Adult Conversational Teaching Techniques

Adult Speech to Toddlers

The affect of a parent's behavior on her child's language acquisition varies with the age of the child (Masur, Flynn, & Eichorst, 2005). Around a child's first birthday, nonverbal behavior seems to influence her infant's vocabulary growth in a positive way. In contrast, maternal verbal behavior is more important for her child's vocabulary growth during the 13- to 17-month ages, especially her verbal responses to her child and her supportive directions. These changes reflect her child's increasing ability to comprehend and use verbal information. Intrusive verbal directions by the mother negatively influence vocabulary growth.

Throughout the first two years of life, parents talk with their children, label objects and events, and respond to their children's communication. It would be simplistic, however, to assume that a child just applies the labels heard to his or her pre-existing internal concepts. Meaning is also derived from the communication process (Levy & Nelson, 1994). Although words are constrained initially by the conversational context, they are later used more flexibly as a child encounters the word in other contexts and gradually modifies its meaning. Within the conversational context, parents aid acquisition by engaging in modeling, cueing, prompting, and responding behaviors that affect the linguistic behaviors of their children.

Modeling: Motherese (Parentese)

Children's speech occurs in conversation and generally serves to maintain the exchange. As noted previously, conversational behavior is well established by the time a child begins to speak. Almost from birth, a child encounters a facilitative verbal environment that enables him or her to participate as a conversational partner.

As a child's communication behaviors develop, its mother unconsciously modifies her own behaviors so that she requires more child participation. For example, the mother may not accept babbled responses once her child begins to use single words. Instead, she may respond to babbling with "What's that?"-a request for a restatement. In a second example, when the child follows her pointing, the mother may ask a question. Whether the child responds with a gesture or a smile, the mother will supply an answer. In a third example, once the child is able to vocalize, the mother "ups the ante" and withholds the names of objects or repeatedly asks the name until the child vocalizes. Then mother gives the label.

Word learning depends on the establishment of joint reference. Mothers are very effective at following their child's line of regard, then labeling the object of the child's attention. The more time allotted to such joint attending, the larger a child's vocabulary as a toddler (Akhtar, Dunham, & Dunham, 1991; Tomasello, Mannle, & Kruger, 1986). In short, a child is more likely to learn a symbol when focused on the referent as he or she is during joint attending (Tomasello & Farrar, 1986). As might be expected, mismatches between the focus of a child's attending and the adult's labeling of that focus occur frequently-approximately 30 to 50 percent of the time (Tomasello & Farrar, 1986; Tomasello & Mannle, 1985). When this occurs, the caregiver attempts to redirect the child both nonverbally and verbally (Baldwin, 1993). As a result, 18-month-old toddlers may learn some words from only one exposure.

Although first words express semantic knowledge and a child's conversational intent, they are learned within interactive contexts. Those structures modeled most frequently by mothers are most likely to be used by their children. Data from English and Modern Hebrew demonstrate that nearly all the utterances of young children mirror patterns used by their mothers.

Initially mothers provide object names, but within a short time they begin to request labels from children. By the middle of the second year, mothers are labeling and requesting at approximately equal rates, and dialog is fully established. This dialog becomes the framework for a new routine. The mother begins to shape the child's speech by distinguishing more sharply between acceptable and unacceptable responses. The child's verbalizations are often responses that fill specific slots within the dialog, such as answering a question. Within the dialog, the mother provides consistency that aids her toddler's learning. These consistencies include the amount of time devoted to dialog, the number of turns, the repetition rate, the rate of confirmation, and the probability of reciprocating (Bruner, 1978).

In addition, mothers make other speech modifications that, taken together, are called motherese (Newport, Gleitman, & Gleitman, 1977) or parentese. The characteristics of motherese are listed in the table below. Compared to adult-adult speech, motherese exhibits (a) greater pitch range, especially at the higher end; (b) lexical simplification characterized by the diminutive ("doggie") and syllable reduplication (consonant-verb syllable repetition); (c) shorter, less complex utterances; (d) less dysfluency; (e) more paraphrasing and repetition; (0 limited, concrete vocabulary and a restricted set of semantic relations; (g) more contextual support; and (h) more directives and questions.

As you know, mothers use a short utterances when conversing with their infants. They use even shorter, less adult utterances with toddlers. The lowering of a mother's MLU, beginning in the second half of a child's first year, is positively related to better receptive language skills by a child at 18 months of age, although there seems to be no measurable effect on expressive language (Murray et al., 1990). Mothers aid their baby's bootstrapping, mentioned previously, by maintaining semantic-syntactic consistency (Rondal & Cession, 1990). For example, in utterances addressed to children, mothers use agents as subjects almost exclusively. A mother's behavior makes it easier for her child to decipher the syntax of mother's utterances.

As her child's language matures, a mother's speech directed to a child likewise changes. Motherese seems well-tuned to the child's language level.

The amount of maternal speech, of partial repetitions of a child, of gestures accompanying speech, and of initiated statements commenting on her child's activity or eliciting attention vary with a child's overall language level. Thus, the dependence on nonlinguistic contextual cues, such as gestures, decreases with an increase in a child's linguistic abilities. These dynamic elements appear to be strongly related to a child's subsequent development. At age 2, the amount of shared attention and maternal gestures and relevant comments are positively correlated with a child's verbal learning a year later (Schmidt & Lawson, 2002).

Slow at first, the rate of both mother's and child's linguistic change increases with age. The length and complexity of a mother's utterances change most between 20 and 27 months, when her child's language changes most rapidly. In contrast, there seems to be little or no change in the structural complexity of motherese between 8 and 18 months. During this period there is also little corresponding change in the complexity of child speech, the changes consisting primarily of the addition of single words.

Mothers fine-tune their language input to their children based primarily on the children's comprehension level. Other factors that influence the level of a mother's language are the conversational situation, the content, and different conversational acts (Snow, 1986). Overall, adults will simplify their input if the child does not seem to comprehend.

The amount of parental labeling or naming in both English and French varies with the age and development of a child. A positive relationship exists between the amount of adult labeling and a child's subsequent vocabulary growth (Poulin-Dubois, Graham, & Sippola, 1995). Both gesturing and the use of noun labels in English decrease with development (Schmidt, 1996). Nouns are replaced by verbs describing the action performed by the object.

Undoubtedly, the influence of a child's characteristics on the interaction and on motherese also has an influence on the language input to which a child is exposed (Yoder & Kaiser, 1989). The toys that a child plays with also influence the amount and types of language produced by her or his parent (O'Brien & Nagle, 1987). In general, toys that encourage role play, such as dolls, elicit more language and a greater variety of language from parents.

If adults simplify their language in order to be understood, then these modifications must reflect cues coming from the child. Apparently, however, adults are not conscious of their modifications, nor are they consciously attempting to teach language. Adult-to-child speech seems to be modified in response to the amount of child feedback and participation. Not only is much of the speech addressed to a child adapted for the child's linguistic level, but speech not adapted is simply ignored or not processed by children (Snow, 1986). In other words, children play an active role in selecting the utterances to which they will attend. A lack of response is important, for it informs a parent there has been a breakdown in communication that, in turn, necessitates linguistic changes by that parent. Although the exact nature of child feedback is unknown, children seem to be the key to adult linguistic changes (Furrow & Nelson, 1984).

The pragmatic aspects of a mother's speech may be related to the talking style of her child. Referential children tend to name frequently while expressive children engage in more conversation. Mothers of referential children seem to use more descriptive words and fewer directives (Della Corte, Benedict, & Klein, 1983). In addition, these mothers use more utterances within a given situation than mothers of children with more expressive speech.

Nonlinguistic behaviors are also critical. In order to use motherese, it's necessary for an adult to see a child. In fact, maternal linguistic modifications are different when her child is absent.

Despite linguistic inadequacies, children can participate effectively because of their mothers, ability to maintain the conversation. The steady, rhythmic flow of the dialog depends on the structural similarity of a mother's and child's utterances and on the correspondence of a mother's speech to events in the environment. She enables her child to participate through her use of turn-passing devices. She does not use turn-grabbing or turn-keeping behaviors, such as "well ... :' "but ... :' or pause fillers.

Mothers maintain control, however, and the dialog is much less symmetrical than it may appear. They maintain the interaction by inferring their children's communication intentions, compensating for the children's communication failures, and providing feedback. After her child reaches age 2, the mother slowly relinquishes her control.

Within the interactional sequence, a mother analyzes, synthesizes, and abstracts language structures for her child. Through word substitutions, she aids her child's learning of language form. A sequence might be as follows:

CHILD: She running.

MOTHER: She's running fast. Oh, she's tired. Now she's running slowly. She's stopping. She's jumping slowly. Now she's jumping quickly.

Note how the mother uses the same forms repeatedly. As a result, her child is not a lone linguist attempting to learn the language code; much of analysis, synthesis, and abstraction is performed by the mother (Moerk, 1985).

Fathers and Other Caregivers. Despite the name motherese, these speech modifications are not limited to mothers. Fathers and other caregivers modify their speech in very similar ways. In fact, fathers seem to provide even more examples of simplified adult speech than mothers.

The range of vocabulary used by fathers and mothers with their young language-learning children is similar, but fathers use fewer common words (Ratner, 1988). In this way, fathers are more lexically demanding than mothers.

Although fathers make modifications similar to those of mothers, they are less successful in communicating with toddlers, as measured by the amount of communication breakdown (Tomasello, Conti-Ramsden, & Ewert, 1990). Fathers use more requests for clarification than mothers. In addition, the form of these requests is more nonspecific ("What?") than those of mothers ("You want what?"). Fathers also acknowledge their children's utterances less frequently ("Um-hm," "Yeah," "Okay"). In return, children tend to persist less in conversation with their fathers than with their mothers. It is possible that fathers serve as a bridge for their children between communication with the mother and with other adults. The child learns how to communicate with those less familiar with his or her style and manner.

Even children as young as 4 years of age make language and speech modifications when addressing younger language-learning children. Adult and peer language modifications differ somewhat. In general, peer speech to toddlers is less complex and shorter and contains more repetition than adult-to-toddler speech, although peers elicit fewer language responses than parents. Peer interaction may provide a "proving ground" where younger children can try new linguistic structures.

Children enrolled in daycare centers and preschools also encounter a variety of motherese that varies with the size of the group and the age of the children (Scopesi & Pellegrino, 1990). In general, the larger the group of children, the less individual adaptation by an adult. Larger groups force teachers to concentrate on keeping attention and control. While use of behavior and turn-taking control techniques by teachers results in little toddler language production, use of child-centered strategies, such as adopting a child's topics and waiting for child initiations, and interaction-promoting behaviors result in high levels of talkativeness by toddlers, especially labeling of entities in the environment (Girolametto & Weitzman, 2002; Girolametto, Weitzman, van Lieshout, & Duff, 2000). There are clear language-learning advantages for children attending preschool when the curriculum emphasizes language and literacy (Craig, Connor, & Washington, 2003).

The presence of older siblings may also influence the language a younger child hears and produces. For example, an older child will usually respond to more of a parent's questions, thereby reducing the number of responses made by a younger child. The younger child will often respond by imitating the older sibling. In this situation, the mother uses fewer rephrased questions, fewer questions with hints and answers, and fewer questions as expansions or extensions when the older child is present. (Expansions and extensions will be explained in more detail in the section on responding behaviors in this chapter.) In addition, the mother uses more direct repetitions of questions.

Summary. Parents who use a more conversational style with less direct instructing are more likely to have children who learn language more quickly. In other words, children benefit more from language input when parents are more concerned with understanding and participation and less so with teaching.

The exact effect of motherese on language acquisition is unknown. The modifications made by mothers may aid language acquisition by bringing maternal utterances into the "processing range" of a child. If nothing else, they increase a mother's chances of getting a response from her child. Since we find similar modifications in many cultures, we can assume that, at least, they somehow facilitate communication between adults and children.

The modifications of motherese seem to be maximally effective with the 18- to 21-month-old child. The child attends selectively, focusing on the best examples of various structures.

Prompting

Prompting includes any parental behaviors that require a child response. Three common types are fill-ins, elicited imitations, and questions. In fill-ins, the parent says "This is a .... " No response or an incorrect response from the child will usually result in additional prompts and recueing.

In elicited imitations, the parent cues with "Say X." Young language-learning children respond to slightly over half of the elicited imitations addressed to them.

Questions may be of the confirmational yes/no type, such as "Is this a bam" or of the wh- variety, such as "What's that?" or "Where's doggie?" Unanswered or incorrectly answered questions are usually reformulated by the adult. Approximately 20 to 50 percent of mothers' utterances to young language-learning children are questions. The individual range varies greatly.

In general, these three types of maternal language-teaching utterances have a shorter average length than the majority of the utterances addressed to the child. Maternal yes/no interrogatives, such as "Are you going home?" appear to correlate with child language-development gains in syntactic complexity, while intonational interrogatives, such as "You going home?" correlate with gains in a child's pragmatic ability. In contrast, maternal directives, such as "go get your coat," seem to correlate highly with child gains in utterance length and semantic-syntactic complexity, although they may slow vocabulary growth (Harris, Jones, Brookes, & Grant, 1986; Tomasello et al., 1986).

Parents employ an interesting technique to give their child an opportunity to produce two related single-word utterances (Schwartz et al., 1985). After a child produces a single-word utterance, his or her parent uses questions to aid the child in producing other elements of a longer utterance. The parent concludes by repeating the whole utterance. The following exchange is an example of this strategy:

CHILD: Daddy.

ADULT: Uh-huh. What's Daddy doing?

CHILD: Eat.

ADULT: Yeah, Daddy eat cookie.

Although prompting and cueing are effective teaching techniques, their exact effect on language development is unknown. Several studies have demonstrated their effectiveness with children with language disorders.

Responding Behaviors

Parents do not directly reinforce the syntactic correctness of toddler's utterances. In fact, less than 10 percent of children's utterances are followed by verbal approval. Generally, such reinforcement is given for truthfulness and politeness, not for the correctness of the syntax.

Feedback by parents, however, does follow their children's language production and varies with its correctness (Bohannon & Stanowicz, 1988; Demetras, Post, & Snow, 1986; Furrow, Baillie, McLaren, & Moore, 1993; Moerk, 1991; Penner, 1987). Imitation, topic changes, acknowledgments, or no response are more frequent following grammatically correct child utterances, while recasts or reformulations, expansions of the child's utterance, and requests for clarification are more likely following ungrammatical utterances. Different responses may signal a child as to the acceptability of the utterance. For example, Japanese mothers facilitate their infants' transition between sounds and words by repeating poorly formed child words correctly, thus signaling errors for the child and providing an alternative (Otomo, 2001). Let's look at some of the strategies used by English-speaking moms in the United States.

Let's assume that a 30-month-old says to you, "Gran'ma car, go zoo, 'morrow with Nuncle Juan." You might reply, "Yes, tomorrow Uncle Juan and you are going to the zoo in grandmother's car." What you just did is called a reformulation or a recast utterance. Your goal is not to teach but to understand the child. That said, what is the effect on the child?

As in the example, children's truncated or ungrammatical utterances can leave caregivers wondering what exactly a child means, so adults frequently check their own understanding against the child's meaning. An adult does this, as you did above, by reformulating the child's utterance into what the adult thinks the child meant to say. In the process, the adult locates the error or errors and embeds a correction. As a result, the child hears a more conventional form for expressing his or her meaning.

With preschoolers, adults reformulate more frequently than they imitate error-free utterances (Chouinard & Clark, 2003). As mentioned previously, imitation among both children and adults decreases markedly as the child passes from toddler to preschooler. In a similar fashion, reformulations decrease as a child passes through the preschool years (Chouinard & Clark, 2003).

We assume from their behavior that children understand reformulations to be corrections. For their part, children repeat the reformulation, acknowledge the correction with yeah or uh-huh and continue the conversation, or reject the reformulation because the adult has misunderstood the child's meaning. Reformulation is a great teaching tool because of its immediacy, timeliness, and the attending of the child.

Some responding behaviors seem to have reinforcing value. Approximately 30 percent of mothers' responses to 18- to 24-month-old children consist of expansions. An expansion is a more mature version of a child's utterance in which the word order is preserved. For example, if a child says "Mommy eat;' mother might respond with "Mommy is eating her lunch." The mother assumes that the child intends to communicate a certain meaning. As a child's average utterance length increases beyond two words, the number of expansions by the mother decreases. Approximately one-fifth of a 2-year-old's ill-formed utterances are expanded by the mother into syntactically more correct versions.

Children seem to perceive expansions as a cue to imitate. Nearly a third of adult expansions are in turn imitated by the child. These imitations are more likely to be linguistically corrrect than the child's original utterance. Let's see how it works:

CHILD: Block fall.

ADULT: Um-hm, blocks fall down.

CHILD: Block fall down.

It is believed that spontaneous productions follow, and rules are generalized to conversational use. As spontaneous production of structures occurs, imitation of these structures decreases. Expansion adds meaning to a child's utterance at a time when the child is attending to a topic he or she has established. In addition, expansion provides evaluative feedback. Expansions continue the topic of conversation and encourage a child to take his or her turn and, thus, to maintain the dialog.

The type of expansion used by the mother may have an effect on the particular form being learned (Farrar, 1990). For example, reformulating the child's previous utterance by adding, substituting, or moving a morpheme may aid learning of plurals and progressives (is eating) but has less effect on the past tense or the verb to be, which seem to benefit from removal of morphemes and restatement of correct forms.

Right now expansion and reformulation probably seem like the same thing. Let's sort it out. Expansions, used primarily with younger children, maintain the child's word order while providing a more mature form of the child's utterance. While both expansion and reformulation seek to preserve the child's meaning, reformulation is a strategy for older children who are beginning to create truly complicated sentences. Think of reformulation as the next step in caregiver teaching after expansion. Reformulations go beyond a mere expansion and can involve considerable rearrangement of the sentence elements while preserving the child's meaning as you did in the example.

Extension, a semantically related comment or reply on a topic established by a child, may be even more helpful. For example, when a child says "Doggie eat;' the partner replies, "Doggie is hungry." Thus, extension provides more semantic information. Its value lies in its conversational nature, which provides positive feedback, and in both its semantic and pragmatic contingency. A semantically contingent utterance is one that retains the focus or topic of the previous utterance. A pragmatically contingent utterance concurs with the intent of the previous utterance; that is, topics invite comments, questions invite answers, requests invite responses, and so on. In short, both types of contingency maintain the flow, which is inherently rewarding to almost all children.

Finally, parents imitate their child's speech. A little over 20 percent of these adult repetitions are imitated, in turn, by the child.

All three consequating behaviors—expansion, extension, and imitation—result in greater amounts of child imitation than adult topic initiation or nonimitative behaviors. Hence, expansion, extension, and imitation appear to be valuable language-teaching devices. Each reinforces a child verbal behavior, and expansion and extension also provide models of more mature language. Maternal extending correlates significantly with changes in the length of a child's utterances. The adult utterance is semantically contingent upon the preceding child utterance. This characteristic decreases the linguistic processing load on a child because the adult utterance is close to the child's utterance in form and content. Parents do not consciously devise these teaching strategies; rather, they evolve within the conversational context of child-caregiver interactions.

Characteristics of Motherese Compared to Adult-to-Adult Speech

Paralinguistic

Slower speech with longer pauses between utterances and after content words

Higher overall pitch; greater pitch range

Exaggerated intonation and stress

More varied loudness pattern

Fewer dysfluencies (one dysfluency per 1,000 words versus 4.5 per 1,000 for adult-adult)

Fewer words per minute

Lexical

More restricted vocabulary

Three times as much paraphrasing

More concrete reference to here and now

Semantic

More limited range of semantic functions

More contextual support

Syntactic

Fewer broken or run-on sentences

Shorter, less complex sentences (approximately 50% are single words or short declaratives)

More well-formed and intelligible sentences

Fewer complex utterances

More imperatives and questions (approximately 60% of utterances)

Conversational

Fewer utterances per conversation

More repetitions (approximately 16% of utterances are repeated within three turns)