Font appropriateness and brand choice

https://doi.org/10.1016/S0148-2963(02)00487-3Get rights and content

Abstract

This paper illustrates how font, viewed as a component of a brand's visual equity, can enhance a brand's identity and build its market share. In a two-alternative forced-choice task, brands, covering 10 different product categories, were chosen twice as frequently when they were in an appropriate font relative to when they were not. This finding was replicated with no diminution of the effect even when brand names were highly connotative. In a subsequent small field study, consumers chose chocolates from a box of chocolates having an appropriate font rather than one having an inappropriate font on 75% of occasions. Our study establishes the importance for marketers to adopt a coherent font policy to cover current and possible future brand extensions.

Introduction

Although the most obvious aspect of a brand's projection is its name (Hart, 1998), other aspects also merit attention: collectively, shape, symbol, color, and lettering contribute to what has been called visual equity (Lightfoot and Gerstman, 1998). Visual equity is the value derived from ‘visual form’, i.e. the ‘look and feel’ of the brand. It contributes to brand awareness (e.g. FedEx's purple and chrimson lettering), to brand image (Keller, 1998), and thus ultimately to sales. Kohli and LaBahn (1997) sum it up neatly for the naming process: “a carefully created and chosen name can bring inherent and immediate value to the brand” (i.e. without the need for prior conditioning). Likewise, visual form should bring value to the brand.

Many elements make up the visual form: pictogram, color, letter font, etc. In this paper, we explore the value of font, where ‘value’ is operationalized as an increased likelihood of the brand being chosen.

Three-quarters of a century ago, Poffenberger and Franken (1923) asked people to rank order the appropriateness of 29 fonts for each of five “commodities” (automobiles, building materials, coffee, jewelry, perfume) and five “abstract qualities” (cheapness, dignity, economy, luxury, strength). They concluded that “differing type faces do vary in appropriateness and that judges are able to ‘feel’ this appropriateness or lack of appropriateness” (p. 328). Grouping products together that shared appropriateness with a given font type, automobiles, building material, and coffee clustered strongly together, while jewelry and perfume formed a second tight cluster. The former cluster was most appropriately represented by fonts that were emboldened, simple and easy-to-read (e.g. Cheltenham Bold, Century Bold). These in turn were associated with the qualities of “cheapness,” “economy,” and “strength.” The latter cluster was most appropriately represented by fonts that were italicized, scripted, ornate (e.g. Caslon Old Style Italic; Typo Slope), and these in turn were associated with the qualities of “luxury” and “dignity.”

The finding that fonts differ in appropriateness for products was replicated by Davis and Smith (1933), and by Schiller (1935)—the latter a close replication in that many of the same qualities and products were used. Schiller found that in the dozen or so years intervening between her study and Poffenberger and Franken's, coffee and automobiles were no longer perceived as merely commodity-like products, but had begun to take on elements of the luxury products perfume and jewelry.

In subsequent studies, fonts have been found to differ in their perceived appropriateness to represent not just different kinds of products, but also different kinds of books (Ovink, 1938), newspaper stories (Haskins, 1958), and professions (Walker et al., 1986). A parallel research stream relates and distinguishes fonts by rating scale descriptors Poffenberger and Franken, 1923, Schiller, 1935, Davis and Smith, 1933, Ovink, 1938, Tannenbaum et al., 1964, Bartram, 1982, Rowe, 1982. In studies conducted in and after the 1960s, Osgood's semantic differential (Osgood et al., 1957) has been frequently used to select rating scale descriptors. So, for instance, italics has been found to be associated with Osgood's activity dimension, whereas bold was associated with Osgood's potency dimension.

Interestingly enough, Poffenberger and Franken (1923) took a pessimistic view of the behavioral consequences of their findings, believing that the differences in appropriateness which they observed would have little import: “…it might be argued that such differences as do exist are too slight to warrant consideration for practical purposes” (p. 328). Our work will show otherwise.

One of the best designed studies in the area, and the one with the most relevant theoretical implications for this paper, was conducted by Lewis and Walker (1989). They theorized that image (of which font is an example) and word offer parallel routes to meaning. (In an analogous way, someone's intonation offers meaning, over and above the actual words spoken.) They pretested a number of fonts to find a pair, one of which connoted “heavy” and the other “light.” People had to press the left key if words “heavy” or “slow” were presented, and the right key if “fast” or “light” were presented. People's response times were faster when the font was appropriate for the word being presented, e.g. “heavy” presented in the font that connoted heaviness.

A second experiment showed that response times were again affected even when the quality (e.g. “fast”) was implicit in an animal name (e.g. “cheetah”) rather than explicitly presented as an adjective/adverb, as in their first experiment. Typical contrasts used were: cheetah (fast) versus tortoise (slow); and elephant (heavy) versus hedgehog (light). Part of the semantic representation of cheetah is “fast”: and according to people pretested before the main experiment, part of the semantic representation of Palatino Italic is also “fast,” while part of the meaning of Cooper Black is “slow.” When meanings are in consonance, response times are speeded up; when they conflict, response times are slowed down. These experiments proved that there are behavioral consequences to font–word pairings. But equally important to take away from this study is the idea that fonts generate their own connotative meaning, which is processed independently of the meaning generated by the word per se.

Of more immediate relevance is the work of Perfect and his colleagues. Perfect and Askew (1994) showed that prior exposure to printed ads may induce people to rate them more favorably on a second presentation, even if they are unable to consciously recognize the ad. In other words, they have demonstrated the now well-established phenomenon of implicit memory Jacoby and Witherspoon, 1982, Graf and Masson, 1993 applied to the recollection of ads. They argued that exposure to the ad leads to increased perceptual fluency on second presentation (i.e. they just see more quickly what is in the ad and what it is about). “Were it tested appropriately, this increased fluency would be expressed as quicker response time on a perceptual identification task” (Perfect and Heatherley, 1997, p. 803). However, under certain circumstances, people may misattribute their own fluency not to prior exposure, but to features inherent in the ad: “…as a result [the ads] are rated as being more memorable, more likeable, more distinctive, and more eye-catching.” (The misattribution theory of implicit memory is due to Jacoby et al., 1989).

A follow-up study (Perfect and Heatherley, 1997) attempted to pinpoint the locus of the effect, using the principle that the greater the perceptual overlap between the first and the second presentation, the greater the perceptual fluency, the greater the misattributed affect (emotional ‘charge’) generated. They hypothesized that affect would be greatest when the second presentation was the ad itself (which included the company logo), and the least when the company name appeared in a standard font (Times New Roman). It would be somewhere in between when the company name (logo) appeared in its usual font, but in the absence of the ad. Although they did replicate their previous study, they found only weakly supportive evidence for the more specific predictions.

Returning to Lewis and Walker's study, congruence between font and name leads to perceptual fluency (response times were faster), which we hypothesize from Perfect's work, should lead to positive affect, which should lead to an increased likelihood to investigate the product further, or even choose it. Satisficing behavior implies that to have your product investigated first will improve its chances of being eventually chosen. Our experimental hypotheses are therefore:

H1

Products presented in an appropriate (consonant) font will be investigated further more frequently than when presented in an inappropriate font.

H2

Products presented in an appropriate (consonant) font will be chosen more frequently than when presented in an inappropriate font.

Note that in this case perceptual fluency is not derived from prior exposure, but from two systems in the brain (image processing and verbal processing) producing congruent meanings. The two etiologies of fluency mirror the two methods described above, by which visual forms may drive visual equity—either by prior exposure (cf. Perfect and Askew), or by carefully choosing the right form to give “inherent and immediate value to the brand” (cf. Lewis and Walker).

Section snippets

Pretest of font appropriateness

Our first task was to identify a pool of fonts appropriate for a pool of products. The 2×2 FCB grid (Vaughn, 1980) distinguishes products by whether they are ‘thinking’ or ‘feeling’ products, and whether they are high or low involvement products. It was used as a sampling frame to select 32 products, eight from each quadrant of the grid. Next, 27 fonts were chosen from those bundled with Microsoft packages, supplemented by some freeware/shareware obtained via the Internet, covering traditional

Pretest of names and fonts

Four product categories were initially chosen and several dozen names were generated by the experimenters. The two product categories where the experimenters felt they had been most productive in generating names were then retained (box of chocolates and bottled water). On the basis of advice from three other people, six names were retained for each of these products. They were, for chocolates: Enigma, Forbidden, Indulgence, Intrigue, Mystery, Temptation; and for bottled water: Aqua-Vitalis,

Results

Combining responses from Questions 1 and 2, it is possible to infer what percentage of people each respondent thought would choose the first alternative. This was used as the dependent measure in an analysis of variance. Three between-subject factors were: first alternative as name A or B (name), first alternative in font X or Y (font), and which of the two products appeared on the top half of the page (order).

For bottled water, the only significant result was for font: F(1,73)=28.42, P<.0001, η

Field test

It must be admitted that many of the choices offered respondents in Experiments 1 and 2 are one step removed from actual choice situations. To remedy this shortcoming, a small field study was undertaken using the same names and fonts that were used for chocolates in Experiment 2. Round, metallic silver cardboard boxes (4 in. diameter, 2.25 in. in height) were used to contain chocolate truffles in a choice task. The box tops were decorated with silver string tied up in a shoelace bow, and a

Summary and discussion

We have shown that the aptness of a font for a particular type of product does have consequences. In Experiment 1, brands presented in appropriate fonts were chosen more often than brands presented in inappropriate fonts (averaged over the 10 product categories, they were chosen in the ratio 2:1). One limitation of that study was that the brand names were connotatively neutral for the respondents. Highly connotative names, it might be argued, might overwhelm the font effect. In Experiment 2,

References (22)

  • D Bartram

    The perception of semantic quality in type: differences between designers and non-designers

    Inf. Design J.

    (1982)
  • R.C Davis et al.

    Determinants of feeling tone in type faces

    J. Appl. Psychol.

    (1933)
  • P Graf et al.

    Implicit memory: new directions in cognition, development, and neuropsychology

    (1993)
  • S Hart

    Developing new brand names

  • J.B Haskins

    Testing the suitability of typefaces for editorial subject-matter

    Journal. Q.

    (1958)
  • L.L Jacoby et al.

    Remembering without awareness

    Can. J. Psychol.

    (1982)
  • L.L Jacoby et al.

    Memory attributions

  • K.L Keller

    Building, measuring, and managing brand equity

    (1998)
  • C Kohli et al.

    Creating effective brand names: a study of the naming process

    J. Advert. Res.

    (1997)
  • C Lewis et al.

    Typographic influences on reading

    Br. J. Psychol.

    (1989)
  • C Lightfoot et al.

    Brand packaging

  • Cited by (91)

    View all citing articles on Scopus
    View full text