Analysis of Presurgical Language in Children with Posterior Fossa Tumours Relative to Postoperative Speech Outcomes: Findings from the European CMS Study.

Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Cerebellum . 2026 Apr 10;25(2):48. doi: 10.1007/s12311-026-01987-3 Search in PMC Search in PubMed View in NLM Catalog Add to search Analysis of Presurgical Language in Children with Posterior Fossa Tumours Relative to Postoperative Speech Outcomes: Findings from the European CMS Study Aliene Reinders Aliene Reinders 1 Center for Language and Cognition Groningen (CLCG), University of Groningen, Groningen, The Netherlands 2 Research School of Behavioural and Cognitive Neurosciences (BCN), University of Groningen, Groningen, The Netherlands Find articles by Aliene Reinders 1, 2 , Cheyenne Svaldi Cheyenne Svaldi 1 Center for Language and Cognition Groningen (CLCG), University of Groningen, Groningen, The Netherlands 3 Division of Psychology and Language Sciences, University College London, London, UK Find articles by Cheyenne Svaldi 1, 3 , Annet Kingma Annet Kingma 4 Department of Pediatrics/Pediatric Oncology, University Medical Center Groningen, Groningen, The Netherlands Find articles by Annet Kingma 4 , Jonathan Kjær Grønbæk Jonathan Kjær Grønbæk 5 Department of Paediatric and Adolescent Medicine, Copenhagen University Hospital Rigshospitalet, Copenhagen, Denmark Find articles by Jonathan Kjær Grønbæk 5 , Ditte Boeg Thomsen Ditte Boeg Thomsen 6 Department of Nordic Studies and Linguistics, University of Copenhagen, Copenhagen, Denmark Find articles by Ditte Boeg Thomsen 6 , Karin Persson Karin Persson 7 Department of Health Sciences, Lund University, Lund, Sweden Find articles by Karin Persson 7 , René Mathiasen René Mathiasen 5 Department of Paediatric and Adolescent Medicine, Copenhagen University Hospital Rigshospitalet, Copenhagen, Denmark Find articles by René Mathiasen 5 , Christine Dahl Christine Dahl 5 Department of Paediatric and Adolescent Medicine, Copenhagen University Hospital Rigshospitalet, Copenhagen, Denmark Find articles by Christine Dahl 5 , Andrea Carai Andrea Carai 8 Neurosurgery Unit, Bambino Gesù Children’s Hospital, Rome, Italy Find articles by Andrea Carai 8 , Bianca Andreozzi Bianca Andreozzi 9 Department of Hematology/oncology, Cell Therapy, Gene Therapies and Hematopoietic Transplant, IRCCS Bambino Gesù Children’s Hospital, Rome, Italy Find articles by Bianca Andreozzi 9 , Angela Mastronuzzi Angela Mastronuzzi 9 Department of Hematology/oncology, Cell Therapy, Gene Therapies and Hematopoietic Transplant, IRCCS Bambino Gesù Children’s Hospital, Rome, Italy 10 Department of Life Sciences and Public Health, Università Cattolica del Sacro Cuore, Milan, Italy Find articles by Angela Mastronuzzi 9, 10 , Barry Pizer Barry Pizer 11 University of Liverpool, Liverpool, UK 12 Department of Oncology, Alder Hey Children’s NHS Foundation Trust, Liverpool, UK Find articles by Barry Pizer 11, 12 , Colin Thorbinson Colin Thorbinson 12 Department of Oncology, Alder Hey Children’s NHS Foundation Trust, Liverpool, UK Find articles by Colin Thorbinson 12 , Kristian Aquilina Kristian Aquilina 13 Department of Neurosurgery, Great Ormond Street Hospital, London, UK Find articles by Kristian Aquilina 13 , Eelco Hoving Eelco Hoving 14 Oncological Pediatric Neurosurgery, Princess Maxima Center for Pediatric Oncology, Utrecht, The Netherlands Find articles by Eelco Hoving 14 , Marianne Juhler Marianne Juhler 15 Department of Neurosurgery, Aarhus University Hospital, Aarhus, Denmark Find articles by Marianne Juhler 15 , Roel Jonkers Roel Jonkers 1 Center for Language and Cognition Groningen (CLCG), University of Groningen, Groningen, The Netherlands 2 Research School of Behavioural and Cognitive Neurosciences (BCN), University of Groningen, Groningen, The Netherlands Find articles by Roel Jonkers 1, 2 , Vânia de Aguiar Vânia de Aguiar 1 Center for Language and Cognition Groningen (CLCG), University of Groningen, Groningen, The Netherlands 2 Research School of Behavioural and Cognitive Neurosciences (BCN), University of Groningen, Groningen, The Netherlands 16 Department of Radiation Oncology, University Medical Center Groningen, Groningen, The Netherlands Find articles by Vânia de Aguiar 1, 2, 16, ✉ Author information Article notes Copyright and License information 1 Center for Language and Cognition Groningen (CLCG), University of Groningen, Groningen, The Netherlands 2 Research School of Behavioural and Cognitive Neurosciences (BCN), University of Groningen, Groningen, The Netherlands 3 Division of Psychology and Language Sciences, University College London, London, UK 4 Department of Pediatrics/Pediatric Oncology, University Medical Center Groningen, Groningen, The Netherlands 5 Department of Paediatric and Adolescent Medicine, Copenhagen University Hospital Rigshospitalet, Copenhagen, Denmark 6 Department of Nordic Studies and Linguistics, University of Copenhagen, Copenhagen, Denmark 7 Department of Health Sciences, Lund University, Lund, Sweden 8 Neurosurgery Unit, Bambino Gesù Children’s Hospital, Rome, Italy 9 Department of Hematology/oncology, Cell Therapy, Gene Therapies and Hematopoietic Transplant, IRCCS Bambino Gesù Children’s Hospital, Rome, Italy 10 Department of Life Sciences and Public Health, Università Cattolica del Sacro Cuore, Milan, Italy 11 University of Liverpool, Liverpool, UK 12 Department of Oncology, Alder Hey Children’s NHS Foundation Trust, Liverpool, UK 13 Department of Neurosurgery, Great Ormond Street Hospital, London, UK 14 Oncological Pediatric Neurosurgery, Princess Maxima Center for Pediatric Oncology, Utrecht, The Netherlands 15 Department of Neurosurgery, Aarhus University Hospital, Aarhus, Denmark 16 Department of Radiation Oncology, University Medical Center Groningen, Groningen, The Netherlands ✉ Corresponding author. Received 2025 Dec 17; Accepted 2026 Mar 20; Issue date 2026. © The Author(s) 2026 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . PMC Copyright notice PMCID: PMC13068704  PMID: 41961445 Abstract Posterior Fossa Syndrome is a common complication in children following posterior fossa tumour surgery, typically marked by transient postoperative speech impairment (POSI; i.e., mutism or reduced speech). Previous studies suggest that children who develop POSI show different language profiles postoperatively compared to those who do not. It remains unclear to what extent these language difficulties exist preoperatively and whether preoperative language difficulties are related to postoperative speech status. This study provides the first comprehensive analysis of preoperative language samples, using data from the European study of Cerebellar Mutism Syndrome. Patients with and without POSI were compared to identify preoperative language characteristics that may be associated with POSI. Preoperative language samples of 34 patients aged 3–16 years were analysed (16 developed POSI; 18 did not). We compared global sample characteristics and language performance across four levels: the semantic, lexical, morphosyntactic, and phonological level. No significant preoperative language differences were found between the groups for the four levels of language processing (all p -values > 0.137). Children who developed POSI produced more unintelligible speech preoperatively (β = -14.455, p = .024), with better intelligibility related to older age (age×group: β = 0.152, p = .007), only for the group with POSI. While the main focus of this study was on language, these findings suggest that risk factors for POSI within the domain of verbal output may lie more in preoperative speech. A comprehensive analysis of preoperative speech may provide valuable insight into speech characteristics potentially related to POSI. Supplementary Information The online version contains supplementary material available at 10.1007/s12311-026-01987-3. Keywords: Cerebellar mutism syndrome, Mutism, Posterior fossa syndrome, Infratentorial neoplasms, Preoperative language impairment, Language disorders Introduction Approximately half of the brain tumours in children occur in the posterior fossa [ 1 ]. Treatment for Posterior Fossa Tumours (PFTs) generally entails neurosurgical resection, which is often followed by chemo- and radiotherapy [ 2 ]. Posterior Fossa Syndrome (PFS), also referred to as Cerebellar Mutism Syndrome (CMS), is a common complication following PFT surgery in children [ 3 ]. It typically arises within days after the neurosurgical resection and affects around 24–34% of patients [ 4 – 7 ]. The most defining symptom of PFS is transient mutism (PFS1; [ 8 ]) or severely reduced speech (PFS2), typically lasting from a couple of days to 6 months [ 9 ], with an average duration of about 8 weeks [ 10 ]. Mutism and severely reduced speech are also jointly referred to as postoperative speech impairment (POSI) [ 5 ]. PFS is further characterised by emotional lability, hypotonia, and a wide range of motor and cognitive deficits [ 11 ]. Although the mutism or reduced speech is transient, speech problems can persist later in life [ 12 ]. Importantly, long-term deficits associated with postoperative mutism or reduced speech extend beyond motor speech production, and affected children exhibit broad neurocognitive vulnerabilities, including in the domain of language [ 13 ], reflecting the cognitive system underlying the representation and processing of linguistic information. This highlights the importance of identifying early (cognitive) markers that may be associated with the development of postoperative mutism or reduced speech. While some studies have examined preoperative language functioning in this population, these have typically relied on broad or composite language measures rather than providing an in-depth profile of language abilities across linguistic processing levels [ 14 , 15 ]. As a result, potential presurgical differences in language functioning between children who go on to develop postoperative mutism or reduced speech compared to those with habitual speech remain insufficiently characterised. In this study, we therefore perform a comprehensive analysis of narrative language samples of children diagnosed with a PFT before they undergo neurosurgical resection. Their language performance will be related to their postoperative speech status (mutism or reduced speech vs. habitual speech), to identify if there are language characteristics that may be related to the emergence of mutism or reduced speech. Speech and Language after Mutism or Reduced Speech When mutism or reduced speech subsides, patients will often present with motor speech disorders, a pattern collectively termed mutism with subsequent dysarthria [ 16 ]. They may also have more impairments in the initiation of voluntary movement, including speech initiation [ 17 ], termed adynamic speech [ 18 ]. However, long-term impairments occur in both speech and language in paediatric PFT survivors regardless of whether they experienced mutism or reduced speech or not [ 13 , 19 , 20 ]. Nonetheless, several studies suggest that there may be differences in either the nature or the severity of motor and cognitive impairments observed, and specifically in speech and language abilities of patients who experienced mutism or reduced speech compared to patients who did not [ 16 – 18 , 21 ]. Patients who experienced mutism were reported to have poorer outcomes in verbal comprehension, receptive and expressive language, verbal memory, and verbal fluency [ 19 ]. Additionally, the presence of mutism or reduced speech was found to be related to slower reading pace [ 22 ] and poorer verbal learning [ 23 ]. In Svaldi et al. [ 13 ], children who experienced mutism were found to show language impairments predominantly in the morphosyntactic and semantic domains of language, while children who did not develop mutism showed a wider spread in language impairments across all language domains. However, in a recent study investigating postoperative word-finding abilities in a large sample, Persson et al. [ 20 ] did not find a relation between postoperative speech impairment (POSI, consisting of mutism or reduced speech) and poorer postoperative word-finding abilities. Still, all children who had experienced a complete absence of speech exhibited a postoperative decline in word finding. Studies examining language differences between patients with or without mutism or reduced speech included either small participant numbers [ 13 ], or focused on a specific aspect of language [ 20 ]. Furthermore, most studies did not account for preoperative language impairment (with the exception of [ 20 ]). This is important, as it may be that preoperative language characteristics can be linked to the emergence of mutism or reduced speech [ 14 , 15 ]. Risk Factors for Mutism or Reduced Speech Extensive research has been done on the risk factors related to the emergence of mutism or reduced speech, see [ 24 ] for an overview. Several studies show that the incidence of mutism or reduced speech decreases with age [ 5 , 6 , 25 ], with rare occurrence in adults [ 26 ]. İldan et al. [ 27 ] proposed that the higher incidence of mutism in younger children could be related to the incomplete maturation of the brain that renders younger children more prone to developing the complication. Additionally, some tumour locations are associated with higher risk of mutism or reduced speech, such as vermal and midline tumours [ 4 , 5 , 28 ] and brainstem tumours [ 7 ]. Cerebellar hemisphere tumours, on the other hand, generate a lower risk [ 5 , 7 ]. Concerning tumour type, patients with high-grade medulloblastomas have consistently been reported to develop mutism or reduced speech more often than those with low-grade astrocytomas [ 4 , 5 ]. Preoperative risk factors in the domain of language are underexplored, as previous research primarily focuses on differential postoperative language outcomes in patients with mutism or reduced speech. Nonetheless, Di Rocco et al. [ 15 ] found that children in their sample without preoperative language problems did not develop mutism, while all children who developed mutism presented with preoperative language problems, characterised by a shorter Mean Length of Utterance (MLU) and problems with verbal fluency and lexical naming. Bianchi et al. [ 14 ] extended the findings of Di Rocco et al. by enlarging the patient cohort, showing that 20 out of 70 patients with PFTs who developed mutism presented with preoperative language impairments. Persson et al. [ 29 ] found that patients with PFTs experience word finding difficulties before neurosurgical resection, characterised by slow and/or inaccurate word finding, but they did not examine the relationship between preoperative word finding difficulties and the emergence of mutism or reduced speech. Research into the preoperative language abilities of children with PFTs is thus limited, and no research has comprehensively compared the language profiles of children who do and do not experience mutism or reduced speech across different levels of language processing. Identifying Language Impairments Through Connected Language The existing literature on language disorders in child survivors of PFTs suggests that these may affect all language domains (e.g., phonology, morphosyntax, lexical and semantic knowledge, pragmatics) and may present themselves in variable combinations and severity across children [ 13 , 30 ]. Preoperative language abilities should thus be evaluated comprehensively in every patient, including all language domains. A highly productive approach to evaluate multiple domains of language is through the evaluation of language samples, either elicited in conversations/interactions, picture descriptions, or through the (re)telling of narratives [ 13 , 31 , 32 ]. This approach has been used previously in children with PFTs, focusing on the macro- and microstructural aspects of language after neurosurgical resection and revealing differences from healthy controls [ 13 , 33 ] as well as differences between subgroups [ 34 ]. This approach combines the use of standard measures (e.g., MLU) and the critical variable approach by Shallice [ 35 ], similarly to Svaldi et al. [ 13 ], see also the Supplementary Material for a detailed description of this approach. In the critical variable approach, Shallice describes that properties of words (i.e., psycholinguistic properties) can impact language performance, and that these properties, which reflect functioning of specific levels of language processing, can reveal impairments at their respective levels. Levels of Language Processing Across models of language processing (e.g. [ 36 – 38 ]), it is commonly agreed that conveying a message (e.g., ‘The cat meows’) through spoken language requires processing at several levels of language. At the semantic level, conceptual information related to the meanings of words is being stored and retrieved (e.g., the knowledge that a cat meows; [ 38 ]). At the lexical level, the word forms which make up our mental dictionary, or lexicon, are stored and retrieved [ 38 ]. The morphosyntactic level relates to the internal structure of words (morphology) and sentences (syntax) [ 39 ]. Morphological rules govern the internal structure of words and establish a relation to other units in the syntactic structure (e.g., the addition of ‘s’ to the verb-stem ‘meow’, to match the third person singular of the subject). Syntax consists of grammatical information such as word classes (e.g., ‘cat’ is a noun, ‘meowing’ is a verb), as well as rules concerning sentence structure (e.g., the subject ‘cat’ must come before the verb ‘meows’). Furthermore, at the phonological level, segmental phonological information is retrieved. This entails information of individual speech sounds, which need to be ordered correctly and stored in phonological short-term memory in preparation for and during speech [ 38 ]. Breakdowns at each of these levels can lead to characteristic patterns of errors (see [ 40 ]). For each of these levels, there are linguistic variables which may be extracted from narrative language to study their functioning. For example, semantic representations may be easier to retrieve depending on how easily a concept evokes a mental image (imageability) [ 41 ] or how concrete/abstract it is [ 42 ], so patients with semantic disorders may be biased to use the words which are highly imageable and concrete (e.g. [ 13 , 43 ]), . Other variables that tap into the semantic system are familiarity [ 44 ] or instrumentality of verbs [ 45 ]. At the lexical level, vocabulary size can be estimated with the ratio of different to total words in a sample (Type-token ratio, TTR), although this measure is known to be sensitive to sample length [ 46 ]. Alternatively, lexical properties, such as lexical accuracy, corpus-based word frequency [ 47 ], or word age of acquisition [ 48 ] may be studied. Morphosyntactic ability may be assessed with measures such as the Mean Length of Utterances (e.g. [ 49 ]), , as well as grammatical accuracy [ 50 ], and proportion of finite verbs (i.e., inflected verbs such as ‘walks’, as opposed to ‘walk’), with the latter two showing impairment in children with PFTs [ 13 ]. Other word properties such as verb transitivity, unaccusativity, and regularity of inflectional paradigms, can also be used to detect atypical verb usage in populations with language impairment (e.g. [ 51 , 52 ]), . At the phonological level, impairments may be revealed by phonological errors [ 53 ], or a bias to produce less complex articulatory patterns (e.g., ‘string’ vs. ‘sing’) [ 54 , 55 ]. Furthermore, a tendency to produce short words may be indicative of phonological short-term memory difficulties [ 56 ]. While most of these variables were included in the work by Svaldi et al. [ 13 ] on postoperative language, such a comprehensive analysis of narrative language has not been reported in studies concerning preoperative language abilities of PFT patients. Current Study In summary, postoperatively, in addition to speech difficulties, there may be language characteristics that distinguish children who have experienced postoperative speech impairment from those who have not. Research into language in the preoperative stage is limited but suggests that linguistic differences between patients with mutism or reduced speech compared to those with habitual speech might already be present before neurosurgical resection. However, knowledge on the exact nature of these preoperative impairments and what language characteristics are related to the emergence of mutism or reduced speech remains unclear. This study will be the first to extensively analyse the preoperative language samples of patients who underwent neurosurgical resection for a PFT and did or did not develop postoperative speech impairment. Language samples of 34 patients will be analysed, replicating and expanding on language analysis procedures used by Svaldi et al. [ 13 ]. By taking postoperative speech status as the grouping factor, we aim to examine whether differences between the groups are present in cognitive domains beyond speech, specifically language functioning, prior to surgery. Namely, we investigate whether preoperative semantic, lexical, morphosyntactic, and phonological characteristics may be related to the emergence of postoperative mutism or reduced speech. Strengthening our understanding of preoperative language abilities as a risk factor for the emergence of this complication will help predict postoperative outcomes and allow better preparation of patients and their parents regarding potential difficulties that might await them after surgery. Method Participants Data from children who underwent surgery for a PFT were retrieved through a database that is part of the prospective European CMS Study [ 57 ]. This is a multicentre prospective study in which approximately 40 treatment centres across 13 European countries collect data on patients treated for a posterior fossa tumour at several stages of the treatment trajectory. The study was registered on November 24, 2014, with ClinicalTrials.gov (registration number: NCT02300766 ). Within the European CMS Study, postoperative mutism or reduced speech is referred to as Postoperative Speech Impairment (POSI), a term we adopt in the current study. Between 2013 and 2024, 794 patients participated in the study. We included patients from treatment centres in The Netherlands, The United Kingdom and Italy, who were assessed in Dutch, English and Italian, respectively. As the number of children undergoing surgery for a PFT within individual countries is limited, inclusion of participants across multiple language backgrounds enabled the investigation of preoperative language differences between children with and without POSI in a sufficiently large cohort. Criteria for inclusion in the current study were: (1) the availability of a language sample collected before the first neurosurgical resection of a PFT, (2) the availability of information on the presence of POSI, (3) the absence of a history of neurological, neurodevelopmental, psychiatric, or learning disorders (with the exception of ADHD with hyperactivity), as well as the absence of pre-existing language disorders that could affect baseline language performance, and (4) the absence of previous chemotherapy and/or radiotherapy treatment. As the focus of the present study was on language characteristics rather than speech motor control, children were not excluded based on speech sound disorders, voice disorders, preoperative dysarthria, or having received therapy for these conditions, provided there was no evidence of an underlying language disorder. In addition to the inclusion criteria, the patient group was further characterised by POSI status, tumour location and histology, and preoperative hydrocephalus, dysarthria and oculomotor abnormalities. POSI was reported by a clinician as present if the patient presented for at least one day postoperatively with mutism (i.e., no production of words or short sentences) or severely reduced speech (i.e., limited to single words or short sentences which can only be elicited after vigorous stimulation). Tumour location was reported after surgical resection, and could include the cerebellar vermis, the right and left hemispheres, the fourth ventricle and/or the brainstem. Tumour histology was reported as medulloblastoma, pilocytic astrocytoma, ependymoma or other. Preoperative hydrocephalus was reported as present or absent. Preoperative dysarthria and oculomotor difficulties were rated by a clinician using a five- and three-point scale, respectively, which we reclassified for the current study as absent (original score 0), or present (any higher score). Participant selection is summarised in Fig. 1 . Sixty-six patients met the inclusion criteria, of whom 16 developed POSI and 50 did not. Given our specific interest in language difficulties in patients with POSI, we performed an additional selection step to create a matched control group of patients without POSI. Eighteen patients who did not develop POSI were matched at the group level to the 16 patients with POSI. As the study included children speaking English and Dutch (West Germanic languages) and Italian (a Romance language), cross-linguistic structural differences (e.g., in morphological richness, pro-drop properties, and function word use) may affect measures such as MLU. Additionally, patients belonged to a wide age band, and a proportion of participants were bi- or multilingual, which may introduce variability in language performance. To minimize potential confounding effects, the groups were matched on age, sex, country and language background (i.e., mono- or multilingual), as well as presence of preoperative hydrocephalus, dysarthria, and oculomotor difficulties. Further clinical characteristics were not considered in the matching procedure to avoid creating atypical samples, given the relationship between POSI and, for instance, tumour histology and location [ 24 ]. Fig. 1. Open in a new tab Process of participant inclusion Materials and Procedures The language samples were collected by a speech and language pathologist, a nurse, or a physician, typically from a few days before surgery to the day of surgery. To elicit a narrative language sample, the ERRNI – Fish Story [ 58 ] was used 1 . All clinicians administering the ERRNI received instructions from a speech and language therapist, based on the ERRNI manual. In this task, the child was presented with a wordless picture book depicting a boy visiting a pet store. Children were instructed to first look through the book silently, after which they were asked to tell the story, with the help of the pictures. The examiner was further instructed to interfere minimally but was allowed to provide encouragement to continue, such as ‘Mhm,’ or general prompts like ‘What happened next?’. However, examiners frequently intervened despite these instructions (to be explained in more detail below). The story told by the child was recorded using an audio recorder. Data Coding Data Preparation: Global Sample Characteristics The narrative language samples were transcribed using the transcription and annotation software ELAN [ 59 ], following a detailed protocol based on the Spontaneous Speech Analysis Procedure (STAP) [ 60 ] and the ERRNI – Fish Story [ 58 ]. Utterances that contained 20% or more unintelligible or hard-to-understand words were excluded from further analysis. Additionally, utterances in which the child mimicked the tester, general comments unrelated to the story and questions asked to the examiner were excluded from further analysis. All remaining utterances were included in the analyses described hereafter. As mentioned above, examiners were instructed not to prompt patients but occasionally deviated from these instructions. Utterances were therefore coded into three categories: (1) elliptic utterances , if they were syntactically dependent on a prompt given by the tester (e.g., Examiner: “What does the mom give to the boy?” , Participant: “Money” ); (2) prompted utterances , if they were prompted by the tester but syntactically independent (e.g., Examiner: “ What is the girl doing with the bags?” , Participant: “The girl is swapping the doll and the fish.“ ); and (3) free utterances , if the child produced them spontaneously or in response to a general, neutral prompt (e.g., Examiner: “ And what do you see here?” , Participant: “ The boy is walking to the pet store.” ). Typically, elliptic and prompted utterances are excluded from further analysis because they do not reflect the child’s independent ability to formulate a sentence and may affect language measures. However, in the current study, neither elliptic nor prompted utterances were excluded, as doing so would result in the loss of a significant portion of data. From this pre-processing step, we calculated several global sample characteristics, which could help enhance the interpretation and contextualisation of the results. First, we calculated the sample size in number of words and number of utterances . Furthermore, the percentage of unclear speech was calculated, reflecting the percentage of utterances that had to be removed from the sample because they contained 20% or more hard-to-understand or unintelligible words. Finally, the percentage of prompted utterances was calculated to reflect the number of utterances produced with support from the examiner (either elliptic or prompted). Psycholinguistic Analyses After preparing the data, a psycholinguistic language sample analysis was performed, similar to the procedures reported by Svaldi et al. [ 13 ]. A total of 24 language measures were extracted from the samples on four levels of language processing (i.e., semantic, lexical, morphosyntactic, and phonological). See Table 1 for an overview of variables per level of language processing. Table 1. Overview of language measures Level of language processing Standard language sample measures Additional psycholinguistic variables Semantics - Concreteness* Familiarity* Imageability* Verb instrumentality (proportion) Lexical Lexical diversity (TTR)* Age of acquisition* Lexical accuracy (percentage) Word frequency* Morphosyntax Mean length of utterance (in words) Verb transitivity (proportion) Grammatical accuracy (percentage) Unaccusativity (proportion) Finiteness index Verb regularity (proportion) Phonology Phonological errors Word length (in phonemes)* Cluster index Open in a new tab Note. Variables marked with * were analysed separately for nouns and verbs. TTR = Type-Token Ratio. See Supplementary Material for detailed descriptions of each variable Standard Language Sample Measures The standard language measures at the lexical level included Type-Token Ratio ( TTR ) and lexical accuracy . TTR was calculated by dividing the number of unique nouns/verbs in the sample by the total number of nouns/verbs, including those uttered as part of hesitations, repetitions and self-corrections. TTR scores could range from 0 to 1, with scores closer to 1 expressing better performance. Lexical accuracy was determined based on whether an utterance contained lexical errors (e.g., semantic paraphasias) and was expressed as the percentage of lexically correct utterances. Morphosyntactic standard language measures included mean length of utterance ( MLU ), grammatical accuracy and finiteness index . MLU was calculated by dividing the total number of words in a language sample by the total number of utterances, with higher scores indicating more syntactically complex language. Grammatical accuracy was determined based on whether an utterance contained grammatical errors (e.g., errors in word order or missing elements) and was expressed as the percentage of grammatically correct utterances. The finiteness index was calculated by dividing the number of correctly produced inflected verbs by the total number of required inflected verbs. Scores for the finiteness index could range from 0 to 1, with scores closer to 1 expressing better performance. At the phonological level, the proportion of phonological errors in the sample was calculated by dividing the total number of errors by the total number of words in the sample. Additional (psycholinguistic) Variables For every unique noun and verb produced by the child, ratings for multiple psycholinguistic variables (e.g., imageability, frequency) were extracted, reflecting different levels of language processing. At the semantic level, concreteness , familiarity , and imageability ratings were extracted. On the lexical level, AoA and word frequency were considered. For the variables verb instrumentality , transitivity , unaccusativity and regularity , at the morphosyntactic level, a trained linguist determined whether the given property could be attributed to every verb produced (e.g., ‘1’ if the verb was instrumental, ‘0’ if the verb was not). Thereafter, the proportion of instrumental/transitive/regular verbs to the total number of verbs was calculated. Unaccusativity was calculated as the proportion of unaccusative verbs to the total number of intransitive verbs. On the phonological level, word length in phonemes was extracted from a database or, if absent, determined by a trained linguist. Additionally, the cluster index was calculated by dividing, per utterance, the total number of correctly produced consonant clusters by the total number of required consonant clusters. Scores could range from 0 to 1, with higher scores reflecting a more accurate production. To ensure accuracy and consistency, the aforementioned data coding was validated by the first and/or second author. See Appendix 6 for an overview of the language-specific databases used. Table 6. Overview of databases used for extracting psycholinguistic values Property English Dutch Italian Concreteness [ 42 ] [ 42 ] Montefinese et al., (2014) Imageability [ 41 ] Van Loon-Vervoorn, (1984) Montefinese et al., (2014) Familiarity [ 41 ] Hermans & De (Houwer, 1994) Montefinese et al., (2014) Age of acquisition [ 48 ] [ 42 ] ItAoA, Montefinesse et al., (2019) Frequency van Heuven et al., (2014) SUBTLEX-NL (Keuleers et al., 2010) Montefinesse et al., (2014) Word length CLEARPOND (Marian et al., 2012) DutchPOND (Marian et al., 2012) Self constructed Open in a new tab Analyses To evaluate the variables extracted from the language samples, we performed a Principal Component Analysis (PCA) [ 61 ]. Given the high number of variables included, this multivariate technique was applied to reduce the number of comparisons to be performed, by clustering variables together that contribute similarly to the variability in the data. Furthermore, considering the PCA is sensitive to the participant-variable ratio, and that we had a relatively high number of variables (24) in relation to the sample size (34), we performed a separate PCA for each level of language processing. Finally, a PCA cannot be performed when variables contain missing values. Therefore, three patients were excluded from the semantic PCA because none of their produced verbs had available concreteness or imageability ratings in the databases used. Subsequently, the variables were evaluated for suitability for the PCA, using the Kaiser–Meyer–Olkin Measure of Sampling Adequacy (KMO) and Bartlett’s test of sphericity; variables with KMO values > 0.5 2 and no violation of the sphericity assumption were retained. Based on the eigenvalues (> 1.0) of the components and the elbow method, we determined how many components to retain; see [ 61 ] for further explanation. Variables with a loading greater than 0.45 or less than − 0.45 were considered to contribute significantly to the variability explained by the component. All variables were then normalised and variables that had a significant contribution to the component were averaged. Where necessary, variables were recoded so that higher scores consistently reflected better performance. Those combined variables were then used for further analyses 3 . Several variables were not eligible for inclusion in the PCA due to insufficient KMO values, indicating that these variables could not be clustered in components. Those variables were therefore compared between groups separately. Extracted components from the PCA and separate variables were compared between groups separately using linear models [ 62 ], including group (POSI vs. no POSI, i.e., habitual speech) as a predictor and age and country as covariates. Additionally, we added group × age interactions to the models, given the higher risk of developing POSI for younger children. For components including TTR nouns and/or TTR verbs, we added number of words as a fixed effect to the model, given TTR’s known sensitivity to sample size [ 46 ]. Additionally, we ran linear models for the global sample characteristics (i.e., sample size in words and utterances, intelligibility and prompting). All statistical analyses were performed using RStudio [ 63 ]. Results Patient Sample Table 2 presents the demographic characteristics of the final patient groups. Across the total sample, participants ranged in age from 3;5 to 16;0 years (median = 8;8, IQR = 6;5–12;3; M = 9;0, SD = 3;5). One child in the POSI group had previously received speech support for a voice disorder. In the no-POSI group, one patient had a diagnosis of ADHD with hyperactivity, and one patient had documented unilateral preoperative damage to the right VIIIth cranial nerve, which was considered a proxy indicator of hearing status in the absence of systematic hearing data. The groups did not differ in tumour histology and tumour location, although a trend towards a group difference in tumour location was observed. See Appendix Table 5 for individual demographic and clinical information. Table 2. Demographic and clinical background information per group POSI ( n = 16) no POSI ( n = 18) χ 2a /t p Age (Y; M) t = - 0.214 0.832 Range 3;9–14;2 3;5–16;0 M ( SD ) 8;10 (3;3) 9;1 (3;8) Median (IQR) 8;7 (6;3–12;3) 8;6 (6;5–12;1) Sex, n (% b ) χ 2 = 0.007 1.000 Male 10 (63) 11 (61) Female 6 (37) 7 (39) Country, n (%) χ 2 = 0.092 1.000 Italy 7 (44) 8 (44) The Netherlands 3 (19) 4 (22) United Kingdom 6 (38) 6 (33) Language background, n (%) χ 2 = 0.993 1.000 Monolingual 13 (81) 14 (78) Bi- and multilingual 3 (19) 3 (17) Unknown 0 (0) 1 (6) Tumour location, n (%) χ 2 = 8.543 0.065 Left cerebellar hemisphere 1 (6) 1 (6) Right cerebellar hemisphere 0 (0) 6 (38) Vermis 5 (28) 7 (44) Fourth ventricle 13 (72) 7 (44) Brainstem 4 (22) 7 (44) Tumour histology, n (%) χ 2 = 6.359 0.156 Medulloblastoma 10 (63) 6 (33) Pilocytic astrocytoma 2 (13) 9 (50) Ependymoma 1 (6) 1 (6) Other 2 (13) 2 (11) Unknown 1 (6) 0 (0) Pre-op dysarthria, n (%) χ 2 = 0.694 0.834 Present 2 (13) 1 (6) Absent 13 (81) 15 (83) Unknown 1 (6) 2 (11) Pre-op hydrocephalus, n (%) χ 2 = 0.034 1.000 Present 12 (75) 13 (72) Absent 4 (25) 5 (28) Pre-op oculomotor difficulties, n (%) χ 2 = 0.174 1.000 Present 9 (56) 9 (50) Absent 5 (31) 6 (33) Unknown 2 (13) 3 (17) Open in a new tab Note. POSI = Postoperative Speech Impairment, defined as mutism or severely reduced speech; No POSI = habitual speech; Y;M = age in years and months; Other = Atypical Teratoid/Rhabdoid Tumour and other tumour types; Pre-op = preoperative. a Chi square test was performed using Monte Carlo simulation (10,000 replicates) to account for small cell counts and limited sample size. b Percentages may not total exactly 100% due to rounding Table 5. Individual demographic and clinical information Open in a new tab Note. ID = participant ID; F = female; M = male; Age = age in years and months; Country = country where patients were treated (NL = Netherlands; UK = United Kingdom; IT = Italy); LB = language background (mono = monolingual; bi-/multi = bi- or multilingual); POSI status = postoperative speech impairment (Habitual = habitual speech; reduced = reduced speech; mute = mutism); MB = medulloblastoma; PA = pilocytic astrocytoma; ED = ependymoma; Other = Atypical Teratoid/Rhabdoid Tumour and other tumour types; VM = vermis; LH = left hemisphere; RH = right hemisphere; FV = fourth ventricle; BS = brain stem; DA = dysarthria; HC = hydrocephalus; OCM = oculomotor difficulties. Global Sample Characteristics In Table 3 , the results of the analysis of the global sample characteristics are reported. The percentage of unclear speech that had to be excluded from further analyses was significantly higher in the group who later developed POSI compared to the group with habitual speech (β = -14.455, p = .024). We also found a significant interaction between group and age . In the POSI group, intelligibility was lower in younger children but improved with increasing age, whereas in children with habitual speech, intelligibility was relatively similar across ages (β = 0.152, p = .007). Additionally, a near-significant difference was observed in the proportion of utterances produced with examiner support (either prompted or elliptic; β = 48.754, p = .064), indicating a tendency for children who later developed POSI to receive more prompts from the examiner. No main effect of group was found in the number of words or utterances that the child produced when telling the story. Age did not show a significant main effect on any of the global sample characteristics. Country significantly impacted the percentage of unclear speech and the sample size in number of utterances. See Appendix Table 8 for the results of the complete models including predictors and covariates. Table 3. Linear models for global sample characteristics POSI No POSI Group β (SE, p ) Age × group β (SE, p ) R ² #Words 105 132 -42.75 (55.20, 0.445) 0.016 (0.48, 0.739) 0.18 #Utterances 18.8 19.9 2.91 (5.57, 0.605) -0.03 (0.05, 0.484) 0.17 %Unclear 3.9% 1.9% -14.46 (6.06, 0.024) 0.15 (0.05, 0.007) 0.22 %Prompted 23.6% 11.6% 48.75 (25.31, 0.064) -0.35 (0.22, 0.127) 0.23 Open in a new tab Note. POSI = postoperative speech impairment, defined as mutism or severely reduced speech; No POSI = habitual speech; #Words = sample size in number of words; #Utterances = sample size in number of utterances; %Unclear = percentage of hard-to-understand and unintelligible speech; SE = standard error Table 8. Complete linear models including predictors and covariates Number of words Number of utterances Predictor β SE t p Predictor β SE t p (Intercept) 105.61 37.23 2.84 0.008 (Intercept) 19.62 3.75 5.23 < 0.001 Group -42.75 55.19 -0.77 0.445 Group 2.91 5.57 0.52 0.605 Age 0.41 0.30 1.37 0.182 Age 0.03 0.03 1.02 0.316 Country (NL) -36.61 24.64 -1.49 0.148 Country (NL) -4.64 2.48 -1.87 0.073 Country (UK) -31.09 21.09 -1.47 0.152 Country (UK) -6.45 2.13 -3.03 0.005 Group × Age 0.16 0.48 0.34 0.739 Group × Age -0.03 0.05 -0.71 0.484 Model fit : R² = 0.30; Adj. R² = 0.18; F(5, 28) = 2.43; p = .060; Residual SE = 52.09 Model fit : R² = 0.30; Adj. R² = 0.17; F(5, 28) = 2.39; p = .063; Residual SE = 5.25 % unclear speech % prompted utterances Predictor β SE t p Predictor β SE t p (Intercept) 6.44 4.09 1.58 0.126 (Intercept) 35.67 17.07 2.09 0.046 Group -14.46 6.06 -2.39 0.024 Group 48.75 25.31 1.93 0.064 Age -0.05 0.03 -1.57 0.127 Age -0.17 0.14 -1.22 0.233 Country (NL) -3.18 2.71 -1.18 0.249 Country (NL) -4.43 11.30 -0.39 0.698 Country (UK) 4.83 2.32 2.08 0.046 Country (UK) -13.88 9.67 -1.44 0.162 Group × Age 0.15 0.05 2.89 0.007 Group × Age -0.35 0.22 -1.57 0.127 Model fit : R² = 0.33; Adj. R² = 0.21; F(5, 28) = 2.71; p = .040; Residual SE = 5.72 Model fit : R² = 0.34; Adj. R² = 0.22; F(5, 28) = 2.92; p = .031; Residual SE = 23.89 Semantics C1 Semantics C2 Predictor β SE t p Predictor β SE t p (Intercept) 0.55 0.24 2.28 0.031 (Intercept) 0.69 0.38 1.81 0.082 Group -0.05 0.36 -0.13 0.900 Group -0.30 0.56 -0.52 0.605 Age -0.00 0.00 -0.43 0.672 Age -0.00 0.00 -0.65 0.524 Country (NL) -0.36 0.16 -2.22 0.035 Country (NL) -1.91 0.25 -7.57 < 0.001 Country (UK) -0.86 0.14 -6.25 < 0.001 Country (UK) -0.10 0.22 -0.46 0.650 Group × Age -0.00 0.00 -0.44 0.666 Group × Age 0.00 0.00 0.38 0.706 Model fit : R² = 0.61; Adj. R² = 0.54; F(5, 28) = 8.82; p < .001; Residual SE = 0.34 Model fit : R² = 0.70; Adj. R² = 0.65; F(5, 28) = 13.07; p < .001; Residual SE = 0.53 Lexical C1 Lexical C2 Predictor β SE t p Predictor β SE t p (Intercept) -0.76 0.28 -2.74 0.011 (Intercept) -0.38 0.59 -0.65 0.524 Group -0.73 0.41 -1.78 0.086 #Words -0.01 0.00 -2.63 0.014 Age 0.00 0.00 0.13 0.898 Group 0.51 0.77 0.66 0.517 Country (NL) 1.19 0.18 6.47 < 0.001 Age 0.01 0.00 1.63 0.115 Country (UK) 1.58 0.16 9.99 < 0.001 Country (NL) 0.56 0.35 1.59 0.123 Group × Age 0.01 0.00 1.53 0.137 Country (UK) 0.59 0.30 1.96 0.061 Model fit : R² = 0.80; Adj. R² = 0.77; F(5, 28) = 22.55; p < .001; Residual SE = 0.39 Model fit : R² = 0.44; Adj. R² = 0.31; F(6, 27) = 3.48; p = .011; Residual SE = 0.72 Age of Acquisition verbs Frequency nouns Predictor β SE t p Predictor β SE t p (Intercept) -0.52 0.43 -1.21 0.237 (Intercept) -1.20 0.69 -1.74 0.093 Group -0.22 0.64 -0.34 0.739 Group 1.25 1.02 1.22 0.232 Age 0.00 0.00 0.58 0.565 Age 0.01 0.01 2.35 0.026 Country (NL) 1.76 0.29 6.12 < 0.001 Country (NL) -0.26 0.46 -0.56 0.578 Country (UK) -0.38 0.25 -1.54 0.135 Country (UK) -0.23 0.39 -0.59 0.561 Group × Age 0.00 0.01 0.63 0.535 Group × Age -0.01 0.01 -1.51 0.143 Model fit : R² = 0.69; Adj. R² = 0.63; F(5, 28) = 12.38; p < .001; Residual SE = 0.61 Model fit : R² = 0.21; Adj. R² = 0.07; F(5, 28) = 1.48; p = .228; Residual SE = 0.97 Morphosyntax C1 Morphosyntax C2 Predictor β SE t p Predictor β SE t p (Intercept) -0.20 0.56 -0.35 0.729 (Intercept) -0.21 0.64 -0.33 0.743 Group -0.96 0.84 -1.15 0.261 Group -1.23 0.95 -1.29 0.206 Age 0.00 0.00 0.66 0.514 Age 0.00 0.01 0.74 0.467 Country (NL) -0.47 0.37 -1.25 0.221 Country (NL) -0.57 0.42 -1.34 0.191 Country (UK) 0.11 0.32 0.35 0.732 Country (UK) -0.29 0.36 -0.79 0.439 Group × Age 0.01 0.01 1.04 0.307 Group × Age 0.01 0.01 1.44 0.162 Model fit : R² = 0.20; Adj. R² = 0.06; F(5, 28) = 1.40; p = .255; Residual SE = 0.79 Model fit : R² = 0.28; Adj. R² = 0.15; F(5, 28) = 2.15; p = .089; Residual SE = 0.90 Verb regularity Phonology C1 Predictor β SE t p Predictor β SE t p (Intercept) 0.86 0.50 1.73 0.095 (Intercept) 0.67 0.15 4.40 < 0.001 Group -0.08 0.74 -0.11 0.916 Group 0.05 0.22 0.24 0.810 Age 0.00 0.00 0.03 0.978 Age 0.00 0.00 1.89 0.069 Country (NL) -1.77 0.33 -5.40 < 0.001 Country (NL) -1.01 0.10 -10.09 < 0.001 Country (UK) -1.32 0.28 -4.69 < 0.001 Country (UK) -2.04 0.09 -23.81 < 0.001 Group × Age 0.00 0.01 -0.01 0.994 Group × Age 0.00 0.00 -0.13 0.895 Model fit : R² = 0.59; Adj. R² = 0.52; F(5, 28) = 8.09; p < .001; Residual SE = 0.69 Model fit : R² = 0.96; Adj. R² = 0.95; F(5, 28) = 128.00; p < .001; Residual SE = 0.21 Phonology C2 Predictor β SE t p (Intercept) 0.45 0.03 16.83 < 0.001 Group -0.05 0.04 -1.16 0.258 Age 0.00 0.00 1.12 0.272 Country (NL) 0.04 0.02 2.11 0.044 Country (UK) 0.03 0.02 1.76 0.090 Group × Age 0.00 0.00 0.76 0.451 Model fit : R² = 0.30; Adj. R² = 0.17; F(5, 28) = 2.35; p = .067; Residual SE = 0.04 Open in a new tab Psycholinguistic Analysis Principal Component Analysis In the Principal Component Analysis, two components (C1 and C2) were extracted per level of language processing. At the semantic level, all variables were included in the PCA. For Semantics C1, the variables with a significant loading were concreteness verbs , imageability nouns , imageability verbs , and instrumentality verbs , and for Semantics C2 these were concreteness nouns and familiarity nouns . At the lexical level, the variables AoA verbs and frequency nouns were not suitable for inclusion in the PCA due to insufficient KMO values. The components extracted from the remaining variables were C1, where lexical correctness , AoA nouns and frequency verbs had a significant contribution, and C2, containing the variables TTR verbs and TTR nouns . At the morphosyntactic level, unaccusativity was excluded from further analysis, as only 6 children produced unaccusative verbs. Regularity was not included in the PCA due to an insufficient KMO value. Two components were extracted from the remaining variables: C1 containing grammatical correctness and finiteness index , and C2 containing MLU and transitivity . At the phonological level, all variables were included in the PCA. In C1, word length verbs and word length nouns had a significant loading. In C2, the variables with a significant loading were phonological errors and cluster index . See Appendix Table 7 for the full overview of the variables and their loadings in the components. Table 7. Components and variable loadings extracted from the PCA Level of Language Processing Variable C1 C2 Semantics Concreteness verbs 0.92 0.16 Concreteness nouns 0.21 0.87 Imageability verbs 0.95 -0.11 Imageability nouns 0.66 0.14 Familiarity nouns 0.02 0.90 Instrumentality verbs -0.78 -0.19 Lexical TTR nouns 0.26 0.80 TTR verbs 0.02 0.90 Lexical correctness -0.79 -0.23 AoA nouns 0.74 0.18 Frequency verbs 0.84 -0.01 Morphosyntax MLU 0.42 0.60 Grammatical correctness 0.96 0.12 Finiteness index 0.92 0.30 Transitivity 0.08 0.92 Phonology Word Length verbs 0.95 -0.03 Word Length nouns 0.95 0.07 Phonological errors -0.04 0.95 Cluster Index -0.11 -0.94 Open in a new tab Note. Values in bold relate to the variables that are normalised and averaged to construct the component; C1 = Component 1; C2 = Component 2 Group Comparisons Linear models were constructed for each component and for the separate variables not included in the PCA, including number of words (C2), age and country as covariates, group as a predictor, and the interaction group × age . See Table 4 for an overview of the models for each component and separate variable and see Appendix Table 8 for the results of the complete models including predictors and covariates. Table 4. Linear model per component or separate variable Component/variable Group β (SE, p ) Age × group β (SE, p ) R ² Semantic C1: Concr. V+Imageab. V+ -0.19 (0.12, 0.103) -0.00 (0.00, 0.666) 0.54 Instrum. V+Imageab. N C2: Concr. N + Fam. N -0.30 (0.57, 0.605) 0.00 (0.00, 0.706) 0.65 Lexical C1: Lex. cor + AoA N+ -0.73 (0.41, 0.086) 0.01 (0.00, 0.137) 0.77 Freq. V C2: TTR N + TTR V 0.51 (0.77, 0.517) -0.00 (0.01, 0.722) 0.31 AoA V -0.22 (0.64, 0.739) 0.00 (0.01, 0.535) 0.63 Freq. N 1.25 (1.02, 0.232) -0.01 (0.01, 0.143) 0.07 Morphosyntax C1: Gram. cor.+Finite. ind. -0.96 (0.84, 0.261) 0.01 (0.01, 0.307) 0.06 C2: MLU+Trans. V -1.23 (0.95, 0.206) 0.01 (0.01, 0.162) 0.15 Regularity V -0.08 (0.74, 0.916) 0.00 (0.01, 0.994) 0.52 Phonology C1: Length V+Length N 0.05 (0.22, 0.810) -0.00 (0.00, 0.895) 0.95 C2: Phon. err.+Clust. ind. -0.05 (0.04, 0.257) -0.00 (0.76, 0.451) 0.17 Open in a new tab Note. C = Component; N = Nouns; V = Verbs; Concr. = Concreteness; Imageab. = Imageability; Instrum. = Instrumentality; Fam. = Familiarity; Lex. cor. = Lexical correctness; AoA = Age of acquisition; Freq. = Frequency; TTR = Type-Token Ratio; Gram. cor. = grammatical correctness; Finite. ind. = finiteness index; Trans. = Transitivity; Phon. err. = Phonological errors; Clust. ind. = Cluster index; SE = standard error Country had a significant effect on the components/variables Semantic C1, Semantic C2, Lexical C1, AoA verbs, verb regularity and Phonology C1 and C2. Additionally, age significantly impacted Frequency nouns, and number of words had a significant effect on the component Lexical C2, including TTR nouns and verbs. Group appeared not to be a significant predictor of the scores for any of the components or separate variables. No significant age × group interactions were found for any of the components or separate variables. Discussion To examine whether preoperative language characteristics are related to the emergence of POSI, we compared preoperative narrative language samples of 16 patients who developed POSI and those of 18 patients who did not following neurosurgical resection for a PFT. Younger children with POSI produced more unclear speech relative to older children with POSI; this effect was not seen for children without POSI. In contrast, at the level of language, the psycholinguistic analyses showed no differences between the two groups in the measures reflecting semantic, lexical, morphosyntactic, and phonological processing, and no interactions between group and age. In this section, we will discuss the group comparisons and how this relates to previous research on this population. Global Sample Characteristics Considering the global sample characteristics, the percentage of unclear speech that had to be excluded from further analysis was higher in the patients who later developed POSI compared to those who did not. Bianchi et al. [ 14 ] characterised preoperative impairments in the vast majority of patients who went on to develop mutism as a phonological disorder , implying a disorder at the level of language. However, they also referred to these difficulties as a phonetic disorder and apraxia of speech , which are motor speech disorders, leaving some uncertainty about the exact nature of these difficulties. While unclear speech was excluded from further analysis, motor speech-related factors potentially contributed to the speech being unintelligible, such as respiration, phonation, resonance, prosody, articulation, and speech rate [ 64 , 65 ]. Several studies reported preoperative difficulties in the domain of speech in children who developed mutism, such as dysarthria [ 15 , 65 ], ataxia [ 65 – 67 ] and apraxia of speech [ 15 ], although these did not extensively assess the nature of these speech difficulties. A more systematic, in-depth analysis of preoperative speech, intelligibility and speech sound errors could provide more insight into the phonological and/or speech impairments these patients experience and whether these could be a risk factor for the development of POSI. Interestingly, age differentially affected the percentage of unclear speech across groups. In children who later developed POSI, unclear speech was more frequent in younger children and decreased with increasing age, whereas no age-related association was observed in children who did not develop POSI. This aligns with the reported higher risk of speech impairments in younger children, namely for the occurrence of mutism or reduced speech [ 5 , 6 ]. Mutism or reduced speech is hypothesised to be a form of cerebello-cerebral diaschisis, characterised by damage to the connections between the cerebellum and cerebrum, resulting in hypoactivity of the cerebral hemispheres [ 68 , 69 ]. These connections may be more vulnerable to damage in children, due to the incomplete maturation of the brain, resulting in a higher incidence of mutism or reduced speech [ 27 ]. The presence of this age effect already before neurosurgical resection suggests that not only the surgical intervention [ 24 ], but also the tumour itself may have a stronger impact on these pathways in younger children. Additionally, we observed an imbalance in the amount of prompting provided by the testers between groups, with more prompting for the POSI group, albeit just above the significance threshold. This occurred despite instructions to interfere minimally and only provide general prompts when the child needed encouragement [ 58 ]. This pattern might suggest that the adynamic verbal output pattern often observed after neurosurgical resection [ 18 ] could already be present to some extent in the preoperative stage, prompting testers to provide more support. However, this prompting should be investigated more systematically, in greater detail and in a larger group of patients before firm conclusions can be drawn. Psycholinguistic Analysis At the level of language, none of the measures in our psycholinguistic analysis differentiated children who developed POSI from those who did not based on their preoperative abilities. These results are in contrast with previous research by Di Rocco et al. [ 15 ], who found a lower MLU and problems with lexical naming and verbal fluency for children who developed mutism before surgery, and Bianchi et al. [ 14 ], who expanded Di Rocco et al.’s sample, suggested a relationship between preoperative language performance and mutism. Critically, MLU was defined by Di Rocco et al. [ 15 ] based on parental report [ 70 , 71 ]. This measure may not be entirely comparable to the typical MLU calculation in linguistic studies, which reflects the length of syntactic units and distinctions between devices such as conjunction vs. subordination, which would be difficult for parents to judge [ 72 ]. Furthermore, motor speech symptoms (rather than language difficulties), such as impaired coordination of respiration, phonation, and articulation, may lead to more frequent pauses, potentially causing sentences to be judged by parents as shorter. In line with this, a perceptual analysis of speech carried out in the European CMS Study [ 73 ] by our team showed that children who went on to develop POSI speak in shorter phrases before surgery, compared to those who do not. Similarly, performance on language tasks such as picture naming or verbal fluency may also be influenced by poor motor speech, especially when the tasks are timed (as is the case of verbal fluency) or reaction times are taken into account, as done in some picture naming tests [ 74 , 75 ]. Additionally, performance on language tasks is not solely determined by language or speech abilities. Verbal fluency, for instance, also relies on neuropsychological processes such as executive functioning and processing speed [ 76 , 77 ]. The group differences in verbal fluency found by Di Rocco and colleagues could therefore also be driven by difficulties in these domains. Horne et al. [ 78 ] and Cámara et al. [ 19 ] reported postoperative impairments in executive functioning and processing speed. Given the overall poorer preoperative neuropsychological status of patients who developed mutism reported by Mariën [ 68 ], these difficulties are likely at least to some extent already present before neurosurgical resection. Combined, this suggests that children with and without mutism or reduced speech may differ in their (motor) speech or overall neuropsychological status before neurosurgical resection, which may affect their performance on language tasks, while differences in language itself may be absent. From a clinical perspective, the present findings therefore do not support the inclusion of preoperative language measures in risk prediction models for postoperative mutism or reduced speech. Instead, the observed preoperative differences in speech intelligibility suggest that speech-related measures may be of interest as potential clinical markers for POSI in future risk prediction models, although this requires further investigation. Nevertheless, children with PFTs in general may still exhibit preoperative language impairments (relative to healthy children, not examined in this study) due to tumour presence and growth. This was shown for word finding by Persson et al. [ 29 ]. Further studies could provide more insight into the extent of difficulty and the specific levels of language affected. Something that has to be taken into account when interpreting the results is the type of tumour diagnosed in patients. The groups were not matched on tumour type, as this would create atypical groups, given the relatedness of tumour type (namely medulloblastomas) to emergence of mutism or reduced speech [ 4 , 24 ]. In line with this, the proportion of medulloblastomas is relatively (albeit non-significantly) higher in our POSI group (63%), compared to the no POSI group (33%). However, Persson et al. [ 29 ] did not find a relation between tumour type and word-finding difficulties. They hypothesize that, for medulloblastomas, mutism or reduced speech may not be related to the tumour itself, but to the high-risk surgery medulloblastomas require. This could also explain the present findings, where a higher proportion of medulloblastomas was found in the group that later developed POSI, but no group differences in language were found before neurosurgical resection; perhaps postoperative language differences found in earlier studies are related to more aggressive surgery in the POSI group, driven by a higher proportion of medulloblastomas. Still, we find more unclear speech in patients with POSI before surgery, indicating that not all impairments can be explained this way. Further research on the relationship between tumour type, location, and preoperative speech and language performance is needed to better understand the impact of tumour characteristics on performance. Limitations and Suggestions for Future Research A limitation of this study concerns the type of test (i.e., a picture-based narrative task) that was used to collect a language sample, which might have affected the quality of the language samples extracted. Compared to storytelling tasks, story retelling tasks have been associated with better narrative macrostructure [ 79 ], and picture-sequence narrative tasks with greater lexical diversity than single-picture tasks [ 80 ]. Furthermore, in a picture-based narrative task, the predetermined content may have limited children in using their linguistic abilities to their full potential and restricted the potential range of psycholinguistic variable values, such as imageability and word length. Compared to no-visual narrative tasks, samples elicited with a picture-based task were found to be less elaborate [ 80 ], less lexically complex [ 81 ], and had shorter MLUs [ 80 ]. Moreover, a free no-visual conversation task was found to better capture atypical patterns in semantic properties such as concreteness and imageability and to generally detect more atypical language profiles than a picture-description task [ 13 ]. Although we believe differences between groups might primarily be in the domain of speech, the task used might have made it difficult to capture subtle language differences. Opting for more naturalistic approaches (e.g., parent-child or examiner-child interactions; see, for instance, Ellis Weismer et al. [ 82 ]) may provide a more ecologically valid and comprehensive measurement of language abilities. Nonetheless, it should be noted that the European CMS study is a large-scale study across many centres and languages. The ERRNI procedure [ 58 ], using a picture-book-based narrative, creates a setting where language samples can be gathered in a standardised way, despite the large variability in settings, languages and professional backgrounds of those administering the task. Although some variability related to these factors cannot be entirely ruled out, other, more naturalistic or interview-based data collection procedures might introduce much more variability and require even greater expertise from examiners, which may have a detrimental effect on the feasibility of the study or the quality of the data. Furthermore, the way patients were classified as experiencing mutism or reduced speech may have impacted our results. In the European CMS study, decisions about whether a child showed mutism or reduced speech were not made by speech and language therapists, but by clinicians such as surgeons, paediatricians or nursing staff. While agreement on identifying mutism as a complete absence of speech is likely high, the categorisation of reduced speech may be more variable. What one rater considers a clinically significant reduction in speech may not be judged as such by another. This heterogeneity may have introduced noise into the POSI/no POSI classification, potentially reducing our ability to detect associations with preoperative language abilities. Future research may benefit from standardised criteria or rater training, or from involving speech and language therapists in the diagnostic process, to improve the reliability of POSI classification. Another limitation concerns the availability of further language and neuropsychological assessment data. Because we performed a language sample analysis, 86 patients for whom there was no sample available had to be excluded. Information on the reason for not performing the language task was often not available, but this could be related to the characteristics or neuropsychological status of the patients. Mariën et al. [ 68 ] proposed a relationship between neuropsychological status and the emergence of mutism, making the excluded group particularly interesting for future research. Although language test administration is difficult in this group, future research could focus on suitable tests to gain a better understanding of the language abilities of children with worse neuropsychological status. In addition, no standardised neuropsychological or formal language test scores were available in the present dataset. The inclusion of such measures may provide additional context for interpreting language findings and could contribute to a more comprehensive understanding of the broader cognitive profile associated with POSI in future studies. Conclusion This study aimed to identify risk factors in the domain of language that could be related to the emergence of POSI, using a comprehensive analysis of preoperative language samples. The global sample characteristics and the linguistic abilities of patients who developed POSI were compared to those of patients who did not develop POSI. Results revealed a higher proportion of unclear speech in the group that later developed POSI. The linguistic analysis of the language samples did not reveal any group differences. Our results indicate that preoperative differences between the two groups may be primarily related to motor speech rather than to the microstructural aspects of language assessed through our psycholinguistic analyses. The language differences between patients with and without POSI observed postoperatively may thus be correlated with the effects of neurosurgical tumour resection. This study adds to the limited body of preoperative research performed in this population and suggests that already at the preoperative stage, there might be speech characteristics that are related to the emergence of postoperative mutism or reduced speech. Additional research is needed to further explore the predictive value of speech characteristics (which we predict will show more prominent group differences, given our findings on intelligibility). Such advances will help form an increasingly accurate risk prediction of the development of mutism or reduced speech. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 (89.3KB, docx) Acknowledgements We thank the CMS study team for their sustained efforts in patient recruitment and longitudinal follow-up, and the patients whose time and participation made this research possible. Appendix Author contributions Conceptualisation : Aliene Reinders and Vânia de Aguiar; Methodology: Aliene Reinders, Vânia de Aguiar and Cheyenne Svaldi; Project administration and resources: Jonathan Kjær Grønbæk, René Mathiasen, Christine Dahl, Marianne Juhler, Barry Pizer, Colin Thorbinson, Kristian Aquilina, Eelco Hoving, Andrea Carai, Angela Mastronuzzi and Vânia de Aguiar; Investigation: Aliene Reinders, Cheyenne Svaldi and Bianca Andreozzi; Data curation and formal analysis : Aliene Reinders and Cheyenne Svaldi; Supervision: Vânia de Aguiar and Roel Jonkers; Writing - original draft preparation: Aliene Reinders; Writing - review and editing: all authors reviewed the manuscript; Funding acquisition: Vânia de Aguiar, Ditte Boeg Thomsen, Marianne Juhler and Jonathan Kjær Grønbæk. Funding This publication is supported by funding awarded to project Verb Processing and Verb Learning in Children With Paediatric Posterior Fossa Tumours (with file number VI.Vidi.201.003) of the research program NWO-Talentprogramma Vidi SGW 2020 financed by the Dutch Research Council (NWO). Jonathan Kjær Grønbæk and Ditte Boeg Thomsen received funding from the Inge Lehmann grant (grant number 10.46540/4302-00027B) from the Independent Research Fund Denmark. Karin Persson received funding from The Swedish Childhood Cancer Foundation, Queen Silvia’s Jubilee Fund, Jonas Foundation. Data Availability The paper reports a secondary analysis of data from the European CMS study. Requests for access to and reuse of the data should be directed to the principal investigator of the European CMS study, René Mathiasen ([email protected]). Declarations Ethical Considerations The current study used patient data from the European CMS Study. Data collection for this project was approved by the Research Ethics Committees of the Capital Region in Denmark (H-6-2014-002). The first participation of Dutch centres was approved by the Medical Ethics Review Committee (CMO) of Radboud University Medical Center, Nijmegen (NL55516.091.15). Following a temporary discontinuation, the second participation of Dutch centres received ethical approval from the Medical Ethics Review Committee NedMec (METC NedMec; NL81967.041.22). Participation of centres in the UK was approved by the North West - Liverpool East Research Ethics Committee (16/NW/0633). Participation of the Italian centre was approved by the Ethical Committee of the IRCCS Bambino Gesù Children’s Hospital (1923/2019). All procedures were conducted in accordance with the Declaration of Helsinki. Human Ethics and Consent to Participate Written informed consent was obtained from all individual participants or from a parent of the participating children. Competing interests The authors declare no competing interests. Footnotes 1 Although the ERRNI provides normative data from 4 years of age and has been standardised in English, the present study used the picture book only as a narrative elicitation tool. It served to generate data which were used for group comparison; no normative scores were calculated. 2 Given that the main goal of performing the PCA was dimensionality reduction, we choose to be lenient with this cut-off. Therefore, if there were any variables that had a slightly lower (e.g., 0.48) KMO value, but did cluster together with other variables in a meaningful way later on in the analyses, we decided to include those variables in the analyses. 3 In a PCA, every variable has a loading onto every component, some stronger than others (expressed by the loading value). It is possible to extract components that incorporate all variables while considering their specific loadings. However, this leads to less interpretable data, as each component becomes a mixture of all variables. Therefore, we choose to use the PCA as a guide to identify which variables can be clustered together and we averaged the scores of only those variables that made a significant contribution to the component. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References 1. Rickert CH, Paulus W. Epidemiology of central nervous system tumors in childhood and adolescence based on the new WHO classification. Child’s Nerv Syst. 2001;17:503–11. 10.1007/s003810100496. [ DOI ] [ PubMed ] [ Google Scholar ] 2. Formentin C, Joaquim AF, Ghizoni E. Posterior fossa tumors in children: current insights. Eur J Pediatr Springer Berlin Heidelberg. 2023;182:4833–50. 10.1007/s00431-023-05189-5. [ Google Scholar ] 3. Wickenhauser ME, Khan RB, Raches D, Ashford JM, Robinson GW, Russell KM, et al. Characterizing Posterior Fossa Syndrome: A Survey of Experts. Pediatr Neurol Elsevier. 2020;104:19–22. 10.1016/j.pediatrneurol.2019.11.007. [ Google Scholar ] 4. Catsman-Berrevoets CE, Aarsen FK. The spectrum of neurobehavioural deficits in the Posterior Fossa Syndrome in children after cerebellar tumour surgery. Cortex. 2010;46:933–46. 10.1016/j.cortex.2009.10.007. [ DOI ] [ PubMed ] [ Google Scholar ] 5. Grønbæk JK, Wibroe M, Toescu S, Frič R, Thomsen BL, Møller LN, et al. Postoperative speech impairment and surgical approach to posterior fossa tumours in children: a prospective European multicentre cohort study. Lancet Child Adolesc Health. 2021;5:814–24. 10.1016/S2352-4642(21)00274-1. [ DOI ] [ PubMed ] [ Google Scholar ] 6. Khan RB, Patay Z, Klimo P, Huang J, Kumar R, Boop FA, et al. Clinical features, neurologic recovery, and risk factors of postoperative posterior fossa syndrome and delayed recovery: a prospective study. Neuro-Oncol. 2021;23:1586–96. 10.1093/neuonc/noab030. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 7. Robertson PL, Muraszko KM, Holmes EJ, Sposto R, Packer RJ, Gajjar A et al. Incidence and severity of postoperative cerebellar mutism syndrome in children with medulloblastoma: a prospective study by the Children’s Oncology Group. 2006 [cited 2025 Feb 3]; 10.3171/ped.2006.105.6.444 8. Khan RB, Patay Z, Klimo P Jr, Huang J, Kumar R, Boop FA, et al. Clinical features, neurologic recovery, and risk factors of postoperative posterior fossa syndrome and delayed recovery: a prospective study. Neuro-Oncol. 2021;23:1586–96. 10.1093/neuonc/noab030. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 9. Catsman-Berrevoets C, Patay Z. Cerebellar mutism syndrome. In: Manto M, Huisman TAGM, editors. Handb Clin Neurol. Elsevier; 2018. pp. 273–88. 10.1016/B978-0-444-64189-2.00018-4. 10. Gelabert-González M, Fernández-Villa J. Mutism after posterior fossa surgery. Review of the literature. Clin Neurol Neurosurg. 2001;103:111–4. 10.1016/S0303-8467(01)00125-1. [ DOI ] [ PubMed ] [ Google Scholar ] 11. Gudrunardottir T, Morgan AT, Lux AL, Walker DA, Walsh KS, Wells EM, et al. Consensus paper on post-operative pediatric cerebellar mutism syndrome: the Iceland Delphi results. Childs Nerv Syst. 2016;32:1195–203. 10.1007/s00381-016-3093-3. [ DOI ] [ PubMed ] [ Google Scholar ] 12. Paquier PF, Walsh KS, Docking KM, Hartley H, Kumar R, Catsman-Berrevoets CE. Post-operative cerebellar mutism syndrome: rehabilitation issues. Childs Nerv Syst. 2020;36:1215–22. 10.1007/s00381-019-04229-6. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 13. Svaldi C, Paquier P, Keulen S, Van Elp H, Catsman-Berrevoets C, Kingma A et al. Characterising the long-term language impairments of children following cerebellar tumour surgery by extracting psycholinguistic properties from spontaneous language. The Cerebellum [Internet]. 2023 [cited 2024 Jan 2]; 10.1007/s12311-023-01563-z 14. Bianchi F, Chieffo DPR, Frassanito P, Di Rocco C, Tamburrini G. Cerebellar mutism: the predictive role of preoperative language evaluation. Childs Nerv Syst. 2020;36:1153–7. 10.1007/s00381-019-04252-7. [ DOI ] [ PubMed ] [ Google Scholar ] 15. Di Rocco C, Chieffo D, Frassanito P, Caldarelli M, Massimi L, Tamburrini G. Heralding Cerebellar Mutism: Evidence for Pre-surgical Language Impairment as Primary Risk Factor in Posterior Fossa Surgery. Cerebellum. 2011;10:551–62. 10.1007/s12311-011-0273-2. [ DOI ] [ PubMed ] [ Google Scholar ] 16. van Dongen HR, Catsman-Berrevoets CE, van Mourik M. The syndrome of cerebellar mutism and subsequent dysarthria. Neurol Wolters Kluwer. 1994;44:2040–2040. 10.1212/WNL.44.11.2040. [ Google Scholar ] 17. Pollack IF, Polinko P, Albright AL, Towbin R, Fitz C. Mutism and pseudobulbar symptoms after resection of posterior fossa tumors in children: incidence and pathophysiology. Neurosurgery. 1995;37:885–93. 10.1227/00006123-199511000-00006. [ DOI ] [ PubMed ] [ Google Scholar ] 18. De Smet HJ, Baillieux H, Wackenier P, De Praeter M, Engelborghs S, Paquier PF, et al. Long-term cognitive deficits following posterior fossa tumor resection: A neuropsychological and functional neuroimaging follow-up study. Neuropsychology. 2009;23:694–704. 10.1037/a0016106. [ DOI ] [ PubMed ] [ Google Scholar ] 19. Cámara S, Fournier MC, Cordero P, Melero J, Robles F, Esteso B, et al. Neuropsychological Profile in Children with Posterior Fossa Tumors with or Without Postoperative Cerebellar Mutism Syndrome (CMS). Cerebellum Springer US. 2020;19:78–88. 10.1007/s12311-019-01088-4. [ Google Scholar ] 20. Persson K, Grønbæk J, Tiberg I, Fyrberg Å, Castor C, Andreozzi B, et al. Postoperative word-finding difficulties in children with posterior fossa tumours: a crosslinguistic European cohort study. Childs Nerv Syst Springer Berlin Heidelberg. 2025;41:1–14. 10.1007/s00381-025-06787-4. [ Google Scholar ] 21. Siffert J, Young Poussaint T, Goumnerova LC, Scott RM, LaValley B, Tarbell NJ, et al. Neurological Dysfunction Associated With Postoperative Cerebellar Mutism. J Neurooncol. 2000;48:75–81. 10.1023/A:1006483531811. [ DOI ] [ PubMed ] [ Google Scholar ] 22. Ait Khelifa-Gallois N, Puget S, Longaud A, Laroussinie F, Soria C, Sainte-Rose C, et al. Clinical Evidence of the Role of the Cerebellum in the Suppression of Overt Articulatory Movements During Reading. A Study of Reading in Children and Adolescents Treated for Cerebellar Pilocytic Astrocytoma. Cerebellum Springer US. 2015;14:97–105. 10.1007/s12311-014-0612-1. [ Google Scholar ] 23. Nagel BJ, Delis DC, Palmer SL, Reeves C, Gajjar A, Mulhern RK. Early patterns of verbal memory impairment in children treated for medulloblastoma. Neuropsychol Am Psychol Association. 2006;20:105–12. 10.1037/0894-4105.20.1.105. [ Google Scholar ] 24. Grønbæk JK, Boeg Thomsen D, Persson K, Mathiasen R, Juhler M. The Cerebellar Mutism Syndrome: Risk Assessment, Prevention and Treatment. Adv Tech Stand Neurosurg [Internet]. Springer, Cham; 2023. pp. 65–94. [cited 2025 Jan 20] 10.1007/978-3-031-28202-7_4. 25. Toescu SM, Samarth G, Horsfall HL, Issitt R, Margetts B, Phipps KP et al. Fourth ventricle tumors in children: complications and influence of surgical approach. 2020 [cited 2025 Jan 13]; 10.3171/2020.6.PEDS2089 26. Mariën P, De Smet HJ, Wijgerde E, Verhoeven J, Crols R, De Deyn PP. Posterior fossa syndrome in adults: A new case and comprehensive survey of the literature. Cortex. 2013;49:284–300. 10.1016/j.cortex.2011.06.018. [ DOI ] [ PubMed ] [ Google Scholar ] 27. İldan F, Tuna M, Erman T, Göçer AI, Zeren M, Çetinalp E. The Evaluation and Comparison of Cerebellar Mutism in Children and Adults after Posterior Fossa Surgery: Report of two Adult Cases and Review of the Literature. Acta Neurochir (Wien). 2002;144:463–73. 10.1007/s007010200067. [ DOI ] [ PubMed ] [ Google Scholar ] 28. Wibroe M, Rochat P, Juhler M. Cerebellar Mutism Syndrome and Other Complications After Surgery in the Posterior Fossa in Adults: A Prospective Study. World Neurosurg. 2018;110:e738–46. 10.1016/j.wneu.2017.11.100. [ DOI ] [ PubMed ] [ Google Scholar ] 29. Persson K, Boeg Thomsen D, Fyrberg Å, Castor C, Aasved Hjort M, Andreozzi B, et al. Preoperative word-finding difficulties in children with posterior fossa tumours: a European cross-sectional study. Childs Nerv Syst. 2024;40:87–97. 10.1007/s00381-023-06119-4. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 30. Svaldi C, Ntemou E, Jonkers R, Kohnen S, de Aguiar V. Language outcomes in children who underwent surgery for the removal of a posterior fossa tumor: A systematic review. Eur J Paediatr Neurol. 2024;48:129–41. 10.1016/j.ejpn.2023.12.005. [ DOI ] [ PubMed ] [ Google Scholar ] 31. Butler LK, Shen L, Chenausky KV, La Valle C, Schwartz S, Tager-Flusberg H. Lexical and Morphosyntactic Profiles of Autistic Youth With Minimal or Low Spoken Language Skills. Am J Speech Lang Pathol. 2023;32:733–47. 10.1044/2022_AJSLP-22-00098. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 32. Govindarajan K, Paradis J. Narrative abilities of bilingual children with and without Developmental Language Disorder (SLI): Differentiation and the role of age and input factors. J Commun Disord. 2019;77:1–16. 10.1016/j.jcomdis.2018.10.001. [ DOI ] [ PubMed ] [ Google Scholar ] 33. Hudson LJ, Murdoch BE. Spontaneously generated narratives of children treated for posterior fossa tumour. Aphasiology Routledge. 1992;6:549–66. 10.1080/02687039208249491. [ Google Scholar ] 34. Docking K, Munro N, Marshall T, Togher L. Narrative skills of children treated for brain tumours: The impact of tumour and treatment related variables on microstructure and macrostructure. Brain Inj. 2016;30:1005–18. 10.3109/02699052.2016.1147602. [ DOI ] [ PubMed ] [ Google Scholar ] 35. Shallice T. Specialisation within the semantic system. Cogn Neuropsychol Routledge. 1988;5:133–42. 10.1080/02643298808252929. [ Google Scholar ] 36. Patterson K, Shewell C. Speak and spell: Dissociations and word-class effects. Hillsdale, NJ, US: Lawrence Erlbaum Associates, Inc;: Cogn Neuropsychol Lang; 1987. pp. 273–94. [ Google Scholar ] 37. Garrett M. Levels of processing in speech production. Butterworth B Ed Lang Prod Vol 1. Academic Press; 1980. 38. Levelt WJM, editor. Speaking: From Intention to Articulation [Internet]. Cambridge, MA: The MIT Press; 1989. 10.7551/mitpress/6393.003.0009. [ Google Scholar ] 39. Croft W, Morphosyntax. Constructions of the World’s Languages [Internet]. High. Educ. Camb. Univ. Press. Cambridge University Press; 2022. [cited 2025 Feb 3]. 10.1017/9781316145289. 40. Svaldi C, Galli J-I, Paquier P, Keulen S, Van Elp H, Catsman-Berrevoets C, et al. Errors in the Spontaneous Language of Survivors of Pediatric Cerebellar Tumors. The Cerebellum. Springer US. 2025;24:1–7. 10.1007/s12311-024-01754-2. [ Google Scholar ] 41. Scott GG, Keitel A, Becirspahic M, Yao B, Sereno SC. The Glasgow Norms: Ratings of 5,500 words on nine scales. Behav Res Methods. 2019;51:1258–70. 10.3758/s13428-018-1099-3. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 42. Brysbaert M, Warriner AB, Kuperman V. Concreteness ratings for 40 thousand generally known English word lemmas. Behav Res Methods. 2014;46:904–11. 10.3758/s13428-013-0403-5. [ DOI ] [ PubMed ] [ Google Scholar ] 43. McMillen S, Anaya JB, Peña ED, Bedore LM, Barquin E. That’s hard! Item difficulty and word characteristics for bilinguals with and without developmental language disorder. Int J Biling Educ Biling Routledge. 2022;25:1838–56. 10.1080/13670050.2020.1832039. [ Google Scholar ] 44. Milburn EA, Warren T, Dickey MW. Effects of familiarity, context, and abstract representations on idiom processing in aphasia. Front Psychol Conf Abstr Acad Aphasia 53rd Annu Meet. Tucson, United States: Frontiers Media S.A.; 2015. 10.3389/conf.fpsyg.2015.65.00005 45. Jonkers R, Bastiaanse R. Action naming in anomic aphasic speakers: Effects of instrumentality and name relation. Brain Lang. 2007;102:262–72. 10.1016/j.bandl.2007.01.002. [ DOI ] [ PubMed ] [ Google Scholar ] 46. Hess CW, Sefton KM, Landry RG. Sample Size and Type-Token Ratios for Oral Language of Preschool Children. J Speech Lang Hear Res Am Speech-Language-Hearing Association. 1986;29:129–34. 10.1044/jshr.2901.129. [ Google Scholar ] 47. Alario F-X, Ferrand L, Laganaro M, New B, Frauenfelder UH, Segui J. Predictors of picture naming speed. Behav Res Methods Instrum Comput. 2004;36:140–55. 10.3758/BF03195559. [ DOI ] [ PubMed ] [ Google Scholar ] 48. Kuperman V, Stadthagen-Gonzalez H, Brysbaert M. Age-of-acquisition ratings for 30,000 English words. Behav Res Methods. 2012;44:978–90. 10.3758/s13428-012-0210-4. [ DOI ] [ PubMed ] [ Google Scholar ] 49. Hewitt LE, Hammer CS, Yont KM, Tomblin JB. Language sampling for kindergarten children with and without SLI: mean length of utterance, IPSYN, and NDW. J Commun Disord. 2005;38:197–213. 10.1016/j.jcomdis.2004.10.002. [ DOI ] [ PubMed ] [ Google Scholar ] 50. Winters KL, Jasso J, Pustejovsky JE, Byrd CT. Investigating Narrative Performance in Children With Developmental Language Disorder: A Systematic Review and Meta-Analysis. J Speech Lang Hear Res Am Speech-Language-Hearing Association. 2022;65:3908–29. 10.1044/2022_JSLHR-22-00017. [ Google Scholar ] 51. Grela BG, Leonard LB. The use of subject arguments by children with specific language impairment. Clin Linguist Phon Taylor Francis. 1997;11:443–53. 10.3109/02699209708985206. [ Google Scholar ] 52. Ruth de Diego B, Costa A, Sebastián-Galles N, Juncadella M, Caramazza A. Regular and irregular morphology and its relationship with agrammatism: Evidence from two Spanish–Catalan bilinguals. Brain Lang. 2004;91:212–22. 10.1016/j.bandl.2004.02.007. [ DOI ] [ PubMed ] [ Google Scholar ] 53. Roepke E, Brosseau-Lapré F. Speech Error Variability and Phonological Awareness in Preschoolers. Am J Speech Lang Pathol. 2023;32:246–63. 10.1044/2022_AJSLP-22-00031. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 54. Nam H, Goldstein L, Saltzman E. Self-organization of syllable structure: a coupled oscillator model. In: Pellegrino F, Marsico E, Chitoran I, Coupé C, editors. Approaches Phonol Complex [Internet]. Walter de Gruyter; 2009 [cited 2025 Jan 28]. pp. 297–328. 10.1515/9783110223958.297 55. Scheffer A, Keij B, Hakvoort B, Ottow-Henning E, Gerrits E, Wijnen F. Speech sound development of young Dutch children with a developmental language disorder: A complex matter. Int J Lang Commun Disord. 2024;59:2706–22. 10.1111/1460-6984.13109. [ DOI ] [ PubMed ] [ Google Scholar ] 56. Baddeley AD, Thomson N, Buchanan M. Word length and the structure of short-term memory. J Verbal Learn Verbal Behav. 1975;14:575–89. 10.1016/S0022-5371(75)80045-4. [ Google Scholar ] 57. Wibroe M, Cappelen J, Castor C, Clausen N, Grillner P, Gudrunardottir T, et al. Cerebellar mutism syndrome in children with brain tumours of the posterior fossa. BMC Cancer. 2017;17:439. 10.1186/s12885-017-3416-0. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 58. Bishop DVM. ERRNI: Expression, Reception and Recall of Narrative Instrument. London: Harcourt Assessment; 2004. [ Google Scholar ] 59. Wittenburg P, Brugman H, Russel A, Klassmann A, Sloetjes H. Jan. ELAN: a professional framework for multimodality research. 2006 [cited 2025 Jan 30]. pp. 1556–9. https://pure.mpg.de/pubman/faces/ViewItemOverviewPage.jsp?itemId=item_60436 . Accessed 30 2025. 60. Verbeek J, van den Dungen L, Bakker A. Spontane Taal analyse procedure. Verantwoording van het STAP-instrument, ontwikkeld door Margreet van Ierland. Universiteit van Amsterdam; 2008. 61. Abdi H, Williams LJ. Principal Component Analysis. WIREs Comp Stat. 2010;2:433–59. [ Google Scholar ] 62. Fox J. Applied Regression Analysis and Generalized Linear Models. SAGE; 2015. 63. Posit Team. RStudio: integrated development environment for R [Internet], Boston MA. 2025 [cited 2025 Jan 30]. http://www.posit.co/ . Accessed 30 Jan 2025. 64. Connie K, Porcaro. Improving speech intelligibility in adults: clinical application of evidence-based strategies [Internet]. San Diego, CA: Plural Publishing, Inc; 2023 [cited 2025 Jan 30]. http://search.ebscohost.com/login.aspx?direct=true&db=nlebk&AN=3385770&site=ehost-live&scope=site . Accessed 30 Jan 2025. 65. Mei C, Morgan AT. Incidence of mutism, dysarthria and dysphagia associated with childhood posterior fossa tumour. Childs Nerv Syst. 2011;27:1129–36. 10.1007/s00381-011-1433-x. [ DOI ] [ PubMed ] [ Google Scholar ] 66. Ozimek A, Richter S, Hein–Kropp C, Schoch B, Gorißen B, Kaiser O, et al. Cerebellar mutism. J Neurol Steinkopff-Verlag. 2004;251:963–72. 10.1007/s00415-004-0472-6. [ Google Scholar ] 67. Richter S, Schoch B, Ozimek A, Gorissen B, Hein-Kropp C, Kaiser O, et al. Incidence of dysarthria in children with cerebellar tumors: A prospective study. Brain Lang. 2005;92:153–67. 10.1016/j.bandl.2004.06.015. [ DOI ] [ PubMed ] [ Google Scholar ] 68. Marien P, Engelborghs S, Fabbro F, De Deyn PP. The Lateralized Linguistic Cerebellum: A Review and a New Hypothesis. Brain Lang. 2001;79:580–600. 10.1006/brln.2001.2569. [ DOI ] [ PubMed ] [ Google Scholar ] 69. Van Baarsen KM, Grotenhuis JA. The anatomical substrate of cerebellar mutism. Med Hypotheses. 2014;82:774–80. 10.1016/j.mehy.2014.03.023. [ DOI ] [ PubMed ] [ Google Scholar ] 70. Fenson L. MacArthur communicative development inventories: User’s guide and technical manual. Paul H. Brookes; 2002. 71. Caselli MC, Bello A, Rinaldi P, Pasqualetti P. Il primo vocabolario del bambino: gesti, parole e frasi. Forme lunghe e forme brevi del questionario e valori di riferimento per la fascia 8–36 mesi. Milano, Italy: Franco Angeli; 2015. [ Google Scholar ] 72. Eisenberg SL, Fersko TM, Lundgren C. The Use of MLU for Identifying Language Impairment in Preschool Children. Am J Speech Lang Pathol. American Speech-Language-Hearing Association; 2001;10:323–42. 10.1044/1058-0360(2001/028) 73. Ahmed R, Boll-Avetisyan N, Kjær Grønbæk J, Boeg Thomsen D, Mathiasen R, De Aguiar V. Preoperative speech deviations and postoperative impairment in children with poster Fossa Tumors [Poster abstract]. Budapest, Hungary; 2025. 74. Coady JA. Rapid Naming by Children With and Without Specific Language Impairment. J Speech Lang Hear Res. 2013;56:604–17. 10.1044/1092-4388(2012/10-0144). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 75. Miller CA, Leonard LB, Kail RV, Zhang X, Tomblin JB, Francis DJ. Response Time in 14-Year-Olds With Language Impairment. J Speech Lang Hear Res. American Speech-Language-Hearing Association; 2006;49:712–28. 10.1044/1092-4388(2006/052) 76. Elgamal SA, Roy EA, Sharratt MT. Age and Verbal Fluency: The Mediating Effect of Speed of Processing. Can Geriatr J CGJ. 2011;14:66–72. 10.5770/cgj.v14i3.17. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 77. Shao Z, Janse E, Visser K, Meyer AS. What do verbal fluency tasks measure? Predictors of verbal fluency performance in older adults. Front Psychol [Internet] Front. 2014. 10.3389/fpsyg.2014.00772. [cited 2025 Jan 30];5. [ Google Scholar ] 78. Horne BM, Attanayake AA, Aquilina K, Murphy T, Malcolm CP. The neurocognitive profile of post-operative paediatric cerebellar mutism syndrome: A systematic review [Internet]. medRxiv; 2025 [cited 2025 Oct 8]. p. 2025.02.21.25322700. 10.1101/2025.02.21.25322700. 79. Lindgren J. Age and task type effects on comprehension and production of narrative macrostructure: storytelling and retelling by Swedish-speaking children aged 6 and 8. Front Commun [Internet] Front. 2023. 10.3389/fcomm.2023.1252260. [cited 2026 Feb 16];8. [ Google Scholar ] 80. Mills MT. The Effects of Visual Stimuli on the Spoken Narrative Performance of School-Age African American Children. Lang Speech Hear Serv Sch. American Speech-Language-Hearing Association; 2015;46:337–51. 10.1044/2015_LSHSS-14-0070 81. Qiu X. Picture or non-picture? The influence of narrative task types on lower- and higher-proficiency EFL learners’ oral production. Int Rev Appl Linguist Lang Teach De Gruyter Mouton. 2022;60:383–409. 10.1515/iral-2017-0094. [ Google Scholar ] 82. Ellis Weismer S, Venker CE, Evans JL, Moyle MJ. Fast mapping in late-talking toddlers. Appl Psycholinguist. 2013;34:69–89. 10.1017/S0142716411000610. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Data Citations Horne BM, Attanayake AA, Aquilina K, Murphy T, Malcolm CP. The neurocognitive profile of post-operative paediatric cerebellar mutism syndrome: A systematic review [Internet]. medRxiv; 2025 [cited 2025 Oct 8]. p. 2025.02.21.25322700. 10.1101/2025.02.21.25322700. Supplementary Materials Supplementary Material 1 (89.3KB, docx) Data Availability Statement The paper reports a secondary analysis of data from the European CMS study. Requests for access to and reuse of the data should be directed to the principal investigator of the European CMS study, René Mathiasen ([email protected]). Articles from Cerebellum (London, England) are provided here courtesy of Springer ACTIONS View on publisher site PDF (1.9 MB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top