Measuring The Brand Categry Through Semantic Differentiaition

Measuring the brand category through semantic differentiation Marcus Abbott Brunel University, Uxbridge, UK

John P. Shackleton Massey University, Auckland, New Zealand, and

Ray Holland Brunel University, Uxbridge, UK Abstract Purpose – This paper aims to explore the cognitive processing mechanisms of concepts and categories by examining the methodologies behind how branded-product concepts behave in the second of two co-incident alternative constructs – as a member of a product category, and in some cases, as a category by itself. General proposals for such mechanisms present language as a facilitator in the process. Therefore, linguistic concept assessment models are proposed to confirm the “brand as category” hypothesis evident in an example brand. Design/methodology/approach – The study extended conventional semantic differentiation (SD) methodologies; sets of bi-polar measures of concept properties describing the concept “semantic space”, to the brand category. Through iteration, the SD tool is refined and the effects of weighted scales understood. Findings – The results provide evidence that some brands do act as categories, with clearly identifiable exemplar positions within the brand-category “semantic space”. Practical implications – This paper offers interesting alternatives to established brand and product development activities concerned with the provision of product features and consumer benefits. Specifically, for many emotive, non-utilitarian products, brand attributes highly influence purchase decision, and therefore brand accuracy and differentiation, measured in the product’s properties, are key – characteristics that can be most saliently depicted in the “brand as category” alternative. Originality/value – This paper applies SD to the brand category for the first time. It provides a new methodology with advantages for brand and product managers concerned with the development of products that are not only “good” but also “right” for the brand. Keywords Brands, Brand management, Semantics, Product differentiation, Cognition Paper type Research paper

interest of cognitive science, while on the other, concept content structures are the domain of semiotics. Therefore, Rosch (1999), Lakoff (1987) and Fodor (1998) posit the mechanism as a categorisation system of determinant and dominant levels of concept entity inclusion, segmented by a taxonomy of examples of varying typicality, stimulated by deconstructable defining concept properties or attributes (Smith et al., 1988). Saussure (in Chandler, 2002), LeviStrauss (1963) and Eco (1976, 1985), however, place syntagmatic, inclusive, contextual relationships, between concept “signifiers” and “signifieds” in the vertical hierarchy (e.g. a and b and c relations, that ultimately form a concept “narrative”) and paradigmatic, substitutional examples in the horizontal (e.g. x or y or z similar alternatives) (see Figure 1). It is widely agreed that through linguistic facilitation, concept interpretation, reality and thought exists (e.g. Chandler, 2002). As a brand is a also concept, contemporary research has explored brand structure, interpretation and characteristics within the framework of both cognitive science (Aaker, 1991; de Chernatony and McDonald, 2003; Czellar, 2003; Keller, 1993; Lindstrom, 2005) and semiotic theory (Alexander, 1996; Karjalainen, 2005; MacFarquhar, 1994; Valentine and Evans, 1993), in order that it can be “leveraged” within the consumer mind-set for competitive advantage. However, where “brand” sits within the model is interpretable. Most

An executive summary for managers and executive readers can be found at the end of this article.

Introduction For decades, philosophers have debated concepts and categories – the basic, yet often complex structure of the cognitive interpretation of our thoughts, environment and products (Margolis and Lawrence, 1999). Most conclude the concept entity is a taxonomic, biological or goal-derived category of exemplars (Barsalou, 1983) that can be learnt and understood, by humans, and possibly other things endowed with the capacity for sensing and cognitive processing. A concept, therefore, could be “a person, place, thing, feeling, state of affairs, sense of foreboding, fantasy, hallucination, hope or idea” (Eco, 1976, p. 67). Equally, it can be a product and a brand (Franzen and Bouwman, 2001). On the one hand, however, concept processing mechanisms are the The current issue and full text archive of this journal is available at www.emeraldinsight.com/1061-0421.htm

Journal of Product & Brand Management 17/4 (2008) 223– 234 q Emerald Group Publishing Limited [ISSN 1061-0421] [DOI 10.1108/10610420810887572]

223

Measuring the brand category through semantic differentiation

Journal of Product & Brand Management

Marcus Abbott, John P. Shackleton and Ray Holland

Volume 17 · Number 4 · 2008 · 223 –234

Figure 1 Concept category model

ensure the thrust of the paper and its salient, broader implications are understood. The results indicate, strongly, that the brand, in this case, does behave as a category and that clear typicality factors emerge when measured within the category “semantic space”.

identify “brand” as a property of a concept example, of more or less typicality, on the horizontal axis in the product category (e.g. Henderson et al., 1998; Kreuzbauer and Malter, 2005; Meyvis and Janiszewski, 2002; Warlop et al., 2006; Karjalainen and Warell, 2005; Warell, 2006). For example, the graphic “tick” signifies the brand “Nike” example in the “sports shoe” category. Alternatively, however, co-incident hypotheses have explored brands as taxonomic categories in themselves (Abbott et al., 2007; Boush, 1993; Joiner and Loken, 1998; Loken et al., 2002); some brand structures seem to adhere to the horizontal and vertical concept structure axes, their constituent members being the products, of more or less typicality, of the brand category sharing degrees of common, deconstructable properties and signifiers. For example, the brand category “Porsche” is centrally defined by its sports-cars and more recently, an SUV (sport utility vehicle), as well as clothing and accessory brand “extensions”, through the possession of mutual product attributes. Brands within non-utilitarian market segments, like luxury vehicles, where purchase is discretionary and emotionally motivated (Abbott et al., 2006), are particularly receptive to the alternative hypothesis, as consumer values and benefits are obtained by the brand referencing specific aspirations and lifestyles (Belk, 1988), in its multi-sensory properties, signifiers and attributes (Carbonaro and Votava, 2005; Karjalainen and Warell, 2005; Ratneshwar and Shocker, 1991). It is important, therefore, that the branded product, in this case, be differentiated as an example of the brand category, as there is evidence that consumer satisfaction is derived from it being so (Abbott et al., 2006). This paper explores the alternative brand concept hypothesis, proposes a methodology for branded product concept measurement, and thus a route-map to the potential influence of concept properties by brand managers and product designers for consumer satisfaction. Linguistic semantic differentiation methodology is applied to products from a luxury vehicle brand, called “Luxori” for the purposes of this paper to protect confidentiality. Therefore, some subject detail is also omitted, although sufficient is included to

Semantic differentiation methodology A central tenet of semiotic theory is the employ of opposites in the cognitive identification of concepts; we tend to understand what something is by first knowing what it is not (Chandler, 2002; Valentine and Evans, 1993). This phenomenon acts on the model above through metaphoric and paradigmatic constructs whereby certain properties are considered inclusive and defining of the concept category and others related but excluded. For example: If I say “apple” and you respond “orange”, “pear” or “banana” you are working on a paradigmatic axis [. . .] If you say “eat”, “juicy” or “Garden of Eden” your response is placing “apple” in a syntax of combined terms (Valentine and Evans, 1993, p. 132).

Semantic differentiation (SD) theory provides a medium to explore such metaphors (initially demonstrated in relationships between music and colour by Osgood et al., 1957). The originators proposed concepts are most comprehensively, consistently and efficiently captured by language. Subsequently, SD has been extensively adopted by psychologists, consumer researchers and others as a highly popular affective evaluation technique due to its correlation reliability and economy in data recording and retrieval (Franzen and Bouwman, 2001; Heise, 1970; Martindale, 1990). Studies have agreed that SD’s antonym relationship scales (Figure 2), balance semantic sensitivities with data recording economy (e.g. Batra et al., 2006; Bhise et al., 2005; Ratneshwar et al., 1987; Schutte, 2005; Zhang and Shen, 1999), whilst, in particular, the seminal Thesaurus Study (Osgood et al., 1957) also provided first evidence of predominant evaluation, potency and activity factors (E, P, A; < 65 per cent of all variance; Osgood et al., 1957) present within semantic assessments for multiple common concepts. 224

Measuring the brand category through semantic differentiation

Journal of Product & Brand Management

Marcus Abbott, John P. Shackleton and Ray Holland

Volume 17 · Number 4 · 2008 · 223 –234

Figure 2 Typical seven point SDS, with example from the “Luxori” brand property assessment model

and, for example, “Rough” and “Smooth” do not correlate. It is therefore suggested that the model extracts most value with standardised scales for multiple concepts, when D is the focus of interest, and when trends are important (Osgood et al., 1957). These criteria effectively eliminate issues between scales and factors by maintaining constant measurement criteria, but do not address the central issue. Nevertheless, by ensuring the “statistical character” of the test by standardising scales, or concepts, or both, statistical significance can be obtained for both individual and group assessments (Osgood et al., 1957).

EPA factors, then, provide an opportunity to describe the concept “semantic space”. Multidimensional in nature, the affective direction and intensity (or “emotionality”) of the concept can be calculated and represented for single or multiple subjects, is highly stable (Heise, 1970), and is therefore an effective analytical and communication tool for concept position. Semantic difference (D) calculates similarity relationships to other concepts: p D ¼ ðe1 2 e2 Þ2 þ ð p1 2 p2 Þ2 þ ða1 2 a2 Þ2 ; where e is evaluation, p is potency and a is activity. As difference is also a measure of similarity, and similarity relates to typicality (Smith et al., 1988), the calculation effectively describes concept category relationships of prototypicality. Some criticisms, however, have been raised against SD as an evaluation technique, the most prominent being the assumption that distances between scale increments are equal for every subject, and correlate between scales. For example, how is it ensured that a þ 3 score for “Precise” (in Figure 2) is the same for every evaluator, given that the “intensity” of “precision” is highly subjective? Collapsing scores across subjects provides an appropriate and commonly adopted solution. Alternatively, in other fields like kansei engineering, scale intensities are more accurately controlled (in some tests) by the utilisation of “true” Likert scales, containing a biased pole of full content verses full absence (agreement that an attribute is missing is more likely to be repeatable), set against a single concept statement (Figure 3) (Schutte, 2005). For both principles, central tendency bias is a risk if understanding is lacking (Schutte, 2005). The disadvantage of the Likert scale, appears, however, to be two-fold: 1 very specific meaning vectors cannot be explored (for example, Wordnet 2.1 identifies seven senses for the word “Solid” with antonyms “Liquid”, “Hollow” and “Soft”, listed amongst others; see http://wordnet.princeton.edu); and 2 opposing properties for concepts that are multi-directional towards different typicality centroids, along common factors, cannot be measured.

Establishing properties and the category “semantic space” For complex concepts, like a luxury vehicle brand’s product, detailed analysis from multiple resources is necessary to acquire a full and accurate property list. In the course of the current “Luxori” brand study, significant properties were earlier identified through analysis of their branded products produced over a number of decades, supplemented by alternative company market-research. Following Osgood et al.’s (1957) recommendation, expert analysis concluded by identifying 20 multi-sensory properties, unequally but appropriately distributed, into six brand differentiating features within the interior: 1 overall materials and appearance; 2 trim panels; 3 carpets; 4 brightware; 5 veneer; and 6 overall control functionality. Against each property, positive adjectives were chosen from the previous research database. Selection of specific property antonym pairs were made by reference to the Thesaurus Study (Osgood et al., 1957), ensuring strong primary factoring against E, P or A, and Wordnet 2.1 (see http:// wordnet.princeton.edu) to confirm the antonym relationship within regular English. Examples of three of the 20 properties and associated word-pairs are given in Figure 4. A check of the mix of EPA primary factors was conducted to ensure the recommended proportion (Heise, 1970); adjustments were made to word-pairs to give a final 9:6:5 ratio. The synonym column effectively describes prototypical focus brand attributes for the branded product concept. Under test conditions it was therefore expected that a brand profile would be obtained that would be linear and to the right (Strong, positive) of the scale for the “Luxori” brand and alternative profiles obtained for competitor examples. Conversely, the multidimensional concept category “semantic space” would be characterised by a cluster along a strongly positive vector for all three factors, with a low D value, for “Luxori” brand products and with other brands distributed elsewhere (see Figure 5).

For example, the concept “Trim panel softness” with the bipolar potency factor adjectives “Hard ! Soft”, along a single SDS may serve to simultaneously measure the direction towards the “Luxori” aspired brand prototype of “Soft” and a competitor brand prototype of, possibly, “Hard”; an advantage to the brand or product manager who strives to differentiate by maximising D for their brand against competitor examples. The second issue of equality between scales also cannot be established; the distances between “Precise” and “Imprecise” Figure 3 Seven point Likert scale, typical of kansei engineering approach

225

Measuring the brand category through semantic differentiation

Journal of Product & Brand Management

Marcus Abbott, John P. Shackleton and Ray Holland

Volume 17 · Number 4 · 2008 · 223 –234

Figure 4 Example SDS and EPA primary factors from the “Luxori” brand

Figure 5 Concept category “semantic space” with expected cluster for the “Luxori” brand

Methodology refinement

Test 1

Three pre-tests were initially conducted to refine properties and word-pairs selected, test methodology, response form layout and to understand initial test repeatability. With these and all subsequent tests, attention was paid to assessments of actual product examples (vehicles), rather than substitute media, to avoid problems with stimuli restriction and to make assessment ratings as “rich” as possible. Four pairs of assessors (eight in total) performed Pre-test 1 on a single vehicle from the “Luxori” brand range. Assessors were “expert” in the commodities rated but were not provided with pre-instructions or help during the assessment. Individual pairs conducted ratings at different times. Pretest 2 examined potential differences between two further products from the “Luxori” range, and the repeatability of the test over time, with two pairs of “expert” assessors rating each vehicle four weeks apart. Pre-test 3 observed a wider vehicle set, one from “Luxori” and three competitor examples, by three assessor pairs (six in total), two “expert” pairs and one “non-expert” pair. Summarising all pre-tests, overall rating repeatability was strong, especially for the “Luxori” brand products studied. Rating spread was greater for competitor vehicles, however, providing some first evidence of brandcategory effects by confirming the property mix and bi-polar scales were more specific and accurate for the brand, but were inconsistently interpreted for examples falling outside the category. It is reasonable to expect, therefore, that such effects may not be present if all examples were assessed against product category properties and a reverse effect would be demonstrated if alternative brand properties were listed.

Test 1 was designed to establish whole-vehicle attribute targets for the “Luxori” brand largest production volume product against seven competitor vehicles (m1-m7) from six alternative brands in advance of a new model programme (see Table I). Two examples of the “Luxori” brand product range were included (ma and mb), providing nine assessed vehicles in total. Fourteen pairs of assessors (28 in total), all employees of the company, rated each vehicle using the SDS described above. Assessments were made over a period of three days, with each assessor-pair spending an average of one and a half hours in each vehicle, which were randomly mixed in order. Assessments were made in the vehicles both statically and dynamically, partly on a pre-defined test-track route and partly on a pre-defined public road route. Weather conditions were dry, sunny and cold (, 108C). Results from the SDS assessment were summarised into a property ðkÞ £ concept (m) matrix whereby subject (n) mean property scores were calculated for each property against each concept (in this case, each vehicle interior). For the two “Luxori” brand products, the scores were averaged again to give a single brand profile, illustrated in Figure 6. The profile generally matches a positive linear inclination towards the expected brand concept profile with other brands exhibiting alternative profiles. Figure 6 includes two for comparison; one “closest” in concept to the focus brand (m4) and one “furthest away” (m2). However, it is also clear that in three property areas (kf:kg, ki:kj, kr), the “Luxori” brand itself drifts from a strong positive position towards neutral. In each of the three neutral clusters, expert analysis subsequently identified actual property degradation from the brand’s aspiration in the 226

Measuring the brand category through semantic differentiation

Journal of Product & Brand Management

Marcus Abbott, John P. Shackleton and Ray Holland

Volume 17 · Number 4 · 2008 · 223 –234

Table I Semantic difference (D) between “Luxori” brand product (ma) and others

Product “Luxori” 2 “Luxori” 3 “Luxori” 4 Porsche 911 Turbo Ferrari 612 BMW M6 Mercedes S600L Mercedes CL500 Aston Martin DB9 Jaguar XKR

mb mc md m1 m2 m3 m4 m5 m6 m7

Semantic difference (D) (“Luxori” 1½ma
! mb ; m1 ; . . . ; – m7 Þ

Semantic difference (Dw) (“Luxori” 1½ma
! mb ; m1 ; . . . ; m7 Þ, weighted

0.54 0.28 0.47 3.6 5.14 3.68 2.3 2.75 2.43 5.13

0.22 0.14 0.37 1.79 2.62 1.82 1.18 1.41 1.74 2.6

Figure 6 Brand profiles against “Luxori” brand properties, with areas of neutral property positions

distances (D ¼ 2:3-5:14, Table I) between the brand’s products and competitors, whilst the tight clustering evident between the brand’s products (D ¼ 0:54, Table I), within the expected zone (Figure 5). Results, therefore, hint at prototypicality effects; by small D values being evident, subjects expressed clear recognition and understanding of the two vehicles belonging to the brand concept category. Therefore, also, large D value differences between ma and mb and competitors builds further evidence, within this research, towards the brand acting as a category.

examples (such tendencies were also evident in each of the three pre-tests). Although at this stage misinterpretation of these specific properties was not eliminated, it was considered unlikely. Each E, P or A factored property score was summed and averaged for each vehicle concept giving a three co-ordinate position within the “Luxori” brand concept category “semantic space”. In this calculation, the two “Luxori” brand products were separated (ma and mb) to provide an opportunity to assess familiarity and prototypicality effects. Although, in the brand profile calculation, some ambiguity was present for some of the “Luxori” brand properties against competitive sets, and some neutrality was detected, the EPA calculation demonstrated clarity of the “Luxori” brand concept positioning (Figure 7), with large semantic

Discussion Test 1 identified clear “brand-category” effects, including prototypicality and category “boundaries”, by assessment against the “Luxori” brand properties, represented in the 227

Measuring the brand category through semantic differentiation

Journal of Product & Brand Management

Marcus Abbott, John P. Shackleton and Ray Holland

Volume 17 · Number 4 · 2008 · 223 –234

Figure 7 “Luxori” brand concept category “semantic space” with competitor positions

and activity properties, except for kf (D1.3; p), kg (D0.6; e), ks (D0.6; e) and ku (D0.9; e). These four properties were expertly compared between Tests 1 and 2, concluding that specific property variations were evident between the products according to the directions assessed (e.g. property kf was evidently stronger (more positive) towards the brand aspiration in mc than ma and mb, whilst ku was evidently weaker in mc than ma and mb). Similarly, expert analysis agreed tangible product property weaknesses for mc against kf:kg, ki:kj and kr and also for ku, which were not evident in either ma or mb. EPA factor property scores were then calculated, according to Test 1 procedures, for concept mc (Figure 7). As a member of the brand-category, mc demonstrated very close brand familiarity (D ! ma ¼ 0:28; Table I) to ma and mb within the brand concept “semantic space”, equalling ma in p (1.5) and a (1.7) but slightly lower in e (1.6 versus 1.9), potentially due to the product being a model older in design, and therefore not executed as well (was not as “good”), or that some modification due to “novelty” effects may be present in the assessments of ma and mb (Hekkert et al., 2003; Snelders and Hekkert, 1999). Discussion The objectives of Test 2 were to understand SDS model repeatability for the branded product concept and to understand if brand-category effects would be evident for an example from a conceptually discreet product from the “Luxori” brand. Results show that high levels of test repeatability were present, with the alternative concept example clustering very closely to other examples from the brand in both brand property profile and the brand-category “semantic space”. Any differences in values recorded were clearly explainable as apparent product property differences between the examples rather than test variation. As such, the SDS model does appear to offer a stable, and therefore valuable, alternative as both a product development and brand development instrument for assessing multi-sensory product properties related to brand aspirations that communicate and reinforce that message with the consumer. As discussed above, this is particularly relevant in market segments where brand values are significant purchase motivators. The model also enables competitive positions to be mapped, thereby alerting the brand custodian to brand concept “overlap” or opportunities for further differentiation or clarity in the brand message communicated by the product. However, Test 1 and Test 2 both concentrated on mature products within the public domain, where necessary multisensory interaction is prearranged. Therefore, questions rise over the model’s value for new product development, when the media employed to develop the product may limit stimuli (e.g. tactile, auditory and smell characteristics are often only assessable when physical prototypes are available), but when, in fact, it would also be most influential in the design process. A further test was designed to understand the effectiveness of the model under limited stimuli on a new product under development.

concept category “semantic space”. The test also identified property areas for the brand’s products which fell short of the brand’s aspirations; a useful aid to future product improvement and brand development activity. However, as the first data gathering exercise, using SDS in this way, repeatability of the category and prototypicality effects described is unproven (although anticipated due to previous studies; for example Heise, 1970). Additionally, the two vehicles from the “Luxori” brand range were very closely related, especially in interior execution, potentially minimising semantic difference by small product variation rather than perceptible differences in property values. Therefore, two further tests were designed to understand SDS and “brandcategory” effect repeatability for alternative products from the “Luxori” brand; one other product (mc), unrelated to (ma) or (mb) and one future-model product (md).

Test 2 Test 2 featured a related, but conceptually discreet, product from the “Luxori” brand range (mc). Thirty-seven “expert” and “non-expert” assessors from the company, working in pairs or in groups of three, spent up to 30 minutes each assessing the example using the SDS from Test 1. The assessments were made with the vehicle in a static condition. None of the assessors from Test 1 were employed in Test 2. Results were analysed in the same manner as Test 1, producing a property profile mapped against that of the “Luxori” brand products, (ma) plus (mb), shown in Figure 8. Although assessments were made by a completely unique group in Test 2 to Test 1, at a different time and against an alternative example, correlation between Tests 1 and 2 supported Heise (1970) repeatability predictions, with all property deviations falling within the expected 0.5 scale units for evaluation properties and 0.7 to 1 scale units for potency

Test 3 Test 3 featured a full-size model of a proposed new vehicle interior concept (md) from the “Luxori” brand. The model was constructed from modeller’s clay but was painted to 228

Measuring the brand category through semantic differentiation

Journal of Product & Brand Management

Marcus Abbott, John P. Shackleton and Ray Holland

Volume 17 · Number 4 · 2008 · 223 –234

Figure 8 Brand profiles for “Luxori” brand products ma, mb and mc

simulate correct colours, materials and finishes. During assessment, concepts for “below vehicle waist” were demonstrated; the model exhibiting without a roof or trunk. Nine assessors viewed the model for 20 minutes using the SDS model from Tests 1 and 2. All were “expert” in the design and engineering of the vehicle but some had not experienced the assessment method previously. The assessors were asked to ignore property word-pairs where the lack of available stimuli did not afford accurate assessment so that following “expert” group prediction of performance, based upon anticipated technology content, could be made. Results were again analysed according to procedures from Tests 1 and 2, with a property profile mapped against the subjects from those tests (ma, mb and mc) (Figure 9). Property points were measured, predicted by agreement against property scores for ma, mb and mc. Data, again, proved repeatable, or alternatively justifiable, based upon actual product variations, e.g. kd (D ¼ 1.6; e) demonstrated significantly weaker property value in md than ma, mb or mc whilst kq (D ¼ 0.6; a) was demonstrably stronger than ma, mb or mc. EPA factor values were then calculated for md according to Tests 1 and 2. Likewise, md clustered close to ma, mb and mc, within the brand category “semantic space” (D ! ma ¼ 0:47) (Table I), although a weaker e value, in this case, positioned the example at the boundary of the target zone due to low assessment values for kd (Figure 7), where some specific property execution concerns heavily influenced the categorisation effect. However, application to non-limited stimuli examples of this design, by non-expert assessors, may correct the assessment, as some “manipulation” of the score in Test 3 was articulated during the assessment in order to affect subsequent design change. In such instances, the model

proved to be highly sensitive to variations of a single property, although the overall semantic difference (D) for md was closer to ma than mc to ma. Discussion Test 3 confirmed previous trends in the brand-category effects of examples from the “Luxori” brand measured by the SDS model, therefore providing further evidence that some brands act as categories. In restricting property stimuli, however, predictive scores were relied upon to provide a full brand profile that resulted in familiarity of the example to the brandcategory in this case. In doing so, the model proves effective for forecasting consumer recognition of a new example to the brand concept during the development process, thereby offering advantages to brand and product managers concerned with the development of products that are not only “good” but also “right” for the brand. However, Test 3 also identified sensitivities in the SDS model that indicate property identification accuracy is particularly significant for brand-concept classification repeatability and accuracy. Whilst the tests suggest that property identification in this case is accurate, no account for property weighting within the set was made. It appears reasonable to agree that, within cognitive processing, more or less emphasis is placed upon certain properties and certain stimuli in the identification and understanding of the brand concept (de Chernatony and McDonald, 2003; Romaniuk and Sharp, 2004). Therefore, as the SDS model involves necessary sensitivity, weighting the 20 properties from Tests 1-3 may result in adjustments in D that contradict the brandcategory hypothesis for the “Luxori” brand, or indeed, for other subjects in other applications or condense D to strengthen the category clustering effect, or both. 229

Measuring the brand category through semantic differentiation

Journal of Product & Brand Management

Marcus Abbott, John P. Shackleton and Ray Holland

Volume 17 · Number 4 · 2008 · 223 –234

Figure 9 Brand profiles for “Luxori” brand products ma, mb, mc and md

Weighting properties in Tests 1-3

tests, with concepts m2 and m7 shifting by 2.52 and 2.53 D scale units respectively, whilst Dw (D, weighted) for the alternative “Luxori” brand examples (mb, mc and md) reduced by between 0.32 and 0.1 (Table I), although the overall distribution (compare Figure 10 and Figure 7) was not significantly transformed. Again, md appeared on the periphery of the brand-category target zone due to its low score in kd combined with the highest weighting of the property set (0.92). It is apparent, however, that the effect of weighting EPA in this test amplifies constructs in the competitor sets that are most salient in their concept positions; for example, m6 is the closest vehicle in “character” to the “Luxori” brand concept (style, materials used, etc.), but it lacks the standard of “quality” of execution (degree of achieved quality; Kano et al., 1984) of m4 or m5. Therefore, the largest relative shift to weighted EPA for m6 was along the activity vector (De ¼ 0:1, Dp ¼ 0:1, Da ¼ 0:3). Similarly, m2 is closer in character to the “Luxori” brand concept than m4 or m5, which is not reflected in the non-weighted positions, but which is by weighting EPA. However, m4 and m5 are closest to the “Luxori” brand concept in both “quality” of execution and the level of “detail” (or “layered elaboration”) that both “character” and “quality” express. Examining concept positions within the focus-brand category “semantic space” concludes in proposed designation of the EPA vectors as relative measures of the “Luxori” brand “quality” (evaluation), “elaboration” (potency) and “character” (activity), closely related to Osgood et al.’s (1957) semantic vectors of “reward”, “size” and “warmth” respectively.

Amongst the established methodologies deployed for understanding consumers’ product attribute preference by marketing and other disciplines is conjoint analysis. Conjoint analysis forces a trade-off between concept properties (Aaker, 1991; Dean, 2004) in an experimental approach that establishes property weightings, or values, for particular sets of subjects by asking consumers to “compensate” one property versus another (Koslow, 1999). By utilising a property £ property (k £ k) matrix a pair-wise conjoint-type analysis (Warell, 2006) was made on the 20 focus brand properties (ka, kb, kc, . . . , ku) in a panel discussion by “expert” assessors from the focus brand company. Dominance of one property over another (e.g. x ¼ ka . y ¼ kg ) was indicated by a value of 1, whereas conversely reverse dominance (e.g. x ¼ ka , y ¼ kg ) was indicated by a value of 2 1. Property pair equality was given a value of 0. Following dominance agreement, eachP column is summed and a relative sum calculated (e.g. ka ; nk), with weightings given by: P a k ; nk 2 1 : nk 2 1; nk 2 1 Weightings calculated for the “Luxori” brand properties are shown in Table II. The EPA factor ratio became adjusted to 8.8:7:4.2 from 9:6:5 in Tests 1-3, de-emphasising e, in kb especially. Each property weighting was then applied to the brandconcept (ma, mb, mc and md) and competitor concepts (m1m7), within the brand concept category “semantic space” (Figure 8). Positioning altered somewhat from the non-weighted 230

Measuring the brand category through semantic differentiation

Journal of Product & Brand Management

Marcus Abbott, John P. Shackleton and Ray Holland

Volume 17 · Number 4 · 2008 · 223 –234

Table II “Luxori” brand concept property weightings with corresponding EPA factors ka

kb

kc

kd

ke

kf

kg

kh

ki

kj

kk

kl

km

kn

ko

kp

kq

kr

ks

kt

0.47 E

0.00 E

0.63 E

0.92 E

0.82 A

0.61 P

0.24 E

0.87 P

0.29 A

0.45 P

0.24 A

0.37 E

0.18 P

0.63 E

0.61 P

0.63 A

0.08 A

0.79 E

0.79 P

0.39 E

product category. However, a small but growing field of research has looked at branded product concepts alternatively; the brand as a category of products which share “necessary and sufficient” conditions for membership. This paper adds to the knowledge by both providing further evidence that some brands do appear to behave as categories, with clear typicality factors, and by offering a route-map for measuring that effect, and therefore providing a potential instrument for manipulation of it, for consumer satisfaction. By developing the established semantic differentiation scale methodology, the linguistic device proves to be quick, reliable, accurate and highly sensitive to both specific brand properties that are congruent with consumer expectations and beliefs, and to discreet concepts. Emphasis is drawn to the full and careful choice of antonym word-pairs for multi-sensory properties that are appropriately allocated to the SD primary evaluation, potency and activity factors and weighted. However, stimuli restriction, whilst being unadvised, does not deteriorate the potency of the model if combined with expert predictive scoring based upon known product feature content. The brand studied has benefited from such analysis, under these conditions, stressing strengths and weaknesses in product propositions and presenting clearly defined categorisation effects that can be further exploited. Reciprocal competitor categorisation studies would only enrich understanding of the brand “semantic space”. Indeed, it is because the luxury vehicle market relies heavily on emotional brand values and associated benefits that are rooted in the product’s familiarity to the category “prototype” that the model is so valuable for brand and product custodians when developing future brand or product propositions and extensions.

Figure 10 “Luxori” brand concept category “semantic space”, with weighted properties; Tests 1, 2 and 3 combined

Discussion The weighting of properties within the tests of the SDS model appears to both amplify methodology sensitivities and to intensify saliency within the brand-category studied. Its effect, then, appears to improve the accuracy of the cognitive representation of the “brand as category” hypothesis. In application, some sophisticated messages also appear to be communicated by following this course; inclusive competitive positioning, brand concept typicality and semantic interpretations flow, which offer potential advantages to brand and product development activities. In a single application, however, questions will always be asked over applicability to alternative brand-categories, which requires supplementary clarification, particularly as sensitivities appear in property, word-pair and weighting choice. Further, reciprocal effects of the “Luxori” brand positioning within competitor brand-category “semantic spaces” would be necessary to add depth of understanding to the “brand as category” proposition.

Managerial implications This paper reflects important trends evident within many mature product segments; “the brand”, in its many definitions, has grown to be a significant product differentiator in the consumer mind-set, not only constituting a recognisable product name, or nomenclature, but a set of values or beliefs, expressed by all aspects of the product, that the consumer wishes to participate with. In such circumstances, careful management of the brand can be a key commercial success factor that not only concerns marketing departments but designers and engineers as well; it is their collective brief to create products that are not only good, but also right for the brand. To understand, however, what is right for the brand, what fits within the “brand category”, it is beneficial to look at the concept of “the brand” in one of two ways: 1 as a constituent element of a product that exists within a product segment, or category; or, more especially 2 as a category which contains examples, of more or less familiarity, to the brand’s actual or target values.

Conclusion Many fields of research and practice have discussed the cognitive processing mechanisms of concepts and categories. In commercial applications, the majority have centred on “the brand” existing as an element of a product that inhabits a 231

Measuring the brand category through semantic differentiation

Journal of Product & Brand Management

Marcus Abbott, John P. Shackleton and Ray Holland

Volume 17 · Number 4 · 2008 · 223 –234

By taking the second view, it may be possible to identify the specific constituent multi-sensory product properties, or attributes, that define the branded product concept within the brand category and manipulate them, through product specification, to better fit the brand proposition. This paper applies semantic differentiation (SD) methodology, used extensively within many fields as an accurate, sensitive and efficient linguistically based concept analysis tool, to the branded product concept to measure, for the first time, how “right” a product is within the brand category (how representative it is of the category “prototype”). The methodology allows competitor positions to be mapped within the focus brand “semantic space”, visually demonstrating unique positions, threats, and overall brand characters, for existing products and new offerings. By weighting attributes, sensitivity is increased and dominant characters emphasised. Further, by discreet adjustments to property specifications within the model, consumer categorisation effects can be predicted, enabling brand extensions to be evaluated for familiarity against the parent brand’s values. For a number of brands, like luxury vehicles, where utilitarian product attributes are not major purchase motivators, and where price is more or less desensitised, nurturing the brand’s values and associations drives consumer satisfaction. SD methodology applied to the “brand as category” in this case (but not exclusively so), presents potential value to brand management and product development activities by maximising design and engineering efficiency (getting the product right for the brand quicker), maximising brand positioning opportunities (by enabling the strengthening of brand character features) and mitigating risk against competitive product positions or brand extensions.

Boush, D.M. (1993), “Brands as categories”, in Aaker, D.A. and Biel, A.L. (Eds), Brand Equity and Advertising: Advertising’s Role in Building Strong Brands, Lawrence Erlbaum Associates, Mahwah, NJ, pp. 299-313. Carbonaro, S. and Votava, C. (2005), “Paths to a new prosperity”, The Nordic Textile Journal, pp. 71-85. Chandler, D. (2002), Semiotics: The Basics, Routledge, Abingdon. Czellar, S. (2003), “Consumer attitude toward brand extensions: an integrative model and research propositions”, International Journal of Research in Marketing, Vol. 20, pp. 97-115. de Chernatony, L. and McDonald, M. (2003), Creating Powerful Brands, Elsevier Butterworth-Heinemann, Oxford. Dean, D.H. (2004), “Evaluating potential brand associations through conjoint analysis and market simulation”, Journal of Product & Brand Management, Vol. 13 No. 7, pp. 506-13. Eco, U. (1976), A Theory of Semiotics, Indiana University Press, Bloomington, IN. Eco, U. (1985), “Producing signs”, in Blonsky, M. (Ed.), On Signs, Basil Blackwell, Oxford, pp. 176-83. Fodor, J.A. (1998), Concepts: Where Cognitive Sciences Went Wrong, Oxford University Press, Oxford. Franzen, G. and Bouwman, M. (2001), The Mental World of Brands – Mind, Memory and Brand Success, World Advertising Research Centre, Oxford. Heise, D.R. (1970), “The semantic differential and attitude research”, in Summers, G.F. (Ed.), Attitude Measurement, Rand McNally, Chicago, IL, pp. 235-53. Hekkert, P., Snelders, D. and van Wieringen, P.C.W. (2003), “‘Most advanced yet acceptable’: typicality and novelty as joint predictors of aesthetic preference in industrial design”, British Journal of Psychology, Vol. 94, pp. 111-24. Henderson, G.R., Iacobucci, D. and Calder, B.J. (1998), “Brand diagnostics: mapping branding effects using consumer associative networks”, European Journal of Operational Research, Vol. 111, pp. 306-27. Joiner, C. and Loken, B. (1998), “The inclusion effect and category-based induction: theory and application to brand categories”, Journal of Consumer Psychology, Vol. 7 No. 2, pp. 101-29. Kano, N., Seraku, N., Takahashi, F. and Tsuji, S. (1984), “Attractive quality and must-be quality”, Hinshitsu (Quality: The Journal of the Japanese Society for Quality Control), Vol. 14 No. 2. Karjalainen, T. (2005), “Safe shoulders and personal faces: transforming brand strategy to product design”, Proceedings of the 12th International Product Development Conference, Copenhagen, 12-14 June. Karjalainen, T. and Warell, A. (2005), “Do you recognize this tea flask? Transformation of brand-specific product identity through visual design cues”, paper presented at the International Design Congress, Taiwan, 1-4 November. Keller, K.L. (1993), “Conceptualizing, measuring, and managing customer-based brand equity”, Journal of Marketing, Vol. 57 No. 1, pp. 1-22. Koslow, S. (1999), “Conjoint analysis”, in Earl, P.E. and Kemp, S. (Eds), Consumer Research and Economic Psychology, Edward Elgar, Cheltenham. Kreuzbauer, R. and Malter, A.J. (2005), “Embodied cognition and new product design: changing product form to influence brand categorisation”, Journal of Product Innovation Management, Vol. 22, pp. 165-76.

References Aaker, D.A. (1991), Managing Brand Equity – Capitalizing on the Value of a Brand Name, The Free Press, New York, NY. Abbott, M., Shackleton, J.P. and Holland, R. (2007), “‘Brand’ as category: an analysis of categorisation and branded product concepts”, Proceedings of IASDR07, Hong Kong, 12-15 November. Abbott, M., Shackleton, J.P., Holland, R., Guest, P. and Jenkins, M.-J. (2006), “Engineering emotional identities in high-luxury vehicles”, Proceedings of the 5th International Conference on Design and Emotion, Gothenburg, 27-29 September. Alexander, M. (1996), “The myth at the heart of the brand”, paper presented at ESOMAR Seminar: The Big Brand Challenge, available at: www.semioticsolutions.com Barsalou, L.W. (1983), “Ad-hoc categories”, Memory & Cognition, Vol. 11 No. 3, pp. 211-27. Batra, R., Lenk, P. and Wedel, M. (2006), “Separating brand from category personality”, University of Michigan, Ann Arbor, MI, available at: www.umich.edu Belk, R.W. (1988), “Possessions and the extended self”, Journal of Consumer Research, Vol. 15 No. 2, pp. 139-68. Bhise, V.D., Hammoudeh, R., Nagarajan, R., Dowd, J.D. and Hayes, M. (2005), “Towards development of a methodology to measure perception of quality of interior materials”, SAE Transactions, Vol. 114 No. 5, pp. 503-9. 232

Measuring the brand category through semantic differentiation

Journal of Product & Brand Management

Marcus Abbott, John P. Shackleton and Ray Holland

Volume 17 · Number 4 · 2008 · 223 –234

About the authors

Lakoff, G. (1987), Women, Fire and Dangerous Things, University of Chicago Press, Chicago, IL. Levi-Strauss, C. (1963), Structural Anthropology, Basic Books, Oxford. Lindstrom, M. (2005), Brand Sense: How to Build Powerful Brands through Touch, Taste, Smell, Sight and Sound, Kogan Page, London. Loken, B., Joiner, C. and Peck, J. (2002), “Category attitude measures: exemplars as inputs”, Journal of Consumer Psychology, Vol. 12 No. 2, pp. 149-61. MacFarquhar, L. (1994), “This semiotician went to market”, Lingua Franca, September/October, pp. 59-79. Margolis, E. and Lawrence, S. (Eds) (1999), Concepts: Core Readings, MIT Press, Cambridge, MA. Martindale, C. (1990), The Clockwork Muse: The Predictability of Artistic Change, Basic Books, New York, NY. Meyvis, T. and Janiszewski, C. (2002), “When are broader brands stronger brands? An accessibility perspective on the formation of brand equity”, available: at: www.cba.ufl.edu (accessed 27 July 2006). Osgood, C.E., Suci, G.J. and Tannenbaum, P.H. (1957), The Measurement of Meaning, University of Illinois Press, Champaign, IL. Ratneshwar, S. and Shocker, A.D. (1991), “Substitution in use and the role of usage context in product category structures”, Journal of Marketing Research, Vol. 28 No. 3, pp. 281-95. Ratneshwar, S., Shocker, A.D. and Stewart, D.W. (1987), “Toward understanding the attraction effect: the implications of product stimulus meaningfulness and familiarity”, Journal of Consumer Research, Vol. 13 No. 4, pp. 520-33. Romaniuk, J. and Sharp, B. (2004), “Conceptualizing and measuring brand salience”, Marketing Theory, Vol. 4 No. 4, pp. 327-42. Rosch, E. (1999), “Principles of categorisation”, in Margolis, E. and Lawrence, S. (Eds), Concepts: Core Readings, MIT Press, Cambridge, MA. Schutte, S. (2005), “Engineering emotional values in product design”, doctoral thesis, Linko¨ping University, Linko¨ping. Smith, E.E., Osherson, D.N., Rips, L.J. and Keane, M. (1988), “Combining prototypes: a selective modification model”, Cognitive Science, Vol. 12, pp. 485-527. Snelders, D. and Hekkert, P. (1999), “Association measures as predictors of product originality”, Advances in Consumer Research, Vol. 26, pp. 588-92. Valentine, V. and Evans, M. (1993), “The dark side of the onion: rethinking ‘rational’ and ‘emotional’ responses”, Journal of the Market Research Society, Vol. 35 No. 2, pp. 125-43. Warell, A. (2006), “Identity recognition in product design: an approach for design management”, paper presented at the 13th International Product Development Management Conference, Milan, 11-13 June. Warlop, L., Ratneshwar, S. and van Osselaer, S.M.J. (2006), “On the role of distinctive brand cues in learning product quality from consumption experience”, available at: www. econ.kuleuven.ac.be (accessed 1 November 2006). Zhang, L. and Shen, W. (1999), “Sensory evaluation of commercial truck interiors”, SAE Transactions, Vol. 108.

Marcus Abbott is a product manager within the automotive industry. For the past 14 years he has practised design, engineering and engineering management at a number of international automotive companies and in a number of countries, including, the UK, the USA and Germany. Previously he received a Master’s degree in industrial design from the Royal College of Art, London. He is currently conducting research into product attribute management at Brunel University and has recently presented at Design and Emotion 2006, International Automotive Interior 2006, Automotive Comfort 2007 and International Association of Societies of Design Research 2007. Marcus Abbott is the corresponding author and can be contacted at: [email protected] John P. Shackleton is Senior Lecturer in Product Development at the Institute of Technology and Engineering, Massey University, New Zealand. After graduating in Aeronautical Engineering from Salford University, he spent several years in industry before pursuing his interests in design and gaining a Master’s degree in industrial design from Teesside Polytechnic. This led to a Japanese government scholarship to undertake research in Japan, and he obtained his PhD in industrial design science from Chiba University. His research work in Japan included a number of collaborations with industry, including Mitsubishi Cars and Sony Creative Design, and he was twice awarded the annual Japanese Society for the Science of Design prize for research. His research interests now centre on modelling the relationship between consumer perceptions and engineered product attributes. Ray Holland directs the Master’s design and branding strategy programmes at Brunel University, UK. These are multi-disciplinary design management courses attracting students from all over the world. His personal research reflects his conviction that design can find its direction through systems thinking and human/cultural issues. He was one of the pioneers of design management education and assists many overseas universities to develop Master’s and PhD programmes in design and branding. Ray also advises on design management-led courses and design research as Visiting Professor to the University of Glamorgan and is an external examiner for leading UK university design schools. He spent his early career as a company lawyer and accountant until he found his home in design. As a consultant, Ray has helped to implement large computer-based information systems and design-led change management programmes.

Executive summary and implications for managers and executives This summary has been provided to allow managers and executives a rapid appreciation of the content of the article. Those with a particular interest in the topic covered may then read the article in toto to take advantage of the more comprehensive description of the research undertaken and its results to get the full benefit of the material present. Getting in touch with emotions, for a purpose The case for developing the emotional attachments consumers feel for brands is overwhelming. This journal has been at the forefront of publishing the requisite research. In 233

Measuring the brand category through semantic differentiation

Journal of Product & Brand Management

Marcus Abbott, John P. Shackleton and Ray Holland

Volume 17 · Number 4 · 2008 · 223 –234

most aspects of brand management the debate is wholly accepted and much of the focus is on tools, techniques and best practice. Arguably an area where older logic applies is in the field of new product development, where a product features and benefits mindset looms large – the “best” new products being those with the “best” new features to meet customer wants, needs and desires, expressed in terms of customer benefits. In using the semantic differentiation research technique to explore brand categories, Abbott, Shackleton and Holland of Brunel and Massey Universities in the UK and New Zealand have done marketers a considerable favour in moving the debate on a step. The technique uses linguistics to examine brands through a series of descriptions. It provides for the notion of “semantic space” which brands can inhabit. It is about emotions, and understanding brand concepts from that perspective. The authors propose this approach as an alternative, but rather like alternative medicine it could become bigger than the mainstream. So many brands are for non-utilitarian products, at least as they impact upon developed societies and the wealthier segments in the developing world. In these categories, brand attributes play a very significant role in the purchasing decision making process – these are emotive decisions. As a result, brand accuracy and differentiation, as measured in the properties of the product, matter. They matter particularly in those luxury goods where the brand is highly conceptual and refers to specific aspirations and life-styles. These characteristics are the ones that can best be depicted through the “brand as category” approach. It is an approach that draws upon much older philosophical traditions. What has been philosophized over centuries has been, in part, missed in a management paradigm still struggling to shake off some of the constraints of pure logic and reason.

based upon the utility of the product. In reality, that is lots of places. The researchers chose a brand within the luxury vehicle market to test their approach. It is a sector in which emotional brand values are heavily relied upon by manufacturers. Within this context they found the method to be quick, reliable and sensitive. It is the concept of the brand’s “semantic space” that they found to be so important, and therefore one that brand managers would do well to explore for their own brand. Not included, but identified as further enriching, is to conduct reciprocal competitor categorization studies to understand what space your brand occupies relative to the competition. From the study conducted, the benefits of the process to future brand propositions, brand extensions or other product propositions are highlighted. In short, this is a cognitive process worth thinking about. Implications in brief In summarizing the implications of this research: . The brand is a key differentiator in many market segments. Given this, marketers, designers and engineers need to consider what is “right for the brand” as well as what is “good” in developing new products. . Understanding what is right for the brand, first needs understanding of the brand category. Within this, attributes can be understood which can be managed through the specification of products to better fit the brand proposition than could be managed by a mere logical, “good product” understanding. . The semantic differentiation technique can be used to understand the “semantic space” of an observed brand and also its competitors. It provides the basis for understanding how customers will perceive brand extensions relative to parent brands, and product development decisions can be taken on this customerfocused basis. . The semantic differentiation approach enables brand managers to get new products “right” for the brand they carry, maximizing efficiencies in design and engineering, and preventing confusion among customers in the marketplace. It enables them to make the most of opportunities to strengthen the brand position, while understanding better the relative merits of competitors.

Thinking about it The majority of research to date has examined the brand as an element of a product within a product category. Abbott, Shackleton and Holland have built upon a smaller but growing body of work that views the brand as being a category of products sharing “necessary and sufficient” conditions for “membership” of a category. It may sound like a small shift in thinking – but it is not. Used diligently, semantic differentiation, within a context of new thinking, could lead to greater success in new product introductions, as well as broader brand management. The place to start would seem to be at the luxury end, or certainly where the importance of the brand is not specifically

It is a process that is built upon strong research foundations, but will be new to many. (A pre´cis of the article “Measuring the brand category through semantic differentiation”. Supplied by Marketing Consultants for Emerald.)

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

234