Languages across the world differ from each other in a number of respects, and one such difference is in terms of how their lexicons are categorized. Compared to most European languages with distinct, functionally dedicated word classes in the traditional sense, quite a few languages are observed to possess lexical items that can fulfill the functions typically associated with more than one traditional word class such as ‘noun’ and ‘verb’. According to Rijkhoff and van Lier (2013), these lexemes exhibit what is called ‘flexibility’. Classical Chinese is observed to feature word-class flexibility, in the sense that there are lexemes that can be used to serve the functions of two or more traditional word classes, without the functional change being marked by any derivational means. For instance, a lexical item like xìn can either function as a verb meaning ‘to be trustworthy [intr.]’ or ‘to believe, to trust [tr.]’ or serve as a noun meaning ‘trust, oath of alliance’. Similarly, a human-denoting lexeme such as yŏu FRIEND cannot only mean ‘a friend’ but also ‘to be a friend, to behave friendly [intr.]’, ‘to make friends with [tr.]’ or ‘to consider as a friend [tr.]’; an instrument word like biān WHIP cannot only mean ‘a whip’ but also ‘to whip’. This situation is often thought to be related to the fact that Classical Chinese does not have any kind of productive morphology in the traditional sense (e.g. Zádrapa 2011). This is reflected in the lack of markedness distinctions across Croft’s (2000, 2001) conceptual space for parts of speech. This study ascribes flexibility of parts of speech in Classical Chinese to precategoriality, in line with Bisang (2008 a, b). Precategoriality can roughly be defined as the absence of the noun-verb distinction in the lexicon; instead, the linking of individual words to the syntactic position of N or V as well as their text frequency in these positions are subject to pragmatics. Precategorial lexical items are those that are not preclassified into parts of speech in the lexicon; rather, their word-class specification is ultimately determined at the syntactic level, according to their position/function in a given word-class indicating construction. From a diachronic viewpoint, this study assumes that precategoriality and categoriality of individual lexical items are not static, but that they are potentialities and tendencies that may change over time. Specifically, (full) precategoriality and (full) categoriality are assumed to constitute a continuum in the lexicon of Chinese throughout its history. In any given historical period, lexical items of the language are distributed between the two extremes on the continuum, according to the intensity of the association between their lexical meaning and the syntactic position/function of e.g. N or V. Generally, along the continuum at a given historical stage, lexemes with a strong association between meaning and function (i.e. lexemes that are normally associated only with one word-class specification for a particular syntactic role) tend to be located close to the extreme of (full) categoriality. In contrast, lexemes that are not necessarily related to one specific association between meaning and function, but can potentially occur in a variety of such associations, are assumed to be placed closer to (full) precategoriality instead. Roughly speaking, the group of lexemes that is located towards (full) precategoriality are flexible lexemes, though with varying degrees of flexibility, whose semantics licenses a syntactic variety and can thus be linked to more than one word-class specification through syntactic specification, a syntactically specified process of category assignment. Based on these considerations, this study aims to present the results of a corpus-based investigation into flexibility of parts of speech in Classical Chinese. The research focuses on two types of syntactic specifications of flexible lexemes, namely, those using action-denoting lexemes in nominal function (the V→N type), and those using object-denoting lexemes in verbal function (the N→V type). The two types of syntactic specifications are investigated for this study in the five Classical Chinese texts (Zuozhuan, Mengzi, Guoyu, Mozi, and Zhanguoce). Based on empirical facts, flexibility of parts of speech in Classical Chinese is addressed at three descriptive levels in this study: First, at the level of syntax, the discussion focuses on the most important syntactic configurations for the use of flexible lexemes and their relations to the basic word order of this language, with flexibility being observed in two positions of an argument structure construction: the V-position and the syntactic position of an argument. The findings of this study demonstrate that as far as the argument structure constructions formed with flexible lexemes are concerned, VO word order is much more frequent than OV. This strong preference for VO is, in connection with lexical flexibility, explained as follows: With the loss of derivational morphology in early stages of Old Chinese (e.g. Sagart 1999), word order became the most important indicator of word class and strongly supported the omission of strict verb-noun distinctions (co-existence of precategoriality and categoriality) in the lexicon of this language. Second, at the level of cognitive semantics (e.g. Lakoff 1987; Kövecses and Radden 1998; Schönefeld 2005), the discussion concentrates on the metonymic relationships that constitute the cognitive-semantic foundation of the use of flexible lexemes in Classical Chinese. In a metonymic mapping of either the V→N or the N→V type, the original semantics of a lexical item (which may typically be associated with a certain syntactic role of N or V) is used as a reference point to provide mental access to the newly derived meaning of the item in another syntactic function. Given the typologically salient characteristics of Classical Chinese discussed in this book, the argument is that the flexible use of an existing word form as a metonymically related but syntactically distinct item is one of the most economic ways in this language to name a new concept or a newly construed situation in discourse. Third, at the level of argument structure constructions (Bisang 2008a, b), the discussion focuses on how the different metonymic relationships mentioned interact with a given argument structure construction (which carries its own meaning within itself), and how these are further concretized into rule-based or metaphorically motivated pragmatic implicatures. A closer examination of an argument structure construction with an object word in the V-position reveals that there are two underlying frameworks for deriving the concrete meaning of the construction. In the rule-based framework, the verbal function of a given object word can basically be derived through grammatical analysis of the whole construction. In the metaphorical framework, the composed semantics of the construction actively interacts with the outside world in our conceptual system, where metaphor (Lakoff 1987, 1993; Kövecses 2010) serves as an essential cognitive principle in establishing and (re-)interpreting relations in the construction. The two mechanisms, rule-based and metaphorical, complement each other and work together to account for flexibility in Classical Chinese. This study argues that flexibility of parts of speech in Classical Chinese can only be fully understood by integrating a wide range of aspects, both linguistic and non-linguistic. The components that are needed to account for it include constructions (form-meaning pairings), semantics (Croft’s conceptual space), pragmatic implicatures, metonymies, metaphors, as well as world knowledge as reflected within a culture. In my view, it is reasonable to argue that these components need not be specific to the language investigated here; they are applicable to any language that shows flexibility in its parts-of-speech system.