CC BY
50
Научная статья
УДК 81-114.2
DO110.52070/2542-2197_2022_6_861_48
Распознавание окказиональных метафор при чтении. Работает ли эффект прайминга?1
М. И. Киосе1, В. А. Ярикова2
1,2Московский государственный лингвистический университет, Москва, Россия
Аннотация. Представлены результаты эксперимента, тестирующего глазодвигательное поведение при чте-
нии предложений, содержащих окказиональные метафоры. Установлено, что определенные параметры пространства источника и цели метафоры (агентивность, «оживленность», телесная семантика) увеличивают глазодвигательную нагрузку. Эффект прайминга, заключающийся во влиянии типа пространства ранее введенной в текст метафоры на чтение последующей метафоры, обнаружен не был.
Ключевые слова: окказиональная метафора, прайминг, пространство-источник, пространство-цель, окулографичес-кий эксперимент
Для цитирования: Киосе М. И., Ярикова В. А. Распознавание окказиональных метафор при чтении. Работает ли эффект прайминга? // Вестник Московского государственного лингвистического университета. Гуманитарные науки. 2022. Вып. 6 (861). С. 48-56. DOI: 10.52070/2542-2197_2022_6_861_48
Original article
Novel Metaphors Recognition in Reading. Do We Face Priming Effects?2
Maria I. Kiose1, Valeria A. Yarikova2
1,2Moscow State Linguistic University, Moscow, Russia 1 [email protected] [email protected]
Abstract.
Keywords: For citation:
This work presents the results of the eye tracking experiment testing the gaze behavior of participants reading the sentences containing novel metaphors. The results show that several types of source and target domains (agentivity, animateness, bodily semantics) of conceptual metaphors produce higher gaze costs. Meanwhile, the priming effects of these domains in case they reappear in novel metaphors in post-position are not observed.
novel metaphor, priming, source domain, target domain, eye tracking experiment
Kiose, M. I., Yarikova, V. A. (2022). Novel metaphors recognition in reading. Do we face priming effects? Vestnik of Moscow State Linguistic University. Humanities, 6(861), 48-56. 10.52070/2542-2197_2022_6_861_48
1 Разделы статьи «Metaphor recognition and priming effects» and «Results and discussion» подготовлены М. И. Киосе в рамках проекта РНФ 2228-01754 «Исследование экономии когнитивных ресурсов человека при интерпретации медиатекстов: Разработка Мультимодального Корпуса Окулографических Реакций MultiCOR», реализуемого в Московском государственном лингвистическом университете.
2 The sections «Metaphor recognition and priming effects» and «Results and discussion» were written by Maria I. Kiose as part of the Project 2228-01754 "Cognitive load economy in media texts interpretation: Multimodal Corpus of Oculographic Reactions MultiCOR" supported by the Russian Science Foundation, which is carried out at the Centre for SocioCognitive Discourse Studies at Moscow State Linguistic University
INTRODUCTION
How readers cope with the use of novel figurative Language has long interested researchers working on cognitive development and Language acquisition; however, in recent years this direction has become one of the main foci in media studies due to the interesting facts the figurative Language ability [Falcum, 2019] reveals about the relations of the readers' linguistic knowledge and reading competence. In the present article, we hope to achieve a degree of reconciliation between the two explanations which address figurative language recognition. The first holds that metaphor recognition depends on the types of input spaces or domains since they structure the world differently [Thibaut, Schyns, 1995]. The contrasting line of research emphasizes the role of priming effects in activating a metaphor in context [French, 2008]. Priming occurs due to prior exposure to a semantically or phonologically related stimulus; it can be observed in reading tasks and registered oculographically [Falikman, Koifman, 2005; Velichkovsky et al., 2020]. Priming in gaze behavior can appear in the repeated use of the types of input domains in the chain of metaphors in context since such cases fall within the group of semantic priming effects [Odekar et al., 2009]. Presumably, the use of similar domains in preposition will decrease the cognitive load produced by metaphors in post-position.
Additionally, metaphor recognition is highly dependent on its character, novel or entrenched. Activating an entrenched metaphor is less demanding; whereas language comprehension which creates new ad hoc categories unfolds via enhanced activation of only those features of input domains that are relevant for the novel interpretation of the lexical item [Wolff, Gentner, 2011]. Therefore, in this study, we will consider only novel metaphors recognition as being sensitive to both types of input domains and their priming effects in context. The aim of the work is to test two hypotheses: 1) the features of source and target domains of novel metaphors influence the readers' gaze behavior, and 2) their semantic priming effects stimulate the decrease in gaze costs.
METAPHOR RECOGNITION AND PRIMING EFFECTS
An issue of importance in metaphor recognition studies is the role of metaphor mapping models in metaphor processing. The features of input domains, or source and target domains are extensively explored in Conceptual Metaphor Theory, CMT,
following Lakoff and Turner, which considers them as components of the mapping process, for instance in DEATH IS A REAPER (where people are plants to be harvested, etc.) and LIFE IS A JOURNEY (where difficulties are obstacles, objectives are destinations) [Lakoff, Turner, 1989]. While these elements activate different mapping models, presumably, they may also stimulate different gaze effects and the semantic priming effects in metaphor recognition in case they appear repeatedly in sequential metaphors in context. As known, semantic priming is displayed in the increase in reading pace or in the accuracy in post hoc reports produced by a stimulus such as a word or a picture, when it is preceded by a semantically related stimulus (e. g., cat-dog) as opposed to the situations when it is preceded by a semantically unrelated stimulus (e. g., table-dog). The stimulus to which responses are made (e. g., dog) is termed the target and the preceding stimulus (e. g., cat) is the prime [McNamara, 2005]. Following Falikman & Koifman (2005) we consider that priming effects are contingent on attention distribution and different types of information processing and can be determined in problem-solving tasks. One of the methods testing the priming effects is eye tracking which allows to explore gaze behavior in terms of the gaze characteristics such as reading time, fixation duration, revisits. The studies exploring priming effects in eye tracking experiments most commonly use the proportion of fixation duration in the areas of interest (AOIs); the average fixation duration for fixations allocated in AOIs, and first fixation duration measures [Odekar et al., 2009]. Importantly, different studies investigating priming effects in gaze behavior report the results which do not admit the presence of consistent semantic priming. For instance, priming effects can appear both in related and unrelated conditions, but with a difference in gaze characteristics when specific positive and / or negative target words are used [Hyona, 1993; Velichkovsky et al., 2020] or when two associates are in the same clause [Carroll, Slowiaczek, 1986]. Still, the experimental studies claim that priming effects are greater in the situations of related primes and targets which in lexical semantic tasks occurs when a word (target) is semantically related to the preceding word (prime).
The question remains, will the semantic priming occur in case of novel metaphor use in context with the primes being the elements activated in the metaphor mapping models? Will the reappearing semantic elements produce lower gaze costs in the metaphors in post-position? Evidence from experimental psychology helps confirm the costs in reaction time which are produced when people
switch from one metaphor mapping model to another [Wolff, Gentner, 2011]. In Camac & Glucksberg it was shown that metaphor comprehension was faster with related pairs of metaphors than with scrambled ones, still there was no priming effect between semantic elements as metaphor mapping models constituents [Camac, Glucksberg, 1984]. A later experiment exploring semantic elements priming effects in metaphor mapping models, produced more detailed results [Nakamoto, 2003]. Metaphor recognition was facilitated if the vehicle (the term used metaphorically) served as prime and the topic (the subject of the metaphor) served as target. In contrast, if the topic preceded the vehicle, no priming effect was found. In this eye tracking study, we specify the priming effects of semantic elements repeatedly activated in the mapping models, as well as the effects of these elements on the gaze behavior during metaphor recognition.
EXPERIMENT DESIGN AND DATA
In the oculographic experiment the text stimuli with 7 sentences each containing a novel metaphor and the first sentence containing a non-metaphoric expression given in bold type in Areas of Interest (AoIs) was presented for all subjects for a fixed time. 14 students (age range = 19-28) from Moscow State Linguistic University participated in the experiment. 114 probes were obtained and subjected to further analysis. The participants faced with a lexical
decision task to identify the correct reference of a novel metaphor. Each sentence containing a novel metaphor was followed by a decision task with three options; the participants had to identify a correct variant by reading it aloud. The stimulus is shown in Figure 1.
All the sentences manifested in the stimulus were selected from the National Corpus of the Russian language (NCRL), where the novelty of metaphors was identified following the same procedure:
Step 1. If the noun was used figuratively, we addressed the dictionary to determine its definitions. Step 2. If the meaning exploited in the context was found out within the list of noun definitions, even if it was marked as figurative, we did not consider this word as a novel metaphor and further addressed only the examples with the nouns whose meanings were not registered in the dictionary. Step 3. If the noun did not display the meaning registered in the dictionary, we initiated a new National Corpus of the Russian language search with this noun as a target word and identified the meanings of its first 100 uses, with each use appearing the first on the new page and having a new author. Step 4. If we did not find out the corpus uses of the noun exploiting the same meaning as in the sentence under consideration, we regarded the metaphor as novel.
To be included into the stimulus, the novel metaphors had to satisfy the following requirements. They had to be nouns not exceeding 7 letters (to
И ОКАЗЫВАЛОСЬ, ЧТО КАЖДОЕ ДЕРЕВО, КАЖДЫЙ КУСТ И ДАЖЕ САМЫЙ МАЛЕНЬКИЙ ЦВЕТОК ИМЕЮТ СВОЕ ИМЯ ИСТОРИЮ. ЦВЕТОК ЧЕЛОВЕК НЕТ ОТВЕТА
НА ДРУГОЙ СТОРОНЕ ПЛОЩАДИ ПОЯВИЛОСЬ МНОГО НОВЫХ РАЗВЯЗОК И ЗСТАКАД, И ЭТО БЕТОННОЕ КОЛЬЦО ГРОЗИЛО ОХВАТИТЬ ВЕСЬ ГОРОД. ЭСТАКАДЫ ВЛАСТИ НЕТ ОТВЕТА
ВСЕ ЭТИ КЛИЕНТЫ БЫЛИ ДЛЯ НЕЕ СЛОВНО ПАПКИ, ТАКИЕ ЖЕ БЕЗЛИКИЕ. СЕГОДНЯ ОН ВНОВЬ РАССПРАШИВАЛ ЕЕ ПО ПОВОДУ ТОЙ ПАПКИ, КОТОРАЯ ПРИХОДИЛА НА ПРОШЛОЙ НЕДЕЛЕ. ДОКУМЕНТ КЛИЕНТ НЕТ ОТВЕТА
ВОТ ЭТО ЗАСТЫВШЕЕ ЛИЦО ОН СИЛЬНО СТАРАЛСЯ ЗАБЫТЬ, ПОТОМУ ЧТО ПАМЯТЬ ЭТО НАСТОЯЩАЯ СЕТЬ, КОТОРУЮ НЕ СЛЕДУЕТ ЧЕРЕСЧУР НАПРЯГАТЬ, ЧТОБЫ УДЕРЖИВАТЬ ТЯЖЕЛЫЕ ГРУЗЫ. ПАМЯТЬ ГРУЗ НЕТ ОТВЕТА
НОВИКОВ ПОШЕЛ ПО ЖЕЛЕЗНОДОРОЖНЫМ ПУТАМ, БОЯСЬ, ЧТО ЭШЕЛОН УЖЕ УШЕЛ. КАК ОКАЗАЛОСЬ, ЭТА ИГЛА МЕШАЛА ЕМУ СОСРЕДОТОЧИТЬСЯ НА
МЫСЛЯХ О ПРЕДСТОЯЩЕМ БОЕ.
СТРАХ ПОЕЗД НЕТ ОТВЕТА
КОГДА-ТО ГОДЫ И ГОРОДА ПОДПОЛЬЯ, ССЫЛКИ ВОЙНЫ БРОСАЛИ ЕМУ НАВСТРЕЧУ СОТНИ РЕДКОСТНО ПРЕКРАСНЫХ ЛЮДЕЙ, И ЭТО МОРЕ ОКРУЖАЛО ЕГО
И ДАВАЛО СИЛЫ ЖИТЬ.
ЛЮДИ ГОРОДА НЕТ ОТВЕТА
ЗИМНЯЯ ДОРОГА. РОВНАЯ, ТВЕРДАЯ, ГЛАДКАЯ: БЕЛЫЙ ФАРФОР. ХРУСТАЛЬ ДОРОГА НЕТ ОТВЕТА
ОН УМЕЛ УЛЫБАТЬСЯ, И, КОГДА УЛЫБАЛСЯ, ТО ЭТО БЫЛ АНГЕЛ, ВТИРАЛСЯ В ДОВЕРИЕ К ЛЮДЯМ И ПОТОМ ИХ ОБМАНЫВАЛ. ЧЕЛОВЕК СВЯТОЙ НЕТ ОТВЕТА
Figure 1. Stimulus
fit into the window span), to have high frequency of corpus use (in direct meaning) to avoid different salience effects [Kiose, 2020], to have a direct non-metaphoric counterpart in pre-position. Still, the metaphors appeared in different syntactic positions (subject and predicate positions), in different positions on the reading line (initial, final and mid-position), were preceded or not preceded by a preposition adjective or pronoun. These effects were intended to alleviate syntactic and graphological priming since it could prevent us from observing the effects caused by lexical semantic primes. Sentence 1 did not contain a metaphor; however, we also introduced the bold type for the noun group to prepare the reader for the task. In each sentence the metaphor received annotation describing its reference and event frame in both source domain and target domain separately.
Following the theories developed in cognitive semantics [Siewerska, 2004; Iriskhanova, 2014], we annotated 6 reference groups; referent agentivity, number, reference, referential integrity, personalization. For instance, the group Agentivity involved Agentive participant (e. g., in the source domain PERSON in той папки, которая приходила на прошлой неделе "that file which came last week"), Object / Recipient (e. g., in the target domain DOCUMENT in той папки, которая приходила на прошлой неделе "that file which came last week" since it could not arrive into the office on its own, it was probably brought in there), Patient / Inactive participant (e. g., both source domain ROAD and target domain CHINA in Зимняя дорога. Ровная, твердая, гладкая: белый фарфор "Winter road. Even, solid, smooth: white china"), contrasted as more and less agentive. Below, we will consider the reference types in source and target domains of novel metaphors.
Despite the fact that sentence 1 did not contain a metaphor, it still activated semantic elements which might provoke further priming effects, that is they might serve as primes facilitating the recognition of novel metaphors in post-position. Therefore, we also considered its semantic elements, but did not differentiate between the source and target domains. In terms of reference types in the source domains of 7 novel metaphors and one noun in sentence 1, we observed variance in the following semantic elements: in Agentivity group we had Agentive participant in AOIs2 3 5 6 8, Object / Recipient in AOIs3 4 7 and Patient / Inactive participant in AOIr in Reference group we had Person-participant in AOI8, Animated object (e. g., nature object) in AOIr
Inanimate participant / object in AOIs,
„. In terms
of reference types in the source domains of 7 novel metaphors and one noun in sentence 1, we observed
variance in the following semantic elements: in Agentivity group we had Agentive participant in AOIs2 3 5 6 8, Object / Recipient in AOIs3 4 7, Patient / Inactive participant in AOIr in Number group we had Single participant / object in AOIst 4 5 7 8, Three and more participants / objects in AOIs2 3 6; in Reference group we had Person-participant in AOIs3 6 8, Animated object (e.g., nature object) in AOIr Inanimate participant / object in AOIs2 3 6 7 8, Abstract object in AOIs4 5; in Referential integrity group we had Integral participant / object in AOIst 3 4 5 6 8 and Component of participant or object in AOIs2 7, Facial movements of participant in AOI5. Overall, the variance of semantic elements as components of target domains was much higher, still the question remains whether any of them influences the gaze behavior.
Following the theories of narratology, syntactic semantics and cognitive semantics [Talmy, 2000], we annotated 12 event frame groups, truth, type, relations, manageability, completeness, instantness, achievement, evaluation, space location, time location, repeatability, cause and effect. For instance, the group Type includes Creation / destruction (e.g, in the target domain HIGHWAYS in это бетонное кольцо грозило охватить весь город "this concrete ring threatened to capture the city"), Shifting / changing (e.g., in the target domain PERSON in втирался в доверие к людям и потом их обманывал "he stimulated people to trust him and then deceived them"), Perception (Sensory) (e. g., in the target domain OBSESSION in эта игла мешала ему сосредоточиться на мыслях "this needle prevented him from concentrating on his thoughts"), Lack of action or dynamics (e. g., in the target domain CHINA in Зимняя дорога. Ровная, твердая, гладкая: белый фарфор "Winter road. Even, solid, smooth: white china"). Below, we will consider the event frame types in source and target domains of novel metaphors in the stimulus.
In terms of event frame types in the source domains of 7 novel metaphors and one noun in sentence 1, we observed variance in all semantic elements apart from Manageability and Instances group; therefore, we will not list all the differences but enumerate only several of them. For instance, in Type group we had Shifting / changing event frame in AOIs2 3 4 5 6 8, Perception in AOIs2 5 6 8, Lack of action or dynamics in AOIst 7. In Relations group we had Interpersonal relations in AOI8, Interactive relations in AOIs3 5 6, No interpersonal or interactive relations in AOIst 2 3 4 5 7. Several event frame types co-occured within one sentence because the sentence manifested two of more propositions employing several predicates. For instance, in sentence 3 все
эти клиенты были для нее словно папки, такие же безликие. Сегодня он вновь расспрашивал ее по поводу той папки, которая приходила на прошлой неделе "all these clients resembled files having no personal faces. Today he again asked her about the file which came in last week", we observed two noun-predicate groups, he asked about the file and the file came in. The first group displayed interpersonal relations, while the second did not; at the same time, they both manifested Shifting / changing event frame. In terms of target domains, we observed variance in all groups apart from Manageability, Instantness; so, there were no differences in target and source domains variance. However, the types of semantic elements within the metaphor mapping models were different.
The stimulus was presented on a 21-inch monitor. The SMI Red-x eye tracker was applied in the experiment (binocular system, frequency = 60 Hz, accuracy = 0.4°, head movement 40x20 cm, operating distance = 60-80 cm). The participants were instructed that they should give their responses choosing one of the three response variants (with the first two naming the source and target domain, the third claiming that there is no answer; so only one variant is correct). Since in this article we will consider only the gaze behavior of the experiment
participants, we will not present the analysis of their reports which might have also been influenced by semantic priming.
RESULTS AND DISCUSSION
Prior to gaze data preparation we had to decide which gaze characteristics should have been considered. We mostly regarded the characteristics specified in the studies focusing on exploring the priming effects [Odekar et al., 2009] and tested the following characteristics: fixation duration in the areas of interest (AOIs), the average fixation duration for fixations allocated in AOIs, first fixation duration measures; the average first fixation duration, and revisits. The following gaze paths (Figure 2b) show that these gaze characteristics displayed variance in different AOIs. In Figure 2a we present the AOIs, so that they could be traced in the gaze paths.
Overall, the gaze paths showed that several AOIs were skipped. Therefore, we had to process the data only of the participants who did not skip the AOIs in each pair of sentences since we had to consider the priming effects of the semantic elements in the mapping models in pre-position.
ИОКАЭЫВАЛОСЬ.что КАКДОС Д№СDO. КАЖДЫЙ KVCTи дажс самыймаяекькийИНТ^мсогск* имя и историю.
ЦВЕТОК ЧЕЛОВЕК НЕТ ОТВЕТА
ИАД^ЧОГЮОКОШ. I 1/ЮЩАДИП0миЛ10СЬМ1Ю(01 iCXbUFAjlWjOKH XlAWW. И ЛОТ Ы IChkoJ 'dlturt jfOJtl/lOOJ.JAIHIb
hicuokw. I— I
ЭСТАКАДЫ ВЛАСТИ ИСТ ОТВЕТА
Я1Ч чту клиенты ьыли ДЛЯ ни CiWHHQ ПАПКИ, ТАК/16 ЖЕ fctJflUKHt CU О ДНЯ ОН В*КЖ> РАССПРАШИВАЛ Н ПО ПОВОДУ ТОЙ t fjtfJlj j КОТОРАЯ ПРИХОДИЛ* ЬА ПРОШЛОЙ НЕДЕЛЕ. ПОКУМСНТ КЛИЕНТ НЕТОТВОА
■ОТ это JACTWBlUf Е ЛИЦО ОН СИЛЬНО СТАРАЛСЯ ЗАБЫТЬ. ПОТОМУ ЧТО ПАМЯТЬ это млстоящАО! ООАоторую НЕ СЛЕДЯТ ЧЕРЕСЧУР НАПРЯГАТЬ, ЧТОВЫ УДЕРЖИВАТЬ ТЯЖЕЛЫЕ ГРУЗЫ ПАМЯТЬ ГРУЗ НЕТ ОТВЕТА
НЛЧИКОЙ "ЮШГЛ ПО ЖЕЛЕЗНОДОРОЖНЫМ ПУТЯМ, КОЯ О,, ЧТО ЭШЕЛОН УЖЕ VlriFi. КАК ОКУПАЛОСЬ. 009" 11АЛЛ FMV СОСРГЛОТОЧНТНСЯ НА мкел«X О ПРГ/ЧТОЯИIFMfiOf. СТРАХ ПОЕЗД ИСТ DTBCTА
, кпмг-го годы к города подпол м. ссылки войны бросали jr.-v навстречу сотни редкостно прекрасных людей, н это АФоДО&кРУжДЛО ЕГО и давало силы жить.
ЛЮДИ ГОРОДА НЕТ ОТВЕТА _
ЗИМНЯЯ ДОРОГА РОСНАЯ. ТВЕРДАЯ. ГЛАДКАЯ: БЕЛЫЙ V« »1ЧН»Р| ХРУСТАЛЬ ДОРОГА HITОТ8СТА [ [
ОН УМЕЛ УЛЫСАТЬСЯ. II. КОГДА УЛЫЕАЛСЯ, ТО >70 ОТИРАЛ СЯОДСОЕРИЕ К ЛЮДЯМ И ЛОТОМ ИХОЕМАНЫВАП.
ЧЕЛО Si К СВЯТОЙ НЕТ ОТЛЕТА
GAZE DATA
Table 1
AOIt aoi2 AOI3 AOL 4 aoi5 AOI6 aoi7 AOIs
Dwell Time, ms 98.1 176.2 229.2 124.1 173 260.3 122.8 155.4
Revisits 0.3 0.4 1.1 0.1 0.4 0.6 0 0.5
Average fixation, ms 76 106.5 80.2 90.7 88.6 124.4 56.2 58.4
First fixation, ms 78.2 107.4 79.6 91.1 93.1 117.7 59.7 63.2
Fixation count 0.6 1.1 1.5 0.8 1 1.5 0.6 0.6
Figure 2. 2a - Areas of Interest in the Stimulus, 2b - one of the gaze paths
To test hypothesis one, which claimed that the features of source and target domains of novel metaphors influence the readers' gaze behavior, we applied the data produced in AOIs Key Performance Indicators Regime. In this case we could identify the differences in gaze behavior in relation to 8 AOIs. In Table 1 we present AOIs Key Performance Indicators for 8 AOIs.
The gaze data shows that there are differences in all characteristics; however, since Revisits would be aliased with Fixation count, and Dwell time would be dependent on both Average fixation and Fixation count, we considered three gaze characteristics, Dwell time, Revisits and First fixation duration. Besides, to avoid collinearity in semantic element effects onto gaze characteristics, we conducted statistical tests separately for two data blocks, for reference semantic elements and for event frame semantic elements, to establish their contingencies on gaze characteristics. The procedure of regression analysis was applied, which allowed first, to identify the predictability of gaze characteristics produced by semantic elements of metaphor mapping models, and second, to identify the semantic elements as best predictors of gaze behavior. Overall, we conducted 12 tests, with 6 tests for regression models of three gaze characteristics and two semantic elements groups of reference and event frame for source domains and target domains.
SIGNIFICANT RESULTS WERE OBTAINED IN THE FOLLOWING CASES
a) Source domain. Significant results, according to the Dwell Time (model value R2 = 1), were obtained in Agentive participant (Estimate (E)=154.7, p<.001), Object / Recipient (E=54.6, p=.001), Inactive participant/ Animated object (E=-56.5, p=.003), Person-participant (E=-19.2, p=.012), Inanimate participant / object (E=-85.7, p<.001). Significant results, according to the First Fixation Duration (model value R2 = .999), were obtained in Shifting / changing (E=32.1, p=.037), Perception (E=14.9, p=.06), Lack of action or dynamics (E=-67.4, p=.023), Interactive relations (E=-23.9, p=.05). Significant results, according to the Revisits (model value R2 = 0.994), were obtained in Agentive participant (E=1.05, p=.003), Object / Recipient (E=0.7, p=.008), Inactive participant/ Animated object (E=0.75, p=.015).
b) Target domain. No significant results, according to the Dwell Time, were obtained. Significant results, according to the First
Fixation Duration (model value R2 = .995), were obtained in Single participant / object (E=-51.57, p=.05). Significant results, according to the Revisits (model value R2 = .99), were obtained in Agentive participant (E=0.85, p=.018), Object / Recipient (E=0.53, p=.023), Inactive participant/ Animated object (E=0.78, p=.02).
The results show that the semantic elements in metaphor mapping models affect gaze behavior in different ways. More significant results were obtained in the source domain than in the target domain; this is partly explained by higher variance of semantic elements in target domains. At the same time, the results prove that semantic elements displaying different reference types (in contrast with event frame types) produce higher effects, with their highest values displayed by Agentivity. The results prove the dominant and consistent role of several semantic elements in metaphor mapping process earlier described in Lakoff & Turner (1989), Thibaut & Schyns (1995), Grady et al. (1999), however they extend them to show which particular elements produce higher gaze costs. We have provided evidence that most accessible referents as stated in [Siewerska, 2004] or more prominent ones as stated in [Iriskhanova, 2014] are the ones that require higher gaze costs. Interestingly, this effect appears in different types of metaphor mapping models, even in the ones which employ different event frames and different source and target domains. The second important result is the attested dominant role of the same feature, Agentivity or Dynamicity as a semantic element of both reference and event frame. The results experimentally prove the methodological views expressed in [Talmy, 2000] where he stresses the importance of dynamicity in event construal.
To test hypothesis two, which claimed that the semantic priming effects of elements as mapping model components stimulate the decrease in gaze costs, we conducted two types of analysis.
First, we applied the gaze data obtained in AOIs Key Performance Indicators Regime to identify whether they displayed a subsequent decrease in gaze costs contingent on the effects of semantic elements repeated use in the preceding metaphor. To do it, we revealed the examples of repeated use of reference and event frame semantic elements. There were 4 examples in the sample which satisfied this condition. 3 examples were found in AOI2 and AOI3 in the semantic elements Single participant / object in source domain and its absence in target domain, and Three and more participants / objects in target domain and its absence in source domain,
and also Event located in space in target domain and its absence in source domain. In AOI4 and AOI5 1 example in the semantic element Inanimate participant / object present in source domain and absent in target domain was found. Three regression tests were made with three gaze characteristics, Dwell time, First Fixation duration and Revisits, and these variables. The highest model predictability was obtained with Revisits (R2=0.662), this model had the semantic elements of AOI2 and AOI3 as good model predictors (E=0.78, p=.026); however, other two regression models did not manifest high R2 or any good predictors. The results show that there are no steady effects of semantic priming on the gaze behavior in AOIs with novel metaphors. Presumably, these effects display a more individual character.
Second, we applied the data received from individual participants to see if their individual gaze behavior displayed the effects of semantic priming. Since several participants missed several AOIs in reading, two gaze characteristics which could be of use in this case were First Fixation Duration and Dwell Time in case a participant had not missed two consecutive AOIs with similar semantic elements in either source or target domains. Consequently, we used First Fixation Duration and Dwell Time in contingency tests (Student's t-tests) to identify the priming effects. First Fixation Duration in all AOIs falls within the range of 130-200 ms; however, its highest values are observed in AOI7 and AOI8, which seems unexpected since the priming effects were to have been at their highest. 7 Student's tests were then carried out which measured the contingency effects of First Fixation Duration in each pair of AOIs, AOIj and AOI2, AOI2 and AOI3, AOI3 and AOI4, AOI4 and AOI5, AOI5 and AOI6, AOI6 and AOI7,AOI7 and AOI8. We did not reveal any significant contingences in the tests, which means that combined semantic priming effects were less potent to produce any stable gaze costs decrease.
Next, we appealed to Dwell time, however in Student's t-tests we considered only the values in two successive AOIs which were not missed by the participants. 7 tests were carried out, still no significant contingencies were revealed, either. Overall, the second hypothesis has not been confirmed: the experiment did not prove the existence of semantic priming effects with novel metaphors. Several possible explanations to this fact can be produced. First, metaphor models displayed high diversity due to different correspondences in semantic elements belonging to source and target domains; this may have caused mixed effects which we have observed. Second, other types of priming, for
instance, syntactic or graphological priming might have influenced the results. Third, due to the fact that several participants did not have fixations in some AOIs, we had to discard their gaze behavior results in all AOIs, otherwise we could not observe the results in successive metaphors. Consequently, the data were limited. All facts considered, we did not observe the priming effect claimed in [Na-kamoto, 2003], although the stimuli AOIs satisfied the same conditions (the vehicles, e.g., развязки и эстакады 'interchanges and flyovers', preceded the topic, here, бетонное кольцо concrete ring'). At the same time, the results are consistent with those described in the earlier study [Camac, Glucksberg, 1984] which did not report the association effects caused by metaphors.
FINAL REMARKS
In the current study, we explored the priming effects caused by repeated semantic elements as components of correspondences in metaphor mapping models. We report the results of the eye tracking experiment where participants were exposed to a lexical task which stimulated their reading for comprehension; therefore, their cognitive costs were expected to bring out different gaze costs depending on the presence or absence of semantic primes in the metaphors in pre-position. The study tested two hypotheses. The first one claiming that the features of source and target domains of novel metaphors influence the readers' gaze behavior, was confirmed. The major semantic element influencing the increase of gaze costs was agentivity in reference and dynamicity in event frame. The second hypothesis claiming that the semantic priming effects of these features stimulate the decrease in gaze costs was not verified. We appealed to two gaze characteristics, First Fixation Duration and Dwell time to explore their contingency on semantic primes, still no significant differences were observed. Accordingly, further studies on semantic priming have to be carried on; presumably, the mixed effects of semantic elements within the mapping models should be alleviated to bring out more consistent results and the effects of other types of primes, syntactic and graphological, should also be balanced. As revealed in the experiment, there is propensity for several semantic elements to produce higher cognitive load which might lead to better metaphor recognition and consequently to semantic priming effects; still, the procedural ways to capture these effects do not allow to explore them extensively.
СПИСОК ИСТОЧНИКОВ
1. Falcum I. L. Metaphor and metonymy in acquisition: a relevance-theoretic perspective // Relevance, Pragmatics and Interpretation / K. Scott, B. Clark, R. Carston (eds.). Cambridge: Cambridge University Press, 2019. P. 205-217.
2. Thibaut J.-P., Schyns P. G. The development of feature spaces for similarity and categorization // Psychologica Belgica. 1995. Vol. 35. P. 167-185.
3. French R. M. Relational priming is to analogy-making as one-ball juggling is to seven-ball juggling // Behavioral & Brain Sciences. 2008. Vol. 31. P. 386-387.
4. Фаликман М. В., Койфман А. Я. Виды прайминга в исследованиях восприятия и перцептивного внимания. Часть 1 // Вестник МГУ. Серия 14. Психология. 2005. Вып. 3. С. 86-97.
5. Velichkovsky B. B. et al. Diffused model of reaction time onto recent negative probes / B. B. Velichkovsky, F. R. Sultanova, D. V. Tatarinov, A. A. Kachina // Experimental psychology. 2020. Vol. 13(1). P. 35-50.
6. Odekar A. et al. Validity of eye movement methods and indices for capturing semantic (associative) priming effects / A. Odekar, B. Hallowell, H. Kruse, D. Moates, C. Lee // Journal of speech, language, and hearing research. 2009. Vol. 52(1). P. 31-48.
7. Wolff P., Gentner D. Structure-mapping in metaphor comprehension // Cognitive Science. 2011. Vol. 35. P. 1456-1488.
8. Lakoff G., Turner M. More than cool reason: A field guide to poetic metaphor. Chicago: University of Chicago Press, 1989.
9. McNamara T. P. Semantic priming: Perspectives from memory and word recognition. Psychology Press, 2005.
10. Hyona J. Effects of thematic and lexical priming on readers' eye movements // Scandinavian Journal of Psychology. 1993. Vol. 34(4). P. 293-304.
11. Carroll P., Slowiaczek M. L. Constraints on semantic priming in reading: A fixation time analysis // Memory and Cognition. 1986. Vol. 14. P. 509-522.
12. Camac M. K., Glucksberg S. Metaphors do not use associations between concepts, they are used to create them // Journal of Psycholinguistic Research. 1984. Vol. 13(6). P. 443-455.
13. Nakamoto K. Semantic priming effect of metaphor constituent terms // Perceptual and motor skills. 2003. Vol. 96. P. 33-42.
14. Kiose M. I. The interplay of syntactic and lexical salience and its effect on default figurative responses // Studies in logic, grammar and rhetoric. 2020. Vol. 61(74). P. 161-178.
15. Siewerska A. On the discourse basis of person agreement // Approaches to cognition through text and discourse / ed. by T. Verhagen. Berlin, New York: Mouton de Gruyter, 2004. P. 33-48.
16. Iriskhanova O. K. Games of focus in language. Moscow: Languages of Slavic culture, 2014.
17. Talmy L. Toward a cognitive semantics: in 2 vols. Cambridge: MIT Press, 2000.
REFERENCES
1. Falcum, I. L. (2019). Metaphor and metonymy in acquisition: a relevance-theoretic perspective. In K. Scott, B. Clark, R. Carston (eds.), Relevance, Pragmatics and Interpretation (pp. 205-217). Cambridge: Cambridge University Press.
2. Thibaut, J.-P., Schyns, P. G. (1995). The development of feature spaces for similarity and categorization. Psychologica Belgica, 35, 167-185.
3. French, R. M. (2008). Relational priming is to analogy-making as one-ball juggling is to seven-ball juggling. Behavioral & Brain Sciences, 31, 386-387.
4. Falikman, M. V., Koifman, A. Y. Vidy prajminga v issledovanijah vosprijatija i perceptivnogo vnimanija = Types of priming in perception and perceptive attention studies. Part 1. Moscow University Psychology Bulletin, 3, 86-97. (In Russ.)
5. Velichkovsky, B. B., Sultanova, F. R., Tatarinov, D. V., Kachina, A. A. (2020). Diffused model of reaction time onto recent negative probes. Experimental psychology, 13(1), 35-50.
6. Odekar, A., Hallowell, B., Kruse, H., Moates, D., Lee, C. (2009). Validity of eye movement methods and indices for capturing semantic (associative) priming effects. Journal of speech, language, and hearing research, 52(1), 31-48.
7. Wolff, P., Gentner, D. (2011). Structure-mapping in metaphor comprehension. Cognitive Science, 35, 1456-1488.
8. Lakoff, G., Turner, M. (1989). More than cool reason: A field guide to poetic metaphor. Chicago: University of Chicago Press.
9. McNamara, T. P. (2005). Semantic priming: Perspectives from memory and word recognition. Psychology Press.
10. Hyona, J. (1993). Effects of thematic and lexical priming on readers' eye movements. Scandinavian Journal of Psychology, 34(4), 293-304.
11. Carroll, P., Slowiaczek, M. L. (1986). Constraints on semantic priming in reading: A fixation time analysis. Memory and Cognition, 14, 509-522.
12. Camac, M. K., Glucksberg, S. (1984). Metaphors do not use associations between concepts, they are used to create them. Journal of Psycholinguistic Research, 13(6), 443-455.
13. Nakamoto, K. (2003). Semantic priming effect of metaphor constituent terms. Perceptual and motor skills, 96, 33-42.
14. 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.
15. Siewerska, A. (2004). On the discourse basis of person agreement. In T. Verhagen (ed.), Approaches to cognition through text and discourse (pp. 33-48). Berlin, New York: Mouton de Gruyter.
16. Iriskhanova, O. K. (2014). Games of focus in language. Moscow: Languages of Slavic culture.
17. Talmy, L. (2000). Toward a cognitive semantics: in 2 vols. Cambridge: MIT Press.
информация об авторах Киосе Мария Ивановна
доктор филологических наук, доцент, ведущий научный сотрудник Центра социокогнитивных исследований дискурса Московского государственного лингвистического университета
Ярикова Валерия Алексеевна
студент Московского государственного лингвистического университета
INFORMATION ABOUT THE AUTHORS Kiose Maria Ivanovna
Doctor of Philology (Dr. habiL.), Assistant Professor, Leading research scientist, Moscow State Linguistic University
Yarikova Valeria Alekseevna
Student, Moscow State Linguistic University
Статья поступила в редакцию 12.04.2022 одобрена после рецензирования 11.05.2022 принята к публикации 16.05.2022
The article was submitted 12.04.2022 approved after reviewing 11.05.2022 accepted for publication 16.05.2022
