Analyzing Claims and Evidence

A 2024 systematic review found no studies of NLA that included any kind of outcome measure. The authors were willing to examine single-subject designs, qualitative studies, and unpublished studies and those that were not peer reviewed, yet they found no studies meeting their inclusion criteria.

Several sources of evidence have been cited in support of GLP/NLA, however, these sources do not back up the claims that are unique to NLA:

• Articles published by Marge Blanc in the Autism and Asperger's Digest in the years 2005-2008 have also been cited as evidence of NLA; however, these magazine articles are an outline of a single case study that did not undergo peer review. The process for analyzing language samples and providing NLA-based therapy are not described in enough detail to be replicable. 

• Marge Blanc's 2012 self-published book, Natural Language Acquisition on the Autism Spectrum, is frequently described as a longitudinal qualitative study of NLA. A 2024 appendix states that the book was based on research on 85 participants seen in the clinic over the course of 15 years; it's not clear why that was not mentioned in the book itself. No information about these participants is provided, such as their ages, length of time in therapy, reason for stopping therapy, and the nature of the diagnoses and other services. Neither the book nor the addendum describes an informed consent procedure or how participants were chosen for inclusion. A few case studies are described and only two language samples are reported in full with scoring. There is no summary data on how the participants' proportion of utterances from the six stages of NLA changed over time. It's also not clear that qualitative research is the best tool for analyzing longitudinal language sample data using NLA scoring. The outcome measure of NLA scoring is a percentage of utterances at each stage, and tracking changes in percentages over time is a question for quantitative methods, not qualitative. 

• Articles by Barry Prizant, PhD, CCC-SLP: Gestalt language and gestalt processing in autism (1982), Language acquisition and communicative behavior in autism: Toward an understanding of the “whole” of it (1983); Analysis of functions of delayed echolalia in autistic children (1984, with Patrick Rydell). These articles demonstrate the communicative value of echolalia, an idea that is not contested by contemporary researchers. Prizant’s papers  hypothesize a gestalt language processing style that develops in four stages and call for further longitudinal research to verify those claims.

• The book Units of Language Acquisition (Peters, 1983) reports a case study of a single child who was typically-developing. It does not test any intervention or make any mention of language delays or autism. 

• Anecdotes from parents and clinicians are another popular source of evidence for NLA. While I have no reason to doubt the anecdotes are true, we must use caution when applying anecdotes to our own caseloads. When a clinician reports that her therapy got more effective after learning NLA, we must ask what the therapist was doing before learning about NLA, what components of NLA they are using now, and for what clients and what goals they have seen change. Similarly, client anecdotes are missing key pieces of information that we need for decision-making: what were the child's specific strengths and needs, what components of NLA were being addressed, what additional services were they receiving. Anecdotes also don't include the voices of families and clinicians who did not find that NLA benefited from them. 

There are components of NLA that have evidence, but most of them are components that are used in other effective therapies (e.g., child-led, honoring echolalia, emphasis on connection). Many of those other therapies predate NLA. but this evidence is rarely cited in discussions of NLA (child-led, honoring echolalia, emphasis on connection). NLA proponents rarely cite that evidence, and it is important to make clear that evidence for components of NLA is not the same as evidence for NLA as a whole. Similarly, evidence that an alternative approach was not effective is not evidence that NLA is effective. Discussions of NLA research often create a false dichotomy of NLA vs "old-school" ABA; proponents of NLA will argue that since ABA is ineffective (per a DOD report that has some methodological flaws), NLA is the only alternative. This is not true! There are many approaches, such as Hanen, Floortime, and NDBIs, that have evidence behind them and do not involve discrete trial training or reward charts. 

Yes! Neurodiversity-affirming just means that we accept and embrace neurodivergent individuals as equal, valid, and not broken or in need of repair. We identify the individual's strengths and build on them. This is in contrast to a deficit-oriented approach that identifies a person's "deficits" and develops a plan to "remediate" them. Rather than focusing on making a person look neurotypical, a neurodiversity-affirming approach celebrates the variation among humans and works with an individual and their family to meet them where they are and support them in ways that are meaningful to them.

Because different individuals and families have different needs, values, strengths, and preferences, it is crucial to take an individualized approach to therapy planning. Neurodiversity-affirming is as much about the values of the clinician as it is about the specific therapy approach, and many different therapeutic approaches can be implemented in ways that are or are not neurodiversity-affirming. The OT Butterfly has a helpful article on Identifying Neurodiversity Affirming Therapists.

What this means: Rather than respond to questions and criticism with relevant information, many large social media accounts will make comments referring to skepticism of NLA/GLP as "naysaying." Sometimes critics are accused of not valuing echolalia.

My Response: NLA is not just about honoring echolalia, which is something we should all be doing and was already part of treatment best practices before NLA was proposed. The things that make NLA unique are its other claims, such as avoiding modeling of referential verbs before stage 4; not only are those claims often unsupported by evidence, as this website describes, but some of them could potentially undermine children’s language acquisition.

If you're reading this far, I hope you see that there are many criticisms of specific claims about NLA/GLP that have nothing to do with not wanting to honor echolalia, or holding negative beliefs about autistic individuals. I think it is unhelpful to paint critics with such a broad brush, and I wish that proponents would actually address the criticisms of NLA/GLP so we can move forward with productive dialogue. And I think it is important for clinicians and parents to be fully informed about not only the state of the research for and against NLA, but of the alternatives to NLA and the ways NLA overlaps with existing evidence-based therapies. It's important for the integrity of the field of speech-language pathology that we ensure our messaging to the public is accurate, and that includes accurately representing the state of the evidence for a therapy, accurately characterizing its criticisms, and accurately characterizing its alternatives. 

What it means: According to supporters of NLA, individuals can be classified as either "Analytic Language Processors" (ALPs) or "Gestalt Language Processors" (GLPs) based on how they process language. Supposedly, ALPs learn language beginning from single words, which they then combine into two-word utterances, than short phrases, and eventually sentences. This is contrasted with GLPs, who learn language beginning as memorized chunks (delayed echolalia, also called gestalts), which they break down to "mix and match" with other chunks, before breaking those down further into single words. They then combine words, produce short phrases, and progress to producing sentences. Other supporters of NLA say that processing is a spectrum from ALP to GLP, with most people falling in the middle- these may be called "dual processors." A person's processing style is determined from analysis of their language samples, clinical observations, and a language history taken by a clinician knowledgeable of NLA. The child's processing style is then used in developing a therapy plan, and determining which therapy techniques are or are not appropriate for that child. Many claims are also made about GLPs, such as things like "most autistic humans are GLPs," "many GLPs are musical" or "GLPs learn language from media."

Unique to GLP/NLA: Yes. The idea of gestalt and analytic processing styles (whether discrete or a continuum) and the fact that they can be distinguished based on expressive language is a core idea of GLP and is unique to it. 

The evidence: There is a great deal of psycholinguistics research on language processing, including research on autistic children. While language processing researchers don't all agree on theories of language processing, a common feature across studies is that expressive language is not adequate for studying language processing. Language processing research uses methods such as eye tracking, response time measures, EEG, and fMRI to study language comprehension in real time. Research on language processing in children, including those with autism, has not found any evidence for "processing styles," in fact, autistic and neurotypical children exhibit the same patterns of language processing development. Venker & Lorang (2024) wrote a letter to the editor that expands on this issue and cites supporting research.

What it means: A common claim made by NLA proponents is that language development and treatment research does not apply to GLPs, because only ALPs are included in research studies.

Unique to GLP/NLA: Yes; as the ALP-GLP distinction is unique to GLP and NLA. The term ALP is not used in language development research.

The evidence: An increasing number of language development studies are including or focusing on autistic children. Researchers studying language development in Autistic children include Courtney Venker, Letitia Naigles, Evan Kidd, Megan Roberts, and Sudha Arunachalam. Additionally, without a clearly-defined way to label a child as a GLP or ALP, it is not possible to say who is included in the research. Most criticisms that NLA proponents make about language acquisition research are valid criticisms of language acquisition research from the 1970s that are now strawman arguments due to the substantial advances in our knowledge of language acquisition since Watergate.

What it means: Echolalia refers to repetitions of words or phrases. Immediate echolalia occurs immediately after an utterance is produced, such as if a parent says, "Can I pick you up?" and the child says, "can I pick you up?" Delayed echolalia occurs minutes or years after the utterance was produced, such as if the child later initiates a request to be picked up by saying, "Can I pick you up?" Delayed echolalia can serve many functions: it can communicate a want, need, or idea, as in the previous example; it can be used to maintain turn-taking in a conversation; and some individuals repeat phrases because they like the way they sound or feel. 

Unique to GLP/NLA: No. The communicative function of delayed and immediate echolalia has long been acknowledged in autistic (Kanner, Prizant) and neurotypical (e.g., Keenan), populations. More recent studies have used conversation analysis to document the meaningfulness of echolalia in autistic children, such as Sterponi et al., Tarplee & Barrow, and Yu & Sterponi. While many clinicians in various fields may have been taught to ignore or extinguish echolalia in autistic clients, that is not a best practice. 

The evidence: Just like we don't conduct studies to prove that self-generated language means something, we don't need a randomized controlled trial or other experimental set-up to show that echolalia is meaningful. And no single theory or approach has a monopoly on the idea of treating echolalia as meaningful. Conversation analysis is one useful tool for investigating and characterizing the meaningfulness of echolalia. Ultimately, we want to find a balance between using context clues and talking to adults who know a child well to determine what an echoed phrase might mean, while avoiding "reading into" a phrase meaning that isn't intended. This is something we already do when deciphering the early utterances of typically-developing children.

What it means: Social media is filled with phrases such as "connection over compliance" to highlight the idea that therapy should start with a trusting relationship where the child feels safe and comfortable interacting with the therapist. While connection is a necessary foundation to the therapeutic relationship, this can be done as part of therapy programs incorporating a number of different techniques and goals.

Unique to GLP/NLA: The idea that safe and trusting relationships are important for children is neither new or unique to NLA. However, there are clinicians who received poor training in working with autistic children, and were taught to focus on training the child to comply with demands or label pictures as the foundation of speech therapy. In these cases, NLA might be the first time the therapist learns what it looks like to build a trusting relationship in the context of a therapy session.

The evidence: There is data showing the value of responsive therapy techniques, which is the technical term for things we do in therapy that focus on the adult adjusting their behavior to observe and recognize the child's communication. A 2016 study in The Lancet by Pickles et al. found, among other things, that increases in parental responsiveness was the primary predictor of increases in the child's communication skills. Responsive therapy techniques are a key component of a group of techniques known as Naturalistic Developmental and Behavioral Interventions (NDBIs), the only therapy shown to be effective for improving communication outcomes for autistic children in a 2020 meta-analysis (this 2022 meta-analysis goes more in depth). While there is evidence backing up the idea that connecting with a child with autism is an important component of therapy, it's worth noting that this evidence does not come from studies of NLA and is generally not referenced by therapists who use the Natural Language Acquisition Protocol with their autistic clients. Many parents also know that they don't need research to justify building a trusting relationship with their child, and seeking out therapists who will do the same - at the same time, having research showing its therapeutic benefit is still a good thing!

What it means: Many proponents of NLA/GLP say that most autistic individuals are GLPs. Some autistic adults self-identify as GLPs on social media, and may report childhood recollections of understanding and using language in chunks. 

Unique to GLP/NLA: Yes, labeling someone (or self-identifying) as a GLP is unique to GLP/NLA.

The evidence: None that I am aware of. Claims about a large group, such as autistic individuals, cannot be made based on collections of anecdotes. Evidence for such a claim would require a systematic study of a representative sample of autistic individuals to determine how many are GLPs. This, in turn, would require a clear definition of GLP, which has not been offered. Using GLP as an identity marker is a different use of the term than using it as a clinical label that comes with claims about a child's internal cognitive/linguistic processing. If using the definition of GLP as an identity label, we would still need a systematic survey of autistic adults to determine whether it is in fact the case that most autistic adults are GLPs. 

What it means: Proponents of GLP state that GLPs understand language in larger "chunks" such as memorized phrases, sentences, or extended dialogues from movies/TV. They do not understand the individual words and grammatical structures that give conventional meanings to sentences. 

Unique to GLP/NLA: Yes. 

The evidence: There is none. There is an subfield of linguistics called psycholinguistics that is devoted to the study of language processing, which is the study of the cognitive processes involved in doing language, whether it is speaking, signing, listening, reading, or writing. There are entire journals, an annual conference, and a society devoted to the study of language processing. Most studies of language processing in children focus on language processing during listening comprehension. Researchers use tools such as eye tracking, fMRI, and EEG to study spoken language processing. Studies of both autistic and neurotypical children have shown that children engage in incremental processing, meaning that children process the sentence as they hear it, forming predictions as they go and revising them as necessary. For example, if children are shown a picture of a ball and a shoe, and hear the sentence "let's wear the shoe," both autistic and neurotypical children will look at the shoe upon hearing the word wear, before the word shoe is spoken. If autistic children were primarily GLPs who process language in chunks, we would not expect them to respond to words within a sentence, as they might instead treat the sentence as whole. So not only do studies such as those contradict the idea of gestalt language processing, they are also not consistent with processing styles more generally, since we see similar patterns of language processing across neurotypical and autistic children. These processing studies are described in more detail and cited in Venker & Lorang (2024).

Crucially, language processing is not something we can see or hear. Nor is it something we can just intuit by thinking about language, or infer from what a person says. The fact that people can not describe how they process language does not mean that people don't have insight into their language use and language experiences. It just means that the things people have insight into from their own language experiences are generally not processing. Not less valuable than processing, just not processing. 

Note: For some autistic adults, GLP has become a term for celebrating echolalia. In this usage, GLP seems more like an identity label that does not make scientific claims about neuroscience or cognitive science. If that's what GLP means to you or someone you care about, that's great! I'm not here to take that away. However, when GLP is used to make population-level claims about therapy strategies, the term is being used as a clinical label and requires reseach backing up its use.

What it means: This claim, made often by proponents of NLA, is not often precisely spelled out. I see a lot of generic statements like "traditional therapy doesn't work for GLPs," however, it is not clear exactly which strategies are deemed ineffective. Since many SLPs receive poor training in supporting autistic children and adults, it makes sense that some children will receive ineffective therapy; however, these ineffective strategies are likely to be just plain outdated, rather than ALP-specific. 

Unique to GLP/NLA: yes.

The evidence: It's not possible to provide evidence for or against vaguely-specified "ALP strategies" or "traditional speech therapy." These conversations will be much more productive if we can describe the specific strategies that have or have not been effective for a particular client. We also need to be mindful that a strategy(e.g., mirroring and mapping) in the context of an activity that is a poor fit for a child will not be effective, even if the same strategy might be effective when delivered in a different activity. Conversations about what works and what doesn't work require more nuance than is typically possible in sound bites. Additionally, since there is no clearly-specified criteria for an ALP vs a GLP, and there is no data linking processing styles to therapy needs, the onus is on proponents of GLP to provide clear definitions and evidence for this claim.

What it means: Some therapists are trained to ignore delayed echolalia, and may make comments to children who echo like "no echoing" or "no movie talk." Proponents of NLA have done an amazing job pushing back on that outdated view of echolalia to remind us that echolalia has meaning and is a valuable way to communicate. Many autistic adults share that their scripts (aka delayed echolalia or gestalts) are an important part of their identity and preferred methods of communicating. 

Unique to GLP/NLA: No. 

The evidence: First of all, we don't prove that communication is valuable. That's a determination we make based on our values. Delayed echolalia is prevalent in all early communication - whether a child is autistic, has delayed language development, or is neurotypical. Neurotypical adults use scripts when we have "inside jokes" with each other - those references to movies or shared experiences that make sense only to a select group of loved ones. It would be quite odd (not to mention ableist) to insist that delayed echolalia is a valuable form of communication for everyone except autistic early communicators, so let's just go ahead and celebrate communication in all forms, okay? We can do that while also providing developmentally-appropriate supports for children to expand their communicative repertoire to include more flexible use of language. If you want to learn more about celebratory perspectives on autistic echolalia use, these two qualitative papers by Cohn et al. are a good start.

What it means: NLA proceeds through six stages (Blanc et al., 2023):

• Stage 1: Language gestalts (wholes, scripts, songs, and language of episodes)

  • Example: Thewheelsonthebusgoroundandround

• Stage 2: Mitigations (mitigated gestalts and partial scripts); Mix and match combinations of partial scripts

  • Example: The wheels on the truck go round and round

• Stage 3: Isolated single words; Two-word combinations of referential single words

  • Example: truck wheels

• Stage 4: Original phrases and beginning sentences; sentences containing structures from levels 1-3 of Lee's Developmental Sentence Scoring

  • Example: truck drived 

• Stage 5: Original sentences with more complex grammar; sentences containing structures from levels 4-6 of DSS

  • Example: Can I drive the truck?

• Stage 6: Original sentences using a complete grammar system; sentences containing structures from levels 7-8 of DSS

  • Example: If you drive the truck on the table, it could fall off.

Language is assessed in GLPs by taking a language sample and determining which stage each of the child's utterances is in. There are criteria for using those results to determine what stage of NLA the child is currently in (Battye, 2025).

Unique to GLP/NLA: Yes.

The evidence: The first four stages were hypothesized by Barry Prizant in several papers from the early 1980s, based on a few in-depth case studies of autistic children. He acknowledged that longitudinal studies would be needed to provide evidence for gestalt langauge development. Marge Blanc's 2012 self-published book, Natural Language Acquisition on the Autism Spectrum, was the first to use the term "Natural Language Acquisition" and is frequently described as the longitudinal data that was needed to "prove" Dr. Prizant's hypothesis; however, no data is reported in the book. A 2024 appendix to the book states that the book was based on research on 85 participants seen in the clinic over the course of 15 years; it's not clear why that was not mentioned in the book itself. Inadequate information about these participants is provided, and the book does not report their ages, length of time in therapy, reason for stopping therapy, and the nature of their diagnoses and other services. Neither the book nor the addendum describes an informed consent procedure or how participants were chosen for inclusion. A few case studies are described and only two language samples are reported in full with scoring. There is no summary data on how the participants' proportion of utterances from the six stages of NLA changed over time. The non-peer-reviewed book is described as qualitative research, but it is not clear that qualitative research is the best tool for analyzing longitudinal language sample data using NLA scoring. The outcome measure of NLA scoring is a percentage of utterances at each stage, and tracking changes in percentages over time is a question for quantitative methods, not qualitative. 

Additionally, there is a lack of data on whether language samples can be scored reliably using the six-stage system. The same utterance could often be scored as several different stages, and clinicians must make subjective judgment calls to decide how to score an utterances. For example, "open the fridge" could be a stage 1 gestalt if the clinician believes the client memorized it; a stage 2 mitigation if the clinician believes the child used a "mix and match" approach by combining "open the" from a gestalt "open the door," with the word fridge, or it could be a stage 4 original sentence if the clinician believes the child "built" the utterance himself.

What it means: NLA treatment involves modeling utterances for a child that are at the child's current stage of NLA. If a child is at stage 1, the clinician will model potential gestalts for the child. If the child is at stage 4, the clinician will model short, simple sentences.

Unique to GLP/NLA: Yes. 

The evidence: As described above, Blanc (2012, 2023) is often cited as evidence but does not contain enough detail about the treatment approach or the participants to be replicable. Additionally, in order for NLA stages to be a means of selecting goals, we need evidence of scoring reliability. If you score two language samples a week apart, will the child be in the same stage at both language samples? If two clinicians score the same language sample, will they arrive at the same stage for the child? If not, then is NLA scoring the best method for analyzing language samples and determining goals?

What it means: When providing NLA-based therapy, clinicians are instructed to model language based on the child's NLA stage about half the time, and model language "naturally" (i.e., talk normally) half the time. Clinicians use the phrase "model without expectation" to describe this approach: the adult models language without putting pressure on the child to respond. While there are sources that provide detailed guidance about what to model at each stage (e.g., Communication Development Center; Battye, 2025), Blanc et al. (2023) provides the following overview in Table 1:

• Stage 1: Follow the child's lead in play; focus on establishing trust and connection with the child by: 

  • Acknowledge all communication modes and attempts

  • Try to determine what the child might be communicating (nonliteral, contextual meaning)

  • Model new language gestalts that provide the soundtrack for new experiences

• Stage 2:

  • Listen and model options for mitigations that can be used in other contexts (e.g., “Let's play + ball!” or “Let's go + school?”)

  • Provide and seek out experiences as opportunities to model mitigations with multiple communication partners

• Stage 3: Acknowledge single words the child isolates, and model a variety of two-word combinations, without regard for grammar or word order (e.g., noun + attribute, noun + location, noun + noun)

• Stage 4: Support and model short phrases that express semantic relationships (e.g., noun + verb + location; verb + adjective) using the pre-sentence grammar of DST and the sentence grammar of DSS Levels 1 through 3 as guides

• Stage 5: Introduce new grammar targets using DSS Levels 4 through 6 in meaningful conversation and narratives

• Stage 6: Introduce new grammar using DSS Levels 7 and 8 in meaningful conversation and narratives

Unique to GLP/NLA: Sort of. Modeling is an important component of all evidence-based language interventions. Modeling at the child's current NLA stage is unique to NLA.

The evidence: Modeling language that is relevant to what the child is a strategy that is widely used among SLPs, and has been before the term NLA was coined. It is also a core component of evidence-based interventions. This includes the stage 1 recommendations to respond to the child's communication attempts and acknowledge all communication. However, there is no evidence for the specific recommendations to model "soundtracks" for new experiences rather than modeling forms of declarative language describing objects and actions in the environment. At stage 3, modeling two-word combinations without regard for grammar such as "shoe table" is contradicted by theory and evidence showing benefits of only using grammatically correct utterances when speaking to children. Stages 4-6 rely on the DSS, which in 1974 was shown to be valid and reliable for diagnosing language disorders, but the levels have not been validated or researched as a way of "staging" language development and were not intended for goal selection. 

What it means: NLA therapy is based around "modeling without expectation," meaning that adults model language but there are no demands placed on the child to imitate or speak. 

Unique to GLP/NLA: Yes.

The evidence: There is no evidence specific to NLA, and I have not encountered any evidence supporting this point that doesn't mention NLA. Various prompting hierarchies are included in the set of evidence-based techniques called Naturalistic Developmental and Behavioral Interventions, and one study showed that prompting according to those hierarchies actually increased spontaneous language in autistic children. Of course, there are ways to prompt badly, that lead to the child depending on prompts; such "prompt dependence" should not happen if the clinician is appropriately adjusting the prompts in response to the child's performance. Adults using prompting need to carefully consider what, when, and how they prompt; further discussion is beyond the scope of this website. NLA proponents often describe outdated practices, such as forcing a child to sit at a table and giving the child a small piece of food every time they imitate the label for a flashcard. However, the fact that such compliance-based methods are not recommended does not mean that all prompts are equally bad. This handout on Enhanced Milieu Teaching describes different kinds of prompts and when to use them. 

What it means: Proponents of NLA often say that teaching single words (as opposed to phrases) to a child in stage 1 or 2 of Natural Language Acquisition will cause these words to be acquired as "stuck single word gestalts," and will prevent the child from being able to use that word later on in self-generated utterances.

Unique to GLP/NLA: Yes. This claim relies on identifying stages of NLA, and using those stages to determine treatment recommendations, which are unique features of NLA.

The evidence: Modeling single words for emergent communicators is a key feature of Enhanced Milieu Teaching, which is backed by research supporting its effectiveness in supporting communication development in autistic children. EMT is not the only evidence-based technique for supporting autistic communicators, nor is it the only technique that involves modeling single words. A full review of the literature on single word modeling is well beyond the scope of this website. I am also not aware of any research that is aligned with a claim that modeling single words in naturalistic contexts harms language development.

What it means: In NLA, 50% of utterances to the child should be presented at their stage of language development. In stages 1 and 2, nouns and verbs should only modeled as part of potential gestalts or mitigated gestalts, such as utterances beginning with "Let's" or "It's time to" (e.g., "Let's swing," "Let's play"). Clinicians may also model phrases for experiences like "I like it!" "This is fun!" or "It's too loud." However, utterances such as "The car is driving," where nouns and verbs are used to describe the environment, are not recommended; nor is labeling of single sounds or verbs. In stage 3, the clinician models two-word utterances consisting of nouns, adjectives, and prepositions, with no verbs. 

Unique to GLP/NLA: Yes. Since the stages of NLA are unique to NLA, therapy recommendations based on the NLA stages are also unique to NLA. 

The evidence: Modeling nouns and verbs in meaningful contexts for emergent communicators is a key feature of all evidence-backed therapies. For example, Enhanced Milieu Teaching (EMT) has research supporting its effectiveness in supporting communication development in autistic children. Toy Talk and the Sentence-Focused Framework also emphasize the importance of nouns and verbs in early communication development. There is also research from language processing that autistic children understand more referential language than their caregivers are aware of, so I worry that depriving them of that referential language in the input is depriving them of learning opportunities.

EMT and Toy Talk are not the only evidence-based technique for supporting autistic communicators, nor are they the only techniques that involves modeling referential nouns and verbs. A full review of the literature on the acquisition of nouns and verbs is well beyond the scope of this website. However, I am not aware of any research supporting a decision to hold off on modeling referential nouns and verbs. Ultimately it is the responsibility of someone making a claim in favor of NLA to provide evidence in support of it.

I'm glad you found something that workd for the kids on your caseload! I'm not here to tell you what to do in therapy. Many elements of NLA overlap with existing evidence-based practices, such as: following a child's lead, building trust, modeling meaningful language during meaningful interactions, and acknowledging the value of echolalia. So if those are the parts of NLA you are using, I'm not sure that's NLA. You may have learned those things in an NLA training, but those characteristics of therapy predate NLA and are not unique to it.

Now maybe you're incorporating features of NLA that are unique to it, such as identifying a child's NLA stage, modeling 2-word phrases at stage 3 without regards for grammar, modeling potential gestalts at stage 1, or setting goals for stages 4-6 based on the DSS. If you are mixing these NLA strategies with other approaches, I'm not sure that's NLA either. The NLA protocol is an approach to therapy that is more than just reframing how you think about echolalia and autistic communication. It also includes a prohibition on certain strategies, such as recasting, expansions, and prompting. 

I'm not here to tell you what counts as NLA or not - and until there is a clear definition or fidelity checklist, I don't know that it's possible to determine what "counts" as NLA. However, an Instagram post by Meaningful Speech states, "the difficulty here is that many people think they can blend the traditional strategies they have learned over the years with what they are learning about GLP. And this doesn't work, especially at the early stages." I think what's most important for having effective conversations is to be really clear about what we are talking about. Instead of saying "I use NLA" or "I use parts of NLA," we need to describe what goals, strategies, and activities we are implementing.