5.1.2　A Working principle of local coherence in the English learning advertisements

Responding to the comment of current research in visual-verbal relations in the previous section，this section attempts to propose a working principle of local coherence with which analysis can be done with systematic，operational criteria and sound empirical basis.As mentioned above，local coherence，in this chapter，refers specifically to the semantic connection between adjacent verbal and visual elements within each stage of complex multimodal discourses.

The central concept in the analytic tool in this book is information linking between verbal and visual modes.As widely acknowledged，a common principle in the design of multimodal discourses is that the ideational meaning represented in language and in image must have some overlapping or linking（see for example，Iedema，2003）.The phenomenon of information link between semiotic modes has been widely analyzed in the existing frameworks of text-image relations in different names，including redundancy，semiotic metaphor，concurrence，among others.Moreover，the variety of terms in the above-mentioned frameworks，such as complementarity and instantiation，has also been proposed on the basis of the assumption that there is a connection in information between language and image at the content level.

This book maintains that the connectivity of ideational meaning，or information linking between different semiotic modes is the essence of the coherence in multimodal discourses.In some cases the connectivity is direct and obvious while in other cases the connectivity is indirect and needs to be made through inferences.In the cases which lack apparent information linking，the inferences made on the reader's part are important for the coherent interpretation of the semiotic construct，and actually these inferences are often intended by the designer.This section will elaborate on the principle of information linking between verbal texts and visual images.

First of all，the principle of visual-verbal relations in this chapter assumes that like language，images can also be analyzed by the unit of meaning—the information unit.First of all，the concept of information unit should be clarified.This concept can be used for both language and image.For language，the information unit is similar to the notion“sememe”.A sememe is a proposed unit of transmitted or intended meaning;it is atomic or indivisible.A sememe can be the meaning expressed by a morpheme，such as the English pluralizing morpheme-s，which carries the sememic feature[+plural].Alternately，a single sememe（for example[go]or[move]）can be conceived as the abstract representation of such verbs as skate，roll，and jump.If extended to the multimodal sense，the unit of meaning should be changed to information unit which can cover a large variety of semiotic modes.That is，the unit of meaning can also be thought of as the semantic counterpart to any of the following：a meme in a culture，a gene in a genetic make-up，or an atom（or，more specifically，an elementary particle）in a substance.Thus，the proposal in this chapter is that it can also be thought of as the semantic unit of the visual images as a semiotic system.Specifically，the entities，actions，and qualities，among others，represented in a picture express information units that make up the whole meaning of the picture.Just as words like nouns，verbs and adjectives represent units of meaning which make up the whole meaning of a verbal text when structured together as participants，processes and attributes，the different elements in a picture also carry units of meaning which combine into the whole meaning of the picture.

Although the term“sememe”is controversial in linguistics for its fuzzy boundaries of identification，it will not be a big issue in the analysis of the English learning advertisements.The reason is that these advertisements are mostly in the Chinese language which has little morphological changes and the boundaries of words are relatively clear.If we regard word（including compound word）as the information unit in verbal texts in the materials at issue，we will have a small number of cases of indeterminacy.As for the visual images in these materials，the situation is relatively easy to handle，too.Considering that the visual images in them include，most of all，pictures of people，and also pictures of discrete objects，the majority of which have clear forms and limited range of meaning，it is easy to identify the information units if we regard the discrete participants and objects in these pictures as information units.Therefore，the notion“information unit”helps break the whole chunk of picture meaning into analyzable parts or elements，and through which，an inter-semiotic analysis between images and language is made possible.

Secondly，the“information”here in this proposal refers to the conceptual meaning represented in the verbal and pictorial texts.In Leech's（1974：23，cited in Hu and Jiang，2002：279）terminology，conceptual meaning means logical，cognitive or denotative content.That is，the associative meanings such as the connotative meaning and affective meaning are not considered.The limit of conceptual or denotative meaning makes the principle of information linking possible in operation.

Thirdly，according to the degree or intensity of information linking between language and image，the visual-verbal combination can be measured along a cline ranging between two ends.On the one end is isomorphism which refers to complete overlapping or correspondence in meaning between verbal and visual.On the other end is juxtaposition which refers to the absence of semantic connection between verbal and visual.

The principle here emphasizes the degree or intensity of information linking between language and image，but not how to label the link in semantic terms.This strategy has two main strengths：

One is that it can circumvent the problem of labeling the nature or type of semantic connection between language and image.As mentioned in Section 5.1.1，labeling semantic relations is becoming a highly challenging task in the modern information age when the advertising is becoming increasingly creative.With the principle of information linking，however，what the researcher needs to do is to analyze the information or conceptual meaning expressed in the verbal and visual modes and compare whether they have any information unit in common.Although there is still some degree of subjectivity，the task is much more manageable.Attempts will be made to moderate the subjectivity，such as inter-rater reliability.

The other strength is that this principle makes possible the measurement of the semantic connection between language and image，and through which quantitative analysis of local coherence with reference to empirical data concerning the processing and comprehension by readers can be conducted.For example，in Chapter 7 the ratings of local coherence will be analyzed in the light of eye movement data and results of memory recognition test.Therefore，this principle of information linking provides a way to check the effects of visual-verbal connection on the cognitive processes in the reading and comprehension of the readers.Since it is held in this book that discourse coherence is partly in the text and partly in the reader's mind，the empirical exploration into the relationship between patterns of local coherence and cognitive processes in reading can give us huge implications in understanding the construction of coherence in multimodal discourses.

The stages of the English learning advertisements differ greatly in the semiotic configurations such as the features of the visual elements.For example，in the stage of Reader Attraction，the verbal texts are short whereas the picture are large in size;in the stage of Statements of Competitiveness，especially in the phase“teaching method”，diagram is the most frequently used visual form;in the stage Course Information，the information is very often shown in the form of tables.This diversity in materials predicts that the patterns of information linking will also show great diversity across the stages of the genre at issue.Due to the complexity of situation，we will not propose an overall pattern or model of information linking between visual and verbal in the English learning advertisements in this section.Instead，we'll generalize several models of information linking for the various stages of the English learning advertisements based on textual analysis in Section 5.2 and Section 5.3.