On the role of creativity in the formation of new complex words

2.1 Creativity

When discussing creativity, many eminent historical figures – ‘creative geniuses’ (Simonton 2019a) – such as Shakespeare, Picasso, Beethoven, or Newton come to mind with their remarkable contributions. These instantiations represent the so-called ‘Big C’ Creativity reserved for extraordinary products made by non-conventional people in their respective fields. Nevertheless, this is only the tip of the iceberg, and many more nuanced forms of creativity can be identified in the everyday activities of ordinary people (Cotter et al. 2019; Kaufman and Beghetto 2009).

Various approaches and methods can be traced in order to understand and assess these forms of creativity (cf. Kaufman and Beghetto 2009; Kozbelt et al. 2010; Plucker et al. 2019). Among them, the Torrance test of creative thinking (TTCT) occupies a prominent position, as, according to Kim (2006), it is considered as the most widely used and referenced test of creativity, and was created by Torrance with the aim to provide “a reliable and valid test of creative thinking abilities that could be administered to individuals from kindergarten through adulthood” (Runco et al. 2010b: 362).

According to a recent integrative bio-psycho-behavioral model (Jauk 2019), three hierarchical levels can be delineated when describing creativity. Real-life creative behavior is situated at the most visible level. This surface level is influenced by a deeper level, comprising psychological constructs related especially to abilities and personality. The middle level is further determined by deeper neuro-biological systems such as the default mode network, the executive control, and the dopaminergic system. Although the neural level is beyond the scope of the present study, the two higher levels are crucial here as, according to Jauk (2019), the cognitive creative potential, in terms of the divergent thinking ability, occupies a prominent place at the center of the middle level. In fact, divergent thinking has been a synonym for creativity for many years, and its role in creative potential and creative problem solving is still considered as crucial (Glăveanu and Kaufman 2019; Jauk 2019; Runco and Acar 2019; Runco et al. 2010a).

Divergent thinking can be conceptualized as a “measure of ideation that fuels creative thinking” (Runco and Acar 2019: 244), and, in comparison to convergent thinking, it captures various directions of thoughts, which are necessary in situations where more than one correct answer exists. Torrance based his notion of creative potential on Guilford’s theory (Guilford 1956, 1986; see also; Cramond et al. 2005; Plucker et al. 2019; Runco et al. 2010b). Accordingly, four main scores, namely Originality, Elaboration, Fluency, and Flexibility, and one additional score, Creative Strengths, can be distinguished. Although various types of the TTCT can be traced depending on the specific version (cf. Torrance 1966, 1974, 1987, 1990, 1998) and the form (figural vs. verbal form), we used the most recent official Slovak version (Jurčová and Szobiová 2008) to find out whether the creative potential of an individual, as captured by various TTCT scores, relates to word formation as an act of creativity by a language user.

2.2 Creativity in word formation

The notion of word-formation ‘creativity’ has traditionally been connected with deviation from the established rules, with a non-rule-governed process. Marchand, as early as in his 1960 groundwork of English word formation, introduced the concept of ‘word-manufacture’, in which “[m]ore or less arbitrary parts of words may be welded into an artificial new word” (1960: 368) and therefore do not rely on any productive morphological process. Conversely, Bauer (1983: 63) emphasizes the factor of motivation, and defines creativity as “the native speaker’s ability to extend the language system in a motivated, but unpredictable (non-rule-governed) way.” Similarly, Renouf (2007: 70) maintains that “[c]reativity is typically thought of as the act or quality of an unpredictable departure from the rules of regular word formation.” Creativity, therefore, denotes cases in which “the (nonce) coined word obviously transgresses the morphological system” (Dal and Namer 2018: 224).

The term ‘creativity’ acquired a broader scope within various treatments of lexical creativity (see, for example, the contributions in Arndt-Lappe et al. [2018] and Munat [2007b]). In this context, it is first and foremost related to nonce-formations like blends, clippings, acronyms, initialisms, truncations and phrasal lexical items, but also to the metaphorical and metonymical shifts of existing words. These lexical items manifest various degrees of deviation from the expected patterns and “different degrees of ‘noteworthiness’” (Hohenhaus 2007: 16). They mostly serve as attention-seeking devices, means of humor, playfulness, ludicity, puns, wordplay, etc., and are therefore, as pointed out by Dal and Namer (2018), mostly shifted to the area of performance and pragmatics.

The extensive scope of the lexical creativity agenda is aptly expressed by López Rúa (2010), who maintains that

[l]anguage is thus used to break away from the norm, and also from what is foreseeable or even politically correct. In morphological terms, this is reflected in the manipulation of morphological rules, which results in the creative or deviant use of word-formation devices, such as affixation (Preprophecy), conversion (Damnwells), compounding (The Lovemongers), or blending (The Beatscuits). Moreover, the fragile correspondence between an orthographic word, a phonological word and a lexeme is constantly challenged by the creative use of graphemes and punctuation, word play, or semantically anomalous word combinations. (López Rúa 2010: 51)

This view is in accordance with Bauer’s (2001) delimitation of the scope of creative coinages: the formation of simple words (complex words are, in his view, the matter of productivity), the figurative extension of existing words, and formations produced by isolated individuals or by the extension of non-productive patterns.

In fact, this understanding of creativity in word formation is closely linked to the long-discussed problem of the relation between creativity and productivity. On the one hand, there are views that identify creativity with productivity (e.g., Chomsky 1964, 1965; Veale 2007). Opposed to this are views that strictly separate these two phenomena and consider linguistic creativity exclusively as a deviation from productive rules, or, in other words, while productivity is rule governed and therefore automatic and unintentional, creativity is non-rule-governed and therefore intentional (Bergs 2019; Botha 1968; Fernández-Domínguez 2010; Lieber 2010; Ronneberger-Sibold 2008; Schultink 1961). Finally, there are authors who treat the relation between linguistic creativity and productivity as a continuum (Chung 2008; Ladányi 2000; Mattiello 2018; Munat 2007a). In principle, this means that creative coinages “tend to be at least rule-related, up to the point where a distinction from ‘governed by productive rules’ becomes problematic” and therefore “it is preferable to consider the difference between creativity and productivity as a cline” (Hohenhaus 2007: 16).

Our comprehension and use of the notion of creativity in word formation is different. Creativity is generally understood as a universal feature of human beings, as their creative potential (Kampylis and Valtanen 2010). Since creative potential is implemented through creative performance (Runco et al. 2010a: 343), we understand word-formation creativity as one of many areas of creative performance, as a manifestation of this creative potential in coining new complex words. We understand it as the ability of any and all language speakers to form a new complex word in response to the specific need of a speech community to give a name to a new object of extralinguistic reality or a new name to an already named object. It is assumed that every act of naming is a creative act that employs a language speaker’s cognitive abilities in order to select and employ one of a number of possible naming strategies. The creativity of word formation in this sense is manifested at each level of the naming process, i.e., at the conceptual level, the onomasiological level and the onomatological level. This fact also makes the onomasiological approach an advantageous tool for the examination of various word-formation strategies employed by different (groups of) language users in terms of the opposing tendencies to semantic transparency and economy of expression.

Our first attempt at treating each act of word formation as an act of creativity by a language user rather than an automatic process was more than a decade ago (see Štekauer et al. 2005). The notion of creativity lays emphasis on the active role of language users in coining new words by reflecting the fact that, in each act of naming, there is considerable space for a coiner’s individual preferences and for a selection from several options. Since language users have unequal experiences, knowledge, intellectual capacities, imaginations, education, ages, professional interests, and so on, one can expect considerable variation in their approaches to the naming task.^[1] These differences can be advantageously observed by examining the way(s) in which language users solve the conflict between the tendency towards economy of expression, which is preferred by a speaker (who is able to say more within a given time period), and the tendency towards semantic transparency, which is preferred by a listener (who is able to understand new complex words faster and more easily). These contradictory preferences were observed as early as the start of the 20th century by von der Gabelentz (1901: 181–185), who used the terms Bequemlichkeit (preferred by the speaker) and Deutlichkeit (preferred by the listener).^[2]

The concept of semantic transparency plays a crucial role in research into the interpretation of compound words, especially Noun + Noun compounds. In psycholinguistic literature, it is usually understood as scalar relation between the meaning of a compound word and its constituents, in particular as a degree of correspondence of the meaning of a compound word with the meanings of its motivating constituents (e.g., Bell and Schäfer 2016; Gagné and Spalding 2009; Ji et al. 2011; Libben 1998; Schäfer 2018: 1). Libben (1998) distinguishes two types of semantic transparency. The first, ‘constituency’, pertains to the use of morphemes in their original/shifted meaning (in shoehorn, shoe is transparent because it is used in its original meaning, while horn is opaque). The second, ‘componentiality’, bears on the meaning of a compound as a whole: for example, bighorn is non-componential because the meaning of this word cannot be inferred from the meanings of its constituents, even if these are related to independent morphemes.

Psycholinguists differ, however, in their views of the importance of the head and modifier, respectively, in their contribution to the interpretation of compounds as a whole. This is reflected in asymmetric models that assign a higher value to one of the constituents in the process of interpretation. Thus, the role of the modifier is emphasized by Gagné and Shoben (1997), Gagné (2001), Spalding and Gagné (2008)and Spalding et al. (2010) in their CARIN model, even if later, in the RICE model, Gagné and Spalding (2014: 103) maintain that the information associated with both head and modifier is crucial to the evaluation phase of interpretation. In addition, as pointed out by El-Bialy et al. (2013), the processing of a compound also depends on whether the compound constituents have similar semantic transparency, i.e., whether the transparency of the first and the second constituents match.

Another factor that plays a role in the semantic transparency of complex words is the semantic relations between the modifier and the head, the relations that various morphologists and psycholinguists have attempted to capture since Lees (1960). The most influential system appeared to be Levi’s system of Recoverably Deletable Predicates (1970), even though it came to be the target of extensive criticism. What matters more with regard to the interpretation of the most common object of meaning-interpretation studies, i.e., N+ N compounds, is the competition between numerous potential semantic relations between the head and the noun of each such compound as a result of the absence of any verbal element. The idea of competition came to be the focal point of the above-mentioned CARIN and RICE models, and it also underlies Štekauer’s theory of meaning predictability (2005a). Nevertheless, these are issues that go beyond the scope of the present research, which explores the influence of creative potential upon word-formation creativity rather than meaning interpretation.^[3]

The conflict between the tendency towards semantic transparency and the tendency towards economy of expression can be analyzed and evaluated by means of an onomasiological theory of word formation (cf. Körtvélyessy 2010; Körtvélyessy et al. 2015; Štekauer 1998, 2005b, 2016). This theory accounts for the way in which new complex words are formed. It reflects a crucial triad of relations between extralinguistic reality (object to be named), a speech community (represented by a ‘coiner’), and the word-formation system, thus emphasizing the fact that each act of naming responds to a specific naming demand on the part of a member (or members) of a speech community. It lays emphasis on the active role of language users in the process of giving names to objects. The naming act is not a purely linguistic act. New complex words do not come into existence in isolation from factors such as human knowledge and experiences, human cognitive abilities, human imagination and creativity, etc. This position is in accordance with Koch’s idea that the onomasiological viewpoint is closer to that of the speaker as a linguistic innovator than the semasiological viewpoint (2001: 17). By implication, the naming act is a cognitively founded phenomenon that starts with a conceptual analysis of a class of objects to be named. This analysis captures the prototypical features of the class of objects by means of logical predicates (noemes). Some of these logical predicates, represented by corresponding semantic categories like Agent, Action, Instrument, Result, Location, Time, Diminutive, Causative, etc., are chosen to ‘motivate’ the future complex word at the onomasiological level, thus constituting an onomasiological structure. In accordance with Dokulil (1962, 1997), we understand the onomasiological structure as involving different forms of structuring the concept in view of its expression in the given language. The onomasiological structure is subsequently represented by morphemes of a language at the onomatological (morphemic) level. In other words, onomasiology establishes the cognitive basis for the naming act by identifying the semantic categories and the relations between them. Onomatology ‘cares’ for the representation of these semantic categories by morphemes of a language by means of the Morpheme-to-Seme-Assignment Principle (Štekauer 1998, 2005b). Then, the onomasiological type reflects the interrelation between the onomasiological and the onomatological (morphemic) levels.

In particular, the structure of the onomasiological level is prototypically constituted by three constituents: the base, the determining mark, and the determined mark. The phenomenon to be named is first classed with a certain conceptual class and functions as the onomasiological base. Within the limits of this class, its scope is narrowed down by the determining mark. In the prototypical ternary structure, these two constituents are related by the determined mark that always stands for the cognitive category of action in one of its three modifications (Action proper, Process and State).^[4] Any of the semantic categories representing the onomasiological structure can but does not have to be represented at the onomatological level. The possibility (not) to represent individual semantic categories by corresponding morphemes determines the form of a new complex word and establishes various onomasiological types. As a result, they reflect different degrees of transparency/economy (see Figure 1).

Figure 1:

Scales of transparency and economy (OTs that occur in our research).

On the omission of OT7 and OT8, see below.

In the following overview of onomasiological types, OT stands for ‘onomasiological type’, DingM for ‘determining mark’, DedM for ‘determined mark’, and R for ‘morphemically represented’.

Only two constituents are represented by morphemes in (2) and (3):

OT4 ranks among the most economical of all onomasiological types because a ternary structure is represented by a single morpheme. It is based on the action-to-substance recategorization:

(4)

OT5 employs the same feature as OT4, i.e., a joint representation of the base and the action category. However, unlike OT4, the determining constituent of the onomasiological structure is represented by a morpheme. For illustration, this type is exemplified with a complex word for ‘a person who cheats tourists’ in (5).

(5)

OT6 is another extremely economical onomasiological type. Neither the base nor the determined mark is expressed, as is the case with exocentric compounds like redskin, i.e., ‘a person with a red skin’ (referring offensively to American Indians). The absence of the base and the determined constituent of the mark at the morphemic level can only be explained by an effort of a coiner with the maximum possible economy of expression. By implication, the semantic transparency of this onomasiological type is poor. Anything can have red skin, as evidenced by another meaning of this word, i.e., ‘a potato that has red skin’.

In OT7, the mark cannot be structured into determining and determined parts, which yields a binary onomasiological structure, comprising a mark and a base. Both base and mark are morphemically represented.^[5][6]

In OT8, the mark of a binary onomasiological structure is not expressed. This can be illustrated with an example of a substance-to-action conversion, such as ‘to bridge (something)’.

In OT9, both mark constituents are expressed with the meaning ‘a person smiling on a billboard’. This type develops OT6, which is restricted to the expression of the determining mark:

It follows from these examples that the differences between the extent of the morphemic representation of the ternary or binary onomasiological structure reflect the competition between the above-mentioned contradictory tendencies and make it possible to determine which way of forming new complex words is preferred by various groups of speakers distinguished by age, profession, education, psychological concept of creativity, etc. Figure 1 illustrates the relations between semantic transparency and economy of expression:

The reasoning that underlies Figure 1 builds on the crucial role of the morphemic representation of the cognitive category action in the concept of onomasiological type. Similar to the verb which, with its valency, is at the core of sentence semantics, the significance of the category action stems from its capacity to semantically relate the other two constituents of the onomasiological structure. Without a morphemic representation of this category, the relation between the polar constituents remains vague owing to the usual diversity of possible interpretations. Its crucial position is also reflected in various morphological and psycholinguistic models of word-interpretation of (primarily) Noun + Noun compounds which are aimed at the identification of the semantic relation between two nominal constituents. The best-known attempt is Levi’s (1978) system of Recoverably Deletable Predicates.^[7]

The system of onomasiological types that ranges over all word-formation processes (prefixation, suffixation, compounding, conversion, etc.) is advantageous for the evaluation of the discussed contradictory tendencies in word formation because it makes it possible to capture various naming strategies that manifest different degrees of preferences for semantic transparency vs. economy. Taking the above-outlined onomasiological theory of word formation as a theoretical framework, four basic criteria can be identified for the classification of individual onomasiological types in terms of semantic transparency and economy of expression:

1. A ternary vs. binary onomasiological structure

2. The number of constituents of the onomasiological structure represented by morphemes at the onomatological structure

3. Constituents of the onomasiological structure that are/are not represented by a morpheme:

   • – Base

   • – Determining mark

   • – Determined mark

4. A shared representation of a single constituent of an onomasiological structure by one common morpheme

Based on these principles, it is assumed that:

1. The transparency of OT1 is the maximum possible because each semantic constituent of the ternary structure is morphemically represented.

2. OT2 is more transparent than OT3 thanks to the morphemic representation of the semantic category of action. The same is true of OT4.

3. OT5 is more transparent than OT4 because, apart from the base and the action being merged, it is also represented by a morpheme standing for the determining mark.

4. OT2 is more transparent than OT4 because both the base and the action are represented by morphemes.

5. OT5 is more transparent than OT2. Although its action and base are represented by a common morpheme, in contrast to OT2, the determining mark is also represented by a morpheme.

6. OT9 is more transparent than OT3 because while it lacks a morpheme for the base, the action and the determined mark are expressed.

7. OT9 is less transparent than OT4, which contains morphemes for the action as well as for the base.

8. OT6 is the least transparent because it only contains the determining element of the mark.

It should be noted that the scales in Figure 1 only represent those onomasiological types that occurred in the respondents’ answers to the specific tasks of the word-formation test. The absence of OT7 and OT8 is understandable because the scope of the naming tasks in the test was restricted to Agent and Patient names. As will be seen in the evaluation of the data obtained, the uses of individual onomasiological types for these kinds of named objects vary considerably.

3.1 Torrance test of creative thinking (figural form)

The TTCT, as an indicator of creative potential, consists of various scores (subscales). These are assessed across three tasks. In every task, participants are encouraged to create something special. Ten minutes are provided to accomplish every task.

In the first task, named Picture construction, a colored sticker in the shape of a jellybean and a blank sheet of paper are provided. The goal is to stick the sticker on the paper to create a picture incorporating the sticker and to name it. The stimulus material and instructions are illustrated below (see Figure 2).

Figure 2:

Sheet of a paper with a sticker as an illustration of the stimulus material for the first task.

Task 1. In front of you, there is a rounded piece of colored paper. Think of a picture of an object that you can draw when you use this piece as a part of the overall drawing. Then, draw lines to create the intended image. Try to think of a picture that you assume no one else would invent. Add elements to your sketch so that the image you create expresses a most interesting phenomenon. When you finish drawing, think of a name for your image and write it at the bottom of the page. Think of a very accurate and unusual name to help explain your drawing.

In the second task, named Picture completion, the goal is to complete 10 incomplete figures and to name them. The stimulus material and instructions for the task are illustrated below (see Figure 3).

Figure 3:

Sheet of a paper with incomplete figures as an illustration of the stimulus material for the second task.

Task 2. By adding lines to the incomplete figure on the next page, you can turn the sketched shapes into interesting objects or pictures. Again, try to think of an object or picture that no one else would invent. Complete and expand your first idea and try to add lines to create the most complete and interesting image. Think of an interesting title for your drawing and write it at the bottom of the frame next to the image number.

In the third task, called Circles, the goal is to complete the circles and to create a picture (or pictures). The circles are illustrated below (see Figure 4).

Figure 4:

Sheet of a paper with circles as an illustration of the stimulus material for the third task.

Task 3. In the next 10 min, see how many objects and pictures you can create from the circles on this and the next pages. Circles should be the main part of everything you create. Draw a line to the circles to create an image. Try to figure out things that no one else could come up with. You can draw between, inside and outside the circles. Create as many different pictures as you can and put as many ideas into each as possible. Try to make them show the most interesting scene. Write names under the pictures.

The above-mentioned tasks were used to assess various aspects of divergent thinking. In particular, scores for Creativity, Fluency, Flexibility, and Elaboration can be derived from the tasks. Moreover, one additional score, called Creative Strengths, is assessed.

Originality captures how unique the creative product is in comparison to the existing standards. Statistically speaking, the less frequent the answer is in the normative sample, the higher the originality score is. For illustration purposes, we will use the fourth drawing of the second task (Picture completion). The task consists of completing two horizontal lines. Although there are numerous possibilities for completing the drawing, we will mention two hypothetical scenarios. As illustrated in Figure 5, in the A variant (the second from the left), the completed picture depicts ‘a rail’. In contrast, in the B variant (the third picture from the left), the completed picture depicts ‘a belt’. Since B is much less common according to the standards specified in the manual (less than 1% in a normative sample in comparison to 3–5% for the second answer), the originality score for B is higher (3 points) in comparison to A (1 point). Note, however, that even the second answer is relatively original. For an answer occurring in more than 5% of all answers, drawings such as ‘a road’ will be considered as non-original, and 0 points will be given to the participants.

Figure 5:

Illustration of (A) less vs. (B) more creative drawings.

Fluency captures the number of relevant answers provided by a participant. The higher the number of completed pictures, the higher the fluency score. For example, Figure 6 illustrates two situations that can hypothetically occur in the second task (Picture completion). In the upper line (A), (only) four legitimate answers are provided. Therefore, the score is 4. In comparison, the lower line (B) provides 10 different answers, and therefore a score of 10 is given to the participant.

Figure 6:

Illustration of (A) less vs. (B) more fluent answers.

Flexibility captures the number of different categories provided in the pictures. The more diverse the answers are, the higher the flexibility. One point is gained for every new category that is presented in the drawings. We can illustrate this with two hypothetical scenarios in solving Task 3 (Circles). As illustrated in the upper line (A) of Figure 7, the circles can be completed in such a way that every circle belongs to the same category. As all pictures capture emoticons, no points for flexibility will be given to the participant. In comparison, in line B, all three drawings differ regarding their content because the football, the seven-spotted ladybird and the yin-yang symbol represent different categories. As the category shifted twice in example B, 2 points are provided to the participant for their flexibility score.

Figure 7:

Illustration of (A) lower vs. (B) higher flexibility.

Elaboration captures how many details are provided in a drawing beyond what is necessary to understand the drawing. The more details are provided beyond what is necessary to capture the idea, the higher the elaboration score. This can be illustrated by the third picture of the second task (Picture completion). While there is a plethora of possible ways as to how to complete the picture, two hypothetical scenarios are illustrated in Figure 8. In B (the third picture from the left), a lot of additional details can be found, such as a skiing site, trees, a star, and a cabin. In contrast to this more elaborate drawing, the picture of a needle in situation A (the second drawing from the left) does not provide any additional details. Therefore, no points for elaboration are given to the participant for this picture.

Figure 8:

Illustration of (A) lower vs. (B) higher elaboration.

Creative Strengths consist of 13 categories (e.g., fantasy, colorfulness of imagery, humor, emotional expressiveness, and so on). The score increases as the number of these categories grows.

3.2 Word-formation test

The word-formation test was aimed at coining names for Agents and Patients. The reason for this focus consists in the availability of numerous options in this field of word formation, comprising suffixation with a number of suffixes (-er, -ist, -ian, -ee, -eer, -ant/-ent), different types of compounding with various possibilities of how to express the onomasiological base (man, woman, person, etc.) and conversion (as in cheat_N). If we add numerous possibilities for the expression of the determining and the determined marks in every specific act of naming to these, this area of naming offers a number of options that can be captured by the system of established onomasiological rules and evaluated in terms of the examined contradictory tendencies towards transparency or economy.

The test itself consisted of three tasks, each of them consisting of three subtasks. The three tasks were different in their nature: the first one was based on multiple-choice questions, the second on verbal descriptions, and the third on drawings. This structure of the word-formation test complies with the domain-specificity/task-specificity theory of creativity, according to which “the skills, traits, and understandings that underlie creative performance … vary from domain to domain” (Baer 2017: 379). In Task 1, the participants were offered a range of options for naming a person described by some text. The last option indirectly prompted them to come up with a non-rule-governed solution by offering a ‘free choice’. The second set of tasks differed from the first by offering no options. The respondents were asked to propose their own new complex word based on a brief specification of the object to be named. Finally, the third set of three tasks replaced the wording with a drawing of a situation in which an object performs an unusual activity.

Since the test was aimed at the formation of new words, the respondents were expressly asked not to use either existing words in a shifted meaning or descriptive phrases (see Figures 9–11).

Figure 9:

Drawing for Task 3/1.

Figure 10:

Drawing for Task 3/2.

Figure 11:

Drawing for Task 3/3.

Task 1. Choose the word that you think is the most suitable for the person described in the task.

Task 1/1. A person whose smiling face is used for billboard advertisements:

1. smiler

2. smilist

3. smileman

4. smile

5. smile-person

6. smile-face

7. billboard smile-face

8. [free choice]

Task 1/2. A person who frequently interrupts other people when they are talking:

1. interrupter

2. cutter-in

3. cutter-inner

4. interrupt

5. talk-interrupter

6. talk-interrupt

7. interrupt-man

8. [free choice]

Task 1/3. A person who believes in miracles

1. miraclist

2. miracler

3. miracle-man

4. miracle-believer

5. miracle-hoper

6. miracle-hope

7. [free choice]

Task 2. Each question describes a person in an unusual situation. If you had to come up with a name or title for the person, what would it be? You may make up a word or choose a word that already exists in English.

Task 2/1. What would you call someone who does research about spiders’ webs?

Task 2/2. Suppose that aliens were about to land on Earth for the first time. What would you call a person who was supposed to meet them as a representative of the human race?

Task 2/3. Suppose that a woman has a clone made of herself. Then suppose that a man has a clone made of himself. Now suppose that the two clones marry each other and have a child. What would you call the child?

Task 3. Each drawing below shows a person performing an unusual action. If you had to come up with a name or title for the person in each drawing, what would it be? You may invent a word or choose a word that already exists in English.

3.3 Transparency and economy indices

An index that captures the general inclination of a language speaker to form words in a more or less transparent or economical fashion was computed across all word-formation tasks (general index). This index facilitates the statistical evaluation of economy and transparency at a more general level. Moreover, since the individual tasks differed in the stimulus material and their answer format, we computed indices for each of the three word-formation tasks (specific task-related indices). These indices enable us to capture the general inclination of a language speaker to form words in a more or less transparent or economical fashion.

To accomplish this task, firstly, all individual answers in the word-formation test were analyzed and ranked according to their level of semantic transparency and economy of expression based on their OT, as described above in Section 2.2 (maximum to minimum transparency – 1(OT1), 2(OT5), 3(OT2), 4(OT4), 5(OT9), 6(OT3), 7(OT6/OT10); and maximum to minimum economy – 1(OT4/6/10), 2(OT2/3/5/9), 3(OT1)).^[8] For instance, hypothetically, if the answers of a participant in the second word-formation task (where participants had to come up with a name for a person in three unusual situations on the basis of a verbal description) were classified as OT4, OT6 and OT5, from the economy point of view, the values of 1, 1, and 2 were assigned to the participant.

Note, however, that this classification, based on scales ranging from 1 to 7 for semantic transparency and 1 to 3 for economy, is categorical in its nature. Although the rankings emerged naturally, the differences in the hypothetical space between two categories could differ across the scale. In comparison to measures such as size, weight, or temperature, it is not possible to say that the difference between 1 and 2 is the same as that between 2 and 3. Moreover, in our scales, there are no absolute zero points in a strict sense unlike, for instance, measures of size or weight, for which plausible zero values exist. Thus, instead of treating these values at an interval or ratio level, they are considered as ordinal level data. An important implication is that the application of algebraic operators for ordinal level data is more problematic than for an interval/ratio level. Therefore, to provide a summary measure of the center in the numerical data (a measure of central tendency), we preferred to use the median over the mean/average values for these scales. For instance, to provide the measure of the central tendency for the above-mentioned hypothetical situation (1, 1, 2), the median value will be 1 (instead of the mean 1.33).

Consequently, based on the summary median value for three tasks and across the whole word-formation test, we identified general tendencies to form words in a more or less transparent or economical fashion (i) for each of the three word-formation tasks (specific task-related indices), and (ii) for all three word-formation tasks as a whole (general index). These values of the median central tendencies were used for a non-parametric correlation analysis with data from the TTCT (ratio-level data). Note, however, that to bolster the intuitiveness of the interpretation, the scales for transparency and economy were inverted so that a higher score represents a stronger tendency. For instance, after inversion, a value of 7 represents more transparent answers than a value of 1.^[9]

3.4 Statistical analysis

The non-parametric Spearman’s Rank-Order Correlation was used to ascertain the pattern of the relationship between the TTCT scores and the general indices of an individual’s tendency towards transparency or economy in word formation. Spearman’s Rho can be understood as a proportion of the variability accounted for with rank data. Since there are both the ordinal level of indices regarding the word-formation test and a violation of the assumption of normality regarding the TTCT, the non-parametric alternative is preferable to the parametric one (Field 2017).

Null hypothesis significance testing was used as the main interpretational framework. More precisely, due to the exploratory nature of the study, a Fisherian approach to null significance testing was used for interpretation purposes. According to this approach, significant results indicate that either H₀ does not explain the research data or a rare event emerged (Perezgonzalez 2015).

Although null hypothesis significance testing is widely used and has significant merits, this approach also has its limits. For instance, multiple comparisons could increase type one errors (false positives) and a null hypothesis could be rejected incorrectly merely by chance. Consequently, to reflect this inherent limit, we use the Benjamini-Hochberg p-value adjustment aimed at controlling false discovery rates.^[10] Additionally, one of the other disadvantages of the null hypothesis significance testing is that it “does not allow one to distinguish three evidential states of affairs, namely evidence for H0 rather than H1, evidence for H1 rather than H0, or not much evidence either way. By contrast, Bayes factors do allow this three-way distinction” (Dienes and Mclatchie 2018: 215). Therefore, as recommended by many authors (e.g., Dienes and Mclatchie 2018; Halsey 2019), the null hypothesis significant testing was accompanied by additional statistical indicators to further bolster the verisimilitude of the results. For this purpose, the method of triangulation was used, where both the effect size (with 95% credibility intervals) and the Bayes factor are reported and interpreted beyond the null hypothesis significance testing (Halsey 2019).

The Bayes factor can be understood as the extent to which the observed data give support to one hypothesis over another. For instance, let us imagine a situation in which H₁ assumes that the effect is present, and H₀ assumes that there is no effect. An alternative hypothesis (H₁) states that there is a relationship between the two variables in question. The null hypothesis (H₀) postulates that there is no relationship between the two variables. A Bayes factor of magnitude 6 represents the information that, according to the available data, the strength of evidence for an alternative hypothesis is six times greater than for the null hypothesis (the data are six times more probable under H₁ than under H₀). As this study is novel and no prior information was available, the default prior and two-sided tests were used. To facilitate the communication and interpretation of the strength of evidence, Jeffrey’s adjusted benchmarks were used (Wagenmakers et al. 2018).^[11]

The effect size represents the magnitude of the effect and, in the case of correlations, it is manifested by the correlational coefficient itself. This is crucial, as this estimate is, in comparison to the p-value, not determined by the sample size and can confirm the practical significance of the results. Moreover, 95% credibility intervals could be understood as (given the observed data) the effect having a 95% probability of falling within the range of the lower and upper thresholds. This information can help us better understand the uncertainty related to the estimated parameters.^[12] The analysis was conducted by means of the R-based software Jamovi 1.2 with JSQ module.

4.1 Descriptive results

The first part of the analysis provides summarized descriptive information for the variables in question. The mean, median, mode, standard deviation, skewness, kurtosis, and Shapiro-Wilk normality test figures for the TTCT subscores are depicted in Table 1.

The median, mode, skewness, and kurtosis figures for the word-formation test are provided in Tables 2 and 3.

4.2 Inferential results

Before proceeding to a more advanced analysis, Table 4 depicts (for the sake of clarity) the correlation matrix based on the null hypothesis significant testing for relationships among the TTCT scores and the general tendencies to economy and transparency. Moreover, since the word-formation test employed three tasks with different types of instruction and stimulus material (word-related material in Tasks 1 and 2 vs. drawing in Task 3) and different types of answers (multiple-choice in Task 1 vs. open-ended answers in Tasks 2 and 3), the relationship between the TTCT and the specific word-formation task is provided as well.

The following sections provide a more detailed analysis and description of the results.

4.3 Discussion

Even though, from the psycholinguistic and the broader psychological perspectives, both word formation and creativity are important and prolific fields of study on their own, the scarcity of evidence connecting these two fields is evident and motivated the present large-scale interdisciplinary research. In a nutshell, the present work indicates that creative potential, understood as the ability to produce novel and useful ideas, as captured by divergent thinking abilities, seems to be related to the way individual language users coin new words, especially in the broader context of the tendency towards economy of expression or semantic transparency. More specifically, from a general perspective, it has been shown that the Economy and Creative Strengths scores are negatively correlated with the general tendency towards economy of expression. Originality has been shown to be significant as well; nevertheless, there is a possibility that this result is only a false positive and should be considered with a grain of salt.

When analyzing the three tasks separately, it was shown that the present pattern of results for transparency is, logically, complementary to economy of expression. In particular, in the third word-formation task, both Elaboration and Creative Strengths were related to the tendency to coin new words more transparently and less economically. Although in the present study, it is not possible to infer causality, our working hypothesis is that various facets of divergent thinking, such as the tendency to elaborate or not, could be considered as the cognitive underpinning of the way how language speakers coin new complex words and, more specifically, how the creative potential is manifested in the word-formation process. For instance, Elaboration captures the number of details provided and “it is suggested when the individual follows an associative pathway for some distance” (Runco and Acar 2012: 67). The ability to follow these associative pathways is reflected in coining more transparent and less economical complex words. This is important for the theory because some other candidate processes could be identified at a deeper neurobiological level (e.g., Executive Control Network, Default Mode Network, and Dopaminergic System (Jauk 2019; Runco and Acar 2019).

Moreover, it seems that the third task was especially suitable in the present research context. In the third word-formation task, participants were asked to propose a name for the person depicted in a drawing. This is crucial with regard to ecological validity, as this task entailed both visual stimuli, as presented in many real-life scenarios where natural naming occurs, and open-ended responses where divergent thinking could be freely manifested.

Moreover, according to the present results, Elaboration is not the only potentially important correlate of word formation. It was observed that the criterion-referenced measures (Creative Strengths), comprising factors like humor and colorfulness of imagery, and, potentially, also the tendency to create rare but still meaningful responses (Originality) are related to the way language speakers coin new complex words (although, the latter could be a false-positive finding).

Fluency and Flexibility seem to be related to the tendency to coin new words less transparently. Fluency captures the number of generated meaningful ideas. Flexibility captures the number of different categories provided. Thus, language speakers that tend to provide more answers and are able to switch categories more easily prefer to coin new words less transparently. Note, however, that these results could also be false positives and that this pattern of results occurred only in the second-word formation task. In this task, participants were given a verbal description of a person in an unusual situation. This indicates that word-formation creativity is task-specific (see above, Section 3.2) and that divergent thinking abilities seem to play more important role in some contexts than in others.

In general, the present pattern of results is in accordance with the approach accenting that “the creative potential of human minds is directly reflected in language structure and use” (Langlotz 2015: 40). In fact, from the psycholinguistic perspective, the present pattern of results is important as it suggests that the word-formation process is not just any ‘blind process’ but rather an act of creativity by a language user, as stressed by Štekauer (1998) and Štekauer et al. (2005). This fact is captured by the principle of Creativity within Productivity Constraints, which lays emphasis on the creativity of any member of a speech community who implements the act of naming by an individual choice of a word-formation process/rule and, within it, its name-motivating constituents. Since each individual has unequal experiences, general knowledge, intellectual capacity, imagination, education, age, professional interests, and linguistic knowledge and experiences, one would expect speakers to bring considerable variation to the naming task. In other words, word-formation creativity in our understanding is based on the availability of a number of options at (i) the onomasiological level that reflects the cognitive processing of an object to be named and identifying those semantic categories that will constitute the basis for the linguistic representation of the object, and (ii) the onomatological (morphemic) level that consists of selecting the linguistic constituents that represent the semantic categories of the onomasiological level. From the psychological perspective, the present pattern of results is in line with the bio-psycho-behavioral model stating that behavioral outcomes (as an act of creativity in coining new words) are determined by a deeper psychological level where abilities, in the form of divergent thinking potential, play an important role (Jauk 2019).