Scala naturae: the impact of historical values on current ‘evolution of language’ discourse

Scala naturae: the impact of historical values on current ‘evolution of language’ discourse Abstract Various complaints about the consistent use of a non-epistemological ‘norm of progress’ (also known as ‘Scala Naturae’) can be found frequently in recent evolution of language and communication literature. Affiliated to earlier studies that addressed quantification of some overt indicators such as word combinations of ‘high + species’, the current account aims to go beyond the obvious in describing the presumed phenomena. Using a mixed-methodology approach, we quantify the general use of vocabulary, range of study species, amount of ‘progressionist attributes’ and subsequently qualify the context of some key words. Investigating 915 peer-reviewed articles from a species-comparative evolution of language and communication discourse, we found that articles focussing on species groups historically regarded as ‘high’ make more use of attributes implying directed progress than otherwise. We subdivided all articles in two distinct corpora. Articles using the term ‘language’ or ‘speech’ in title, abstract or keywords were labelled ‘language’. Those using other terms than language were labelled ‘communication’. We could identify a more diverse focus on studied species groups and a more behaviouristic vocabulary in corpus ‘communication’ as compared to the corpus ‘language’. Additionally, articles from the latter corpus tend to stress a narrative of human uniqueness. Our results, taken together, do not provide clear evidence for a structural and active promotion of a ‘norm of progress’, but hint towards historical aftermaths exercising indirect influence and worthy of further study. 1. Background Over the last decade, a growing number of articles discuss the so called ‘replication crisis’ in psychology and other scientific disciplines (Ioannidis, 2012). A subsequent boost in meta-research found that many empirical results are not as robust as they originally seemed. Publication biases (Fanelli 2010), insufficient replication (Makel et al., 2012), lack of data sharing (Wicherts et al. 2006), questionable research practices (John et al. 2012), or low statistical power (Bezeau and Graves 2001) are just some of the factors identified as possible explanations for the crisis. However, the efforts so far seem to underestimate subjective influences such as personal expectations, use of terminology, as well as physical and psychological constraints in designing hypotheses and conducting experiments. In order to investigate these factors, meta-research must reach beyond quantitative methods and supplement them with a mixed-method approach. This means that the current investigation takes as a basis quantitative data (e.g. from text mining) and mixes it with qualitative material (e.g. analysis of context). Such an approach may be especially helpful in identifying and reflecting upon the structure and function of non-epistemological values or norms, such as ethical, social, or political considerations (Douglas 2016). Given those non-epistemological norms on the one side, epistemological norms, such as ‘reproducibility’, ‘scope’, and ‘transparency’, on the other side, are an accepted integral part of scientific reasoning (Douglas 2009: 17). According to this, it is only the first group of norms that is sometimes perceived as threat to scientific objectivity (see Hudson 2016). From the perspective of a value-free ideal, non-epistemological norms (i.e. moral considerations) should not affect scientific practice, because their subjective element is a ‘remit of art, not of science’ (Mogie 2000: 869). However, several classic studies (Feyerabend 1975; Latour and Woolgar 1979) as well as some recent publications (Elliott and McKaughan 2009, Davis 2013; Douglas 2016; Mascolo 2016) question the value-free ideal. In general, so the criticism goes, scientists are part of society and therefore inextricably linked to its values (Douglas 2016). As a consequence, any description of human or non-human behaviour that goes beyond mere observations draws inevitably on the bias of preconceptions: ‘The privileging of measurement over meaning puts the empirical cart before the conceptual horse.’ (Mascolo 2016: 5). Following this line of reasoning, not only is data input influenced by subjective values such as ‘preference for similar others’ (‘homophily’; Haun and Over 2013) or a priori rejection of ‘human-animal similarity’ (‘anthropodenial’; de Waal 1999), but scientists’ data output also has consequences in the social and ethical domain (Douglas 2009: 115). Illustrative examples can be found with reference to sign languages. Until the mid-20th century, a dominant preconception in science understood the oral modality as a necessary prerequisite for having ‘language’, which itself was supposed to be responsible for rationality and flexible communication (Ullrich 2016: 185). As a consequence, deaf humans were forced to learn oral forms of communication instead of better-suited sign languages, with various negative consequences for decades (Ullrich 2016: 189). One cannot simply blame scientific concepts and the use of terminology for those developments, but they did play a major role. Values in science become more important where ‘social categories and the images they embed are inescapably value-laden’ (Davis 2013: 554). We believe this also to be the case with ‘evolution of language and communication’ discourse, where scientists try to create a valid human self-conception with reference to a supposedly human unique characteristic, namely language (e.g. Berwick et al. 2013; Hauser et al. 2014; Scott-Phillips 2015). Given the entanglements between science and non-epistemological values or norms, it appears to us more productive to monitor norms instead of combating them. In that respect, we want to qualify and quantify one potentially lasting norm in order to enable future investigations towards experimental design, the formulation of questions and subsequent interpretation of data. As such, the study contributes to the process of scientific self-correction. The potential norm at issue is the ‘norm of progress’ (also known as ‘Scala Naturae’ or ‘Great Chain of Being’), which assumes that evolution proceeds in a linear ‘upward’ way from a simple/primitive condition towards an ‘improved’ state. Although modern evolutionary theory rejects this prediction (Johnson et al. 2012), a number of scientists complain about the persistence of the norm (Chittka et al. 2012: 2678; Cimatti and Vallortigara 2015: 6; de Waal 1999: 257; Emery and Clayton 2004: 37; Fitch et al. 2010: 796; Nee 2005). A number of qualitative studies focus on the history and current influence of the ‘norm of progress’ (Ghiselin 2005; Hodos and Campbell 1969; Lovejoy 1936; Ruse 1996). By design they do not quantify the phenomenon in recent discourse. Thus, despite the frequent complaints regarding the persistence of the ‘norm of progress’, to date there are only two attempts to study the existence of this norm in more quantitative ways. In 2000, Mogie searched scientific papers published between 1995 and 1999 using the attributes ‘higher’ or ‘lower’ in descriptions of species within the title. A low-tech query returned over 700 positive hits, mostly in studies of plants (n = 665) (Mogie 2000). Following that study, in 2013, Rigato and Minelli performed a scientometric analysis of over 67,413 biological articles published between 2005 and 2010 in 16 different scientific journals. Their queries on journal websites identified 1,287 out of 67,413 articles (1.91%) using ‘Scala Naturae language’ (Rigato and Minelli 2013). Another query in the course of the same study on PubMed confirmed that >55% of all positive hits derive from Botany (Rigato and Minelli 2013). Yet despite providing first evidence for possible implications of a non-epistemic norm within a discourse, neither study continues beyond overtly quantifiable issues, and both fail to identify any of the phenomena other academic peers have attributed to the realm of the norm. For instance, the historical exclusion of birdsong as a model of language (Sereno 2014: 5), addressed by qualitative research, escaped these quantitative accounts. Given the prevailing value-free ideal (Reiss and Sprenger 2014: Chapter 3.1), it is assumed that non-epistemological norms are mostly deployed unintentionally and not overtly, and are therefore difficult to identify. For these reasons, the current study aims to go beyond easily accessible ‘higher/lower classifications’. Instead, it quantifies implicit indicators of a ‘norm of progress’ in peer-reviewed publications on language/communication across a variety of species groups. In order to do that we intent to divide articles from the evolution of language discourse into two distinct corpora: ‘language’ and ‘communication’. The only reason we count an article to the corpus ‘language’ lies in the presence of the predefined terms ‘language’ or ‘speech’ in abstract, title, or keywords. On the other side, articles of corpus ‘communication’ use terms like ‘signal’, ‘song’, ‘vocalization’, ‘gesture’, or ‘communication’. The categorization makes no assertion about the actual focus of a publication. We are aware that sometimes the terms ‘language’ and ‘communication’ are used in similar ways within the current study. It is not our attempt to equate these terms. However, definitions of ‘language’ are notoriously diverse (Botha 2000). That includes perspectives which see ‘communication’ as mere side-effect of ‘language’ (e.g. Chomsky 2011: 264–65), as well as the opposite claim that regards communication as main driver for ‘language evolution’ (e.g. Okanoya 2017; Zuberbühler 2013: 188). Still, others interpret ‘language’ as part of a broader ‘communicative toolkit’ which also includes music and animal song (Rohrmeier et al. 2015). In general, we use a broad definition of ‘language’ that includes various cognitive (e.g. learning and memory) and physiological mechanisms (e.g. perception and motor control) (Fitch 2017: 5). The reason for dividing all articles in two corpora is the following: we hypothesize that authors using the word ‘language’ at prominent sections of an article, implicitly tie their research to a more human-centred perspective of research than researchers avoiding the term. If a ‘norm of progress’ exists, we would expect an increase of ‘progressionist vocabulary’ in the corpus ‘language’. ‘Progressionist vocabulary’, like ‘higher’ or ‘sophisticated’, implies the existence of an improved, more sophisticated or more complex ‘end state’ (mostly realised in humans). Since evolutionary theory is not based on a teleological framework, an ‘end state’ cannot exist and the ranking of structures or abilities along a scale of improvement appears mostly human-centred and/or arbitrary. Therefore, use of ‘progressionist language’ is not only ineloquent, but also value-laden. Hence, we assume that if the ‘norm of progress’ exists, we should find biased sampling of study species in the corpus ‘language’ compared to the corpus ‘communication’. Within the total of 915 journal articles, we expect to identify value-laden ‘progressionist’ vocabulary, dependent on species, article format or corpus group. 2. Material and methods In order to gather corpus material, we performed search queries on the citation database ‘Scopus’. Our aim was to identify a specific fraction of articles concerned with evolution of language from a species-comparative point of view. We chose to select those specific articles for two reasons: first, in both past and current debates it is notoriously difficult to identify a generally accepted definition of ‘language’ (Botha 2000), which makes the whole field of research an ideal candidate for speculation and value-laden narratives. Second, one controversial point in research on the ‘language origin’ concerns the question as to whether ‘language’ evolved either continuously across species (Wilcox 1999; Hurford 2014) or abruptly in human beings (Berwick et al. 2013). An answer might have wide-ranging implications for the human self-concept and, thus giving reason to expect value-loading on that issue in particular. For the years 2005–15, we selected from 16 Journals that have a high impact in the particular field of research (Table 1). Articles using ‘language’ or ‘speech’ in their abstract, title, or as keywords are collected in a corpus termed ‘language’ (n = 890). To contrast the results, we also wanted to identify publications focussing on communication, signal, song, gesture, or vocalization. Articles using one of those terms in their abstract, title, or as keywords are collected in a corpus termed ‘communication’ (n = 1,107).1 All articles examined were manually checked for relevance by reading abstracts and key words. Articles were included in the corpus of investigation when they fulfilled the following requirements: they (i) use a comparative, cross-species approach; (ii) focus on language/communication (not cognition in general); (iii) focus on biological evolution (i.e. exclude machines); (iv) consider multicellular organisms (but not plants, fungi), and (v) focus on inter-individual communication. Table 1. Composition of corpus ‘language’ and ‘communication’ by journal. Since two journals were founded in 2010 and 2011, respectively, they were not available for analysis before that year. Furthermore, publications from Behav. Brain. Sci. were not available as full text HTML before 2006 and thereby excluded for 2005. Journal name  No. of papers in corpus ‘language’  No. of papers in corpus ‘communication’  Anim. Behav.  36  205  Anim. Cogn.  22  17  Behav. Brain. Sci.  106  7  Curr. Anthropol.  37  4  Curr. Biol.  44  43  Evol. Hum. Behav.  9  2  Evol. Psychol.  4  1  Front. Psychol.  22  2  J. Comp. Psychol.  8  1  Nat. Commun. (*2010)  9  7  Phil. Trans. R. Soc. B  33  11  PLoS Biol.  9  2  PLOS ONE  42  70  PNAS  33  18  Proc. R. Soc. B  19  74  Sci. Rep. (*2011)  6  12  Journal name  No. of papers in corpus ‘language’  No. of papers in corpus ‘communication’  Anim. Behav.  36  205  Anim. Cogn.  22  17  Behav. Brain. Sci.  106  7  Curr. Anthropol.  37  4  Curr. Biol.  44  43  Evol. Hum. Behav.  9  2  Evol. Psychol.  4  1  Front. Psychol.  22  2  J. Comp. Psychol.  8  1  Nat. Commun. (*2010)  9  7  Phil. Trans. R. Soc. B  33  11  PLoS Biol.  9  2  PLOS ONE  42  70  PNAS  33  18  Proc. R. Soc. B  19  74  Sci. Rep. (*2011)  6  12  Relevant articles (‘language’ n = 439; ‘communication’ n = 476) were supplemented with meta-information such as (a) species focus, (b) modality, and (c) full-text download link. With regard to (a) nine groups of species were identified (1. human primate, 2. non-human primate, 3. non-primate mammals, 4. marine mammal, 5. bird, 6. other vertebrates, 7. invertebrate, 8. fish, 9. unspecified). With reference to (b) seven modalities were identified (1. acoustic, 2. visual, 3. chemical, 4. tactile, 5. thermal, 6. cross-modal, 7. multimodal). Most articles were automatically retrieved2 based on their link, converted from source HTML into a raw text format, and broken down to the level of individual words. Specific word classes were attributed automatically via TreeTagger using default settings (Schmid 1995). In addition to obvious lemmas like ‘high’ and ‘low’ used by Rigato and Minelli 2013, we consider a greater number of terms as contributing to a ‘norm of progress’. We created two groups of 56 handpicked lemmas (see Supplementary Material) to investigate the use of ‘progressionist vocabulary’, i.e. words that in a broader sense allow a linear differentiation between ‘high’ and ‘low’. Those potentially value-laden lemmas were identified by earlier research as relating to the ‘norm of progress’ (Güntürkün and Bugnyar 2016; Jarvis et al. 2005; Karten 2015; McShea 2011; Ruse 1996; Ullrich 2016) or were mentioned within an open survey by members of the Comparative Developmental Psychology group in Berlin (see Supplementary Material). For brevity, we named those word groups ‘high’ and ‘low’, respectively and used them in order to compare the appearance of lemmas between corpora and various meta-data. All quantitative analyses were performed using R 3.2 (R Development Core Team, 2016). A list of additional R-packages in use can be found at the Supplementary Methods section. To capture even subtle indicators of the ‘norm of progress’, the study combines quantitative text analysis and a qualitative audit of context (a mixed-methods approach). For qualitative analysis of context, we extracted respective text snippets into Excel Sheets and rated for context manually (i.e. ‘opposite meaning’, ‘species related’, ‘neutral’). All R-Scripts used and consulted material are open and can be downloaded (DOI 10.17605/OSF.IO/EGFHV). 3. Results and discussion 3.1 Authors mostly avoid direct linkage of ‘high/low’ to various species groups Rigato and Minelli (2013) concluded that ‘the great chain of being is still there’. When we reproduced their methodology for 915 publications from our corpus, we could identify 8 cases of direct linkage between ‘high’ and several species, but could not find any incidence with ‘low’. Hits from ‘higher’ linked to either ‘vertebrates’ (Earley 2010: 2676; Hauser et al. 2014: 1; Iriki and Taoka 2012: 18) or ‘primates’ (Cunningham and Ramos 2014: 806; Glickstein 2007: 824; Jablonka and 2012: 2155; Sadagopan et al. 2015: 10) with one exception of ‘plants’ (Caulier et al. 2013: 1). By definition, publications from the field of botany were excluded, whereby 0.87% positive hits from 915 articles nearly resembles those botany-free results presented by Rigato and Minelli (2013). Contrary to their interpretation, we do not conclude that results can lead us to state that researchers adhere to a ‘norm of progress’. In all affected articles, we could identify only one or two singular events linking ‘high + species’. When checking those papers manually, we could not identify a systematic use of ‘Scala Naturae language’. Instead, we consider those findings as singular cases of ‘historical baggage’ (Mogie 2000: 868) where expressions and metaphors echo a long tradition of teleological thinking. However, as previously mentioned, we did not assume that the linking of overt ‘high’ and ‘low’ classifications with various species would occur at a high frequency, since we expected non-epistemological norms to be mostly used unintentionally and therefore not overtly expressed in the text. This is why we started exploratory investigations for more implicit indicators that might impact the discourse. 3.2 Primates dominate corpus ‘language’ In 2014, Sereno claims that ‘birdsong has often been dismissed as a model of human language for the reason that monkeys seem much smarter than some birds’ (Sereno 2014: 5). We wanted to quantify his complaint regarding ‘Scala Naturae thinking’ by checking its substance in current literature. As described in our methods section, we divided all articles into two corpora labelled ‘language’ and ‘communication’, respectively. Subsequently, we decided to compare the range of studied species groups between the corpus ‘language’ and that of ‘communication’. We found a substantially wider range of studied species groups in the corpus ‘communication’ as compared to the corpus ‘language’ (Fig. 1). About 70% of all 439 articles using the terms ‘language’ or ‘speech’ in title, abstract or keywords focussed on primates. Broken down to specific groups we observed for the corpus ‘language’ that the majority of articles focussed specifically on human primates (38%), followed by non-human primates (32%), birds (11%), and finally publications without definite species focus (7.2%). In contrast, articles within the corpus ‘communication’ mostly focussed on invertebrates (26.89%) and birds (26.68%), followed by other vertebrates and non-human mammals (both 10.29%). From an overall 476 articles investigating communication and its evolution, only 11 focussed on human (2.3%) and 40 on non-human primates (8.4%). Figure 1. View largeDownload slide Comparison of the range of studied species groups in corpus ‘language’ and ‘communication’. White horizontal line in primate box subdivides the group into human (above broken line) and non-human primates (below). Figure 1. View largeDownload slide Comparison of the range of studied species groups in corpus ‘language’ and ‘communication’. White horizontal line in primate box subdivides the group into human (above broken line) and non-human primates (below). The results pertaining to humans in the corpus ‘language’ may not surprise, since many researchers regard language to be unique to them (e.g. Berwick et al. 2013; Hauser et al. 2014; Scott-Phillips 2015). Nonetheless, almost 62% of all articles using ‘language’ in their title, abstract, or keywords do focus on non-human animals, most of them on non-human primates. For our study, it is of no importance to distinguish if those articles investigate the origin of ‘language’ or ‘communication’. We also cannot distinguish between the opposite use of the term or its context. Apart from these issues, it strikes us that articles focussing on invertebrates, fish, or other vertebrates avoided almost completely the term ‘language’ in their opening sections. Even when most articles on non-human primates in the corpus ‘language’ investigated the ‘origin of communication’ instead of ‘language’, it surprises us that only 8.4% of articles from the corpus ‘communication’ are dedicated to the non-human primate group. That leads us to the conclusion that researchers studying primates are more likely to use the term ‘language’ when investigating communicative behaviours than researchers concerned with other species groups. Given Sereno’s statement (Sereno 2014), we indeed conclude that articles from the corpus ‘language’ tended in relative numbers to ‘dismiss’ birds as a model for investigating the evolution of language. What might explain this phenomenon? Researchers tend to see abilities that they value, which is more easily done in species that closely resemble humans, e.g. primates. For instance, the oral/acoustic modality of human communication is the subject of 58.5% of the studies within the corpus ‘language’. Modalities presumably less relevant to average humans or multimodal accounts that received attention only recently are covered comparably less (cross-modal: 22.3%; visual: 11.2%; multimodal: 7.1%; chemical: <1%; see Supplementary Fig. S1). Earlier studies have traced some historical sources of the phenomenon’s origin such as an ’oral norm’ (Ullrich 2016), ‘a priori biases’ (Slocombe et al. 2011), ‘Primatocentrism’ (Cimatti and Vallortigara 2015; Emery and Clayton 2004), or ‘Chimpocentrism’ (Vaesen 2014). Frequent focus on primates’ unimodal behaviour in the early days of comparative communication studies might have caused an underestimation of the communicative abilities of non-primates, which in turn makes non-primate research look less interesting. The circle creates its own evidence and fuels a view of ascending ‘complexity’ over the course of evolutionary development. In accordance with this interpretation, we examined if both corpora would differ in their use of directional language. Since in the corpus ‘language’, there are more species investigated historically considered ‘high’ than in the corpus ‘communication’, we expect to find more adjectives representing ‘high’ in the corpus ‘language’ than the other way around. We thus decided to identify the 80 most common adjectives used in both corpora. 3.3 A selection of the 80 most common adjectives hints towards substantially different narratives across the two corpora Adjectives give more precise information about a particular object of interest. Therefore, in case the ‘norm of progress’ influences scientific publications, we would expect more adjectives implying ‘high’ in the corpus ‘language’ as compared to the corpus ‘communication’. This expectation is based on the following reflection: if articles from the corpus ‘language’ focus mostly on species groups that were considered ‘high’ under the terms of a ‘norm of progress’, and articles from the corpus ‘communication’ deal with ‘lower’ ones, adjectives implying ‘high’ should appear more frequently in corpus ‘language’. However, the analysis for the 80 most frequent adjectives did not meet our initial expectations. Indeed, the adjective ‘complex’ occurs more often in ‘language’ as compared to ‘communication’, while adjective ‘low’ followed the opposite pattern (Fig. 2). However, we were not able to detect any structural regularity that would systematically ascribe ‘high’ or ‘low’ value-laden adjectives to any of the corpora. Instead, we became interested in those adjectives without respective counterparts within the list. Figure 2. View largeDownload slide List of the 80 most common adjectives of the respective corpus, ordered by their occurrence. Adjectives that appear on either side are linked by lines. Adjectives without line do not have a respective counterpart among the most frequent 80. Figure 2. View largeDownload slide List of the 80 most common adjectives of the respective corpus, ordered by their occurrence. Adjectives that appear on either side are linked by lines. Adjectives without line do not have a respective counterpart among the most frequent 80. With regard to the corpus ‘language’, examples of some frequently employed adjectives are: cognitive, linguistic, communicative, neural, functional, cultural, syntactic, gestural, and semantic. With regard to the corpus ‘communication’, some examples are: sexual, reproductive, sensory, aggressive, conspecific, facial, territorial, and dominant. It appears to us that those words tell very different stories about similar observations. One (corpus ‘communication’) investigates the communicative behaviour of a species for the sake of the species itself, while the other corpus (‘language’) aims to compare communicative behaviour between non-human and human animals. Articles using the term ‘language’ in the abstract, title or key words tend to link and compare their findings to cognition and linguistic concepts, aspects that were investigated in former times under the umbrella term ‘animal psychology’. Articles avoiding ‘language’ concentrate on ecology and ethology, aspects that are investigated under the umbrella term ‘behaviourism’. However, since extracting the 80 most common adjectives did not answer the question as to whether one of the corpora would feature the more frequent deployment of ‘high’ or ‘low’ classifications, we then decided to directly create a list of target words with the objective of comparing them accordingly. 3.4 No difference in directional vocabulary between corpus, but between species group and articles type Due to the different emphasis on species groups between the corpora and the identification of two diverging uses of vocabulary when writing up results, we were interested in whether a selected list of words could also reveal a difference in the use of lemmas classified as ‘high’ or ‘low’. We predicted that under the terms of a persistent ‘norm of progress’, articles in the corpus ‘language’ would use more lemmas implying values of ‘high’ while avoiding those implying ‘low’, as compared to the corpus ‘communication’. To quantify frequencies of word appearances, we created a list of 58 words which either imply evolutionary ‘improvement’ or ‘simplicity’. The choice of words was based on earlier research (Güntürkün and Bugnyar 2016; Jarvis et al. 2005; Karten 2015; McShea 2011; Ruse 1996; Ullrich 2016) and an open survey among researchers in comparative psychology (see Supplementary Material). To account for different text length, we corrected all hits by the total number of words per article. Altogether we found that words of the category ‘high’ are used 40% more often in the corpus ‘language’ and 32% more often in the corpus ‘communication’ as compared to words of category ‘low’. However, the difference of direct hits between the corpora was not as clear as expected. Indeed, publications of corpus ‘language’ did use words classified as ‘high’ 10.51% more often and words classified as ‘low’ 1.55% less often as compared to the corpus ‘communication’. When related to other factors such as ‘species group’ and ‘article type’, these results shift in weight and appear rather comparable. Indeed, relative frequency differed substantially between various groups of species (Fig. 3). Publications in the corpus ‘language’ focussing on non-human primates use vocabulary from the word group ‘high’ more often than, for instance, articles focussing on birds (+27% in corpus ‘language’; +23% in corpus ‘communication’). Similarly, articles in the corpus ‘language’ without any focus on a species group used words classified as ‘high’ with increasing frequency as compared to articles with a focus on birds (+35% in corpus ‘language’; +18% in corpus ‘communication’). In general, we observed a tendency by which articles focussing on species groups ranking ‘high’ according to a ‘norm of progress’ increase their use of words valuing ‘high’. The small sample for articles focussing on humans (n = 10) in the corpus ‘communication’ constitute an exception to this observation. Since many articles with ‘unspecified’ species groups appear to consist of comments, review or theory pieces, we also wanted to quantify differences for that factor. We found that not only does the species group influence linguistic usage, but also the article format (Fig. 4). In general, we observed a tendency whereby articles with less experimental or empirical focus increase their use of words defined as ‘high/low’. For instance, in articles classified as theoretical publications we identified an increase of words categorised as ‘high’ by 23.6% (corpus ‘language’) and 26.5% (corpus ‘communication’), respectively. Following various conference discussions, most scientists eagerly deny reference to any such ‘norm of progress’.3 As said earlier mostly there is no active promotion for such a norm as it appears ‘unscientific’ (Mogie 2000: 869). Of course, a quantitative text analysis cannot determine to what degree vocabulary is used deliberately, or in which context. For that reason we checked context for one specific word that scientists usually value: ‘unique’. Figure 3. View largeDownload slide Occurrence of words classified as ‘high’ (dark-grey) and ‘low’ (grey) computed per 1,000 words of original article. Results are broken down for species group and ordered by their summarized mean for ‘high’. Horizontal bars indicate mean for all articles of a document type. Numbers in brackets represent articles under investigation. Error bars depict the standard error. Figure 3. View largeDownload slide Occurrence of words classified as ‘high’ (dark-grey) and ‘low’ (grey) computed per 1,000 words of original article. Results are broken down for species group and ordered by their summarized mean for ‘high’. Horizontal bars indicate mean for all articles of a document type. Numbers in brackets represent articles under investigation. Error bars depict the standard error. Figure 4. View largeDownload slide Occurrence of words classified as ‘high’ (dark-grey) and ‘low’ (grey) computed per 1,000 words of original article. Results are broken down for article type and ordered by their summarized mean for ‘high’. Horizontal bars indicate mean for all articles of a species group. Numbers in brackets represent articles under investigation. Error bars depict the standard error. Figure 4. View largeDownload slide Occurrence of words classified as ‘high’ (dark-grey) and ‘low’ (grey) computed per 1,000 words of original article. Results are broken down for article type and ordered by their summarized mean for ‘high’. Horizontal bars indicate mean for all articles of a species group. Numbers in brackets represent articles under investigation. Error bars depict the standard error. 3.5 ‘Language’ more unique than ‘communication’ In order to evaluate our previous results, we wanted to approach the problem of context blindness for one case example. We chose the lemma ‘unique’, because usually its usage does not imply directional connotations. Furthermore, from a biological point of view there is nothing special about being ‘unique’, since every species is defined by its autapomorphy—a derived trait that defines the status as a species. However, based on previous qualitative research we hypothesized that when publications repeatedly state something as uniquely human, but do not mention anything else as unique in non-humans, than this one-sided view might hint towards values in use. In order to test the hypothesis we first quantified the phenomenon and subsequently qualified the results. We found that 52% of all articles in the corpus ‘language’ make use of the lemma ‘unique’, but only 37% of articles in the corpus ‘communication’. If the lemma is used, articles in the corpus ‘language’ use it on average 3.2 times, while articles in the corpus ‘communication’ employ it 1.9 times. Altogether we identified 57% more instances of the lemma ‘unique’ in the corpus ‘language’ as compared to ‘communication’. In order to investigate the qualitative context of the lemma, we extracted all occurrences, including the context, and validated its usage. When ‘unique’ referred to any species group, we labelled it accordingly. When used to the contrary (e.g. ‘not unique’), we labelled it ‘opposite’. When used without reference to any species, we labelled it ‘neutral’. When used in context of an unanswered question or within quotations, we labelled it ‘undecided’. We found that in nearly half the cases from the corpus ‘language’ the lemma ‘unique’ referred to humans, while in 78% of all incidences in the corpus ‘communication’ it was used in neutral manner (Fig. 5). Figure 5. View largeDownload slide Comparison of context from the lemma ‘unique*’ {including: ‘uniquely’ and ‘uniqueness’} between corpora. In the corpus ‘communication’, the lemma ‘unique*’ is mostly used in neutral context, whereas its use in the corpus ‘language’ refers in almost half of the incidences to humans. See text for details and definition of individual labels. Figure 5. View largeDownload slide Comparison of context from the lemma ‘unique*’ {including: ‘uniquely’ and ‘uniqueness’} between corpora. In the corpus ‘communication’, the lemma ‘unique*’ is mostly used in neutral context, whereas its use in the corpus ‘language’ refers in almost half of the incidences to humans. See text for details and definition of individual labels. Certainly, it might come of no surprise that the term ‘unique’ appears frequently compared to ‘human’, since language is regarded as one of the important autapomorphies of the species. However, all species-specific forms of communication are unique by definition. Either someone takes the view that human language is unique and thus not comparable to any non-human form of communication, or one conducts species comparative research and therefore allows a comparison of language and animal communication. When following the second strategy, the consequence is that not only language is unique to humans, but also ultrasonic social communication to bats, electric communication signals to electric fish, and multimodal chemo-acoustic signals to lemurs. Still, in only 44 cases, ‘unique’ relates to the behaviour of a species in the corpus ‘communication’, while we could find 388 such cases (mostly in reference to humans) in the corpus ‘language’. That might constitute a scientific narrative that justifies human speciality as an evolutionary ‘achievement’. As such it hints towards a somewhat chauvinistic function where non-human species are not actively discriminated, but implicitly eclipsed. While scientists highlight human uniquely features, they also feel the urge to find biological ‘roots’ of behaviours and thus start testing and observing ‘downwards’ along the ‘evolutionary tree’. In this respect, such a research agenda could be classified as motivated by the vestiges of a historical ‘norm of progress’. After all, non-epistemological norms do indeed play a role within scientific reasoning (see Douglas 2000). However, the task of monitoring them is always valuable and never completed, enabling readers to develop a critical view of hypotheses, questions, and results. 4. Conclusion In order to quantify a possible non-epistemological ‘norm of progress’ within a current scientific discourse of language evolution, we applied a quantitative text and qualitative context analysis to a corpus consisting of 915 articles. Historically one can find clear evidence for the existence of a ‘norm of progress’ in scientific publications. A reproduction of a study by Rigato and Minelli (2013), however, could show only minor evidence for an open and active promotion of that norm. Hence the focus of subsequent tests was put on implicit factors such as species range, use of vocabulary and values in language. Although papers from corpus ‘language’ and ‘communication’ focus on a similar phenomenon, their narratives appear strikingly different as indicated by the frequency of 80 of the most commonly employed adjectives. In addition, both corpora differ widely in their range of studied species groups and the usage of the lemma ‘unique’. In all cases, the corpus ‘language’ establishes a narrative of human speciality, compared to other species, as could be additionally shown by qualifying all uses of the lemma ‘unique’ within the corpus. However, both corpora use more frequently words in the category ‘high’ with reference to primates as compared to birds or insects. Taken together, there is no evidence for a structural and overt promotion of a non-epistemological ‘norm of progress’ within the discourse. Still, several implicit factors hint at the lingering historical aftermaths of norm-related ideas and an associated subconscious function as leading forces in identifying and formulating current and future research questions. Supplementary data Supplementary data is available at Journal of Language Evolution online. Conflict of interest statement. None declared. Authors’ contributions R.U. conceived and designed the study, collected corpus material, did qualitative analysis, drafted the manuscript, analysed and interpreted quantitative data and wrote parts of R code. M.M. substantially participated in quantitative data analysis & wrote the majority of R code. K.L. did critical revision of article drafts and provided important intellectual content. All authors gave final approval for publication. Funding R.U. and M.M. received no specific funding for this work. K.L. receives a DFG Excellence Initiative Grant. Research ethics The study did not require ethical approval from a local ethics committee. Data availability Data and research materials supporting the results in the article are open and available at the Open Science Framework: doi 10.17605/osf.io/egfhv. Footnotes 1 One example (for more see DOI 10.17605/OSF.IO/EGFHV) of a ‘Scopus’ query for the Journal Animal Cognition contributing to the corpus ‘communication’: TITLE-ABS-KEY (communication OR song OR signal* OR vocali?ation OR gesture AND evol* AND NOT language AND NOT speech) AND ISSN (1435-9456) OR ISSN (1435-9448) AND PUBYEAR AFT 2004 AND PUBYEAR BEF 2016 2 Due to technical oddities this procedure had to be done by hand for two Journals: “Journal of Comparative Psychology” & “Current Anthropology”. 3 e.g. personal communication to Andrew Whiten, T. Scott-Phillips. Acknowledgements We thank Prof. Markus Wild and participants from the Workshop ‘Minds of Animals’ in Bern 2016 for helpful comments. References Berwick R. C. et al.   ( 2013) ‘ Evolution, Brain, and the Nature of Language’, Trends in Cognitive Sciences , 17/ 2: 98. http://doi.org/10.1016/j.tics.2012.12.002. Google Scholar CrossRef Search ADS   Bezeau S., Graves R. ( 2001) ‘ Statistical Power and Effect Sizes of Clinical Neuropsychology Research’, Journal of Clinical and Experimental Neuropsychology , 23/ 3: 399– 406. http://doi.org/10.1076/jcen.23.3.399.1181. Google Scholar CrossRef Search ADS PubMed  Botha R. ( 2000) ‘ Discussing the Evolution of the Assorted Beasts called Language’, Language & Communication , 20: 149– 60. http://doi.org/10.1016/S0271-5309(99)00022-1. Google Scholar CrossRef Search ADS   Caulier G. et al.   ( 2013) ‘ When a Repellent Becomes an Attractant: Harmful Saponins are Kairomones Attracting the Symbiotic Harlequin Crab’, Scientific Reports , 3: 2639. http://doi.org/10.1038/srep02639. Google Scholar CrossRef Search ADS PubMed  Chittka L. et al.   ( 2012) ‘ What is Comparable in Comparative Cognition?’, Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences , 367/ 1603: 2677– 85. http://doi.org/10.1098/rstb.2012.0215. Google Scholar CrossRef Search ADS PubMed  Chomsky N. ( 2011) ‘ Language and Other Cognitive Systems. What Is Special About Language?’, Language Learning and Development , 7/ 4: 263– 78. http://doi.org/10.1080/15475441.2011.584041. Google Scholar CrossRef Search ADS   Cimatti F., Vallortigara G. ( 2015) ‘ So Little Brain, so much Mind. Intelligence and behaviour in Nonhuman Animals’, Reti, Saperi, Linguaggi , 4/ 7: 5– 20. http://doi.org/10.12832/81287. Cunningham C. L., Ramos M. F. ( 2014) ‘ Effect of Training and Familiarity on Responsiveness to Human Cues in Domestic Dogs (Canis familiaris)’, Animal Cognition , 17: 805– 14. http://doi.org/10.1007/s10071-013-0714-z. Google Scholar CrossRef Search ADS PubMed  Davis J. E. ( 2013) ‘ Social Science, Objectivity, and Moral Life’, Society , 50/ 6: 554– 9. http://doi.org/10.1007/s12115-013-9710-9. Google Scholar CrossRef Search ADS   de Waal F. B. M. ( 1999) ‘ Anthropomorphism and Anthropodenial: Consistency in Our Thinking about Humans and Other Animals’, Philosophical Topics , 27/ 1: 255– 80. Google Scholar CrossRef Search ADS   Douglas H. ( 2000) ‘ Inductive Risk and Values in Science’, Philosophy of Science , 67/ 4: 559. http://doi.org/10.1086/392855. Google Scholar CrossRef Search ADS   Douglas H. ( 2009). Science, Policy, and the Value-free Ideal . Pittsburgh: University of Pittsburgh Press. Google Scholar CrossRef Search ADS   Douglas H. ( 2016). ‘Values in Science’, in Humphreys P. (ed.) The Oxford Handbook of Philosophy of Science , 609– 32. Oxford, New York: Oxford University Press. http://doi.org/10.1093/oxfordhb/9780199368815.013.28. Earley R. L. ( 2010) ‘ Social Eavesdropping and the Evolution of Conditional Cooperation and Cheating Strategies’, Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences , 365/ 1553: 2675– 86. http://doi.org/10.1098/rstb.2010.0147. Google Scholar CrossRef Search ADS PubMed  Elliott K. C., McKaughan D. J. ( 2009) ‘ How Values in Scientific Discovery and Pursuit Alter Theory Appraisal’, Philosophy of Science , 76/ 5: 598– 611. http://doi.org/10.1086/605807. Google Scholar CrossRef Search ADS   Emery N., Clayton N. S. ( 2004). ‘Comparing the Complex Cognition of Birds and Primates’, in Rogers L.J., Kaplan G. (eds) Comparative Vertebrate Cognition , pp. 3– 55. New York: Springer Science+Business Media. Google Scholar CrossRef Search ADS   Fanelli D. ( 2010) ‘ Do Pressures to Publish Increase Scientists’ Bias? An Empirical Support from US States Data’, PLoS ONE , 5/ 4, p. e10271. http://doi.org/10.1371/journal.pone.0010271. Feyerabend P. ( 1975). Against Method (3rd 1993) . London: Verso. Fitch W. T. ( 2017) ‘ Empirical Approaches to the Study of Language Evolution’, Psychonomic Bulletin & Review , http://doi.org/10.3758/s13423-017-1236-5. Fitch W. T., Huber L., Bugnyar T. ( 2010) ‘ Social Cognition and the Evolution of Language: Constructing Cognitive Phylogenies’, Neuron , 65/ 6: 795– 814. http://doi.org/10.1016/j.neuron.2010.03.011. Google Scholar CrossRef Search ADS PubMed  Ghiselin M. T. ( 2005) ‘ The Darwinian Revolution as Viewed by a Philosophical Biologist’, Journal of the History of Biology , 38/ 1: 123– 36. http://doi.org/10.1007/s10739-004-6513-2. Google Scholar CrossRef Search ADS PubMed  Glickstein M. ( 2007) ‘ What does the Cerebellum really do?’, Current Biology , 17/ 19: 824– 7. http://doi.org/10.1016/j.cub.2007.08.009. Google Scholar CrossRef Search ADS   Güntürkün O., Bugnyar T. ( 2016) ‘ Cognition without Cortex’, Trends in Cognitive Sciences, Xx , 1– 13, http://doi.org/10.1016/j.tics.2016.02.001. Haun D., Over H. ( 2013). ‘Like me: a Homophily-Based Account of Human Culture’, in Richerson P. J., Christiansen M. H. (eds), Cultural Evolution: Society, Technology, Language, and Religion ,pp. 75– 85. Cambridge: MIT Press. Hauser M. D. et al.   ( 2014) ‘ The Mystery of Language Evolution’, Frontiers in Psychology , 5(MAY): 1– 12. http://doi.org/10.3389/fpsyg.2014.00401. Hodos W., Campbell C. G. B. ( 1969) ‘ Scala Naturae: Why there is no Theory in Comparative Psychology’, Psychological Review , 76/ 4: 337– 50. Google Scholar CrossRef Search ADS   Hudson R. ( 2016) ‘ Why we should not Reject the Value-Free Ideal of Science’, Perspectives on Science , 24/ 2: 167– 91. http://doi.org/10.1162/POSC_a_00199. Google Scholar CrossRef Search ADS   Hurford J. R. ( 2014). Origins of Language: A Slim Guide . Oxford: Oxford University Press. Ioannidis J. P. A. ( 2012) ‘ Why Science is not Necessarily Self-Correcting’, Perspectives on Psychological Science , 7/ 6: 645– 54. http://doi.org/10.1177/1745691612464056. Google Scholar CrossRef Search ADS PubMed  Iriki A., Taoka M. ( 2012) ‘ Triadic (Ecological, Neural, Cognitive) Niche Construction: a Scenario of Human Brain Evolution Extrapolating Tool Use and Language from the Control of Reaching Actions’, Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences , 367/ 1585: 10– 23. http://doi.org/10.1098/rstb.2011.0190. Google Scholar CrossRef Search ADS PubMed  Jablonka E., Ginsburg S., Dor D. ( 2012) ‘ The co-evolution of Language and Emotions’, Philosophical Transactions of the Royal Society B: Biological Sciences , 367/ 1599: 2152– 9. http://doi.org/10.1098/rstb.2012.0117. Google Scholar CrossRef Search ADS   Jarvis E. D. et al.   ( 2005) ‘ Avian Brains and a New Understanding of Vertebrate Brain Evolution’, Nature Reviews. Neuroscience , 6(February): 151– 9. http://doi.org/10.1038/nrn1606. Google Scholar CrossRef Search ADS   John L. K., Loewenstein G., Prelec D. ( 2012) ‘ Measuring the Prevalence of Questionable Research Practices with Incentives for Truth Telling’, Psychological Science , 23/ 5: 524– 32. http://doi.org/10.1177/0956797611430953. Google Scholar CrossRef Search ADS PubMed  Johnson N. A., Lahti D. C., Blumstein D. T. ( 2012) ‘ Combating the Assumption of Evolutionary Progress: Lessons from the Decay and Loss of Traits’, Evolution: Education and Outreach , 5/ 1: 128– 38. http://doi.org/10.1007/s12052-011-0381-y. Google Scholar CrossRef Search ADS   Karten H. J. ( 2015) ‘ Vertebrate Brains and Evolutionary Connectomics: On the Origins of the Mammalian “Neocortex’, Philosophical Transactions of the Royal Society B: Biological Sciences , 370/ 1684: 20150060. http://doi.org/10.1098/rstb.2015.0060. Google Scholar CrossRef Search ADS   Latour B., Woolgar S. ( 1979). Laboratory Life. The Construction of Scientific Facts . Princeton, New Jersey: Princeton University Press. Lovejoy A. O. ( 1936). The Great Chain of Being. A Study of the History of an Idea (26th repri) . Cambridge, Massachusetts, London: Harvard University Press. Makel M. C., Plucker J. A., Hegarty B. ( 2012) ‘ Replications in Psychology Research: How Often Do They Really Occur?’, Perspectives on Psychological Science , 7/ 6: 537– 42. http://doi.org/10.1177/1745691612460688. Google Scholar CrossRef Search ADS PubMed  Mascolo M. F. ( 2016) ‘ How Objectivity Undermines the Study of Personhood: Toward an Intersubjective Epistemology for Psychological Science’, New Ideas in Psychology , 44: 41– 8. Google Scholar CrossRef Search ADS   McShea D. W. ( 2011). ‘Evolutionary Progress’, in Ruse M., Travis J. (eds), Evolution: The First Four Billion Years , pp. 550– 57. Cambridge, Massachusetts: Harvard University Press. Mogie M. ( 2000) ‘ Historical Baggage in Biology: The Case of “Higher” and “Lower” Species’, BioEssays , 22/ 9: 868– 69. http://doi.org/10.1002/1521-1878(200009)22:9<868::AID-BIES13>3.0.CO;2-A. Google Scholar CrossRef Search ADS PubMed  Nee S. ( 2005) ‘ The Great Chain of Being’, Nature , 435(May): 429. http://doi.org/10.1177/00221678930333006. Google Scholar CrossRef Search ADS   Okanoya K. ( 2017). Sexual Communication and Domestication may give rise to the Signal Complexity Necessary for the Emergence of Language: An Indication from Songbird Studies. Psychonomic Bulletin & Review, 24: 106– 10. R Development Core Team, R. ( 2016). R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing. Retrieved from http://www.r-project.org/ accessed 7 July 2016. Reiss J., Sprenger J. ( 2014). ‘Scientific Objectivity’, in Zalta Edward N. (ed.) The Stanford Encyclopedia of Philosophy  (Fall 2014). Retrieved from http://plato.stanford.edu/entries/scientific-objectivity/ accessed 10 May 2016. Rigato E., Minelli A. ( 2013) ‘ The Great Chain of being is still here’, Evolution: Education and Outreach , 6/ 18: 1– 6. http://doi.org/10.1186/1936-6434-6-18. Rohrmeier M. et al.   ( 2015) ‘ Principles of Structure Building in Music, Language and Animal Song’, Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences , 370/ 1664: 20140097. http://doi.org/10.1098/rstb.2014.0097. Google Scholar CrossRef Search ADS PubMed  Ruse M. ( 1996). Monad to Man. The Concept of Progress in Evolutionary Biology . Cambridge, Massachusetts, London, England: Harvard University Press. Sadagopan S., Temiz-Karayol N. Z., Voss H. U. ( 2015) ‘ High-field Functional Magnetic Resonance Imaging of Vocalization Processing in Marmosets’, Scientific Reports , 5: 10950. http://doi.org/10.1038/srep10950. Google Scholar CrossRef Search ADS PubMed  Schmid H. ( 1995). Improvements in Part-of-Speech Tagging with an Application To German. In Proceedings of the ACL SIGDAT-Workshop, pp. 1–9. Dublin, Ireland. Retrieved from http://www.cis.uni-muenchen.de/∼schmid/tools/TreeTagger/ accessed 8 September 2016. Scott-Phillips T. C. ( 2015) ‘ Nonhuman Primate Communication, Pragmatics, and the Origins of Language’, Current Anthropology , 56/ 1: 56– 80. http://doi.org/10.1086/679674. Google Scholar CrossRef Search ADS   Sereno M. I. ( 2014) ‘ Origin of Symbol-Using Systems : Speech, but not Sign, without the Semantic Urge’, Philosophical Transactions of the Royal Society B , 369(August): 20130303. http://dx.doi.org/10.1098/rstb.2013.0303. Google Scholar CrossRef Search ADS   Slocombe K. E., Waller B. M., Liebal K. ( 2011) ‘ The Language Void: the Need for Multimodality in Primate Communication Research’, Animal Behaviour , 81/ 5: 919– 24. http://doi.org/10.1016/j.anbehav.2011.02.002. Google Scholar CrossRef Search ADS   Ullrich R. ( 2016) ‘ From “Speech” to “Gesture”: The “Oral” as Norm in “Language” Research’, Interdisziplinäre Anthropologie Jahrbuch: Wahrnehmung , 4: 179– 208. http://doi.org/10.1007/978-3-658-14264-3. Vaesen K. ( 2014) ‘ Chimpocentrism and Reconstructions of Human Evolution (a Timely Reminder)’, Studies in History and Philosophy of Science Part C :Studies in History and Philosophy of Biological and Biomedical Sciences , 45/ 1: 12– 21. http://doi.org/10.1016/j.shpsc.2013.12.004. Google Scholar CrossRef Search ADS   Wicherts J. M. et al.   ( 2006) ‘ The Poor Availability of Psychological Research Data for Reanalysis’, The American Psychologist , 61/ 7: 726– 728. http://doi.org/10.1037/0003-066X.61.7.726. Google Scholar CrossRef Search ADS PubMed  Wilcox S. ( 1999). ‘The Invention and Ritualization of Language’, in King B. (ed.) The Origins of Language: What Nonhuman Primates Can Tell Us , pp. 351– 84. Oxford: James Currey Ltd. Zuberbühler K. ( 2013). ‘Primate Communication’, in Lefebvre C., Comrie B., Cohen H. (eds) New Perspectives on the Origins of Language , pp. 187– 210. Amsterdam: John Benjamins Publishing Co. http://doi.org/10.1075/slcs.144. Google Scholar CrossRef Search ADS   © The Author 2017. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Language Evolution Oxford University Press

Scala naturae: the impact of historical values on current ‘evolution of language’ discourse

Loading next page...
 
/lp/ou_press/scala-naturae-the-impact-of-historical-values-on-current-evolution-of-hZJrc8urPw
Publisher
Oxford University Press
Copyright
© The Author 2017. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com
ISSN
2058-4571
eISSN
2058-458X
D.O.I.
10.1093/jole/lzx017
Publisher site
See Article on Publisher Site

Abstract

Abstract Various complaints about the consistent use of a non-epistemological ‘norm of progress’ (also known as ‘Scala Naturae’) can be found frequently in recent evolution of language and communication literature. Affiliated to earlier studies that addressed quantification of some overt indicators such as word combinations of ‘high + species’, the current account aims to go beyond the obvious in describing the presumed phenomena. Using a mixed-methodology approach, we quantify the general use of vocabulary, range of study species, amount of ‘progressionist attributes’ and subsequently qualify the context of some key words. Investigating 915 peer-reviewed articles from a species-comparative evolution of language and communication discourse, we found that articles focussing on species groups historically regarded as ‘high’ make more use of attributes implying directed progress than otherwise. We subdivided all articles in two distinct corpora. Articles using the term ‘language’ or ‘speech’ in title, abstract or keywords were labelled ‘language’. Those using other terms than language were labelled ‘communication’. We could identify a more diverse focus on studied species groups and a more behaviouristic vocabulary in corpus ‘communication’ as compared to the corpus ‘language’. Additionally, articles from the latter corpus tend to stress a narrative of human uniqueness. Our results, taken together, do not provide clear evidence for a structural and active promotion of a ‘norm of progress’, but hint towards historical aftermaths exercising indirect influence and worthy of further study. 1. Background Over the last decade, a growing number of articles discuss the so called ‘replication crisis’ in psychology and other scientific disciplines (Ioannidis, 2012). A subsequent boost in meta-research found that many empirical results are not as robust as they originally seemed. Publication biases (Fanelli 2010), insufficient replication (Makel et al., 2012), lack of data sharing (Wicherts et al. 2006), questionable research practices (John et al. 2012), or low statistical power (Bezeau and Graves 2001) are just some of the factors identified as possible explanations for the crisis. However, the efforts so far seem to underestimate subjective influences such as personal expectations, use of terminology, as well as physical and psychological constraints in designing hypotheses and conducting experiments. In order to investigate these factors, meta-research must reach beyond quantitative methods and supplement them with a mixed-method approach. This means that the current investigation takes as a basis quantitative data (e.g. from text mining) and mixes it with qualitative material (e.g. analysis of context). Such an approach may be especially helpful in identifying and reflecting upon the structure and function of non-epistemological values or norms, such as ethical, social, or political considerations (Douglas 2016). Given those non-epistemological norms on the one side, epistemological norms, such as ‘reproducibility’, ‘scope’, and ‘transparency’, on the other side, are an accepted integral part of scientific reasoning (Douglas 2009: 17). According to this, it is only the first group of norms that is sometimes perceived as threat to scientific objectivity (see Hudson 2016). From the perspective of a value-free ideal, non-epistemological norms (i.e. moral considerations) should not affect scientific practice, because their subjective element is a ‘remit of art, not of science’ (Mogie 2000: 869). However, several classic studies (Feyerabend 1975; Latour and Woolgar 1979) as well as some recent publications (Elliott and McKaughan 2009, Davis 2013; Douglas 2016; Mascolo 2016) question the value-free ideal. In general, so the criticism goes, scientists are part of society and therefore inextricably linked to its values (Douglas 2016). As a consequence, any description of human or non-human behaviour that goes beyond mere observations draws inevitably on the bias of preconceptions: ‘The privileging of measurement over meaning puts the empirical cart before the conceptual horse.’ (Mascolo 2016: 5). Following this line of reasoning, not only is data input influenced by subjective values such as ‘preference for similar others’ (‘homophily’; Haun and Over 2013) or a priori rejection of ‘human-animal similarity’ (‘anthropodenial’; de Waal 1999), but scientists’ data output also has consequences in the social and ethical domain (Douglas 2009: 115). Illustrative examples can be found with reference to sign languages. Until the mid-20th century, a dominant preconception in science understood the oral modality as a necessary prerequisite for having ‘language’, which itself was supposed to be responsible for rationality and flexible communication (Ullrich 2016: 185). As a consequence, deaf humans were forced to learn oral forms of communication instead of better-suited sign languages, with various negative consequences for decades (Ullrich 2016: 189). One cannot simply blame scientific concepts and the use of terminology for those developments, but they did play a major role. Values in science become more important where ‘social categories and the images they embed are inescapably value-laden’ (Davis 2013: 554). We believe this also to be the case with ‘evolution of language and communication’ discourse, where scientists try to create a valid human self-conception with reference to a supposedly human unique characteristic, namely language (e.g. Berwick et al. 2013; Hauser et al. 2014; Scott-Phillips 2015). Given the entanglements between science and non-epistemological values or norms, it appears to us more productive to monitor norms instead of combating them. In that respect, we want to qualify and quantify one potentially lasting norm in order to enable future investigations towards experimental design, the formulation of questions and subsequent interpretation of data. As such, the study contributes to the process of scientific self-correction. The potential norm at issue is the ‘norm of progress’ (also known as ‘Scala Naturae’ or ‘Great Chain of Being’), which assumes that evolution proceeds in a linear ‘upward’ way from a simple/primitive condition towards an ‘improved’ state. Although modern evolutionary theory rejects this prediction (Johnson et al. 2012), a number of scientists complain about the persistence of the norm (Chittka et al. 2012: 2678; Cimatti and Vallortigara 2015: 6; de Waal 1999: 257; Emery and Clayton 2004: 37; Fitch et al. 2010: 796; Nee 2005). A number of qualitative studies focus on the history and current influence of the ‘norm of progress’ (Ghiselin 2005; Hodos and Campbell 1969; Lovejoy 1936; Ruse 1996). By design they do not quantify the phenomenon in recent discourse. Thus, despite the frequent complaints regarding the persistence of the ‘norm of progress’, to date there are only two attempts to study the existence of this norm in more quantitative ways. In 2000, Mogie searched scientific papers published between 1995 and 1999 using the attributes ‘higher’ or ‘lower’ in descriptions of species within the title. A low-tech query returned over 700 positive hits, mostly in studies of plants (n = 665) (Mogie 2000). Following that study, in 2013, Rigato and Minelli performed a scientometric analysis of over 67,413 biological articles published between 2005 and 2010 in 16 different scientific journals. Their queries on journal websites identified 1,287 out of 67,413 articles (1.91%) using ‘Scala Naturae language’ (Rigato and Minelli 2013). Another query in the course of the same study on PubMed confirmed that >55% of all positive hits derive from Botany (Rigato and Minelli 2013). Yet despite providing first evidence for possible implications of a non-epistemic norm within a discourse, neither study continues beyond overtly quantifiable issues, and both fail to identify any of the phenomena other academic peers have attributed to the realm of the norm. For instance, the historical exclusion of birdsong as a model of language (Sereno 2014: 5), addressed by qualitative research, escaped these quantitative accounts. Given the prevailing value-free ideal (Reiss and Sprenger 2014: Chapter 3.1), it is assumed that non-epistemological norms are mostly deployed unintentionally and not overtly, and are therefore difficult to identify. For these reasons, the current study aims to go beyond easily accessible ‘higher/lower classifications’. Instead, it quantifies implicit indicators of a ‘norm of progress’ in peer-reviewed publications on language/communication across a variety of species groups. In order to do that we intent to divide articles from the evolution of language discourse into two distinct corpora: ‘language’ and ‘communication’. The only reason we count an article to the corpus ‘language’ lies in the presence of the predefined terms ‘language’ or ‘speech’ in abstract, title, or keywords. On the other side, articles of corpus ‘communication’ use terms like ‘signal’, ‘song’, ‘vocalization’, ‘gesture’, or ‘communication’. The categorization makes no assertion about the actual focus of a publication. We are aware that sometimes the terms ‘language’ and ‘communication’ are used in similar ways within the current study. It is not our attempt to equate these terms. However, definitions of ‘language’ are notoriously diverse (Botha 2000). That includes perspectives which see ‘communication’ as mere side-effect of ‘language’ (e.g. Chomsky 2011: 264–65), as well as the opposite claim that regards communication as main driver for ‘language evolution’ (e.g. Okanoya 2017; Zuberbühler 2013: 188). Still, others interpret ‘language’ as part of a broader ‘communicative toolkit’ which also includes music and animal song (Rohrmeier et al. 2015). In general, we use a broad definition of ‘language’ that includes various cognitive (e.g. learning and memory) and physiological mechanisms (e.g. perception and motor control) (Fitch 2017: 5). The reason for dividing all articles in two corpora is the following: we hypothesize that authors using the word ‘language’ at prominent sections of an article, implicitly tie their research to a more human-centred perspective of research than researchers avoiding the term. If a ‘norm of progress’ exists, we would expect an increase of ‘progressionist vocabulary’ in the corpus ‘language’. ‘Progressionist vocabulary’, like ‘higher’ or ‘sophisticated’, implies the existence of an improved, more sophisticated or more complex ‘end state’ (mostly realised in humans). Since evolutionary theory is not based on a teleological framework, an ‘end state’ cannot exist and the ranking of structures or abilities along a scale of improvement appears mostly human-centred and/or arbitrary. Therefore, use of ‘progressionist language’ is not only ineloquent, but also value-laden. Hence, we assume that if the ‘norm of progress’ exists, we should find biased sampling of study species in the corpus ‘language’ compared to the corpus ‘communication’. Within the total of 915 journal articles, we expect to identify value-laden ‘progressionist’ vocabulary, dependent on species, article format or corpus group. 2. Material and methods In order to gather corpus material, we performed search queries on the citation database ‘Scopus’. Our aim was to identify a specific fraction of articles concerned with evolution of language from a species-comparative point of view. We chose to select those specific articles for two reasons: first, in both past and current debates it is notoriously difficult to identify a generally accepted definition of ‘language’ (Botha 2000), which makes the whole field of research an ideal candidate for speculation and value-laden narratives. Second, one controversial point in research on the ‘language origin’ concerns the question as to whether ‘language’ evolved either continuously across species (Wilcox 1999; Hurford 2014) or abruptly in human beings (Berwick et al. 2013). An answer might have wide-ranging implications for the human self-concept and, thus giving reason to expect value-loading on that issue in particular. For the years 2005–15, we selected from 16 Journals that have a high impact in the particular field of research (Table 1). Articles using ‘language’ or ‘speech’ in their abstract, title, or as keywords are collected in a corpus termed ‘language’ (n = 890). To contrast the results, we also wanted to identify publications focussing on communication, signal, song, gesture, or vocalization. Articles using one of those terms in their abstract, title, or as keywords are collected in a corpus termed ‘communication’ (n = 1,107).1 All articles examined were manually checked for relevance by reading abstracts and key words. Articles were included in the corpus of investigation when they fulfilled the following requirements: they (i) use a comparative, cross-species approach; (ii) focus on language/communication (not cognition in general); (iii) focus on biological evolution (i.e. exclude machines); (iv) consider multicellular organisms (but not plants, fungi), and (v) focus on inter-individual communication. Table 1. Composition of corpus ‘language’ and ‘communication’ by journal. Since two journals were founded in 2010 and 2011, respectively, they were not available for analysis before that year. Furthermore, publications from Behav. Brain. Sci. were not available as full text HTML before 2006 and thereby excluded for 2005. Journal name  No. of papers in corpus ‘language’  No. of papers in corpus ‘communication’  Anim. Behav.  36  205  Anim. Cogn.  22  17  Behav. Brain. Sci.  106  7  Curr. Anthropol.  37  4  Curr. Biol.  44  43  Evol. Hum. Behav.  9  2  Evol. Psychol.  4  1  Front. Psychol.  22  2  J. Comp. Psychol.  8  1  Nat. Commun. (*2010)  9  7  Phil. Trans. R. Soc. B  33  11  PLoS Biol.  9  2  PLOS ONE  42  70  PNAS  33  18  Proc. R. Soc. B  19  74  Sci. Rep. (*2011)  6  12  Journal name  No. of papers in corpus ‘language’  No. of papers in corpus ‘communication’  Anim. Behav.  36  205  Anim. Cogn.  22  17  Behav. Brain. Sci.  106  7  Curr. Anthropol.  37  4  Curr. Biol.  44  43  Evol. Hum. Behav.  9  2  Evol. Psychol.  4  1  Front. Psychol.  22  2  J. Comp. Psychol.  8  1  Nat. Commun. (*2010)  9  7  Phil. Trans. R. Soc. B  33  11  PLoS Biol.  9  2  PLOS ONE  42  70  PNAS  33  18  Proc. R. Soc. B  19  74  Sci. Rep. (*2011)  6  12  Relevant articles (‘language’ n = 439; ‘communication’ n = 476) were supplemented with meta-information such as (a) species focus, (b) modality, and (c) full-text download link. With regard to (a) nine groups of species were identified (1. human primate, 2. non-human primate, 3. non-primate mammals, 4. marine mammal, 5. bird, 6. other vertebrates, 7. invertebrate, 8. fish, 9. unspecified). With reference to (b) seven modalities were identified (1. acoustic, 2. visual, 3. chemical, 4. tactile, 5. thermal, 6. cross-modal, 7. multimodal). Most articles were automatically retrieved2 based on their link, converted from source HTML into a raw text format, and broken down to the level of individual words. Specific word classes were attributed automatically via TreeTagger using default settings (Schmid 1995). In addition to obvious lemmas like ‘high’ and ‘low’ used by Rigato and Minelli 2013, we consider a greater number of terms as contributing to a ‘norm of progress’. We created two groups of 56 handpicked lemmas (see Supplementary Material) to investigate the use of ‘progressionist vocabulary’, i.e. words that in a broader sense allow a linear differentiation between ‘high’ and ‘low’. Those potentially value-laden lemmas were identified by earlier research as relating to the ‘norm of progress’ (Güntürkün and Bugnyar 2016; Jarvis et al. 2005; Karten 2015; McShea 2011; Ruse 1996; Ullrich 2016) or were mentioned within an open survey by members of the Comparative Developmental Psychology group in Berlin (see Supplementary Material). For brevity, we named those word groups ‘high’ and ‘low’, respectively and used them in order to compare the appearance of lemmas between corpora and various meta-data. All quantitative analyses were performed using R 3.2 (R Development Core Team, 2016). A list of additional R-packages in use can be found at the Supplementary Methods section. To capture even subtle indicators of the ‘norm of progress’, the study combines quantitative text analysis and a qualitative audit of context (a mixed-methods approach). For qualitative analysis of context, we extracted respective text snippets into Excel Sheets and rated for context manually (i.e. ‘opposite meaning’, ‘species related’, ‘neutral’). All R-Scripts used and consulted material are open and can be downloaded (DOI 10.17605/OSF.IO/EGFHV). 3. Results and discussion 3.1 Authors mostly avoid direct linkage of ‘high/low’ to various species groups Rigato and Minelli (2013) concluded that ‘the great chain of being is still there’. When we reproduced their methodology for 915 publications from our corpus, we could identify 8 cases of direct linkage between ‘high’ and several species, but could not find any incidence with ‘low’. Hits from ‘higher’ linked to either ‘vertebrates’ (Earley 2010: 2676; Hauser et al. 2014: 1; Iriki and Taoka 2012: 18) or ‘primates’ (Cunningham and Ramos 2014: 806; Glickstein 2007: 824; Jablonka and 2012: 2155; Sadagopan et al. 2015: 10) with one exception of ‘plants’ (Caulier et al. 2013: 1). By definition, publications from the field of botany were excluded, whereby 0.87% positive hits from 915 articles nearly resembles those botany-free results presented by Rigato and Minelli (2013). Contrary to their interpretation, we do not conclude that results can lead us to state that researchers adhere to a ‘norm of progress’. In all affected articles, we could identify only one or two singular events linking ‘high + species’. When checking those papers manually, we could not identify a systematic use of ‘Scala Naturae language’. Instead, we consider those findings as singular cases of ‘historical baggage’ (Mogie 2000: 868) where expressions and metaphors echo a long tradition of teleological thinking. However, as previously mentioned, we did not assume that the linking of overt ‘high’ and ‘low’ classifications with various species would occur at a high frequency, since we expected non-epistemological norms to be mostly used unintentionally and therefore not overtly expressed in the text. This is why we started exploratory investigations for more implicit indicators that might impact the discourse. 3.2 Primates dominate corpus ‘language’ In 2014, Sereno claims that ‘birdsong has often been dismissed as a model of human language for the reason that monkeys seem much smarter than some birds’ (Sereno 2014: 5). We wanted to quantify his complaint regarding ‘Scala Naturae thinking’ by checking its substance in current literature. As described in our methods section, we divided all articles into two corpora labelled ‘language’ and ‘communication’, respectively. Subsequently, we decided to compare the range of studied species groups between the corpus ‘language’ and that of ‘communication’. We found a substantially wider range of studied species groups in the corpus ‘communication’ as compared to the corpus ‘language’ (Fig. 1). About 70% of all 439 articles using the terms ‘language’ or ‘speech’ in title, abstract or keywords focussed on primates. Broken down to specific groups we observed for the corpus ‘language’ that the majority of articles focussed specifically on human primates (38%), followed by non-human primates (32%), birds (11%), and finally publications without definite species focus (7.2%). In contrast, articles within the corpus ‘communication’ mostly focussed on invertebrates (26.89%) and birds (26.68%), followed by other vertebrates and non-human mammals (both 10.29%). From an overall 476 articles investigating communication and its evolution, only 11 focussed on human (2.3%) and 40 on non-human primates (8.4%). Figure 1. View largeDownload slide Comparison of the range of studied species groups in corpus ‘language’ and ‘communication’. White horizontal line in primate box subdivides the group into human (above broken line) and non-human primates (below). Figure 1. View largeDownload slide Comparison of the range of studied species groups in corpus ‘language’ and ‘communication’. White horizontal line in primate box subdivides the group into human (above broken line) and non-human primates (below). The results pertaining to humans in the corpus ‘language’ may not surprise, since many researchers regard language to be unique to them (e.g. Berwick et al. 2013; Hauser et al. 2014; Scott-Phillips 2015). Nonetheless, almost 62% of all articles using ‘language’ in their title, abstract, or keywords do focus on non-human animals, most of them on non-human primates. For our study, it is of no importance to distinguish if those articles investigate the origin of ‘language’ or ‘communication’. We also cannot distinguish between the opposite use of the term or its context. Apart from these issues, it strikes us that articles focussing on invertebrates, fish, or other vertebrates avoided almost completely the term ‘language’ in their opening sections. Even when most articles on non-human primates in the corpus ‘language’ investigated the ‘origin of communication’ instead of ‘language’, it surprises us that only 8.4% of articles from the corpus ‘communication’ are dedicated to the non-human primate group. That leads us to the conclusion that researchers studying primates are more likely to use the term ‘language’ when investigating communicative behaviours than researchers concerned with other species groups. Given Sereno’s statement (Sereno 2014), we indeed conclude that articles from the corpus ‘language’ tended in relative numbers to ‘dismiss’ birds as a model for investigating the evolution of language. What might explain this phenomenon? Researchers tend to see abilities that they value, which is more easily done in species that closely resemble humans, e.g. primates. For instance, the oral/acoustic modality of human communication is the subject of 58.5% of the studies within the corpus ‘language’. Modalities presumably less relevant to average humans or multimodal accounts that received attention only recently are covered comparably less (cross-modal: 22.3%; visual: 11.2%; multimodal: 7.1%; chemical: <1%; see Supplementary Fig. S1). Earlier studies have traced some historical sources of the phenomenon’s origin such as an ’oral norm’ (Ullrich 2016), ‘a priori biases’ (Slocombe et al. 2011), ‘Primatocentrism’ (Cimatti and Vallortigara 2015; Emery and Clayton 2004), or ‘Chimpocentrism’ (Vaesen 2014). Frequent focus on primates’ unimodal behaviour in the early days of comparative communication studies might have caused an underestimation of the communicative abilities of non-primates, which in turn makes non-primate research look less interesting. The circle creates its own evidence and fuels a view of ascending ‘complexity’ over the course of evolutionary development. In accordance with this interpretation, we examined if both corpora would differ in their use of directional language. Since in the corpus ‘language’, there are more species investigated historically considered ‘high’ than in the corpus ‘communication’, we expect to find more adjectives representing ‘high’ in the corpus ‘language’ than the other way around. We thus decided to identify the 80 most common adjectives used in both corpora. 3.3 A selection of the 80 most common adjectives hints towards substantially different narratives across the two corpora Adjectives give more precise information about a particular object of interest. Therefore, in case the ‘norm of progress’ influences scientific publications, we would expect more adjectives implying ‘high’ in the corpus ‘language’ as compared to the corpus ‘communication’. This expectation is based on the following reflection: if articles from the corpus ‘language’ focus mostly on species groups that were considered ‘high’ under the terms of a ‘norm of progress’, and articles from the corpus ‘communication’ deal with ‘lower’ ones, adjectives implying ‘high’ should appear more frequently in corpus ‘language’. However, the analysis for the 80 most frequent adjectives did not meet our initial expectations. Indeed, the adjective ‘complex’ occurs more often in ‘language’ as compared to ‘communication’, while adjective ‘low’ followed the opposite pattern (Fig. 2). However, we were not able to detect any structural regularity that would systematically ascribe ‘high’ or ‘low’ value-laden adjectives to any of the corpora. Instead, we became interested in those adjectives without respective counterparts within the list. Figure 2. View largeDownload slide List of the 80 most common adjectives of the respective corpus, ordered by their occurrence. Adjectives that appear on either side are linked by lines. Adjectives without line do not have a respective counterpart among the most frequent 80. Figure 2. View largeDownload slide List of the 80 most common adjectives of the respective corpus, ordered by their occurrence. Adjectives that appear on either side are linked by lines. Adjectives without line do not have a respective counterpart among the most frequent 80. With regard to the corpus ‘language’, examples of some frequently employed adjectives are: cognitive, linguistic, communicative, neural, functional, cultural, syntactic, gestural, and semantic. With regard to the corpus ‘communication’, some examples are: sexual, reproductive, sensory, aggressive, conspecific, facial, territorial, and dominant. It appears to us that those words tell very different stories about similar observations. One (corpus ‘communication’) investigates the communicative behaviour of a species for the sake of the species itself, while the other corpus (‘language’) aims to compare communicative behaviour between non-human and human animals. Articles using the term ‘language’ in the abstract, title or key words tend to link and compare their findings to cognition and linguistic concepts, aspects that were investigated in former times under the umbrella term ‘animal psychology’. Articles avoiding ‘language’ concentrate on ecology and ethology, aspects that are investigated under the umbrella term ‘behaviourism’. However, since extracting the 80 most common adjectives did not answer the question as to whether one of the corpora would feature the more frequent deployment of ‘high’ or ‘low’ classifications, we then decided to directly create a list of target words with the objective of comparing them accordingly. 3.4 No difference in directional vocabulary between corpus, but between species group and articles type Due to the different emphasis on species groups between the corpora and the identification of two diverging uses of vocabulary when writing up results, we were interested in whether a selected list of words could also reveal a difference in the use of lemmas classified as ‘high’ or ‘low’. We predicted that under the terms of a persistent ‘norm of progress’, articles in the corpus ‘language’ would use more lemmas implying values of ‘high’ while avoiding those implying ‘low’, as compared to the corpus ‘communication’. To quantify frequencies of word appearances, we created a list of 58 words which either imply evolutionary ‘improvement’ or ‘simplicity’. The choice of words was based on earlier research (Güntürkün and Bugnyar 2016; Jarvis et al. 2005; Karten 2015; McShea 2011; Ruse 1996; Ullrich 2016) and an open survey among researchers in comparative psychology (see Supplementary Material). To account for different text length, we corrected all hits by the total number of words per article. Altogether we found that words of the category ‘high’ are used 40% more often in the corpus ‘language’ and 32% more often in the corpus ‘communication’ as compared to words of category ‘low’. However, the difference of direct hits between the corpora was not as clear as expected. Indeed, publications of corpus ‘language’ did use words classified as ‘high’ 10.51% more often and words classified as ‘low’ 1.55% less often as compared to the corpus ‘communication’. When related to other factors such as ‘species group’ and ‘article type’, these results shift in weight and appear rather comparable. Indeed, relative frequency differed substantially between various groups of species (Fig. 3). Publications in the corpus ‘language’ focussing on non-human primates use vocabulary from the word group ‘high’ more often than, for instance, articles focussing on birds (+27% in corpus ‘language’; +23% in corpus ‘communication’). Similarly, articles in the corpus ‘language’ without any focus on a species group used words classified as ‘high’ with increasing frequency as compared to articles with a focus on birds (+35% in corpus ‘language’; +18% in corpus ‘communication’). In general, we observed a tendency by which articles focussing on species groups ranking ‘high’ according to a ‘norm of progress’ increase their use of words valuing ‘high’. The small sample for articles focussing on humans (n = 10) in the corpus ‘communication’ constitute an exception to this observation. Since many articles with ‘unspecified’ species groups appear to consist of comments, review or theory pieces, we also wanted to quantify differences for that factor. We found that not only does the species group influence linguistic usage, but also the article format (Fig. 4). In general, we observed a tendency whereby articles with less experimental or empirical focus increase their use of words defined as ‘high/low’. For instance, in articles classified as theoretical publications we identified an increase of words categorised as ‘high’ by 23.6% (corpus ‘language’) and 26.5% (corpus ‘communication’), respectively. Following various conference discussions, most scientists eagerly deny reference to any such ‘norm of progress’.3 As said earlier mostly there is no active promotion for such a norm as it appears ‘unscientific’ (Mogie 2000: 869). Of course, a quantitative text analysis cannot determine to what degree vocabulary is used deliberately, or in which context. For that reason we checked context for one specific word that scientists usually value: ‘unique’. Figure 3. View largeDownload slide Occurrence of words classified as ‘high’ (dark-grey) and ‘low’ (grey) computed per 1,000 words of original article. Results are broken down for species group and ordered by their summarized mean for ‘high’. Horizontal bars indicate mean for all articles of a document type. Numbers in brackets represent articles under investigation. Error bars depict the standard error. Figure 3. View largeDownload slide Occurrence of words classified as ‘high’ (dark-grey) and ‘low’ (grey) computed per 1,000 words of original article. Results are broken down for species group and ordered by their summarized mean for ‘high’. Horizontal bars indicate mean for all articles of a document type. Numbers in brackets represent articles under investigation. Error bars depict the standard error. Figure 4. View largeDownload slide Occurrence of words classified as ‘high’ (dark-grey) and ‘low’ (grey) computed per 1,000 words of original article. Results are broken down for article type and ordered by their summarized mean for ‘high’. Horizontal bars indicate mean for all articles of a species group. Numbers in brackets represent articles under investigation. Error bars depict the standard error. Figure 4. View largeDownload slide Occurrence of words classified as ‘high’ (dark-grey) and ‘low’ (grey) computed per 1,000 words of original article. Results are broken down for article type and ordered by their summarized mean for ‘high’. Horizontal bars indicate mean for all articles of a species group. Numbers in brackets represent articles under investigation. Error bars depict the standard error. 3.5 ‘Language’ more unique than ‘communication’ In order to evaluate our previous results, we wanted to approach the problem of context blindness for one case example. We chose the lemma ‘unique’, because usually its usage does not imply directional connotations. Furthermore, from a biological point of view there is nothing special about being ‘unique’, since every species is defined by its autapomorphy—a derived trait that defines the status as a species. However, based on previous qualitative research we hypothesized that when publications repeatedly state something as uniquely human, but do not mention anything else as unique in non-humans, than this one-sided view might hint towards values in use. In order to test the hypothesis we first quantified the phenomenon and subsequently qualified the results. We found that 52% of all articles in the corpus ‘language’ make use of the lemma ‘unique’, but only 37% of articles in the corpus ‘communication’. If the lemma is used, articles in the corpus ‘language’ use it on average 3.2 times, while articles in the corpus ‘communication’ employ it 1.9 times. Altogether we identified 57% more instances of the lemma ‘unique’ in the corpus ‘language’ as compared to ‘communication’. In order to investigate the qualitative context of the lemma, we extracted all occurrences, including the context, and validated its usage. When ‘unique’ referred to any species group, we labelled it accordingly. When used to the contrary (e.g. ‘not unique’), we labelled it ‘opposite’. When used without reference to any species, we labelled it ‘neutral’. When used in context of an unanswered question or within quotations, we labelled it ‘undecided’. We found that in nearly half the cases from the corpus ‘language’ the lemma ‘unique’ referred to humans, while in 78% of all incidences in the corpus ‘communication’ it was used in neutral manner (Fig. 5). Figure 5. View largeDownload slide Comparison of context from the lemma ‘unique*’ {including: ‘uniquely’ and ‘uniqueness’} between corpora. In the corpus ‘communication’, the lemma ‘unique*’ is mostly used in neutral context, whereas its use in the corpus ‘language’ refers in almost half of the incidences to humans. See text for details and definition of individual labels. Figure 5. View largeDownload slide Comparison of context from the lemma ‘unique*’ {including: ‘uniquely’ and ‘uniqueness’} between corpora. In the corpus ‘communication’, the lemma ‘unique*’ is mostly used in neutral context, whereas its use in the corpus ‘language’ refers in almost half of the incidences to humans. See text for details and definition of individual labels. Certainly, it might come of no surprise that the term ‘unique’ appears frequently compared to ‘human’, since language is regarded as one of the important autapomorphies of the species. However, all species-specific forms of communication are unique by definition. Either someone takes the view that human language is unique and thus not comparable to any non-human form of communication, or one conducts species comparative research and therefore allows a comparison of language and animal communication. When following the second strategy, the consequence is that not only language is unique to humans, but also ultrasonic social communication to bats, electric communication signals to electric fish, and multimodal chemo-acoustic signals to lemurs. Still, in only 44 cases, ‘unique’ relates to the behaviour of a species in the corpus ‘communication’, while we could find 388 such cases (mostly in reference to humans) in the corpus ‘language’. That might constitute a scientific narrative that justifies human speciality as an evolutionary ‘achievement’. As such it hints towards a somewhat chauvinistic function where non-human species are not actively discriminated, but implicitly eclipsed. While scientists highlight human uniquely features, they also feel the urge to find biological ‘roots’ of behaviours and thus start testing and observing ‘downwards’ along the ‘evolutionary tree’. In this respect, such a research agenda could be classified as motivated by the vestiges of a historical ‘norm of progress’. After all, non-epistemological norms do indeed play a role within scientific reasoning (see Douglas 2000). However, the task of monitoring them is always valuable and never completed, enabling readers to develop a critical view of hypotheses, questions, and results. 4. Conclusion In order to quantify a possible non-epistemological ‘norm of progress’ within a current scientific discourse of language evolution, we applied a quantitative text and qualitative context analysis to a corpus consisting of 915 articles. Historically one can find clear evidence for the existence of a ‘norm of progress’ in scientific publications. A reproduction of a study by Rigato and Minelli (2013), however, could show only minor evidence for an open and active promotion of that norm. Hence the focus of subsequent tests was put on implicit factors such as species range, use of vocabulary and values in language. Although papers from corpus ‘language’ and ‘communication’ focus on a similar phenomenon, their narratives appear strikingly different as indicated by the frequency of 80 of the most commonly employed adjectives. In addition, both corpora differ widely in their range of studied species groups and the usage of the lemma ‘unique’. In all cases, the corpus ‘language’ establishes a narrative of human speciality, compared to other species, as could be additionally shown by qualifying all uses of the lemma ‘unique’ within the corpus. However, both corpora use more frequently words in the category ‘high’ with reference to primates as compared to birds or insects. Taken together, there is no evidence for a structural and overt promotion of a non-epistemological ‘norm of progress’ within the discourse. Still, several implicit factors hint at the lingering historical aftermaths of norm-related ideas and an associated subconscious function as leading forces in identifying and formulating current and future research questions. Supplementary data Supplementary data is available at Journal of Language Evolution online. Conflict of interest statement. None declared. Authors’ contributions R.U. conceived and designed the study, collected corpus material, did qualitative analysis, drafted the manuscript, analysed and interpreted quantitative data and wrote parts of R code. M.M. substantially participated in quantitative data analysis & wrote the majority of R code. K.L. did critical revision of article drafts and provided important intellectual content. All authors gave final approval for publication. Funding R.U. and M.M. received no specific funding for this work. K.L. receives a DFG Excellence Initiative Grant. Research ethics The study did not require ethical approval from a local ethics committee. Data availability Data and research materials supporting the results in the article are open and available at the Open Science Framework: doi 10.17605/osf.io/egfhv. Footnotes 1 One example (for more see DOI 10.17605/OSF.IO/EGFHV) of a ‘Scopus’ query for the Journal Animal Cognition contributing to the corpus ‘communication’: TITLE-ABS-KEY (communication OR song OR signal* OR vocali?ation OR gesture AND evol* AND NOT language AND NOT speech) AND ISSN (1435-9456) OR ISSN (1435-9448) AND PUBYEAR AFT 2004 AND PUBYEAR BEF 2016 2 Due to technical oddities this procedure had to be done by hand for two Journals: “Journal of Comparative Psychology” & “Current Anthropology”. 3 e.g. personal communication to Andrew Whiten, T. Scott-Phillips. Acknowledgements We thank Prof. Markus Wild and participants from the Workshop ‘Minds of Animals’ in Bern 2016 for helpful comments. References Berwick R. C. et al.   ( 2013) ‘ Evolution, Brain, and the Nature of Language’, Trends in Cognitive Sciences , 17/ 2: 98. http://doi.org/10.1016/j.tics.2012.12.002. Google Scholar CrossRef Search ADS   Bezeau S., Graves R. ( 2001) ‘ Statistical Power and Effect Sizes of Clinical Neuropsychology Research’, Journal of Clinical and Experimental Neuropsychology , 23/ 3: 399– 406. http://doi.org/10.1076/jcen.23.3.399.1181. Google Scholar CrossRef Search ADS PubMed  Botha R. ( 2000) ‘ Discussing the Evolution of the Assorted Beasts called Language’, Language & Communication , 20: 149– 60. http://doi.org/10.1016/S0271-5309(99)00022-1. Google Scholar CrossRef Search ADS   Caulier G. et al.   ( 2013) ‘ When a Repellent Becomes an Attractant: Harmful Saponins are Kairomones Attracting the Symbiotic Harlequin Crab’, Scientific Reports , 3: 2639. http://doi.org/10.1038/srep02639. Google Scholar CrossRef Search ADS PubMed  Chittka L. et al.   ( 2012) ‘ What is Comparable in Comparative Cognition?’, Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences , 367/ 1603: 2677– 85. http://doi.org/10.1098/rstb.2012.0215. Google Scholar CrossRef Search ADS PubMed  Chomsky N. ( 2011) ‘ Language and Other Cognitive Systems. What Is Special About Language?’, Language Learning and Development , 7/ 4: 263– 78. http://doi.org/10.1080/15475441.2011.584041. Google Scholar CrossRef Search ADS   Cimatti F., Vallortigara G. ( 2015) ‘ So Little Brain, so much Mind. Intelligence and behaviour in Nonhuman Animals’, Reti, Saperi, Linguaggi , 4/ 7: 5– 20. http://doi.org/10.12832/81287. Cunningham C. L., Ramos M. F. ( 2014) ‘ Effect of Training and Familiarity on Responsiveness to Human Cues in Domestic Dogs (Canis familiaris)’, Animal Cognition , 17: 805– 14. http://doi.org/10.1007/s10071-013-0714-z. Google Scholar CrossRef Search ADS PubMed  Davis J. E. ( 2013) ‘ Social Science, Objectivity, and Moral Life’, Society , 50/ 6: 554– 9. http://doi.org/10.1007/s12115-013-9710-9. Google Scholar CrossRef Search ADS   de Waal F. B. M. ( 1999) ‘ Anthropomorphism and Anthropodenial: Consistency in Our Thinking about Humans and Other Animals’, Philosophical Topics , 27/ 1: 255– 80. Google Scholar CrossRef Search ADS   Douglas H. ( 2000) ‘ Inductive Risk and Values in Science’, Philosophy of Science , 67/ 4: 559. http://doi.org/10.1086/392855. Google Scholar CrossRef Search ADS   Douglas H. ( 2009). Science, Policy, and the Value-free Ideal . Pittsburgh: University of Pittsburgh Press. Google Scholar CrossRef Search ADS   Douglas H. ( 2016). ‘Values in Science’, in Humphreys P. (ed.) The Oxford Handbook of Philosophy of Science , 609– 32. Oxford, New York: Oxford University Press. http://doi.org/10.1093/oxfordhb/9780199368815.013.28. Earley R. L. ( 2010) ‘ Social Eavesdropping and the Evolution of Conditional Cooperation and Cheating Strategies’, Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences , 365/ 1553: 2675– 86. http://doi.org/10.1098/rstb.2010.0147. Google Scholar CrossRef Search ADS PubMed  Elliott K. C., McKaughan D. J. ( 2009) ‘ How Values in Scientific Discovery and Pursuit Alter Theory Appraisal’, Philosophy of Science , 76/ 5: 598– 611. http://doi.org/10.1086/605807. Google Scholar CrossRef Search ADS   Emery N., Clayton N. S. ( 2004). ‘Comparing the Complex Cognition of Birds and Primates’, in Rogers L.J., Kaplan G. (eds) Comparative Vertebrate Cognition , pp. 3– 55. New York: Springer Science+Business Media. Google Scholar CrossRef Search ADS   Fanelli D. ( 2010) ‘ Do Pressures to Publish Increase Scientists’ Bias? An Empirical Support from US States Data’, PLoS ONE , 5/ 4, p. e10271. http://doi.org/10.1371/journal.pone.0010271. Feyerabend P. ( 1975). Against Method (3rd 1993) . London: Verso. Fitch W. T. ( 2017) ‘ Empirical Approaches to the Study of Language Evolution’, Psychonomic Bulletin & Review , http://doi.org/10.3758/s13423-017-1236-5. Fitch W. T., Huber L., Bugnyar T. ( 2010) ‘ Social Cognition and the Evolution of Language: Constructing Cognitive Phylogenies’, Neuron , 65/ 6: 795– 814. http://doi.org/10.1016/j.neuron.2010.03.011. Google Scholar CrossRef Search ADS PubMed  Ghiselin M. T. ( 2005) ‘ The Darwinian Revolution as Viewed by a Philosophical Biologist’, Journal of the History of Biology , 38/ 1: 123– 36. http://doi.org/10.1007/s10739-004-6513-2. Google Scholar CrossRef Search ADS PubMed  Glickstein M. ( 2007) ‘ What does the Cerebellum really do?’, Current Biology , 17/ 19: 824– 7. http://doi.org/10.1016/j.cub.2007.08.009. Google Scholar CrossRef Search ADS   Güntürkün O., Bugnyar T. ( 2016) ‘ Cognition without Cortex’, Trends in Cognitive Sciences, Xx , 1– 13, http://doi.org/10.1016/j.tics.2016.02.001. Haun D., Over H. ( 2013). ‘Like me: a Homophily-Based Account of Human Culture’, in Richerson P. J., Christiansen M. H. (eds), Cultural Evolution: Society, Technology, Language, and Religion ,pp. 75– 85. Cambridge: MIT Press. Hauser M. D. et al.   ( 2014) ‘ The Mystery of Language Evolution’, Frontiers in Psychology , 5(MAY): 1– 12. http://doi.org/10.3389/fpsyg.2014.00401. Hodos W., Campbell C. G. B. ( 1969) ‘ Scala Naturae: Why there is no Theory in Comparative Psychology’, Psychological Review , 76/ 4: 337– 50. Google Scholar CrossRef Search ADS   Hudson R. ( 2016) ‘ Why we should not Reject the Value-Free Ideal of Science’, Perspectives on Science , 24/ 2: 167– 91. http://doi.org/10.1162/POSC_a_00199. Google Scholar CrossRef Search ADS   Hurford J. R. ( 2014). Origins of Language: A Slim Guide . Oxford: Oxford University Press. Ioannidis J. P. A. ( 2012) ‘ Why Science is not Necessarily Self-Correcting’, Perspectives on Psychological Science , 7/ 6: 645– 54. http://doi.org/10.1177/1745691612464056. Google Scholar CrossRef Search ADS PubMed  Iriki A., Taoka M. ( 2012) ‘ Triadic (Ecological, Neural, Cognitive) Niche Construction: a Scenario of Human Brain Evolution Extrapolating Tool Use and Language from the Control of Reaching Actions’, Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences , 367/ 1585: 10– 23. http://doi.org/10.1098/rstb.2011.0190. Google Scholar CrossRef Search ADS PubMed  Jablonka E., Ginsburg S., Dor D. ( 2012) ‘ The co-evolution of Language and Emotions’, Philosophical Transactions of the Royal Society B: Biological Sciences , 367/ 1599: 2152– 9. http://doi.org/10.1098/rstb.2012.0117. Google Scholar CrossRef Search ADS   Jarvis E. D. et al.   ( 2005) ‘ Avian Brains and a New Understanding of Vertebrate Brain Evolution’, Nature Reviews. Neuroscience , 6(February): 151– 9. http://doi.org/10.1038/nrn1606. Google Scholar CrossRef Search ADS   John L. K., Loewenstein G., Prelec D. ( 2012) ‘ Measuring the Prevalence of Questionable Research Practices with Incentives for Truth Telling’, Psychological Science , 23/ 5: 524– 32. http://doi.org/10.1177/0956797611430953. Google Scholar CrossRef Search ADS PubMed  Johnson N. A., Lahti D. C., Blumstein D. T. ( 2012) ‘ Combating the Assumption of Evolutionary Progress: Lessons from the Decay and Loss of Traits’, Evolution: Education and Outreach , 5/ 1: 128– 38. http://doi.org/10.1007/s12052-011-0381-y. Google Scholar CrossRef Search ADS   Karten H. J. ( 2015) ‘ Vertebrate Brains and Evolutionary Connectomics: On the Origins of the Mammalian “Neocortex’, Philosophical Transactions of the Royal Society B: Biological Sciences , 370/ 1684: 20150060. http://doi.org/10.1098/rstb.2015.0060. Google Scholar CrossRef Search ADS   Latour B., Woolgar S. ( 1979). Laboratory Life. The Construction of Scientific Facts . Princeton, New Jersey: Princeton University Press. Lovejoy A. O. ( 1936). The Great Chain of Being. A Study of the History of an Idea (26th repri) . Cambridge, Massachusetts, London: Harvard University Press. Makel M. C., Plucker J. A., Hegarty B. ( 2012) ‘ Replications in Psychology Research: How Often Do They Really Occur?’, Perspectives on Psychological Science , 7/ 6: 537– 42. http://doi.org/10.1177/1745691612460688. Google Scholar CrossRef Search ADS PubMed  Mascolo M. F. ( 2016) ‘ How Objectivity Undermines the Study of Personhood: Toward an Intersubjective Epistemology for Psychological Science’, New Ideas in Psychology , 44: 41– 8. Google Scholar CrossRef Search ADS   McShea D. W. ( 2011). ‘Evolutionary Progress’, in Ruse M., Travis J. (eds), Evolution: The First Four Billion Years , pp. 550– 57. Cambridge, Massachusetts: Harvard University Press. Mogie M. ( 2000) ‘ Historical Baggage in Biology: The Case of “Higher” and “Lower” Species’, BioEssays , 22/ 9: 868– 69. http://doi.org/10.1002/1521-1878(200009)22:9<868::AID-BIES13>3.0.CO;2-A. Google Scholar CrossRef Search ADS PubMed  Nee S. ( 2005) ‘ The Great Chain of Being’, Nature , 435(May): 429. http://doi.org/10.1177/00221678930333006. Google Scholar CrossRef Search ADS   Okanoya K. ( 2017). Sexual Communication and Domestication may give rise to the Signal Complexity Necessary for the Emergence of Language: An Indication from Songbird Studies. Psychonomic Bulletin & Review, 24: 106– 10. R Development Core Team, R. ( 2016). R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing. Retrieved from http://www.r-project.org/ accessed 7 July 2016. Reiss J., Sprenger J. ( 2014). ‘Scientific Objectivity’, in Zalta Edward N. (ed.) The Stanford Encyclopedia of Philosophy  (Fall 2014). Retrieved from http://plato.stanford.edu/entries/scientific-objectivity/ accessed 10 May 2016. Rigato E., Minelli A. ( 2013) ‘ The Great Chain of being is still here’, Evolution: Education and Outreach , 6/ 18: 1– 6. http://doi.org/10.1186/1936-6434-6-18. Rohrmeier M. et al.   ( 2015) ‘ Principles of Structure Building in Music, Language and Animal Song’, Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences , 370/ 1664: 20140097. http://doi.org/10.1098/rstb.2014.0097. Google Scholar CrossRef Search ADS PubMed  Ruse M. ( 1996). Monad to Man. The Concept of Progress in Evolutionary Biology . Cambridge, Massachusetts, London, England: Harvard University Press. Sadagopan S., Temiz-Karayol N. Z., Voss H. U. ( 2015) ‘ High-field Functional Magnetic Resonance Imaging of Vocalization Processing in Marmosets’, Scientific Reports , 5: 10950. http://doi.org/10.1038/srep10950. Google Scholar CrossRef Search ADS PubMed  Schmid H. ( 1995). Improvements in Part-of-Speech Tagging with an Application To German. In Proceedings of the ACL SIGDAT-Workshop, pp. 1–9. Dublin, Ireland. Retrieved from http://www.cis.uni-muenchen.de/∼schmid/tools/TreeTagger/ accessed 8 September 2016. Scott-Phillips T. C. ( 2015) ‘ Nonhuman Primate Communication, Pragmatics, and the Origins of Language’, Current Anthropology , 56/ 1: 56– 80. http://doi.org/10.1086/679674. Google Scholar CrossRef Search ADS   Sereno M. I. ( 2014) ‘ Origin of Symbol-Using Systems : Speech, but not Sign, without the Semantic Urge’, Philosophical Transactions of the Royal Society B , 369(August): 20130303. http://dx.doi.org/10.1098/rstb.2013.0303. Google Scholar CrossRef Search ADS   Slocombe K. E., Waller B. M., Liebal K. ( 2011) ‘ The Language Void: the Need for Multimodality in Primate Communication Research’, Animal Behaviour , 81/ 5: 919– 24. http://doi.org/10.1016/j.anbehav.2011.02.002. Google Scholar CrossRef Search ADS   Ullrich R. ( 2016) ‘ From “Speech” to “Gesture”: The “Oral” as Norm in “Language” Research’, Interdisziplinäre Anthropologie Jahrbuch: Wahrnehmung , 4: 179– 208. http://doi.org/10.1007/978-3-658-14264-3. Vaesen K. ( 2014) ‘ Chimpocentrism and Reconstructions of Human Evolution (a Timely Reminder)’, Studies in History and Philosophy of Science Part C :Studies in History and Philosophy of Biological and Biomedical Sciences , 45/ 1: 12– 21. http://doi.org/10.1016/j.shpsc.2013.12.004. Google Scholar CrossRef Search ADS   Wicherts J. M. et al.   ( 2006) ‘ The Poor Availability of Psychological Research Data for Reanalysis’, The American Psychologist , 61/ 7: 726– 728. http://doi.org/10.1037/0003-066X.61.7.726. Google Scholar CrossRef Search ADS PubMed  Wilcox S. ( 1999). ‘The Invention and Ritualization of Language’, in King B. (ed.) The Origins of Language: What Nonhuman Primates Can Tell Us , pp. 351– 84. Oxford: James Currey Ltd. Zuberbühler K. ( 2013). ‘Primate Communication’, in Lefebvre C., Comrie B., Cohen H. (eds) New Perspectives on the Origins of Language , pp. 187– 210. Amsterdam: John Benjamins Publishing Co. http://doi.org/10.1075/slcs.144. Google Scholar CrossRef Search ADS   © The Author 2017. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com

Journal

Journal of Language EvolutionOxford University Press

Published: Jan 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off