The central focus of this dissertation is the semantic and morphosyntactic analysis of tense, aspect, modality, and evidentiality (TAME) categories in Korean, using the frameworks of cognitive linguistics and construction grammar. Which of these grammatical categories Korean verbal markers belong to has been the subject of much debate. In this work, I pose the following questions:

- What are the systematic markers of TAME in the Korean verbal complex (KVC)?

- What are the meanings of the markers, and how do they fit into the larger system?

- How are markers of evidentiality and epistemic modality fitted into the broader grammatical system of the KVC?

- How does the analysis of these categories depend on an understanding of viewpoint; or what analysis of viewpoint do we need, underlying this analysis?

The answers to the first two questions are significant in that the grammatical categories tense, aspect, modality, and evidentiality in Korean have been contested and are in need of systematic accounts. Such an inquiry is even more significant in that the functions and distributions the Korean TAME elements have never been analyzed within the field of Cognitive Semantics. Starting from the assumption that meanings and functions are primary in linguistic analysis, this study begins by examining the paradigmatic units and syntagmatic relationships between the various tense and aspect elements in the KVC. I argue that grammatical units involving tense and aspect in Korean cannot be assigned purely to one category of the other. Rather, they should be given a more motivated analysis that takes into account their hybrid and versatile functional properties I therefore develop a multilayer approach to represent how information about tense, aspect, modality, and evidentiality is simultaneously accessed and exploited by interlocutors. In addition to demonstrating the interrelatedness of tense and aspect, this study further argues that tense and aspect information cannot be grasped without incorporating modality and evidentiality information.

After introducing the multilayer framework in Chapter 1, Chapter 2 provides a foundation for modeling the construal of TAME elements in a cognitively motivated way by introducing some key temporal reference points previously described in the literature on tense and aspect, including event time, speech-act time, and reference time, and positing a fourth key reference point, secondary viewpoint time (which may be instantiated as either assessment time or perception time). That chapter describes in detail the roles of and constraints on these key temporal parameters in the multilayer framework.

In Chapter 3, I apply my proposed multi-layered framework to a description of the temporal suffixes of the Korean verbal complex, exploring two major groups of nonterminal temporal elements. The first group is comprised of four suffixes, the non-past imperfective marker -nun-, the anteriority marker -ess-, the presumptive/volitive epistemic-modal -keyss-, and the retrospective firsthand evidential marker -te-. The second group is comprised of two durative markers, the progressive marker -koiss- and the resultative marker -eiss. I argue that none of these elements can be described simply as tense markers or aspect markers, but that each also involves the semantics of modality and evidentiality.

The answers to the second set of research questions are significant in that most previous analyses of Korean have subsumed evidentiality under the category of modality (Strauss 2005). I argue that the Korean grammar does in fact have a distinct grammatical system for evidentiality and give a systematic account of that system, approaching it from the angle of multiple frameworks within the general field of cognitive semantics (cf. formal semantics approach, D. Lim 2010, J. Lee 2010, K.-S. Chung 2005). The second half of this dissertation provides an overall picture of the evidentiality and epistemic modality systems in Korean and gives detailed descriptions from various angles of the three elements I identify as evidential markers, the firsthand evidential -te-, the inferential evidential -napo-, and the quotative/reportive evidential -ay. These descriptions highlight the fact that the nature of these markers cannot be grasped without understanding how the deictic settings of the origo's sensory perception, conceptualization, and subjective assessment of the relevant focal events affect the construal of utterances - i.e., understanding the ubiquitous effects of viewpoint. The importance of viewpoint motivates my use of cognitive approaches such as Mental Spaces Theory (Fauconnier 1997, Fauconnier and Turner 2002) and construction grammar (Fillmore and Kay 1988, 1999).

In Chapter 4, I provide a qualitative and theoretical overview of the modality and evidential elements in the KVC. First, I describe the four major terminal epistemic-modal suffixes, the indicative marker -e, the committal marker -ci, the mirative marker -kwun, and the factive/realization marker -ney. Then I claim that the KVC can express three types of modes of access to information using the direct/firsthand evidential marker -te-, the inferential evidential marker -napo-, and the quotative/reportive evidential -ay. I argue that contemporary Korean can therefore be described as having a typologically-rare scattered three-term evidentiality system (contra K.-S. Chung 2005 and J.-M. Song 2007).

In Chapter 5, I provide cognitive semantic accounts of some viewpoint phenomena related to the three Korean evidentiality markers, using Mental Spaces Theory. The three markers share the syntactic/semantic property that utterances containing them show asymmetrical distribution of first- and non-first-person subject depending on whether they are paried with activity predicates or experiential predicates. I argue that the best way to describe these asymmetries is not using abstract formal syntactic rules per se, but that they are a matter of semantics, namely, of the interactions between the viewpoints of the speaker (the origo) and the experiencer of the focal event (the grammatical subject). I claim that MST can be used to clearly model and capture these interactions and their effects. The discussion of these phenomena leads to a discussion of distancing effects (Dancygier and Sweetser 2005), contending that the cognitive distance between the multiple viewpoints relevant in the evidential constructions results in various syntactic and semantic constraints. To account for the conceptually distinct nature of grammatical evidentiality as compared to periphrastic descriptions of modes of access to information, Chapter 5 also proposes two novel means of mental-space elaboration, Backgrounded-Information Accommodation and Indirect Epistemic-Space Triggering. The former models how backgrounded information about mode of access is accommodated in the construal of utterances. The latter captures how inferential and quotative/reportive evidentials frame how the speaker's reasoning processes are triggered by evidence related to the focal event.

Finally, Chapter 6 argues that cognitive-linguistics approaches can better account for the TAME system of Korean in comparison with other formal frameworks in that they take into account the relevance of the cognizer's viewpoint, which is indispensible in understanding how the target constructions are construed. That chapter shows how my multilayer approach to TAME is based in cognitively grounded frameworks such as Cognitive Grammar (Langacker 1991) and x-schema models of aspect (Narayanan 1994). To further clarify my analysis of the Korean evidentials, I also provide construction grammar notations for the relevant constructions. The discussion in this chapter also includes a deeper exploration of the nature of inferential evidentiality, which has not received much attention in the literature, examining and schematizing its internal event structure.