Coulmas begins with the history of ideas about writing, from Aristotle down to the present. Modern linguistics has been afflicted by scriptism, defined as "the tendency of linguists to base their analyses on writing-induced concepts such as phoneme, word, literal meaning and sentence, while at the same time subscribing to the principle of the primacy of speech for linguistic inquiry".
The obvious options for writing are representation of meaning or sound, but Coulmas explains why neither a language-neutral "representation of ideas" nor a perfect transcription system is possible, and why writing systems are not even approximations to such things. He suggests three basic principles for approaching real writing systems. They are autonomous systems, with their own functional units and structures. They map onto other levels of linguistic structure ("phonetic, phonemic, morphophonemic and lexical") in complex ways, requiring interpretation at different levels. And they are historical entities, agents as well as results of language change.
With these preliminaries out of the way, Coulmas begins with "signs of words". It is not just the notion of an ideographic, "idea representing" writing system that is problematic — the notion of a logographic, "word writing" system has problems as well. The status of the word as "a clearly delimited unit of a given language is a result rather than a prerequisite of writing". This is illustrated by a close look at the Sumerian and Chinese writing systems.
Next Coulmas considers "signs of syllables", presenting theoretical notions of a syllable and analysis into onset, rhyme, and tone. An expert on Japanese himself, he treats Japanese kana at length and touches on the Akkadian, Cherokee, Cree, and Vai syllabaries. Syllabaries range from a few dozen signs to nearly a thousand, with most "defective or incomplete if we take the encoding of all distinct speech syllables of the language as a measure of completeness". That is, however, irrelevant to the practical balance between economy and ease of interpretation, which is dependent on the features of individual languages.
Turning to segments, Coulmas touches on the theory of phonemes and polyvalence between graphemes and phonemes. Not all alphabetic systems function in the same manner. Among the most prominent is the Latin alphabet, but this "has never been the neutral writing tool it is sometimes thought to be [and] which its modern offspring, the IPA [International Phonetic Alphabet], still strives to be" — it functions as "both model and image".
Next comes a look at differences between consonants and vowels. That is followed by a study of Semitic writing systems, "consonantal alphabets" which focus on consonants and often omit vowels, or include them only optionally. Case studies here are of Hebrew, Arabic, and early Greek.
The status of Indian scripts is debated, with labels like "neosyllabary", "semisyllabary", "syllabic alphabet", and "alphasyllabary" used to avoid a choice between "alphabet" and "syllabary". Coulmas surveys the theory and history of vowel incorporation in Brahmi-derived scripts, focusing on Devanagari, Tamil, Tibetan, and Thai.
Linguistic analysis interacts in complicated ways with linearity — the basic functional units of writing may not appear in the same linear order as the linguistic units they encode. Coulmas looks briefly at Mangyan and Ethiopic and in detail at Korean Han'gul.
"The creators of Han'gul wanted it to be a script that is easy to learn and easy to read. These requirements are met by keeping the number of basic graphs — for segments and subsegmental features — very low to meet the requirements of the learner and writer, while creating enough graphic complexity — in the syllable blocks — to meet the reader's requirements for contrast and discernibility."
And then there are mixed systems, "without a predominant principle for the interpretation of their basic units". In this category Coulmas puts ancient Egyptian, Akkadian, Japanese, and English.
"English writing makes use of different units. At least three levels must be distinguished: (1) individual letters and digraphs and trigraphs, which are regularly interpreted as phonemes, allophones often being ignored; (2) letter sequences interpreted as morphemes; (3) orthographic words that cannot be reduced to grapheme-phoneme correspondences or morphophonological alternations. In addition, there are strata of the written lexicon that are held distinct by virtue of different spelling conventions or the absence thereof. Anglo-Saxon words, Romance (and Greek-origin) words, and etymological cognates follow different rules."
All of this includes some history of individual languages and language families, but Coulmas now presents a more general overview. He explains that monogeneticism is dead — "writing was invented more than once: to the best of our knowledge, at least four times, in Mesopotamia, Egypt, China and Mesoamerica" — and why unidirectional evolutionism (most commonly an alphabetic triumphalism) is untenable. The reinterpretation of signs in the adaptation of scripts to new languages is a significant source of change. And there is no simple dependency between speech and writing in diachronic language change.
Coulmas includes a brief overview of work on the psycholinguistics of reading and writing. And he ends with a chapter on the sociolinguistics of writing, looking at its long association with social status, at diglossia (two coexisting varieties, often with a written/spoken contrast) and digraphia (the use of different scripts for the one language), and at spelling reforms and the role of the state in setting standards.