Clay material characterization is of importance for many geo-engineering and environmental applications, and geo-electrical methods are often used to detect them in the subsurface. Spectral induced polarization (SIP) is a geo-electric method that non-intrusively measures the frequency-dependent complex electrical conductivity of a material, in the mHz to the kHz range. We present a new SIP dataset of four different types of clay (a red montmorillonite sample, a green montmorillonite sample, a kaolinite sample, and an illite sample) at five different salinities (initially de-ionized water, 10-3, 10-2, 10-1, and 1 mol/L of NaCl). We propose a new laboratory protocol that allows the repeatable characterization of clay samples. The complex conductivity spectra are interpreted with the widely used phenomenological double-Pelton model. We observe an increase of the real part of the conductivity with salinity for all types of clay, while the imaginary part presents a non monotonous behavior. The decrease of imaginary conductivity (sensitive to polarization mechanisms) over real conductivity with salinity is interpreted as evidence that conduction due to electromigration processes increases with salinity faster than conduction due to polarization processes. We test the empirical petrophysical relationship between σ”surf and σ’surf and validate this approach based on our experimental data and two other datasets from the literature. With this dataset we can better understand the frequency-dependent electrical response of different types of clay. This unique dataset of complex conductivity spectra for different types of clay samples is a step forward toward better characterization of clay formations in situ.