You will have to choose a compound with similar physical/chemical characteristics as methanol (boiling point, polarity, etc.). A higher molecular weight alcohol such as ethanol, n-propanol or iso-propanol could work provided that methanol is not present in the higher alcohol, the higher alcohol is not present in your samples, and you have sufficient baseline chromatographic separation between the methanol, the internal standard, and other compounds in the matrix.
I think you better build a calibration curve using methanol at known concentrations and then, analyze your sample to know the exact concentration of methanol in your sample. if you use this type of analysis avoid any other complication such as separation, etc.
I agree with Alberto and Yanju. Unless there is a particular reason you have to use an internal standard I would suggest just going with external calibration using methanol standards at concentrations that bracket the response you're seeing in your samples. You can also spike known amounts of methanol into your samples to check percent recovery.
Exactly right. Use an external standard calibration to get a working calibration for your system, and then use the internal standard to account for run-to-run variability.