PTL – Philosophical Tag Language

The Philosophical Tag Language (PTL) is a fictitious and freely invented markup language developed for the structured analysis of philosophical texts. It uses tags whose names consist exclusively of CAPITAL LETTERS (e.g. <ZIEL>, <METHODE>, <REGEL1>, <PERSPEKTIVE>, etc.), in order to mark various aspects of textual analysis. The PTL grammar is organised into main categories – epistemic aim, topic, method, rules, perspective and meta-reflection – and is intended to offer linguistically clear, methodologically robust and didactically productive support for a targeted, AI-assisted analysis of texts in philosophy.

In what follows, all PTL tags are explained according to categories, each with its purpose, an application example within the analysis workflow, and a concrete reading instruction for working with philosophical texts. The most important point is this: one may always have the AI explain everything that follows and let it work out alternatives. The language is "fictional" in the sense that it is presented here as an analytical structure. The PTL language serves solely the didactic perfection of prompting. Students should generate templates for recurring occasions.

A template begins, in rudimentary form, as follows:

Here is an initial AI prompt in this sense:

The Development and Optimisation of the PTL Structure

The Philosophical Tag Language (PTL) simulates a markup language for philosophical thought and analysis of texts in the humanities. The structure presented above has two levels: that of contents focused on "knowledge" ('Who is the murderer?' 'Summarise this!' 'What does the author say?') and that of a structure focused on "competencies" (<TAG>). One ought to keep these two levels separate. In its first version, the structure was oriented towards a rule-governed, sequential mode of working: the analytical instruction consists of a series of consecutively numbered <REGEL> elements (e.g. <REGEL1>, <REGEL2> and so forth), supplemented by additional checking tags such as <PRUEFEKOHAERENZ /> or <PRUEFETEXTBEZUG />. These were intended to prescribe concrete working steps for the AI, such as the definition of central concepts, the reconstruction of theses, or the reflection upon social relevance.

The concepts of "competency" and "knowledge" can, within academic and educational-policy discussions, quickly generate an ideologically charged atmosphere — and this often leads to very little. In working with AI, the matter at hand is simply to prompt competently. This competency one must acquire for oneself: it means extricating oneself, at the outset of the learning process, from one's own morass. The AI will not "pull you out of the morass" — no more than teachers can. Know-how and know-that stand in a complementary relation to one another. Yet one always begins with one's own incompetence. The productive path is to ask the AI specifically how one might methodically approach a task. The AI will explain the methodology. This methodology is then to be applied, its results commented upon, and improvements specifically sought. By developing better prompts iteratively, and by having explained to oneself why they are methodically superior, one acquires — beyond all ideological smokescreens — a disciplinary competency of competency. To resist one's own incompetence in this manner is a learning process within which one acquires both know-that and know-how. And this is precisely what one's studies are about: not merely to accumulate knowledge or skills, but to acquire the ability to pose questions precisely and thereby to obtain methodically well-founded answers. A heap of knowledge is merely a heap. Competency of competency ferments. This is why an AI on academiccloud.de is called "Sauerkraut".

Although the first structure in PTL permits a clear and traceable working-through of individual analytical tasks, it proves problematic in application: the AI interpreted the instructions as a linear list of tasks, which frequently led to fragmented, list-like, and poorly integrated output. The analysis was not presented in the form of an argumentatively coherent continuous text, but rather in separate sections that followed the rules yet formed no coherent argumentative unity. The structure thus possessed a high degree of formalisation, but a low degree of textual coherence and stylistic naturalness.

An Improved Structure with TEXTERSTELLUNG, ANALYSEKRITERIEN, and METAPRUEFUNGEN

To address this shortcoming, a revised PTL structure is introduced. These pages ought to be read as a didactic tool by means of which one can acquire the competency to distinguish between "knowledge" and "competency". What is at stake, then, is a "competency of competency". This new version replaces the sequential rule structure with three central components:

1. <TEXTERSTELLUNG> This element explicitly defines the text type, style, and argumentative character of the expected answer. It formulates an overarching meta-instruction that ensures the answer is generated as a coherent, scientifically formulated continuous text. The AI is thereby guided not merely towards working through individual tasks, but towards the compositional integration of all analytical aspects into a coherent text.
2. <ANALYSEKRITERIEN> In place of individual commands, structured analytical stipulations appear here in the form of substantively defining <KRITERIUM> tags. These designate the substantive focal points of the analysis (e.g. clarification of concepts, examination of coherence, social relevance), without being understood as sequential tasks. They function as substantive guiding rails, which are to be embedded within the argumentation.
3. <METAPRUEFUNGEN> These optional self-closing tags (e.g. <PRUEFEKOHAERENZ />, <FINDEARGUMENTALTERNATIVE />) declare metacognitive operations that are to be carried out in the background of the text's composition. They do not appear as explicit textual parts, but rather as silent processes of evaluation within the AI-generated analysis. Thus, the text remains stylistically fluent, without the checking tasks having to be visibly enumerated.

From the critique of the preceding <TAG> structure, the following improvement may be developed:

Advantages of the new structure. The new PTL configuration offers a number of clear advantages over the original version:

• Increased coherence: The TEXTERSTELLUNG component ensures a stylistically and logically coherent presentation.
• Flexible extensibility: New criteria or verification tasks can be added modularly without fragmenting the text structure.
• Separation of content and control logic: The analytical objectives are linguistically integrated, while the meta-verifications operate in the background.
• Didactic connectivity: The new structure allows analytical processes to be modelled as implicitly reflexive operations, which is particularly advantageous for teaching.

Overall, the second PTL variant leads to a marked improvement in the functional adequacy, stylistic quality, and analytical depth and precision of the AI-generated texts.

PTL (Philosophical Tag Language) is a specialised markup language developed to systematically structure philosophical analyses and textual interpretations. It employs specific tags to precisely define the substantive focus, the methodological approach, and the reflective dimensions of an analysis.

The most important elements of PTL are:

1. Objective tags, Theme tags, Perspective tags: These indicate the substantive focus of an analysis and determine which thematic emphases, objectives, and viewpoints are to be taken into account when interpreting a text.
2. Method tags: These specify the methodological approach to be applied in the analysis, e.g. hermeneutic interpretation, conceptual analysis, or discourse analysis.
3. Rule tags: These structure the analytical process by prescribing concrete working steps, such as the definition of central concepts or the examination of argumentative coherence.
4. Analytical criteria tags: These provide precise indications of which aspects of the text are to be elaborated and critically reflected upon, e.g. theses, concepts, or normative implications.
5. Meta tags: These tags support reflection on the effectiveness of the prompt. They serve quality assurance, methodological clarification, and the systematic improvement of the analysis.
6. Style tags: These govern the desired tone, linguistic precision, and argumentative rigour of the response. They help ensure a scholarly style and adapt the analysis to an academic standard.
7. Text-composition tags: These tags instruct the AI on how the finished text is to be formulated, e.g. in the form of a coherent continuous text, with complete sentences and logical transitions.
8. Target-audience tags: In a school context, one should tailor the AI's response to age groups or competence levels. At university, one should be familiar with the preferences of one's instructors.

Unlike HTML, there is no consortium or standardising body for PTL that authoritatively establishes and further develops the tag system. An AI therefore does not automatically expect a particular structure such as that envisaged by the PTL template system. Instead, the PTL tags developed here serve as orientational tools – they are not a rigid set of rules, but rather offer points of reference that can be individually adapted and expanded.

There are no objective constraints regarding syntax and semantics: you must develop for yourself the competence to ask the right questions and create meaningful structures. Everything else – that is, mere knowledge of tags or terms – remains secondary. Scholarship always means: generating knowledge, not merely reproducing it.

If you are uncertain how to approach a problem, first formulate your problem as vividly as possible. Then ask the AI to develop questioning strategies, to justify them, and to explain how the proposed questions can help solve your problem. In this way, your initial uncertainty becomes the starting point for a systematic development of competence.

The only important thing is this: the tags you use must be substantively "comprehensible" to the AI. Whereas HTML relies on clearly defined, standardised syntactic and semantic principles – which is why browsers know precisely how to render HTML code – PTL remains an open language. Its effectiveness depends entirely on your ability to think creatively, precisely, and in a problem-oriented manner – and this we must convey to the AI in the prompt in a comprehensible way.