This thesis investigates the structure, function, and diachronic development of the Chinese grammatical morpheme le within the frameworks of Construction Grammar (e.g., Croft 2001; Goldberg 1995, 2006) and Diachronic Construction Grammar (e.g., Traugott 2014; Traugott and Trousdale 2013; Trousdale 2010). In the existent research, the grammatical marker le is usually recognized as two separate morphemes that encode distinct functional meanings in view of the two distinct syntactic positions it occurs, e.g., the verb-final and sentence-final positions. This research re-examines the problems unsolved in previous proposals and explores the usage patterns of le from synchronic and diachronic perspectives. It aims to answer three key questions: 1) What is the core grammatical function of le? 2) Do one or two le’s exist? 3) Why does le display the diverse usages as observed? Three main issues on the function, constructionalization, and distribution of le that arise from different perspectives are addressed to figure out the actual relationship between the verbal le and the sentential le. Instead of treating le in the two positions as two distinct morphemes, this research proposes that one single morpheme le exists in terms of the form-meaning mapping unit, but it may occur in two different positions and also render an interpretational implicature in discourse relevance. The core function of le is to serve as a temporal anchor, i.e., it is an eventualization marker. This specific function indicates the actualization of an event in a broad sense, including action, state, or foreseeable situation. The proposed function can be utilized to its diverse usages and signals a Chinese-specific way of marking tense-aspect, which is distinct from the Indo-European languages.

The thesis first reviews the major previous proposals and distinguishes the inadequacies of previous studies in dealing with the issues of le, which encompass the notions of completion, perfectivity, and boundedness. Counterexamples to previous proposals are given to show that a more adequate and revealing account is still needed to explicate its diverse uses. Moreover, the review of the main previous proposals on the grammaticalization pathways and interaction of the verbal suffix le and sentence-final particle le may show that the traditional grammaticalization approach ignores that the aspects of form and meaning are encoded by the aspectual particle le/liǎo itself. Thus, its emergence and positional change issues have been unceasingly debated. It then introduces the data used for the synchronic and diachronic analyses, as well as the basic notions and principles of the theoretical framework in Chapter 3.

A constructional analysis of the grammatical function of le is provided in Chapter 4. A construction-based approach is used to analyze the essential grammatical function and temporal features that apply to the different uses of le in the two different positions. It proposes that le should be taken as the constructional operator marking the eventualization of the preceding X in the form-meaning pairing construct [X-le], where X can be any predicative constituent that may occur with le. This principled and well-motivated constructional account can encompass different usages of le in different syntactical positions, as it can be typically instantiated with varied VP combinations, such as [V], [V NP], or [V le NP]. These VP combinations can be viewed as constructional variants of the same [X-le] construction. The specific double use of le in a clause, e.g., 我看了電視了 wǒ kàn le diànshì le ‘I have watched TV.’, is considered an embedded structure [[V-le-NP]X-le], which contains two layers of [X-le]. The semantic distinctions emerging in the constructional variants can be attributed to the interaction between the basic function of le and its co-occurring X at varied positions, e.g., the clause-internal or clause-final position. Specifically, the [X-le] construction encodes the semantic notions of temporal realization and anteriority (X is realized by a reference time). In addition, the use of le in the clause-final position obtains the notion of discourse relevance by discourse implicature, which gives rise to the “currently relevant” reading (Li and Thompson 1981). This functionally unified constructional account can help solve some puzzling issues and benefit teaching and learning Chinese grammar by avoiding the unnecessary burden of constantly differentiating the two different le’s in the traditional view.

The basic consensus of previous research on the grammaticalization of the aspectual marker le is that it evolved from the completion verb (henceforth CV in short) 了 liǎo ‘to complete, to finish’, which was first observed in the 3rd century CE. The framework of constructionalization is used to examine the emergence and development of the aspectual particle liǎo and the event-actualization construction [X-liǎo], as discussed in Chapter 5. Agreeing with Mei (1981, 1999)’s proposal regarding the precursor of the structure V+liǎo+NP, this study proposes that the anterior-posterior temporal relationship encoded in the continuous compound sentence (CCS) in the form of [VP₁+CV, VP₂] was the original form-meaning constraint that needs to be surpassed for the constructionalization of an aspectual particle. It argues that the development of the proto-[X-le] construction mainly includes the constructionalization of the aspectual particle yǐ and the relay of liǎo to the aspectual yǐ in terms of form-meaning re-association with the aspectual function. Specifically, the frequent use of the CV 已 yǐ ‘to complete, to finish’ in Chinese-translated Buddhist scriptures initially paved the way for the later emergence of the aspectual particle liǎo in two aspects. First, the collocations of yǐ with different verb types triggered the semantic extension of yǐ, which ranged from signaling event completion to signaling event occurrence. Second, the string VP+yǐ went beyond the constraint pattern of CCS and began to be used more independently when it was employed to translate different tense-aspects other than absolutive gerund in Sanskrit, and yǐ was neoanalyzed as an aspectual particle to signal eventive occurrence. In the later period of the Tang Dynasty, liǎo modeled upon the usage patterns of yǐ in the re-association of form and meaning to render the function of indicating event-actualization. Along with the constructionalization of the aspectual particle liǎo, it was shown to develop into a full-fledged aspectual particle from being attached to the whole preceding VP in [VP-liǎo] to a smaller verbal unit in [V-liǎo-NP].

After the proto-construction [X-le] is defined in Chapter 4 and the constructionalization process of the aspectual particle liǎo is addressed in Chapter 5, a long-discussed issue of the different performances of the verbal le and the sentential le proposed in making sentences complete is dealt with in Chapter 6. That is, sentences like 他吃了藥 tā chī le yào ‘He took medicine’, where a bare noun is used after le, are usually considered incomplete or ungrammatical in previous studies. Saying 他吃了藥了 tā chī le yào le ‘He took medicine’ or adding another subsequent clause as in 他吃了藥就睡了 tā chī le yào jiù shuì le ‘He took medicine and fell asleep’ is more natural. Therefore, the verbal le has been claimed as inadequate in completing or ending a sentence (Guo 2015; Hu and Shi 2005; Lü 1983; Shen 1995; Sybesma 2019; Tang and Lee 2000; Tsai 2008). Nevertheless, the issue can be solved by considering the informativeness of the sentence. As long as the utterance is sufficiently informative in a given context, [V-le-NP] can be well used and understood, as in 他吃了毒藥 tā chī le dúyào ‘He took poison’ (Kong 1994: 436). Instead of contributing the incompleteness of sentences containing the verbal le to the inadequacy of the verbal le in completing or ending a sentence, the thesis expounds on the incompleteness of sentences with the verbal le from a historical perspective. It proposes that, after the constructionalization and positional shift of the aspectual particle liǎo, the string V+liǎo+NP gradually replaced the string VP+liǎo to signal the actualization of an event in an anterior-posterior temporal relationship with another event in the CCS. Moreover, the diachronic distribution of the aspectual particle liǎo in the forms of [V-liǎo-NP] and [VP-liǎo] are investigated. It further suggests that the inclined reference time type may help explain the incomplete reading of the verbal le. In particular, the predominant usage of the aspectual liǎo in the form of [V-liǎo-NP] in the CCS suggests that the verb-final liǎo tends to employ the time of a subsequent event as the reference time. By contrast, the predominant use of the aspectual particle liǎo in the form of [VP-liǎo] in the non-CCS implies that the sentence-final liǎo tends to use the time of speech as the reference time. The incompleteness of sentences with the verbal le can be eliminated when a reference time is specified in the context.

In summary, the proposed analysis of le as one form-meaning pairing morpheme, which can occur in different positions to form different constructional variants, offer a principled constructional account that may solve most of the unsolved issues in the literature. This research is significant in proposing a functional correlation between the varied uses of le with different co-occurring constituents in the [X-le] construction. It also explores the synchronic and diachronic evidence for the constructionalization of le based-on corpus data. Synchronically, the construction [X-le] is clearly defined, which may account for recent novel usages of le; diachronically, the shift of position in the uses of le is well motivated with a functional adaptation from parallel structures. The constructional account in this thesis provides a new and effective solution to tackle the notoriously difficult issue in Chinese grammar, which eventually contributes to research in theoretical and applied Chinese linguistics.