A Kansei Study on the Style Image of Fashion Design Yi-Ching Chang*, Ming-Chuen Chuang**, Shu-Hui Hung**, Shu-Ju Chang Shen*, Bing-shu Chu*

* Fashion Design Department Shih Chein University No.70 Da-zhi St.,Taipei, Taiwan300,R.O.C. [email protected], , [email protected], [email protected] ** Institute of Applied Arts National Chiao Tung University No.1001 Ta-Hsueh Rd.,Hsinchu City,Taiwan 300,R.O.C. [email protected], [email protected]

Abstract: Fashion design generally puts more emphasis on style than on function consideration. However, fashion designers depend largely on their instinct and experience in creating the style images of their works. There is no way to guarantee these images to be perceived by customers as the ones the designers want to express.

For the designers with less experience, there is no tool to assist them for creating demanded images

in their fashion designs.

Thus, this kansei engineering investigation was designed and conducted to find out

solutions for filling these gaps.

In this study, fashion experts were invited to analyze varied fashions

collected from current fashions magazines and documents worldwide to identify some distinct fashion style images and to define the design language for each of the style images.

The morphological analysis of all the

investigated fashions then was conducted by a focus group of fashion designers.

The differences among

these fashions were judged and were analyzed by cluster analysis and MDS analysis.

Each clusters in the

perceived space of fashions derived from the about analysis were tabbed with a identified style image by the focus group.

The membership of each fashion to its specific style image then was calculated and treated as

dependent variable, while the constitutional elements of the fashion derived from the above morphological analysis were treated as independent variables.

A conjoint analysis was conducted accordingly to find out

the relationship between the constitutional elements of fashion designs and their style image. This relationship can be served as a guide for fashion designers to design or check the demanding images in their designs.

On the other hand, the sensibility to fashion image is different from person to person. Designers

who are highly creative in general and with professional education and/or experience may be gifted with better ability to manipulate form for representing specific images than general consumers.

Therefore,

questionnaire survey was also conducted in this study to find out the difference in sensing fashion style image between designers and consumers. Keywords: fashion design, fashion style image, cluster analysis, MDS, Kansei Engineering

1. Introduction This research is mainly to explore the cognition of modeling of fashion design styles and impressions. We analyzed and organized our data by means of relevant literature collection, comments and analysis of design examples, and experience of experts. Furthermore, we established the classification of fashion design styles and impressions, and the database of fashion 1

components for styles and impressions based on interviews of experts and questionnaire surveys. Then through the methods of kansei engineering for styles and impressions, we developed the principles for designing each particular style of fashion design as the references for practical fashion design. The degree to which the fashion design is transmitted and perceived varies by individuals. For those who have the nature of rich imagination as well as experience or background of studying design, they have better reinterpretations for modeling of designs and icons. On the contrary, those who lack such strengths may have weaker reinterpretations. Hence, there may be differences in psychological cognitions of fashion of the same type and shaping of design style in different designers and wearers (users). Therefore, when we conduct a multi-phase study according to the basic types of clothing, we should try to have a thorough understanding of basic cognitions of designers and users for all kinds of specific styles/impressions of designers and users, and then minimize the difference in their cognitions. Then we will extract all sorts of specific designs styles and impressions with relatively common cognitions shared by designers and users, in which molding components and design semantics can serve as references for fashion design practice. Since the interval between the rise and disappearance of a fashion trend is roughly 3 to 5 years (Chie Koike, 1992), this study targeted female fashion in 1995-1999. The range of fashion design style is so wide that we cannot discuss each style. Thus we focused on examples of fashion design styles presented in the four world fashion centers: Paris, Milan, London and New York from 1995-1999.

2. Method This study first chose test examples of fashion pictures periodically published in journals after they were reported in the media. These pictures were sampled from the works of internationally celebrated fashion designers, and these works were presented every other season annually. These samples were assessed by perception similarity in styles of designers and non-designers. We implemented cluster analysis on the statistical results, and thereby found the representative samples in each style. Then senior fashion designing experts summarized the components of all fashion styles and impressions according to the analysis on all test samples. Finally, we summarized the designing principles for each specific style and impression by means of statistical analysis. These principles would be the basis on which the complete facilitating computer system for designing is established in the future. The structure and experimental algorithm were composed of four phases: 1. Extract pictures of fashion styles and impressions. 2. Assess the similarity in styles and impressions. 3. Analyze and summarize the components of impressions. 4. Analyze the results. These phases are delineated as follows. 2-1. Extract pictures of fashion styles and impressions 1. The reports about fashion designs in the world fashion centers: Paris, Milan, London and New

2

York, were collected from the journals published from 1995-1999. The reference sources of pictures are COLLEZIONI (Italy). After screening, the 100 most representative fashion models were chosen. 2. Screening principles of the pictures (see figure 1)： (1) The front side of the model. (2) The model is standing with stretched arms. (3) The dress is “whole-body and front-side” with clear contours (4) The dress has only a plain texture. This study is to study the “shape” of fashion, and therefore we will remove the clothes with calico, stripes, plaids, and special textiles, as these features influence the visual linear judgment. (5) The clothing had no accessories or ornaments, because these items would disturb the subject’s visual judgment on lines and contours. (6) The types of fashion were chosen evenly, and the types

figure.1.

The

picture

belonging to specific seasons, designers, brands and models

sample of fashion style

were eliminated.

sand impressions

(7) 500 pictures were chosen first, and then 100 of them were chosen after being considered more representative by experts. These 100 pictures were used in the final questionnaire survey. 1. 3. Number the 100 pictures and arrange them in a loose-leaf sample booklet. The digital image data were then stored. 4. The colors of backgrounds of pictures were removed, and thereby the pictures centered on the fashion model. 2-2. Assess the similarity of styles and impressions 1. In order to understand the psychological perception similar in designers and non-designers, and to judge the possible components of impressions, we conducted the perception similarity test for two kinds of subjects. 2. The subjects in this experiment had 33 users, and 37 designers. Each subject was asked to group the pictures according to styles and impressions based on his or her intuition. Grouping the pictures into 8~10 groups was preferred. Each cell was filled in with the numbers from each group. 3. Testing conditions: we used the normal sun lamp as the source of illumination. The visual distance was around 30~40 CM, and the testing time is 15~20 minutes. 2-3. Questionnaire data entry and analysis of results We organized the results of the experimental questionnaires, and entered the data into SPSS (version 8). We conducted the multi-dimension cognition spacing analysis on the assessment of perceived similarity for the subjects. Then we carried out each relevant analysis on correlations among components of fashion styles and impressions, and thereby constructed the space and application principles for fashion styles and impressions. 3

3. Result and Discussion 3-1. Multi-dimension assessment of cognition space of fashion styles and impressions In order to explore which important components can be used to distinguish all kinds of fashion styles and impressions, we conducted MDS analysis on the similarity assessment of each fashion sample for designer subjects and non-designer subjects as a whole and separately.

First,

the pair-wise comparison data were obtained from the 100 picture samples completed by 70 subjects. The data were then transformed into a matrix of heterogeneity. Through the formula calculation in MDS (ALSCAL) of the statistical program SPSS, we obtained the distribution of 100 picture samples distributed in the cognition space. Meanwhile, we conducted cluster analysis and obtained 10 clusters from all of the samples in the tree graph. Based on the questionnaire survey on 20 designers with more than 3 years of experience, and some common adjective terms for fashion styles from fashion journals, we chose 12 adjectives for styles and impressions (renascent, splendid, avant-garde, modern, leisurely, folk, elegant, technological, romantic, sexy, young, and neutral). Two senior fashion designing experts with experience of more than 10 years paired the 10 clusters of style sample pictures with 12 styles/impressions. They chose 10 types of fashion styles/impressions: renascent, avant-garde, modern, leisurely, folk, elegant, romantic, sexy, young and neutral. The cognition of each style / impression was summarized below. The results of MDS analysis showed that the higher degree the dimension of cognition space was, the more difficult the interpretation was. Due to the complex components of fashion styles/impressions, the R² value derived from the five-dimension analysis was only 0.50. Hence, we had to choose five dimensions to construct our cognition space of fashion styles and impressions. MDS was also used to calculate the coordinates of a fashion picture sample in this space. According to these values, we marked each sample on the cognition plane and its corresponding cognitive meaning at each axis, in order to have our initial examination.

3-2. The sample distribution of each fashion style/impression and axial corresponding analysis (1) The sample distribution of the neutral style/impression and axial corresponding analysis ( see fig. 2 ) The sample distribution of neutral style/impression was most prominent at axis I, while the distribution was more concentrated at axis II. The sample #51 was the one closest to the average point at both the axes I and II. (2) The sample distribution of the renascent style/impression and axial corresponding analysis ( see fig.3 ) The sample distribution of the renascent style was most prominent at axis I, while more concentrated at the axis II. The sample #18 was the one closest to the average ideal point at axis II.

4

fig.2

neutral style/impression and axial analysis

fig.3

renascent style/impression and axial analysis

(3) The sample distribution of the modern style/impression and axial corresponding analysis ( see fig. 4 ) The sample distribution of the modern style was more prominent and more concentrated at axis IV. The distribution was difficult to be distinguished at other axes. The sample #15 and #12 were closest to the average ideal point at axis II. (4) The sample distribution of leisurely style/impression and axial corresponding analysis ( see fig. 5) The sample distribution of leisurely style was scattered at each axis. The distribution was relatively more concentrated at axis IV. The sample #96 was close to the average point of the axis IV; the sample #57 was close to the axle center of axis I; the sample #64 was more close to the average point of axis II.

fig.4

modern style/impression and axial analysis

fig. 5

leisurely style/impression and axial analysis

(5) The sample distribution of young style/impression and axial corresponding analysis.( see fig. 6) The sample distribution of the young style was most concentrated at axis II, but the correlation was negatively significant. The sample #27 was more close to the axle center; the sample

5

distribution at the axis III had a significantly positive correlation, and sample #23 was more close to the average point. (6) The sample distribution of the romantic style/impression and axial corresponding analysis. ( see fig.7 ) The sample distribution of the romantic style was more concentrated at axis II with a significantly positive correlation. The sample #25 was closer to the average point. The sample distribution had negative correlations at the other four axes. It had significantly negative correlation at axis V., where the sample #60 was closer to the axle center. The sample #5 was closer to the average point at axis I; the sample #61 was closer to axes III and IV.

fig.6 youngstyle/impression and axial analysis

fig.8

fig. 7

elegant style/impression and axial analysis

romantic style/impression and axial analysis

fig. 9

sexy style/impression and axial analysis

(7) The sample distribution of the elegant style/impression and axial corresponding analysis( see fig. 8 ) The sample distribution of the elegant style was more concentrated at axis II with significantly positive correlation, and the sample #62 was closer to the axle center. The sample

6

distribution had significant negative correlation at axis I, where sample #35 was closer to the average point. (8) The sample distribution of the sexy style/impression and axial corresponding analysis ( see fig. 9) The sample distribution of the sexy style was more concentrated at axis I with significant negative correlation. The sample #67 was closer to the axle center. The distribution had positive correlations at other axes. The sample distribution had significantly positive correlation at axis III, where sample #39 was close to the average point. The distribution at the axis II was quite concentrated as well, where sample #70 was closer to the average point. Sample #67 was closer to the axle center of the axis V. (9) The sample distribution of the folk style/impression and axial corresponding analysis ( see fig. 10 ) The sample distribution of the folk style was most concentrated at the axis V with significantly positive correlation; the sample #80 was closer to the axle center. The distribution had negative correlations at other axes. The sample distribution had significantly negative correlation at the axis III, where sample #89 was closer to the average point. Among the distributions with negative correlations, the one at the axis II was more concentrated, where sample #92 was closer to the average point. The distribution was most scattered at axis IV. (10) The sample distribution of the avant-garde style/impression and axial corresponding analysis ( see fig.11 ) The sample distribution of the avant-garde style was most concentrated at axis V with significantly positive correlation, and sample #99 was closer to the axle center. Among the distributions with negative correlations, the distribution at axis I was more concentrated, where sample #77 was closer to the average point. The distributions at other axes were quite scattered.

fig.10

folk style/impression and axial analysis

fig. 11

analysis

7

avant-garde style/impression and axial

4. Conclusions After the initial multi-dimension assessment of cognition space of styles and impressions was finished, we asked six senior fashion designers (with more than 10 years average experience) to provide professional analysis of fashion constitution on the 100 fashion sample pictures. First, the constitutions were divided into five categories: contour, shape of collar, shape of shoulder, shape of the sleeve, and the front open lapel, coded with A, B, C, D, and E respectively. Detailed analysis was provided for each category. Then the five categories were coupled with10 style/impression terms respectively, and thereby a component table was laid out. We conducted the correlation analysis for these two domains. After the correlation analysis, we obtained a table with fashion styles/impressions and their corresponding constitution components. Lastly, we “iconized” each component for each style according to our study results, which meant we standardized all possible components that constituted each style/impression in order to complete the picture of each fashion model. This study aims to finish all basic components constituting the style/impression cognition, with the hope to diminish the difference in cognitions for all kinds of fashion styles/impressions of designers and users, and thereby offer a good basic understanding and application of fashion design styles for beginners. 4-1. Pictures of models of fashion styles and impressions ( see fig. 12 – 31 ) According to our research on 10 kinds of fashion styles and impressions and their components, there were 20 types of fashion models. We drew them one by one, illuminated in the pictures as follows. 1. Uni-sex style A. uni-sex style’s garment component one = loose waist, straight bottom & full length + notch collar + square shoulder + long narrow sleeves + center front opening ( fig. 12 ) B. uni-sex style’s garment component two = loose waist, straight bottom & full length + mandarin collar + square shoulder + long narrow sleeves + center front opening ( fig. 13 ) C. uni-sex style’s garment component three = l loose waist, straight bottom & full length + V neckline + square shoulder + long narrow sleeves + center front opening ( fig. 14 ) D. uni-sex style’s garment component four = loose waist, straight bottom & full length + shirt collar + square shoulder + long narrow sleeves + center front opening (fig. 15 ) 2. Retro style Retro style’s garment component = loose waist, straight bottom & full length + notch collar + square shoulder + long narrow sleeves + center front opening double breasted ( fig. 16 ) 3. Modern style A. modern style’s garment component one = nature waist, flare bottom & full length + mandarin collar + bare shoulder + without sleeves + without center front opening ( fig. 17 ) B. modern style’s garment component two = nature waist, straight bottom & full length + mandarin collar + bare shoulder + without sleeves + without center front opening ( fig. 18 ) C. modern style’s garment component three = nature waist, flare bottom & full length + boat

8

collar + bare shoulder + without sleeves + without center front opening ( fig. 19 ) D. modern style’s garment component four = nature waist, straight bottom & full length + boat collar + bare shoulder + without sleeves + without center front opening ( fig. 20 )

fig. 12 Uni-sex style A

fig. 13 Uni-sex style B

fig. 14 Uni-sex style C

fig. 15 Uni-sex styled

fig. 16 Retro style

4. Casual style casual style’s garment component one = nature waist, straight bottom & full length + mandarin collar + nature shoulder + without sleeves + without center front opening ( fig. 21 ) A. casual style’s garment component two = nature waist, straight bottom & full length + mandarin collar + nature shoulder + without sleeves + center front opening ( fig. 22 ) B. casual style’s garment component three = nature waist straight bottom & full length + shirt collar + nature shoulder + without sleeves + center front opening ( fig. 23 )

9

fig. 17 Modern style A

fig. 18 Modern style B

fig. 19 Modern style C

fig. 22 Casual style B

fig. 23 Casual style C

fig. 20 Modern

style D

fig. 21 Casual style A 5. Young-fashion style

A. young-fashion style’s garment component one = nature waist, straight bottom & full length + jewel neckline + nature shoulder + without sleeves + without center front opening ( fig. 24 ) B. young-fashion style’s garment component two = nature waist, straight bottom & full length + shirt collar + nature shoulder + without sleeves + without center front opening

10

( fig. 25 )

fig. 24

Young-fashion style A

fig. 26

Romantic style

fig. 25

fig. 27

Young-fashion style B

Chic style

fig. 28

Sexy style A

fig. 29 Sexy style

B

6. Romantic style Romantic style’s garment component = tight waist, flare bottom & full length + V neckline + nature shoulder + without sleeves + without center front opening ( fig. 26 ) 7. Chic style Chic style’s garment component = tight waist, straight bottom & full length + V neckline + nature shoulder + without sleeves + without center front opening ( fig. 27 )

11

8. Sexy style A. Sexy style’s garment component one = tight waist, flare bottom & full length + without collar + narrow shoulder strap + without sleeves + without center front opening ( fig. 28 ) B. Sexy style’s garment component two = tight waist, flare bottom & full length + asymmetrical neckline + asymmetrical shoulder strap + without sleeves + without center front opening ( fig. 29 ) 9. Folk style Folk style’s garment component = loose waist, straight bottom & half length + mandarin collar + nature shoulder + shirring cuff, long sleeves + center front opening ( fig. 30 ) 10. Avant-garde style Avant-garde style’s garment component = nature waist, straight bottom & full length + jewel neckline + nature shoulder + without sleeves cap, short sleeve + without center front opening ( fig. 31 )

fig. 30 Folk style

fig. 31 Avant-garde style

References 1. Chuang, Ming-Chuen, Applications of multi-dimension scaling to design, Theses of Technology and Teaching Seminars, Taipei, 1990. 2. Shih, Wei-Ming, The study of the Corresponding Relationship between Forms and Images of Eye-glasses, thesis, Institute of Applied Arts National Chiao-Tung University, Hsinchu, 1996. 3. Hung, Shu - Huei, A Study on the Relationship between Texture Image and Textile Fabrics of Bags, thesis, Institute of Applied Arts National Chiao-Tung University, Hsinchu, 1999 4. Kombayasi Shigenobu, Psychology of Design Constitution (by Ochi Hiroshi, translated by Shiu-Shiung Wang, Fundamentals of Arts Design, Taiwan Bookstore, Taipei Yi Fong Publisher Edition), Yi Fong Publisher, Taipei, 1991.

12

5. Chie Koike, The theories of Fashion Design, Arts and Crafts Book Co., Ltd, Taipei, 1992 6. Cheng, Chao-Ming, Cognitive Psychology – Theories and Practice, Laureate Book Co., Ltd, Taipei, 1993. 7. Chao Ping, Lui, I-Hua, Introduction to Psychology of Fashion Design, China Textile Press, Beijing, 1995.

8. Gordon I. E., Theories of Visual perception, John Wiley ＆ Sons , New York, 1989. 9. Zanfi Editori, Collezioni Donna, Vol.46,47,49, 1995; Vol.50,52,53,55, 1996; Vol.56,58,59, 61, 1997; Vol.62, 64,65,67, 1998; Vol.68,70,71, 1999; Vol.73,74, 2000, Italy.

13