We follow up emission line galaxies identified through the near-infrared slitless HST/WFC3 WISP survey with VLT/FORS2 optical spectroscopy. Over 4 WISP fields, we targeted 85 of 138 line emission objects at $0.4\lt z\lt 2$ identified in WFC3 spectroscopy. Half the galaxies are fainter than $H_{AB}=24$ mag, and would not have been included in many well-known surveys based on broad-band magnitude selection. We confirm 95 per cent of the initial WFC3 grism redshifts in the 38 cases where we detect lines in FORS2 spectroscopy. However, for targets which exhibited a single emission line in WFC3, up to 65 per cent at $z\lt 1.28$ did not have expected emission lines detected in FORS2 and hence may be spurious (although this false-detection rate improves to 33 per cent using the latest public WISP emission line catalogue). From the Balmer decrement, the extinction of the WISP galaxies is consistent with $A($H $\alpha)=1$ mag. From SED fits to multiband photometry including Spitzer$3.6\, \mu$m, we find a median stellar mass of $\log _{10}(M_\star /{\rm M}_{\odot })=8.94$. Our emission-line-selected galaxies tend to lie above the star-forming main sequence (i.e. higher specific star formation rates). Using [O iii], [O ii], and H β lines to derive gas-phase metallicities, we find typically sub-solar metallicities, decreasing with redshift. Our WISP galaxies lie below the $z=0$ mass–metallicity relation, and galaxies with higher star formation rates tend to have lower metallicity. Finally, we find a strong increase with redshift of the H α rest-frame equivalent width in this emission-line selected sample, with higher $EW_0$ galaxies having larger [O iii]/H β and O32 ratios on average, suggesting lower metallicity or higher ionization parameter in these extreme emission line galaxies.