Cases and Thematic Roles. Ergative, Accusative and Active


Free download. Book file PDF easily for everyone and every device. You can download and read online Cases and Thematic Roles. Ergative, Accusative and Active file PDF Book only if you are registered here. And also you can download or read online all Book PDF file that related with Cases and Thematic Roles. Ergative, Accusative and Active book. Happy reading Cases and Thematic Roles. Ergative, Accusative and Active Bookeveryone. Download file Free Book PDF Cases and Thematic Roles. Ergative, Accusative and Active at Complete PDF Library. This Book have some digital formats such us :paperbook, ebook, kindle, epub, fb2 and another formats. Here is The CompletePDF Book Library. It's free to register here to get Book file PDF Cases and Thematic Roles. Ergative, Accusative and Active Pocket Guide.
Teaching & Learning Guide for: Typology of Ergativity

For example, although agent, patient , and instrument are English words, there is only a loose connection between the commonly known meanings of these words and the event categories that researchers have labeled Agent, Patient, and Instrument. Instead, role categories have been thought to reveal themselves through a range of linguistic structures, both lexical e. In addition, lexical markers of thematic roles are often closed-class and have wide semantic extension, seemingly encompassing multiple meanings and multiple thematic roles. The English preposition with , for example, marks instruments, as in 1a , but it also ranges over other roles, as in I want the room with two beds , Marnie delivered the talk with confidence , and I walked to the store with my brother.

Given that event participant categories are not as self-evident as categories provided by nouns and verbs, we review studies demonstrating the psychological reality of abstract roles. For example, the sentence the journalist checked the spelling is processed more quickly than the mechanic checked the spelling Bicknell et al. Given this detailed event- and verb-specific knowledge, in combination with the difficulty of defining and identifying thematic roles, the question arises as to whether abstract participant categories have a place in a theory of cognition.

The analysis in Croft , for example, involves direct mappings between specific event structures and syntactic positions e. In addition, Rissman and Rawlins analyze the meanings of the English instrumental markers with and use only in terms of the roles Agent and Patient, without appealing to the role Instrument. Theories of language and cognition that do without thematic roles, or employ a more restricted set of roles, may be simpler and therefore preferred. It is crucial, therefore, to put forward evidence that proposed thematic roles have psychological reality.

Such evidence is found in experiments on event and sentence processing by adults and children, which we review below. The second step of our argument is to address whether abstract roles in specific languages and cultures are shaped by universal biases in conceptual and semantic structure. Fillmore , for example, proposed that the repertoire of role categories is the same across languages.

This hypothesis is not possible to test directly at least not yet; see fifth section , because experimental studies demonstrating abstract roles have not been conducted in every language. Nonetheless, we review four types of evidence that bear on this question indirectly: linguistic typology, studies of conceptual representation in infancy, sign language emergence, and studies of child language development.

As core knowledge is a source of universal cognitive biases, this evidence can be used to support the claim that thematic roles are part of core knowledge. Linguistic typology involves comparing large sets of diverse languages. Through patterns of colexification across languages, this research sheds light on the extent to which there are universal biases to categorize event roles: when a single adposition or case marker is frequently used to encode two distinct meanings, this indicates that the two meanings are semantically related; when two meanings are frequently marked in different ways linguistically, this indicates that the meanings are semantically more distant.

This colexification evidence thus shows common patterns in what types of roles are often distinguished and what roles are often conflated. As the very nature of categories is to distinguish in-group from out-group members, these patterns suggest what notions are regularly distinguished in diverse languages and consequently give clues as to what the categories in individual languages are likely to be.

Common patterns cross-linguistically could reflect cognitive biases, social structure, historical and environmental influences, or some combination of these factors. We therefore review evidence from infancy, emerging sign language, and child development: developmental data are critical to reveal notions distinguished early in life before language has been fully acquired, and thus provide insight into the cognitive biases that individuals begin their learning with.

For example, representations that develop early in life are likely to reflect fundamental biases that structure cognition. So studies with infants can reveal conceptual knowledge about event roles that is in place before children have fully learned the grammar of any particular language. Studies of newly emerging sign languages also reveal cognitive predispositions regarding concepts that lend themselves to be readily encoded in language, but in this case in the context of language being created rather than learned.

Finally, child language learning can reveal whether particular role categories are more salient than others. In this paper, to the contrary, we suggest the answer to this question may differ depending on the particular role considered. For this reason, we review three types of roles, addressing whether there is evidence for abstract categorization, as well as evidence for universal cognitive biases. We find strong evidence for a universal bias to distinguish agents and patients.

Here we address the question of whether there is a universal bias for goals to be categorized more robustly than recipients, an asymmetry that would support the hypothesized primacy of spatial cognition. We find minimal evidence for such a bias, raising the possibility that categorization of Goals and Recipients is more variable than categorization of Agents and Patients. We discuss instrumental events in the fourth section, which are typically analyzed as an Agent acting on an Instrument in order to affect a Patient.

We ask whether there is a universal bias to construct an instrument category around the notion of a tool, a physical object wielded by an agent in order to achieve an outcome. Across the three types of roles considered here, we find the weakest evidence that universal biases shape role categorization for instrumental events. In addition to questions of category abstraction and universal biases, a third crucial question for theories of thematic roles is how these categories are structured. The experimental work critical to prototype theories of objects has largely not been conducted with thematic roles, for the simple reason that word meanings provide only an indirect cue to the structure of role categories.

In the fifth section, we discuss how this gap could be filled in the future. In the fifth section, we also provide a roadmap of the studies that should be conducted to fully test the nativist view of thematic roles. We argue in this paper that for multiple roles, there is only weak evidence for the nativist view. We stress, however, that the literature is more often characterized by absence of evidence the critical studies have not been conducted rather than negative evidence findings contrary to the nativist view.

We conclude this paper by discussing what some of these critical studies would be. At the same time, thematic roles are also event participant categories, which although reflected in language through specific linguistic structures, are not exclusively represented at a linguistic level. It therefore follows that evidence from sources outside of the confines of linguistics — including studies of infant and adult event cognition — are pertinent to understanding the full nature of thematic roles. Logically it is possible, of course, that there are multiple domain-specific thematic role representations, one that modulates syntactic behavior and another that modulates conceptual event representation, for example.

In this review, we assume as a null hypothesis that, even if there are such domain-specific roles, there is also a domain-general system of event participant categories, and both linguistic and non-linguistic behavioral evidence is relevant toward understanding the nature of these categories.

For example, adults rapidly extract agent and patient role information from visual presentation of events. Hafri, Papafragou, and Trueswell presented English-speaking adults with scenes of transitive events, such as a girl pushing a boy, for brief intervals either 37 or 73 ms. Although participants could succeed at this task if they only extracted event-specific roles i.

The findings of Hafri, Trueswell, and Strickland support this interpretation. Participants viewed simple transitive events and performed a task unrelated to roles they had to spatially locate a particular gender or shirt color of the people in the event. Nonetheless, participants were slower when the role Agent vs.

Patient of the target e. This indicates that people extract role categories specifically Agent and Patient categories during visual event apprehension, even when they are not asked to attend to event roles. For example, Mauner and Koenig found that readers could easily interpret a sentence like the vase was sold to collect money for the charity.

As the rationale clause to collect money for the charity requires an agent semantically, this finding indicates that readers activated an Agent concept upon hearing the verb sold. By contrast, readers did experience difficulty interpreting the vase sold to collect money for the charity. Nonetheless, the idea that agent- and patient-like roles are assigned relatively rapidly during comprehension is uncontroversial.

Knowledge of abstract Agent and Patient categories is also manifest through studies of novel verb interpretation. For example, Kako asked English-speaking adults to read sentences without contentful open-class words, such as the rom mecked the zarg. Even for such seemingly meaningless sentences, participants rated the Subject as having more agentive properties than the Object, and the Object as having more patient-like properties than the Subject, indicating that abstract role information is linked to syntactic positions.

This array of evidence shows that agent-like and patient-like participants are represented in terms of abstract categories. The status of putative Agent and Patient roles is not necessarily identical, however. For a range of English verbs, White, Rawlins, and Van Durme clustered arguments by semantic features considered definitive of Agent and Patient roles, such as having intention and undergoing a change of state Dowty, While a clear Agent cluster emerged, the patient-like arguments clustered into four separate roles, subtypes of a more diffuse Patient role.

The idea that Patient is a more diffuse category than Agent is consistent with findings in Hafri et al. These authors hypothesized that agents, when physically instantiated, tend to be characterized by perceptual features such as outstretched limbs and leaning forward. By contrast, patients were hypothesized to be characterized by the absence of these features. For still images of an animate agent acting on an animate patient, agents were more consistently judged as having these features than patients were judged to not have them, suggesting that the Patient role is more heterogeneous than the Agent role.

Although intriguing, it is not clear to what extent this finding would generalize to other sorts of events or stimuli, such as events with inanimate agents. Behavioral effects from individual languages, such as priming from one type of event to another, constitute evidence for abstract event participant categories. As most of the studies reviewed in the section above were conducted with speakers of English, the question of whether the same abstract role categories are shared across the languages of the world has largely been unanswered.

Nonetheless, typological evidence provides an indirect clue as to what meanings are commonly distinguished across languages, and therefore how role categories are likely to be structured.

Aggregating evidence from a wide range of languages reveals a robust tendency to distinguish agents from patients. Siewierska sampled languages and found that among those languages that use verbal person marking to code arguments, none attested a single person marker for both agents and patients, while using a different person marker for the single argument of an intransitive verb.

In his study on case marking, Comrie found a similar tendency to distinguish agents from patients: across languages, no language received the same case marker for agent and patient arguments of a transitive verb, but a different case marker for the single argument of an intransitive verb see also Comrie, This dispreference for grouping agents and patients together may not be truly universal, however: Payne documents a case-marking pattern in which agents and patients are colexified for some Iranian languages, albeit for only some pronouns.

While such counter-examples suggest that distinguishing agents from patients is not a descriptive universal, these counter-examples are still compatible with a strong tendency to distinguish agents from patients. Finally, in a study of lexically-specific roles e. They used multidimensional scaling to analyze the similarity between each of the specific roles and found that agent-like roles e.

Before proceeding further, some caution is warranted. For this reason, the distinction between agents and patients that we see across diverse languages is not truly independent of the particular role categories in the minds of the researchers see Newmeyer, Nonetheless, the fact that these rough-and-ready notions were still highly likely to be encoded by two distinct forms across languages, and that other notions do not show the same proclivity as reviewed later , can be taken as indicative of a strong pan-human bias to draw this distinction.

While the typological evidence reveals a bias to distinguish agents from patients, evidence from emerging sign languages and infancy suggests this bias is rooted in cognition. Goldin-Meadow and Mylander found that homesigning children from the USA and Taiwan encoded a distinction between agents and patients in their homesign. When these children were describing events with two participants, such as a mouse eating cheese, they were more likely to produce a sign for the patient than for the agent. More importantly, agents and patients were produced using consistent word orders, for example, producing patients before acts e.

These findings show a tendency to distinguish agents from patients that emerges even among children who are not learning a linguistic system from their caregivers. Rissman and Goldin-Meadow found that a single child homesigner from the USA developed a morphological form for expressing causation, again indicating sensitivity to the category of Agent and propensity to encode this category in language all of the causers that the child described were animate, thus agents.

Grammatical devices for distinguishing agents from patients also emerge early in sign language communities. New sign languages come into being when deaf people join a community where they are not taught an existing sign language but are free to sign with each other, as when a community opens a school for special education, or high rates of congenital deafness lead to many deaf people within extended families. Subsequent generations learn the language that the older deaf members of the community had been using. Nicaraguan Sign Language NSL emerged about 50 years ago: even the earliest cohort of NSL uses word order to distinguish animate agents from inanimate patients although word orders were more variable when both the agent and patient are animate; Flaherty, The first generation of CTSL did not appear to use a consistent linguistic device to distinguish agents from patients, but the second and third generations did, using word order, spatial reference, character assignment, and causal chaining.

Together, these results demonstrate that when signers in different cultures around the world create new languages, distinguishing agents from patients has high priority. Experiments with infants provide a final piece of evidence for a cognitive bias to distinguish agents from patients. For example, Woodward showed that by 6 months of age, infants analyze reaching events in terms of the goals of the agent e. Crucially, many of these experiments feature events that would be unfamiliar to infants, such as blocks and balls moving in a seemingly autonomous manner.

This suggests that infants have an abstract agency schema, allowing them to interpret novel, unfamiliar events as goal-directed. Even if the roles Agent and Patient cannot be defined in terms of necessary and sufficient conditions, the evidence reviewed above suggests a universal bias to encode Agent and Patient categories distinctly from each other. In the first section, we sketched a nativist view of thematic roles: they are cross-culturally universal, present early in life, and change little over the course of development. The strength of the evidence presented here suggests that this strong view may be true for Agent and Patient roles.

It would be premature, however, to endorse this position, as too little is yet known about the structure of these roles in individual languages. Recall, for example, the research indicating the Patient role in English is more diffuse and heterogeneous than the Agent role. This may be indicative of a broader cross-linguistic trend, where a universal bias to distinguish agents from patients is in fact a universal bias to distinguish agents from all other types of participants.

Alternatively, other languages may differ from English in having a more tightly clustered Patient role or a more heterogeneous Agent role. More in-depth cross-linguistic research is needed to address these possibilities, and their implications for the nativist view of thematic roles, as we discuss in the fifth section. The strength of the findings on agents and patients have led some to the more general conclusion that thematic roles as a class are part of core knowledge see Strickland, The range of events that we are able to represent and describe, however, is much more varied than the prototypical case of an agent acting on a patient, as demonstrated by example 1.

We next consider events involving goals, recipients, and sources, and ask whether our representations of these events are shaped by universal cognitive biases. Across proposed lists of thematic roles, frequently listed candidates are Recipients e. Studies of adult sentence processing have shown that adults activate information about recipients and goals upon hearing verbs that encode these participants e.

For example, Chang et al. Infants, even as young as 10 months, represent an abstract Goal category when viewing motion events. For example, Lakusta, Spinelli, and Garcia a familiarized and At test, infants looked longer at an event of a duck moving out of a bowl a source path than moving into a bowl an IN-path event , indicating the infants had generalized IN-path events as being part of the same category as AT-path and ON-path events. Synthesizing the adult and infant studies, we find robust evidence that for motion and transfer events, people represent event participants in terms of abstract Goal and Recipient categories.

A thornier question concerns the relationship between the Goal and Recipient categories. These two thematic roles occupy a similar semantic space: in Reba threw the ball to Ronnie , for example, Ronnie could be construed as either a Goal or a Recipient. Nonetheless, Recipients are typically characterized in terms of transfer of an object, or of information to an animate participant, whereas Goals are typically characterized in terms of the endpoint of a spatial path.

Individual Page (Not Available)

Are Goals and Recipients distinct categories that happen to overlap, or are they subtypes within a single overarching category? We might therefore expect that Goal constitutes the central member of this overarching category, with Recipient a less prototypical instance of this category — that transfer is a metaphorical extension of motion along a path. Ziegler and Snedeker tested these two hypotheses about the relationship between goals and recipients. In a language production study with English-speaking adults, they found the order in which a goal was mentioned in a sentence did not prime the order in which a recipient was mentioned, or vice versa e.

Minimally, these results indicate that in English, priming draws on distinct Goal and Recipient categories; but the results are also consistent with the stronger hypothesis that for adult English speakers, Goal and Recipient are representationally distinct. Along these lines, de Cuypere analyzed the English preposition to and found evidence against the proposal that to has a core spatial meaning: the diachronic record of to shows that the earliest uses of to were not restricted to spatial meanings. Existing behavioral and corpus studies of English do not provide support for the proposal that adults represent goals and recipients in terms of a single abstract category, with Recipient being a metaphorical extension of Goal.

It is possible, however, that there is a general bias to represent recipients in terms of goals, even if this asymmetry is not manifest in English. The typological evidence does not support this possibility. For example, Rice and Kabata investigated the range of meanings expressed by allative markers across 44 languages where the core use of an allative was defined as the goal of a motion event, as in Jane walked to the store. Allative markers encode many senses beyond that of a spatial goal: in Japanese, for example, the allative marker ni is extended to 20 different senses see Kabata Some extensions are more common than others cross-linguistically.

Beyond marking goals, the allative extends most commonly to recipients e. Similarly, dative markers regularly extend beyond recipients to goals, beneficiaries e. Researchers have used this sort of data i. Across the board, goals and recipients are positioned close to each other on these semantic maps, although the central node of the map varies depending on whether the focus is goal or recipient extension. Other studies also do not support a strong distinction between these categories. For example, Bickel, Zakharko, Bierkandt, and Witzlack-Makarevich analyzed a genetically diverse sample of languages using cluster analysis, asking whether thematic role categories could explain common patterns of non-default case marking.

For nine out of ten children, two word-order patterns were found: recipients were more likely to be ordered after agents, and they were also more likely to be ordered after patients. In addition, they show the cognitive robustness of these categories, as even child homesigners encode them through patterns of word order.

In a later study, Zheng and Goldin-Meadow distinguished recipients from goals in their investigation of homesigners, and found both American and Chinese homesigning children were more likely to produce gestures for goals than recipients. This suggests goals are cognitively more central than recipients, as predicted by the idea that spatial representations are the foundation for action representation Gruber, ; Heine et al. Several studies of emerging sign languages have documented how signers describe events of transfer for CTSL: Ergin et al. Nonetheless, we are not aware of studies that have tested whether grammatical devices for encoding goals emerge before grammatical devices for encoding recipients or vice versa.

If there is a general bias for Goal to be a more robust category than Recipient, we would predict that sensitivity to goals emerges in infancy before sensitivity to recipients. Evidence from infants does not support this: sensitivity to goals and recipients develops around the same time roughly the first birthday.

Infants subsequently dishabituated to a switch in the roles of the puppets who was giver and who was recipient , but not to a switch in the spatial locations of the puppets. This asymmetry held for Considering the close relationship we observe between goals and recipients across a range of paradigms, a first possibility is that humans are biased to represent goals and recipients in terms of a single thematic role centered around an animate goal, as in Reba threw the ball to Ronnie. If so, this super-category would have a vast extension, so as to capture the range of meanings commonly encoded by allative and dative markers cross-linguistically: from purposes and locations to experiencers, possessors, and patients.

One problem with such an account is that it may be too inclusive to have any explanatory power. We propose a second possibility here: our representations of goals and recipients rely on two fundamental, but distinct cognitive systems — spatial cognition on the one hand and social relationships between animates on the other.

The ability to represent spatial relationships is a core cognitive ability, and spatially based metaphors and abstractions are widespread in language Lakoff, Nonetheless, social cognition is also fundamental, and the dative may best be understood as involving a relationship between animate individuals where there is some intermediary between the individuals, as in Marnie taught John Spanish.

Events of entities crossing space and events of two animates interacting often overlap, as in Reba threw the ball to Ronnie. An animate participant may be the endpoint of a spatial path, and an animate participant may become affected by virtue of receiving an object that has crossed space.

Under this second possibility, we observe categories of Goal and Recipient because there are universal biases to represent events in terms of movement along a spatial path, as well as social interaction between animates. The evidence on goals and recipients is compatible with the nativist view of thematic roles — they may be distinct, universal categories that are present early in life and change little over development.

An alternate possibility is more likely, however, given the diverse array of languages in which goals and recipients are colexified. The high semantic overlap of these roles may lead to more cross-linguistic diversity. As we reviewed above, there is evidence that Goal and Recipient categories are present in infancy. Presuming that infants across cultures have similar conceptual knowledge, it may be that categorization of goals and recipients shifts over the course of development, with adults categorizing these roles in more diverse ways than infants.

The evidence reviewed above suggests the category of Goal is not more cognitively robust than Recipient. Such an asymmetry is present, however, between Goals and Sources, as in Tyrell got a book from the library. Sources are represented in terms of an abstract category. In addition, Lakusta and Landau found adults described source paths more often when they were primed with a source-encoding verb such as unhook. This asymmetry is echoed by the typological literature. I walked from the store and the transfer domain e.

Bickel et al. Most importantly, Kabata surveyed 24 languages and found sources had a narrower scope of semantic extension than goals, consistent with the behavioral evidence showing Source is a less robust category. Evidence from emerging sign languages and infancy supports these findings. Zheng and Goldin-Meadow studied descriptions of motion events from homesigning children in the USA and Taiwan, and found children from both groups produced gestures for goals and sources; but gestures for goals were about five times as common as gestures for sources.

In addition, infants encode goals of motion events more robustly than sources. Tatone et al. Taken together, the typological evidence and studies with adults, homesigners, and infants provide strong evidence for a universal cognitive bias such that goals are represented more robustly than sources. The consensus that Sources are less robust than Goals, however, is potentially inconsistent with Clark and Carpenter's study of children learning English. Clark and Carpenter interpret this result as indicating the primacy of spatial cognition, and conclude there is a broad category of Source that includes Agent.

This interpretation is difficult to reconcile with the infant research that shows infants encode agents roughly six months before they encode sources. An alternative possibility is that English-learning children overgeneralize from because they are uncertain about the category of meanings from picks out, rather than because they represent a primary Source category.

The psycholinguistic literature provides robust evidence that English-speaking adults represent events in terms of Goal and Recipient categories that abstract beyond the verb-specific and event-specific level. These categories appear to be an important part of our conceptual and semantic repertoire: infants less than a year of age represent these or similar notions, and deaf children lacking exposure to a language model produce signs for these entities.

In addition, humans robustly distinguish sources from goals, as shown through cross-linguistic research, adult psycholinguistic experiments, and studies with infants. Nonetheless, it is unclear how our concepts about recipients and goals are structured: are these subtypes of the same category, or are they distinct categories drawing on two distinct cognitive systems? Although the nativist view of thematic roles is not decisively falsified by the available evidence on recipients and goals, this evidence suggests that there is greater variability in how these event participants are categorized than in how agents are categorized.

We now turn to the final type of thematic role reviewed here — Instrument, introduced in 1 — and explore the evidence for an Instrument category. Fillmore proposed that Instrument is one of a small set of universal roles. That is, the function of an instrument is thought to be essential to its meaning: being used by an agent to achieve something.

This tool prototype is exemplified by 1a , Janine ate the custard with a spoon , and the other examples in 1 are assumed to be less prototypical members of the category. There is surprisingly little empirical evidence for the Instrument category. Psycholinguistic studies have shown specific verbs can activate instrumental concepts Andreu et al.

For example, in Koenig et al. When the verb encoded the presence of an instrument e. This result indicates verb-specific encoding of instruments rather than an abstract Instrument category. The two words in English that most commonly introduce instruments are with and use. Problematically, this diagnostic assumes a priori an Instrument category centered around the tool prototype.

The existence of an Instrument category is under-supported empirically. Nevertheless, there could be a universal bias to construct a participant category around the concept of a tool. Children learning English also appear to be sensitive to the relationship between agents and instruments. But in trials with only two participants, such as the cake was cut with a knife , children were uncertain as to whether the non-object here, the knife was an Actor or an Instrument. Narrog and Ito found that having a single form for instruments and comitatives was one of the most frequent colexifications in their data set.

The behavior of English with lead Lakoff and Johnson to propose a cross-culturally universal metaphor in which an instrument is conceptualized as a companion. In addition to agents and comitatives, instrumental meanings colexify with themes, as in Gabe filled the glass with orange juice. In their sample of languages, Bickel et al. In other words, the object being thrown e. In sum, the typological literature exhibits a swathe of semantic relationships between instruments and other meanings.

Given this, it is not obvious how one could capture all cross-linguistically attested patterns within a single semantic notion. The typological data are consistent with the proposal that there are no universal biases shaping how we construct instrumental categories. On the other hand, the typological data could also be consistent with the proposal that humans are biased to represent instrumental events in terms of a tool prototype, a possibility we consider below.


  • Cases and thematic roles: Ergative, accusative and active.!
  • Global leaders in Islamic finance : industry milestones and reflections!
  • Firefox For Dummies.
  • Basic English Syntax with Exercises.
  • A Companion to Greek and Roman Historiography.
  • TeX for the Impatient.
  • Antenna Theory & Design?

The idea that tools are cognitively central is supported by work in the infancy literature. Stavans and Baillargeon found that 4-month-old infants were able to individuate objects when they had seen those objects used as tools, suggesting infants use information about tool use to assist individuation. Crucially, infants did not individuate objects when they were not used in a typical tool-like fashion e.

When to month-old infants observe novel objects being used functionally i. Moreover, The evidence that infants can represent the functions of tools, and that they use this to help them individuate objects, suggests tools are a cognitively robust category. Relatively few studies have addressed how instrumental meanings are encoded in emerging sign languages, although we do know that homesigning children label tools: Rissman, Horton, and Goldin-Meadow studied nine children from Guatemala, Nicaragua, the USA, and Taiwan age range: 2;11 — 12;0 , and found all children produced signs referring to tools.

This evidence does not, however, reveal the categories that these homesigners used to represent tools. To summarize, the nativist view of thematic roles may be correct for Instruments, such that the tool is the prototype within the Instrument category, and it could be a category with universal relevance. Although the current evidence does not contradict the nativist hypothesis, there is little linguistic or behavioral evidence that directly supports it either. Instead, we see evidence for linkages between instruments and a range of other roles, from agents and comitatives to locatives and themes.

In many ways, the current literature provides a detailed view of event participant categories and how they influence linguistic and cognitive behavior across a range of human populations. But there are a variety of gaps in the literature that prevent us from knowing whether the nativist view of thematic roles is correct for even a single role. In this section, we discuss how these gaps could be filled. One of the most notable omissions concerns the structure of the categories. In the section below, we discuss the prominent proposal that thematic roles are structured in terms of prototypes, and some of the limitations in extending prototype theory to thematic roles.

We also provide a roadmap for the types of studies that should be done to fully test the nativist view of thematic roles. For example, both nominative and ergative case markers introduce agents, but these markers have different extensions: only nominative case marks the single argument of an intransitive verb such as run. Analyzing thematic roles in terms of prototypes is the most common response to the observation that roles are not easily defined in terms of necessary and sufficient conditions. And, given extensive cross-linguistic variability, the proposal that thematic roles are part of core knowledge may depend on thematic roles having prototype structure.

The evidence reviewed illustrates that roles are represented in terms of abstract categories and are present in infants and deaf signers creating new languages. But the literature reveals little about whether these categories have prototype structure, and whether the prototypes are the same across languages. We summarize the relevant work here, and suggest how it could be applied to the study of thematic roles in the future. The behavioral phenomena supporting prototype representations are well-documented see Geeraerts, ; Hampton, ; Murphy, ; Rosch, ; for review.

For example, when people judge whether item X is a member of category Y, response times are faster when X is a more prototypical member of the category e. Similarly, when people are asked to list members of a category, prototypical members are mentioned earlier, and more often across people. Also, hedging language is more acceptable for non-prototypical members of a category than prototypical members a penguin is technically a bird vs. Although the idea that thematic roles have prototypes is widespread, the classic methods used to establish prototypes have rarely been applied to the study of thematic roles, for the critical reason that these methods rely on lexical labels e.

As described in the introduction, role categories often do not map cleanly onto words. For example, neither English with nor use on their own point to the category Instrument, as discussed above. Only the intersection of these terms corresponds to intuitions about what a prototypical Instrument might be. Given this asymmetry between event participant categories and lexical meanings, the same methods that support prototype theories cannot be directly extended to the study of thematic roles.

As a consequence, claims about thematic role prototypes are often put forth without extensive behavioral or cross-linguistic evidence, as pointed out by Rice Geeraerts : summarizes four types of prototypicality effects: 1 differences of typicality and membership, 2 membership uncertainty, 3 clustering into family resemblances, and 4 absence of necessary-and-sufficient definitions. Not all categories demonstrate all four of these effects. The category fruit , for example, demonstrates both the first and second effects: first, apples and pomegranates are both fruit, but an apple is a more typical exemplar than a pomegranate.

Demonstrating membership uncertainty, many English speakers are unsure whether olives and coconuts are fruits. The English category bird , however, demonstrates the first characteristic but not the second: a robin is a better example of a bird than a penguin, but for English speakers with a particular educational background, the boundaries of the category bird are discrete rather than fuzzy: a bat is not a bird, a flying squirrel is not a bird.

Which of these four characteristics do thematic role categories demonstrate? As described in the first section, they cannot be convincingly defined in terms of necessary and sufficient conditions. In the linguistics literature, thematic roles are often thought to demonstrate the first effect, graded typicality among members. Most notably, Dowty invokes prototypes to explain the puzzle that in English, agents generally appear as Subject and patients as Object Tasha kissed the baby has to mean that Tasha was the one doing the kissing.

Nonetheless, many types of participants appear as Subject that are not particularly agentive, such as in Tasha believed the news. An argument does not have to be a good example of an agent to appear in Subject position; it just has to have some Proto-Agent properties. This type of linguistic graded typicality effect has not, however, been demonstrated for a wider range of behavioral data. Nonetheless, existing behavioral methods can be extended to test for graded typicality of thematic roles, as event processing studies reveal asymmetries in how people identify and remember different types of participants.

As described in the third section, Lakusta and Landau found in a change-detection task that goals were remembered better than sources. Although both the agent and the recipient were animate, participants were less accurate at naming the recipient than the agent in the to ms window. If thematic roles are structured such that some members of the category are more typical than others, then a reasonable linking hypothesis is that in such tasks, processing and memory will be more robust for the typical members than for the atypical ones, ceteris paribus. Consistent with this reasoning, note that the asymmetry between source and goal demonstrated by Lakusta et al.

When Lakusta and Carey showed month-olds events of a balloon crossing from a source to a goal, infants did not increase their looking to either the different-goal events or the different-source events. This suggests the most robustly encoded notion of Goal is narrower than the endpoint of a crossing of space — it is the destination reached by a self-directed entity. This in turn might suggest that the latter is the more prototypical instance of the Goal category than the former.

In the Recipient domain, one possibility is that physical transfer is central: cross-linguistically, verbs of giving are by far more common than other types of three-argument verbs Newman, Along similar lines, Tatone et al. In the third section, we proposed that the Goal and Recipient categories are underpinned by two fundamental cognitive systems spatial vs. We are not aware of any studies that have directly tested whether processing of goals is more robust than processing of recipients, so this remains a matter for future investigation. The second prototypicality characteristic discussed by Geeraerts is that categories may have fuzzy boundaries, leading to uncertainty about whether a particular exemplar is a member of the category.

In practice for researchers trying to delimit where one thematic role ends and another begins, this uncertainty is all too familiar. Consider, for example, the contrast between Martha ate the custard with a spoon and Martha sprayed the ferns with water. Some researchers have analyzed the two underlined participants as members of different categories, Instrument and Locatum, respectively, under the justification that the latter sentence is fundamentally an event of a substance crossing space, rather than an event of tool manipulation Jackendoff, Other researchers have classified both the spoon and the water as Instruments, as they are both used by an agent to achieve a goal Koenig et al.

This issue of membership uncertainty has led to the widespread confusion observed by Dowty and Newmeyer at the outset of this paper. Drawing on the current example, the water may be an atypical member of the category Instrument, a typical member of the distinct category Locatum, or it may not fit neatly into any category, as with the noun olive , discussed above.

Priming experiments such as in Ziegler and Snedeker can address whether instruments and locatum participants are part of the same category. The third prototypicality characteristic is family resemblance. A well-known example concerns the different senses of the English preposition over e. Brugman, ; Lakoff, ; Taylor, These types of meaning chains are explicit in the Proto-Role proposals of Dowty and Grimm , among others.

The typological literature reviewed in the second to fourth sections also reveals which semantic relationships are particularly relevant for category membership across languages. Goals, for example, are closely related to locations and purposes, and instruments are closely related to agents and themes. So we have a relatively well-developed understanding of how similarity is defined within the semantic space of event participant roles.

In fact, the clearest alternative to the proposal that thematic roles have prototype structure is that participant roles are semantically related through chaining, without the presence of a central reference point. Prices are subject to change without notice. Prices do not include postage and handling if applicable. Free shipping for non-business customers when ordering books at De Gruyter Online. Please find details to our shipping fees here. Print Flyer Recommend to Librarian. Overview Aims and Scope This study examines the mapping of thematic roles, such as agent and patient, onto syntactic cases, such as nominative or ergative, or onto structural relations in a cross-linguistic survey that is supplemented with German data.

It would be premature, however, to endorse this position, as too little is yet known about the structure of these roles in individual languages. Recall, for example, the research indicating the Patient role in English is more diffuse and heterogeneous than the Agent role. This may be indicative of a broader cross-linguistic trend, where a universal bias to distinguish agents from patients is in fact a universal bias to distinguish agents from all other types of participants.

Alternatively, other languages may differ from English in having a more tightly clustered Patient role or a more heterogeneous Agent role. More in-depth cross-linguistic research is needed to address these possibilities, and their implications for the nativist view of thematic roles, as we discuss in the fifth section. The strength of the findings on agents and patients have led some to the more general conclusion that thematic roles as a class are part of core knowledge see Strickland, The range of events that we are able to represent and describe, however, is much more varied than the prototypical case of an agent acting on a patient, as demonstrated by example 1.

We next consider events involving goals, recipients, and sources, and ask whether our representations of these events are shaped by universal cognitive biases. Across proposed lists of thematic roles, frequently listed candidates are Recipients e. Studies of adult sentence processing have shown that adults activate information about recipients and goals upon hearing verbs that encode these participants e.

For example, Chang et al. Infants, even as young as 10 months, represent an abstract Goal category when viewing motion events. For example, Lakusta, Spinelli, and Garcia a familiarized and At test, infants looked longer at an event of a duck moving out of a bowl a source path than moving into a bowl an IN-path event , indicating the infants had generalized IN-path events as being part of the same category as AT-path and ON-path events.

Synthesizing the adult and infant studies, we find robust evidence that for motion and transfer events, people represent event participants in terms of abstract Goal and Recipient categories. A thornier question concerns the relationship between the Goal and Recipient categories.

These two thematic roles occupy a similar semantic space: in Reba threw the ball to Ronnie , for example, Ronnie could be construed as either a Goal or a Recipient.

[Introduction to Linguistics] Theta Roles / Thematic Roles

Nonetheless, Recipients are typically characterized in terms of transfer of an object, or of information to an animate participant, whereas Goals are typically characterized in terms of the endpoint of a spatial path. Are Goals and Recipients distinct categories that happen to overlap, or are they subtypes within a single overarching category? We might therefore expect that Goal constitutes the central member of this overarching category, with Recipient a less prototypical instance of this category — that transfer is a metaphorical extension of motion along a path.

Essays sakuntala LISSIM 6

Ziegler and Snedeker tested these two hypotheses about the relationship between goals and recipients. In a language production study with English-speaking adults, they found the order in which a goal was mentioned in a sentence did not prime the order in which a recipient was mentioned, or vice versa e. Minimally, these results indicate that in English, priming draws on distinct Goal and Recipient categories; but the results are also consistent with the stronger hypothesis that for adult English speakers, Goal and Recipient are representationally distinct.

Along these lines, de Cuypere analyzed the English preposition to and found evidence against the proposal that to has a core spatial meaning: the diachronic record of to shows that the earliest uses of to were not restricted to spatial meanings. Existing behavioral and corpus studies of English do not provide support for the proposal that adults represent goals and recipients in terms of a single abstract category, with Recipient being a metaphorical extension of Goal.

It is possible, however, that there is a general bias to represent recipients in terms of goals, even if this asymmetry is not manifest in English. The typological evidence does not support this possibility. For example, Rice and Kabata investigated the range of meanings expressed by allative markers across 44 languages where the core use of an allative was defined as the goal of a motion event, as in Jane walked to the store.

Allative markers encode many senses beyond that of a spatial goal: in Japanese, for example, the allative marker ni is extended to 20 different senses see Kabata Some extensions are more common than others cross-linguistically. Beyond marking goals, the allative extends most commonly to recipients e. Similarly, dative markers regularly extend beyond recipients to goals, beneficiaries e. Researchers have used this sort of data i. Across the board, goals and recipients are positioned close to each other on these semantic maps, although the central node of the map varies depending on whether the focus is goal or recipient extension.

Other studies also do not support a strong distinction between these categories. For example, Bickel, Zakharko, Bierkandt, and Witzlack-Makarevich analyzed a genetically diverse sample of languages using cluster analysis, asking whether thematic role categories could explain common patterns of non-default case marking.

For nine out of ten children, two word-order patterns were found: recipients were more likely to be ordered after agents, and they were also more likely to be ordered after patients. In addition, they show the cognitive robustness of these categories, as even child homesigners encode them through patterns of word order. In a later study, Zheng and Goldin-Meadow distinguished recipients from goals in their investigation of homesigners, and found both American and Chinese homesigning children were more likely to produce gestures for goals than recipients.

This suggests goals are cognitively more central than recipients, as predicted by the idea that spatial representations are the foundation for action representation Gruber, ; Heine et al. Several studies of emerging sign languages have documented how signers describe events of transfer for CTSL: Ergin et al. Nonetheless, we are not aware of studies that have tested whether grammatical devices for encoding goals emerge before grammatical devices for encoding recipients or vice versa.

If there is a general bias for Goal to be a more robust category than Recipient, we would predict that sensitivity to goals emerges in infancy before sensitivity to recipients. Evidence from infants does not support this: sensitivity to goals and recipients develops around the same time roughly the first birthday. Infants subsequently dishabituated to a switch in the roles of the puppets who was giver and who was recipient , but not to a switch in the spatial locations of the puppets. This asymmetry held for Considering the close relationship we observe between goals and recipients across a range of paradigms, a first possibility is that humans are biased to represent goals and recipients in terms of a single thematic role centered around an animate goal, as in Reba threw the ball to Ronnie.

If so, this super-category would have a vast extension, so as to capture the range of meanings commonly encoded by allative and dative markers cross-linguistically: from purposes and locations to experiencers, possessors, and patients. One problem with such an account is that it may be too inclusive to have any explanatory power.

We propose a second possibility here: our representations of goals and recipients rely on two fundamental, but distinct cognitive systems — spatial cognition on the one hand and social relationships between animates on the other. The ability to represent spatial relationships is a core cognitive ability, and spatially based metaphors and abstractions are widespread in language Lakoff, Nonetheless, social cognition is also fundamental, and the dative may best be understood as involving a relationship between animate individuals where there is some intermediary between the individuals, as in Marnie taught John Spanish.

Events of entities crossing space and events of two animates interacting often overlap, as in Reba threw the ball to Ronnie. An animate participant may be the endpoint of a spatial path, and an animate participant may become affected by virtue of receiving an object that has crossed space. Under this second possibility, we observe categories of Goal and Recipient because there are universal biases to represent events in terms of movement along a spatial path, as well as social interaction between animates. The evidence on goals and recipients is compatible with the nativist view of thematic roles — they may be distinct, universal categories that are present early in life and change little over development.

An alternate possibility is more likely, however, given the diverse array of languages in which goals and recipients are colexified. The high semantic overlap of these roles may lead to more cross-linguistic diversity. As we reviewed above, there is evidence that Goal and Recipient categories are present in infancy.

Presuming that infants across cultures have similar conceptual knowledge, it may be that categorization of goals and recipients shifts over the course of development, with adults categorizing these roles in more diverse ways than infants. The evidence reviewed above suggests the category of Goal is not more cognitively robust than Recipient. Such an asymmetry is present, however, between Goals and Sources, as in Tyrell got a book from the library.

Sources are represented in terms of an abstract category. In addition, Lakusta and Landau found adults described source paths more often when they were primed with a source-encoding verb such as unhook. This asymmetry is echoed by the typological literature.


  • Synonyms and antonyms of ergative in the English dictionary of synonyms.
  • Oh no, there's been an error;
  • The Complete Reloading Manual for the .243 Winchester.
  • Subscribe to RSS!
  • The Barrakee Mystery (Inspector Napoleon Bonaparte 1).
  • Semantics of case;

I walked from the store and the transfer domain e. Bickel et al.

Language Index

Most importantly, Kabata surveyed 24 languages and found sources had a narrower scope of semantic extension than goals, consistent with the behavioral evidence showing Source is a less robust category. Evidence from emerging sign languages and infancy supports these findings. Zheng and Goldin-Meadow studied descriptions of motion events from homesigning children in the USA and Taiwan, and found children from both groups produced gestures for goals and sources; but gestures for goals were about five times as common as gestures for sources.

In addition, infants encode goals of motion events more robustly than sources. Tatone et al. Taken together, the typological evidence and studies with adults, homesigners, and infants provide strong evidence for a universal cognitive bias such that goals are represented more robustly than sources. The consensus that Sources are less robust than Goals, however, is potentially inconsistent with Clark and Carpenter's study of children learning English.

Clark and Carpenter interpret this result as indicating the primacy of spatial cognition, and conclude there is a broad category of Source that includes Agent. This interpretation is difficult to reconcile with the infant research that shows infants encode agents roughly six months before they encode sources.

Meaning of "ergative" in the English dictionary

An alternative possibility is that English-learning children overgeneralize from because they are uncertain about the category of meanings from picks out, rather than because they represent a primary Source category. The psycholinguistic literature provides robust evidence that English-speaking adults represent events in terms of Goal and Recipient categories that abstract beyond the verb-specific and event-specific level.

These categories appear to be an important part of our conceptual and semantic repertoire: infants less than a year of age represent these or similar notions, and deaf children lacking exposure to a language model produce signs for these entities. In addition, humans robustly distinguish sources from goals, as shown through cross-linguistic research, adult psycholinguistic experiments, and studies with infants.

Nonetheless, it is unclear how our concepts about recipients and goals are structured: are these subtypes of the same category, or are they distinct categories drawing on two distinct cognitive systems? Although the nativist view of thematic roles is not decisively falsified by the available evidence on recipients and goals, this evidence suggests that there is greater variability in how these event participants are categorized than in how agents are categorized.

We now turn to the final type of thematic role reviewed here — Instrument, introduced in 1 — and explore the evidence for an Instrument category. Fillmore proposed that Instrument is one of a small set of universal roles. That is, the function of an instrument is thought to be essential to its meaning: being used by an agent to achieve something.

This tool prototype is exemplified by 1a , Janine ate the custard with a spoon , and the other examples in 1 are assumed to be less prototypical members of the category. There is surprisingly little empirical evidence for the Instrument category. Psycholinguistic studies have shown specific verbs can activate instrumental concepts Andreu et al. For example, in Koenig et al. When the verb encoded the presence of an instrument e. This result indicates verb-specific encoding of instruments rather than an abstract Instrument category.

The two words in English that most commonly introduce instruments are with and use. Problematically, this diagnostic assumes a priori an Instrument category centered around the tool prototype. The existence of an Instrument category is under-supported empirically. Nevertheless, there could be a universal bias to construct a participant category around the concept of a tool. Children learning English also appear to be sensitive to the relationship between agents and instruments. But in trials with only two participants, such as the cake was cut with a knife , children were uncertain as to whether the non-object here, the knife was an Actor or an Instrument.

Narrog and Ito found that having a single form for instruments and comitatives was one of the most frequent colexifications in their data set. The behavior of English with lead Lakoff and Johnson to propose a cross-culturally universal metaphor in which an instrument is conceptualized as a companion. In addition to agents and comitatives, instrumental meanings colexify with themes, as in Gabe filled the glass with orange juice. In their sample of languages, Bickel et al. In other words, the object being thrown e. In sum, the typological literature exhibits a swathe of semantic relationships between instruments and other meanings.

Given this, it is not obvious how one could capture all cross-linguistically attested patterns within a single semantic notion. The typological data are consistent with the proposal that there are no universal biases shaping how we construct instrumental categories. On the other hand, the typological data could also be consistent with the proposal that humans are biased to represent instrumental events in terms of a tool prototype, a possibility we consider below.

The idea that tools are cognitively central is supported by work in the infancy literature. Stavans and Baillargeon found that 4-month-old infants were able to individuate objects when they had seen those objects used as tools, suggesting infants use information about tool use to assist individuation. Crucially, infants did not individuate objects when they were not used in a typical tool-like fashion e. When to month-old infants observe novel objects being used functionally i. Moreover, The evidence that infants can represent the functions of tools, and that they use this to help them individuate objects, suggests tools are a cognitively robust category.

Relatively few studies have addressed how instrumental meanings are encoded in emerging sign languages, although we do know that homesigning children label tools: Rissman, Horton, and Goldin-Meadow studied nine children from Guatemala, Nicaragua, the USA, and Taiwan age range: 2;11 — 12;0 , and found all children produced signs referring to tools. This evidence does not, however, reveal the categories that these homesigners used to represent tools. To summarize, the nativist view of thematic roles may be correct for Instruments, such that the tool is the prototype within the Instrument category, and it could be a category with universal relevance.

Although the current evidence does not contradict the nativist hypothesis, there is little linguistic or behavioral evidence that directly supports it either. Instead, we see evidence for linkages between instruments and a range of other roles, from agents and comitatives to locatives and themes. In many ways, the current literature provides a detailed view of event participant categories and how they influence linguistic and cognitive behavior across a range of human populations.

But there are a variety of gaps in the literature that prevent us from knowing whether the nativist view of thematic roles is correct for even a single role. In this section, we discuss how these gaps could be filled. One of the most notable omissions concerns the structure of the categories. In the section below, we discuss the prominent proposal that thematic roles are structured in terms of prototypes, and some of the limitations in extending prototype theory to thematic roles.

We also provide a roadmap for the types of studies that should be done to fully test the nativist view of thematic roles. For example, both nominative and ergative case markers introduce agents, but these markers have different extensions: only nominative case marks the single argument of an intransitive verb such as run. Analyzing thematic roles in terms of prototypes is the most common response to the observation that roles are not easily defined in terms of necessary and sufficient conditions.

And, given extensive cross-linguistic variability, the proposal that thematic roles are part of core knowledge may depend on thematic roles having prototype structure. The evidence reviewed illustrates that roles are represented in terms of abstract categories and are present in infants and deaf signers creating new languages. But the literature reveals little about whether these categories have prototype structure, and whether the prototypes are the same across languages. We summarize the relevant work here, and suggest how it could be applied to the study of thematic roles in the future.

The behavioral phenomena supporting prototype representations are well-documented see Geeraerts, ; Hampton, ; Murphy, ; Rosch, ; for review. For example, when people judge whether item X is a member of category Y, response times are faster when X is a more prototypical member of the category e. Similarly, when people are asked to list members of a category, prototypical members are mentioned earlier, and more often across people.

Also, hedging language is more acceptable for non-prototypical members of a category than prototypical members a penguin is technically a bird vs. Although the idea that thematic roles have prototypes is widespread, the classic methods used to establish prototypes have rarely been applied to the study of thematic roles, for the critical reason that these methods rely on lexical labels e. As described in the introduction, role categories often do not map cleanly onto words. For example, neither English with nor use on their own point to the category Instrument, as discussed above.

Only the intersection of these terms corresponds to intuitions about what a prototypical Instrument might be. Given this asymmetry between event participant categories and lexical meanings, the same methods that support prototype theories cannot be directly extended to the study of thematic roles. As a consequence, claims about thematic role prototypes are often put forth without extensive behavioral or cross-linguistic evidence, as pointed out by Rice Geeraerts : summarizes four types of prototypicality effects: 1 differences of typicality and membership, 2 membership uncertainty, 3 clustering into family resemblances, and 4 absence of necessary-and-sufficient definitions.

Not all categories demonstrate all four of these effects. The category fruit , for example, demonstrates both the first and second effects: first, apples and pomegranates are both fruit, but an apple is a more typical exemplar than a pomegranate. Demonstrating membership uncertainty, many English speakers are unsure whether olives and coconuts are fruits.

The English category bird , however, demonstrates the first characteristic but not the second: a robin is a better example of a bird than a penguin, but for English speakers with a particular educational background, the boundaries of the category bird are discrete rather than fuzzy: a bat is not a bird, a flying squirrel is not a bird. Which of these four characteristics do thematic role categories demonstrate? As described in the first section, they cannot be convincingly defined in terms of necessary and sufficient conditions.

In the linguistics literature, thematic roles are often thought to demonstrate the first effect, graded typicality among members. Most notably, Dowty invokes prototypes to explain the puzzle that in English, agents generally appear as Subject and patients as Object Tasha kissed the baby has to mean that Tasha was the one doing the kissing. Nonetheless, many types of participants appear as Subject that are not particularly agentive, such as in Tasha believed the news.

An argument does not have to be a good example of an agent to appear in Subject position; it just has to have some Proto-Agent properties. This type of linguistic graded typicality effect has not, however, been demonstrated for a wider range of behavioral data. Nonetheless, existing behavioral methods can be extended to test for graded typicality of thematic roles, as event processing studies reveal asymmetries in how people identify and remember different types of participants.

As described in the third section, Lakusta and Landau found in a change-detection task that goals were remembered better than sources. Although both the agent and the recipient were animate, participants were less accurate at naming the recipient than the agent in the to ms window. If thematic roles are structured such that some members of the category are more typical than others, then a reasonable linking hypothesis is that in such tasks, processing and memory will be more robust for the typical members than for the atypical ones, ceteris paribus.

Consistent with this reasoning, note that the asymmetry between source and goal demonstrated by Lakusta et al. When Lakusta and Carey showed month-olds events of a balloon crossing from a source to a goal, infants did not increase their looking to either the different-goal events or the different-source events. This suggests the most robustly encoded notion of Goal is narrower than the endpoint of a crossing of space — it is the destination reached by a self-directed entity. This in turn might suggest that the latter is the more prototypical instance of the Goal category than the former.

In the Recipient domain, one possibility is that physical transfer is central: cross-linguistically, verbs of giving are by far more common than other types of three-argument verbs Newman, Along similar lines, Tatone et al. In the third section, we proposed that the Goal and Recipient categories are underpinned by two fundamental cognitive systems spatial vs. We are not aware of any studies that have directly tested whether processing of goals is more robust than processing of recipients, so this remains a matter for future investigation. The second prototypicality characteristic discussed by Geeraerts is that categories may have fuzzy boundaries, leading to uncertainty about whether a particular exemplar is a member of the category.

In practice for researchers trying to delimit where one thematic role ends and another begins, this uncertainty is all too familiar. Consider, for example, the contrast between Martha ate the custard with a spoon and Martha sprayed the ferns with water. Some researchers have analyzed the two underlined participants as members of different categories, Instrument and Locatum, respectively, under the justification that the latter sentence is fundamentally an event of a substance crossing space, rather than an event of tool manipulation Jackendoff, Other researchers have classified both the spoon and the water as Instruments, as they are both used by an agent to achieve a goal Koenig et al.

This issue of membership uncertainty has led to the widespread confusion observed by Dowty and Newmeyer at the outset of this paper. Drawing on the current example, the water may be an atypical member of the category Instrument, a typical member of the distinct category Locatum, or it may not fit neatly into any category, as with the noun olive , discussed above.

Priming experiments such as in Ziegler and Snedeker can address whether instruments and locatum participants are part of the same category. The third prototypicality characteristic is family resemblance. A well-known example concerns the different senses of the English preposition over e. Brugman, ; Lakoff, ; Taylor, These types of meaning chains are explicit in the Proto-Role proposals of Dowty and Grimm , among others.

The typological literature reviewed in the second to fourth sections also reveals which semantic relationships are particularly relevant for category membership across languages. Goals, for example, are closely related to locations and purposes, and instruments are closely related to agents and themes.

So we have a relatively well-developed understanding of how similarity is defined within the semantic space of event participant roles. In fact, the clearest alternative to the proposal that thematic roles have prototype structure is that participant roles are semantically related through chaining, without the presence of a central reference point. In a study of how the senses of English words changed over time, Ramiro, Srinivasan, Malt, and Xu found that a model based on nearest neighbor semantic chaining was more efficient than a model based on a central prototype. Whether the same holds for thematic roles is not known.

An additional question concerns the types of semantic features that comprise the family resemblance structures in thematic roles. Dowty and Grimm, inter alia , propose abstract features such as intentionality and causation. In the event cognition experiments of Hafri et al. Verb-specific typicality information e. Each of these types of features may be part of role prototypes — it is unknown, however, whether different cognitive and linguistic behaviors rely on different sets of features.

The previous section described methods for testing whether event participant categories have prototype structure.

Your Answer

These methods would enable researchers to establish not only whether event participants are represented in terms of abstract categories, but also to propose analyses of how the categories are structured. In this section, we propose a variety of steps that should be taken to more decisively answer the question of whether event participant categories are influenced by universal cognitive biases and whether the nativist view of thematic roles is correct.

The majority of experimental studies on event cognition and memory have been conducted with speakers of English. Experimental work on thematic roles should also be expanded to a broader range of languages to address both the question of role abstraction as well as prototypicality. Although Ziegler and Snedeker did not find syntactic priming between goals and recipients in English, such priming may well be observed for other languages. As described above, experimental work could also test whether thematic roles demonstrate prototype effects, such as typicality of membership and membership uncertainty.

It is entirely possible that, even if all the thematic roles summarized in this paper are represented in terms of prototypes in a particular language, the prototypes are not identical across languages and cultures. Much of the typological and language emergence evidence reviewed in the second section documented that agents and patients are strongly distinguished across languages. However, there may be a universal bias to distinguish agents and patients, without these categories having the same prototypes and category structure in each culture.

In addition, the modeling in White et al. If such variability is attested, it opens a range of questions that have scarcely been asked: for example, how are event participant categories molded by culture, environment, and history, and whether the particular distinctions made in our language affect our conceptual representations of thematic roles. Universal biases can influence not only structure within a category but also relationships across categories. We reviewed a variety of evidence that some types of roles appear to be more cognitively prominent than others.

Asymmetries such as these are not only reflected in language. As described in the third section, infants encode sources less robustly than goals, and adults have poorer memory for sources than goals. Dobel et al. Experimental work across a diverse range of languages is needed to test whether role asymmetries can be explained by universal cognitive biases.

Another crucial question for future research concerns the relationship between thematic roles as linguistic objects and thematic roles as conceptual categories. This variability is also manifest within a single language. Nonetheless, use and with encode different instrumental features: use requires an intentional agent. This variability does not, however, falsify the hypothesis that the category Agent is universal.

If we allow that Ag e has prototype structure, where an unintentional agent is still an Agent, then both use and with are compatible with this category. For Agent to truly be universal, however, it must first be the case that in English, the categories demonstrated by non-linguistic event cognition experiments are the same as the categories needed to account for natural language semantics e.

It must then be the case that this parallelism holds in all languages. If we find any language where this parallelism does not hold, this suggests domain-specific thematic role representations that may be structured in different ways. A category such as Ag e , for example, may need to be relativized to a particular language or linguistic construction.

One of the fundamental questions of cognitive science concerns the ways in which people categorize the world around them, and the extent to which humans form similar or different categories across languages and cultures. From a domain-general perspective, the study of thematic roles is the study of event participant categories, and studying these categories provides an opportunity to gain insight into the interface between conceptual and semantic knowledge.

Many psychologists working on concept representation do not distinguish between the semantic category picked out by the particular English word dog , and categorization of furry, four-legged domestic animals at a conceptual level. As thematic roles are largely not labeled through open class lexical items, such conflation is not possible in the domain of event participant categories. At the outset of this review, we described two poles in theorizing about thematic roles — thematic roles are part of core knowledge or they are scholarly fictions. The evidence we have reviewed from adult psycholinguistics, development, typology, and emerging sign languages makes it untenable to maintain the view that thematic roles are fiction.

What the data do show is that for the best-studied roles, there is evidence for abstraction, scaffolded by universal cognitive biases. The exact nature of these abstractions — i. In sum, despite the substantive literature reviewed herein, many foundational questions about thematic roles remain unanswered: what the structure of event participant categories is in individual languages, and whether, for lesser-studied roles such as Instrument, there are universal biases. Nonetheless, through an integration of diverse methods and sources of evidence, answers to these foundational questions are well within reach.

In this paper, role labels such as "Instrument" are used intensionally to refer to a proposed mental category. Lower-case labels such as "instrument" are used extensionally to refer to a particular entity which may or may not be a member of a role category. Skip to main content Skip to sections. Advertisement Hide. Download PDF. Thematic roles: Core knowledge or linguistic construct? Open Access. First Online: 09 July Introduction Thematic roles such as Agent, Patient, and Goal have a long-standing presence in theories of linguistics and cognitive science.

A universal bias to distinguish agents from patients Behavioral effects from individual languages, such as priming from one type of event to another, constitute evidence for abstract event participant categories. Summary Even if the roles Agent and Patient cannot be defined in terms of necessary and sufficient conditions, the evidence reviewed above suggests a universal bias to encode Agent and Patient categories distinctly from each other.

Behavioral evidence for abstract recipient and goal categories Across proposed lists of thematic roles, frequently listed candidates are Recipients e. No universal bias to represent recipients in terms of goals Existing behavioral and corpus studies of English do not provide support for the proposal that adults represent goals and recipients in terms of a single abstract category, with Recipient being a metaphorical extension of Goal.

Summary The psycholinguistic literature provides robust evidence that English-speaking adults represent events in terms of Goal and Recipient categories that abstract beyond the verb-specific and event-specific level. Evidence for an instrument category? Universal biases The existence of an Instrument category is under-supported empirically. The future of research on thematic roles In many ways, the current literature provides a detailed view of event participant categories and how they influence linguistic and cognitive behavior across a range of human populations.

Compliance with ethical standards Open practices statement We did not collect new data or conduct new data analysis for this paper. Ackerman, F. Proto-properties and grammatical encoding. Stanford Monographs in Linguistics.

Cases and Thematic Roles

Stanford: CSLI. Google Scholar. Adam, M. Cognitive Development, 43 Supplement C , Altmann, G. Thematic Role Assignment in Context. Journal of Memory and Language, 41 1 , Andreu, L. Frontiers in Psychology, 6 Angiolillo, C. Experimental evidence for agent-patient categories in child language.

Journal of Child language, 9 03 , Arunachalam, S. Meaning from syntax: Evidence from 2-year-olds. Cognition, 3 , Baker, M. Incorporation: a theory of grammatical function changing. Chicago: University of Chicago Press. Bickel, B. Semantic role clustering: An empirical assessment of semantic role types in non-default case assignment. Studies in Language. Bicknell, K. Effects of event knowledge in processing verbal arguments. Journal of Memory and Language, 63 4 , Biro, S. Developmental Science, 10 3 , Blake, B.

Cambridge: Cambridge University Press. Blansitt, E. Datives and allatives Studies in syntactic typology Vol. Boland, J. Visual arguments. Cognition, 95 3 , Journal of Memory and Language, 34 6 , Bowerman, M. Crosslinguistic perspectives on argument structure: implications for learnability. New York: Lawrence Erlbaum.

Cases and Thematic Roles. Ergative, Accusative and Active Cases and Thematic Roles. Ergative, Accusative and Active
Cases and Thematic Roles. Ergative, Accusative and Active Cases and Thematic Roles. Ergative, Accusative and Active
Cases and Thematic Roles. Ergative, Accusative and Active Cases and Thematic Roles. Ergative, Accusative and Active
Cases and Thematic Roles. Ergative, Accusative and Active Cases and Thematic Roles. Ergative, Accusative and Active
Cases and Thematic Roles. Ergative, Accusative and Active Cases and Thematic Roles. Ergative, Accusative and Active
Cases and Thematic Roles. Ergative, Accusative and Active Cases and Thematic Roles. Ergative, Accusative and Active
Cases and Thematic Roles. Ergative, Accusative and Active Cases and Thematic Roles. Ergative, Accusative and Active
Cases and Thematic Roles. Ergative, Accusative and Active Cases and Thematic Roles. Ergative, Accusative and Active

Related Cases and Thematic Roles. Ergative, Accusative and Active



Copyright 2019 - All Right Reserved