Is there documentation of research and clinical evidence to support claims about Natural Language Acquisition?
Marge Blanc’s 2012 book, Natural Language Acquisition on the Autism Spectrum, is widely referred to as a description of the research supporting NLA. Marge Blanc makes frequent reference to “15 years of clinical research” backing NLA, and this claim is often repeated in promotion of NLA and GLP.
Although Marge Blanc has often referred to her book as a report of her research, she has also rejected that characterization, writing in a comment on my Instagram account: “for anyone who has not read the book, let me first comment, as the author, that the book was never intended to be a description of the research.”
Clinicians and parents making choices about treatment based on a belief that Blanc has conducted research deserve access to evidence that the research actually occurred, and the opportunity to see the results of that research. Clinicians promoting NLA have a duty to ensure that their statements about its evidence are true.
In this blog post, I will analyze two sources that have been described by some as research in support of NLA. To examine the methodological quality of these sources, I will use the relevant research appraisal tools from JBI, which can be downloaded for free (no login required!) at jbi.global. While these tools are not intended for making a binary decision about whether something constitutes a report of research, they can be a useful starting point for thinking through such a decision.
The purpose of this post is to answer the question: Are the documents that get described as clinical research in support of NLA, actually reports of research? Regardless of where an individual stands on NLA and GLP, I think we can all agree that we should only call something research if it is, in fact, research.
Blanc (2005-2006): Finding the Words…
In a series of magazine articles published in Autism & Asperger’s Digest (now called Autism Digest), Blanc introduced gestalt language processing and natural language acquisition through a series of case vignettes. Blanc tells vignettes of five children: Cam, Dylan/Daniel (the same child is referred to with different names), Bevin, Nicky, and Will.
Although Blanc has recently described these articles as a case study, that descriptor is not contained in the articles themselves. We can use the JBI Critical Appraisal Checklist for Case Reports to assess the methodological quality of these articles as research. JBI offers a list of 8 questions for assessing case reports, and each question can be answered “yes,” “no,” or “unclear.” I have retyped the questions verbatim in italics, with my answers in boldface, followed by my rationale.
Were patient’s demographic characteristics clearly described? No. There are references to the ages of Daniel/Dylan and Will. I did not find additional demographic information, such as ages of other children, race, sex, socioeconomic status, medical and developmental history, diagnoses, previous and concurrent treatments, and previous assessments.
Was the patient’s history clearly described and presented as a timeline? No.
Was the current clinical condition of the patient on presentation clearly described? No. There are brief descriptions of children’s communication skills when Blanc first met them, such as the following description of Daniel: “When we first met him, he was nearing four years old, and he could recite numerous lines from favorite videos. His mother knew all the lines, and could say the preceding and following lines, a strategy she used to create verbal interactions with her son, and to ‘keep the conversation going.’” A publishable case study would have more detail about the ways Daniel used his echolalia and his other communication skills.
Were diagnostic tests or assessment methods and the results clearly described? No. Blanc defines the stages of NLA and makes reference to the children in her case studies moving through the stages. In a case study using the stages of NLA as an outcome measure, I would expect a table for each child, showing the percent of utterances at each stage at each time point in the study so that the reader could see the progression. There is no report of numerical data for any child.
Was the intervention(s) or treatment procedure(s) clearly described? No. Blanc’s vignettes include snippets of her treatment approach, but the treatment is not described in enough detail that someone could replicate the treatment. For example, Blanc wrote about Daniel: “We helped [his mom] learn to build useful, flexible ‘gestalts’ into their daily lives. Recognizing that all language that surrounds a child constitutes inadvertent ‘models,’ we wanted to make sure that Daniel heard not only Walt Disney’s language, but plenty of other language that would become useful in everyday communication. ‘Let’s get out of here!’ works much of the time, but ‘Come on!’ is more generally communicative.”
Was the post-intervention clinical condition clearly described? No. Brief descriptions of children’s progress lacked detail. Since the children’s progress was reportedly tracked by the stages of utterances in their language samples, this would have been a straightforward post-intervention measure to include, but it was not reported.
Were adverse events (harms) or unanticipated events identified and described? No. There is also no mention of procedures for determining parental and client permission to publish these case studies.
Does the case report provide takeaway lessons? Yes. Among other takeaways, Blanc provides suggestions for acknowledging and building on echolalia, providing meaningful language models, and building on a child’s interests. To the extent that these recommendations overlap with existing best practices, that overlap is not named or cited.
Total “yes” scores: 1/8
Source for checklist items: Moola S, Munn Z, Tufanaru C, Aromataris E, Sears K, Sfetcu R, Currie M, Qureshi R, Mattis P, Lisy K, Mu P-F. Chapter 7: Systematic reviews of etiology and risk. In: Aromataris E, Munn Z (Editors). JBI Manual for Evidence Synthesis. JBI, 2020. Available from https://synthesismanual.jbi.global
Blanc (2012): Natural Language Acquisition on the Autism Spectrum
Blanc published this book in 2012 and released an addendum to the book on her website in 2024. The book and the addendum will be analyzed together here. The book does not describe itself as a report of research, which makes me wonder if it was intended as research at the time of publication, or if it has been retroactively labeled as such in response to criticism of the evidence base for NLA. The 2024 appendix states, “the NLA book was published in 2012, the result of 15 years of longitudinal clinical research in our clinic,” and states that “Everything about NLA is from longitudinal, qualitative research!”
Because Blanc describes her research as qualitative, I first analyze it using the JBI Critical Appraisal Checklist for Qualitative Research. JBI offers a list of 10 questions for assessing qualitative research and each question can be answered “yes,” “no,” or “unclear.” I have retyped the questions verbatim in italics, followed by my answer in boldface, followed by my rationale.
Is there congruity between the philosophical or theoretical positioning of the study and the methodological approach that has been adopted? Unclear. The book does not report a philosophical/theoretical positioning, nor does it report a research methodology. Elements of a description of methodology that are missing include: frequency of language sampling, procedures for checking accuracy of language sample transcription and scoring, procedures for storing the scored language samples, and replicable description of treatment procedures.
Is there congruity between the research methodology and the research question or objectives? Unclear. The book contains no report of a research question, research objectives, or research methodology.
Is there congruity between the research methodology and the methods used to collect the data? Unclear. The book reports that language samples were taken, includes snippets of the language samples, describes some language sampling contexts, and describes general best practice strategies for obtaining language samples. However, the book does not report details of data collection such as how often language samples were taken, how it was determined when to take a language sample, procedures for transcribing language samples and checking transcription accuracy, procedures for scoring language samples and checking scoring accuracy, or data storage procedures.
Is there congruity between the research methodology and the representation and analysis of data? Unclear. There is no representation or analysis of data reported in the book. The 2024 appendix states that “All 85 progressed from at least one stage to the next while they were in the clinic (some moved or left the clinic when they were seen at school)” and “The vast majority (more than 80%) progressed from Stage 1 or 2 to at least a solid Stage 4.” A research paper should describe the steps of data collection, storage, organization, and analysis that led to these numerical claims. Importantly, claims involving percentages are quantitative claims, not qualitative ones.
Is there congruity between the research methodology and the interpretation of results? Unclear, because the book contains no report of research methodology or results.
Is there a statement locating the research culturally or theoretically? No.
Is the influence of the researcher on the research, and vice-versa, addressed? No.
Are participants, and where appropriate, their voices, adequately represented? Unclear. The book contains case vignettes of nine of Blanc’s clients. The appendix states there were 85 total clients represented in the research. We don’t know anything about the other 76 clients. The book makes statements about the children’s internal thoughts, feelings, and cognitive processes, which may or may not truly represent the children’s internal states. On pages 259-260 of the book, Blanc describes a client, Will, who used the Rapid Prompting Method, which raises authorship concerns about messages attributed to Will.
Is the research ethical according to current criteria, and for recent studies, is there evidence of ethical approval by an appropriate body? No. The appendix states that Blanc began the research while working at a university, yet there is no description of IRB involvement. The book and appendix contain no descriptions of an informed consent process. This is particularly important since Blanc was conducting research on the clients at her own clinic.
Do the conclusions drawn in the research report flow from the analysis, or interpretation, of the data? No. There is not adequate data reported in the book to support conclusions about stages of language development or therapy effectiveness.
Total “yes” scores: 0/10
Source for checklist items: Porritt, K., Evans, C. Bennett, C. Loveday, H., Bjerrum, M. Salmond, S., Munn, Z., Pollock, D., Pang, D., Vineetha, K., Seah Betsy. Lockwood, C. Systematic reviews of qualitative evidence (2024). Aromataris E, Lockwood C, Porritt K, Pilla B, Jordan Z, editors. JBI Manual for Evidence Synthesis. JBI; 2024.
Although Blanc has said her research is qualitative, her description of a longitudinal study involving language sampling at multiple time points to show children moving through stages of development sounds more like a quantitative study. Her outcome measure is determined numerically: the percent of utterances a child produces at each stage of NLA. A study looking at changes in those percentages over time might be better described as an observational longitudinal study, similar to a cohort study. Therefore, I will also analyze the book using the JBI Critical Appraisal Checklist for Cohort Studies. JBI offers a list of 11 questions for assessing cohort studies and each question can be answered “yes,” “no,” “unclear,” or “Not Applicable.” I have retyped the questions verbatim in italics, with my answers in boldface, followed by my rationale.
Were the two groups similar and recruited from the same population? No. There is no comparison group, which can be a defensible decision for certain research questions. In the absence of a stated research question, I will give Blanc the benefit of the doubt and proceed as if the lack of comparison group is justified. However, per JBI guidelines, a “yes” answer to this question would still require the following information to be available, which is absent from the book: inclusion/exclusion criteria for participants (which must be developed prior to the beginning of the study), and a description of characteristics of participants, which I interpret to include: age, race, SES, diagnoses, prior and concurrent treatment, language background, and medical and developmental history. Even allowing for no comparison group, there is inadequate information reported about study participants.
Were the exposures measured similarly to assign people to both the exposed and unexposed groups? No. While I do not penalize the paper for not having a comparison group, a “yes” for this item requires that the exposure measures are clearly defined and described in detail. While exposure feels like a weird word here, in this context all it means is the diagnosis or “the thing that participants experienced.” Depending on the research question, the “exposure” in this study would either be being identified as a GLP, or as the language therapy provided at Blanc’s clinic. GLP is not clearly defined, and the therapy approach is not described in a level of detail that would allow a reader to replicate the therapy. We don’t have clear criteria we could use to observe a therapy session and determine if it “counts” as an NLA-based therapy session.
Was the exposure measured in a valid and reliable way? No. No method for identifying GLPs or tracking information about the therapy provided is reported. There is no mention of inter-rater reliability calculations to measure the accuracy of identifying kids as GLPs, transcribing language samples, or scoring language samples. There is no measure of a fidelity procedure to ensure that all therapy provided actually followed NLA principles, such as modeling language at the child’s NLA stage and refraining from prompting. There is no connection to the literature on language processing to justify categorizing children by language processing styles.
Were confounding factors identified? No. We do not have any information about factors other than the CDC-provided therapy that may have influenced children’s language development, such as other therapies received before or during the study, child age, additional diagnoses, home and school environment, parent involvement for carryover of therapy, and age and language skills at the start of the study.
Were strategies to deal with confounding factors stated? No.
Were the groups/participants free of the outcome at the start of the study (or at the moment of exposure)? Not applicable. While an outcome measure is not explicitly stated in the book, the 2024 appendix suggests that the child’s NLA stage was the outcome measure. It would not make sense for participants to be “free from language development” at the start of the study, so I chose to mark this item N/A.
Were the outcomes measured in a valid and reliable way? No. One might use the NLA stage of a child’s language samples as an outcome measure, and track this change over time. Using the NLA stages at an outcome measure would require the authors to establish the reliability of NLA scoring, such as by assessing inter-rater scoring reliability. Then they would actually need to measure each child’s NLA stage at each sampling point, and report their procedure for doing so, and report the results. Validity could be established in a number of ways, including using data to show that kids do progress from one stage to another or showing with data that NLA stages capture changes in development better than other language sample analysis methods.
Was the follow-up time reported and sufficient to be long enough for outcomes to occur? No. There is no report of therapy dose, frequency, or duration. At a minimum, it is expected to report the mean, range, and standard deviation of the length of time participants received therapy. An example statement could be something like, “Participants attended weekly hour-long sessions for between 6 months and 6 years (mean = 1.4 years, sd = 0.7 years). During their time in the study, participants attended between 12 and 278 hours of therapy (mean = 80 hours, sd = 25 hours).” Even better would be to provide a table with a row for each participant that shows how many sessions they attended and the duration of time between their first and last session. We would also want to know if the researchers followed up with participants after their last therapy session to assess whether their progress continued or was at least maintained.
Was follow-up complete, and if not, were the reasons for loss to follow-up described and explored? No. There is no report of why participants stopped therapy beyond this parenthetical explanation in the appendix: “some moved or left the clinic when they were seen at school.” At a minimum, it would be expected to list the reasons for ending therapy along with the number of participants who stopped therapy for that reason.
Were strategies to address incomplete follow-up utilized? No.
Was appropriate statistical analysis used? No. The book contains no report of statistical analysis procedures. There are two statements of percentages in the appendix: “All 85 progressed from at least one stage to the next while they were in the clinic” and “The vast majority (more than 80%) progressed from Stage 1 or 2 to at least a solid Stage 4.” In order to calculate these numbers, one would need, at a minimum, some kind of table or list that shows each participant’s starting and ending stage. No such table is reported or described.
Total “yes” scores: 0/10
Source for checklist items: Barker TH, Hasanoff S, Aromataris E, Stone JC, Leonardi-Bee J, Sears K, et al. The revised JBI critical appraisal tool for the assessment of risk of bias for cohort studies. JBI Evid Synth. 2025;23(3). Supplemental Digital Content 1. https://doi.org/10.11124/JBIES-24-00103
Conclusion
Marge Blanc’s 2005-2006 magazine articles in the “Finding the Words…” series do not report the kind of information that is necessary in a research study, even a case study. The series contained takeaway messages for parents of children who use delayed echolalia, which was the only of the eight JBI criteria for a case study that the articles met.
Blanc’s oft-cited 2012 book did not receive any “yes” responses on the JBI checklists for a qualitative study or a longitudinal (cohort) study. If clinical research was completed that led to the publication of that book, the appraisal tools above provide an outline for the kind of information that should be shared.
While the JBI checklists do not state a minimum number of “yes” responses that need to be present in a high-quality study, a complete absence of “yes” responses suggests to me that we do not have anywhere near enough information to appraise any research that may have taken place at Blanc’s clinic.
The fact that Blanc’s writings do not report the elements listed on the JBI checklists does not mean that research didn’t happen or that those documents have no clinical value. It simply means that the documents I analyzed do not include the kind of information that would be necessary in a report of research. I do not think it is accurate to call these documents reports of research when they lack the basic elements required to assess a study, whether qualitative or quantitative: descriptions of participants, methods, and results.
I am aware that many parents report benefiting from NLA and from their interactions with Marge Blanc. I know that many families have felt let down by outdated and ineffective therapies and interactions with therapists who do not have the necessary skills to support their child, and I am always glad to hear when parents and children get they help they need or want. I want to honor those lived experiences.
In honoring that lived experience, it is important to remain accurate about the types of evidence that are available for NLA. We should not be calling something research if it is not research. Misrepresenting the evidence for an approach can erode trust in our field and undermine the research that is being done. Many proponents of NLA have privately shared with me their concerns about the evidence for NLA being overblown. I have had conversations with clinicians who believed NLA was backed by evidence, and later changed their minds when they read the research themselves and realized it was either not research or did not support claims specific to NLA. I also know clinicians who use NLA and are transparent about the lack of research for it.
It is not my place to speculate about why a given SLP might say that Blanc (2005-2006) and Blanc (2012) are reports of research. However, I believe this is an inaccurate statement, regardless of intent.
It matters that what we say is true.
Don’t miss my next blog post!
Sign up to receive emails about future blog posts and speaking engagements.
More about the JBI
The JBI is a research institute at the University of Adelaide in Australia. It was established in 1996 with the goal of promoting evidence-based decision-making in healthcare.
I chose the JBI tools for this blog post because they are available for free online, and the JBI has created tools that are specific to different types of evidence. While appraisal tools created by different authors may include slightly different items, I do not think that using a different tool would lead to substantially different ratings for the documents discussed in this post.
The JBI appraisal tools were written to aid authors of systematic reviews and meta-analyses in assessing the methodological quality of studies. The tools can also be used for educational purposes, such as teaching students what to look for in a study. In this post, I used the tools to highlight what is missing from certain publications cited as “research.” While JBI does not offer specific guidance on how many “yes” scores constitute acceptable study quality, I think many researchers would agree that a study that only scored one or no “yes” responses is not a high-quality study, and might not even be considered research.
I have no affiliation with JBI, and encourage you to go to their website if you want to learn more about it: jbi.global
