Soundforge can do it by itself, just select "Average RMS level (loudness)" rather than "Peak level" in the normalize window.
Check "Use equal loudness contour", click on "Scan Levels" to get the RMS level of your unprocessed recording, then accordingly select the target RMS level for the normalization . Make sure that you have "Apply dynamic compression" selected in front of "If clipping occurs", and then hit OK.
For instance if your recording has an RMS level of -16 dB and you select -10 dB, the average loudness of your recording will be increased by 6dB, and every clipping peak will go through a limiter, lowering them just under 0 dBfs without making them sounds ugly.
You should also know that RMS level is not an exact mathematical transcription of how our ears perceive loudness, so it might happen that two records normalized to the same RMS level have a slight perceived loudness difference, depending on the audio material. Other ways of measuring loudness and producing results closer to our perception exists, such as Leq (RLB) and LKFS (R2LB), but you should be mostly ok with RMS.
If you feel that the way Soundforge peak limiter process audio is not transparent or good enough, you should use other limiting plug-ins, such as the Wave L-serie Chuck told you about.