DSLs are proliferating, as system modelers and designers are finding them useful for their 
purposes, and as tool support starts to be readily available for them (editors, validators, 
code generators, etc.). The previous “one language fits all” approach (e.g. Java, UML) has 
given path to “one language for each purpose”, and this is where the need to make public, 
coordinate and combine languages (i.e., globalize them) has risen. 

“Globalizing modeling languages” was defined in the original GEMOC paper as “The use of 
multiple languages to support coordinated development of diverse systems aspects”. 
However, I only agree in part with such definition. First, I see it is more adequate for 
defining what Multi-Viewpoint Modeling is/should be about: “The combination of multiple 
languages to support coordinated specification, analysis and development of diverse 
systems aspects” Thus, in my view, “Globalizing a Modeling Language” means “Making a 
Modeling Language amenable for integration into a (standard) Multi-Viewpoint Modeling 
environment.”
 
In this context, it is important to note that (a) Globalized MLs need to be combinable and 
integrable (b) Interfaces at different levels should be defined , and (c) Standardization 
should play a key role here.

In our group we have been working in this area, in the context of the Reference Model for 
Open Distributed Processing (RM-ODP), an ISO & ITU-T international standard that provides 
a mature framework for the specification of large complex systems, using viewpoints. RM-
ODP defines five viewpoints and their associated Viewpoint Languages (VPL), as well as 
explicit correspondences between the VPLs.  This is an example of the coordination and 
integration of separate languages, focusing on disparate concerns, and using 
correspondents to relate them.In addition, we have been working on the combination of DSMLs, 
studying the problems, issues and challenges involved in this area. 

The three major challenges that we see in the globalization of modeling languages are the 
following. First, there is the need for defining mechanisms, process and tools for the 
Combination/Integration/Unification of languages (which needs establishing 
correspondences between them, at all levels: Abstract Syntax, Concrete Syntax and 
Semantics); and needs to deal with heterogeneous (and not always combinable) semantics. 
Second, correspondences between metamodels, and between models, needs to be specified 
in an efficient, correct, usable and maintainable manner, and using different approaches 
(depending on the use we want to make of them). The third challenge that we want to 
highlight is about being able to reason about the information expressed across the different 
models, so that properties of the overall system (including emergent properties) can be 
inferred, proved or denied. 

Tool support is essential in this context for achieving these goals, given the complexity of 
the domain and of the systems being specified. Without tools any proposed solution will be 
useless.