The grammar of audio-visual language is mainly divided into "sequence" and "juxtaposition". The concepts of these two categories mainly come from Saussure's linguistic theory (editing is the grammar of this language). Saussure believes that there are two "axes" in the formation of language, one is the horizontal axis and the other is the vertical axis. He called the horizontal axis "combination" and the vertical axis "selection". Later, through the development of Sioux people in Jakobson, this theory was extended to other fields besides linguistics, and "choice" was also changed. Generally speaking, the combination axis means that different positions in the language seem to form a sentence in the horizontal direction, while the aggregation axis means that each word must identify itself according to a group of words with related meanings. According to Metz, the combination in film language refers to the orderly connection between shots: "Several local viewpoints in the same space are placed end to end;" This end-to-end connection is positional or inference = combination, because it actually constitutes this fragment. "Aggregation is the parallel connection between shots:" Aggregation can include a single semantic "unit" distributed at various distances (for example, in "temperature/heat/boiling" or as a general phenomenon of counterpoint synonyms), and it can also include those terms that are combined because of opposition ("heat/cold" and all antonyms); In either case, language provides us with a limited field in which we must choose-this is the definition of aggregation. " From Metz's narration, we can find his fascination with linguistics, because the idea of applying the concepts of combination and aggregation to movies originally came from linguist Jakobson. This fascination not only makes Metz's theory not fit perfectly with the film itself, but also makes him see the similarities between the film image narrative and the general language, so we can absorb his strengths on the basis of Metz. Through the study of Metz, we find that all film languages can be encapsulated by two structural forms: combination and aggregation, so we have reason to think that this is the basic structural mode of the film language system.