Domains usually need to fold a certain way in order to be functional, and so if they are connected by residues that should be buried then the structure could get pulled apart. A linker between them could help keep each individual domain folded properly despite being connected to each other.

Linkers (usually containing combinations of alanine, serine or glycine) are used typically for fluorescent protein fusions, as Beverley pointed out correctly they help giving both proteins better chances to fold independently. If the N-terminus or C-terminus of a specific protein under investigation is buried inside the protein (in the hydrophobic environment) even linkers may be insufficient to render the resulting protein fusion functional. Linkers are also notoriously at risk of being cleaved by proteases, so when people make GFP fusions, they may observe the localisation of GFP after it has been cleaved from the fusion protein. Complementation studies in genetics, or fluorescence microscopy to verify localisation, should always be accompanied by western blots. If there is only one band on the blot that corresponds to the size of the fusion protein, so far so good, but if one also sees a second lower molecular weight free GFP, then one should always ask if the complementation is caused by the cleavage product of the original protein without the GFP, and if the subcellular localisation reflects on free GFP rather than the fusion protein. Does that sound super critical? Well, it is perfectly reasonable and recommended to ask these questions. The problem is that most plant molecular biologists don't do westerns, instead they do RT-PCRs which really do not report on what matters most, the gene product, its stability, its localisation, and its mode of action. People need to ask more questions about proteins, thanks for raising an interesting question within the Research gate forum. :)  

Thank you Dr Beverley Marie Dancy and Prof Jürgen Denecke for detailed explanation.