The first sentence on page 348 is incomplete. You have to start with the last line on page. 347
Colors have been the favorite stimulus material because their phys-ical nature can described relatively easily, standard stimulus material is readily available, the relative frequency of occurrence in the environment is not too likely to affect subjects` reactions in an experiment, and perceptual qualities may also be controlled relatively easily. (2) Acuity of Discrimination
There is a collection of colors, known as the Farnsworth-Munsell Test(Farnsworth, 1949), consisting of 86 chips of equal brightness and saturation but differing in hue. These colors were chosen by the authors of the test so as to constitute an instrument for the testing of hue discrimination. Subjects are presented with two chips, say one green and one blue, and are asked to put a disarranged collection of shades between these two colors into the proper order so that an array results with all chips finely graded from green to blue. Similar tasks are required with shades between blue and purple, purple and red, and red and green. When the test is administered to a noncolor-blind standard population of young American adults, a certain average number of sorting mistakes occur, but the mistakes occur with equal likelihood anywhere in the spectrum (and the mistakes for any one color chip have a frequency distribution that is the same for all colors.)
The question now arises whether the construction and standardization of the test might have been biased by the language habits of its authors who were English speaking individuals. Since the color vocabulary superimposes a classification upon the physical color continuum might it not have predisposed us to pay more attention to the borderline cases, thus sharpening our acuity across wordclass boundary, and to pay relatively less attention to the clear-cut cases, thus dulling our acuity within the word classes?
Lenneberg and J. Bastian (unpublished data) administered the Farnsworth-Munsell Hue Discrimination Test to a group of Zuni and Navaho Indians whose color vocabulary had been mapped into the stimulus continuum by Lennebery and Roberts(1956) and Landar et al.(1960) respectively. The different locations of the Zuni and Navaho color-name boundaries did not predict in either case systematic difference in discriminatory acuity between the Indians and a control group of Anglo-American farmers living in the same region. Thus there is little ground to assume that the peculiarities of English color words have affected the construction of the Farnsworth-Munsell Test in any basic way.
There is, however, some other type of evidence that may cause us to reconsider our conclusion. Beare(1963) elicited English color words for a series of monochromatic lights. Her stimulus material had the advantage of being perfectly controlled in its physical properties and each color could be specified in term of wavelength. She required subjects to name every stimulus presented as red, orange, yellow, green, blue, violet and to make the response as fast as possible. After an appropriate decision had been made the subject was allowed to quality the name given by further adjectives. If a stimulus fell within a name map the response was made relatively quickly and few or no qualifiers were used. A stimulus that fell between maps had a longer response latency and elicited a greater number of qualifiers. Thus, she used Approach B to map words into the particular referent space represented by her stimuli.
The same stimulus material had been used earlier for another experiment (Judd, 1932) designed to discover acuity of hue-discrimination. In that experiment it was found that the differential limen was not constant for the visible spectrum. In some spectral areas our eye was shown to discover smaller hue differences than in other spectral areas. In Fig. 8.5 Judd`s hue discrimination curve is shown; low points indicate relatively high acuity. The other curve plotted in Fig. 8.5 is based on Beare`s naming data; it is the frequency of occurrence of name-qualifiers. This frequency is high when a stimulus is ambiguous with respect to its word-class assignment. The curve reflects what we have called name-indeterminacy. Figure 8.5 suggests that there is a relationship between the two variables. Within the large, clear-cut word classes acuity seems to be slightly less sharp than in the areas of relatively large name-indeterminacy. These findings are merely suggestive, but not conclusive.
Beare used a slightly different procedure to compare her naming data with Judd`s acuity data, but she was generally skeptical about a causal relationship between the two. In her own words;”...what fit there is between the two functions is tenuous, and it is possible that the configuration of the naming curve will change (if a different stimulus array is used), or with changes in instructions with regard to categories of judgment.” Even if Beare shall eventually be proved to have been too modest in her conclusion, we can still not be certain whether the name boundaries of the English words are entirely arbitrary; only if other languages can be shown to have markedly different boundaries for this particular stimulus array and if the speakers distribute their loci of greatest acuity in a predictable direction may we conclude that nonphysiological factors have affected hue-discrimination. For the time being, we must withhold judgment.
(3) Memory and Recognition
The first experiment on the effects of certain language habits on memory and recognition was carried out by Brown and Lenneberg(1954). Until that time investigators had attempted to manipulate language variables by quite ephemeral conditions such as teaching subjects nonsense names for nonsense objects or influencing their verbal habits by instructions given immediately prior to the cognitive task (Carmichael, Hogan and Walter, 1932; Kurtz and Hovland, 1953; and others cited in carroll, 1964), but the semantic structure of the subjects` native language was neither known to the experimenter nor utilized as a variable in the experiment. Yet the habits induced by one`s native language have acted upon an individual all his life and are, therefore, much more relevant to the basic problems that concerns us here than the verbal habits that can be established during a single experimental session. Thus, experiments that build upon the properties of a natural language are in a sense more crucial than the other type.