Abstract

Stimuli

Both traditional screens and modern screens have two types: zuoping, or the standing screen with an undetachable panel connected with the pedestal, and chaping, or the dismountable screen with detachable panel. With the representative screens of the two types as stimuli, which focus on the screen frame and component linear type, the extraction of linear elements, image vocabulary selection, and variable analysis are carried out. The Kansei image is based on the theory and method of Kansei engineering and reflects the image expression of the user to the screen.

First, 54 screen stimuli were selected from domestic and foreign museum collections or private collections. By using the Kawakita Jiro (KJ) method (Kawakita 1970) to remove those stimuli of similar height, same type, and vague information, 26 stimuli were finally selected for research and were classified according to the types of standing screen (including multileaf standing screen, single-leaf dismountable screen, small tabletop screen/inkstone screen) and folding screen, of which there are nine standing screens, six dismountable screens, five small tabletop screens (including inkstone screen), and six folding screens (Fig. 1).

View Figure

• Download as Powerpoint
• Download Figure

Scholars selected suitable Kansei words according to the characteristics and styles of furniture and completed the evaluation research on the basis of modern design aesthetics. Zhang and Xu (2020) made a Kansei evaluation of the morphologic characteristics of important parts such as the backrest, armrest, and legs of chairs in the Tang Dynasty. Wan et al. (2021) selected five pairs of perceptual words to analyze the surface characteristics of composite bamboo furniture and people's subjective preferences and used the quantified data in modern design. Song et al. (2016) studied the visual effects of different wooden decoration and furniture arrangements. Jing et al. (2015) studied the evaluation system of modernization design of traditional furniture. This experiment mainly involves the linear outline of Chinese screens. The evaluation results and quantitative data will provide a reference and basis for modern design. To avoid the influence of color, pattern, and process of the screen on its evaluation results, the inlaid or carved patterns on the screen were removed and only the outline of the main component was retained. The linear outlines of the 26 screen stimuli were drawn and are shown in Figure 1.

Participants

The screens are Chinese traditional furniture that are widely used in people's daily life, so the subjects of the experiment include both professionals of design and art and nonprofessionals. Among the 75 persons that participated in the questionnaire, there are 21 males and 54 females, including 41 professionals with education background of design and art or experience in a design-related field, and 34 nonprofessionals.

Experimental process

Three groups of representative Kansei image words were selected according to the shape, decoration, and weight of the screen: (1) traditional style–modern style, (2) luxury style–plain style, and (c) heavy style–light style. Each pair of the Kansei image words has a strong preference tendency. In the experiment we made a questionnaire of three groups of Kansei words. By using the semantic differential analysis method with a five-point scale (Osgood et al. 1957), 26 screen stimuli were evaluated and scored in the questionnaire.

The questionnaires were distributed online. The subjects who received the questionnaires looked through the outlines of the 26 stimuli and gave scores to the three groups of Kansei words. Scoring rules: −2 points represents very traditional, very luxurious, and very heavy; −1 point represents traditional, luxurious, and heavy; 0 points represents neutral; 1 point represents modern, plain, and light; 2 points represents very modern, very plain, and very light (see Fig. 2).

View Figure

• Download as Powerpoint
• Download Figure

Calculation and analysis of Kansei words scoring results

After calculating the scoring results, the data in Table 1 were obtained. Under the traditional–modern type, stimulus 2 scores −1.28, representing the most traditional stimulus, and stimulus 22 is the most modern one, scoring 0.99; under the luxury–plain type, stimulus 6 scores −0.96, representing the most luxurious stimulus, and stimulus 1 scores 1.16, representing the plainest stimulus; under the heavy–light type, stimulus 2 scores −0.99, representing the heaviest stimulus, and stimulus 22 scores 1.07, representing the lightest stimulus.

Table 1. Experimental sample image vocabulary scoring results.

View Fullscreen

All statistical analyses were performed using SPSS 22.0. In addition, the analyzed data of this study have passed the reliability test with Cronbach's alpha = 0.946, Kaiser–Meyer–Olkin test = 0.705 > 0.6, and the data have good structural validity. The above findings are the basis for further analysis.

Establishment of quantitative type I and mathematical model

Quantification theory is used for the prediction of quantitative benchmark variables. Its advantage is that qualitative variables can be quantitatively studied (Jing 2015). The linear elements in this experiment are qualitative variables, which are usually called items in quantitative theory, and the values contained in the items are called types.

Quantification theory type I is used to establish the relationship between a set of qualitative variables and a set of quantitative variables by using the mathematical models established on the basis of multiple regression analysis. In this experiment, the linear element was set as x (qualitative variable, item), and the evaluation value of the stimuli Kansei was set as y (quantitative variable, type). The partial correlation coefficient represents the contribution of each item to the Kansei evaluation value. The solution method is as follows: Suppose that the matrix of the Kansei evaluation value y and item j is P:

Then, the partial correlation coefficient between the Kansei evaluation value y and the jth item is as follows:

The contribution of item j to Kansei evaluation value y is R_yj.

Suppose item x has r types and δ_i (j, k) is the response of type k of item j to the ith stimulus (i = 1, 2..., n; j = 1, 2..., m; k = 1, 2..., r_j), then

Suppose that there is a linear relationship between the Kansei evaluation value and the linear elements (items) and each type response; then, the linear model is as follows: where a_jk is the kth constant that depends only on the jth item, and ε_i represents the random error of the ith sampling. The least-squares estimate a_jk of the coefficient a_jk is obtained by using the least-squares method. Then, the prediction equation is as follows:

ŷ is the predicted value of the benchmark variable y, and δ(j, k) is the response of any stimulus in the type k of the item j. â_jk is the score of the type k of the item j. The model is then further assumed to be where y is the average value of the Kansei evaluation value and a_jk is called the standard coefficient, representing the contribution of each type to the Kansei evaluation value of the stimulus.

Extraction of the linear elements of the top section, straight bracket, and pedestal of the screen

In the experiment, four professors and two designers used the KJ method (Kawakita 1970) to extract and classify the linear elements of the experimental stimuli, and finally selected three linear elements that are of greatest importance for the 26 screen stimuli. These three linear elements are (1) pingmao, or pingtou: the top section of the screen connected by tenon at the top of the screen with both functions of stabilization and decoration; (2) zhanya: the straight brackets that are commonly found on both sides of the pedestal of the standing screen, which is used to reinforce the panel and connect it with the pedestal; (3) dunzuo: the pedestal of the screen that is used to maintain the stability of the screen. It should be noted that the pedestal of the standing screen is called xumi zuo (Sumitomo base), and that of the standing screen with removable panel is dunzi (the foundation), which is also the linear element extracted in this experiment. The three linear elements are shown in stimulus 4 and stimulus 19 in Table 2.

Table 2. Extraction of linear elements of stimuli.

View Fullscreen

Classification of linear elements of top section, straight bracket, and pedestal

Each table must be typed on a separate sheet. In this experiment, the extracted linear elements are classified respectively, that is, similar elements are classified into one type, so 26 top sections (pingmao or pingtou) have 8 types; 14 straight bracket, 6 types; and 26 pedestals, 8 types. Suppose the linear style is X, then that of the top section is X₁, and its types are expressed as X₁₁, X₁₂, …X₁₈; the linear style of the straight bracket is X₂, and its types are X₂₁, X₂₂, …X₂₆; the linear style of the pedestal is X₃, and its types are X₃₁, X₃₂, …X₃₈. (See Table 3 for details.)

Table 3. Classification of linear elements.

View Fullscreen

Data analysis

According to the above analysis of the stimuli and types by quantifying class I, the final results of this experiment are as follows: the standard coefficients in Table 4 reflect the contributions of the top section (eight types), straight bracket (six types), and pedestal (eight types) to the image evaluation values of traditional–modern style, luxury–plain style, and heavy–light style. The closer the standard coefficient is to 2, the greater the contribution of the type to the evaluation value of modern style, plain style, and light style. Conversely, the type closer to −2 contributes more to the traditional style, luxury style, and heavy style. Taking the top section as an example, the standard coefficient of the X₁₆ type is 1.14, which has the largest contribution to the modern style of the stimuli in the top section type. As for the straight bracket, the X₂₄ type has the largest contribution to the luxury image of the stimuli with the standard coefficient of −0.96.

Table 4. Experimental results of screen sample Kansei image.^a

View Fullscreen

The partial correlation coefficient reflects the degree of correlation between the linear element and the stimulus in terms of the images of traditional–modern. The closer the result is to 1, the greater the correlation between the two. From the data in Table 4, the linear type of the top section (pingmao or pingtou) has the greatest correlation with the traditional type, luxury type, or heavy type of the screen, followed by the linear type of the straight bracket, and that of the pedestal has the smallest correlation. The complex correlation coefficient R represents the accuracy of the model. The larger the value, the higher the degree of linear correlation between qualitative variables and quantitative variables. R = 0.875 in this experiment indicates a higher accuracy.