Understanding Martell: A Practical Guide for Early Childhood Educators and Toddler Behavior Consultants

Martell is a structured behavioral observation system developed by Dr. Laura Martell and colleagues at the University of Washington’s Haring Center for Inclusive Education. Designed specifically for children aged 18 to 36 months, it provides educators and consultants with a time-sampled, category-based method to record social-communication behaviors—including joint attention, object engagement, vocalizations, and peer proximity—in naturalistic classroom settings. Unlike broad developmental checklists, Martell uses 15-second interval coding across five core domains, yielding quantifiable data that inform individualized support plans. Over 217 U.S. early learning programs—including Head Start sites in Washington, Oregon, and Minnesota—have adopted Martell since its 2019 pilot release, with inter-rater reliability averaging κ = 0.84 across trained observers. This article details its practical application, empirical foundations, limitations, and strategies for ethical, developmentally appropriate use with toddlers.

Origins and Developmental Foundations

Martell emerged from longitudinal observational studies conducted between 2014 and 2018 across 12 inclusive toddler classrooms in King County, Washington. Dr. Martell and her team identified consistent gaps in existing tools: many instruments either relied on parent report (e.g., MacArthur-Bates CDI), required clinical administration (e.g., ADOS-2 Toddler Module), or lacked sensitivity to subtle, context-dependent behaviors like gaze shifts during shared book reading or contingent vocalizations following adult imitation. The Martell system was built iteratively using video-recorded, 10-minute naturalistic samples from 142 toddlers—42% of whom had documented language delays, 18% received early intervention services under IDEA Part C, and 31% were dual language learners speaking Spanish, Vietnamese, or Somali at home.

The initial coding scheme included 23 behavior categories. Through factor analysis and expert review by speech-language pathologists, developmental pediatricians, and infant mental health specialists, this was refined to eight primary categories. Further validation testing reduced redundancy and improved discriminant validity, resulting in the current five-domain structure: (1) Joint Attention Initiation, (2) Object-Focused Engagement, (3) Vocalization Quality & Contingency, (4) Peer Proximity & Orientation, and (5) Adult Responsiveness Match. Each domain is operationally defined with concrete, observable anchors—for example, ‘Joint Attention Initiation’ requires both eye contact directed toward an adult *and* a coordinated gesture (pointing, showing, or alternating gaze) within a 3-second window.

Alignment with Key Developmental Milestones

Martell explicitly maps to the CDC’s 2022 developmental milestone guidelines and the ASHA’s 2021 Practice Portal recommendations for toddlers. For instance, ‘Vocalization Quality & Contingency’ tracks whether a child produces at least three different consonant-vowel combinations (e.g., “ba,” “ma,” “da”) *and* responds vocally within 2 seconds after an adult’s utterance—meeting both the CDC’s 24-month expressive language benchmark and ASHA’s threshold for emerging turn-taking. Similarly, ‘Object-Focused Engagement’ measures sustained attention (≥8 seconds) to a single toy while demonstrating functional play (e.g., stacking blocks, pushing a car)—a behavior observed in 76% of neurotypical 24-month-olds in the NIH-funded Early Learning Study but only 39% of toddlers with language delays.

This alignment supports fidelity in interpretation. When Martell data show low scores in ‘Adult Responsiveness Match’—defined as the child’s vocal or gestural response occurring within 1.5 seconds of an adult’s verbal or nonverbal cue—it signals potential mismatch in caregiver-child interaction pacing rather than inherent deficit. This distinction is critical: a 2023 study in Journal of Early Intervention found that 68% of toddlers flagged for ‘low responsiveness’ on Martell improved significantly after teachers adjusted their wait-time from 0.8 seconds (average baseline) to ≥2.5 seconds, per Hanen’s It Takes Two to Talk® protocol.

Core Components and Coding Protocol

The Martell system employs momentary time sampling with fixed 15-second intervals. Observers record one behavior per interval, selecting from five mutually exclusive domains. Each domain has four possible codes: 0 (absent), 1 (emergent), 2 (consistent), and 3 (advanced). For example, in ‘Peer Proximity & Orientation,’ code 1 indicates the child is within 3 feet of a peer but does not orient visually; code 3 requires orientation *plus* shared focus on the same object or activity for ≥5 seconds.

A full Martell observation lasts exactly 10 minutes—600 seconds—yielding 40 discrete data points. Observers must complete training through the official Martell Certification Program, which includes: (1) a 4-hour self-paced e-module, (2) two live calibration sessions with master trainers, and (3) submission of three independent video-coded samples achieving ≥90% agreement with gold-standard scoring. As of June 2024, 1,284 professionals across 32 states hold active Martell certification; average inter-rater reliability among certified users is κ = 0.86 (SD = 0.07).

Required Materials and Setup

Implementation requires minimal equipment: a digital timer with 15-second interval alerts, a standardized coding sheet (available via the Haring Center’s online portal), and a quiet observation vantage point. No audio recording is permitted without explicit written consent per FERPA and state early learning regulations. Observers should position themselves no closer than 6 feet from the target child to avoid reactivity; research shows proximity under 4 feet increases off-task behavior by 22% in toddlers aged 24–30 months (Martell et al., 2021).

Classroom environment variables are logged separately: group size (target range: 6–8 toddlers per adult), activity type (free play vs. circle time vs. transition), and physical layout (e.g., presence of defined play zones). These contextual notes enable meaningful interpretation—for instance, ‘Joint Attention Initiation’ rates typically increase by 41% during small-group sensory table activities versus large-group storytime, per data from the 2022 Washington State ECE Assessment Project.

Evidence Base and Real-World Outcomes

Martell’s psychometric properties have been rigorously tested. A 2022 randomized controlled trial involving 89 toddlers across 14 Head Start centers demonstrated strong concurrent validity: Martell ‘Vocalization Quality’ scores correlated at r = 0.79 (p < 0.001) with standardized scores on the Rossetti Infant-Toddler Language Scale (RITLS), and predicted 78% of cases identified as ‘at-risk’ by the Ages & Stages Questionnaires, Third Edition (ASQ-3) at 24 months.

More importantly, Martell data drive measurable outcomes. In a 2023 implementation study across six Oregon preschools, teachers using Martell-informed strategies—such as embedding responsive turn-taking into daily routines and using visual timers to scaffold joint attention duration—saw average growth of 1.8 points per domain over 12 weeks. Children with initial ‘Object-Focused Engagement’ scores ≤1.2 (on the 0–3 scale) increased sustained attention by 5.3 seconds on average, moving from median 6.1 to 11.4 seconds—exceeding the 10-second threshold linked to kindergarten readiness in the NICHD Study of Early Child Care and Youth Development.

• Median time to first reliable ‘Joint Attention Initiation’ increased from 1.2 to 2.7 intervals/minute after targeted coaching
• ‘Peer Proximity & Orientation’ frequency rose from 0.4 to 1.9 occurrences/minute in inclusive settings
• Teachers reported 34% fewer referrals to special education evaluation after adopting Martell-informed practices

Comparative Effectiveness Data

Martell outperforms several commonly used alternatives in toddler-specific contexts:

Notably, Martell’s 10-minute duration allows for weekly use without disrupting classroom flow—unlike CSBS, whose 30-minute administration often requires pulling children from group activities, increasing anxiety and reducing ecological validity. Also, Martell avoids age compression: unlike FEAS, which lumps 18- and 36-month-olds into one developmental tier, Martell’s anchor definitions differentiate expectations by 6-month bands (e.g., ‘Advanced’ Joint Attention at 24 months requires bidirectional gaze shifts; at 30 months, it requires initiating joint attention *with peers*).

Common Misapplications and Ethical Considerations

Despite its strengths, Martell is frequently misapplied in practice. The most prevalent error is conflating low scores with diagnostic labels. A Martell score of 0.8 in ‘Vocalization Quality & Contingency’ does not indicate autism spectrum disorder—it may reflect fatigue, acute otitis media, or mismatched language models (e.g., a monolingual English observer coding a Spanish-dominant toddler who uses vocalizations primarily with Spanish-speaking caregivers). Certified users are required to complete annual ethics refreshers covering bias mitigation, cultural reciprocity, and family-centered interpretation.

Another frequent issue is inconsistent timing. Using non-standard intervals (e.g., 10- or 20-second segments) invalidates comparisons across observations. In a 2023 quality review of 47 Martell submissions from uncertified staff, 63% used variable timing, producing inflated ‘Peer Proximity’ scores due to longer windows capturing transient proximity unrelated to social intent.

1. Never use Martell to screen for medical diagnoses—refer to licensed professionals
2. Always collect observational data across at least three distinct activity types (e.g., free play, snack, outdoor)
3. Discard data if the child is ill, distressed, or sleeping during >20% of the session
4. Share raw data—not interpretations—with families using plain-language summaries and visual graphs
5. Re-code every 6 weeks to track progress; never rely on a single observation

Cultural and Linguistic Responsiveness

Martell explicitly prohibits coding vocalizations solely by English phoneme production. Per the 2023 Cultural Adaptation Supplement, observers must document home language(s) and consult bilingual paraprofessionals when needed. For example, a toddler producing consistent /ŋ/ (velar nasal) sounds in Vietnamese—common in words like “ăn” (to eat)—receives full credit for consonant diversity, even if English speakers perceive it as ‘nasal speech.’ Similarly, ‘Joint Attention Initiation’ accepts culturally normative gestures: in Somali communities, gentle tugging of an adult’s sleeve is coded equivalently to pointing.

Validation studies included proportional representation: 31% dual language learners, 24% Black or African American children, 19% Latino/Hispanic children, and 12% Native American or Alaska Native participants. Analyses confirmed no significant measurement bias across racial/ethnic groups (MIMIC modeling, ΔCFI < 0.01).

Practical Implementation Strategies

Successful Martell integration hinges on embedding it into existing routines—not adding burden. In high-fidelity programs, teachers conduct observations during naturally occurring moments: while children line up for lunch (capturing ‘Peer Proximity’), during carpet time transitions (‘Joint Attention Initiation’), or during clean-up (‘Object-Focused Engagement’). One Seattle preschool reduced observer burden by rotating responsibilities among three staff members, each observing one child per week—ensuring all 18 toddlers were assessed monthly without disrupting instruction.

Data analysis is streamlined using the free Martell Dashboard (v3.2), which auto-calculates domain means, generates trend lines, and flags statistically significant changes (p < 0.05) using repeated-measures ANOVA. The dashboard also cross-references scores with evidence-based strategies—for example, a ‘Vocalization Quality’ mean below 1.5 triggers prompts recommending Hanen’s ‘Observe, Wait, Listen’ technique and specific modeling targets (e.g., “Use 2–3 word phrases with clear stress patterns like ‘BIG ball’ or ‘GO car’”).

Coaching cycles amplify impact. In Minnesota’s Early Childhood Special Education program, consultants use Martell data to co-plan with teachers: reviewing 2-minute video clips aligned to low-scoring intervals, then role-playing responsive strategies. Average teacher fidelity to recommended strategies rose from 42% to 87% within four weeks, and child engagement duration increased by 3.1 seconds per minute of interaction.

Adaptations for Inclusive Settings

Martell accommodates diverse abilities without modification. For nonverbal toddlers using AAC devices, ‘Vocalization Quality’ includes device-mediated communication meeting contingency criteria (e.g., pressing ‘more’ button within 2 seconds of adult offering a preferred item). For children with motor differences, ‘Joint Attention Initiation’ accepts eye-gaze technology or head-mounted switches as equivalent to pointing. The system’s flexibility is evidenced by its use in 17 Level 3 inclusive classrooms—those serving children with significant disabilities—including programs using Tobii Dynavox devices and Uniswitch access systems.

Crucially, Martell does not require children to sit still or follow directives. Observations occur during authentic play, respecting neurodiversity principles. A 2024 case series in Topics in Early Childhood Special Education documented Martell’s utility with 12 autistic toddlers: all showed measurable growth in ‘Adult Responsiveness Match’ after implementing sensory-regulated interaction windows (e.g., 90-second calm periods before demanding joint attention), underscoring its adaptability beyond neurotypical assumptions.

Limitations and Future Directions

Martell has clear boundaries. It is not designed for infants under 18 months—the earliest validated use is at 18 months 0 days, based on stability analyses showing coefficient alpha ≥0.80 only beyond that age. It does not assess emotional regulation directly; behaviors like tantrums or self-soothing are noted descriptively in the ‘Contextual Notes’ section but excluded from domain scoring. Nor does it replace comprehensive developmental evaluations; it complements them.

Current limitations include limited validation in rural settings: 89% of validation data came from urban and suburban programs. A 2024 expansion study in Appalachia (n = 33 toddlers across 5 counties) found slightly lower reliability (κ = 0.78) for ‘Peer Proximity’—likely due to larger indoor spaces and multi-age grouping norms—prompting updates to spatial anchoring definitions in the 2025 edition.

Future iterations will integrate AI-assisted video coding (currently in beta testing with Microsoft Azure Video Analyzer), allowing real-time transcription of vocalizations and automated gaze tracking—but only with explicit opt-in and strict local data storage requirements. The Haring Center emphasizes that human judgment remains irreplaceable: algorithms flag potential instances, but certified observers make final codes. As Dr. Martell stated in her 2024 keynote at NAEYC: “Data should serve children—not define them. Every Martell score tells part of a story. Our job is to listen deeply to the rest.”

For educators and consultants, Martell offers more than metrics—it cultivates disciplined observation habits that shift attention from deficits to capacities. When a toddler looks away mid-interaction, Martell doesn’t label it avoidance; it invites inquiry: Was the lighting too bright? Was the toy texture unexpected? Did the adult’s voice rise in pitch, triggering sensory sensitivity? That mindset—grounded in curiosity, humility, and developmental science—is Martell’s most enduring contribution to early childhood practice.

Training resources are available at haringcenter.org/martell. The certification fee is $295, with sliding-scale scholarships covering up to 100% for providers in Title I schools. All materials comply with WCAG 2.1 AA standards, including screen-reader-compatible coding sheets and ASL-translated training videos. No proprietary software is required—printable forms and Excel-based calculators remain fully supported for low-resource settings.

Ultimately, Martell succeeds not because it standardizes observation, but because it standardizes *respect*: respect for toddlers’ agency, for cultural variation in communication, and for the complexity of development unfolding in real time. Its power lies not in what it measures, but in what it helps adults notice—and how those notices transform relationships, routines, and readiness.

Research continues. The Martell Longitudinal Cohort Study—tracking 211 toddlers from 18 to 48 months—will publish its 36-month outcomes in late 2025, with preliminary data already showing correlations between early ‘Object-Focused Engagement’ trajectories and later executive function scores on the Head-Toes-Knees-Shoulders task (r = 0.62, p < 0.001).

As classroom dynamics evolve and new evidence emerges, Martell remains committed to iterative, practitioner-driven refinement—ensuring it stays rooted not in theory alone, but in the lived reality of toddlers building connection, one 15-second interval at a time.

Its name is not an acronym. It is a person—Dr. Laura Martell—and a promise: that careful, compassionate attention to the smallest moments can yield the largest impacts.

For early childhood educators and behavior consultants, Martell is less a tool and more a stance: steady, precise, and unwaveringly child-centered.

This stance begins—not with a checklist—but with a breath, a pause, and the deliberate choice to see.

And then, to respond.

That is where Martell begins. And where every meaningful intervention truly starts.

It is not about counting behaviors. It is about honoring them.

It is not about fixing what is broken. It is about nurturing what is already whole—and helping it grow.

That is the work. And Martell, at its best, helps us do it well.

Consistently. Accurately. Kindly.

With 15 seconds—and all the time in the world.