2.2Multimodality
As a new concept emerging in recent years, multimodality is widely applied to various fields, especially in systemic-functional and social linguistics. Earlier due to the limitation of science and technology, scholars barely analyzed discourse from a multimodal aspect. As Hu Zhuanglin (2007:1-10) mentioned that “now the rapid development of new technology and corpus have made the multimodal study of natural discourse possible”. Kress & Leeuwen thought that "Multimodality is to use several kinds of semiotics modes respectively or comprehensively to strengthen the expression of the same meaning, or to exercise supplementary function, or to sort the meaning in a hierarchical order" (Kress & Leeuwen, 2001:20). While another famous multimodal theory experts O’Halloran, combining essence of various point of views, gave the multimodality a new definition as "a theoretical analysis and practice comprehensively integrated of language, visual images, other symbolic resources, construction paper text, digital media and daily life text, things and events" (O'Halloran,2008:231). It means that in multimodal texts, the receiver needs to use multiple senses to absorb the external symbols in a variety of modal forms, and meanwhile, the information received through these different sensory modes will interact effectively or produce one image as part of the written text and it is also true of the other way around, etc. (Kress, 2001:7). It has become a new issue to discuss how to use this newly-arise way to convey meanings, to construct metaphors, to combine the multimodality with metaphors and explore the metaphors in multimodality.
In terms of multimodal metaphors, in Pictorial Metaphor in Advertising (Forceville, 1996), Charles Forceville analyzed pictorial metaphors in advertisements from a visual aspect. In Multimodal Metaphors-Application of Cognitive Linguistics in 2009, he expounded the multimodal metaphors from various perspectives. In the multimodal metaphor theory, Forceville pided metaphors into 2 categories: one is the mono-modal metaphor and the other is the multimodal one. The mono-modal metaphor means two domains using only one modality to represent the metaphorical meaning while the multimodal metaphor is one choosing different modalities respectively to characterize the hidden meaning in both source domain and target domain (Forceville，1998:384), of which the construction and understanding involve all kinds of modal symbols: images and sounds, demanding the participation of people’s multiple senses: vision and hearing, in order to realize the effective interaction among the symbolic information.
3. Theoretical framework: systemic view to multimodal metaphor
As an approach of linguistics, systemic functional linguistics regards language as a social semiotic system. The founder Michael Halliday, 'liberated' the dimension of choice from structure in a certain sense, making it the main organizing dimension of the theory. For Halliday, one principal theoretical principle is that “any act of communication involves choices”. Halliday thinks “linguistics is also ‘functional’ because he believes that languages have evolved under the pressure of the particular functions that the language system has to serve”. Systemic functional linguistics holds that language has three metafunctions including “the ideational metafunction, the interpersonal metafunction and the textual metafunction” (Halliday，2010：35). The idea of “context” is important in the study of Halliday and it is created with the help of the construed experience and the enacted social relations.来!自~751论-文|网www.751com.cn
The multimodal discourse symbols theory by Kress and Leeuwen is based on Halliday's theory. By introducing this sound theory into this field, Kress and Leeuwen's Reading Images (1996:20) rose attention to discourses that are of not solely verbal words but combinations of various semiotic resources. This framework provides a theoretical basis and analytical method for the visual image’s multimodal discourse analysis. Under the premise of considering the images as social symbols, they extended the ideas of metafunctions to visual patterns, creating the visual grammar centered on the “representational meaning, interactive meaning and compositional meaning”. Since then MDA has become a new sphere of Discourse Analysis and its study objects covers extensive materials, which includes “visual art, painting, sculpture and architecture” (O'Tool, 2010：78). Most of MDA researches later follow this social semiotic approach and analyzes the materials mentioned earlier. 多模态警示语的隐喻性表达(4):http://www.751com.cn/yingyu/lunwen_76688.html