Reading glossary

I don’t plan to pre-post glossary items for the week 8 and 9 readings, which are not tested on either midterm. If I find time, I’ll post them when I’m able, before the end of the quarter (but don’t depend on it). Thanks for your understanding!:

A few notes of context:

• These glossary entries are meant to help students understand jargon terms/opaque expressions as they read. Most students in the class don’t have a background in linguistics, cogsci, or infant psychological methods!
• This is an extra task for me and supplemental information for students, so I’m not closely checking for typos etc.. Please do let me know if there’s something especially unclear and I’ll do what I can to correct it.
• We covered some glossary items for Saffran et al. (1996) and Saffran (2020) in class. So, at least for now, I’m not going back and adding them.
• If I’ve covered a term in one reading, I likely won’t list it again for the following readings. So this page is meant to be reviewed from top to bottom.

Happy reading!

- Marisa

Paper index:

1. Maye et al. (2002)
2. Smith & Yu (2008)
3. Cristia & Casillas (2022)
4. Goldin-Meadow (2020)
5. He & Arunachalam (2017)
6. Frank et al. (2020)
7. Gergely et al. (2002)
8. Herrmann et al. (2007)
9. Kail (1991)
10. Vlach & Johnson (2013)
11. Kuhl et al. (2003)
12. Smith et al. (2018)
13. Liszkowski et al. (2012)
14. Yu & Smith (2013)
15. Gaskins (2006)
16. Casillas et al. (2020)
17. Rowe & Weisleder (2020)

Back to main page

Requested additions

• domain-general versus domain-specific: “Domain” here refers to cognitive domains (e.g., language, attention, social cognition, …). Domain-general indicates a mechanism or feature of cognition or learning that crosses multiple domains. Domain-specific, in contrast, refers to mechanisms that are only available to one area of cognition. In the case of our course, “domain-specific” will almost always refer to “language-specific” cognition.

Maye et al. (2002)

• acoustic dimensions: When we describe speech acoustically, we can measure the signal in many different ways (each of these measures is a ‘dimension’ of hypothetical acoustic space). Some measures that are crucial for measuring one set of sounds (e.g., “voice onset time” for consonants p/b, t/d, and k/g) are irrelevant for others (consonants f,z). Maye et al. vary their stimuli in a single acoustic dimension: VOT (see below).
• (acoustic) tokens: “Token” contrasts with “type”—tokens are individual instances whereas types are categories. So you might have an abstract category (i.e., type—here a phonological type) “p” but when you speak it or hear it, you only get access to individual examples from that category (i.e., tokens—here phonetic tokens).
• bimodal vs. unimodal: This refers to the mode(s) of a statistical distribution. The mode of a distribution is the most common value. So if you imagine a bell curve, the mode is the x-axis value under the peak. Distributions can have no mode (“uniform”) or have multiple (“multimodal”; note this term is used elsewhere in the language literature to refer to multiple communication signals). Maye et al.’s experiment uses two artificial distributions of phonetic forms: one unimodal (one mode) and one bimoday (two modes), which the idea that kids might perceive each mode and its surrounding distributional mass as a phonetic category.
• category structure: In this paper, category structure refers to the probabilistic distributions of tokens around their category centers (i.e., a mode and the body of mass around it). The broader use of this term refers to the form and organization of mental categories (i.e., representations).
• discriminable contrasts: Pairs of sounds that are or can be perceived as different (e.g., ‘p’ versus ‘b’). Adults largely lose the ability to hear the contrast between sounds that are non-native to their language , especially if those sounds are not especially acoustically salient (e.g., for monolingual English speakers… unaspirated ’t’ made as in English vs. ’t’ with the tongue tip curled back as in Hindi).
• familiarization: As discussed last week, this refers to the period in an experiment during which infants are passively exposed to the stimuli; in this case another artificial language. This gives them the ‘data’ upon which to act in the test phase.
• interstimulus interval (ISI): The period of “silent” time between playback tokens in the experiment.
• masking music: The adults accompanying the infant participants (often including the experimenter!) are often asked to wear headphones over which the scientists play music that interferes with their ability to hear what the infant is hearing. The adult may also be asked to wear masking glasses (usually sunglasses with the lenses blocked out). This keeps the adult from (even unintentionally) influencing the infants’ behavior during the experiment.
• minimal pairs: This phrase isn’t explicitly mentioned, but is implied by “speech sounds that distinguish word meanings”. A minimal pair for a speech sound is like “bear” and “pear”, which are only distintuished on the basis of the first consonant. Linguists take this to mean that the ‘b’-‘p’ contrast bears function in English (otherwise we’d lose the ability to hear the difference between these two words). Such pairs of linguistic units—with only one difference—are called minimal pairs. Minimal pairs are an important analytical tool available to linguists in describing the meaningful units of a language.
• native language input: “Native langage” is also referred to as home language or mother tongue. Infants may of course have more than one of these. “Input” is the experience they have hearing (or in the case of visual sign languages, seeing) that language. Some theories mostly care about “input” addressed to the child (“child-directed input”) and others think more broadly about what children learn from the ambient language around them.
• perception of speech sounds: In this paper and other work we read on infant speech perception, this phrase refers to whether, when, and how infants come to think that individual instances of sounds fall into the same sound CATEGORY (e.g., two spoken productions of “p”) versus not (e.g., one “p” and one “k”). This has to do with the infants’ growing mental (INTERNAL) representations of which categories are relevant in their language environment.
• (phonetic) continuum: A gradient (non-discrete) path between two phonetic endpoints. In this case, think of it a bit like phonetic ‘mixing’: so an ’t’-’d’ gradient with three points might be a 100% t sound, then a 50% t-50% d sound, then a 100% d sound.
• phonetic organization of native language: This refers to the way in which an individual language groups some instances of sounds together into categories and separates other sounds into instances of separate categories. So, while both English and Spanish have ‘p’ and ‘b’ sounds (a phonological distinction—having to do with abstract representation of the sound system), the way this pair of sounds is actually produced (a phonetic distinction) is fairly different. In fact, the ‘p’ in Spanish is phonetically quite close to the ‘b’ in English (in both case, these are unaspirated stops). So “phonetic organization” here refers to the way that a language’s abstract phonological categories result in clusters of closer and further-apart categories in phonetic (acoustic and articulatory) space. For the purposes of the present study, you can just treat “phonetic organization” as a set of as phonetic distributions that infants, via exposure, can then mentally internalize and as phonetic categories.
• preferential looking procedure: A measure of infants’ interest in hearing the test sounds. Using a light or another attention getter, the experimenter orients the infant’s visual attention to a sound source (in this study, a screen) and then starts playing a test sound. When the infant looks away, the sound playback is stopped. This is done again and again for all the sounds in the test phase, and infants’ interest in listening is assumed to correlate with their total listening duration. You can see a similar set up, but with the infant listening to sounds on the left and right (Maye et al. just use a center screen) in this extremely 90s video (5m28s, “Head Turn Preference Procedure”)
• stochastic: Term often used in the statistical learning and computational modeling literatures of cognitive science; refers not to randomness per se, but the probabilistic nature of learning processes and outcomes.
• voiced and voiceless unaspirated stops: Voiced = if your vocal folds are vibrating; voiceless = if the’re not. Unaspirated = no substantial puff of air when transitioning between the consonant and its following vowel. Stop = a consonant that involves full closure of the airway (e.g., by closing the lips, or putting the tongue on the roof of the mouth).
• VOT: VOT = voice onset time. VOT is the measurement, in milliseconds, of the period between the release of a stop’s closure and the very beginning of vocal fold vibration in the following vowel. So think about the consonant ‘pa’. There’s a very short period of transition between the ‘p’ and the ‘a’. This transition time is VOT. VOT can be really long in English, where ‘p’ sounds at the start of many syllables have a puff of air following them (‘p__a’); in this case, voicing starts when the ‘a’ begins. VOT is close to zero in English ‘b’, which is otherwise identical to English ‘p’ (‘pa’). In Spanish and many other languages, VOT is actually negative for ‘b’, because the vocal folds are already vibrating while the lips are still closed (’–ba’); this is sometimes referred to as ‘prevoicing’. In the Maye et al. paper, English-learning infants are exposed to the Spanish-like system: VOT that ranges from negative to zero VOT. Why might that matter for their research aims (for a clue, check footnote 4)?

Back to main page / Back to top

Smith & Yu (2008)

• associative learning: Contrasts with hypothesis-driven learning in that learners never need to “commit” to a set of hypotheses. So, when I hear “zimble” with two novel objects, I they each become equally associated with “zimble” (e.g., the probability that I see object A when I hear “zimble” is 1/1, same as object B). When I hear “zimble” again later, and I only see object B, the associations begin to diverge: 1 instance of object A across two mentions of “zimble” (1/2 = 50%) but 2 instances of object B across two mentions of “zimble” (2/2 = 100%). There is never any “hypothesis” in this framework—just stronger or weaker associations built up by co-occurrence over time.
• co-occurrence frequency: The frequency with which two stimuli occur together.
• constraints: In this use, constraints refers to mechanisms that helpfully restrict children’s understanding of the world. An example of a constraint is the Whole Object Assumption, which basically states that children, by default, assume that words refer to whole objects (and not, e.g., individual parts of objects). This may save the learner a lot of trouble in deciding whether ‘dog’ refers to the whole creature, the creature and the ball in his mouth, or just his wagging tail—the Whole Object Assumption may nail down the candidate referents to just the dog or the ball objects.
• cross-trial statistics: Also referred to as “cross-situational” statistics elsewhere. In this study, this phease refers to the probability that a label is used in the presence of a referent over the course of the multiple trials. Here, multiple trials (in this case, exposures to label-referent groupings) gives infants a sense of emerging, cumulative evidence for the correct mappings between words and objects.
• formal simulations: This refers to computational modeling, which is broadly used in cognitive science, including language development research. Simulation modeling implies a computational model meant to investigate one or more specific mechanisms by which a learner draws inferences from finite input data.
• hypothesis-driven learning: A view of learning in which individual learners hold a set of hypotheses about how the world works that can then be assessed when new evidence is presented. So, for example, if I hear “zimble” and see two novel objects, under this learning framework, I’d randomly pick one of the two objects as a zimble. Then, on a later trial, if I’m wrong, I’ll re-assign the meaning of “zimble” to the other object (or if I’m right, I’ll keep it).
• main vs. interaction effect: This kind of expression may often be used in the results sections of our papers when describing statistical outcomes. “Effect” refers to a difference in the measurement of interest—historically researchers will be seeking evidence for “significant” effects, which indicate that the difference is unlikely to have arisen by chance. “Main” effects are sometimes called “simple” effects—they indicate some overall difference in measurement. “Interaction” effects are differences that only show up in some cases. So, for example, if we see a difference in infant looking time between Condition A and Condition B, but only in our 8mo sample and not our 10mo sample, we are likely to have an interaction between Age and Condition—the effect of condition is conditional on which age we are considering.
• mutual exclusivity: We use what we know to make inferences about what we don’t know. When infants and young children are shown a novel or unfamiliar object (e.g., nut and bolt) and a familiar one (e.g., a ball), they will assume that a new word (“zimble”) refers to the less familiar object. Making this inference is referred to as a “mutual exclusivity” effect.
• randomization: You’ll see a lot of talk about randomization in the experimental studies we read. Basically experimenters do what they can to guarantee that, if infants show differences in behavior across conditions of interest (e.g., bimodal versus unimodal familiarization), there’s no way to attribute that difference to some uninteresting property of the study, such as the order in which the audio files were played or where they appeared on a screen.
• referent: The meaning that a word or expression refers to. When I say “look at that dog”, the referent of the linguistic symbol ‘dog’ is the canine I’m drawing attention to. The mapping (i.e., established linkage) between referring expressions and their referents is complex. For example, I can refer to the same canine referent as ‘Fido’ or ‘my dog’ or ‘him’. Mappings depend on context, but can also show considerable consistency (the word ‘dog’ has a decent tendency in referring to canine creatures).
• reliability: When measurements are made on the basis of human judgments, it’s often a good idea to have a second human judge independently measure the data and make sure that the two sets of judgments are reasonably similar. This is typically what is happening when researchers refer to reliability in our papers. There are other kinds of reliabiity measures (e.g., measuring a machine’s performance against a human’s) but we won’t cover them.
• target vs. distracter vs. filler: Target refers to a stimulus that is of interest for hypothesis testing. Distracter refers to a stimulus that acts as a foil to a target (e.g., a plausible alternative if the infant doesn’t perceive some key difference). A filler is a stimulus that is not of interest for hypothesis testing, but may be useful for other reasons, such as getting infants’ attention, providing evidence that learning is restricted to target items, or masking the hypothesis from the participants.

Back to main page / Back to top

Cristia & Casillas (2022)

Okay there are a lot of specialized concepts in this paper, so I’ve organized them by general domain

phonology and phonetics words

• continuant: A consonant that can be produced with continuous airflow. Compare s, z, m, and ch to p, d, and k.
• deletion/insertion/substitution/metathesis: Mechanisms by which strings might differ when measuring Levenshein difference: deletion (leaving a sound out), insertion (adding a sound in that wasn’t in the original word), substitution (adding an alternative sound in place of a sound from the original word), and metathesis (switching the order of two sounds in the original word).
• graphemes: Representations of sound categories in written language (written letters).
• lateral consonant: Consonants produced by having a central point of closure in the oral tract while allowing continuous airflow around the sides of the tongue. English has one of these ’l’. Other languages may have more, including Ku Waru, the Papuan language mentioned in the paper.
• Levenshtein distance (a “phoneme-level score”): A method for calculating how different two strings of sounds are for each other. Can be defined in slightly different ways—look in the paper for our way of defining it!
• nasal/oral and short/long vowels: Vowels can contrast in whether their airflow goes primarily through the mouth (oral) or the nose (nasal). In French, for example, beau /bo/ versus bon /bõ/ (sounds shown between slashes indicate phonological representation; sounds shown between square brackets indicate phonetic ones). Vowel nasalization isn’t contrastive in English. Vowels can also contrast in length (short/long), for example in Finnish: ’tuli’ (fire) versus ’tuuli’ (wind).
• non-word repetition (NWR): A method by which children’s phonological system is tested by having them listen to and repeat back phonologically legal strings of sound that have no meaning. The meaningless legal strings (“non-words”) can be easier or harder. This task involves children’s phonological working memory (for decoding, remembering, and encoding what they hear) as well as their stable phonological representations.
• phonological development: The process by which children acquire the phonology of the languages in their home environments.
• phonological inventory: The full collection of abstract sound representations that can be combined to create meaningful forms (e.g., words) in a language.
• phonological working memory: A special memory mechanism we have in our brains that holds on—for a short period—onto acoustically detailed signals that a hearer encountered in their sound environment. This can include rehearsed speech that the hearer is preparing to speak aloud. Sometimes called the “phonological loop”.
• phonologically legal / phonotactics: In any given language, there are limitations on what phonological sequences are permitted. “stsots” is a normal-sounding word in the Mayan language Tseltal (it means ‘his/her hair’), while “fry” is a normal-sounding word in English but impossible in Tseltal phonology. Legal sequences include both legal sounds that are combined in a legal way. So English has ’t’ and ’s’ but never combines them into ‘sts’ at the start of words; therefore it’s not phonotactically legal in English. The permitted ones are sometimes called “legal” or “grammatical” and non-permitted “illegal” or (more commonly) “ungrammatical”.
• phonologization: A language change process by which some variation in phonetic forms becomes integrated into the phonology of the language; this kind of process typically takes several (perhaps many) generations.
• phonology: The set of abstract representations and processes for a language’s sound system; also the name of the discipline that studies such sound systems. “sound” system is actually a little wrong, because sign languages also have phonology—the core phenomenon in phonology is the unit of language that is combined to create minimal meanings (e.g., sounds within a spoken word, handshape of a signed word). contrasts with phonetics, which concerns how (abstract) phonological systems are realized in the physical world (especially articulatorily and acoustically). We observe phonetic facts and infer phonological ones.
• place of articulation (phonetics): A feature of consonants that refers to where, in the oral tract, there is full or partial closure that gives the consonant its distinct quality. “Labial” consonants are produced with full/partial closure on the lips; “Alveolar” consonants are produced with full/partial closure on the alveolar ridge (the bump on the roof of your mouth, just behind your teeth), etc. Organized into column names in the International Phonetic Alphabet: https://en.wikipedia.org/wiki/International_Phonetic_Alphabet.
• plosive/stop consonant: These two words are used jointly to refer to consonants that have a full closure of air at some point in their production (stop = no airflow; plosive = burst of air when the closure is released): in English, p, b, t, d, k, g.
• pre-nasalized, nasally released, labialized, palatalized: Consonants can have a ‘secondary’ articulation, which is basically like having a tagalong consonant right before or after the consonant is produced. For reasons we won’t get into, the tagalong doesn’t get to count as a full sound on its own. But prenasalized means nasal airflow before the consonant starts ’ndoo’ (tooth; Yélî Dnye); nasally released means nasal airflow after the consonant ends ‘dnye’ (sound; Yélî Dnye); labialized release means lip rounding on the consonant end (like the tw in ’twin’); palatalized repease means a ‘y’ like sound on the consonant end (like the py sound at the start of ‘puny’).
• simplex versus doubly-articulated: Consonants that are produced with only one closure point (simplex) versus two closure points (complex - specifically here ‘doubly-articulated’, sometimes ‘co-articulated’), such as kp and tp. These are not produced as distinct sequences of consonant closures, but as near-simultaneous consonant productions. These consonants are common in languages of Africa and Papua New Guinea. English only has simplex consonants.
• voicing contrast: These are pairs of sounds that differ only in when and whether the vocal folds are vibrating. In English, the following pairs have voicing contrasts (the second one is ‘voiced’): p-b, t-d, k-g, f-v, th-th (thigh-thy), s-z, ch-dj (cheese-jeez), sh-zh (mesher-measure; tbh there’s nore really a good minimal pair for this one…).

other linguistic unit words

• CV and CVC syllable formats (and syllable complexity): The abstract structure of a syllable, where C = consonant and V = vowel. The most complex syllable in English would be a CCCVCCC: strengths /stɹɛŋθs/.
• lexical neighbor: All the words that are similar to a target word (often limited to those with a one-sound difference). So for ‘coffee’, close neighbor would be ’toffee’ and ‘coughing’, and further neighbors would be ‘coffer’ and ‘iffy’.
• lexicon: “the lexicon” is the mental dictionary; at least including words, their meanings, and their interconnections, but also information about their use and associations in actual interaction.
• monosyllables: one-syllable form (‘ba’); disyllable (‘baba’); trisyllable (‘bababa’), etc..
• morphemes: the smallest meaningful unit of language. For example, the word ‘dogs’ has two morphemes: dog (noun) and -s (plural marker).
• resyllabification: When the boundary between two syllables moves because of the conditions under which they are combined, which means consonants may attach to vowels in articulation in a different way. Take, e.g., how in ‘run’ the ‘-n’ is attached to the ‘u’ vowel, but in ‘running’ it becomes attached to the ‘i’ of ‘ing’ (ruh-ning not run-ing).
• word types: see the (acoustic) tokens entry above—this is the same, but with words.

language evolution and variation

• diachronic language change: Change in language over chronological time.
• isolate: Languages evolve and spread in ‘families’. For example, English is part of the Germanic family, which includes English, Dutch, German, Icelandic, Danish, and many more. A language with no known ‘family’ is referred to as an ‘isolate’.
• typologically frequent: Broadly refers to the fact that some linguistic structures recur frequently across diverse language families. If, in contrast, a structure is rare across language families, it is typologically infrequent.

community description

• ethnographic: pertaining to the scientific description of a social or cultural group.
• hamlet: a small collection of households; smaller than a village, which may include several hamlets.
• patrilocal: a system in which women, upon marriage, move into their husband’s family’s land/dwellings; the complimentary system type is called ‘matrilocal’.
• swidden horticulturalists: small-scale farmers who move their garden plots from newly-burned clearing to newly-burned clearing in a forested area.

statistics and open science

• descriptive statistics: statistics summarizing the distribution of data, or the relationship between sources of data in a dataset; contrasts with “inferential statistics”, which are used to extrapolate beyond the data itself.
• exploratory analysis: analysis from which no inferential claims will be made—the scientist is planning purely on observing the patterns in the data descriptively, perhaps with the plan to later form hypothesis and test them more rigorously in future work.
• meta- or mega-analysis: quantitative analyses in which the datasets are a collection of prior empirical findings (meta-analysis) or a collection of prior meta-analyses (mega-analysis). This approach is used to assess the evidence for a phenomenon across all the relevant work produced in a field.
• monotonic: changing in one direction (always increasing/decreasing)
• natural logarithm: ln (x) - practically used in statistical analyses to help make skewed distributions approximate a normal (Gaussian) distribution. This is done to better align with the assumptions of the statistical models being used (i.e., that the data are normally distributed).
• pre-registering (a study): declaring, in some public scientific forum, the hypotheses and planned analyses for a study before analysis is done. Sometimes pre-registration may also involve declaring planned methods for data collection. Pre-registration is a practice that developed to counter the practice of ‘p-hacking’, which is widely attributed as a factor underlying the replication crisis in psychology. more here and here.

other

• elicitation: Methods by which linguists get participants to produce language data of one kind or another (e.g., asking participants to read a list aloud or to talk to each other, or asking questions for participants to answer, etc.)
• informant: In field linguistics, a native speaker of the language in question whose expertise is key to understanding linguistic patterns in the data and ensuring the quality of linguistic stimuli.
• MacArthur-Bates Communicative Development Inventory (“CDI”): A vocabulary checklist that is used in many languages to assess how many words children know; it’s typically filled out by the kid’s parents and to record which words the parents think their kids understand and say.
• “online” processes: Cognitive processes that aid the listener in processing unfolding language live (in real time).
• spontaneous speech: Speech that is not pre-rehearsed and/or produced in the moment.
• (un)attested: Attested = observed; unattested = not observed.
• usage frequency: How often something is used in everyday speech.

Back to main page / Back to top

Goldin-Meadow (2020)

• Bayesian statistical model: A statistical approach used sometimes for analysis and sometimes for simulation that generates inferential conclusions using Bayes’ Rule.
• (cognitive) biases: Tendencies or predispositions that lead us to be more likely to come to some conclusions over others when faced with ambiguous information. Think of these like shortcuts to the ‘right’ answer for learners.
• cognitively basic: used in the paper to imply a cognitive bias (see above)
• co-speech gesture: Gestures produced in the course of speaking.
• consistent word order/canonical word order: word order has to do with the default ways in which words are strung together to form sentences. some languages, like English, are pretty strict (consistent) about their word order. other languages, like Turkish, are not strict at all—and it’s still easy to follow meaning because Turkish has other cues to tell you who did what to whom in any given sentence (English lacks these other cues!). Canonical word order is another way of saying ‘default’ word order for a language.
• conventional language model: Sufficient input from a fluent user of a language for a learner to acquire the core representations and processes of that language.
• depictive forms: Representations of objects that are designed to remind the addressee of the perceptual properties of the referent (e.g., referring to a tree by stretching out the arms and fingers above one’s head so they look like brances)
• deictic gestures/pointing gestures: Deictic gestures canonically highlight a referent in the here and now (e.g., pointing to my cat while referring to her).
• determiner-noun construction: I’m going to restrict this to English to keep it simple… determiners in English tyipcally refer to the/a/this/that/these/those and numerals. You hopefully know what nouns are! A “construction” is a linguistic unit made by combining two or more components (often considered a type of syntax). So a det-noun construction would be like: the-ball, this-shoe, two-apples.
• established sign language: A sign language that is stably used among a community of fluent signers.
• handshape versus motion forms: handshape refers to a phonological feature of signs: the ways in which hand is arranged to create a recognizable shape that can be moved and/or placed in different ways to produce different meanings. Motion refers to another phonological feature: the ways in which articulators are moved in conventional ways to produce different meanings (e.g., the same handshape with motion A means something different from motion B).
• hierarchy/hierarchical structure: structure whereby some elements are grouped as sub-parts of higher-level elements (like branches on a tree).
• homesigns: Semi-conventionalized gestures used to communicate between deaf children and their non-deaf, non-signing family members.
• iconic gestures: See depictive forms above; gestures that seek to convey the perceptual features of a referent (e.g., referring to a tree by holding out the hand in a splayed high-five shape with the elbow flexed so the fingers point toward the sky in front of the gesturer).
• internal processes: In this work, processes that occur mentally—in the head of the speaker/addressee/learner.
• language-learning models (simulations): Computational models used to examine theoretical assumptions and hypotheses in fully controlled circumstances.
• language emergence: the process by which new languages come to arise; this can and has happened for a variety of reasons; this paper concerns sign language emergence in deaf communities without access to a pre-existing sign language.
• multigesture nominals: noun expressions created by the combination of multiple gestures.
• narrowly referential: picking out specific aspects of a scene to convey to an addressee (contrasted with full depictive coverage of a scene).
• number marking: A kind of inflection indicating the plurality or number class of a referent.
• ontongenetic versus historical time: another way of contrasting individual learning and development (i.e., in the human lifespan) versus language evolution and change (i.e., over many generations).
• (O)VS word order: a default word order in which objects come first, then verbs, then subjects. an example would be apple-eats-Susan.
• productivity: an indication of how ‘abstract’ children’s linguistic knowledge is. if they have learned a generalized pattern (e.g., is VERB-ing) or an abstraction (determiner-noun) then they can produce novel versions of those structures (e.g., is BLICKETing and the BLICKET).
• resilient properties of language: Features of linguistic structure that emerge in new languages even if there is no conventional language model from which learners can directly observe those features.
• silent gesture: gesture produced in the absence of speech.
• subject (syntax): the noun phrase that controls a verb - often (but not always) the agent of a clause.
• symmetrical versus reciprocal relations: contrasts actions that are construed as joint endeavors (e.g., high-fiving) versus simultaneous actions that are not joint but may perceptually resemble symmetrical relations (e.g., punching each other).
• thematic roles: various actors and objects involved in events. an agent is the doer of an action. a patient is the recipient of an action, etc.
• transmitting language: the transmission of language happens when new generations of learners encounter a pre-existing language system and then learn it sufficiently to become fluent users themselves.

Back to main page / Back to top

He & Arunachalam (2017)

• causative: verbs that describe an action brought about by an agent; classically contrasted with non-causative verbs, with this difference encoded as transitive (causative)-intransitive (non-causative) pairs. for example: fall versus drop (“I dropped it” is okay but “I falled it” is not). English-speaking children use non-causative verbs causatively quite often. Melissa Bowerman has some really nice examples of this (“giggle me” = “make me giggle”) if you want to learn more.
• communicative intent: the meaning that a speaker wants to convey to their addressee (regardless of what linguistic form they actually use or their success in communicating it)
• core knowledge: inborn knowledge about how the world is structured; if you’re very interested, check out this short paper
• extension (of word meaning): broadening a referent’s category to include a variety of members (e.g., unseen cases, non-prototypical cases, etc.)
• free and bound morphology: morphemes are the smallest units of meaning in a language. free morphemes tend to have word-like content, and can stand alone in the language (e.g., “dog”, “walk”, “here”). bound morphemes have meaning, but can’t stand on their own—they must be bound onto a free morpheme, like -ing and -s in “walking” and “walks”. Sometimes these bound morphemes are called “function” morphemes because they do inflection. This meaning of “function” is also sometimes extended to free morphemes whose work is primarily grammatica (e.g., “the” and “from”).
• heuristic: in cognitive science, a shortcut principle for reasoning about something/making an inference about experiences
• hypothesis space: in cognitive science, the possible inferences a learner might draw, given data about the world. the word learning literature, this most often refers to the space of possible label-referent mappings.
• ideal learners: this refers to the computational modeling literature, in which learners are simulated. the IDEAL learner is typically a rational learner that uses some proposed constraint in learning from in input in a predictable way.
• LENA: a recording device with accompanying software that allows researchers to record children’s home audio environments for up to 16 hours on a given day and to automatically analyze several aspects of the audio file.
• lexical constraint: some bias, mechanism, or heuristic that makes a learner more likely to entertain some mapping hypotheses over others (e.g., the whole object assumption)
• open versus closed class: “open class” refers to categories of words (e.g., nouns and verbs) that readily add new members. for example, new technologies are a constant source of new nouns and verbs, and we have a number of processes for innovating new nouns and verbs in English (e.g., thinking of our course name, “to BBC something” or to “pull a BBC”). “closed class” refers to categories of words (e.g., number words, determiners) that have fixed vocabularies to which new words are rarely added (if so, only in historical time).
• (overt) arguments: in this paper they are talking about verb arguments, such that arguments are the things that verbs link together. like how a transitive verb links together a subject and an object. overt arguments are arguments that are explicitly included in the sentence “we eat apples” and dropped (aka elided) arguments are implied but not explicitly stated like “eat apples”. English generally requires subjects, so “eat apples” is not a good way to say “we eat apples”, but lots of languages allow this kind of argument dropping, like Spanish: “comemos manzanas”.
• phonological form: the “label” to which a referent should be mapped
• productive/expressive versus receptive vocabulary: productive (aka expressive) vocabulary is the set of words children are known to produce (now or previously), while receptive vocabulary is the set of words children appear to understand, even if they do not yet produce them themselves.
• prosody: approximately, the rhythm and intonation of language
• scaffolding: in developmental psychology, guidance from a more knowledgeable other that is designed to help the recipient demonstrate their potential ability. typically exceeds their “actual” ability. see vygotsky!
• sentential complement: basically, a sentence that tags onto another sentence as a sub-structure. you can keep it simple for English by thinking of it as full sentences within sentences that start with that, e.g., “she wanted to know that they were actually using the glossary”. the “they were using the glossary” bit is the sentential complement. note that it contains its own inflected verb + and arguments (e.g., subject, object).
• speaker’s reliability: in this context, the authors are referring to experiments in which speakers have a history of being ‘unreliable’ (e.g., calling a ‘shoe’ by the wrong name) versus ‘reliable’ (calling it by its right name: ‘shoe’). children know to trust reliable speakers over unreliable ones.
• taxonomic level: the level at which a concept is sorted. in this framework, concepts are interrelated within hierarchies of types: animal > mammal > dog > Dalmatian. for each of these taxonomic hierarchies, there is a kind of intuitive “basic” taxonomic level (in this case, “dog”), that has a super-categories (“animal”, “mammal”), and sub-categories (“Dalmation”, “Fido”)
• transitive versus intransitive: transitive verbs take a subject and an object (subject-“unlock”-object) while intransitive verbs only take a subject (subject-“sleep”).

Back to main page / Back to top

Frank et al. (2020)

• administration (of a task): how an instrument or task is carried out by the person running it.
• caregiver: a person, typically an adult, who is responsible for a child’s wellbeing
• instrument (psychometric): a tool and/or method used to conduct a test, in this case a parental questionnaire.
• morphologically complex: words with a lot of bound morphology (see “free and bound morphology” above)
• nominals: another word for nouns
• partition (concepts): think of a conceptual space (e.g., actions) as being cut into pieces and assigned labels (e.g., walking versus running). this process of dividing up the perceptual space is sometimes referred to as ‘partitioning’.
• predicates: predicates are words that link concepts together. in this paper we’re only concerned about verbs (link nouns), adjectives (modify nouns), and adverbs (modify verbs).
• sentence-medial: not at the start of the sentence! not at the end! but… in the middle :)
• WG and WS: two versions of the CDI (see “MacArthur-Bates Communicative Development Inventory (“CDI”)” above). WG = Words and Gestures—this is the version for younger kids who don’t talk much. WS = Words and Sentences—this is the version for older kids who are definitely talking. Kids usually hit ceiling (i.e., know all the words) on the WS version around age 2.5 years.

Back to main page / Back to top

Gergely et al. (2002)

• emulation: goal-directed action that relies on a pre-existing set of behaviors the organism already has command of; a kind of semi-flexible problem solving. If infants emulated in the task, they would simply turn on the light with their hands.
• goal-directed action: actions planned and executed as a means to an end (i.e., to get something done).
• imitative learning: a flexible type of learning by which the learner reproduces the behavior of someone else; a type of social learning. In this paper, infants’ imitative learning implies that they think there is some advantage/reason for turning the light on with their head.
• principle of rational action: the assumption that actions are planned so as to achieve goals in an efficient or effective manner.

Back to main page / Back to top

Herrmann et al. (2007)

• cladistic: relating groups of organisms by their evolutionary ancestry (think of ’trees’ of evolution that you’ve seen.
• counterbalancing: dividing some (scientifically uninteresting) feature of stimuli across participants, conditions, or testing groups so that any differences found are not attributable to the uninteresting thing. For example, if you have a battery of tasks (Task A and Task B), you want to counterbalance the order of the tasks, so that one half of your participants do A then B and one half do B than A; in this case you are trying to make sure that any conclusions you draw don’t depend on a specific ordering of the tasks.
• ecological intelligence hypothesis: humans got their smarts from the evolotionary advantages gained by becoming better foragers.
• “live” coding: noting participant responses in real time during an experimental or observational study.
• meta-analysis: a scientific framework that involves a quantitative analysis of prior empirical findings. so basically the scientist comprehensively gathers prior experimental articles and extracts quantitative measures of the experimental findings from each. then they quantitatively analyze the literature as a whole to infer how reliable some experimental effect is, given all available past work. the scientist can also sometimes use these data to reason about what causes an experimental effect to show up more strongly or more weakly. fun fact: a meta-analysis of meta-analyses is called a mega-analysis :)
• mediators: some secondary explanatory factor that changes how we understand the relationship between a primary explanatory factor and an outcome. for example, in the US, more parental talk to 1.5 year-old kids is associated with larger child vocabulary sizes a half year later, but there’s a twist! child-directed speech also predicts how quickly children recognize the referents of familiar words. if we make a statistical model that lets child-directed speech predict recognition speed, and then let recognition speed predict vocabulary, we have a mediation model (i.e., here recognition speed is mediating the relationship between child-directed speech and vocabulary size). in fact, in this case, once recognition speed is taken into account, child-directed speech has nothing else to explain about vocabulary size. (see Weisleder & Fernald, 2013 for more on this specific example)
• ontogeny (vs. phylogeny): individual development of an organism (vs. evolutionary change of a group of organism).
• social intelligence hypothesis: humans got their smarts from the evolotionary advantages gained by cooperating on complex tasks.
• socio-cultural cognition: social cognition includes the mental processes and representations dedicated to understanding and interacting with others. socio-cultural cognition implies that this social cognition is sophisticated enough to support the evolution and maintenance of ‘culture’, which involves the accumulation of relatively stable (and inter-generationally transmitted) knowledge systems that include social customs and social institutions (e.g., language, political systems, cultural artifacts)
• spatial language: linguistic phenomena relating to locating objects, relations, and trajectories in space. most commonly refers to spatial deixis (English: here/there and this/that) or spatial relations (English: in/on, left/right, north/south). languages vary quite a bit in how they encode spatial relations!
• spatial orientation: how objects are arranged, as it relates to each other (left-right), the arranger (my left/my right), or some external reference (west-east).
• theory of mind: we could spend a whole class on this, but generally refers to the ability we have to consider others’ psychological states (i.e., what they know and what they’re thinking about). you can consider this a kind of ‘mindreading’. children gradually become able to model others’ mental states, eventually becoming able to recognize that others can hold beliefs about the world that contradict their own, certain, beliefs.

Back to main page / Back to top

Kail (1991)

• cognitive activities: mental tasks to which cognitive resources must be allocated.
• cognitive resources: mental capacities for perceiving, processing, storing, and retrieving information. Kail refers to “processing resources”—in his case, processing speed, and “attentional resources” which focuses more on the person’s capacity for allocating attention on tasks (e.g., selectively, for a sustained period, etc.).
• cross-sectional: a study technique that examines change in age (or time) by recruiting different groups of participants for each age-point (or timepoint) sampled.
• cycle time: Kail draws on analogy from computer science here; think of it sort of like the basic unit of information processign as slower (longer cycles) or faster (shorter cycles); as kids get older, they appear to have shorter cycle time, such that they can keep up with responses to more and more rapidly unfolding stimuli.
• declarative knowledge: knowledge that can be consciously experienced or examined (e.g., knowledge you can tell to yourself/others).
• developmental function: “function” here is used in the mathematical sense; Kail is interested in creating a mathematical model of developmental chance in processing speed. His models are expressed as equations that fit observed data with minimal error. To achieve that, his models have some parameters (variables) that he can input to get a more accurate fit to his data. Late in the paper he also mentions a “step function”, which looks like this. You can read a little more about continuous change (as opposed to step functions) in this blogpost.
• dual-task: when a participant is asked to engage in two distinct, simultaneous cognitive tasks (e.g., remembering a number and listening for a word in a speech stream).
• empirical journals: Academic journals that publish empirical studies (like this one)! AKA the journals that publish all the “JA” articles on our syllabus.
• group mean: the average of a score across a group of participants.
• information processing: mechanisms for analyzing and storing information (e.g., as received through sensory channels, like speech perception) or as created by internal processes (e.g., conscious reasoning)
• instance theory: the idea that practice effects (i.e., increased speed with practice) happens because participants start up with slow, algorithmic approaches (stepping through a problem step by step) and eventually shift to semi-automatic approaches (from diret access of a preformulated appropriate response).
• longitudinal: a study technique that examines change in age (or time) by recruiting a groups of participant and observing them repeatedly over age (or time).
• negatively decelerate: slowing down more quickly at first, and then continuing to slow down, but more slowly later on.
• null hypothesis: a statistical term that, here, specifically implicates the situation in which a researcher’s theory predicts NO difference between two conditions. The problem is that, with traditional frequentist statistics (e.g., p-values), the lack of statistical evidence for difference (i.e., a null outcome) is NOT satisfactory evidence for similarity. So Kail is saying that he’s putting himself in a tough situation by predicting the null hypothesis. Contemporary researchers in the psychological sciences now occasionally use Bayesian statistics to get evidence for similarity rather than evidence for difference.
• positive transfer: when practice or improvement in one task benefits performance in another task, typically via cognitive processes that are shared between tasks.
• R^2^ (R-squared): a measure that tells you how well a statistical or computational model fits observed data; that’s why it’s often referred to as measuring “explained variance”—what’s leftover is the observations that don’t fit the model’s predictions. Basically, higher R-squared is usually better. Read more and see examples here!
• Raven’s Progressive Matrices: an IQ test with a wide age range of potential test takers (summary here)
• recall (free vs. cued): retrieving some information with no support, versus with some kind of memory cue as support.
• Wechsler Intelligence Scale for Children: IQ test for kids (summary here).

Back to main page / Back to top

Vlach & Johnson (2013)

• eye-tracker: device that records participants eye movements in real time while they observe stimuli.
• attention getter: trials that serve to get the participant’s attention back to an optimal location before a target trial begins. In this experiment, the attention getters not only gave kids a ‘break’ from the main stimuli, but also helped to re-center gaze on the screen before the target-trial LEFT-RIGHT objects appeared.
• block (experimental): a preformulated set of trials to which participants are exposed during an experiment (e.g., 4 blocks of 10 trials each).
• dwell time: the time a participant spent fixating on a region of interest (e.g., on a computer monitor).
• correct for multiple comparisons: when an analysis requires lots of statistical tests, the researcher can avoid spurious positive statistical outcomes by proportionally lowering their significance threshold. this is known as “correcting” for multiple comparisons, typically using the Bonferroni method.
• selective attention: the ability to foreground one’s attention to some specific aspect of incoming information.
• spacing effect: the meta-analytically supported finding that—at least in adults—learning outcomes tend to improve when recall of information is spread out over time, rather than clustered together in time.

Back to main page / Back to top

Kuhl et al. (2003)

• alveolo-palatal fricative/affricate: you do NOT need to know this for any test. this is a phonetic description of two kinds of consonants that are native and commonly used in Mandarin. alveolo-palatal describes where the tongue touches the roof of the mouth (a bit higher up the palate, after the ‘bump’ behind the teeth). fricative means the tongue is open JUST enough to create a kind of fuzzy staticky sound. in contast to fricative, affricate means that the tongue starts out all the way closed and then, subsequently, is opened up to fricative position. an example of a palatal fricative-affricate minimal pair in English is sh-ch.
• frication: you do NOT need to know this for any test. this just describes the key characteristic of a fricative (which is defined under alveolo-palatal fricative/affricate)
• fundamental frequency: you do NOT need to know this for any test. the lowest component of a complex wave… but unless you’re really into phonetics or physics, this is always going to refer to something simpler: pitch (i.e., note). so you know what high and low pitch sound like if you’re talking? that’s also high and low fundamental frequency. sometimes noted as “F0”.
• head turn (HT) conditioning procedure: when we learned about speech perception before (e.g., with Maye) we heard about preferential looking. head turn is very different, but is used to achieve the same thing: finding out what infants can and can’t discriminate between. the HT procedure requires a pretty long and involved training period (~15 min) where you gradually train infants to look to one side if they detect a change in an auditory signal. you do that by having something really attractive (e.g., a moving toy) that only appears on the infant’s side just after a change in the signal happens. after a little while, the infant learns the association and begins to look anticipatorily at the toy BEFORE it actually appears. once you’ve solidly trained them to turn their head anticipatorily after hearing change, you can play the target stimuli to see if they turn their heads when they hear the sound you’re interested in. HT has a larger infant drop-out rate than other methods, because it’s arduous! however, meta-analyses show that the discrimination data you get from this method can be clearer to interpret.
• neural commitment: (permanent) entrenchment of neural pathways that are optimized for responding to stimuli. So in Kuhl’s case, this would be neural pathways for recognizing different phonetic categories in spoken language; she’s arguing that some neural pathways become permanently expert at processing the listener’s native sound system, and as a trade off, the listener loses some flexibility in their ability to perceive differences among non-native sounds.
• neural readiness: Kuhl is arguing that, around the time of perceptual narrowing (8–10mo), infants’ minds are maturationally in a state that is highly sensitive to environmental information, so as to effectively make neural commitments to the relevant sounds in their linguistic environment. While infants at this age may appear to be “committing” behaviorally, she argues that her work shows that they are still highly responsive to social language use, which rapidly opens their perceptual ability to non-native sound contrasts.
• sensitivity: in Kuhl’s case, this refers to the listener’s ability to detect differences between sounds. Thus “sensitivity” here means “discrimination” as we saw it in the Saffran and Maye work earlier.
• sound-treated room: a room for running experiments and/or for recording audio stimuli. the rooms are designed for situations in which background noise might hinder the quality of science being conducted. the rooms are “sound treated” by various means, but often some amount of sound-diffusing insulation material and air-tight chambers separating the room from other parts of the building. in this case, Kuhl exposes the English-speaking infants to Mandarin in a sound-treated room, therefore minimizing interferring background noises that might distract or overlap with the Mandarin speakers’ voices.
• spectrogram: you do NOT need to know this for any test. this is a way of analyzing the energy profile of an acoustic signal (here, speech). you can see an example in Fig 1. the darker the point on the figure, the more acoustic energy the speech has at that point in time. some sounds—like fricatives—have a kind of ‘raincloud’ looking energy, with lots of black-to-gray energy spread across a wide band of the y-axis (i.e., the upper left-hand raincloud forms of panels C and D in figure 1). other sounds, like vowels, show distinct horizontal bands of energy (these horizontal energy bands on vowels are called “formants”), with bands of low energy between formants. we can primarily hear the difference between vowels thanks to changes in the position of these formants from one vowel type to another (e.g., see formants for the vowels in “see” [i], “Sue” [u], and “saw” [a] here).
• steady-state vowel formant frequencies: you do NOT need to know this for any test. FORMANT is defined under spectrogram. so the additional information here is “steady state”, which is just referring to the portion of the vowel where the horizontal lines are staying pretty constant in their frequency (i.e., while they’re flat). this is often in the middle of the vowel.
• visual attention: in this study, “visual attention” refers to a coder’s perception of how often the infant participant is looking at the speaker in the exposure sessions.

Back to main page / Back to top

Smith et al. (2018)

• bootstrapping: from a learning standpoint, using something you know to learn something you don’t know. here posed as a process by which children might take advantage of skewed distributions in their visual input. the example given is that children can more effectively segment objects (and thus potential referents!) from their environment by having a grasp on what is NOT being referred to (the familiar thing, for which they may already know the name).
• consistency: reliance on repeated exposure to some items (likely within similar, repeated contexts). here posed as a process by which children might take advantage of skewed distributions in their visual input. smith et al. point out here that consistent datasets change with infant age, calling them “small lessons”.
• desirable difficulty: aspects of the input that make it harder for the learner to understand what is happening in the short term, but result in more robust learning in the long term. here posed as a process by which children might take advantage of skewed distributions in their visual input.
• egocentric: in this work, ’egocentric’ refers to the location of recording sensors on the infant, which gives the researcher a ‘child-centered’ view onto the world.
• infant sensors: recording devices that the infant can wear, which give the researchers data on what the child is perceptually experiencing as they navigate their environment from a first-person perspective. this can include visual data (as in smith), audio data, movement data, heart rate, and more.
• ordered datasets: here smith refers to training data, specifically data that is fed to a learner in such an order that learning is (a) better suited to the child’s maturational state, (b) makes the learning task easier or more robust, or (c) all of the above. the relevant example from smith’s study is the faces before hands content of infants’ visual experiences.
• sensitive period: kind of like a critical period, but typically implies a more gradual onset and offset of the sensitive window, possibly a much longer overall window of sensitivity, and a less fundamental effect on the related innate capacity (e.g., something that can still be supported in learning by supplementary means).
• skewed distribution: think of a bell curve—that’s a “normal” distribution. a skewed distribution has most of its mass on only one side (e.g., the left or the right), leaving a long tail on the other side. see an example here.
• “starting small” hypothesis: this has to do with the idea that infants’ immature processing capacity is possibly helpful for learning, such that they begin learning by picking out only the most unmissable, frequent statistical patterns in their environment, forming a reliable and useful foundation for learning more nuanced statistical patterns (once their general processing abilities mature further). smith points out that this hypothesis is typically used to point to similar statistical learning capacity and similar inputs in adults and infants, with only differences in general processing; she is arguing instead that the input changes dramatically with development.
• timepoint: in a developmental study, the ages at which data are sampled. for a study of infants’ mobility in the first year, we might propose monthly samples, which would create 12 timepoints (one for each month) in our analysis.

Back to main page / Back to top

Liszkowski et al. (2012)

• conventionalized: in language, the process by which some behavior gains a shared, recognized, and used meaning among a community of language users. words are one type of of linguistic convention: English speakers agree that “dog” refers to dogs (e.g., and not all mammals)
• cooperative (pragmatics): there’s at least one citation of “cooperative” here that refers indireclty to the Cooperative Principle, which (broadly) is the idea that communication is only possible if the interactants can assume that the other is producing their communicative signals in a way that is intended to be understood by the addressee in the context in which it is produced. Successful communication is proposed to rely on interactants’ trust that they are all following the cooperative principle; even during conflict (like in a verbal disput), interactants depend on cooperation and coordination to engage in organized interaction.
• deictic/deixis: communicative forms (often words or gestures) whose referential meaning depends on context. so “here” is an example of a deictic word—it can refer to literally anywhere, and thus can only be interpreted meaningfully in context.
• dyadic: involving two people.
• elicitation: any task whose primary purpose is to get participants to produce linguistic or communicative behavior. in this case, Liszkowski and colleagues are trying to elicit spontaneous communicative behavior.
• industrialized societies: societies that rely on mass production and technology; often used for post-industrial society which more emphasizes economic production on the basis of services and information creation.
• language socialization: this is a field of study AND a theoretical approach. it examines how children learn to become social agents in their community in part by learning language, and ALSO how children learn language in part by learning to become social agents.
• modality: in language, the production system being used to create communicative signals. for example, contrast the “gestural” modality" with the “spoken” modality. sometimes in the readings, signals are referred to as “multimodal”, which simplt means that more than one modality is being used to create the communicative signal.
• morphology (gesture): here, the FORM of the gesture.
• perspective taking: in language, refers to the individual’s ability to attend to (and appropriately utilize) another person’s mental state in their language understanding, language production, or mental reasoning. expression choice often involves perspective taking; depending on whether they know her or not I might use “Edna” or “my daughter” to refer to my child in conversation.
• spatial cognition: a set of mental and linguistic capacities and processes related to tracking and relating objects in space.
• subsistence farming: refers to agriculturalists who farm the food and many of the other materials they consume.
• vocalization: a sound produced with the vocal folds; typically, but not necessarily to communicate something. a vocalization may be linguistic or not, e.g., “bababa” and a high-pitched squeal are both examples of a vocalization an infant might make; “hello baby” is also a vocalization (by a more linguistically mature speaker).

Back to main page / Back to top

Yu & Smith (2013)

• cross-recurrence analysis: a method for measuring pattern recurrences across time-organized data. it can be used to study the temporal coordination of behavior between two individuals. here, cross-recurrence analyses allow the researchers to reason about how reliably the child and parent’s coordinated visual fixations occur across time.
• gaze following: as covered in lecture, the behavior in which a person (e.g., a child) observes the gaze of another (e.g., a parent) and then follows the target of that person’s gaze when it shifts laterally (e.g., the child looking off to the side where the parent is looking).
• regions of interest (ROI): the portion of a visual scene in which the researcher is interested to know whether the participant fixates or not (and for how long). in the Yu & Smith study, the ROIs are the three toys on the table betwen the child and the parent, as well as the face of each interactant.
• (triadic) joint attention: as covered in lecture, the attentional state in which two individuals are attending to the same, third, object of attention, AND are aware that the other is aware that they are attending to the same object of attention.

Back to main page / Back to top

Gaskins (2006)

• interactional taboo: taboos are typically behaviors to be strenuously avoided (often associated with some supernatural consequence). when she uses this term in the paper, Gaskins is referring back to the case of eye contact avoidance by Gusii moms to their infants.

(All the other terms I found are common or already defined above!)

Back to main page / Back to top

Casillas et al. (2020)

• canonical babbling: production of syllables that have a “CV” format—a consonant (C) followed by a vowel (V). canonical babble is considered “speech-like” babble, and its consistent use indicates that the infant has the basic motor skills needed to begin producing first words.
• daylong recording: this is a non-literal term, because the vast majority of daylong recordings are <24 hours. but typically it refers to a type of recording that aims to continuously capture most of a waking day. the most common tool for making daylong recordings is a system called LENA, which is described in an earlier glossary entry.
• ethnolinguistic: characterized by not just a shared language but also a shared membership in an ethnic group.
• input-consolidation cycle: a learning mechanism proposed in Casillas et al. suggesting that infants learn efficiently from sporadic bouts of language input by then falling asleep, facilitating consolidation of the short information bout.
• language delay: a developmental pattern in which an child’s language skills appear significantly less mature than their same-aged peers. “language delay” is a description of this apparent difference in language skill, and does not identify a cause for the difference.
• lexical diversity: a measure of the variety of words used in a span of speech. a three word sentence might have low lexical diversity (“dog dog dog”; 1 word type/3 words total) or high lexical diversity (“dog cat cow”; 3 word types/3 words total).
• non-child-centered: a term referring to the cultural tendency to not center children in mixed-age social interaction. in such communities, children are encouraged to attend to the interactions going on around them, including adult interactions, rather than expecting to be the center of social attention.
• patrilineal: pertaining to the father’s ancestral line; see also patrilocal above.
• polysynthetic (language): a language that tends to have more than one morpheme per word (e.g., by using inflections). Tseltal is mentioned as mildly polysynthetic, due to having lots of obligatory inflection types. Chintang and Inuktitut are a famous cases of high polysynthesis in the language development literature, in which a single word (with its inflections) can express what would take a sentence in English.
• TCDS/CDS/ODS: these acronyms stand for different potential sources of language input: Target Child Directed Speech (speech that is exclusively directed to the child being studied); Child Directed Speech (speech that is directed to any child); ODS (speech that is directed to anyone other than the child being studied).
• vocal maturity: a rating that indicates how linguistically mature a given infant vocalization is. the same infant can (and will) produce a variety of babble types of varying vocal maturity. from least to most mature, the types are: noncanonical babble, canonical babble, single-word vocalizations, and multi-word vocalizations.

Back to main page / Back to top

Rowe & Weisleder (2020)

• attunement: Also sometimes called “fine-tuning”, this refers to the process by which communicative or linguistic signals are tailored to capture the attention and comprehension of the recipient. For example, with a young child you might choose to use simpler syntactic structures and higher frequency words to increase the likelihood that the child understands what you mean—when you do so on the basis of your familiarity with that specific child or situation (e.g., this car-obsessed two-year-old knows “Ford F150”, so I won’t just say “truck”), that is fine-tuning.
• bioecological model: a model of development inspired by Bronfenbrenner that captures the nested, multi-layered nature of environmental effects on development. for example, this model distinguishes macro effects (e.g., at the global or national level) to micro ones (e.g., within the child’s own daily interactions) to more holistically understand complex processes in development (e.g., bilingual first language development).
• contingency (interaction): when one action is produced in a way that is influenced by a prior action, it is contingent. social contingency resembles interactional ’exchanges’ (e.g., taking turns, giving responses, etc.). the paper also mentions contingent feedback, which in this case refers to parents’ reactions to their infants’ vocalizations.
• cultural logics: culturally specific reasoning frameworks. in this case, the logics raised include “concerted cultivation” and “natural growth” from Lareau’s (2011) study of class and race in the US. “concerted cultivation” refers to a logic common in her middle-class families, whereby parents seek to control and augment children’s skill development to prepare them for professional life. “natural growth” refers to a logic more common in her working-class families, whereby children are thought to thrive if parents focus on giving love and basic care.
• decontextualized language: language that does not refer to objects and actions in the presently shared perceptual space (i.e, not in the “here-and-now”). this may, for example, include narratives (about what happened in the past or in an alternate world) or explanations (including information about abstract, or other non-visible entities).
• developmental cascade: we saw this phrase before, when looking at maternal responses to infants’ moving versus stationary bids. in a nutshell, developmental cascades describe the dynamic relationship between the developing individual and their environment: when a child learns to walk, that milestone reconfigures certain aspects of their experience (e.g., ability to hold objects while mobile, verbal responses from their parents) that initiate new developmental changes as a result.
• directive: a command, an imperative.
• forms of participation: this phrase is hinting at work that has examined how children are invited to participate in adult-centered interaction. for example, children may be encouraged to participate as onlookers or side participants, or as full-fledged interactional partners.
• individual differences: this phrase typically refers to studies that focus on person-by-person differences (in our case, child-by-child or family-by-family). so, for example, we could examine how variation in individuals’ working memory capacity relates to their massed versus interleave statistical learning performance.
• laboratory studies: studies conducted in a research lab, typically experimental ones (but not exclusively!)
• language variety: a distinct instance of a language, often associated with a social group; this concept overlaps quite a bit with “dialect”. for example, Chicano English and African American Language are both varieties of American English.
• maternal education: a measure of how many years of education a child’s mother has had; typically used as an index (or partial index) of socioeconomic status.
• mediated by: statistical term, indicating a secondary explanatory factor that critically explains variation; see mediators above.
• natural language corpora: collections (= corpora) of video/audio/transcriptions of spontaneous conversation or interaction.
• parentese: child-directed speech produced by parents; replaces “motherese” (mother-specific) and “babytalk” (which now has some pejorative usage).
• person-centered analysis: Rather than focusing on individual variables (e.g., SES), this kind of analysis seeks distinct groups of individuals that may share configurations of multiple variables, and analyzes variation as an outcome of those clustered characteristics. This approach is contrasted in this paper with variable-centered analysis.
• pragmatic inference: coming to a conclusion about a speaker’s intended meaning that is not directly evidenced in their utterance. this would include non-literal meaning and indirect meaning, e.g., understanding “it’s awfully noisy out there” as a request to shut a window.
• quality versus quantity (input): much research on children’s input discusses how MUCH input children get (the quantity) or how UNDERSTANDABLE the input might be (the quality). in principle, “quality” can be expanded much further to incorporate the CONTENTS of what children encounter in their input (and thereby what kinds of representations they might internalize in the process of learning).
• randomized control trial: a type of intervention study in which participants are randomly assigned to groups that either receive the target “treatment” or some “control” treatment to see whether the target treatment produces the desired outcomes. The random assignment ensures that any outcomes are not due to participant self-selection. For example, participants might be randomly assigned to a condition that encourages book reading (treatment) or a condition that encourages tooth brushing (control) and then later have their child’s vocabulary measured. The researchers might expect to see vocabulary changes specifically in the bookreading group, and if so they would be able to infer that any changes are not due to the behavior of parents who are already particularly interested in bookreading.
• routinized situations: everyday activities, or activity or situational contexts that are encountered frequently and in a highly distinct and predictable format.
• social cohesion: an indicator of a sense of community, solidarity, and deeply shared values and identity. typically relates to between-neighbor trust, use of non-family childcare, and collective socialization behaviors (i.e., socialization by non-family).
• socio-interactionist theory of language development: approaches to understanding child language development that anchor on meaningful, everyday activities and children’s interactions with others within those activities. we saw an example of this with Ratner and Bruner’s social games in class.
• sociocultural theories of development: approaches to understanding child development (not JUST language) that focus on children’s growth as encultured beings into the social roles made relevant to them by their developmental context.
• speech register: a distinct and conventionalized style of speaking that is used in a context-specific way. “Infant-directed speech” is often cited as a register, if it includes high and variable pitch, simplified words and syntax, and multimodal cues (voice, touch, gaze) to elicit attention. Formal or academic speech registers may also be familiar to you.
• spoken word processing: in this paper, this phrase simply refers to the time it takes for children to recognize the referent of a word by visually fixating on its picture in a looking-while-listening task.
• syntactic complexity: an index of how many parts (words and morphemes) and hierarchical nodes make up an utterance. more of these things typically means the utterance has higher syntactic complexity for example, ‘i appreciate that you read the glossary’ has two inflected main verbs (appreciate, read), and so has greater hierarchical complexity than a sentence of equal length with one main verb like, ’the orange cat prefers cheesy warm lasagna’.
• uptake: as discussed extensively in the He & Arunachalam paper, ‘uptake’ refers to the information from the environment that the learner actually notices and processes (i.e., just because the information is there does not necessarily mean it is attended to or learned about). uptake can be affected by cognitive biases, maturation, and environmental support.
• usage-based theory of language development: a theory of language development that emphasizes how children’s linguistic representations are built in a “bottom-up” way, from observing and making inferences about real language usage patterns in the child’s environment rather than in a “top-down” way, which would typically include innate linguistic knowledge. this theoretical perspective usually therefore emphasizes domain-general learning mechanisms and the role of environment—NOT language-specific mechanisms and the role of inborn knowledge.
• variable-centered analysis: A prototyipcal approach to analysis in psychology, which examines how individual variables (e.g., SES) relate to the outcome of interest (e.g., parent input syntactic complexity). This approach is contrasted in this paper with person-centered analysis.
• visual field: the span of visual information available to the perceiver—the whole space that a person can see when they look with both eyes at a single point in space. for humans this might be around 150+ horizontal degrees for each eye (maybe around200 degrees between both eyes) and around 120+ vertical degrees for each eye, with some individual variation.
• word specificity: in this case, “word specificity” is used to refer to the context-specificity of word use. the authors provide the example of location and word use in the Roy et al. Speechome data, so, like “ducky” and “shampoo” might appear almost exclusively when the kid is in the bathroom, while “coat” and “shoes” almost exclusively in the entryway, while other words (e.g., “beautiful”) are used more generally across a variety of places in the house. you can imagine a parallel case for activity: what words are particularly likely during mealtimes, bathtimes, departure times, etc..

Back to main page / Back to top