224 Zhiying Xin
metaphors and metonymies. Then I will elaborate on some essential problems revealed by the
contributions with respect to corpus research.
In the initiating chapter, Anatol Stefanowitsch gives a critical overview of data-retrieving
methodologies and significant results of the corpus research concerning conceptual mappings.
Meanwhile, as one of the editors, he is careful to succinctly introduce the reader to the
methods adopted and major findings of each chapter. He starts by reflecting on the ways
metaphors and metonymies are extracted from the corpus by pointing out both advantages
and problems of each strategy. Early ―manual searching‖ limits the size of the corpus. The
method of ―searching for source domain vocabulary‖ can be exhaustive but criteria for
identifying target domains are badly needed. ―Target domain orientation‖ such as ―keyword-
based method‖ will not identify metaphorical expressions exhaustively or systematically. The
method of ―searching for sentences containing lexical items from both the source domain and
the garget domain‖ allows fast annotation but predicts problems of imperfect recall and
partial identification, etc. ―Searching for metaphors based on ̳markers of metaphor‘‖ realizes
automatic retrieval, but the markers are not always signals of metaphors. The strategy of
―extraction from a corpus annotated for semantic fields/domains‖ extends the first three
strategies. But here problems concerning semantic fields arise. Stefanowitsch is for the
method of ―extraction from a corpus annotated for conceptual mappings‖ which is,
unfortunately, not available yet. Despite the problems of these data extracting strategies,
results of the corpus-based approaches are fairly important and noticeable. Relating to the
achievements of the volume, Stefanowitsch highlights the following aspects: the
reexamination and evaluation of certain conceptual mappings; new insights into structural as
well as textual features of metaphor and metonymy; and discoveries relating to ―cross-
linguistic and diachronic issues‖. Stefanowitsch then proposes three criteria for appropriate
annotation of metaphorical mappings. In his view, such annotation must establish workable
and reliable procedures to spot metaphors and metonymies. In addition, it must define
relevant features for metaphor and metonymy, such as the source domain, the target domain,
and ―degree of metaphricity, metonymicity, and inter-rater reliability.‖ Concerning
―annotation formats‖, he recommends ―SGML (Standard Generalized Markup Language) /
XML (extensible Markup Language)‖ and several others. Stefanowitsch concludes his
chapter by drawing attention to the necessity of ―strict quantification and sophisticated
statistical methods‖ and the importance of putting empirical results into theoretical
considerations.
In the following chapter ―Metaphoricity is gradable‖, Patrick Hanks argues for a
continuum of metaphoricity instead of the traditional dichotomous distinction. Inspired by
Max Black‘s semantic ―resonance‖ theory that metaphoricity is determined by ―resonance
between at least two concepts, in which one (the primary subject) is interpreted in terms of
the other (the secondary subject)‖, Hanks explores the ―gradability‖ of metaphoricity by
focusing on sea and oasis in the British National Bank. The empirical investigations reveal
that there exists a ―metaphoricity cline‖ (29) in the usages of these words. At one extreme of
the cline is the literal meaning, and at the other extreme are highly metaphorical uses. To use
Black‘s terms, the fewer semantic features shared by two concepts, the more resonance, and
the greater metaphoricity. Therefore, some metaphors are more metaphorical than others (31).
To take oasis as an example, in ―one of several splendid oases of green in the city‖, the use of
oases is more metaphorical than its literal meaning (a place in desert), for it refers to a place
without any attributes of a desert. However, in ―it‘s about oases of control where there should