Picture languages are on attractive basis for communication. However, it is difficult to represent and understand accurate meaning in such languages. So, we are developing an approach where pictures with captions are allocated on positions of a 2D structure and special translators generate corresponding sentences in natural languages. These positions have pre-defined meaning and represent a multimedia hieroglyph. The hieroglyph is a complex picture (a multimedia sentence) including more simple pictures (multimedia words). In this approach the "hieroglyphic writing" is reduced to selecting appropriate multimedia words through special interface panels without necessity to use keyboards. As a result, the users can send and receive multimedia messages, though they have never used a computer before. In this paper, methods and implementation techniques for translating multimedia hieroglyphs into English are explained.