There is always a history to words and utterances.
After some weeks or months of one word at a time, they graduate to longer utterances as they express more complex meanings.
They are theories about products, not theories about the processes used in producing and understanding utterances.
All one can observe is the actual use of utterances; and from such observations one may infer some underlying structure.
Constraints on bilingual children's code-mixing who produced mixed utterances across more than one time interval produced violations at each time interval.
The task may be regarded as more complex when the utterances contain more words.
Here the judgment was made regardless of the way in which the materials were handled or treated, or any utterances that the child made.
In our analysis, it is not the relationship between two utterances that is constrained, but the relationship between two pragmatic sources.