Academia.eduAcademia.edu

Figure 6 – uploaded by Mary Dalrymple

See full PDFdownloadDownload figure
TABLE 1 Model parameters for the American and Australian ratings for communication verb double object datives

Table 1 Model parameters for the American and Australian ratings for communication verb double object datives

Related Figures (35)

TABLE 1 Animacy categories with examples 26 / BETH LEVIN AND JASON GRAFMILLER
TABLE 1 Animacy categories with examples 26 / BETH LEVIN AND JASON GRAFMILLER
FIGURE 1 Distribution of stimulus animacy types by verb (collapsed)
FIGURE 1 Distribution of stimulus animacy types by verb (collapsed)
TABLE 1 Distribution of postverbal bare plural subjects. The mismatch between the PVBS test and the GCC test for unac- cusatives can be accounted for in semantic terms. Postverbal bare plu- ral subjects can appear with telic predicates like regresar, desaparecer, arrojarse, or asfixiarse, but not with verbs like heder, trabajar, arras- trarse, or reflejarse. which are atelic. This is summarized in Table 1.
TABLE 1 Distribution of postverbal bare plural subjects. The mismatch between the PVBS test and the GCC test for unac- cusatives can be accounted for in semantic terms. Postverbal bare plu- ral subjects can appear with telic predicates like regresar, desaparecer, arrojarse, or asfixiarse, but not with verbs like heder, trabajar, arras- trarse, or reflejarse. which are atelic. This is summarized in Table 1.
TABLE 2 Distribution of predicates in the generic causee construction.
TABLE 2 Distribution of predicates in the generic causee construction.
Dowty, thematic roles are sets of entailments, determined by the lexi- cal semantics of the predicate. The two prototypical thematic roles, or proto-roles, are characterized by the following entailments: In a sentence with two arguments (i.e. a transitive verb), the subject will display a heavier proportion of proto-agent properties than the ob- ject. This principle allows for different combinations of semantic prop- erties, defining degrees of prototypicality for agents and patients. This approach can be extended to account for patterns of split intransitivity. Subjects of unergative verbs, Dowty says, normally have more proto- agent properties than subjects of unaccusatives. The subject of a verb like run, often classified as unergative, must be volitionally involved in the event, and has the property of moving. A verb like die, on the other hand, which is typically unaccusative, is associated with the entailment that its subject undergoes a change of state, and is not volitionally in- volved in the event.
Dowty, thematic roles are sets of entailments, determined by the lexi- cal semantics of the predicate. The two prototypical thematic roles, or proto-roles, are characterized by the following entailments: In a sentence with two arguments (i.e. a transitive verb), the subject will display a heavier proportion of proto-agent properties than the ob- ject. This principle allows for different combinations of semantic prop- erties, defining degrees of prototypicality for agents and patients. This approach can be extended to account for patterns of split intransitivity. Subjects of unergative verbs, Dowty says, normally have more proto- agent properties than subjects of unaccusatives. The subject of a verb like run, often classified as unergative, must be volitionally involved in the event, and has the property of moving. A verb like die, on the other hand, which is typically unaccusative, is associated with the entailment that its subject undergoes a change of state, and is not volitionally in- volved in the event.
6 As demonstrated in Her (1999, 2010), while languages like English allow this option, languages like Chinese do not. The LMT model proposed here can account for this distinction; due to space limitations, I will not go into this interesting typological issue. Tmt: ... et De a, a a ee ee ee, ee ae ee: ee ‘This analysis does not bring in the co-variation in word order associated with this construction. See Siewierska and Hollmann (2007) for a corpus-based study.
6 As demonstrated in Her (1999, 2010), while languages like English allow this option, languages like Chinese do not. The LMT model proposed here can account for this distinction; due to space limitations, I will not go into this interesting typological issue. Tmt: ... et De a, a a ee ee ee, ee ae ee: ee ‘This analysis does not bring in the co-variation in word order associated with this construction. See Siewierska and Hollmann (2007) for a corpus-based study.
FIGURE 1 Facets and Dimensions across Frameworks FIGURE 2 Cross-framework Analyses of Coordination: left: HPSG (upper), LFG (lower); right: CCG (upper), LTAG (lowe:
FIGURE 1 Facets and Dimensions across Frameworks FIGURE 2 Cross-framework Analyses of Coordination: left: HPSG (upper), LFG (lower); right: CCG (upper), LTAG (lowe:
TABLE 2 Model parameters for the American and Australian ratings for transfer verb double object datives 100 / MARILYN FoRD AND JOAN BRESNAN
TABLE 2 Model parameters for the American and Australian ratings for transfer verb double object datives 100 / MARILYN FoRD AND JOAN BRESNAN
TABLE 3 Model parameters for all main effects and 2-way interactions for Australian RTs at the determiner after the recipient in double object datives THEY WHISPERED ME THE ANSWER / 103
TABLE 3 Model parameters for all main effects and 2-way interactions for Australian RTs at the determiner after the recipient in double object datives THEY WHISPERED ME THE ANSWER / 103
TABLE 4 Model parameters for all main effects and 2-way interactions for American RTs at the noun after the recipient in double object datives
TABLE 4 Model parameters for all main effects and 2-way interactions for American RTs at the noun after the recipient in double object datives
TABLE 1 Acceptability Judgments for New Impersonal in (1a)
TABLE 1 Acceptability Judgments for New Impersonal in (1a)
TABLE 2 Age-related Variation in Acceptability Judgments for the NI Example in (1a) ing ground. Because this is an on-going change, it provides historical linguists with a rare opportunity to observe and document a syntac- tic change at a relatively early stage; the evidence available in written records can be complemented with native speaker judgments of ac- ceptability. On-going research supports the hypothesis that the NI is acquired by young children and not adopted by adults. Forty of the sub- jects tested by M&S in 1999-2000 were retested in 2010; children who acquire the NI do not seem to outgrow it (Thrainsson, Sigurjonsdottir and Eythorsson, 2010).
TABLE 2 Age-related Variation in Acceptability Judgments for the NI Example in (1a) ing ground. Because this is an on-going change, it provides historical linguists with a rare opportunity to observe and document a syntac- tic change at a relatively early stage; the evidence available in written records can be complemented with native speaker judgments of ac- ceptability. On-going research supports the hypothesis that the NI is acquired by young children and not adopted by adults. Forty of the sub- jects tested by M&S in 1999-2000 were retested in 2010; children who acquire the NI do not seem to outgrow it (Thrainsson, Sigurjonsdottir and Eythorsson, 2010).
TABLE 3 Syntactic properties of constructions in Polish and Ukrainian
TABLE 3 Syntactic properties of constructions in Polish and Ukrainian
TABLE 4 Acceptability judgments for NI of a verb that does not form the canonical passive
TABLE 4 Acceptability judgments for NI of a verb that does not form the canonical passive
TABLE 5 Impersonal Passive of Reflexive Verb
TABLE 5 Impersonal Passive of Reflexive Verb
TABLE 1 Instrumentals in the Corpus of Contemporary American English
TABLE 1 Instrumentals in the Corpus of Contemporary American English
[TABLE 2 Social verbs in the Corpus of Contemporary American Englis] for the lemma only (including all inflected forms), and then for ‘verb lemma plus with’ sequences. The results are summarized in Table 2.
[TABLE 2 Social verbs in the Corpus of Contemporary American Englis] for the lemma only (including all inflected forms), and then for ‘verb lemma plus with’ sequences. The results are summarized in Table 2.
However, given the configurations in (29), none of these definitions cor- rectly discriminates among the three cases: it is impossible to distin- guish the acceptable nesting cases from the unacceptable crossing ones with VV f-precedence, or the acceptable non-overlapping cases from the unacceptable crossing ones with V4 f-precedence. SV f-precedence dis- tinguishes none of the cases. 11.5 Redefining the problem
However, given the configurations in (29), none of these definitions cor- rectly discriminates among the three cases: it is impossible to distin- guish the acceptable nesting cases from the unacceptable crossing ones with VV f-precedence, or the acceptable non-overlapping cases from the unacceptable crossing ones with V4 f-precedence. SV f-precedence dis- tinguishes none of the cases. 11.5 Redefining the problem
Instead, we use the definition of fcommand in (38). However, this definition does not always make the right predictions when two attributes share the same value, which is the situation we ar« interested in (R. Kaplan, p.c.; Dalrymple 2001). In (37), the f structure f does not f-command the f-structure g, because there is an f-structure h that contains f but does not contain g.
Instead, we use the definition of fcommand in (38). However, this definition does not always make the right predictions when two attributes share the same value, which is the situation we ar« interested in (R. Kaplan, p.c.; Dalrymple 2001). In (37), the f structure f does not f-command the f-structure g, because there is an f-structure h that contains f but does not contain g.
The operator O2 is the value of the upF attribute (Unbounded De- pendency Function: Asudeh 2010, 2012; see also Alsina 2008), and is also the object of play. There is no constituent in the c-structure that corresponds to O2. This means that the suprRiorITy relation between operators cannot be defined in terms of position in the c-structure tree; the relevant relation must be /-command, a hierarchical relation de- fined at f-structure. The standard definition of fcommand, shown in (36), works for many cases.
The operator O2 is the value of the upF attribute (Unbounded De- pendency Function: Asudeh 2010, 2012; see also Alsina 2008), and is also the object of play. There is no constituent in the c-structure that corresponds to O2. This means that the suprRiorITy relation between operators cannot be defined in terms of position in the c-structure tree; the relevant relation must be /-command, a hierarchical relation de- fined at f-structure. The standard definition of fcommand, shown in (36), works for many cases.
The specification of path in terms of location() allows for VerbNet to specify that the THEME is at the specified location at the start of the event (i.e., the table), is set in motion during (i.e., during(E)) the event, and is located at DESTINATION (i.e., the floor) by the end of the event. However, it is not always the case that only change of location type events will have location() predicates marked as the start or end of an event. In the example sentence in (5), the predicate location () is used as a means of indicating that the decoration (i.e., THEME; the name) is made on the ring (i.e., DESTINATION) and does not indicate In general, for classes in which the realized piece of the path refers to either endpoint of the change of location, VerbNet makes use of the location() predicate in conjunction with the motion() predicate to represent path information. Consider the slide-11.2 class:
The specification of path in terms of location() allows for VerbNet to specify that the THEME is at the specified location at the start of the event (i.e., the table), is set in motion during (i.e., during(E)) the event, and is located at DESTINATION (i.e., the floor) by the end of the event. However, it is not always the case that only change of location type events will have location() predicates marked as the start or end of an event. In the example sentence in (5), the predicate location () is used as a means of indicating that the decoration (i.e., THEME; the name) is made on the ring (i.e., DESTINATION) and does not indicate In general, for classes in which the realized piece of the path refers to either endpoint of the change of location, VerbNet makes use of the location() predicate in conjunction with the motion() predicate to represent path information. Consider the slide-11.2 class:
The most obvious problem is that unlike the location() predicate that represents the location of an event, the Prep() predicate, on its own, does not represent a specific meaning. That is, given the Prep() predicate alone, it is difficult to determine what relationship THEME and LOCATION have with each other and how they relate to event E. Even if we set this problem aside, this representation of path is some- what problematic since the same predicate Prep() is used to represent a static location of an object such as in the corner in the sentence “The statue stood in the corner” (spatial_ configuration-47.6), which is unlike the trajectory Prep( as used in example 6. The way to distinguish such instances is seemingly to look for the motion() predicate along with the Prep(). However, such a heuristic is not fail-safe either. Consider example 7 as a comparison to example 6, in which the representation (e.g., the use of the Prep() in the presence of motion()) seems to look If the sentence expresses the trajectory of the motion, rather than the starting or ending location, VerbNet currently tends to favor the use of the Prep() accompanied by the motion() predicate as seen in exam- ple 6:
The most obvious problem is that unlike the location() predicate that represents the location of an event, the Prep() predicate, on its own, does not represent a specific meaning. That is, given the Prep() predicate alone, it is difficult to determine what relationship THEME and LOCATION have with each other and how they relate to event E. Even if we set this problem aside, this representation of path is some- what problematic since the same predicate Prep() is used to represent a static location of an object such as in the corner in the sentence “The statue stood in the corner” (spatial_ configuration-47.6), which is unlike the trajectory Prep( as used in example 6. The way to distinguish such instances is seemingly to look for the motion() predicate along with the Prep(). However, such a heuristic is not fail-safe either. Consider example 7 as a comparison to example 6, in which the representation (e.g., the use of the Prep() in the presence of motion()) seems to look If the sentence expresses the trajectory of the motion, rather than the starting or ending location, VerbNet currently tends to favor the use of the Prep() accompanied by the motion() predicate as seen in exam- ple 6:
Although LOCATION roles in examples 6 and 8 specify the trajectory along which the THEME moves, the creators of this class chose to treat (8) as a case of viaQ instead of Prep() as seen in (6). The motivation behind this treatment was likely a syntactic one — the LOCATION in (6) is found in a prepositional phrase, while the same role appears as the noun phrase in (8). The predicate via() appears in two other classes, vehicle-51.4.1 (e.g., She skated the canals.) and nonvehicle- 51.4.2 (e.g., She rowed the canals.); both of which are also classes where the syntactic frame in which the path of motion appears is taken into account in the creation of the semantic frames. Less frequent representations of the path of motion are the direction() and the via() predicates. Like location() and Prep(), they are used in conjunction with the motion() predicate. The following is an exam- ple of the use of the direction() predicate:
Although LOCATION roles in examples 6 and 8 specify the trajectory along which the THEME moves, the creators of this class chose to treat (8) as a case of viaQ instead of Prep() as seen in (6). The motivation behind this treatment was likely a syntactic one — the LOCATION in (6) is found in a prepositional phrase, while the same role appears as the noun phrase in (8). The predicate via() appears in two other classes, vehicle-51.4.1 (e.g., She skated the canals.) and nonvehicle- 51.4.2 (e.g., She rowed the canals.); both of which are also classes where the syntactic frame in which the path of motion appears is taken into account in the creation of the semantic frames. Less frequent representations of the path of motion are the direction() and the via() predicates. Like location() and Prep(), they are used in conjunction with the motion() predicate. The following is an exam- ple of the use of the direction() predicate:
The direction() predicate is generally found to represent the mean. ng of the trajectory in a given path of motion as seen in example 9 [his representation is comparable to the one for (6). One advantage thi ‘epresentation has over the use of Prep() is that while with Prep() th ictual preposition is instantiated from the context, the direction( specifies that the THEME is directed towards the LOCATION during she event. Outside of this factor, it is not clear that VerbNet needs tc listinguishes the path information in (9) from that in (6). VerbNet, as discussed above, currently does not have an established method of representing the path of motion. The representations for source, trajectory and destination are currently spread over several dif- ferent representations, and they are generally specific to the needs of the class in which they are found. Furthermore, another difficulty in representing path of motion comes from the fact that VerbNet classes Jo not contain the necessary semantic roles to properly represent the path of motion. For example, consider the run-51.3.2 class seen in ex- umple 6. Currently this class specifies AGENT, THEME and LOCATION. allowing for sentences such as The horse jumped over the river or The horse jumped onto the shore. However, when more than one piece of the path of motion is realized, these roles are no longer adequate.
The direction() predicate is generally found to represent the mean. ng of the trajectory in a given path of motion as seen in example 9 [his representation is comparable to the one for (6). One advantage thi ‘epresentation has over the use of Prep() is that while with Prep() th ictual preposition is instantiated from the context, the direction( specifies that the THEME is directed towards the LOCATION during she event. Outside of this factor, it is not clear that VerbNet needs tc listinguishes the path information in (9) from that in (6). VerbNet, as discussed above, currently does not have an established method of representing the path of motion. The representations for source, trajectory and destination are currently spread over several dif- ferent representations, and they are generally specific to the needs of the class in which they are found. Furthermore, another difficulty in representing path of motion comes from the fact that VerbNet classes Jo not contain the necessary semantic roles to properly represent the path of motion. For example, consider the run-51.3.2 class seen in ex- umple 6. Currently this class specifies AGENT, THEME and LOCATION. allowing for sentences such as The horse jumped over the river or The horse jumped onto the shore. However, when more than one piece of the path of motion is realized, these roles are no longer adequate.
The path() predicate, thus, would replace both the Prep() (see ex. 6) and the via() (see ex. 8). With the replacement of Prep(), we are achieving a more semantically explicit representation of the path. The replacement of via() allows us to represent the semantics of the tra- jectory independently of the constituent’s syntactic expression. The proposed predicate includes the event EF that links the path to the E of the motion() predicate, and by association, the THEME in motion. Additionally, the path() predicate includes all three roles that indicate the different parts of the path of motion, namely, the SOURCE, TRAJECTORY and DESTINATION roles. The path() predicate would al- low for the representation of all possible combinations in the expression of path, which thus far has not been possible in VerbNet. Here is the revised view of the run-51.3.2 class:
The path() predicate, thus, would replace both the Prep() (see ex. 6) and the via() (see ex. 8). With the replacement of Prep(), we are achieving a more semantically explicit representation of the path. The replacement of via() allows us to represent the semantics of the tra- jectory independently of the constituent’s syntactic expression. The proposed predicate includes the event EF that links the path to the E of the motion() predicate, and by association, the THEME in motion. Additionally, the path() predicate includes all three roles that indicate the different parts of the path of motion, namely, the SOURCE, TRAJECTORY and DESTINATION roles. The path() predicate would al- low for the representation of all possible combinations in the expression of path, which thus far has not been possible in VerbNet. Here is the revised view of the run-51.3.2 class:
Finally, the advantage of a consistent and explicit representation of path of motion is that we get the desired change of location inferences for free. In other words, in the sentence in example 4, the inference we want to retrieve is that the the book is on the table before the event begins and is no longer there at the end of the event. In the same way. the book is not on the floor before the event, but is there at the end of the event. Currently, VerbNet generally, albeit inconsistently, handles this by using the start() and end() predicates to show where the THEME is located at the specified point in time. Take the banish-10.2 class as an example:? Furthermore, the path usage of the location() predicate (as op- posed to the static location usage as shown in ex. 5) and the direction () predicate would be united under the same representation.
Finally, the advantage of a consistent and explicit representation of path of motion is that we get the desired change of location inferences for free. In other words, in the sentence in example 4, the inference we want to retrieve is that the the book is on the table before the event begins and is no longer there at the end of the event. In the same way. the book is not on the floor before the event, but is there at the end of the event. Currently, VerbNet generally, albeit inconsistently, handles this by using the start() and end() predicates to show where the THEME is located at the specified point in time. Take the banish-10.2 class as an example:? Furthermore, the path usage of the location() predicate (as op- posed to the static location usage as shown in ex. 5) and the direction () predicate would be united under the same representation.
TABLE 2 A special LMT analysis of present and its Hungarian counterparts
TABLE 2 A special LMT analysis of present and its Hungarian counterparts
Both (5) and (6) have the same structure, since they differ only in the valence that the author assigns to the experience of day in the first two nodes. Yet, although the text never indicates explicitly at the sentence level what those sales figures were like, we have no trouble
Both (5) and (6) have the same structure, since they differ only in the valence that the author assigns to the experience of day in the first two nodes. Yet, although the text never indicates explicitly at the sentence level what those sales figures were like, we have no trouble
References 204 / LiviA POLANYI AND MARTIN HENK VAN DEN BERG
References 204 / LiviA POLANYI AND MARTIN HENK VAN DEN BERG
TABLE 1 Abbreviations used for Manhattan neighborhoods
TABLE 1 Abbreviations used for Manhattan neighborhoods
FIGURE 2 Multidimensional scaling (MDS) of Manhattan neighborhods based on word space similarities Figure 2 is a multidimensional scaling (MDS) of the 16 x 16 similarity matrix of the 16 neighborhoods shown in Table 1.
FIGURE 2 Multidimensional scaling (MDS) of Manhattan neighborhods based on word space similarities Figure 2 is a multidimensional scaling (MDS) of the 16 x 16 similarity matrix of the 16 neighborhoods shown in Table 1.
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved