Empirical Methods in Language Studies

Part One: Experimental and Survey Methods

← 11 | 12 →

← 12 | 13 →

Janusz Badio
University of Łódź, Poland

Events and sentences in story construal of English-native and Polish–foreign language users: experimental methodology and outcomes

Abstract: The article presents an experimental study and its three constituent analyses of narrative events and their construal-coding in sentence format of a story. The theoretical position that provides the starting point for the research questions posed in this work is traced down to Chafe (1994), who concluded that, especially in writing, the sentence can be understood as a centre of interest with scope often much larger than a single focus of consciousness. The author investigates how Polish-native and English-foreign language users construe and code events presented in a children’s cartoon story. The article presents the materials, questions, descriptive statistics, hypotheses, and inferential tests together with some follow-up qualitative observations of the written interpretations of the story. In general, the findings support the more general hypothesis (cf. Badio 2014) that the subordinate goal of construal and coding operations is to properly adjust the salience of entities (e.g., objects, characters, locations, reified processes) and relations (e.g., features of objects, processes). It is, however, the methods used in this study that are highlighted, in agreement with the general theme of the present volume, which is empirical methodology in studying language.

Keywords: story, event, construal, coding, ANOVA, Chi-square, t-test, sentence.

1. Introduction

This work describes a part of a larger project (Badio 2014) on construal and coding of narrative events. Its foremost hypothesis stated that achieving cognitive prominence of selected entities is the language users’ superordinate goal. The empirical studies in this work looked at ways events are perceived and delineated from a stream of video, whether their linguistic coding is isomorphic to their perception and memory, or how they are selected to construe the causal structure of a story neutrally or in such a way as to blame one participant for what has happened. Events’ natural habitat is a story, or a narrative, that is, larger discourse coherence. Construal operations refer both to the level of a single event and discourse. It is ← 13 | 14 → the discourse level, the whole story level, which constitutes the research focus presented in this article. Three related general research questions were asked:

• a)  How will the participants construe and code a cartoon story presented in the form of a silent video, as opposed to a static sequence of pictures (screenshots) from the same video?
• b)  What sentence types are preferably used by native-Polish and English-foreign language users?
• c)  How will the participants in the study package events and their parts in sentence format?

As already mentioned, the participants (N = 30) retold a cartoon story in writing. Data collected in this way were subsequently coded in the format of a Microsoft Access dBase with three tables: SENTENCES, PARTICIPANTS, and CONDITIONS. The first of these tables contained such descriptors of the sentence unit as: the actual sentence, type (clause, coordinate, subordinate, complex), number of events coded in a sentence, and other comments. The table holding information about the participants contained only the number that uniquely identified them, whereas the conditions table included all the conditions (independent variables), in which the story descriptions took place. Importantly, the main sentence table was related to the participants table with the relation one-to-many, i.e., one participant was referred to all the sentences s/he produced in writing, and the same was true of the relation between the condition and sentence tables, i.e., one experimental condition stored all and only those sentences that were produced in it. The format of a relational dBase allowed multiple ways of accessing the created corpus of texts.

Before the presentation of the actual study with its analyses, some theoretical background is offered in section two.

2. Theoretical Background

Construal constitutes the heart of Cognitive Linguistics’ theorizing about language. Its theoretical foundations were laid by Langacker (e.g. 1987, 2008), Talmy (2000 and earlier), or Croft and Cruse (2004). Perhaps its most influential definition was offered by Langacker (ibid.), who claimed that construal is a relation between the conceptualizer and the object that is conceptualized. This relation is described in the works of the above linguists under different, more detailed categories of: selection, perspective, abstraction, prominence, specificity (Langacker ibid.), schematization, perspective, attention, force dynamics (Talmy ibid.), and attention, judgment/comparison, perspective, constitution/gestalt, force dynamics (Croft ← 14 | 15 → and Cruse ibid.). Despite the differences in categorizing construal operations, their underlying common denominator is the belief that it is the so-called cognitive salience/prominence, under the control of attention and its multiple strands¹ that is the super-ordinate goal of any human attempt to construe some content and linguistically code it.

An event, one of the key units, is defined here after Zacks and Tversky (2001: 7) to be a “segment of time at a given location perceived by an observer to have a beginning and end”. Its perceptions are correlated with specific linguistic coding in different languages. Moreover, events, by analogy to physical objects, contain parts, and they also come in variable types. They frequently involve some evolution through time of a certain state of an object, or a set of objects in the case of complex scenes, and they are typically, though not necessarily, coded by verbs in English (e.g. He jumped v.s. He gave a jump). Any fragment of the conceptual content of an event (time, space, manner, etc.) can, in theory, be focused on and coded with an appropriate linguistic unit.

The sentence is also important for this work because after Chafe (1994), it is understood to code the so-called centre of interest, a human attempt to expand the limited scope of the active focus of conscious experience. The following discourse exchange (diacritics provided JB)² illustrates the idea.

• 1)  a) .. I was on the ^bus ^toda=y,

• b) .. ^a=nd there was this w^oman s^aying,
• c) .. that her s^on,
• d) .. ^works .. for the r^anger s^ervice or whatever.
• e) .. and .. there was sn^o=w,
• f) .. ch^est high,
• g) .. at Tu^olome M^eadows. Chafe (1994)

In speech, this fragment consists of [a-g] intonation units, each unit coding the content of a separate, changing focus of attention. However, units [a-d], and [e-g] form larger coherences that could be coded by only one sentence in writing each, I was on a bus today, and there was this woman saying that her son works for the ranger service, and There was snow chest high at Tuolome Meadows, accordingly. ← 15 | 16 →

Moreover, it is important to explain that the story retellings used in this study come from a native form of language (Polish), as well as a foreign language (English). With interest on selected aspects of processing, rather than formal aspects of grammar and vocabulary, this was possible following the so called psychological tradition in SLA (cf. Hulstijn (2007: 197). While the linguistic tradition places emphasis on linguistic representations of knowledge, the psychological tradition, which this paper follows, accentuates the question of how learners process information in a foreign language. In agreement with the psychological tradition, the present work argues that a foreign language is an autonomous and legitimate object of research, though comparisons to a native language are expected and natural. Such a perspective is at least partly concomitant with research on second language variability, some aspects of transfer, cognitive descriptions of second language production, status of knowledge, and cognitive control over access to the first/second language, which also Ellis (2008) discusses.

Having introduced the key concepts of this study, i.e.: construal, coding, event and sentence, the author will proceed to the next part, where the materials, procedures and specific hypotheses together with the remaining, and more detailed theoretical points, will be explained.

3. Materials and procedures

The video selected for this study is the Bolek and Lolek: First day of summer holidays³, a popular children’s silent cartoon in Poland. The choice of a cartoon story was dictated by its accessibility on internet, but also the observation that it shows clearly the characters and their actions. Only the main characters are shown without unnecessary clatter, which is another advantage, and they are contrasted well against the background. The actions performed by the boys are easy to interpret, and the plot is simple to follow.

In one group the viewers were asked to watch the video with sound off in the classroom on TV, while the participants of the experiment in the other group were only shown the screenshots from the video arranged chronologically. Both groups first spent the same amount of time viewing their respective inputs, and then they had to retell the story in writing. All participants were granted fifteen minutes to finish this writing task. The instructions provided on their worksheets in Polish were translated by the author into English, and they are presented below: ← 16 | 17 →

Stage 1 – watching

You will watch a silent video, 8 minutes and 58 seconds long. Watch carefully. After watching only once, please write in Polish [the other group, “write in English”] what happened in the story. The story must be written in the 3^rd person. How long the written retelling should be, and what to include depends on your choice. The point is to recreate in writing the events presented in the film.

Stage 2 – writing

Write the story. You have 15 minutes to complete the task, and there is no word limit.

The instructions for those who watched the screenshots from the same film were as follows:

Stage 1 – watching

You will watch screenshots from a silent video, 8 minutes and 58 seconds long. You will have 8 minutes and 58 seconds to look at the pictures arranged chronologically from left to right and top to bottom. After watching the pictures, please write in Polish [the other group, “write in English”] what happened in the story. The story must be written in the 3^rd person. How long the written retelling should be, and what to include depends on your choice. The point is to recreate in writing the events presented in the film.

Stage 2 – writing

Write the story. You have 15 minutes to complete the task, and there is no word limit.

The participants of the study were all advanced / proficient students of English as a foreign language, all during their first or second year at M.A. level of English Philology in Łódź, Poland, and they agreed voluntarily to take part in the experiment for partial credit on their seminar course.

4. Analysis one – sentences in story retellings

This analysis, i.e., the first part of the cartoon story retellings experiment, had the goal of answering the following research question. Will speakers of Polish-native and English-foreign languages code events presented by either a video or picture sequence differently? Will the four independent groups (i.e., PL-Vid, PL-Pic, Eng-Vid, Eng-Pic) differ as regards the number of sentences they will decide to code in writing. As one can see, apart from language, the other independent variable was input type (continuous/dynamic-video or static-pictures). It was predicted ← 17 | 18 → (cf. Skehan 1998) that different task types could affect language production. Here, the choice of continuous video rather than the static sequence of screenshots from the same video was predicted to result in a larger number of centres of interest, and thus also sentences because, in theory, there is no limit to what can attract the viewers’ attention during a continuous stream of video. Also, it was predicted that the Polish-native language users would produce more sentences (centres of interest), as one can expect, the use of one’s native language should pose fewer language coding problems.

A little repetitively perhaps, the following hypotheses were formulated as regards the two experimental conditions:

a) Video-Pictures

b) Polish-English languages

c) Interaction
Moreover, it was predicted that there may be interaction between the choice of language and input type, such that the choice of the native language coupled with the video input should lead to a greater number of sentences in the written interpretations of the input story.

Hence, the whole design was an ANOVA 2x2 because two independent variables were tested, language and input type, each variable with two levels represented by the choice of either Polish or English, and Video or Static Pictures (screenshots from the same video).

The analysis of descriptive statistics showed that the largest mean number of sentences was produced in the PL-Vid condition, followed by Eng-Pic, Eng-Vid and last, PL-Pic conditions. It was also apparent that the Polish language condition rendered more sentences per retelling (M-PL = 16.75) compared to the choice of English (M-Eng = 13.71). The comparison between the video and picture conditions showed that the former (video) condition invoked more (M-Vid = 17.53) sentences than the picture condition (M-Pic = 13.30). ← 18 | 19 →

So that these results could be generalised, ANOVA 2x2 was run with the following table presenting the important parameters:

Table 1: ANOVA results of the study of sentences in pictures/video vis a vis Polish/English conditions

A significant main effect was obtained only for input type (F (1, 26) = 5.2; p <.05). ANOVA further revealed a significant interaction between language and input type: (F (1, 26) = 11; p<.05). However, there was no significant main effect for language choice: (F (1, 26) = 2.47; p >.05).

To sum up, language choice was not connected with a difference in the number of sentences the writers used to retell the story, but the selection of video over pictures led to their significantly greater numbers. There was also a robust interaction between language choice and video; in was in the video condition, where the writers used Polish (their native language), that resulted in a greater number of sentences.

4.1 Discussion of results

Language choice had no effect on the number of written sentences. The participants reported the input story to be relatively easy to write about, and also that they did not avoid coding in language any specific aspect of the input story. In their post-task commentaries collected immediately after the main task, they also stated that there were problems connected with coding the details of the scenes around the episode of car repair. This shows that time is not a good memory index. The details did not form a cause-effect chain, which normally leads to enhanced memory of events (cf. Barsalou 1998).

Moreover, though the instruction required the participants to retell the story, many of them stated the task required creativity. It was especially effortful in the “picture” condition, as the static pictures disturbed the normal continuity ← 19 | 20 → presented on the video, and the writers in this condition must have been compelled to impose interpretations on how to connect the pictures to form some coherent whole. The effect of language choice (Polish) surfaced only when it was coupled with the video presentation, and hence interaction was reported (see above). The next section presents the report of the second analysis into the type of sentences and events used in the study.