SE367 (Paper Review) "Semantic structure in improvised communication" - Marieke Schouwstra Q. What is the question addressed? Vidur Kumar (Y8560) Ans. The question addressed in the given paper, is regarding the structure in language. The 3 basic components of any simple communication task are classifiable as : 1. The Actor(Ar) / Subject (S) 2. The Patient(P) / Object (O) 3. The Action(A) / Verb (V) Example - The rat ate the cheese. (English) (predominantly an SVO language) - Ram ne machchli pakkdi. (Hindi) (predominantly an SOV language) Now, when a language is developing (or an improvised language requirement is forced) - and there is no completely defined syntax for the language - then, is there a preferential ordering of these components? And if so, then what are the rules for the structuring? Does it depend on the semantic content to be communicated? And alternatively, does the ordering affect interpretation of communication? These are the questions addressed by this field of study, and the research paper being presented. Table 1: Summary of Frequency of a particular ordering in Speech vs. Gestured communication by test-subjects. (Ref 1) (NOTE - these results were obtained with the 'events' being different kinds of 'motion events' - the type that children learn to 
talk about in the early stages of learning language) From Previous Studies : It has been demonstrated in earlier studies, that for simple communication tasks - mostly involving 'motion events' - the dominant structure in communication using non-verbal (gesturing) language - is SOV (as shown in the table above) (Ref 1) Other studies have also shown that the SOV structure is preferred in comprehension tasks, using gesture-based non-verbal communication. (Ref 2) And it has also been hypothesized, that the SOV order is not simply a dominant order by virtue of a communicative function - but might be a consequence of a fundamentally 'natural cognitive representation of events'. This argument is based on the intuitive reasoning that - persons/objects are simpler and less complex, and hence are represented before whatever action/event is occurring between them. (Ref 1) (This study was done via a stacking-of-transparencies-task, which did not involve communicative requirements, nor consistency requirements, in ordering of a transparency sheets (which had the whole 'event', that was to be described, in parts of the components S, V, O drawn on each transparency - illustration given below). Fig.1 : Transparency stacking experiment (Ref 1) However, this research article attempts to prove that the structure in the improvised communication is not independent of the semantic meaning trying to be conveyed - and that the semantic will affect the structure of the improvised language based communication task. Methods and Results : Using volunteers with Dutch as their native language, and no prior experience in sign-languages - the studies were carried out using two different kinds of events that required describing via gestures. The two types of events differed in their verbs - as being either "extensional" or "intensional". Extensional verbs being those which require the object of their 'doing' to exist, and obviously the 'do-er' to exist as well. Intensional verbs however, must have a 'do-er', but the object of the 'verb' may or may not be an existing entity - and it's existence is infact, secondary to 'intension' of the verb, as ascribed to the 'do-er'. Examples of intensional verbs are - "want", "seek", "admire", etc. Which differ in their requirements and specificity of 'objects', as compared to extensional verbs such as - "hit", "pick", etc. The volunteers were asked to describe certain events (shown to them as pictures of the event), which could be either 'motion events' or 'intensional events', and their gestures were taped and transcribed to establish the predominance of a given structure in the communication task. (Called the 'production experiment). They were later asked to describe the same picture in a Dutch sentence. Fig 2 : Results of the production experiment (Reviewed Paper) Interpretation Study: To study if the structuring/ordering of gestures can affect the interpretation of the communication made - the volunteers were shown video clips of gestures, and asked to describe them in a sentence - with the verb-component of the clips being designed as ambiguous (interpretable as intensional or extensional - example given below). Fig 3: Ambiguous action. Interpretable as 'build' or 'climb'. (Reviewed Paper) The two groups of volunteers were shown clips with a different ordering of gestures (S-O-V') and (S-V'-O), and the interpreted results were as follows. (V' = ambiguous verb clip) Fig 4: Results of Interpretation Experiment. (Reviewed Paper) The above set of experiments revealed that the SVO ordering did lead to a greater probability of interpretation of the event as being an 'intensional even'. Discussion : The above study's results corroborated the previous experiments, in terms of re-emphasizing the dominance of SOV ordering in the production and interpretation of 'motion/extensional events'. However, it illustrates that the original hypothesis - which neglected the contribution of semantics, in the structuring of non-verbal communication (and in general, simple-language development) - is wrong, and requires re-analysis. The inference of this study (as illustrated in the title of the paper) cannot be denied - that the semantic meaning of the communication to be made, will affect the structure of communication (when using an improvised language). Hence, in utilization of languages without complete syntax-structure - the semantics of the communication, will affect the organization of the language components. And also, varying the structure of the components would lead to differences in interpretation (given the syntax of the language used is ill-defined). Quite possibly, these results could be tested for other languages/communication methods, which have varying degrees of syntax- structure - to further understand the role of semantics in structure of a language. References : (1) Goldin-Meadow, S., So, W. C., O¨ zyu¨rek, A., & Mylander, C. (2008). The natural order of events: How speakers of different languages represent events nonverbally. PNAS, 105(27), 9163–9168. (2) Langus, A., & Nespor, M. (2010). cognitive systems struggling for word order. Cognitive Psychology, 60(4).
