and the action gestures were the primary gestures considered. Thus, for example, the noun phrase was coded as occurring in clause-initial position if the gesture was the first noun or noun phrase, regardless of whether it was preceded by other nonargument gestures [for example, gestures indicating the location of the event (like adverbs) or referring to an attribute of an object (like adjectives)].

To ensure that the stimulus materials elicited the semantic roles we intended and would elicit the expected grammatical devices in a fully developed language, we also tested a control group of eight hearing native English speakers by using the same materials and procedure, but instead, we asked them to respond in English. For each item, we noted whether the English speakers described the event by using the semantic roles we had designated and whether they marked the primary argument as the grammatical Subject. Forty-six items (17 one-argument and 29 two-argument items) perfectly matched our expectations, eliciting subject marking in at least seven of the eight English speakers’ responses in a language (English) in which we know there is a grammatical category Subject. The remaining items were described appropriately by most English speakers but were described in a slightly different way by some. For example, some used a different verb to describe a particular scene, resulting in a different choice of Subject (e.g., "The ball falls," rather than, "The woman drops the ball"). Out of the 46 matching items, 19 had primary arguments that were Agents and 27 had non-Agent primary arguments. We then conducted an analysis of the home signers’ data on only these 46 items, which produced essentially the same results as in the main analysis: for all three home signers, virtually all of the primary arguments were produced in clause-initial position, for both Agents and non-Agents.

Study 2: Coding and Analysis of Responses. As in Study 1, the first response containing the requisite target clause was analyzed. Following the coding procedure for Study 1, clauses were defined by using prosodic criteria standardly used in sign language research. In this study, however, a response had to contain two complete target clauses: the first had to include an explicit expression referring to Character 1 and the action of the initial Set-up event, and the second had to include an explicit expression referring to Character 2 and the action of the final Test event. Responses that did not meet these criteria could not be coded for whether Characters 1 and 2 were marked with the same word order (or other syntactic marking) and were therefore excluded from analysis. For the three home signers, two, six, and six responses to Test items met this criterion (of eight possible Test items). (In the remaining items, one or both characters were referred to only implicitly through spatial modulations on action gestures or with an ambiguous point at an unlabeled space.) The type of device used to mark each character was coded: lexical noun (e.g., man or woman), phrase, pose, point, or spatial modulation. The word order position of the device was also coded as in Study 1.

Also as in Study 1, the same materials and procedure were used with a control group of eight hearing native English speakers asked to respond in English to an English-speaking listener to ensure that the materials elicited the expected outcome in a well developed language known to contain linguistic devices marking subject and topic. All eight English speakers produced the expected devices in virtually all items. They marked both Character 1 in the initial Set-up event and Character 2 in the final Test event as the primary argument (Subject) of the clause in all but one subevent (63/64). In addition, they used pronouns to refer to Character 1 in events subsequent to the initial Set-up event, indicating that they regarded Character 1 as "old" information (Topic). These results showed that the vignettes effectively conveyed their intended event structures and also elicited appropriate devices for marking Subject and Topic in a language that contained such devices. Thus, all Test items were considered eligible for the home sign analysis.

• Early Edition
• Archives
• Online Submission

• Feature Articles

  (Expanded research articles of exceptional breadth)

• Commentaries

  (Expert commentary on leading research)

• Letters

  (Online-only letters to the editor)

• Inaugural Articles

  (Articles by recently elected NAS members)

• PNAS Profiles

  (Biographical profiles of Academy members)

• Sustainability Science

  (Interactions of human and environmental systems)

• Collected Papers

  (Editorials, Perspectives, Reviews, Colloquia, etc.)

• Special Features

  (Solicited articles on exceptional research topics)

• Sackler Colloquia

  (Scientific reports held under NAS auspices)