Article Text

Original research
In-depth characterisation of a cohort of individuals with missense and loss-of-function variants disrupting FOXP2
  1. Lottie D Morison1,
  2. Elisabeth Meffert2,
  3. Miriam Stampfer3,
  4. Irene Steiner-Wilke4,5,
  5. Brigitte Vollmer6,7,
  6. Katrin Schulze8,
  7. Tracy Briggs9,10,
  8. Ruth Braden1,
  9. Adam Vogel11,12,
  10. Daisy Thompson-Lake13,
  11. Chirag Patel14,
  12. Edward Blair15,
  13. Himanshu Goel16,
  14. Samantha Turner1,17,
  15. Ute Moog18,
  16. Angelika Riess19,
  17. Frederique Liegeois13,
  18. David A Koolen20,
  19. David J Amor1,21,22,
  20. Tjitske Kleefstra20,23,
  21. Simon E Fisher24,25,
  22. Christiane Zweier26,27,
  23. Angela T Morgan1,28,29
  1. 1 Speech and Language, Murdoch Children's Research Institute, Melbourne, Victoria, Australia
  2. 2 SRH University of Applied Health Sciences, Gera, Germany
  3. 3 Medical Genetics Center, Munich, Germany
  4. 4 Centre for Cleft Palate & Craniofacial Malformations, Tuebingen University Hospital, Tubingen, Germany
  5. 5 Department of Paediatric Neurology and Developmental Medicine, University Children’s Hospital Tuebingen, Tuebingen, Germany
  6. 6 Clinical and Experimental Sciences, Faculty of Medicine, University of Southampton, Southampton, UK
  7. 7 Paediatric Neurology, Southampton Children’s Hospital, University Hospital Southampton NHS Foundation Trust, Southampton, UK
  8. 8 Department of Psychology, Heidelberg University, Heidelberg, Germany
  9. 9 Division of Evolution and Genomic Sciences, Faculty of Biology, Medicine and Health, School of Biological Sciences, Manchester Academic Health Science Centre, University of Manchester, Manchester, UK
  10. 10 Manchester Centre for Genomic Medicine, St Mary’s Hospital, Manchester University NHS Foundation Trust, Manchester, UK
  11. 11 Centre for Neuroscience of Speech, University of Melbourne, Melbourne, Victoria, Australia
  12. 12 Redenlab Pty Ltd, Melbourne, Victoria, Australia
  13. 13 Great Ormond Street Institute of Child Health, University College London, London, UK
  14. 14 Genetic Health Queensland, Royal Brisbane and Women's Hospital, Brisbane, Queensland, Australia
  15. 15 Oxford Centre for Genomic Medicine, Oxford University Hospitals NHS Foundation Trust, Oxford, UK
  16. 16 Hunter Genetics, Waratah, New South Wales, Australia
  17. 17 School of Allied Health, Human Services and Sport, La Trobe University, Melbourne, Victoria, Australia
  18. 18 Institute of Human Genetics, Heidelberg University, Heidelberg, Germany
  19. 19 Institute of Medical Genetics and Applied Genomics, University of Tübingen, Tübingen, Germany
  20. 20 Department of Human Genetics, Donders Institute for Brain, Cognition and Behaviour, Radboud University Medical Center (Radboudumc), Nijmegen, The Netherlands
  21. 21 Department of Pediatrics, Royal Children's Hospital, Parkville, Victoria, Australia
  22. 22 Department of Pediatrics, The University of Melbourne, Melbourne, Victoria, Australia
  23. 23 Center of Excellence for Neuropsychiatry, Vincent van Gogh Institute for Psychiatry, Venray, The Netherlands
  24. 24 Language and Genetics, Max Planck Institute for Psycholinguistics, Nijmegen, The Netherlands
  25. 25 Donders Centre for Neuroscience, Donders Institute for Brain, Cognition and Behaviour, Radboud University, Nijmegen, The Netherlands
  26. 26 Institute of Human Genetics, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany
  27. 27 Department of Human Genetics, Inselspital Bern, University of Bern, Bern, Switzerland
  28. 28 Department of Audiology and Speech Pathology, University of Melbourne, Melbourne, Victoria, Australia
  29. 29 Department of Speech Pathology, Royal Children's Hospital, Melbourne, Victoria, Australia
  1. Correspondence to Professor Angela T Morgan, Speech and Language, Murdoch Children’s Research Institute, Melbourne, Victoria, Australia; angela.morgan{at}


Background Heterozygous disruptions of FOXP2 were the first identified molecular cause for severe speech disorder: childhood apraxia of speech (CAS), and yet few cases have been reported, limiting knowledge of the condition.

Methods Here we phenotyped 28 individuals from 17 families with pathogenic FOXP2-only variants (12 loss-of-function, five missense variants; 14 males; aged 2 to 62 years). Health and development (cognitive, motor, social domains) were examined, including speech and language outcomes with the first cross-linguistic analysis of English and German.

Results Speech disorders were prevalent (23/25, 92%) and CAS was most common (22/25, 88%), with similar speech presentations across English and German. Speech was still impaired in adulthood, and some speech sounds (eg, ‘th’, ‘r’, ‘ch’, ‘j’) were never acquired. Language impairments (21/25, 84%) ranged from mild to severe. Comorbidities included feeding difficulties in infancy (10/26, 38%), fine (13/26, 50%) and gross (13/26, 50%) motor impairment, anxiety (5/27, 19%), depression (6/27, 22%) and sleep disturbance (10/24, 42%). Physical features were common (22/27, 81%) but with no consistent pattern. Cognition ranged from average to mildly impaired and was incongruent with language ability; for example, seven participants with severe language disorder had average non-verbal cognition.

Conclusions Although we identify an increased prevalence of conditions like anxiety, depression and sleep disturbance, we confirm that the consequences of FOXP2 dysfunction remain relatively specific to speech disorder, as compared with other recently identified monogenic conditions associated with CAS. Thus, our findings reinforce that FOXP2 provides a valuable entry point for examining the neurobiological bases of speech disorder.

  • genotype
  • genetics
  • phenotype
  • paediatrics

Data availability statement

Data are available on reasonable request. All data relevant to the study are included in the article or uploaded as supplementary information. The datasets generated and analysed during this study are not publicly available because participants have not given permission for data to be made public but may be requested from the corresponding author (AM) who could go back to the participants to request data sharing. Genotypic data were submitted to Decipher (

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See:

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • Heterozygous disruptions of FOXP2 were the first identified molecular cause for severe speech disorder: childhood apraxia of speech (CAS), and yet few cases have been reported, limiting knowledge of the condition.


  • Here we provide the most comprehensive characterisation of individuals with pathogenic FOXP2 variants, almost doubling the number of published families to date. We provide the first cross-linguistic analysis of speech and language across German and English. We show that the phenotype for pathogenic FOXP2 variants remains relatively specific to speech disorder, compared with phenotypes associated with other monogenic conditions involving CAS.


  • This study guides the identification of cases with a FOXP2-related disorder for a clinical genetic diagnosis, improves prognostic counselling and leads to a better targeted clinical management.


FOXP2 was the first gene implicated in a developmental speech and language disorder in the absence of intellectual disability.1 A private heterozygous missense variant in FOXP2 was found to cosegregate with childhood apraxia of speech (CAS) in 15 members of the multigenerational British ‘KE family’, while being absent from all unaffected relatives and healthy controls.1 The study also characterised a balanced chromosomal translocation with a 7q31.2 breakpoint disrupting FOXP2 in an unrelated proband with similar speech deficits.1 CAS is a disorder of speech motor planning and programming that manifests in impaired sequencing of speech sounds into syllables and words with the correct prosody.2 The condition is associated with delayed and protracted speech development.

Pathogenic single-nucleotide variants (SNVs) and intragenic indels that disrupt FOXP2 are rare. To our knowledge, there have been a dozen of these SNVs/indels reported in the literature to date: the original missense variant in the large KE family,1 a non-sense (stop-gain) variant in two siblings and their mother,3 a frameshift in a sporadic patient4 and eight variants across eight small families (intragenic deletions, non-sense, missense and frameshift variants)5 (online supplemental table 1), each occurring in a heterozygous state. The limited number of cases reported may in part be due to the relatively mild speech-focused phenotype associated with pathogenic SNVs/indels of FOXP2, compared with other neurogenetic childhood disorders. While debilitating for affected probands and families, a CAS phenotype does not often lead to clinical genetic testing or to ascertainment in gene discovery cohorts for other neurodevelopmental disorders such as autism spectrum disorder (ASD) or intellectual disability. All probands with pathogenic FOXP2 variants reported to date share a severe speech disorder presentation, most commonly CAS. Yet there is preliminary evidence that, in some cases, SNVs or indels of FOXP2 may cause a broader phenotype including subtle dysmorphology and co-occurring neurodevelopmental features such as ASD.5

Supplemental material

There are other variants that affect FOXP2 that are not SNVs or indels. A large deletion downstream of FOXP2 was hypothesised to have a position effect on expression.6 There have also been case series of large heterozygous 7q31 deletions or reciprocal-balanced translocations associated with more complex phenotypes involving disruptions of FOXP2 in addition to neighbouring genes.7–15 Such phenotypes are now clinically defined as having a FOXP2-plus-related disorder.16

A systematic prospective cohort study of the phenotype(s) associated with FOXP2 variants is warranted to guide better identification of cases for clinical genetic diagnosis, improve prognostic counselling and lead to better targeted intervention. Here we examine speech, language, health and broader developmental phenotypes associated with pathogenic FOXP2 SNVs/indels in a cohort of 28 probands from 17 unrelated families (7 previously published but not deeply characterised for speech and language3–5 and 10 novel) expanding the genetic and clinical spectrum of the disorder. For the first time in any phenotypic study of FOXP2, the specificity of a linguistic phenotype, relative to broader neurodevelopmental skills (eg, communication ability compared with domains of social, motor and daily living skills), was examined using standardised tests. A novel cross-linguistic comparison of speech diagnoses in German-speaking versus English-speaking participants was also conducted to determine homogeneity of the speech phenotype across languages.



Inclusion criterion was a molecular diagnosis of pathogenic variants (SNVs or intragenic deletions/duplications) in FOXP2, in individuals aged ≥6 months. Participants were recruited via clinical genetics colleagues or family self-referral in the Netherlands, France, Britain, Germany, the USA, the UK and Australia. Adult participants and caregivers of child participants provided informed consent to participate in the study. The assessment battery was tailored to cover a wide range of ages and languages.

Health and development

Health and medical information, including developmental milestones and existing diagnoses of neurodevelopmental conditions, were collected via an established direct (adult) or caregiver survey.17–19 Health professional reports and consults confirmed caregiver survey responses. Feeding (Child Oral and Motor Proficiency Scale20) and drooling (Drooling Impact Scale21) measures were collected where age appropriate.


In English-speaking participants, phonology and articulation were assessed using standardised tools (Diagnostic Evaluation or Articulation and Phonology22) or Goldman Fristoe Test of Articulation—Second Edition.23 Phonological delay versus disorder was delineated. Both for English-speaking (authors LDM, ATM) and German-speaking (author EM) participants, phonological and articulation errors were also analysed from a phonetic transcription of a 5 min conversational speech sample. Across both languages, CAS features were rated across three core diagnostic criteria17 18 24: inconsistent speech errors, lengthened and disrupted coarticulatory transitions and impaired prosody. These three criteria were further operationalised into rateable speech errors (see online supplemental table 2). Similarly, dysarthria was assessed using the Mayo Clinic Dysarthria Classification System rating scale25 and evaluating oral motor structure and function.26 The Intelligibility in Context Scale27 documented how well an individual was understood by conversational partners, with a five-point scale ranging from ‘never’ to ‘always’ understood.

Supplemental material

Language and literacy

Receptive vocabulary and expressive vocabulary were assessed using clinician-administered standardised tools dependent on an individual’s age and language (see table 1 for assessments). Likewise, caregiver-administered standardised tools were used to measure speech and language skills. Assessment tools used were dependent on an individuals’ communicative ability, age and language (table 1). Literacy abilities were documented by direct (adult) or caregiver reports, the Vineland Adaptive Behaviour Scales—Third Edition28 written communication subdomain and academic records.

Table 1

Language skills and cognition in this cohort with pathogenic missense/loss-of-function variants disrupting FOXP2

Adaptive behaviour and intellectual ability

The Vineland Adaptive Behaviour Scale, Third Edition (VABS-3) provided scores across language, socialisation, self-care, daily living, motor skills and a composite total adaptive behaviour score.28 A non-parametric Kruskal-Wallis test examined the relative involvement of VABS-3 subdomain scores, and a Wilcoxon signed-rank test was performed between VABS-3 receptive and expressive language scores to highlight any differences between these domains.

General intellectual abilities were assessed with the age-appropriate Wechsler assessment (see table 1 for assessments). Where Full-Scale Intelligence Quotient (FSIQ) could not be obtained, Perceptual Reasoning Indexes and Matrix Reasoning subtest scores were calculated from the relevant Wechsler assessment. A further three assessment tools were used to assess intellectual abilities in four children (table 1). Diagnoses of neurodevelopmental disorders (eg, autism) were identified by caregiver report and confirmed with clinical records.



Twenty-eight participants with pathogenic FOXP2 variants were recruited from 17 families, comprising 10 unreported families (families 1–3, 6, 12–17) and 7 that were previously reported but not deeply characterised for speech and language abilities (families 4, 5, 7–11). Participants had a median age of 16 years 4 months, range 2 years 7 months to 62 years 7 months, and 14 (50%) were male (table 2). Most participants were referred for genetic testing due to the proband’s striking speech and language impairment, except for family 3 (n=2) and participant 17. Family 3 underwent genetic testing as part of research testing for preterm children (ID 3a), and subsequently a variant was also identified in the father (ID 3b). Participant 17 was referred due to microcephaly and dysmorphic facial features.

Table 2

Genotypes and phenotypes of this cohort

Genetic results

The 17 families each had a unique FOXP2 variant, comprising 12 loss-of function and 5 missense variants (table 2). The 12 loss-of-function variants included 3 frameshifts, 4 stop-gain/non-sense variants, 1 variant abolishing the translational start site, 3 intragenic deletions and 1 intragenic duplication. All missense variants were located in the forkhead-box DNA-binding domain of the encoded protein (figure 1). Eleven of the 28 participants had confirmed de novo variants, 12 inherited their variant from a parent, and for 5 participants the inheritance status was unknown (table 2; online supplemental figure 1). Participant 9 was previously reported to have inherited the FOXP2 variant from their father who had the same variant in a mosaic state, and who was unavailable to take part in the present study. Deletions and sequence variants were submitted to Decipher (

Supplemental material

Figure 1

Schematic representation of 17 pathogenic FOXP2 variants in this cohort from 28 individuals in 17 families (NM_014491).

Health and development

Over a third of the assessed participants had feeding difficulties in infancy (10/26), and some had excessive drooling (5/21) in the early years of life (online supplemental table 3; table 2). Gross motor impairments during early development (13/26) and fine motor impairment (13/26) were also present, with a subset of participants having both fine and gross motor impairment (7/26). Of those with fine motor impairment, participant 12b had a formal diagnosis of developmental coordination disorder. In a small number of individuals, hypotonia (IDs 6, 12a) or microcephaly (ID 17) was also noted. Two participants had hearing impairment: mild, conductive hearing loss (25-39dBHL, ID 15) and moderate mixed hearing loss (40-69dBHL, ID 1d). Sleep disturbances were relatively common (10/24), mostly characterised as difficulty falling asleep (5/24) or frequent waking (4/24). Visual impairments (8/26) were present (online supplemental table 3). Physical features were reported in most participants (22/27, four previously reported (online supplemental table 3)), but with no distinct morphological profile across families. Recorded physical features involved the nose (upturned nose: ID 1a; prominent nose: IDs 1b, 1c; hypoplastic alae nasae: ID 4b, 4c; high nasal root: ID 8b; rounded, fleshy or prominent nasal tip: IDs 1b, 1c, 3a, 5, 8b), philtrum (short/flat philtrum: IDs 2, 17), ears (prominent/protruding ears: IDs 1a, 1b, 12b; anteverted ears: ID 14), eyes (periorbital fullness: IDs 3a, 5, 13; prominent eyes: ID 16b), jaw (retrognathia: IDs 1a, 1b, 3a, 13) and lips (full lips IDs: 1a, 1b, 1c, 4a, 4b, 13; thin upper lip: 3a). In individual cases, mild finger pads (ID 10), tapering fingers (ID 8b), single palmar crease (ID 17) and clinodactyly (ID 12a) were also noted.

Supplemental material

Co-occurring diagnoses

A quarter of participants had a diagnosis of ASD (7/27; 2 diagnosed in adulthood) and one had autistic traits but did not meet the criteria. Participant 1c had a diagnosis of attention-deficit hyperactivity disorder (ADHD). Hyperactivity (2/27), attention difficulties (2/27) and restricted interests and behaviour (2/27) were also noted in further participants without formal ASD or ADHD diagnoses. Mental health conditions such as anxiety (5/27), depression (6/27) and obsessive–compulsive disorder (2/27) were reported in five adults and one adolescent.

Most participants (school aged or older) with pathogenic FOXP2 variants attended mainstream schools (14/23); seven attended special education schools and two attended a mix of special education and mainstream schools. Learning support (eg, teaching aide, individualised learning plan) was common (15/23) across all settings. All five preschool participants attended specialist preschool settings for children with additional learning needs. All caregivers of school-aged children and adolescents reported that their child’s academic progress had been most impacted by their speech and language impairments.

Communication development

Speech development was characterised by limited babbling and a reduced phonetic (sound) inventory relative to peers across the first 7 years of life when a full inventory is typically acquired. Some developed first words around the typical age of development (12–15 months, 9/25), whereas others were slightly (15–18 months, 1/26) or more significantly (>18 months, 14/26) delayed (table 3). One participant (ID 14) had not said their first words yet in early childhood (3–5 years old). Eight participants had not yet mastered combining words (IDs 1a, 6, 9, 12a, 13, 14, 15, 17). Only three participants combined words in line with the typical development milestone of 2–3 years of age. The remaining participants combined words between 4 to 5 years (4/22), 6 to 7 years (2/22) and 8 years or older (5/22), representing protracted development relative to the typical developmental milestone of 2–3 years.

Table 3

Speech features and educational placement of individuals with pathogenic SNVs/indels disrupting FOXP2


CAS was the most common speech diagnosis (88%, 22/25) (figure 2, table 3), with features including frequent sound omissions, the same consonant or vowel being produced differently across different words, impaired sequencing of phonemes and syllables, voicing errors, syllable segregation, difficulty achieving initial articulatory configurations, equal stress, altered suprasegmental features and slow rate (online supplemental table 3). CAS was present in most English (86.7%, 13/15) and German (87.5%, 7/8) speakers.

Figure 2

Speech disorders of participants with pathogenic missense/loss-of-function variants disrupting FOXP2 (n=25, two participants had no speech disorder).

Some participants had multiple co-occurring speech sound disorders (figure 2). Dysarthria (8.3%, 2/24) was infrequent and mild in severity, characterised as mixed nasality and harsh vocal quality (IDs 4a, 5). Phonological disorder was common (58.3%, 14/24), especially in children (<18 years); although for some (IDs 4a, 4b) this persisted into adulthood (table 3). A further two adults had a severe phonological delay, that is, typical phonological error patterns that appear in the speech of younger individuals but that should have resolved by 7 years of age (online supplemental table 4a; IDs 1b, 1c). Disordered oral motor movements were present both on speech (eg, say ‘pataka’) and non-speech tasks (eg, bite then blow) (92.9%, 13/14).

Supplemental material

Two adults and one child did not have signs of CAS at the time of testing (IDs 3a, 3b, 11a). Participant 11a had a speech and language disorder in childhood for which they received therapy, but could not recall having CAS. The other adult participant (ID 3b) reported being a ‘quiet’ child but did not have speech therapy, and his son (ID 3a) had a phonological disorder without CAS.

Phonetic inventories were analysed for 12 English-speaking participants (online supplemental table 4a,b). Strikingly, 66.7% (8/12) of English-speaking participants did not have the affricate /d͡ʒ/ (eg, ‘j’ in ‘jump’) and 58.3% (7/12) did not have its voiceless counterpart, /t͡ʃ/ (eg, ‘ch’ in ‘chair’). Many were also missing the later developing sounds of /ɹ/ (eg, ‘r’ in ‘rabbit’, 66.7%, 8/12), /θ/ (eg, voiceless ‘th’ in ‘thin’, 58.3%, 7/12) and /ð/ (eg, voiced ‘th’ in ‘this’, 58.3%, 7/12) (online supplemental table 4a,b). Other phonemes absent in some English speakers’ inventories were /ʃ/ (‘sh’, 41.7%, 5/12), /ŋ/ (‘ng’, 41.7%, 5/12), /l/ (27.3%, 3/11) and /s/ (27.3%, 3/11). The phonemes /d͡ʒ/ and /t͡ʃ/ were not present in most German participants (4/6).

Average intelligibility for children, assessed via the ICS, ranged from ‘never’ understood (20%, 2/10), to ‘rarely’ (33.3%, 3/10) to ‘sometimes’ understood (50%, 5/10). For adults, average intelligibility ranged from ‘sometimes’ (16.7%, 1/6) to ‘usually’ understood (66.7%, 4/6). Only one participant was ‘always’ understood (ID 3b).

Language and literacy

More than half of the cohort (56%, 14/25) had mild to severe receptive vocabulary impairment (table 3; online supplemental figure 2). Receptive grammar was commonly affected (72.2%, 13/18) ranging from moderate to severely impaired. Expressive vocabulary (83.3%, 15/18) and expressive grammar impairments were also common (86.7%, 13/15).

Supplemental material

Speech was the most severely affected communication domain (mean score=1.7) for the Children’s Communication Checklist, Second Edition (CCC-2) (completed for IDs 1a, 2, 3a, 5, 6, 15, 16b; figure 3, normative mean=10, SD=3).

Figure 3

Children’s Communication Checklist subdomains61 in participants with pathogenic FOXP2 variants (n=7, average=10, SD=3). Scores ≤6 are within normal limits, and scores <5 are low. Line=median, x=average, ·=outlier.

Five participants were minimally verbal (<30 words; IDs 9, 12a, 13, 14, 15; between 2 and 18 years old). Three of the five used high-tech Augmentative and Alternative Communication (AAC) devices (eg, a speech-generating application on a tablet), while the remaining two used some sign language.

Spelling (17/24) and reading (17/24) impairments were common (table 3). This was reflected in the results from the VABS-3 (written subdomain mean=9.25, normative mean=15, SD=2). Four participants were at the preliteracy stage (<5 years of age).

Adaptive behaviour and cognition

Receptive and expressive language performance was low in those assessed (n=11, mean=9.5 in both domains; online supplemental figure 3) and not significantly different between the two domains (p=0.79). Language performance (mean=70.73) was significantly different from (p=0.01) and higher than socialisation (mean=61.18) and daily living (mean=57.45). Motor skills were an area of relative strength, although impaired compared with norms (mean=77.18), although the sample size was small and results should be interpreted with caution. Further, normative data for motor skills are only available up to 9 years 11 months; however, none of the older participants reached the ceiling on motor skills.

Supplemental material

Language skills were incongruent with intellectual skills for many, as seven participants with severe language impairment scored within the average range on non-verbal subtests (table 3). IQ was formally assessed in 20 participants (FSIQ n=10, non-verbal IQ n=10). In the non-verbal testing, seven performed in the average range, one was borderline and two were moderately impaired. For an FSIQ, two performed within the average range, four in the borderline range (70–85 IQ) and four had a mild intellectual disability (50–69 IQ).


We systematically delineated the speech and cognitive phenotype in 28 probands from 17 unrelated families (10 of which are novel) with heterozygous pathogenic missense/loss-of-function variants disrupting FOXP2, and completed the first cross-linguistic analysis of this disorder. Our data confirm aberrant speech and language development as a central feature. While speech presentation improves over time, with a reduction in CAS severity and improvement in phonological production, the disorder is characterised by impaired speech intelligibility that persists into adulthood, with most adults in our cohort being understood only ‘sometimes’ or ‘usually’ (rather than ‘always’) understood.

In terms of intellectual ability, scores ranged from below average to average in our cohort. For individuals with FSIQ data available, most (8/10) were scored as having borderline or mild intellectual disability, while for those who had only non-verbal IQ data available, most (7/10) were average. This range and profile are in line with previous findings for individuals with pathogenic SNVs/indels in FOXP2, with most individuals falling in the low average range and below for FSIQ and non-verbal IQ. The critical point of note here is that FSIQ takes into account the language metric of vocabulary knowledge, hence why FSIQ is generally more impaired than non-verbal IQ skills for individuals with this speech and language phenotype. At the same time, we also confirm observations from prior literature that the profile of this disorder differs from classic intellectual disability syndromes, in that severe speech and/or language impairments can occur against a background of non-verbal cognition within the normal range29 as observed for seven of our probands with available data.

ASD has previously been reported in only a small number of individuals carrying pathogenic SNVs/indels of FOXP2 (n=2/46; online supplemental table 1). The findings in our cohort indicate that there may be a higher prevalence of ASD in this disorder (25.9% of our cohort) than in the general population (~1%–2%),30 although further research is needed to account for the discrepancy between our current findings and the prior literature. Of note, pathogenic variants in the closely related orthologue FOXP1 are known to substantially increase the risk for ASD.31 Common non-coding polymorphisms in introns of FOXP2 have shown associations with ADHD in large-scale genome-wide association studies, in the context of a multifactorial framework.32 The current study clearly shows that, by contrast, high-penetrance SNVs/indels disrupting this same locus do not yield elevated susceptibility to ADHD, with a prevalence in our cohort (1/27=4%) that is similar to that in the general population.33

Sleep disturbances were common in our cohort and have been previously associated with idiopathic CAS34 and other neurodevelopmental disorders, such as ASD, intellectual disability and ADHD.35 The aetiology of sleep problems in such disorders is currently unknown, but they are posited to have biological and psychopathological causes. Although ASD, intellectual disability and ADHD were present here to varying degrees (as discussed above), within our cohort, sleep disturbance is also noted in children without those diagnoses.

We provide novel insights into other clinical diagnoses of mental health conditions that might be associated with pathogenic FOXP2 variants. In particular, anxiety (19%) and depression (22%) had a higher prevalence than in the general population (between 2% and 4%)33 and than in other neurodevelopmental disorders which were also present in our cohort, such as mild intellectual disability (~3%–4%).36 Anxiety has previously been associated with idiopathic CAS,37 38 and speech and language disorders are known to have possible negative impacts on mental health.37–40 It is difficult to ascertain whether mental health disorders are part of the phenotypic spectrum due to pathogenic FOXP2 variants, or occur as a secondary consequence of the communication deficits experienced by affected individuals, as is seen in other speech disorders, such as stuttering.41 All participants with anxiety and depression were older than 16 years old, perhaps indicating that these mental health conditions arise later in life due to the impact of the communication impairment.

Gross motor impairment is thought to be relatively uncommon in individuals with pathogenic SNVs/indels disrupting FOXP2.42 However, two-thirds of the assessed participants in this study indicated having difficulty with gross motor skills during development. FOXP2 disruption therefore appears to impact brain circuits involved in fine as well as gross motor development. Gross motor skill learning deficits have been identified in knock-out animal models.43

We did not find convincing evidence of a dysmorphism phenotype in individuals with pathogenic FOXP2 variants. Although physical features were noted for 81% of participants, most were minor and only shared among individuals from the same family. There was no consistent pattern of morphology seen across multiple unrelated probands in the cohort.

Regardless of the associated developmental features noted here, CAS was the most striking and consistent phenotypic characteristic in the present cohort. Dysarthria was far less common than CAS, clarifying the role of FOXP2 in the planning and programming of movement sequences, as supported by animal models.43 Adult participants typically had more intact, although incomplete, phonetic inventories than younger participants. In our study, more than half of all English-speaking participants with pathogenic missense/loss-of-function variants in FOXP2 were missing one or more of the phonemes: /t͡ʃ, d͡ʒ, ɹ, θ, ð/ from their inventory. Three of these phonemes (/ɹ, θ, ð/) are in the ‘late eight’ sounds of English speech development, while affricates /t͡ʃ, d͡ʒ/ sit within the ‘middle eight’, referring to whether they are acquired earlier or later during typical phoneme acquisition.44

There may be a window for plasticity and acquisition of new phonemes. Children with speech sound disorders are more likely to have persistent speech error patterns if these are not resolved by 8½ years old.44 We might speculate that FOXP2 dysfunction has the greatest impact between 2 and 7 years old when most phonemes are acquired.45 46 Intriguingly, neural expression of orthologues of FOXP2 in model organisms has been shown to vary during different periods of vocal development,47–49 for example being upregulated in parts of the brain of the male Zebra finch during a developmental window that is important for vocal learning.50 Reduced expression of this gene in mice alters the development and continuing plasticity of neuronal networks,51 impairs synaptic plasticity in striatal and cerebellar circuits and affects the learning of motor skills.52 53 Perhaps, the lack of acquisition of ‘late eight’ sounds and affricates in children with FOXP2 disruptions may relate to the closing of the relevant developmental window. Other theories to explain the lack of acquisition of these phonemes include reduced functional load and ambient frequency for these phonemes54 or the motoric complexity of these sounds55 which rely heavily on tongue coordination and movement, known to be impaired in children with CAS.55–57 Further research is required to disentangle these relationships.

Speech impairment in the three participants without CAS was minimal or even absent, contrasting with unaffected, previously reported cases of FOXP2 disruption.13 The speech of participant 17 was also less impaired than that of other children, although he had a diagnosis of CAS. Participant 11a was not referred for genetic testing for his speech, but rather to determine whether the variant in his child was de novo or inherited. Of note, family 3 and participant 17 were the only participants who had not been referred for testing primarily on the basis of speech and learning impairments. Thus, it is possible that there is a broader range of speech phenotypes associated with pathogenic FOXP2 variants, and that individuals with milder presentations are unlikely to be referred for genetic testing.

Although there was no statistically significant difference between receptive and expressive language as reported by caregivers, other standardised tests indicated receptive language was more intact than expressive language. Literacy skills were also low across the cohort, in line with the high rate of literacy challenges for individuals with idiopathic CAS.58

The speech domain on the CCC-2 confirmed that speech was the most severely impaired form of communication in children. AAC systems should be considered for children with pathogenic FOXP2 variants due to the protracted speech development and severe speech impairment which persists for many throughout their lifetime.

We were unable to identify any clear phenotype–genotype correlation in the present cohort as we did not have sufficient power due to too few cases of missense and loss-of-function variants. The severity of speech and language disorder differed even among individuals with the same FOXP2 variant in the same family. Family 3 had a relatively mild presentation compared with the other individuals in the study, despite having an intragenic deletion encompassing all FOXP2 exons. Participant 17 had a translational start-site variant and a mild phenotype, with a larger phonemic repertoire and expressive vocabulary than other participants of similar ages in the cohort. This variant may not cause a clear-cut loss of function since there are alternative transcription start sites, potentially leading to a shorter protein.

In conclusion, CAS and language impairments are the most discernable features associated with heterozygous pathogenic missense/loss-of-function variants disrupting FOXP2. We also provide the evidence of additional neurodevelopmental features in subsets of our cohort, such as mild intellectual disability, ASD, anxiety, depression and sleep disturbances. There appear to be no distinctive physical features consistently associated with FOXP2 disruptions. The phenotype associated with pathogenic variants that directly disrupt FOXP2 remains relatively specific to speech disorder, compared with phenotypes associated with other monogenic conditions involving CAS.58 Thus, our findings demonstrate that FOXP2 provides an especially valuable entry point for examining the neurobiological bases of speech disorder.

Data availability statement

Data are available on reasonable request. All data relevant to the study are included in the article or uploaded as supplementary information. The datasets generated and analysed during this study are not publicly available because participants have not given permission for data to be made public but may be requested from the corresponding author (AM) who could go back to the participants to request data sharing. Genotypic data were submitted to Decipher (

Ethics statements

Patient consent for publication

Ethics approval

This study involves human participants and was approved by Royal Children’s Hospital, Melbourne, Human Research Ethics Committee HREC 37353A. Participants gave informed consent to participate in the study before taking part.


Sincere thanks to the children, families and clinicians who took part in this project. We would like to thank Professor Faraneh Vargha-Khadem for early work on the FOXP2 phenotype which served as an inspiration for this paper. Several authors of this publication are members of the European Reference Network on congenital malformations and rare intellectual disability (ITHACA).


Supplementary materials


  • CZ and ATM are joint senior authors.

  • Twitter @mcri_speechlang

  • Correction notice The article has been corrected and republished: participants' data in the Abstract and Results sections have been edited.

  • Contributors LDM: generated data, analysed data, interpreted data, wrote manuscript; EM, RB, AV, ST: generated data, analysed data, interpreted data, revised manuscript; MS, IS-W, BV, KS, TB, CP, EB, UM, AR: generated data, analysed data, revised manuscript; DT-L: collected data, analysed data, revised manuscript; FL: contributed data, revised manuscript; DAK, TK: generated data, revised manuscript; DJA: analysed data, interpreted data, revised manuscript; SEF: designed and conceptualised study, directed project, analysed data, interpreted data, wrote manuscript; CZ: designed and conceptualised study, directed project, generated data, analysed data, interpreted data, revised manuscript; AM: designed and conceptualised study, directed project, generated data, analysed data, interpreted data, wrote manuscript, guarantor.

  • Funding This work was supported by a National Health and Medical Research Council (NHMRC) Centre of Research Excellence in Speech and Language Neurobiology (CRE-SLANG) number 1116976 and NHMRC Project grant number APP1127144, awarded to AM and SEF; NHMRC Practitioner Fellowship number 1105008 and Investigator grant number 1195955 awarded to AM. This work is also supported by the Victorian Government’s Operational Infrastructure Support Program. This work was financially supported by the Dutch Research Council grant to TK (015.014.036 and 1160.18.320) and Netherlands Organization for Health Research and Development to TK (91718310). SEF is supported by the Max Planck Society.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.