Research Article
Print
Research Article
From visual perception to comprehension: Variations in construal and gaze behavior
expand article infoYana Novikova, Maria Kiose§
‡ Moscow State Linguistic University, Moscow, Russia
§ Institute of Linguistics RAS, Moscow State Linguistic University, Moscow, Russia
Open Access

Abstract

The study explores the construal shifts from VISUAL PERCEPTION to COMPREHENSION image schemas (Grady, 2005) which the readers may face when they deal with the contexts employing the verb of visual perception видеть (see) in the Russian language. The aim of the study is to find out how different image schema types affect the readers’ gaze behavior and default interpretation. It develops a two-stage procedure which involves the corpus-based semantic analysis followed by the oculographic experiment and participants’ reports. The corpus-based search and semantic annotation help define a range of linguistic attractors (Gibbs, 2006) contingent on the image schema types of visual perception and comprehension with the latter exhibiting a more abstract character (Grady, 2005). To proceed, we appeal to the morphological, lexical, and syntactic attractors of the verbs of perception and abstract semantics outlined in (Kustova, 2004; Baicchi & Digonne, 2018; Béligon et al., 2019) and apply them for further samples annotation and contingency analysis. The typical attractors for VISUAL PERCEPTION image schema involve the verb in the perfective, visual semantics in pre-position, direct object in post-position, negation marker, animated direct object in post-position; whereas the COMPREHENSION image schema attractors are scarce and are restricted to attributes of evaluation in pre-position and several adverb types. The oculographic experiment tests the effects of perceptual construal priming and the construal entrenchment on gaze behavior and default interpretation. The experiment did not reveal significant schema type contingency on interpretation, however in terms of gaze behavior COMPREHENSION image schema was more demanding.

Key Words

image schema, attractors, entrenchment, VISUAL PERCEPTION image schema, COMPREHENSION image schema, ambiguity, oculographic experiment

Introduction

The semantics of visual perception verbs in Russian, with most studies exploring the verb видеть (see) has received a lot of attention in formal and functional linguistics (for instance, in Apresyan, 1995; Kustova, 2004; Paducheva, 2004). These semantic studies as well as the lexicographic studies (DRL, 2006; ADRL, 2016) distinguish between two basic meanings of видеть, perceptual and non-perceptual (видеть1 as ‘visually perceive’ and видеть2 as ‘understand’, ‘imagine’, ‘dream’). The same two basic meanings of the verbs of visual perception are explored in other languages (Essa, 2010; Blendea, 2015; San Roque et al., 2015, among many) which proves that the construal shifts from visual perception to comprehension are common when we encounter the verbs of visual perception in context. To describe these construal shifts, we will address the notion of image schemas developed in cognitive linguistics and cognitive psychology and associated with any particular type of both perceptual and non-perceptual experience. Image schemas are the construal patterns of events or their components, referents, actions, time, or place (Lakoff, 1987; Johnson, 1990; Gibbs & Colston, 1995; Cienki, 1997), therefore they may suffice to study the differences in sensory-motor (perceptual) and non-imagistic (abstract, non-perceptual) construal (Sinha & Jensen de Lopez, 2000; Grady, 2005; Mandler, Cánovas, 2014). However, whereas the types of image schemas have been explored in multiple studies, there is still scarce consistent experimental research on how perceptual and abstract image schemas are construed and what guides their selection, whether the selection of abstract image schemas requires higher cognitive load (for instance, in terms of comprehension or behavior reactions (Ginns & Leppink, 2019)) and how people make their choice in case of context ambiguity when the context does not offer explicit cues to the schema selection (MacArthur et al., 2015). In the study, we offer a possible methodological solution to the problem which consists of applying parametric semantic analysis to exploring the two basic meanings of видеть in context which may evoke two image schemas, and oculographic experiment followed by the participants’ responses to test the gaze and default interpretation reactions (Giora et al., 2018). We expect to find the contextual semantics cues which in some cases attract either VISUAL PERCEPTION or COMPREHENSION image schema, and in other cases may attract both.

Prior experimental studies exploring the construal processes have revealed several factors which might affect both gaze behavior and default interpretations, with the factors of perceptual construal priming (the image schemas of perception might be selected first) and construal entrenchment (Shmid, 2016; Tsaroucha, 2020) being the most important. Following prior research, we have expected their effects, however it was difficult to predict their combined influence and their specific influence onto gaze patterns and default interpretations of image schemas. These are the questions to be answered in our study.

Methods and procedure

In the studies exploring construal processes, abstract schemas are shown as secondary (in relation to perception schemas), although being both grounded in our bodily experience (Grady, 2005; Bergen, 2015; Pecher, 2018). It suffices to assume that the factor of perceptual construal priming might affect the changes in the reader’s cognitive load in terms of default interpretations and gaze behavior; the priming effects of this type can be measured considering the fixation duration, first fixation duration in the Area of Interest or other indicators of steady gaze position (Rayner et al., 2010; Holmqvist, 2011; Huettig et al., 2011; Rayner & Liversedge, 2011; Barabanschikov, 2015, among many). Since we expect that semantic variation in context can precipitate the schema selection, we also consider the role of entrenchment in the construal (Langacker, 2016) which might be stimulated by several linguistic patterns. These patterns contingent on two basic meanings of the verb видеть may be revealed by means of corpus-based methods (Divjak, 2015; Divjak & Caldwell-Harris, 2015), where higher frequency of the language patterns with видеть as ‘see’ and as ‘understand’ evidences in favor of a more entrenched construal pattern which will most probably evoke VISUAL PERCEPTION or COMPREHENSION image schema. At the same time multiple experimental studies prove that more entrenched patterns contribute to the decrease in cognitive load measured in gaze duration (indicators of steady gaze position) and default interpretations (Becker & Giora, 2018; Filik et al., 2018; Kiose, 2020). Therefore, both factors (perceptual construal priming and entrenchment) affect the cognitive load redistribution, however their combined influence still needs to be attested.

To explore these effects, we develop a complex procedure which comprises two stages 1) the corpus-based analysis (the search was performed in the National Corpus of the Russian Language (NCRL / НКРЯ)) of contextual uses of the verb видеть) followed by manual annotation of semantic features which in all probability evoke two image schemas VISUAL PERCEPTION and COMPREHENSION; 2) the oculographic experiment which is aimed at verifying the significance of perceptual construal priming and entrenchment. We were particularly interested in the cases of construal ambiguity which appeared when the context of видеть displayed the semantic features of both perceptual and abstract construal and consequently might be “read” and “interpreted” as either of them. Since the Areas of Interest in the reading stimulus which might have influenced interpretation could not be restricted to single verbs and had to be largely extended, we decided to consider the gaze duration in the zones starting from the target verb and extending till the end of the sentence comprising two clauses. This method has its disadvantages as some other factors and other context components might influence gaze duration and comprehension, still we assume that 1) the verb видеть in the predicate position will govern the construal processes in its subordinate clause (e.g. I saw that she was unhappy), at the same time the semantics of this clause will contribute to its perceptual or abstract construal (Kaschak & Glenberg, 2000) and 2) we do not assess the default interpretation of the whole Area of Interest, but of видеть only (which will be formulated in the task we give the experiment participants), and therefore they are expected to give account of the preferred image schema and not of the whole Area of Interest event. We will consider the role of both factors in the gaze duration indicators and default interpretation index (the number of correctly identified meanings).

At the first stage we explore the semantic cues of two meanings of the verb видеть, visual and abstract, which might evoke VISUAL PERCEPTION and COMPREHENSION image schemas. We adopt the view that they have to foreground some features of referents and events (Iriskhanova, 2018; Kiose, 2019) which may be either perceived or comprehended, and have to demonstrate variation in further contrastive analysis of the patterns with visual and abstract verb meaning. Following R. Gibbs (Gibbs, 2006) who develops a system of semantic parameters attracting the construal patterns (“attractors”), we apply the foregrounding semantic elements to contrast the patterns which might also serve as attractors for the image schema selection. Presumably, these elements attract the attention of the reader (which might influence gaze behavior) and provide smoother construal (which will influence default interpretation index).

In (1–2) we offer an example of the attractors and the procedure of their elicitation. Contrastive analysis of the patterns employing the verb видеть has shown that even conjunctions and pronouns may serve as attractors, for instance что (that) in post-position (following the verb видеть) and как (in similar cases in English, Complex Object is used, e.g. I saw him swimming or the construction with the way, e.g. I see the way it goes).

(1) Она засмеялась. Но я видел, что известие про комара с ДНА Дракулы ее расстроило (В. Пелевин, 2013, НКРЯ).

she laugh-PST but I see-PST that news about mosquito-OBJ with DNA Dracula-POSS she-ACC disappoint-PST

She burst out laughing. However, I saw that the news about the mosquito with Dracula’s DNA disappointed her (V. Pelevin, 2013, NCRL)

(2) Вот, например, я хочу что-то нарисовать, и не что-то, а конкретно, срисовать какую-то картинку. И я вижу, как там нарисованы несложные линии (Е. Гришковец, 2004, НКРЯ).

Well, for instance I want something-OBJ draw-INF and not something but concretely copy-INF some picture-OBJ. and I see that there draw-PL-PST-PASS simple lines-SUBJ

For instance, I want to draw something, to be more specific, I want to make a copy of a picture. And I see the way simple lines are drawn there (Ye. Griskovets, 2004, NCRL)

In (1) the verb видел with что in post-position may activate either a VISUAL PERCEPTION image schema, since the speaker can actually visualize a saddened face, or a COMPREHENSION one, since the speaker can infer from what he preliminarily saw that the woman was disappointed and not immediately visualize it. However, we found that the use of что is more frequent with видеть in its abstract meaning and will probably attract COMPREHENSION image schema. At the same time the use of как in the cases similar to (2) is more typical of видеть in its visual perception meaning and is likely to attract VISUAL PERCEPTION schema.

To select the relevant attractors, at this stage we addressed the studies on verbal semantics in (Apresyan, 1995; Kustova, 2004; Paducheva, 2004; Baicchi & Digonne, 2018; Béligon et al., 2019, among many) which list the verb aspect, object type, adverbial modifier type as relevant for contrasting the semantics of visual perception and abstract semantics. These elements were further verified (and some earlier unnoticed attractors were added to the list) to see whether they display variation in the patterns. We finally included the attractors which were active in some patterns and inactive in others, so that they might display contingency with either VISUAL PERCEPTION or COMPREHENSION image schemas of possible construal. The frequency (activity) of attractors was further assessed after the procedure of manual annotation of the corpus-based selected patterns employing the verb видеть in several of its finite forms. Statistical processing of the results of the manual annotation allowed to identify the attractors more and less typical of either verb meaning and consequently of two image schemas (where the image schema type could be identified) and the attractors more and less typical of fuzzy meaning in cases of context ambiguity. These results were further applied to select the stimuli samples for the oculographic experiment, since they had to display verified variation in the activity of attractors as contingent with the types of image schemas.

At the second stage we conducted the oculographic experiment with the selected stimuli demonstrating VISUAL PERCEPTION, COMPREHENSION, and “fuzzy” image schemas. The oculographic experiment was performed with the participants’ responses elicitation, where the responses were given after reading each sentence containing the lexeme видеть. We selected the samples with the attractors highly contingent on one or the other of the schemas as well as the sample with the attractors which were contingent on both schema types (a “fuzzy” image schema). To do it, we applied the results of statistical processing of corpus samples (in terms of the distribution of the attractors in the samples displaying two image schemas) received at the first stage of the study. Since we expected the effects of perceptual construal priming, we hypothesized that VISUAL PERCEPTION image schemas would require less effort in gaze behavior and would receive better default interpretations. To test the effects of entrenchment, we selected the samples demonstrating more and less typical distribution patterns of attractors. We hypothesized that less typical (although clearly contingent on either of the image schemas) distributions of attractors (in terms of R. Giora et al. (2018) less salient) will stimulate longer gaze duration and have lower default interpretation index. Among multiple gaze parameters which might suffice to test gaze duration we applied fixation count and duration and saccade duration following (Rayner et al., 2010; Rayner & Liversedge, 2011) who explored the gaze behavior in similar tasks of reading for extracting specific information in the Areas of Interest. However, since our Areas of Interest had different length, we used the relative values of these data (for instance, fixation duration per one sign of the Area of Interest). To study default interpretation, we introduced (following Giora et al., 2018; Kiose, 2020) the default interpretation index which corresponded to the ratio of adequately identified image schema types to the total number of image schemas (or rather the samples evoking image schemas) in the stimulus. Since in their responses several participants had doubts when making their choice in favor of this or that schema (there were hesitation pauses, additional comments), in coding the participants’ responses we used a three-point scale, default interpretation (1), non-default (incorrect) interpretation (0) and vague response (0.5).

The gaze data and default interpretation data were further subjected for T-tests to evaluate the effects of perceptual construal priming and entrenchment.

Image schemas and their attractors

The samples of the contexts with the verb видеть in the finite forms вижу, увижу, видел, увидел were preliminarily selected from the National Corpus of the Russian Language (from the sub-corpus of Russian-language fiction texts published after 1980 since we had to avoid outdated samples). Since the samples were to be manually annotated for the presence or absence of attractors, their number had to be sufficient for the study to be accomplished. At the first step when we had to develop the system of attractors; we formed the first samples’ sub-corpus of 800 samples from the whole sub-corpus found in NCRL 4,120 documents, 31,920 samples (with вижу appearing in 1,063 documents, 5,941 samples, увижу in 360 documents, 610 samples, видел in 1,411 documents, 13,679 samples, увидел in 1,286 documents, 11,960 samples). The search procedure was terminated (and the samples sub-corpus was compiled) due to the discovery of a “stable” set of image schemas and the absence of the emergence of new schemas. These included the following image schemas: VISUAL PERCEPTION, COMPREHENSION, IMAGINATION, MEETING or SEEING AGAIN, DREAM or SEEING IN A DREAM, the examples of possible selection of more than one image schema at a time were also found (see (1–2)). For the most part, the selection of the schema type was context-dependent, for instance, we might observe the schema of COMPREHENSION in (3), IMAGINATION in (4) and MEETING in (5):

(3) А теперь я вижу, что это начало никуда не годится (В. Белоусова, 2001, НКРЯ).

and now I see-PR-SG-1 that this start nowhere not go-PR-SG-3

And now I see that this start is going nowhere (V. Belousova, 2001, NCRL)

(4) Ну, что такое бабушка, говорил он, наклоняя голову, чтобы закурить, наконец, эту короткую ядовитую сигарету (страшнее «Шипки» был, кажется, только «Дымок»), поднося к ней (как ясно вижу я теперь все это…) французскую черную, с золотистым окончанием и кружочком, цеплявшимся за кремень, одноразовую зажигалку фирмы BIC (А. Макушинский, 2012, НКРЯ).

well what be-PR-SG-3 Grandma say-PST-SG-3 he tilt-PTCP head-ACC to light finally this short poisonous cigarette-ACC (scary-COMP Shipka-ACC be-PT-SG-3, seem-PR-SG-INDF-3 only Dymok) bring-PTCP to it-ACC (how clearly see-PR-SG-1 I it all now) French black with golden ending and circle cling-PTCP to flint disposable lighter-ACC company-GEN BIC

Well, what is the matter Grandma, he said, tilting his head to finally light that short, poisonous cigarette (the only thing scarier than Shipka seemed to be Smoke), bringing to it (how clearly I see it all now...) a French black, with a gold-tipped, flint-clinging circle, disposable BIC lighter (A Makushinsky, 2012, NCRL).

(5) Эти прекрасные женщины с этих фотографий… я таких не встречал, я не видел таких ни на улице, ни в магазинах, и даже на пляже таких не видел (Е. Гришковец, 2001, НКРЯ).

This-PL beautiful woman-PL in this-PL photo-ABL-PL… I they-ACC never meet-PST-SG, I never see-PST-SG they-ACC not in street-PL not in shop-PL and even on beach them not see-PST-SG

The beautiful women in these photos... I’ve never met them, I’ve never seen them in the streets or in shops, and even on the beach I’ve never seen them (E. Grishkovets, 2001, NCRL).

As can be seen, these image schemas vary in their degree of abstraction, from more abstract, related to comprehension, to less abstract, demonstrating a repeated (multiple) manifestation of visual perception.

With a certain degree of simplification, we considered all the schemas detected within the two main poles of perceptual and abstract construal. In doing so, we observed the frequency (activity) of the attractors being contingent with one or the other of the image schemas.

The complete list included 59 attractors on three levels: morphological, lexical and syntactic. Morphological attractors included the tense of the verb, its aspect and the presence of negation не; lexical attractors included the presence of attributes (ясно, отчетливо, etc.), direct object, intensifier (ни разу, никогда), object with animated referent semantics, action repetition, change of object location, verb phase, visual semantics in pre-position or post-position, etc. Syntactic attractors included the presence of syntactic words like как, что, чтобы, indicators of a new microevent (вдруг, тут, снова, etc.), parallel constructions, modifiers of place and time in the pre- and post-position, etc. The system of attractor coding was used to annotate the samples: codes 101–105 were used for the attractors of the morphological level, 201–229 for the attractors of the lexical level and 301–326 for the attractors of the syntactic level. Next, we subjected the first sample sub-corpus to randomization procedure, selecting finally 100 samples for annotation including into the second sub-corpus only the first sample from the search page and only one sample from one author. Each sample was manually annotated for the presence and absence of attractor markers and the image schema type (with further specification). The data were processed with HETEROSTAT software (Kiose, Efremov, 2020) to identify the attractor activity in relation to image schema types and the presence of significant attractor correlations.

First, we will outline the results on attractor activity. In Figure 1 we present the image schemas profiles which show the relevant activity of attractors.

Figure 1. 

Image schema profiles.

For VISUAL PERCEPTION image schemas (58 out of 100 samples) the following attractors showed the highest activity (relative frequency of their use is given in brackets): the perfective of the verb видеть and its aspectual derivatives (0.31), the negative prefix не (0.4); object with animated referent semantics (0.29), the presence of a microevent in post-position, associated with a change of location, movement (0.22), the reference to visual mode in pre-position (0.47) and to visual mode in post-position (0.28), the repeated use of a lexeme with visual mode semantics in post-position (words with the semantics of vision) (0.28), the reference to perception in post-position (colour, shape of the object, distance from it, material) (0.31), the involvement of several modes of perception (sight, hearing, smell, touch, taste) (0.28); the presence of a direct object in pre-position (0.22), the presence of a direct object in post-position (0.41), the conjunction что (0.17), the pronoun что (0.24). The following attractors are characteristic of the COMPREHENSION image schemas (31 out of 100): attributes in pre-position отчетливо, ясно, четко, прямо-таки, etc. (0.77), the presence in pre-position of a non-visual microevent (0.1), the adverb как (0.23), the pronoun что (0.13), a modifier of manner in pre-position (0.61). Thus, COMPREHENSION schemas have significantly fewer rigid attractors. This proves that a much larger context is required for their construal.

The remaining 11 samples were categorized as ambiguous, i.e., demonstrating both perceptual and abstract construal possibilities. Next, we applied Pearson correlation analysis to detect the statistically significant attractor correlations. With the critical r-Pearson at p ≤ 0.01 equal to 0.37, R-values exceeding it were considered significant. There were multiple attractor pairs with significant R-values, 34 for perceptual schemas and 42 for non-perceptual schemas; therefore we had to dismiss the idea of ranging the samples considering their R-values. However, among the most significant attractor pairs for VISUAL PERCEPTION schemas were the attributes in pre-position отчетливо, ясно, четко, прямо-таки, etc. + the presence of the pronoun себя (myself) in pre-position + the presence a non-visual microevent in pre-position. The following correlation pair was most typical of COMPREHENSION schemas: inversion + a modifier of time in post-position. Therefore, we decided to include the samples demonstrating these attractor correlation pairs into the stimulus data.

To test whether the gaze behavior and the defaultness of interpretation are actually determined by the image schema type, we selected the samples of the two types of image schema and one sample demonstrating the attractors’ contingent on both schema types to be included into the stimulus. Since we expected the effects of the entrenchment factor, we selected the samples whose attractor distribution was more and less concomitant with the attractor distribution of the two poles.

Gaze behavior and default interpretation

While developing the design of the experiment we considered similar studies contrasting gaze behavior and comprehension (Just & Carpenter, 1980; Hyönä et al., 2003; Pollatsek & Treiman, 2015). To compile the stimulus, we included 7 samples (taken from NCRL) with the form вижу (see-Pr-SG-1). After reading each sample, the participants had to identify the image schema used. Three answer options were provided: вижу as perceived with the eyes, вижу as understood and It is difficult to say / I don’t know (in case the participant was not certain). The Area of Interest was the text fragment starting from the verb вижу to the end of the sample. The stimulus involved 7 Areas of Interest with a target word given in bold type; the Areas differed in length (49, 42, 41, 61, 73, 32, and 48 signs respectively) which should have been considered when assessing the reading time. The experiment was conducted with SMI Red eye tracker with a recording rate of 60 Hz, the participants were seated 60 cm from the screen. The whole experiment for each participant lasted about 7 or 8 minutes with the instruction, calibration time and the gaze record. 12 students participated in the experiment; the average age was 22 years. Oculographic data processing was carried out considering the fixation count, fixation duration per sign, saccade duration per sign in the Areas of Interest. The decision to consider the gaze characteristics per sign (and not per word) was necessitated to balance the differences in the length and semantic roles of the words in post-position to the target word; for instance, in вижу ее в темноте we found it inappropriate to consider the semantic potential of the words ее, в and темноте on equal terms. This explains why we did not apply the dwell time per word as an index of gaze costs, which is most common in gaze experiments aimed at specifying the characteristics of visual foregrounding in words (Rayner et al., 2010; Rayner & Liversedge, 2011), but to view gaze behavior within AoI as is commonly done in experiments testing gaze characteristics in images (Divjak et al., 2020). To identify the default interpretation index, we applied the procedure described above. In Figure 2 we present the stimulus.

Figure 2. 

Stimulus with 7 Areas of Interest.

Below we show the samples with their glosses (the vertical left-alignment was not kept due to the fact that the sentences are long; still, the glosses allow to explore the context in pre-and post-position to the verb) and translations.

КОГДА Я ВОЗВРАЩАЮСЬ, НАВСТРЕЧУ МНЕ ПОПАДАЕТСЯ МАША. Я ОЧЕНЬ ЯСНО ВИЖУ ЕЕ В ТЕМНОТЕ. МЫ МОЛЧА ГЛЯДИМ ДРУГ НА ДРУГА.

when i come back-PR-SG, towards i-ACC appear-PR-SG-3 Masha i very clearly see-PR-SG-1 she-ACC in dark we silently look-PR-PL each at other

WHEN I COME BACK, MASHA COMES TOWARDS ME. I CAN SEE HER VERY CLEARLY IN THE DARK. WE LOOK AT EACH OTHER IN SILENCE.

ВСЕ ВРЕМЯ КАЖЕТСЯ, ЧТО МЕНЯ НЕ ПОНИМАЮТ ТЕ, КТО РЯДОМ. ОДНАКО, ЗАДУМЫВАЯСЬ О ТОМ, КАК ВЫГЛЯДЯТ СО СТОРОНЫ ВСЕ МОИ СВЕРХТОНКИЕ ИСКАНИЯ, Я ЧЕТКО ВИЖУ, ЧТО ПОНИМАТЬ-ТО, В СУЩНОСТИ, НЕЧЕГО.

all time seem-PR-SG-INDF-3 that i-ACC not understand-PL-NEG this-NOM-PL who around however, think-PTCP about how look-PR-PL from outside all i-POSS superfine quest-PL i clearly see-PR-SG that understand-INF, in fact, nothing-ACC

IT SEEMS TO ME ALL THE TIME THAT I AM MISUNDERSTOOD BY THOSE AROUND ME. HOWEVER, WHEN I THINK ABOUT HOW ALL MY SUPERFINE SEARCH LOOKS FROM ASIDE, I CAN CLEARLY SEE THAT THERE IS IN FACT NOTHING TO UNDERSTAND.

ГЕНЕРАЛ ПРОСИТ ПОКАЗАТЬ ЕМУ, ЧТО Я С ТАКИМ ИНТЕРЕСОМ ЧИТАЮ. Я ПРОТЯГИВАЮ КНИГУ И ДОКЛАДЫВАЮ, ЧТО ВЫПОЛНЯЛ НАРЯДЫ ПО УБОРКЕ ГОСПИТАЛЯ. И ВОТ ТЕПЕРЬ, В СВОБОДНОЕ ВРЕМЯ, ПОВЫШАЮ СВОЮ БОЕВУЮ ПОДГОТОВКУ. Я ВИЖУ, ЧТО ВИЛЬЧЕНКО ДОВОЛЕН МОИМ ОТВЕТОМ.

general ask-PR-SG-3 show-INF he-ACC what i with such interest read-PR-SG-1 i give-PR-SG-1 book and report that complete-PST-SG duty-PL on cleaning-DAT hospital-GEN and now on in spare time improve-PR-SG-1 i-POSS combat training i see-PR-SG-1 that vilchenko happy-ADJ i-POSS answer

THE GENERAL ASKS ME TO SHOW HIM WHAT I AM READING WITH SUCH INTEREST. I HOLD OUT THE BOOK AND REPORT THAT I HAVE BEEN DOING HOSPITAL CLEANING DUTY. AND NOW IN MY SPARE TIME I AM IMPROVING MY COMBAT SKILLS. I SEE THAT VILCHENKO IS PLEASED WITH MY ANSWER.

А СТЕНА ТАКАЯ ЖЕ, НЕКРАШЕНАЯ, ПРЕЖНЯЯ, ИЗ КРАСНОГО КИРПИЧА, И КИРПИЧИ, ТОЖЕ КАРТИННО, ВЫСТУПАЛИ ИЗРЕДКА НЕБОЛЬШИМИ СТУПЕНЬКАМИ. ПО ЭТИМ «СТУПЕНЬКАМ» МОЛОДОЙ ПАРЕНЬ ― Я ВИЖУ ОТЧЕТЛИВО ДО СИХ ПОР ― ВЗБИРАЛСЯ И ДОЛЕЗ СНИЗУ ДО КРЫШИ.

and wall the same paint-PST-PTSP-NEG old from red brick and brick-PL also picturesquely protrude-PT-PL occasionally small step-ABL-PL on this-PL step-DAT-PL young lad - i see-PR-SG-1 clearly to this day - climb-PT-SG and reach-PT-SG below to roof

STILL THE WALL WAS THE SAME, UNPAINTED, FROM OLD RED BRICK, AND THE BRICKS, ALSO PICTURESQUE, PROTRUDED OCCASIONALLY IN SMALL STEPS. ON THESE “STEPS” A YOUNG LAD - I CAN STILL SEE CLEARLY TO THIS DAY - WAS CLIMBING AND REACHING FROM BELOW TO THE ROOF.

ПОТОМУ ЧТО Я ЗНАЮ, ЧТО ДЕНЕГ МНЕ БОЛЬШЕ НИГДЕ НЕ НАЙТИ. И ТУТ Я ВИЖУ В УГЛУ СОВСЕМ МАЛЕНЬКОГО МАЛЬЧИКА. ГОДА ЧЕТЫРЕ ЕМУ, ИЛИ ЧУТЬ БОЛЬШЕ.

because that i know-PR-SG-1 that money-ACC i-ACC more anywhere not find-INF-NEG and then i see-PR-SG-1 in corner too little boy year-PL four he-DAT or little more

BECAUSE I KNOW I CAN’T FIND THE MONEY ANYWHERE ELSE. AND THEN I SEE A LITTLE BOY IN THE CORNER. HE’S ABOUT FOUR YEARS OLD, OR A LITTLE MORE.

Я РАССМАТРИВАЮ ЕЕ УВЯДАЮЩУЮ КОЖУ. И Я УЖЕ ВИЖУ, ЯСНО ВИЖУ, КАКОЙ ОНА СТАНЕТ СТАРУХОЙ!

i watch-PR-SG she-POSS wither-PTCP skin and i already see-PR-SG-1 clearly see-PR-SG-1 what she become-FUT-SG-3 old-ABL!

I LOOK AT HER WITHERING SKIN. AND I CAN ALREADY SEE, I CAN CLEARLY SEE WHAT AN OLD WOMAN SHE WILL BECOME!

ФИКСИРУЕТ ИЗЛИШНЮЮ ЗАДУМЧИВОСТЬ, РАССЕЯННОСТЬ, БЕСПРИЧИННЫЕ ПРИСТУПЫ ВЕСЕЛЬЯ ИЛИ ПЕЧАЛИ; ОНА ЗНАЕТ ВСЕ. Я ЧЕТКО ВИЖУ, ЧТО ОНА ЕЩЕ ЛЮБИТ МЕНЯ, НО УЖЕ НЕ УВАЖАЕТ.

fix-PR-SG-3 excessive thoughtfulness absent-mindedness explain-PST-PTCP-NEG bout-PL merriment-GEN or sadness-GEN; she know-PR-SG-3 everything i clearly see-PR-S1 that she still love-PR-SG-3 i-ACC but yet not respect-PR-SG-3

SHE FIXES EXCESSIVE THOUGHTFULNESS, ABSENT-MINDEDNESS, UNEXPLAINED BOUTS OF MERRIMENT OR SADNESS; SHE KNOWS EVERYTHING. I CAN CLEARLY SEE THAT SHE STILL LOVES ME, BUT NO LONGER RESPECTS ME.

In the stimulus, the first and fifth samples illustrate VISUAL PERCEPTION image schema, the second, fourth, sixth and seventh samples illustrate COMPREHENSION schema, and the third sample illustrates an ambiguous case. The default interpretation indexes are shown in Table 1.

Table 1.

Default interpretation indexes.

Sample number 1 2 4 5 6 7
Image schema type VISUAL PERCEPTION COMPREHENSION COMPREHENSION VISUAL PERCEPTION COMPREHENSION COMPREHENSION
Default Interpretation Index 0.917 0.875 0.333 0.917 0.909 1.0

If we consider the data, the mean value for default interpretation of VISUAL PERCEPTION image schemas is equal to 0.92 and of COMPREHENSION image schemas it is equal to 0.78, therefore in general VISUAL PERCEPTION image schemas seem to be interpreted better. However, as it may be seen, the participants faced the problem interpreting only one sample, which is sample 4, the one illustrating COMPREHENSION image schema. Considering it in detail we may observe that the sample demonstrates a number of attractors typical of the VISUAL PERCEPTION image schema, e.g. the presence in post-position of a microevent associated with a change of location, movement (взбирался и долез), the reference to visual mode in pre-position (кирпичи, тоже картинно, выступали изредка), the reference to visual mode in post-position (object angle, distance from it to the bottom to the roof), the reference to perception in post-position (type of paint, small steps). At the same time, only one attractor contingent on COMPREHENSION image schema is active, it is the presence of a modifier of time до сих пор. Obviously, this attractor, even despite the presence of a large number of attractors of VISUAL PERCEPTION image schema, can play a dominant role in the schema selection. However, the presence of multiple attractors of VISUAL PERCEPTION schema proved to be a constraint: one quarter of subjects (3 of 12) expressed their doubt about the schema type, only 2 out of 12 opted for COMPREHENSION schema. Therefore, schema selection is influenced by 1) the presence of attractors of a certain type of image schemas, 2) the presence of certain (most significant) attractors, and 3) the absence of constraining factors, i.e., the presence of significant attractors of another image schema. It is noticeable that we found that the default interpretation index was not largely affected by the type of an image schema itself – both VISUAL PERCEPTION and COMPREHENSION schemas are recognized equally easily.

As for sample 3, a different procedure of assessing interpretation was used, which employed the preference scale: the preference index for VISUAL PERCEPTION or COMPREHENSION image schema is the ratio of responses for one or the other schema to the total number of responses. The preference index for VISUAL PERCEPTION image schema is found to be 0.21 (the preference index for COMPREHENSION image schema is 0.79, respectively). The findings suggest that in ambiguous cases (when there are no significant attractors in favor of VISUAL PERCEPTION schema) the reader chooses COMPREHENSION schema type. However, this statement needs further verification. The priming effects of reading should not be excluded, since the previous sample (sample 2) presents COMPREHENSION schema.

The analysis of the participants’ gaze behavior was then carried out. At the first step we examined the character of the gaze path to see whether the gaze paths would follow serial or parallel lexical processing (E-Z Reader (Reichle et al., 2003) vs. SWIFT (Richter et al., 2006) reading models), which conform to more and less linear information construal with the latter more typical of the search tasks. The analysis of gaze paths on the samples in total did not give strong evidence in favor of parallel processing. In Figure 3 we show the gaze paths of several experiment participants.

Figure 3. 

Gaze paths of participants.

As we can see, the first fixations appear in the central area of the stimulus which is a common case. Next, the gaze path follows to the top left corner and serial processing starts. However, in most cases in the Areas of Interest we face several reverse saccadic movements as well as chaotic movements in several fragments of the Areas, mostly contingent on the target verbs. This can be explained by the fact that the participants were to present their responses in the time of reading. That is why despite the general observation that the lexical processing is serial we have to admit that there are multiple return gaze movements and short fixations typical of the parallel processing in the Areas of Interest, however we cannot claim whether this happens due to the search processes involved or simply due to having to give the response at the time of reading. Consequently, the schema contingency on fixation count with the participants (with M = 102, SD = 44.6 not displaying normal distribution with Shapiro-Wilk p = 0.526) was not explored since it was irrelevant.

Now we will proceed to the gaze duration data. Regarding fixation and saccade durations (we have to remind here that the data were calculated taking into account the duration of fixations on a single sign in the Area of Interest), we can claim that the gaze data on VISUAL PERCEPTION and COMPREHENSION schemas showed significant differences. In Table 2, we present the results of gaze behavior in terms of fixations and saccade duration per sign (some missing data on P 05, participant 5, mean that he did not manage to cope with the task in the allotted time being very slow with the reading and responses; and P 02 gaze results appeared difficult to decipher due to calibration problems).

Table 2.

Duration of fixations and saccades, ms per sign in Areas of Interest.

Participant number / Sample number 1 2 3 4 5 6 7
VISUAL PERCEPTION COMPREHENSION Fuzzy / ambiguous COMPREHENSION VISUAL PERCEPTION COMPREHENSION COMPREHENSION
Р 01 5.6 / 0.4 13.7 / 6 32.6 / 45.9 14.1 / 41.9 7.8 / 22.6 16 / 23.6 52.1 / 153.5
Р 03 17.6 / 117 73.8 / 49.2 59.9 / 27.5 39.9 / 11.8 16 / 3 152.1 / 170.9 28.6 / 25.5
Р 04 23 / 89.2 18.4 / 115.5 16.4 / 13.5 17.2 / 37.4 15 / 29.1 44.6 / 53.3 23.5 / 15.6
Р 05 243.8 / 231.6 54.8 / 25.8 44.5 / 39.6 46.2 / 27.6 69.8 / 48.2 - -
Р 06 53.8 / 83.8 89.3 / 117.2 48.4 / 55.6 91.4 / 51.2 52.2 / 26.7 146.9 / 108.2 67.8 / 47.3
Р 07 23.6 / 17.7 31.6 / 17.9 36.8 / 26 23.7 / 14.6 18.2 / 13.5 22.3 / 17.2 47.5 / 47.5
Р 08 44.9 / 41.6 60 / 26.8 113.9 / 48.2 47.2 / 20.4 43.5 / 18.9 88 / 54.9 167.5 / 49
Р 09 53.8 / 57 68.9 / 67.9 122.9 / 40.8 61.1 / 22.9 46.5 / 23 54.9 / 37.7 43.9 / 28
Р 10 16.6 / 458.4 25 / 73.8 26.6 / 45 58.9 / 69.9 9.8 / 15 93 / 63.7 76.9 / 41.6
Р 11 6.1 / 18.7 10.4 / 155.3 16 / 51.9 24.1 / 79.5 11.7 / 15 72 / 93.6 36.8 / 53.3
Р 12 25.5 / 32.3 68.5 / 48.6 72.8 / 22.5 47.2 / 26 70.4 / 36 79.3 / 35.2 65.3 / 45.3

Since it is the fixation duration which evidences of the information elicitation, we considered it to see if fixation duration was contingent on the image schema type being longer in the cases of COMPREHENSION image schemas. Paired T-test (P 05 data were not considered) verifying the contingency of image schema types and fixation durations did reveal it with t(69) = 11.3 at p < 0.001. Further exploration analysis has shown that these were VISUAL PERCEPTION schemas which required shorter fixations and consequently the COMPREHENSION ones required longer fixations. Paired T-test verifying the contingency of image schema types and saccade durations also revealed it with t(69) = 7.49 at p < .001. Further exploration analysis has shown that VISUAL PERCEPTION schemas also required shorter saccades and consequently the COMPREHENSION ones required longer saccades.

We also hypothesized that there might be several attractors which play a dominant role in gaze behavior, especially in terms of fixation duration. Entertaining this idea, we first conducted an ANOVA test to identify the attractors whose activity displayed contingency on fixation duration values in the image schemas. Next, we tried to explore the integral activity of attractors which might have affected the fixation duration applying the procedure of vector regression using the model with non-aliased parameters. This was a three-step procedure, which assessed the contingency and regression in application to the gaze data received from both schema types; it was followed by similar procedures assessing contingency and regression for each of the schema types. Since we faced multiple alignment effects, we could apply the data from 25 attractors assessing the contingency on both schema types, only 6 attractors on VISUAL PERCEPTION image schema, and 14 attractors on COMPREHENSION image schema (the rest were aliased). In terms of contingency results with gaze behavior, ANOVA has revealed only one attractor contingent on fixation duration, which is the syntactic identifier of a new microevent тут, теперь, вдруг, когда, уже with F(1, 21.7) = 5.72 at p = 0.026. In terms of contingency results with image schemas, ANOVA has not shown any statistically significant results. The following regression analysis has revealed that the attractors displayed very high alignment effects (which happened with the attractors 101–105), for that reason we carried it out separately only for two attractor groups, lexical and syntactic.

Table 3 shows the vector regression model for the attractors of the lexical level, and Table 4 shows the vector regression model for the attractors of the syntactic level.

Table 3.

Regression Model for lexical attractors with non-aliased coefficients predicting Fixation duration.

Predictor Fixation Duration R2 = 0.0967
Estimate SE Z P
Intercept 53.71 12.1 4.443 < .001
Attributes in pre-position отчетливо, ясно, четко, прямо-таки 6.63 17.1 0.388 0.699
Object with animated referent semantics -20.9 17.1 -1.223 0.226
Presence of a microevent in post-position (with a change of location, movement) -3.95 17.1 -0.231 0.818
Reference to visual mode in pre-position -13.58 17.1 -0.794 0.43
Repeated use of a lexeme with visual semantics in post-position 28.05 17.1 1.641 0.105
Reference to visual semantics in post-position 7.32 24.2 0.303 0.763
Table 4.

Regression Model for syntactic attractors with non-aliased coefficients predicting Fixation duration.

Predictor Fixation Duration R2 = 0.09
Estimate SE Z P
Intercept 42.82 12.1 3.542 < .001
Adverbial modifier in pre-position 6.63 17.1 0.388 0.699
Exclamatory sentence 25.36 24.2 1.049 0.298
Direct object in post-position -2.69 24.2 -0.111 0.912
Conjunction что 10.89 17.1 0.637 0.526
Syntactic identifier of a new microevent тут, теперь, вдруг, когда, уже -13.95 17.1 -0.816 0.417

The results in both cases show that the models have low predictability, with R2 equal to 0.0967 in the model for lexical attractors, and R2 equal to 0.09 in the model for syntactic attractors. There are no single predictors within the models. The results suffice to claim that neither single attractors nor the groups of attractors differentiating VISUAL PERCEPTION and COMPREHENSION image schemas can account for gaze behavior, at least in terms of fixation duration which is the most evident indicator of information construal. Considering the fact that two types of image schemas do display contingency on fixation duration, we may still assume that low model predictability can be explained by the high attractor alignment which has made us disregard the majority of them in regression analysis.

Conclusion

In the current study, we explored the effects of two factors, perceptual construal priming and entrenchment as potentially influencing the gaze behavior and default interpretation of two image schema types, VISUAL PERCEPTION and COMPREHENSION. Multiple studies claim the importance of sensory modes as affecting the schema selection (Grady, 2005; Mandler, Cánovas, 2014; MacArthur et al., 2015) and the influence of entrenched patterns onto the construal and gaze reactions (Giora et al., 2018; Shmid, 2016; Tsaroucha, 2020), however they do not specify the linguistic construal means of the image schemas under consideration. Hypothesizing that 1) linguistic construal, here analyzed in terms of linguistic attractors (Gibbs, 2006) might stimulate and hamper VISUAL PERCEPTION and COMPREHENSION image schemas selection, and 2) linguistic construal will affect gaze behavior and default interpretation, we designed a two-stage study, involving corpus and data annotation procedure and oculographic experiment.

The corpus search procedure and samples annotation have revealed 59 attractors which demonstrated higher or lower variation (in frequency) within two types of image schemas. The list of attractors was compiled following the semantic studies of the verbs of perception (Apresyan, 1995; Kustova, 2004; Paducheva, 2004, among many), however we explored their relative activity within the two image schemas. This allowed to develop a procedure of ranging the samples in accordance with the attractor activity demonstrated. It also helped reveal the samples which displayed ambiguity and could be attributed to either VISUAL PERCEPTION or COMPREHENSION image schema. We finally selected the samples displaying different attractor distribution to be further used as stimuli samples in the oculographic experiment.

The experiment has shown that both factors of perceptual construal priming and entrenchment play a role in image schema construal, however their effects are not so straightforward. For instance, in terms of default interpretations of entrenched image schemas the difference is insignificant, but when we come to ambiguous samples which display non-entrenched construal, we face a different situation. Participants’ responses have revealed that in most cases they preferred a COMPREHENSION image schema. It suggests that in the situations of uncertainty people might most likely put away the VISUAL PERCEPTION image schema in favor of a more abstract one; this observation may indirectly indicate the priority in VISUAL PERCEPTION schema. In terms of gaze behavior, we revealed significant variance with two schema types, with COMPREHENSION schema being more demanding in terms of reading time (both fixation and saccade duration). We further expected to specify the role of single attractors or their groups in this variance and conducted both variance and regression analysis which in all the cases (but one) gave unsatisfactory results. The only significant attractor turned out to be the syntactic identifier of a new microevent тут, теперь, вдруг, когда, уже, which is easily explicable since it introduces a new textual event and focuses the reader’s attention onto it. At the same time, we tend to think that the results fail to prove the role of the attractors because of high attractor alignment and because of other construal schemas activated in context.

Considering the results obtained, we may claim that gaze behavior tends to be more contingent on the schema type, VISUAL PERCEPTION or COMPREHENSION, whereas default interpretations are more sensitive to the entrenchment factor of attractor distribution within the samples.

Further research in this area might develop new methods to specify the role of single attractors and attractor groups in image schemas as affecting the gaze behavior, which we failed to explore in this study due to high attractor alignment. This could help develop prognostic analysis in terms of both construal patterns and gaze behavior.

Acknowledgements

The chapters “Methods and Procedure”, “Gaze behavior and default interpretation” of the research were written by Maria Kiose and were financially supported by the Russian Foundation for Basic Research, project No. 20-012-00370 “Textual heterogeneity and the factors of its successful reading comprehension” at Moscow State Linguistic University.

References

  • ADRL – Academic Dictionary of the Russian Language (2016). L. P. Krysin (Ed.). Vol. 1. A – vil’at’. Moscow, Nauka.
  • Apresyan Y. D. (1995). Selected Works. In 2 volumes. Vol. 1. Lexical semantics. Мoscow.
  • Barabanschikov V. A. (Ed.). (2015). Eye-tracking in psychological science and practice. Moscow, Cogito-Centre.
  • Becker I., Giora R. (2018). The Defaultness Hypothesis: A quantitative corpus-based study of non/default sarcasm and literalness production. Journal of pragmatics, 138, 149–164. https://doi.org/10.1016/j.pragma.2018.09.013
  • Bergen B. (2015). Embodiment, simulation and meaning. The Routledge Handbook of Semantics. Abingdon, 142–157.
  • Blendea S. (2015). Prototypical meanings in perception verbs. Letter and Social Science Series, 2, 92–97.
  • Cienki A. (1997). Some properties and groupings of image schemas. In M. Verspoor, K. Dong Lee, & E. Sweetser (Eds.) Lexical and Syntactical Constructions and the Construction of Meaning (pp. 3–15). Amsterdam/Philadelphia, John Benjamins. https://doi.org/10.1075/cilt.150.04cie
  • DRL – Dictionary of the Russian Language (2006). S. I. Ozhegov, N. Yu. Shvedova (Eds.). Мoscow, Nauka.
  • Divjak D. (2015). Exploring the grammar of perception. A case study using data from Russian. Sensory perceptions in language and cognition. Special issue of functions of language, 22(1), 44–68. https://doi.org/10.1075/fol.22.1.03div
  • Divjak D., Milin P. (2020). Construal in language: A visual-world approach to the effects of linguistic alternations on event perception and conception. Cognitive Linguistics, 19, 37–72. https://doi.org/10.1515/cog-2018-0103
  • Essa D. (2010). The semantic range of the verbs of perception in English and Arabic: A contrastive study. Journal of Human Sciences, 1(2), 622–662.
  • Filik R., Howman H., Ralph-Nearman C., Giora R. (2018). The role of defaultness and personality factors in sarcasm interpretation: Evidence from eye-tracking during reading. Metaphor and Symbol, 33 (3), 148–162. https://doi.org/10.1080/10926488.2018.1481258
  • Giora R., Jaffe I., Becker I., Fein O. (2018). Strongly attenuating highly positive concepts: The case of default sarcastic interpretations. Review of Cognitive Linguistics, 6(1), 19–47. https://doi.org/10.1075/rcl.00002.gio
  • Holmqvist K., Nystrom M., Andersson R., Dewhurst R., Jarodzka H., Weijer J. (2011). Eye Tracking. A comprehensive Guide to Methods and Measures. N.Y., Oxford University Press.
  • Huettig F., Rommers J., Meyer A. S. (2011). Using the visual world paradigm to study language processing: A review and critical evaluation. Acta Psychologica, 137(2), 151–171. https://doi.org/10.1016/j.actpsy.2010.11.003
  • Hyönä J., Lorch Jr.R.F., Rinck M. (2003). Eye movement measures to study global text processing. In J. Hyönä, R. Radach, & H. Deubel (Eds.), The mind’s eye: Cognitive and applied aspects of eye movement research (pp. 313–334). Amsterdam: Elsevier Science. https://doi.org/10.1016/B978-044451020-4/50018-9
  • Iriskhanova O. K. (2018). On the dynamic approach to explore the narrative. Images of language and zigzags of discourse: Collection of papers for the 70th anniversary of Demyankov V.Z., 192–215.
  • Johnson M. (1990). The Body in the mind: the perceptual basis of meaning, imagination, and reason. Chicago, London, The University of Chicago press.
  • Kaschak M.P., Glenberg A.M. (2000). Constructing meaning: The role of affordances and grammatical constructions in sentence comprehension. Journal of Memory & Language, 43, 508–529. https://doi.org/10.1006/jmla.2000.2705
  • Kiose M. I. (2020). The interplay of syntactic and lexical salience and its effect on default figurative responses. Studies in Logic, Grammar and Rhetoric, 61 (74), 161–178. https://doi.org/10.2478/slgr-2020-0004
  • Kiose, M. I. & Efremov A. A. HETEROSTAT. Program for discourse parameters variance calculation. Registered 21.09.2020. Registration Number 2020661240.
  • Kustova G. I. (2004). Type, visibility, essence (on the semantic potential of words with the meaning of visual perception). The hidden meanings: Word. Text. Culture: Collection of articles in honor of Arutyunova N.D. Мoscow, 155–175.
  • Langacker R.W. (2016). Entrenchment in cognitive grammar. In H-J. Schmid (Ed.). Entrenchment and the Psychology of Language Learning. How We Reorganize and Adapt Linguistic Knowledge (pp. 39–56). Washington DC, Mouton de Gruyter. https://doi.org/10.1037/15969-003
  • MacArthur F., Krennmayr , Littlemore J. (2015). How basic is “UNDERSTANDING / KNOWING IS SEEING” when reasoning about knowledge? Asymmetric uses of sight metaphors in office hours consultations in English as academic lingua franca. Metaphor and Symbol, 30(3), 184–217. https://doi.org/10.1080/10926488.2015.1049507
  • Paducheva E.V. (2004). Dynamic models in lexical semantics. Moscow.
  • Rayner K., Slattery T. J., Bélanger N. N. (2010). Eye movements, the perceptual span, and reading speed. Psychonomic Bulletin and Review, 17(6), 834–839. https://doi.org/10.3758/PBR.17.6.834
  • Reichle E. D., Rayner K. (2003). The EZ Reader model of eye-movement control in reading: Comparisons to other models. Behavioral and Brain Sciences, 26, 445–476. https://doi.org/10.1017/S0140525X03000104
  • Roque L., Kendrick K. H., Norcliffe E., Brown P., Defina R., Dingemanse M., Majid A. (2015). Vision verbs dominate in conversation across cultures, but the ranking of non-visual verbs varies. Cognitive Linguistics, 26(1), 31–60. https://doi.org/10.1515/cog-2014-0089
  • Schmid H-J. (2016). Entrenchment and the psychology of language learning. How we reorganize and adapt linguistic knowledge. Washington DC. https://doi.org/10.1037/15969-000
login to comment