📑 The Dutch verb-spelling paradox in social media: A corpus study〈Tijn Schmitz, Robert Chamalaun and Mirjam Ernestus〉

Title: The Dutch verb-spelling paradox in social media: A corpus study

Author: Tijn Schmitz, Robert Chamalaun and Mirjam Ernestus

URL: https://doi.org/10.1075/avt.00008.sch

Date: 2018-12

Although the Dutch verb spelling system seems very straightforward, many spelling errors are made, both by children and adults (e.g., Sandra, Frisson, & Daems 2004). ¹ These errors mainly occur with verbs with two or more homophonous forms in their inflectional paradigms. Ample experimental research has been carried out on this topic, but these studies hardly reflect everyday language behavior. In the current corpus study, we reassessed previously found experimental results, but now in a Twitter corpus containing 17,432 tweets with homophonous verb forms. In accordance with previous results, we found a clear preference for the suffix -<d> compared to both -<dt> and -<t> , as well as a frequency effect, resulting in fewer errors for more frequent word forms. Furthermore, the results revealed that users with more followers make fewer errors, and that more errors are made during the evening and night.

Takeaways

#

• if spelling rules are relatively easy to grasp ², but many errors in writing homophones are made anyway, it might mean that these errors come from “the nature of cognitive processes underlying the spelling process, rather than negligence or laziness.”
• writers often decide how to write a verb form based on their mental lexicons instead of applying spelling forms – and it can cause problems when writers use homophones and their spelling choices are “only based on sound-spelling correspondence, not taking the grammatical function of the verb form into account.”
• the choice between the computational procedure (i.e. applying rules) vs. the retrieval procedure (i.e. using mental models) depends on which one is faster
• tweets are good for studying spontaneous writing, since Twitter is used by people from various population groups in terms of age and education, who don’t think too long about what they write
• the overall percentage of mistakes was 6.8%. In reality it’s expected to be even higher, since pseudo-homophones (non-existent words created by illegal spellings) weren’t considered in this study & many users tweet from smartphones, which means that autocorrection might have corrected some of the errors. Still, the 6.8% is significantly lower than 25% found in another study by Sandra (2010) ³

Quotes

#

Previous experimental work on Dutch and French showed that when a single pronunciation is spelled differently depending on its grammatical function, the spelling process of the correct form can be impeded (e.g., Sandra & Fayol 2003). ⁴

A study by Hutto, Yardi, and Gilbert (2013) ⁵ showed that Twitter users seek out well-written content over poorly written content when deciding whether to follow another user. As people with better general language skills tend to be better spellers, we expect that the higher the number of followers, the fewer mistakes in spelling.

Several explanations can be given for this pattern. First, it can be a matter of prestige. Having many followers means that tweets are judged by many people, on the basis of content but also spelling. This could mean that twitterers are more cautious and perhaps more often double-check the spelling of their tweets when they have more followers. A second possibility is that people with many followers tend to have better language skills (including spelling) in general. When someone is talented in writing, people better like what they write, resulting in more followers.

Questions

#

• What factors contribute to low linguistic complexity, aside from morphographic spelling rules?
• What internet slang exists in Dutch?
• How “Twitter Dutch” differs from Standard Dutch / Central Dutch dialect?
• How does forced short format (140 / 280 characters) affect spontaneous writing?
• After the demise of Twitter, which platforms can provide similarly rich material for spontaneous writing research?
• Do mental models influence spelling errors less in languages with less pronounced sound-spelling correspondence (e.g. Chinese)?