Impressions of nine semantic words expressing abstract meanings were measured by a color selection process using twelve hues from vivid tone in the Practical Color Coordinate System (PCCS) as well as White, Grey and Black (as a normal color stimulus set) in a paired comparison method (Experiment 1). Color impressions were rated using 35 paired words in a semantic differential (SD) method (Experiment 2). The data of ten color vision normal (CVN) and four deuteranopic observers were analyzed separately by principal component analysis (PCA). Our previous study [J. Opt. Soc. Am. A37, A181 (2020)JOAOD60740-323210.1364/JOSAA.382518] using these methods indicated that deuteranopes could understand the impression of all colors if the color names can be recognized, even if deuteranopes cannot perceive redness and greenness. In this study, we also used a simulated deutan color stimulus set in which colors were modified to simulate the color appearance in deuteranopes by using Brettel-Viénot-Mollon's model to investigate how these simulated deutan colors would be treated by the deutan observers. In the CVN and deutan observers, color distributions of the principal component (PC) loading values in Experiment 1 were close to the PCCS hue circle in the normal colors, and those of the simulated deutan colors could be fitted by ellipses, but there were wide gaps of 73.7° (CVNs) and 89.5° (deutan) where only "White" existed. The word distributions as the PC score values could also be fitted by ellipses and are moderately similar between stimulus sets, but the fitting ellipses were considerably compressed in the minor axis directions in the deutan observers, although categories of words were similar between observer groups. The word distributions in Experiment 2 were not statistically different between the observer groups and the stimulus sets. The color distribution of the PC score values were statistically different, but tendencies of the color distributions were similar between observers. The color distributions of the normal colors could be fitted by ellipses, similar to the hue circle, and those of the simulated deutan colors could be fitted by cubic function curves. These results suggest that both of the stimulus sets were perceived as one dimensional and as a monotonic series of colors by a deuteranope, but the deuteranope can recognize the difference of the stimulus sets and recall the color distributions of each set, similar to those in the CVN observers.