Heather van der Lely and developmental language disorders: Introduction to the special issue

This special issue honours the work and memory of Heather van der Lely, who died of cancer on 17 February 2014, at the too-young age of 58. Heather pioneered the study of children with Grammatical-Specific Language Impairment using a range of innovative behavioural and neuroimaging methods, and became one of the best-known researchers in the field of developmental language disorders.

When Heather left school at 16 with few qualifications, her severe dyslexia was still unrecognised. She was apparently told that she was only capable of working at a children’s day nursery near her home village in Herefordshire, UK. However, she had a genuine interest in children and their development, and within two years she was running the nursery. She then applied to train as a nursery nurse, which meant first going to night school to get A-levels. By the time she passed her A-levels, she had decided to become a speech and language therapist, which, of course, required a degree. She graduated from Birmingham City University three years later, with First Class honours. She worked as a speech and language therapist for two years before being accepted by the Department of Psychology at Birkbeck College, University of London, to study for a PhD. Her topic was sentence comprehension in children with Specific Language Impairment (SLI). In 1990, as she was completing her PhD, she was diagnosed with a terminal brain tumour. Given only months to live, she gave up work to enjoy what little time she had left.

Thankfully, the tumour was misdiagnosed. A year later she was back at Birkbeck, in receipt of a post-doctoral fellowship from the British Academy which enabled her to continue studying the underlying nature and cause of SLI in children. A career-development fellowship from the Wellcome Trust followed. In 2001, she was offered the new Chair of Developmental Language Disorders and Cognitive Neuroscience at University College London and founded a Centre to further her work. There she gathered together psychologists, linguists, speech and language therapists and cognitive neuroscientists from all over the world in order to study language disorders using the best tools from a range of disciplines. For PhD students and early career researchers such as myself, it was a very stimulating place to work. In 2009, Heather left UCL but, despite difficult circumstances surrounding her departure, she expanded her research with new collaborators; she became vice-chair of a 25-country research project funded by the European Union, and she held Visiting Professorships in Berlin, Louvain and Harvard.

Heather’s most significant scientific contribution was the discovery and characterisation of a form of SLI that she termed ‘Grammatical-SLI’. She claimed, at a time when the prevailing view was that SLI is caused by an underlying auditory processing and/or phonological short-term memory deficit (Gathercole & Baddeley, 1990; Joanisse & Seidenberg, 1998; Tallal & Piercy, 1973), that a small proportion of children with SLI have a deficit that is specific to syntax and morphosyntax and does not affect other aspects of language processing such as speech perception, vocabulary and pragmatics. Over the years she refined this claim, locating the deficit specifically in those aspects of grammar that are non-local, hierarchical, abstract and composed (van der Lely & Pinker, 2014). She initially proposed a ‘Representational Deficit in Dependent Relations’ to account for the inconsistent (‘optional’) use of grammatical structures requiring ‘movement’ (e.g. wh-questions, the passive construction and anaphoric and pronominal reference; van der Lely, 1998). When she discovered that aspects of phonological grammar were also impaired in SLI, she widened her proposal to a deficit in ‘Computational Grammatical Complexity’ (van der Lely, 2005). The claim that such a domain-specific deficit exists was, and remains, controversial and polarising. It tended to find greater favour in Europe and parts of the USA than in the UK but, wherever she was, she thoroughly enjoyed the ensuing debate. At the same time, she never forgot her speech and language therapist and nursery teacher roots, and in 2006 she and her colleagues published the Grammar and Phonology Screening (GAPS) Test, a 7-minute screen for those aspects of syntax, morphology and phonology that are particularly vulnerable in young children with SLI (Gardner, Froud, McClelland, & van der Lely, 2006).

Heather was diagnosed with an aggressive form of cancer in December 2012. This time the diagnosis was accurate. Despite making an apparent recovery the following summer, she was too ill to attend the Boston University Conference on Language Development in November 2013 to which she had been invited as a Keynote speaker. However, she was able, just before her death three months later, to submit a review paper to Trends in Cognitive Sciences, which her colleague Steven Pinker subsequently steered through the revisions. The final line of the abstract perfectly summarises the ambitions of her research programme: ‘Delineating neurolinguistic phenotypes promises a better understanding of the effects of genes on the brain circuitry underlying normal and impaired language abilities’ (van der Lely & Pinker, 2014, p. 586). Heather aimed at nothing less than fully characterising the heterogeneity of SLI, at the behavioural, cognitive, neural and genetic levels, because doing so would, she argued, offer a window onto typical language development and processing, as well as leading to better forms of diagnosis and intervention for children with developmental language disorders. The eight articles that appear in this special issue contribute to that endeavour.

The issue starts with a study authored by Uli Sauerland and colleagues, and on which Heather collaborated before her death. The article investigates subject and object wh-questions, which have been well studied in the SLI literature (e.g. Deevy & Leonard, 2004; Friedmann & Novogrodsky, 2011; Levy & Friedmann, 2009; van der Lely & Battell, 2003; van der Lely, Jones, & Marshall, 2011). Sauerland et al.’s study is not an investigation of SLI – its participants are typically-developing children – but the novel contributions are that between them the children speak 18 different European languages (from seven different sub-families) and the wh-questions they are required to comprehend are systematically manipulated with respect to whether (a) the question contains a simple interrogative word like ‘who’ or a complex one like ‘which princess’, and (b) the question word is related to the subject position or object position of the verb. The authors find that three linguistic factors facilitate children’s understanding of questions: having overt case morphology, having a single lexical item for both ‘who’ and ‘which’, and the use of synthetic verbal forms. The authors caution that, although object questions do tend to be harder for children to comprehend and therefore might be good structures for inclusion in diagnostic tests for SLI, there is also variation in the acquisition of object questions due to properties of the language, and this variation warrants careful consideration and further study. Cross-linguistic studies of this sort are particularly important given the growing migration within and to Europe and the increase in the proportion of children learning two or more, often unrelated, languages.

Flavia Adani, Maja Stegenwallner-Schütz, Yair Haendler and Andrea Zukowski study another syntactic construction that has been shown to be problematic for children with SLI: relative clauses. They elicit a range of relative clauses in German-speaking children with SLI in order to test Heather’s movement optionality account of SLI (van der Lely, 1998). The authors find that German-speaking children with SLI are indeed impaired in relative clause production compared to typically-developing children: they produce alternative structures consisting of simple main clauses, and also produce nominal and prepositional phrases in isolation (sometimes contextually appropriate, sometimes not). Crucially for evaluating the movement optionality account, however, children with SLI produce very few instances of embedded clauses where the relative clause head noun is pronounced in situ; in fact, such responses are more common among the typically-developing group. The authors argue that although their findings underscore the difficulty German-speaking children with SLI have with structures involving syntactic movement, they provide no specific support for the movement optionality account.

In contrast to the first two articles, Rosalind Thornton, Kelly Rombough, Jasmine Martin and Linda Orton study an area of grammar that has been almost neglected in SLI, namely negation. In particular, they elicit negative sentences from English-speaking children with SLI that require the negative auxiliary form ‘doesn’t’. This allows them to study how negative markers (not, n’t) interact with the 3rd person singular ‘s’ in the grammars of children with SLI. Their work fits very much within a theme that motivated some of Heather’s own research, namely how different levels of language interact in typical development and in children with SLI (Marshall & van der Lely, 2007a; Marshall, Marinis, & van der Lely, 2007). Thornton and her colleagues find that the majority of 5-year-old children with SLI in their sample produced non-adult negative sentences, such as ‘it not fit’, ‘it’s not fit’, ‘it not fits and ‘it’s not fits’. They conclude that children with SLI undergo an extended period in the development of sentential negation during which they produce a range of non-adult sentences that are similar to those previously reported in younger typically-developing 2- and 3-year-olds.

Katherine Demuth and Ekaterina Tomas continue the theme of how different aspects of language interact by reviewing how prosodic morphology impacts the acquisition of inflectional morphology. They claim that their ‘Prosodic Licensing Hypothesis’, which proposes that language learners first produce ‘unmarked’ phonological structures, and only later acquire more ‘marked’ phonological structures, explains much of the variability in typically-developing children’s notoriously lengthy mastery of inflection. Inflectional morphology in children with SLI has an even more protracted period of development, if it is mastered at all (van der Lely & Ullman, 2001), and the Prosodic Licensing Hypothesis also makes predictions for SLI. In their review, Demuth and Tomas draw on their own work and on previous work by Heather that has investigated the interaction between phonology and morphology (Marshall & van der Lely, 2007b, 2012).

In the next article, Marina Lalioti, Stavroula Stavrakaki, Christina Manouilidou and Ioanna Talli investigate the performance of Greek-speaking children with SLI on verbal short-term memory and subject–verb agreement tasks. Their results suggest that deficits in verbal short-term memory constitute a robust marker for SLI in Greek, but that subject–verb agreement can be acquired by Greek-speaking individuals with SLI even in the presence of memory limitations. The authors conclude that the linguistic input and the properties of each language can significantly affect the linguistic development of children with SLI, a theme which emerges from much other current work in the field (e.g. Leonard, 2016).

A topic that has attracted considerable research interest in recent years is the comorbidity of SLI with other developmental disorders, including dyslexia (Messaoud-Galusi & Marshall, 2010; Ramus, Marshall, Rosen, & van der Lely, 2013). Fabrizio Arosio, Elena Pagliarini, Maria Perugini, Lina Barbieri and Maria Teresa Guasti address this topic by investigating the production of direct object clitics, a marker of SLI in Italian, by Italian-speaking children who have a diagnosis of dyslexia but whose language abilities are apparently age-appropriate according to available standardised tests. Nearly half of those children demonstrate difficulties with the production of clitics, suggesting that a significant proportion of Italian-speaking children with dyslexia might have undiagnosed SLI. The authors recommend that when children are referred for dyslexia, their language abilities (including their production of clitics) are also assessed.

Gayaneh Szenkovits, Quynliaan Darma, Isabelle Darcy and Franck Ramus present a study of the phonology of French dyslexic adults using innovative methods that inspired Heather to collaborate with Franck on a study of English-speaking children with dyslexia (Marshall, Ramus, & van der Lely, 2010). The phonological deficit in dyslexia is well-known and well-characterised, but what is novel about this work is that adults with dyslexia who still have difficulty with some aspects of phonological processing are found not to differ from unaffected peers in one aspect of phonological grammar – phonological assimilation rules. Some language-dependent phonological rules change the identity of a segment, whether through substitution, deletion or addition, in certain phonological contexts. In French, coronal stops may undergo voicing assimilation, taking on the voicing of the following obstruent: ‘football’, for example, may be realised as ‘foo[d]ball’. That this rule is language-specific – and therefore needs to be learnt as part of the phonological grammar of the speaker – is shown by the contrast with English, where ‘football’ may be realised as ‘foo[p]ball’, an example of place assimilation. One might predict that people with dyslexia who have a phonological deficit – whatever its underlying nature – would have difficulty learning such rules. Szenkovits and colleagues show that their participants have managed to learn those rules and can employ them accurately in production and ‘undo’ them (i.e. compensate for assimilation) in perception. They conclude with the suggestion that the phonological deficit in dyslexia must lie somewhere other than in the quality of phonological representations and in the phonological grammar.

Finally, Jaime Balladares, Chloë Marshall and Yvonne Griffiths adapt Heather’s Grammar and Phonology Screening (GAPS) test for Spanish, for a study of typically-developing children. They investigate whether Chilean Spanish-speaking preschoolers from two socio-economic status (SES) groups – high and low SES – show group differences in performance on the two subtests of the GAPS, sentence repetition and non-word repetition. They show that children from high SES backgrounds repeat sentences more accurately than children from low SES backgrounds, but that the two groups do not differ in how accurately they repeat non-words. This is in contrast to Gardner et al. (2006), who found no effects of SES for their original, English, version of the test. The results are consistent with findings from previous studies that non-word repetition is not affected by SES. The authors suggest, however, that care should be taken when interpreting sentence repetition test scores, as those scores might be affected not only by an underlying language impairment but also by socio-economic factors, making diagnosis of a language impairment less certain.

As those who had the good fortune to observe her energetic participation in professional meetings can attest, Heather relished an intellectual argument. The aim of this special issue has been to compile a set of articles that Heather would have enjoyed discussing with the authors, preferably on a lovely English summer evening in her garden, accompanied by a light supper and a glass of wine. She wouldn’t necessarily have agreed with the authors’ interpretation of their results, but there is much in these pages that would, I think, have piqued her curiosity and led to fruitful conversations.