Everyday language input and production in 1,001 children from six continents.

Language is a universal human ability, acquired readily by young children, who otherwise struggle with many basics of survival. And yet, language ability is variable across individuals. Naturalistic and experimental observations suggest that children's linguistic skills vary with factors like socioeconomic status and children's gender. But which factors really influence children's day-to-day language use? Here, we leverage speech technology in a big-data approach to report on a unique cross-cultural and diverse data set: >2,500 d-long, child-centered audio-recordings of 1,001 2- to 48-mo-olds from 12 countries spanning six continents across urban, farmer-forager, and subsistence-farming contexts. As expected, age and language-relevant clinical risks and diagnoses predicted how much speech (and speech-like vocalization) children produced. Critically, so too did adult talk in children's environments: Children who heard more talk from adults produced more speech. In contrast to previous conclusions based on more limited sampling methods and a different set of language proxies, socioeconomic status (operationalized as maternal education) was not significantly associated with children's productions over the first 4 y of life, and neither were gender or multilingualism. These findings from large-scale naturalistic data advance our understanding of which factors are robust predictors of variability in the speech behaviors of young learners in a wide range of everyday contexts.

Linking acoustic variability in the infants' input to their early word production.

Exposure to talker variability shapes how learning unfolds in the lab, and occurs in the everyday speech infants hear in daily life. Here, we asked whether aspects of talker variability in speech input are also linked to the onset of word production. We further asked whether these effects were redundant with effects of speech register (i.e., whether speech input was adult- vs. child-directed). To do so, we first extracted a set of highly common nouns from a longitudinal corpus of home recordings from North-American English-learning infants. We then used the acoustic variability in how these tokens were said to predict when the children first produced these same nouns. We found that in addition to frequency, variability in how words sound in 6-17 month's input predicted when children first said these words. Furthermore, while the proportion of child-directed speech also predicted the month of first production, it did so alongside measurements of acoustic variability in children's real-world input. Together, these results add to a growing body of literature suggesting that variability in how words sound in the input is linked to learning both in the lab and in daily life. RESEARCH HIGHLIGHTS: Talker variability shapes learning in the lab and exists in everyday speech; we asked whether it predicts word learning in the real world. Acoustic measurements of early words in infants' input (and their frequency) predicted when infants first said those same words. Speech register also predicted when infants said words, alongside effects of talker variability. Our results provide a deeper understanding of how sources of variability inherent to children's input connect to their learning and development.

The role of vision in the acquisition of words: Vocabulary development in blind toddlers.

What is vision's role in driving early word production? To answer this, we assessed parent-report vocabulary questionnaires administered to congenitally blind children (N = 40, Mean age = 24 months [R: 7-57 months]) and compared the size and contents of their productive vocabulary to those of a large normative sample of sighted children (N = 6574). We found that on average, blind children showed a roughly half-year vocabulary delay relative to sighted children, amid considerable variability. However, the content of blind and sighted children's vocabulary was statistically indistinguishable in word length, part of speech, semantic category, concreteness, interactiveness, and perceptual modality. At a finer-grained level, we also found that words' perceptual properties intersect with children's perceptual abilities. Our findings suggest that while an absence of visual input may initially make vocabulary development more difficult, the content of the early productive vocabulary is largely resilient to differences in perceptual access. RESEARCH HIGHLIGHTS: Infants and toddlers born blind (with no other diagnoses) show a 7.5 month productive vocabulary delay on average, with wide variability. Across the studied age range (7-57 months), vocabulary delays widened with age. Blind and sighted children's early vocabularies contain similar distributions of word lengths, parts of speech, semantic categories, and perceptual modalities. Blind children (but not sighted children) were more likely to say visual words which could also be experienced through other senses.

Analyzing the effect of sibling number on input and output in the first 18 months.

Prior research suggests that across a wide range of cognitive, educational, and health-based measures, first-born children outperform their later-born peers. Expanding on this literature using naturalistic home-recorded data and parental vocabulary reports, we find that early language outcomes vary by number of siblings in a sample of 43 English-learning U.S. children from mid-to-high socioeconomic status homes. More specifically, we find that children in our sample with two or more-but not one-older siblings had smaller productive vocabularies at 18 months, and heard less input from caregivers across several measures than their peers with less than two siblings. We discuss implications regarding what infants experience and learn across a range of family sizes in infancy.

A cross-linguistic examination of young children's everyday language experiences.

We present an exploratory cross-linguistic analysis of the quantity of target-child-directed speech and adult-directed speech in North American English (US & Canadian), United Kingdom English, Argentinian Spanish, Tseltal (Tenejapa, Mayan), and Yélî Dnye (Rossel Island, Papuan), using annotations from 69 children aged 2-36 months. Using a novel methodological approach, our cross-linguistic and cross-cultural findings support prior work suggesting that target-child-directed speech quantities are stable across early development, while adult-directed speech decreases. A preponderance of speech from women was found to a similar degree across groups, with less target-child-directed speech from men and children in the North American samples than elsewhere. Consistently across groups, children also heard more adult-directed than target-child-directed speech. Finally, the numbers of talkers present in any given clip strongly impacted children's moment-to-moment input quantities. These findings illustrate how the structure of home life impacts patterns of early language exposure across diverse developmental contexts.

Establishing the reliability of metrics extracted from long-form recordings using LENA and the ACLEW pipeline.

Long-form audio recordings are increasingly used to study individual variation, group differences, and many other topics in theoretical and applied fields of developmental science, particularly for the description of children's language input (typically speech from adults) and children's language output (ranging from babble to sentences). The proprietary LENA software has been available for over a decade, and with it, users have come to rely on derived metrics like adult word count (AWC) and child vocalization counts (CVC), which have also more recently been derived using an open-source alternative, the ACLEW pipeline. Yet, there is relatively little work assessing the reliability of long-form metrics in terms of the stability of individual differences across time. Filling this gap, we analyzed eight spoken-language datasets: four from North American English-learning infants, and one each from British English-, French-, American English-/Spanish-, and Quechua-/Spanish-learning infants. The audio data were analyzed using two types of processing software: LENA and the ACLEW open-source pipeline. When all corpora were included, we found relatively low to moderate reliability (across multiple recordings, intraclass correlation coefficient attributed to the child identity [Child ICC], was < 50% for most metrics). There were few differences between the two pipelines. Exploratory analyses suggested some differences as a function of child age and corpora. These findings suggest that, while reliability is likely sufficient for various group-level analyses, caution is needed when using either LENA or ACLEW tools to study individual variation. We also encourage improvement of extant tools, specifically targeting accurate measurement of individual variation.

Talking to talkers: Infants' talk status, but not their gender, is related to language input.

Prior research points to gender differences in some early language skills, but is inconclusive about the mechanisms at play, providing evidence that both infants' early input and productions may differ by gender. This study examined the linguistic input and early productions of 44 American English-learning infants (93% White) in a longitudinal sample of home recordings collected at 6-17 months (in 2014-2016). Girls produced more unique words than boys (Cohen's d = .67) and this effect grew with age, but there were no significant gender differences in language input (d = .22-.24). Instead, caregivers talked more to infants who had begun to talk (d = .93-.97), regardless of gender. Therefore, prior results highlighting gender-based input differences may have been due, at least partly, to this talking-to-talkers effect.

Talker variability is not always the right noise: 14 month olds struggle to learn dissimilar word-object pairs under talker variability conditions.

Seminal work by Stager & Werker (1997) finds that 14-month-olds can rapidly learn two word-object pairings if the words are distinct (e.g. "neem" and "lif") but not similar (e.g. the minimal pair "bih" and "dih"). More recently, studies have found that adding talker variability during exposure to new word-object pairs lets 14-month-olds succeed on the more challenging minimal pair task, presumably due to talker variability highlighting the "relevant" consistencies between the similar words (Rost & McMurray, 2009; Galle et al., 2015; Hohle et al., 2020). It remains an open question, however, whether talker variability would be similarly useful for learning new word-object pairings when the words themselves are already distinct, or whether instead this extra variability may extinguish learning due to increased task demands. We find evidence for the latter. Namely, in our sample of 54 English-learning 14-month-olds, training infants on two word-object pairings (e.g. "neem" with a dog toy and "lof" with a kitchen tool) only led them to notice when the words and objects were switched if they were trained with single-speaker identical word tokens. When the training featured talker variability (from one or multiple talkers) infants failed to learn the pairings. We suggest that when talker variability is not necessary to highlight the invariant differences between similar words, it may actually increase task difficulty, making it harder for infants to determine what to attend to in the earliest phases of word learning.

Language input to infants of different socioeconomic statuses: A quantitative meta-analysis.

For the past 25 years, researchers have investigated language input to children from high- and low-socioeconomic status (SES) families. Hart and Risley first reported a "30 Million Word Gap" between high-SES and low-SES children. More recent studies have challenged the size or even existence of this gap. The present study is a quantitative meta-analysis on socioeconomic differences in language input to young children, which aims to systematically integrate decades of research on this topic. We analyzed 19 studies and found a significant effect of SES on language input quantity. However, this effect was moderated by the type of language included in language quantity measures: studies that include only child-directed speech in their language measures find a large SES difference, while studies that include all speech in a child's environment find no effect of SES. These results support recent work suggesting that methodological decisions can affect researchers' estimates of the "word gap." Overall, we find that young children from low-SES homes heard less child-directed speech than children from mid- to high-SES homes, though this difference was much smaller than Hart & Risley's "30 Million Word Gap." Finally, we underscore the need for more cross-cultural work on language development and the forces that may contribute to it, highlighting the opportunity for better integration of observational, experimental, and intervention-based approaches.

Talker variability shapes early word representations in English-learning 8-month-olds.

Infants must form appropriately specific representations of how words sound and what they mean. Previous research suggests that while 8-month-olds are learning words, they struggle with recognizing different-sounding instances of words (e.g., from new talkers) and with rejecting incorrect pronunciations. We asked how adding talker variability during learning may change infants' ability to learn and recognize words. Monolingual English-learning 7- to 9-month-olds heard a single novel word paired with an object in either a "no variability," "within-talker variability," or "between-talker variability" habituation. We then tested whether infants formed appropriately specific representations by changing the talker (Experiment 1a) or mispronouncing the word (Experiment 2) and by changing the trained word or object altogether (both experiments). Talker variability influenced learning. Infants trained with no-talker variability learned the word-object link, but failed to recognize the word trained by a new talker, and were insensitive to the mispronunciation. Infants trained with talker variability dishabituated only to the new object, exhibiting difficulty forming the word-object link. Neither pattern is adult-like. Results are reported for both in-lab and Zoom participants. Implications for the role of talker variability in early word learning are discussed.

Vocal development in a large-scale crosslinguistic corpus.

This study evaluates whether early vocalizations develop in similar ways in children across diverse cultural contexts. We analyze data from daylong audio recordings of 49 children (1-36 months) from five different language/cultural backgrounds. Citizen scientists annotated these recordings to determine if child vocalizations contained canonical transitions or not (e.g., "ba" vs. "ee"). Results revealed that the proportion of clips reported to contain canonical transitions increased with age. Furthermore, this proportion exceeded 0.15 by around 7 months, replicating and extending previous findings on canonical vocalization development but using data from the natural environments of a culturally and linguistically diverse sample. This work explores how crowdsourcing can be used to annotate corpora, helping establish developmental milestones relevant to multiple languages and cultures. Lower inter-annotator reliability on the crowdsourcing platform, relative to more traditional in-lab expert annotators, means that a larger number of unique annotators and/or annotations are required, and that crowdsourcing may not be a suitable method for more fine-grained annotation decisions. Audio clips used for this project are compiled into a large-scale infant vocalization corpus that is available for other researchers to use in future work.

A thorough evaluation of the Language Environment Analysis (LENA) system.

In the previous decade, dozens of studies involving thousands of children across several research disciplines have made use of a combined daylong audio-recorder and automated algorithmic analysis called the LENAⓇ system, which aims to assess children's language environment. While the system's prevalence in the language acquisition domain is steadily growing, there are only scattered validation efforts on only some of its key characteristics. Here, we assess the LENAⓇ system's accuracy across all of its key measures: speaker classification, Child Vocalization Counts (CVC), Conversational Turn Counts (CTC), and Adult Word Counts (AWC). Our assessment is based on manual annotation of clips that have been randomly or periodically sampled out of daylong recordings, collected from (a) populations similar to the system's original training data (North American English-learning children aged 3-36 months), (b) children learning another dialect of English (UK), and (c) slightly older children growing up in a different linguistic and socio-cultural setting (Tsimane' learners in rural Bolivia). We find reasonably high accuracy in some measures (AWC, CVC), with more problematic levels of performance in others (CTC, precision of male adults and other children). Statistical analyses do not support the view that performance is worse for children who are dissimilar from the LENAⓇ original training set. Whether LENAⓇ results are accurate enough for a given research, educational, or clinical application depends largely on the specifics at hand. We therefore conclude with a set of recommendations to help researchers make this determination for their goals.

From babble to words: Infants' early productions match words and objects in their environment.

Infants' early babbling allows them to engage in proto-conversations with caretakers, well before clearly articulated, meaningful words are part of their productive lexicon. Moreover, the well-rehearsed sounds from babble serve as a perceptual 'filter', drawing infants' attention towards words that match the sounds they can reliably produce. Using naturalistic home recordings of 44 10-11-month-olds (an age with high variability in early speech sound production), this study tests whether infants' early consonant productions match words and objects in their environment. We find that infants' babble matches the consonants produced in their caregivers' speech. Infants with a well-established consonant repertoire also match their babble to objects in their environment. Our findings show that infants' early consonant productions are shaped by their input: by 10 months, the sounds of babble match what infants see and hear.

Nature and origins of the lexicon in 6-mo-olds.

Recent research reported the surprising finding that even 6-mo-olds understand common nouns [Bergelson E, Swingley D (2012) Proc Natl Acad Sci USA 109:3253-3258]. However, is their early lexicon structured and acquired like older learners? We test 6-mo-olds for a hallmark of the mature lexicon: cross-word relations. We also examine whether properties of the home environment that have been linked with lexical knowledge in older children are detectable in the initial stage of comprehension. We use a new dataset, which includes in-lab comprehension and home measures from the same infants. We find evidence for cross-word structure: On seeing two images of common nouns, infants looked significantly more at named target images when the competitor images were semantically unrelated (e.g., milk and foot) than when they were related (e.g., milk and juice), just as older learners do. We further find initial evidence for home-lab links: common noun "copresence" (i.e., whether words' referents were present and attended to in home recordings) correlated with in-lab comprehension. These findings suggest that, even in neophyte word learners, cross-word relations are formed early and the home learning environment measurably helps shape the lexicon from the outset.

Familiarity plays a small role in noun comprehension at 12-18 months.

Infants amass thousands of hours of experience with particular items, each of which is representative of a broader category that often shares perceptual features. Robust word comprehension requires generalizing known labels to new category members. While young infants have been found to look at common nouns when they are named aloud, the role of item familiarity has not been well examined. This study compares 12- to 18-month-olds' word comprehension in the context of pairs of their own items (e.g., photographs of their own shoe and ball) versus new tokens from the same category (e.g., a new shoe and ball). Our results replicate previous work showing that noun comprehension improves rapidly over the second year, while also suggesting that item familiarity appears to play a far smaller role in comprehension in this age range. This in turn suggests that even before age 2, ready generalization beyond particular experiences is an intrinsic component of lexical development.

Look who's talking: A comparison of automated and human-generated speaker tags in naturalistic day-long recordings.

The LENA system has revolutionized research on language acquisition, providing both a wearable device to collect day-long recordings of children's environments, and a set of automated outputs that process, identify, and classify speech using proprietary algorithms. This output includes information about input sources (e.g., adult male, electronics). While this system has been tested across a variety of settings, here we delve deeper into validating the accuracy and reliability of LENA's automated diarization, i.e., tags of who is talking. Specifically, we compare LENA's output with a gold standard set of manually generated talker tags from a dataset of 88 day-long recordings, taken from 44 infants at 6 and 7 months, which includes 57,983 utterances. We compare accuracy across a range of classifications from the original Lena Technical Report, alongside a set of analyses examining classification accuracy by utterance type (e.g., declarative, singing). Consistent with previous validations, we find overall high agreement between the human and LENA-generated speaker tags for adult speech in particular, with poorer performance identifying child, overlap, noise, and electronic speech (accuracy range across all measures: 0-92%). We discuss several clear benefits of using this automated system alongside potential caveats based on the error patterns we observe, concluding with implications for research using LENA-generated speaker tags.

Day by day, hour by hour: Naturalistic language input to infants.

Measurements of infants' quotidian experiences provide critical information about early development. However, the role of sampling methods in providing these measurements is rarely examined. Here we directly compare language input from hour-long video-recordings and daylong audio-recordings within the same group of 44 infants at 6 and 7 months. We compared 12 measures of language quantity and lexical diversity, talker variability, utterance-type, and object presence, finding moderate correlations across recording-types. However, video-recordings generally featured far denser noun input across these measures compared to the daylong audio-recordings, more akin to 'peak' audio hours (though not as high in talkers and word-types). Although audio-recordings captured ~10 times more awake-time than videos, the noun input in them was only 2-4 times greater. Notably, whether we compared videos to daylong audio-recordings or peak audio times, videos featured relatively fewer declaratives and more questions; furthermore, the most common video-recorded nouns were less consistent across families than the top audio-recording nouns were. Thus, hour-long videos and daylong audio-recordings revealed fairly divergent pictures of the language infants hear and learn from in their daily lives. We suggest that short video-recordings provide a dense and somewhat different sample of infants' language experiences, rather than a typical one, and should be used cautiously for extrapolation about common words, talkers, utterance-types, and contexts at larger timescales. If theories of language development are to be held accountable to 'facts on the ground' from observational data, greater care is needed to unpack the ramifications of sampling methods of early language input.

What Do North American Babies Hear? A large-scale cross-corpus analysis.

A range of demographic variables influences how much speech young children hear. However, because studies have used vastly different sampling methods, quantitative comparison of interlocking demographic effects has been nearly impossible, across or within studies. We harnessed a unique collection of existing naturalistic, day-long recordings from 61 homes across four North American cities to examine language input as a function of age, gender, and maternal education. We analyzed adult speech heard by 3- to 20-month-olds who wore audio recorders for an entire day. We annotated speaker gender and speech register (child-directed or adult-directed) for 10,861 utterances from female and male adults in these recordings. Examining age, gender, and maternal education collectively in this ecologically valid dataset, we find several key results. First, the speaker gender imbalance in the input is striking: children heard 2-3× more speech from females than males. Second, children in higher-maternal education homes heard more child-directed speech than those in lower-maternal education homes. Finally, our analyses revealed a previously unreported effect: the proportion of child-directed speech in the input increases with age, due to a decrease in adult-directed speech with age. This large-scale analysis is an important step forward in collectively examining demographic variables that influence early development, made possible by pooled, comparable, day-long recordings of children's language environments. The audio recordings, annotations, and annotation software are readily available for reuse and reanalysis by other researchers.

Young Infants' Word Comprehension Given An Unfamiliar Talker or Altered Pronunciations.

To understand spoken words, listeners must appropriately interpret co-occurring talker characteristics and speech sound content. This ability was tested in 6- to 14-months-olds by measuring their looking to named food and body part images. In the new talker condition (n = 90), pictures were named by an unfamiliar voice; in the mispronunciation condition (n = 98), infants' mothers "mispronounced" the words (e.g., nazz for nose). Six- to 7-month-olds fixated target images above chance across conditions, understanding novel talkers, and mothers' phonologically deviant speech equally. Eleven- to 14-months-olds also understood new talkers, but performed poorly with mispronounced speech, indicating sensitivity to phonological deviation. Between these ages, performance was mixed. These findings highlight the changing roles of acoustic and phonetic variability in early word comprehension, as infants learn which variations alter meaning.

HomeBank: An Online Repository of Daylong Child-Centered Audio Recordings.

HomeBank is introduced here. It is a public, permanent, extensible, online database of daylong audio recorded in naturalistic environments. HomeBank serves two primary purposes. First, it is a repository for raw audio and associated files: one database requires special permissions, and another redacted database allows unrestricted public access. Associated files include metadata such as participant demographics and clinical diagnostics, automated annotations, and human-generated transcriptions and annotations. Many recordings use the child-perspective LENA recorders (LENA Research Foundation, Boulder, Colorado, United States), but various recordings and metadata can be accommodated. The HomeBank database can have both vetted and unvetted recordings, with different levels of accessibility. Additionally, HomeBank is an open repository for processing and analysis tools for HomeBank or similar data sets. HomeBank is flexible for users and contributors, making primary data available to researchers, especially those in child development, linguistics, and audio engineering. HomeBank facilitates researchers' access to large-scale data and tools, linking the acoustic, auditory, and linguistic characteristics of children's environments with a variety of variables including socioeconomic status, family characteristics, language trajectories, and disorders. Automated processing applied to daylong home audio recordings is now becoming widely used in early intervention initiatives, helping parents to provide richer speech input to at-risk children.