Literal Language Scale: Development & Association With Social Intelligence

Literal Language Scale: Development & Association With Social Intelligence

Autism is known to involve use of language that is “literal” (Happe 1995), and while many scales have been developed to assess autistic traits, there has not (to my knowledge) been a scale developed to assess autistic use of language. Such a scale would be useful to study the role of language in social cognition and how its affected in autism, and to better understand autism. To address this, I created an 8 item questionnaire with questions intended to cover different aspects of literal language [Table 1].

Table 1

The Literal Language Scale	Explanation for item
People tell me I "take things too seriously"	Literal language may involve approaching social interaction in a serious, non-social way
I often miss the point of jokes	Most forms of humor involve complex social information, causing autistic persons to have difficulty understanding them
I often don't pick up when someone is being sarcastic	The detection of sarcasm involves the integration of various points of social information, which autistic persons are less able to pick up
Most words or phrases have only one meaning	Autistic semantic processing is “overly rigid” (Faust & Kenett 2014) which impairs ability to recognize semantic ambiguity
I have trouble "reading between the lines" of what other people say	Autistic social cognition is “hypo-intentional”, and because of this language is interpreted semantically rather than intentionally (Ciaramidaro et al. 2015)
Unless otherwise specified, I assume that people mean exactly what they say.	Autistic persons often interpret figurative language as literal (Happe 1995)
I stick to the technical meaning of words rather than what other people think they mean.	Language is a tool for social interaction, and autistic persons socially interact in a “non-social” way, which leads to language being treated as if it were technical
The meaning of a word is determined by its dictionary definition	^

I then conducted an online survey (n=1122) with PsyToolkit (Stoet 2010, 2017) which included: the literal language scale, the social intelligence scale (TSIS; Silvera et al. 2001), and the ten item autism quotient (AQ-10; Allison et al. 2012). Participants were recruited via an instagram post.

The literal language scale had borderline excellent internal consistency (a=0.89), and factor analysis found a single factor with all items loading highly on the scale (0.62-0.83). The scale correlated highly with the AQ-10 (r=0.60), and this remained for all items of the scale individually (r=0.32-0.64). The literal language scale was negatively correlated with social intelligence (r=-0.71), and this remained for the individual items (r=-0.41-0.68). [Table 2]

Table 2

Item	rTSIS	rAQ-10	PC1
People tell me I "take things too seriously"	-0.44	0.34	0.62
I often miss the point of jokes	-0.62	0.53	0.80
I often don't pick up when someone is being sarcastic	-0.55	0.49	0.76
Most words or phrases have only one meaning	-0.51	0.40	0.76
I have trouble "reading between the lines" of what other people say	-0.68	0.64	0.80
Unless otherwise specified, I assume that people mean exactly what they say.	-0.58	0.49	0.83
I stick to the technical meaning of words rather than what other people think they mean.	-0.49	0.39	0.77
The meaning of a word is determined by its dictionary definition	-0.41	0.32	0.66
Total scale	-0.71	0.60	a=0.89

The literal language scale and AQ-10 were both strong predictors of social intelligence and its three factors, both having almost equal correlations, except for the social awareness factor which literal language predicted r=0.13 higher than the AQ-10 [Table 3]. 

Table 3

	Literal Language	AQ-10
AQ-10	0.60
Social information processing	-0.58	-0.62
Social skills	-0.51	-0.48
Social awareness	-0.67	-0.55
Total social intelligence	-0.71	-0.69

Considering the effects of literal language and the AQ-10 independently, this difference becomes r=0.23, and literal language predicts overall social intelligence slightly better than the AQ-10 by r=0.09, literal language accounting for 15% of the variance in social intelligence and the AQ-10 accounting for 9%. 

Table 4

	Literal Language (independent of AQ-10)	AQ-10 (independent of language)
Social information processing	-0.26	-0.34
Social skills	-0.28	-0.22
Social awareness	-0.42	-0.19
Total social intelligence	-0.39	-0.30

The social awareness factor of the TSIS is similar in content to the social information processing factor, both seeming to reflect theory of mind or cognitive empathy. Based on item content, it seems that social awareness reflects the tendency for someone to automatically empathize with others and be implicitly aware of the mental states of others, whereas social information processing seems to reflect whether someone is able to understand another person's mental state. Low social awareness likely better reflects autism than other factors of the TSIS, as autism is not an inability to recognize another mental state, but an “egocentric” thinking style (Frith & Vignemont 2005) with reduced social brain activity in social situations (Stanfield et al. 2017) where autistic persons pay less attention to mental states, and have trouble accurately making sense of them. 

I argue that the literal language scale better reflects autism than other scales. For example, the autism quotient, while being scored higher in autistic persons, measures generalized social deficits which are present in many mental disorders (Crespi 2019) and does not assess the cause or qualitative nature of these social deficits. The literal language scale may more specifically measure social deficits related to autism. For example, both schizotypal and autistic persons have difficulty interpreting what others say to them, however in schizotypal personality disorder this is largely because of an over awareness of and difficulty resolving ambiguity (Grimshaw et al. 2010)(Faust & Kennet 2014), whereas in autism there is less awareness of ambiguity in meaning behind words and phrases (Faust & Kennet 2014), which is reflected in the item “Most words or phrases only have one meaning”. 

1. Happe (1995) Understanding Minds and Metaphors: Insights from the Study of Figurative Language in Autism

2. Stoet (2010). PsyToolkit - A software package for programming psychological experiments using Linux.

3. Stoet (2017). PsyToolkit: A novel web-based method for running online questionnaires and reaction-time experiments

4. Silvera et al. (2001) The Tromsø Social Intelligence Scale, a self-report measure of social intelligence

5. Allison et al. (2012) Toward brief “Red Flags” for autism screening: The Short Autism Spectrum Quotient and the Short Quantitative Checklist for Autism in toddlers in 1,000 cases and 3,000 controls

6. Faust & Kenett (2014) Rigidity, chaos and integration: hemispheric interaction and individual differences in metaphor comprehension

7. Ciaramidaro et al. (2015) Schizophrenia and autism as contrasting minds: neural evidence for the hypo-hyper-intentionality hypothesis

8. Frith & Vignemont (2005) Egocentrism, allocentrism, and Asperger syndrome

9. Stanfield et al. (2017) Dissociation of Brain Activation in Autism and Schizotypal Personality Disorder During Social Judgments 

10. Crespi (2019) How is quantification of social deficits useful for studying autism and schizophrenia?

11. Grimshaw et al. (2010) Semantic ambiguity resolution in positive schizotypy: a right hemisphere interpretation