Once the acoustic analysis is complete, so that reliable between-subject comparisons are made, the data has to be normalised.

Labov (2001) outlines the rationale behind normalising for Sociolinguistics here:
Labov, W. (2001) Principles of Linguistic Change: Social Factors, London: Blackwell – Chapter 5, 152-164

There’s a range of different methods of normalisation and there are several good articles comparing their use:
Clopper, C. (2009) Computational methods for normalising acoustic vowel data for talker differences Language and Linguistic Compass 3(6), 1430-1442.
This article is based on US English.

Flynn, N. (2011) Comparing vowel formant normalisation procedures York Papers in Linguistics (Series 2)  11. pp. 1-28
This article provides a summary of previous comparisons as well as a comparing 20 different normalisation techniques, it uses British English speech as the data.

NORM: The vowel plotting and normalisation suite
This site is an online interface for normalising your vowel data. It also provides a guide to choosing the right method and how to format your data for the site as well as an extensive reading list and presentation/document-ready plotting tools.