Lexical aspect
Lexical aspect
Main page
L

Lexical aspect

logo
Community Hub0 subscribers
Read side by side
Lexical aspect
View on Wikipedia
from Wikipedia
Grammatical features
Related to nouns
Related to verbs
General features
Syntax relationships
Semantics
Phenomena

In linguistics, the lexical aspect, situation type or Aktionsart (German pronunciation: [ʔakˈtsi̯oːnsˌʔaːɐ̯t], plural Aktionsarten [ʔakˈtsi̯oːnsˌʔaːɐ̯tn̩]) of an event is part of the way in which that event is structured in relation to time. For example, the English verbs arrive and run differ in their lexical aspect since the former describes an event which has a natural endpoint while the latter does not. Lexical aspect differs from grammatical aspect in that it is an inherent semantic property of a predicate, while grammatical aspect is a syntactic or morphological property. Although lexical aspect need not be marked morphologically, it has downstream grammatical effects, for instance that arrive can be modified by "in an hour" while believe cannot.

Theories of aspectual class

[edit]

Although all theories of lexical aspect recognize that verbs divide into different classes, the details of the classification differ. An early attempt by Vendler recognized four classes, which has been modified several times.

Vendler's classification

[edit]

Zeno Vendler classified verbs into four categories on whether they express "activity", "accomplishment", "achievement" or "state". Activities and accomplishments are distinguished from achievements and states in that the first two allow the use of continuous and progressive aspects. Activities and accomplishments are distinguished from each other by boundedness. Activities do not have a terminal point (a point before which the activity has taken place and after which it cannot continue: "John drew a circle"), but accomplishments have one. Of achievements and states, achievements are instantaneous, but states are durative. Achievements and accomplishments are distinguished from one another in that achievements take place immediately (such as in "recognise" or "find"), but accomplishments approach an endpoint incrementally (as in "paint a picture" or "build a house").[1][2]

Comrie's classification

[edit]

In his discussion of lexical aspect, Bernard Comrie included the category semelfactive or punctual events such as "sneeze". His divisions of the categories were as follows: states, activities, and accomplishments are durative, but semelfactives and achievements are punctual. Of the durative verbs, states are unique as they involve no change, and activities are atelic (that is, have no "terminal point") whereas accomplishments are telic. Of the punctual verbs, semelfactives are atelic, and achievements are telic. The following table shows examples of lexical aspect in English that involve change (an example of a state is 'know').[1][3]

Punctual Durative
Telic Achievement
(to release) 		Accomplishment
(to drown)
Atelic Semelfactive
(to knock) 		Activity
(to walk)
Static State
(to know)

Moens and Steedman's classification

[edit]

Another classification is proposed by Moens and Steedman, based on the idea of the event nucleus.[4]

Event nucleus
Preparatory

phase

Culminating

event

Consequent

phase

Semelfactive
State
Activity
Achievement
Accomplishment

Syntactic analyses of event structure

[edit]

Aspectual classes can be analyzed as differing in their event structure, and this has led to the development of syntactic analyses of event structure, with each aspectual class treated as having a distinct syntactic structure.

See also

[edit]

References

[edit]
Revisions and contributorsEdit on WikipediaRead on Wikipedia

Lexical aspect

View on Grokipedia
from Grokipedia
Lexical aspect, also known as aktionsart, is a linguistic category that refers to the inherent temporal properties encoded in the meaning of verbs or verb phrases, distinguishing how events or states unfold over time without reference to grammatical markers of tense or viewpoint.[1] These properties include features such as telicity (whether an event has a natural endpoint or culmination), durativity (whether it extends over time or is punctual), and dynamism (whether it involves change or is static).[2] Unlike grammatical aspect, which imposes a perspective on the event (e.g., ongoing or completed), lexical aspect is a semantic classification determined by the verb's lexical meaning and its arguments.[1] The foundational framework for lexical aspect was proposed by philosopher Zeno Vendler in 1957, who classified English verbs into four primary categories based on their temporal schemata: states, activities, accomplishments, and achievements.[2] States, such as know or love, describe static situations without inherent change or duration, resisting progressive forms (e.g., She knows the answer, but not She is knowing the answer).[2] Activities, like run or sing, denote dynamic processes that are atelic (lacking an endpoint) and durative, compatible with continuous tenses (e.g., He ran for an hour).[2] Accomplishments, such as build a house or run a mile, involve a process leading to a telic endpoint, taking a measurable duration to complete (e.g., She painted the room in two hours).[2] Achievements, exemplified by win or notice, are punctual events that occur instantaneously at a climax, queried by time points rather than durations (e.g., He arrived at 5 p.m.).[2] This classification has proven influential in semantics, syntax, and language acquisition studies, though later refinements include additional categories like semelfactives (punctual, non-telic events such as cough) and account for how arguments or context can shift a verb's aspectual class.[1] Lexical aspect interacts with grammatical aspect to shape sentence interpretation, affecting phenomena like the distribution of adverbials (e.g., for an hour with atelic predicates, in an hour with telic ones) and cross-linguistic variations in event encoding.[1] In computational linguistics, automatic classification of lexical aspect remains a challenge due to ambiguity and context-dependence, with datasets supporting research in natural language processing.[1]

Fundamentals

Definition and core concepts

Lexical aspect, also known as Aktionsart, refers to the inherent temporal properties encoded in the meanings of verbs and predicates, classifying situations in terms of their duration, completion, and internal structure, independent of grammatical tense or mood.[3] This intrinsic classification describes how events or states are structured temporally, focusing on features such as whether they involve change, have endpoints, or extend over time.[4] Core concepts in lexical aspect revolve around several key binary distinctions that capture the semantic properties of predicates. The telic-atelic distinction differentiates events with a natural endpoint or goal (telic, e.g., build a house) from those without such a boundary (atelic, e.g., run).[5] The dynamic-static opposition separates predicates involving internal change or activity (dynamic, e.g., run) from those denoting unchanging conditions (static, e.g., know).[3] Additionally, the punctual-durative contrast distinguishes instantaneous occurrences (punctual, e.g., die) from those that unfold over a period (durative, e.g., sleep).[4] These distinctions influence event interpretation, such as compatibility with adverbials like for an hour (favoring durative, atelic predicates) versus in an hour (favoring telic ones).[5] Illustrative examples highlight these properties: sleep is durative and atelic, portraying an ongoing activity without inherent completion, while die is punctual and telic, marking an instantaneous, bounded change of state.[3] Such semantic implications affect how predicates combine with arguments or modifiers, shaping the overall temporal contour of sentences. The term Aktionsart originated in 19th-century Slavic linguistics, where scholars like Nikolaj Greč (1827) distinguished aspectual properties from tense, and was formalized by Karl Brugmann (1885) in Germanic contexts before Sigurd Agrell (1908) separated it from grammatical aspect in Slavic studies.[6] It gained prominence in English-language semantics after the 1950s, notably through Zeno Vendler's (1957) systematic classification building on these foundational binaries.[7]

Distinction from grammatical aspect

Lexical aspect refers to the inherent temporal structure encoded in the semantics of individual verbs or verb phrases, determined by the verb's meaning and its arguments, while grammatical aspect involves viewpoint-based encodings added through morphological or periphrastic constructions that frame the event's presentation. This distinction, foundational to the study of verbal aspect, was systematically outlined by Bernard Comrie in his 1976 monograph, emphasizing that lexical aspect pertains to situation types intrinsic to the verb (such as telic or atelic), whereas grammatical aspect overlays a speaker's perspective, like ongoingness or completion, via language-specific mechanisms. For instance, in English, the verb run carries an atelic lexical aspect, but grammatical aspect shifts it to progressive in is running or, less commonly, to a completed viewpoint in certain contexts; in contrast, Slavic languages employ dedicated perfective affixes to mark completion morphologically.[8] The interaction between the two arises prominently in how lexical properties constrain grammatical interpretations, particularly with telicity affecting perfective forms. Atelic verbs, lacking an inherent endpoint, often resist perfective grammatical aspect without additional delimiters that alter their lexical profile to telic, such as prefixes in Slavic languages or measure phrases like "in an hour." In Russian, for example, the atelic imperfective čitat' knigu ("read a book") becomes compatible with perfective aspect only when prefixed as pro-čitat' knigu, imposing a telic reading of completion; bare atelic uses with perfective marking yield infelicitous results unless contextually bounded. This interplay ensures that grammatical aspect does not override but builds upon lexical foundations, preventing mismatches in event portrayal.[9] Common confusions stem from languages where morphological distinctions are minimal, blurring the lexical-grammatical boundary through constructional means. In Mandarin Chinese, serial verb constructions frequently imply aspectual shifts—such as adding resultative or directional elements to telicize atelic verbs—without overt affixes, effectively merging inherent verb semantics with viewpoint encoding.[10] For instance, tā pǎo jìn wūzi ("he run enter room") combines an atelic motion verb with a path complement to convey a perfective, bounded event, relying on lexical composition rather than dedicated grammatical markers.[10] Such patterns, prevalent in isolating languages, highlight how the universal distinction adapts to typological variation while maintaining conceptual separation.

Major Classifications

Vendler's verb classes

Zeno Vendler introduced a foundational classification of verbs into four categories based on their inherent temporal properties in his 1957 paper "Verbs and Times."[11] This system distinguishes verbs according to whether they denote static or dynamic situations, atelic or telic events, and durative or punctual processes, providing a framework for understanding lexical aspect in English.[11] The four classes are as follows:
  • States: These are static, atelic, and durative, describing unchanging conditions without inherent endpoints or internal structure, such as "know the answer" or "love music." They lack dynamism and do not progress over time.[11]
  • Activities: Dynamic, atelic, and durative, these verbs denote ongoing processes without a natural culmination, exemplified by "run" or "walk." They involve effort or change but can continue indefinitely.[11]
  • Accomplishments: Dynamic, telic, and durative, these involve processes leading to a defined endpoint, like "paint a picture" or "run a mile." The situation unfolds over time until completion.[11]
  • Achievements: Dynamic, telic, and punctual, these mark instantaneous changes or culminations, such as "recognize someone" or "reach the summit." They lack duration and focus on the moment of attainment.[11]
Vendler proposed diagnostic criteria rooted in compatibility with temporal adverbials to differentiate these classes. States and activities accept phrases like "for an hour" (e.g., "She knew the answer for an hour"), indicating duration without completion, but reject "in an hour" (e.g., *She knew the answer in an hour). Accomplishments and achievements, conversely, pair with "in an hour" to denote the time span to culmination (e.g., "She painted a picture in an hour"), but not typically with "for an hour" for the entire event (*She painted a picture for an hour). These tests highlight the telic/atelic distinction underlying the classes, where telic verbs imply boundedness.[11][12] Published in The Philosophical Review, Vendler's work drew from Aristotelian notions of change and time, influencing subsequent linguistic theories, including generative semantics, where verb meanings were decomposed into aspectual primitives reflecting these classes.[11][13] Despite its influence, Vendler's classification reveals ambiguities, as the same verb root can shift classes depending on context or complements; for instance, "run" is an activity, but "run a mile" becomes an accomplishment due to the telic object imposing an endpoint.[11]

Comrie's refinements and tests

Bernard Comrie, in his 1976 book Aspect: An Introduction to the Study of Verbal Aspect and Related Problems, expanded upon Vendler's four verb classes by introducing refinements to better capture the nuances of lexical aspect, particularly through the addition of semelfactives as a distinct category or subcategory of achievements.[14] Semelfactives denote punctual, atelic events that occur once without implying duration or a resultant state, such as "cough" in English or the Russian verb kašljanut' meaning "to cough once."[14] These differ from achievements, which are punctual but telic (e.g., "notice"), by lacking an inherent endpoint or change, thus allowing for easy iteration without altering the event's atomic nature.[14] Comrie proposed several aspectual tests to identify and distinguish verb classes more precisely, drawing on temporal and semantic compatibility. Time adverbial tests include: (1) Durativity: durative situations (states, activities, accomplishments) compatibly combine with "for X time" phrases (e.g., "She sang for an hour"; "She built a house for two years"), while punctuals (achievements, semelfactives) do not (*He noticed for an hour). (2) Telicity: telic verbs (accomplishments, achievements) pair with "in X time" to indicate completion (e.g., "She built a house in an hour"), whereas atelic verbs (states, activities, semelfactives) pair with "for X time" for duration.[14] Iteration tests assess repeatability: activities and semelfactives permit adverbials like "again" or repeated actions without implying completion (e.g., "She ran again"), whereas states resist such iteration (e.g., ?? "She knew the answer again") and telic events like accomplishments imply finality upon repetition.[14] Conative tests, involving attempt constructions such as "try to V," highlight duration in accomplishments by allowing partial progress (e.g., "She tried to run the marathon" implies effort over time), in contrast to instantaneous achievements (e.g., ?? "She tried to notice the bird").[14] Comrie also addressed overlaps and shifts between classes, noting that lexical aspect is not fixed but can change with modifiers or context; for instance, the activity verb "run" becomes an accomplishment when specified with a goal, as in "run to the store," introducing telicity and a resultant state.[14] Such shifts underscore the compositional nature of aspect, where bare verbs may belong to one class but combine with arguments or adverbials to alter their temporal structure.[14] These refinements were empirically grounded in analyses of English alongside Slavic languages like Russian and Bulgarian, where morphological aspect (perfective/imperfective) interacts with lexical classes to validate the tests; for example, Russian semelfactives often use imperfective forms for single events, contrasting with iterative uses.[14] This cross-linguistic data helped Comrie demonstrate the universality of the diagnostics while highlighting variations in how languages encode punctuality and telicity.[14]

Moens and Steedman's model

In 1988, Marc Moens and Mark Steedman proposed a computational model of lexical aspect that decomposes verb meanings into structured event templates composed of nuclear predicates, aiming to capture the temporal structure of events for natural language processing.[15] This approach treats events as composites built from basic predicates such as PROCESS, which denotes ongoing, extended activities without inherent endpoints (e.g., "climb" or "run"), and BECOME, which indicates a punctual transition or change of state (e.g., "reach the top").[15] These nuclear predicates can combine with additional elements like culminations (result states) or preparations (preceding processes) to form more complex event types, allowing the model to represent aspectual properties dynamically through contingency relations among event subparts.[15] The model defines three primary event templates: the simple process, corresponding to atelic activities that lack a natural boundary and can extend indefinitely; the transition, akin to achievements, where a direct BECOME predicate marks an instantaneous change without prior extension; and the culminated process, which models accomplishments as a PROCESS leading to a BECOME culmination, incorporating both duration and a terminal point.[15] For instance, "Harry climbed the mountain" is analyzed as a culminated process: an extended PROCESS of climbing that culminates in the BECOME state of having reached the summit.[15] This template structure also accommodates preparations (initial phases leading to the main event) and consequent states (post-event results), providing a flexible ontology for event decomposition.[15] Formally, the model employs predicate logic to represent these structures, integrating nuclear predicates with logical operators to denote temporal relations. A classic example is "John ate the apple," formalized as PROCESS(eat, John, apple) ∧ BECOME(eaten, apple), where the PROCESS captures the incremental consumption and the BECOME denotes the completion, resulting in a culminated process with an incremental theme (the apple's gradual disappearance).[15] Such representations enable precise modeling of aspectual composition, including how arguments influence telicity.[15] Originally developed for computational parsing in natural language understanding systems, the model facilitates the interpretation of temporal references and event sequences by linking lexical aspect to syntactic structures, influencing subsequent work in AI semantics and discourse analysis.[15] Compared to Vendler's verb classes, which rely on static categories, Moens and Steedman's framework offers advantages in handling dynamic aspectual shifts, such as coercion (e.g., viewing an accomplishment as an activity under the progressive) and incremental themes, through its decompositional templates that adapt to contextual contingencies.[15]

Syntactic and Semantic Analyses

Event structure decomposition

In formal semantics, lexical aspect is analyzed through the decomposition of verb meanings into structured events composed of atomic semantic primitives, allowing for a precise representation of temporal properties such as duration, change, and culmination. This approach posits that verbs are not semantically primitive but can be broken down into basic predicates like DO (for agentive actions), BECOME (for changes of state), and CAUSE (for causation), which combine to form complex event structures. For instance, the verb "kill" is decomposed as CAUSE(BECOME(dead, x)), where an agent causes a change in the theme x to a dead state.[16] David Dowty's 1979 aspectual calculus provides a foundational framework for this decomposition, integrating it with Montague grammar to model how verbal predicates encode aspectual distinctions through logical forms involving these primitives and temporal operators. This calculus distinguishes between static relations (states) and dynamic processes by specifying conditions on event progress, such as the requirement for a BECOME predicate to mark a transition. Extensions in event semantics, particularly Terence Parsons' 1990 work, further develop this by reifying events as entities in the domain of discourse, enabling subatomic analysis where verbs quantify over event parts (e.g., initial and final subevents) to capture aspectual behavior. Parsons' neo-Davidsonian approach represents verb meanings as relations between events and their participants, with aspectual properties emerging from the event's internal structure.[17] Telicity, the boundedness of an event, arises in these decompositions through the interaction of verbal structure with thematic arguments, particularly incremental themes that measure event progress. An incremental theme, such as "an apple" in "eat an apple," contributes to telicity by establishing a homomorphism between the mereological structure of the object (its parts) and the temporal stages of the event, ensuring the event culminates when the object is fully consumed. Manfred Krifka's 1989 analysis formalizes this via mapping functions that align part-whole relations in the theme with subevents, rendering the predicate telic when the homomorphism covers the entire argument. In formal semantics, aspectual composition is often represented using lambda calculus to combine decomposed predicates with arguments and measures. For example, the telic reading of "eat an apple" can be glossed as λe.∃x [eat(e)∧θ(e,x)∧apple(x)∧μ(x)=1]\lambda e. \exists x \, [\textit{eat}(e) \land \theta(e,x) \land \textit{apple}(x) \land \mu(x) = 1], where θ\theta denotes the theme relation and μ\mu is a measure function that reaches a maximum value at the event's endpoint. This representation highlights how lexical decomposition interfaces with quantification to derive aspectual entailments. Evidence for such decompositions comes from diagnostic properties like the subinterval test: states hold uniformly over all subintervals of their runtime (e.g., "know" is true throughout any part of the knowing interval), whereas accomplishments only culminate at the final subinterval (e.g., "build a house" does not entail completion in proper subintervals). This precursor to full decomposition appears in template-based models like Moens and Steedman's 1988 approach to event templates.[18]

Interaction with syntax and arguments

In generative grammar, lexical aspect interfaces with syntax through the projection of functional categories such as the little vP (v-bar projection), which encodes causation and introduces external arguments for telic predicates. Telic verbs, particularly accomplishments and achievements, typically project a vP shell to license the causative component of the event, thereby linking the external argument—often an agent—to the theta-role of initiator. For instance, in the sentence "John built the house," the telic verb "build" requires a vP projection to causally relate the agent John to the internal event of house-building, ensuring the event's bounded nature is syntactically realized. This projection distinguishes unaccusative structures (lacking an external argument) from transitive ones, where vP enforces the assignment of an agent theta-role.[19] Lexical aspect also plays a crucial role in argument structure by influencing theta-role assignment, particularly in how internal arguments contribute to telicity. Accomplishments, as a Vendlerian class, inherently require a patient or theme argument to delimit the event's endpoint, thereby achieving telicity; without such an argument, the predicate defaults to an atelic activity. For example, "John pushed the cart" (accomplishment, telic due to the bounded patient "the cart") contrasts with "John pushed" (activity, atelic), where the absence of a measuring-out patient prevents culmination. This aspectual constraint on theta-role assignment ensures that the patient theta-role is syntactically projected as a direct object in vP complements, regulating the verb's argument-linking possibilities.[20] A seminal theory integrating lexical aspect with syntactic argument structure is Hale and Keyser's (1993) conflation approach, which posits that verbs are formed via syntactic incorporation and conflation within a lexical-syntactic (l-syntax) framework. This accounts for the manner/result alternation, where manner verbs like "sneeze" resist resultative complements (*"John sneezed the handkerchief clean") unless incorporation conflates the manner root into a causative vP structure, as in "cough up phlegm," yielding a telic result. Hale and Keyser argue that such alternations arise from syntactic constraints on argument realization, with telic results requiring vP-mediated causation to link the external argument to the event's endpoint.[21] Empirical evidence for these interactions appears in phenomena like aspectual coercion, where syntactic contexts force a shift in a verb's lexical aspectual class. For example, the activity verb "run" in "He stopped running a marathon" undergoes coercion to an accomplishment interpretation, implying completion of the bounded event; this shift is licensed by the aspectual verb "stop," which syntactically selects for a telic complement, overriding the verb's inherent atelicity through vP recomposition. Such coercions highlight how syntax enforces aspectual compatibility, often resolving mismatches via type-shifting operators in the argument structure. This syntactic perspective ties into broader classifications, as seen in Levin's (1993) verb classes, which group verbs by their aspectual behavior in alternations, such as the causative/inchoative alternation where telic change-of-state verbs like "break" alternate based on external argument presence. Levin's framework demonstrates how lexical aspect determines syntactic transitivity and argument options, with classes like "verbs of putting" requiring patient arguments for telic projections, thus bridging aspectual properties to generative argument linking.[22]

Cross-Linguistic and Theoretical Extensions

Variations across languages

Lexical aspect manifests differently across non-Indo-European languages, often challenging the universality of classifications like Vendler's verb classes, which require adaptation for morphological and typological variations. In Mandarin Chinese, telicity is frequently influenced by classifiers and measure words that bound events, interacting with lexical properties of verbs. For instance, the activity verb chī yàozǐ ('eat medicine') denotes an atelic event without a natural endpoint, but adding the perfective marker le in chī-le yàozǐ renders it telic by signaling completion, highlighting how grammatical elements override or modify inherent lexical aspect in classifier languages.[23][24] Similarly, in Akan, a Kwa language, serial verb constructions encode aspectual phases of events through sequences of verbs that represent initiation, progression, and termination as a single predicate, without overt coordination; for example, constructions like kɔ̀ sɛ̀ɛ́ bɔ́ ('go take come') depict a chained event with shared tense but aspect marking primarily on the initial verb, thus layering lexical aspectual information across the series.[25][26] Typological diversity further complicates standard diagnostics, as seen in languages lacking clear Vendlerian distinctions. In Navajo, an Athabaskan language, aspect is classifier-based rather than strictly verb-class driven; the four classifiers (∅, D-, ł-, yi-) not only indicate transitivity but also interact with stem sets to mark imperfective or perfective modes, rendering traditional Vendler categories like achievements or accomplishments less applicable without considering the verb base's morphological template.[27] In Bantu languages, boundedness emerges through noun class agreement rather than inherent verbal telicity; for example, in Nyamwezi, the interaction of aspectual markers like -ř-…-íle with noun class prefixes on objects can shift atelic activities toward telic interpretations by delimiting the event via referential specificity, underscoring the limits of Vendler-inspired models in agglutinative systems.[28][29] In polysynthetic languages like Inuktitut, Comrie's diagnostic tests for lexical aspect—such as the subinterval test for progressives or the imperative test for statives—often fail due to the heavy incorporation of roots, affixes, and arguments into single words, necessitating analysis at the verb root level to isolate inherent aspectual properties from derivational morphology. For instance, viewpoint aspect in Inuktitut is not wholly determined by lexical telicity, as polypersonal agreement and postbases obscure boundaries between situation and grammatical aspect, requiring root-focused elicitation to apply adapted diagnostics.[30][31] Post-2000 research has highlighted gaps in understanding lexical aspect in creoles and sign languages, where understudied data reveal unique encodings. In creoles like those derived from French or English contact, such as Haitian Creole, lexical aspect restructures through substrate influences, with telicity emerging via serialized verbs or particles rather than inflection, as explored in studies on tense-aspect realignment during creolization.[32] In sign languages, post-2000 investigations into emerging systems like Nicaraguan Sign Language show lexical aspect conveyed through non-manual markers and classifier handshapes that depict event phases, often blending iconicity with temporal bounding in ways absent from spoken language models.[33] These findings expose research gaps, particularly Wikipedia's undercoverage of non-European data, emphasizing the need for more cross-modal and contact-language studies. Such variations pose challenges to Eurocentric models of lexical aspect, which prioritize Indo-European verb-centric classifications, prompting calls for universal primitives adjusted to accommodate morphological diversity like classifiers, seriality, and polysynthesis. This typological lens reveals biases in applying Vendler or Comrie frameworks without language-specific adaptations, advocating for inclusive methodologies that integrate non-Western structures to refine aspectual theory.[34][35]

Applications in formal semantics

In formal semantics, lexical aspect has been integrated into frameworks extending Montague grammar, where verb meanings are composed with temporal operators to capture event structures. Early extensions treated aspectual classes as part of the denotation of verbs within intensional logics, allowing for the representation of telicity and duration through lambda calculus abstractions.[36] Beyond this, Discourse Representation Theory (DRT), developed by Hans Kamp, incorporates lexical aspect to handle aspectual anaphora, such as references to ongoing or completed subevents in discourse. In Kamp and Rohrer's 1983 work, tense and aspect are modeled dynamically, enabling anaphoric links between events based on their aspectual properties, like the progression from an atelic activity to a telic achievement. In computational linguistics, lexical aspect plays a key role in natural language processing (NLP) tasks involving temporal inference, particularly through annotation schemes that mark event types for downstream applications like timeline extraction. TimeML, a standard developed in the early 2000s and refined post-2005, annotates texts with aspectual features such as perfective or imperfective interpretations, facilitating the resolution of temporal relations in corpora. This scheme integrates Vendlerian classes to tag events, supporting algorithms for inferring sequence and overlap in narratives.[37][38] Automatic TimeML annotators, evaluated on benchmarks like TimeBank, demonstrate how aspectual tagging supports temporal ordering accuracy in English texts.[39] Cognitively, lexical aspect links to event cognition through models of force dynamics, where aspectual properties encode how forces influence event unfolding. Leonard Talmy's 2000 framework integrates aspect with motion events, positing that atelic processes reflect sustained force application, while telic accomplishments denote force overcoming resistance to reach a boundary. This approach extends to psychological domains, explaining how aspect shapes conceptual representations of causation and persistence in human cognition.[40][41] Recent advances in vector semantics leverage contextualized embeddings to model aspectual implicatures, such as the inference that a progressive form implies incompleteness. BERT-based models capture these through layered representations, where verb embeddings encode telicity based on contextual cues, achieving high performance in aspect classification tasks. Adversarial testing reveals that BERT embeddings robustly represent aspectual features like duration and boundedness, outperforming static embeddings in zero-shot settings.[42] In construction grammar, critiques of Vendler's classes highlight their rigidity, arguing that aspect emerges from construction-verb interactions rather than inherent lexical properties, as seen in analyses of aspectual shifts in idiomatic expressions.[43] Future directions emphasize integrating lexical aspect detection into AI systems for multilingual applications, using large language models to classify aspect across languages with limited resources. Evaluations show that models like GPT-4 achieve F1-scores of around 72-76% in aspect detection tasks.[44]

References

  1. [PDF] A Kind Introduction to Lexical and Grammatical Aspect, with a ...
  2. [PDF] Verbs and Times
  3. https://onlinelibrary.wiley.com/doi/10.1002/9781118788516.sem038
  4. Aktionsart - Glottopedia
  5. [PDF] LEXICAL ASPECT - ResearchGate
  6. [PDF] Aspect and Aktionsart: some History - DSpace
  7. Verbs and Times - jstor
  8. (PDF) Lexical Aspect - ResearchGate
  9. [PDF] On the Grammaticalization of Mandarin Aspect Markers - DiVA portal
  10. [PDF] Verbs and Times Zeno Vendler The Philosophical Review, Vol. 66 ...
  11. [PDF] “Verbs and Times” Vendler (1957) - LaBRI
  12. [PDF] The Lexical Semantics of Verbs II: Aspectual Approaches
  13. Aspect | Cambridge University Press & Assessment
  14. https://aclanthology.org/J88-2003
  15. [PDF] chapter 2
  16. Word Meaning and Montague Grammar - SpringerLink
  17. Events in the Semantics of English - MIT Press
  18. Toward the Logic of Tense and Aspect in English - ResearchGate
  19. [PDF] THEORIES OF ARGUMENT STRUCTURE
  20. [PDF] Aspect, Lexical Semantic Representation, and Argument Expression1
  21. [PDF] HaleAndKeyser1993.pdf - The University of Arizona
  22. English Verb Classes and Alternations: A Preliminary Investigation ...
  23. Telicization in Mandarin Chinese | Journal of Linguistics
  24. Grammatical and Lexical Aspect in Mandarin Chinese - Academia.edu
  25. https://zenodo.org/record/6393762/files/306-Sibanda-2022-15.pdf
  26. [PDF] Serial verb constructions and their event representations in Akan
  27. Aspectual Categories in Navajo
  28. (PDF) Aspectuality in Bantu: on the limits of Vendler's categories
  29. https://www.degruyterbrill.com/document/doi/10.1515/lingty-2019-0017/html
  30. [PDF] THE MANIFESTATION OF VIEWPOINT ASPECT IN INUKTITUT*
  31. [PDF] Viewpoint Aspect in Inuktitut: The Syntax and ... - bac-lac.gc.ca
  32. [PDF] the restructuring of tense/aspect systems in creole
  33. [PDF] EMERGING SIGN LANGUAGES OF THE AMERICAS
  34. A Decade of Language Processing Research: Linguistic Diversity?
  35. [PDF] Towards a fieldwork methodology for eliciting distinctions in lexical ...
  36. [PDF] Lecture 9: Tense, Aspect, Events 0. Introduction.
  37. TimeML Specification 1.2.1
  38. (PDF) TimeML: A Specification Language for Temporal and Event ...
  39. [PDF] Holistic Evaluation of Automatic TimeML Annotators - ACL Anthology
  40. [PDF] This excerpt from Toward a Cognitive Semantics - Vol. 1. Leonard ...
  41. Toward a Cognitive Semantics - MIT Press Direct
  42. [PDF] Investigating Aspect Features in Contextualized Embeddings with ...
  43. [PDF] Aspectuality in Bantu: On the limits of Vendler's categories
  44. (PDF) Evaluating Lexical Aspect with Large Language Models
User Avatar
No comments yet.