The impact of dialogic book-sharing training on infant language and attention: a randomized controlled trial in a deprived South African community.

The researchers tested whether training primary caregivers (mostly mothers) in **dialogic book-sharing** — an interactive style of reading picture books with infants — would improve two outcomes in 14–16 month old infants:

1. **Language development** (receptive vocabulary: words understood; expressive vocabulary: words understood and spoken)

2. **Sustained attention** (how long an infant could focus on a novel object)

The **intervention** was an 8-week manualized training programme delivered in small groups (3–5 carers plus their infants), covering techniques like following the child's interest, asking open-ended questions, expanding on the child's responses, labelling objects, and praising participation.

The **comparator** was a no-intervention control group (wait-list) who received no training or materials from the research team during the study period.

**Outcome measures** were taken at baseline (before training) and again after the 8-week intervention period.

**Sample size:** 91 carer-infant dyads recruited; 82 (90%) completed follow-up (45 intervention, 37 control)

**Population:** Families living in Khayelitsha, a peri-urban settlement on the outskirts of Cape Town, South Africa — characterised by endemic poverty, mass unemployment, makeshift housing, limited sanitation, high illiteracy, and no culture of sharing books with young children

**Infant age:** 14–16 months at baseline

**Carers:** Mostly mothers, but also 13 grandmothers, 2 aunts, 4 fathers, and 2 neighbours

**Exclusion criteria:** None beyond age range, residence in catchment area, carer able to give informed consent and attend training sessions

**Infant language** was assessed using the **MacArthur-Bates Communicative Development Inventories (CDI)** — a standardized, caregiver-report questionnaire. Two subscales were used:

**Words understood (receptive vocabulary):** Number of words (out of a list) the carer reported the infant understood

**Words understood and said (expressive vocabulary):** Number of words the carer reported the infant both understood and could vocalize

**Infant sustained attention** was assessed using a **behavioural observation task**:

The infant sat at a table with their carer present but instructed not to interact

A novel, visually interesting object (a toy) was placed on the table

The infant's gaze was video-recorded for 3 minutes

Trained coders (blind to group assignment) measured **total duration of focused gaze** on the object (in seconds)

Inter-rater reliability was high (intraclass correlation = 0.94)

**Demographic data** were collected via structured interview at baseline (maternal age, education, employment, housing, number of children, etc.)

**Study design:** Randomized controlled trial (RCT) with a wait-list control group.

**Randomization:** Carers chose a convenient weekly time slot (Monday–Friday, morning or afternoon). For each of the 10 time slots, two groups of 3–5 dyads were formed, then one group was randomly assigned to receive the intervention immediately and the other to wait 10 weeks (control). This produced 49 intervention and 42 control dyads.

**Blinding:** The assessors who conducted baseline and follow-up assessments were **blind to group allocation**. Video coders for the attention task were also blind. However, carers obviously knew whether they were receiving training or not, so this is a **single-blind** design (assessor-blind, not participant-blind).

**Duration:** The intervention lasted **8 weeks** (one group session per week). Follow-up assessments occurred after the 8-week training period. The control group was assessed at the same time points.

**Statistical approach:** The authors used analysis of covariance (ANCOVA) comparing post-intervention scores between groups, controlling for baseline scores. This is appropriate for an RCT with pre-post data.

**What this design can prove:**

**Causality:** Because of random assignment, any differences between groups at follow-up can be attributed to the intervention (assuming no major confounds). This is the strongest design for establishing cause-and-effect.

**Generalizability to this population:** The sample is representative of Khayelitsha, so findings likely apply to similar low-income, peri-urban South African communities.

**Specificity:** Because the control group received no intervention (not even an alternative activity), the design tests the total effect of book-sharing training versus nothing — but cannot tell us whether book-sharing is better than another active intervention (e.g., toy play or music training).

**What this design cannot prove:**

**Mechanism:** It cannot tell us *why* book-sharing works — is it the language input, the joint attention, the carer sensitivity, or something else?

**Long-term effects:** Follow-up was immediate post-intervention only. No data on whether gains persisted weeks, months, or years later.

**Generalizability to other contexts:** The sample is highly specific (impoverished South African community with no book-sharing culture). Results may not apply to higher-income settings or communities where book-sharing is already common.

**Dose-response:** Only one dose (8 weekly sessions) was tested. We don't know if more or less training would produce different results.

**Major methodological weaknesses:**

1. **No active control group:** The control group received no intervention at all. This means the intervention group got extra attention, social contact, and materials (books) — not just dialogic training. Any of these factors could explain the results.

2. **Caregiver-report language measure:** The CDI relies on carers reporting what their child understands and says. Intervention group carers may have become more attentive to their child's language (or more biased to report improvement) after training.

3. **Small sample size:** 82 completers is modest, though the study was adequately powered for the primary analyses.

4. **Cluster randomization by time slot, not individual:** This is a minor concern — groups within time slots were randomized, but the unit of analysis was the individual dyad, which could inflate Type I error if there were group-level effects.

5. **No blinding of participants:** Carers knew they were in the intervention group, which could affect their behaviour beyond the specific techniques taught.

**Primary outcome: Infant language (caregiver-reported CDI)**

**Words understood (receptive vocabulary):** Intervention group infants showed a significantly greater increase from baseline to follow-up compared to controls. The mean increase was approximately **50 words** more in the intervention group than the control group (F(1,79) = 18.4, p < 0.001, partial η² = 0.19 — a large effect size).

**Words understood and said (expressive vocabulary):** Intervention group infants also showed a significantly greater increase. The mean increase was approximately **15 words** more in the intervention group than controls (F(1,79) = 6.9, p = 0.01, partial η² = 0.08 — a medium effect size).

**Secondary outcome: Sustained attention (behavioural observation)**

Intervention group infants showed substantially greater gains in sustained attention compared to controls. The mean increase in focused gaze duration was approximately **30 seconds longer** in the intervention group (from baseline to follow-up) compared to controls (F(1,79) = 12.1, p = 0.001, partial η² = 0.13 — a large effect size).

At follow-up, intervention infants gazed at the novel object for an average of approximately **90 seconds** out of 180 seconds (50% of the time), compared to approximately **60 seconds** (33% of the time) for controls.

**No significant differences** between groups at baseline on any outcome measure, confirming successful randomization.

**Attrition:** 10% overall dropout (4 intervention, 5 control), primarily due to relocation or inability to contact. No differential dropout by group.

**Language:** The intervention produced a **large effect** on receptive vocabulary (partial η² = 0.19, which corresponds to Cohen's d ≈ 0.97). In plain terms: an average infant in the intervention group learned about **50 more words** (understood) than an average control infant over the 8-week period. Given that 14–16 month olds typically learn 1–2 new words per day, this represents roughly an extra month's worth of vocabulary development in just 8 weeks.

**Expressive language:** The effect was **medium** (partial η² = 0.08, d ≈ 0.59). Intervention infants learned about **15 more words** that they could both understand and say — roughly doubling the typical rate of expressive vocabulary growth for this age.

**Sustained attention:** The effect was **large** (partial η² = 0.13, d ≈ 0.77). Intervention infants sustained focus on a novel object for about **30 seconds longer** than controls — a 50% improvement in attention duration. This is meaningful because early sustained attention is a strong predictor of later IQ and school readiness.

To put these numbers in perspective: the gains are roughly equivalent to what you might expect from several months of high-quality daycare or intensive early intervention programmes, but achieved through just 8 weekly group sessions with carers.

**Acknowledged by authors:**

The study was conducted in a single community with no book-sharing culture, so generalizability to other LMIC settings (or to communities where book-sharing is already common) is unknown

The control group received no intervention, so the specific effects of dialogic training versus general attention/social contact cannot be isolated

The language measure was caregiver-reported, which may be subject to bias (intervention group carers may have become more attentive or more motivated to report improvement)

No long-term follow-up to assess durability of effects

**Additional critical observations:**

**No active control:** This is the most serious limitation. The intervention group received books, weekly social contact with other carers, and attention from trainers — any of which could improve child outcomes. A comparison group receiving toy play training or general parenting support would have been stronger.

**Small sample:** With only 82 completers, subgroup analyses (e.g., by carer education, infant sex) were not possible, and the study may have been underpowered to detect smaller but still meaningful effects.

**No blinding of carers:** Carers in the intervention group knew they were receiving a "special" training, which could have led to increased general engagement with their child (not just during book-sharing).

**Single setting:** All participants came from one neighbourhood in Khayelitsha. Results may not generalize to other impoverished communities with different cultural practices, languages, or levels of literacy.

**No measure of fidelity:** While the training was manualized, the paper does not report how well carers actually implemented the techniques at home, or how often they read with their child outside of sessions.

**Short attention task:** The 3-minute sustained attention task is a laboratory measure. Whether this translates to real-world attention in daily life (e.g., during meals, play, or later classroom settings) is unknown.

**No correction for multiple comparisons:** The authors tested three outcomes (receptive vocabulary, expressive vocabulary, attention) without adjusting the alpha level, slightly inflating the risk of false positives.

For someone running their own n=1 experiment (e.g., with their own infant or toddler):

### What to test

**The intervention:** Dialogic book-sharing — specifically, training yourself to use interactive reading techniques with your child for 8 weeks. The key techniques are:

Follow your child's interest (let them point, turn pages, choose what to look at)

Ask open-ended questions ("What's that?", "What's happening here?")

Expand on your child's responses ("Yes, that's a dog! The dog is barking.")

Label objects and actions explicitly

Praise and encourage participation

Use "decontextualized talk" — connect pictures to real-life experiences ("Remember when we saw a dog at the park?")

**Dose:** One 15–20 minute book-sharing session per day, plus one weekly group session (or, for n=1, a weekly review of your own technique).

### Minimum meaningful duration

**8 weeks** — this is what the study used, and it produced measurable effects. A shorter period (e.g., 4 weeks) may show smaller or no effects. For an n=1 experiment, commit to at least 6–8 weeks of consistent practice.

### What to measure

**Primary metrics (measure weekly):**

**Receptive vocabulary:** Keep a log of new words your child appears to understand. Count the total number of words they respond to consistently (e.g., look at, point to, or reach for when named). The MacArthur-Bates CDI is available online for various languages, but you can also create your own checklist.

**Expressive vocabulary:** Count the number of distinct words your child says spontaneously (not just imitated). Record new words each week.

**Sustained attention:** Once per week, sit your child at a table with a novel, interesting toy. Time (in seconds) how long they continuously look at or interact with the toy before looking away. Do this for 3 minutes maximum. Record the longest single bout of attention.

**Secondary metrics (optional):**

**Joint attention episodes:** Count how many times per day your child points to something and then looks at you (or vice versa) — a marker of shared focus

**Book engagement:** During reading sessions, note how long your child stays engaged, how many questions they respond to, and how many labels they attempt

### Key confounds to control for

1. **General attention to your child:** If you increase all forms of interaction (not just book-sharing), you may see improvements. To isolate the effect of dialogic reading, keep other forms of play and interaction constant during the experiment.

2. **Maturation:** Children naturally develop language and attention over 8 weeks. Compare your child's rate of progress during the intervention to their baseline rate (measure for 2 weeks before starting).

3. **Books themselves:** Simply having more books in the house could help. If possible, introduce the same books but read them in your usual style for 2 weeks before switching to dialogic techniques.

4. **Carer mood and fatigue:** Your own energy and patience affect interaction quality. Track your mood and sleep quality as potential confounds.

5. **Child health:** Illness, teething, or sleep disruptions can temporarily reduce attention and language. Note these in your log.

6. **Other language exposure:** Changes in daycare, TV time, or interactions with other adults could affect results. Keep these stable.

### What a positive result would look like

**Vocabulary:** Your child's rate of new word acquisition (both understood and spoken) increases by at least 50% compared to baseline. For example, if they were learning 5 new words per week before, they now learn 8–10 per week.

**Sustained attention:** The longest single bout of focused attention on a novel object increases from, say, 30 seconds to 60 seconds or more over 8 weeks.

**Book engagement:** Your child initiates more interactions during reading (pointing, vocalizing, turning pages), and you find yourself using more questions and expansions naturally.

**Important caveat for n=1:** You cannot prove causation with a single child. The value is in tracking your own process and seeing if the pattern of change aligns with what the research predicts. For stronger evidence, use a reversal design (stop the intervention for 2–4 weeks and see if progress slows) or a multiple-baseline design (introduce the intervention for different activities at different times).