The<!DOCTYPE> is not a HTML tag. It is an instruction to web browser about what version of HTML the page is written in.
In HTML 4.01, the<!DOCTYPE> declaration refers to a DTD, because HTML 4.01 based on SGML (Standard Generalized Markup Language). So that the browsers render the content correctly.
It is a top level tag known as a Public Text Identifier
HTML5 is not based on SGML so its not require any reference for DTD.
Explanation of Doctype ?
HTML DOCTYPE statement:-
A pair of forward slash characters ("//") is used as delimiters between keyword fields in the FPI (Formal Public Identifier) declaration.
(A : Top Element):- Its Indicates the top level element type declared in the DTD e.g:- For HTML it is <html>. For XHTMl it is <XHTML >
(B : Availability):- field indicates whether the identifier is a publicly accessible object (PUBLIC) or a system resource (SYSTEM) such as a local file or URL. HTML/XHTML DTDs are specified by PUBLIC identifiers.
(C : Registration) - Indicated by either a plus ("+") or minus ("-"). A plus symbol indicates that the organization name that follows is ISO-registered. A minus sign indicates the organization name is not registered. The IETF and W3C are not registered ISO organizations and thus use a "-".
(D : Organization] - This is the "OwnerID" - a unique label indicating the name of the entity or organization responsible for the creation and/or maintenance of the artifact (DTD, etc.) being referenced by the DOCTYPE. The IETF and W3C are the two originating organizations of the official HTML/XHTML DTDs.
(E : Type) - This is the "Public Text Class" - the type of object being referenced. There are many different keywords possible here, but in the case of an HTML/XHTML DTD, it is "DTD" - a Document Type Definition.
(F : Label) - This is the "Public Text Description" - a unique descriptive name for the public text (DTD) being referenced. If the public text changes for any reason, a new Public Text Description string should be created for it.
(G : Language) - This is the "Public Text Language"; the natural language encoding system used in the creation of the referenced object. It is written as an ISO 639 language code (uppercase, two letters.) HTML/XHTML DTDs are usually (always?) written in English ("EN".)
(H : URL) - This is the optional explicit URL to the DTD being referenced.