Considering the problems of conducting empirical research into language teaching, how might linguistics be of benefit in this specific application of empirical research?
The two parts of this (empirical research into langauge teaching and linguistics) are closely connected to each other and are a little confused together.
When analyzed through an empirical method, language studies follow the qualitative research design. This design consists on the compilation of data that answers questions about a specific phenomenon. While the answers that researchers get are worthy of further investigation, they do not necessarily establish a correlation between two variables. This is done through quantitative research.
Through the application of a quantitative research design a strategy, or an intervention, is put to the test in a systematic way where all variables are considered and parameters have been clearly established. For this reason, empirical research can only provide information based on theoretical linguistics. What this means is that empirical studies are often founded on philosophical and theoretical tenets that apply to language studies.
When we apply a formula to language studies we actually move away from theoretical and onto applied linguistics. With applied linguistics you can choose from a range of best practices, strategies, and methodologies that can be applied to the study of language and thus a correlation of events can be created. These correlations tell more about language usage and production than mere theoretical empiricism. This is where the problem actually starts.
Do not get confused though: The term "Empiricism" under the scope of language studies and the term "linguistics" are not the same thing. Empiricism is a theory itself which bases its tenets on the analysis of information using sensorial resources such as observation and personal experience. The term "Linguistics" corresponds to a field of study that specializes in the scientific study of language, from its acquisition to its production. This last fact makes it even more evident that the need to apply a quantitative research design produces data that can be triangulated, preventing as much as possible the influence of bias or partiality which could hurt the validity of the findings. This latter fact is what makes empiricism viable and acceptable but by no means a solitary mean of language data analysis.
Linguistic analysis can be used in empirical studies of language teaching to describe and analyze. For example, if a study is conducted on second language learners in a kindergarten classroom, linguistic terms can be used to describe progress the children are making in acquiring their second language. One area would be applied linguistics
Applied Linguistics is concerned with how second languages are learned and taught. (see Georgetown.edu link)
For example, the researcher may analyze phonemes (sound units) that the kindergarteners are learning in English and how they compare to their native language. Researchers might theorize that languages with phoneme meanings similar to English will be learned more easily. They could then test the students to see if this is the case.