At what age does a child begin to use reasonably accurate syntax using subject verb object?

Lang Speech Hear Serv Sch. 2020 Apr; 51(2): 317–328.

Abstract

Purpose

As is the case with children who rely on spoken language, speech-language pathologists must support and track the expressive language development of children with complex communication needs who use graphic symbols to communicate. This research note presents a framework of the progression of expressive English sentence development using graphic symbols and introduces possible approaches for measuring and analyzing graphic symbol use.

Method

Current issues in measuring graphic symbol utterances are explored, and a range of measures designed to analyze individual graphic symbol utterances as well as larger samples of utterances are presented.

Results

Both the Graphic Symbol Utterance and Sentence Development Framework and suggested measures are based on years of graphic symbol intervention research, including two large ongoing research studies of preschoolers with severe speech impairments. Our framework adapts the work of Hadley (2014) to depict expressive language progression from early symbol combinations to childlike and adultlike sentences and highlights developmental patterns unique to graphic symbol productions. Adaptations of existing measures (such as mean length of utterance) as well as measures unique to graphic symbol analyses are presented and discussed.

Conclusion

To accurately track changes in early graphic symbol utterance growth and complexity, a multidimensional approach, which includes analyses such as symbol relevance, word class diversity, and lexical diversity, is recommended.

Many preliterate children with augmentative and alternative communication (AAC) needs use graphic symbols to communicate; it is equally important for these children to learn to communicate in sentence format as it is for children who rely on speech or sign language. Toward this end, this research note presents a framework of the progression of graphic symbol utterance and sentence development and introduces possible approaches for measuring and analyzing graphic symbol use.

Some young children who require AAC have unintelligible speech secondary to a range of conditions, such as autism, cerebral palsy, or Down syndrome, while others have no reliable diagnosis beyond speech and/or language impairments and reduced speech intelligibility. Irrespective of the etiology of the children's speech and/or language impairments, this group of children shares a fundamental need to supplement natural speech to meet the full range of their communication needs. Graphic symbol communication can be a viable option for a wide range of preliterate individuals across a wide range of ages. These options can be used in low-tech format (e.g., pointing to graphic symbols) and within high-tech communication device options (e.g., using a communication app that “speaks” selected symbols).

The term graphic symbol refers to both photographs and line drawings. For the purposes of the current article, we also refer to symbols that use letters to represent words or grammatical markers such as the words is, am, and are; plural –s; or past tense –ed as graphic symbols. These symbols are, essentially, graphic symbols of orthographic words/morphemes, as children do not need to spell out these symbols letter by letter; instead, they select each symbol that contains the orthographic spelling. When used for communication, graphic symbols can be used in three main ways: (a) A single symbol can be used to represent a single word or morpheme, such as selecting a drawing of a dog to communicate the word DOG and selecting a symbol labeled with “–s” to represent the plural in DOG -S; (b) a single symbol can be used to represent a phrase or sentence, such as selecting a drawing of a dog to communicate the entire phrase I SEE A DOG; and (c) sequences of multimeaning symbols can be used to represent concepts via iconic encoding, such as selecting a sequence of line drawings to communicate a word or phrase. An example of this latter approach is to first select a drawing of a lamb—which represents all animals—and then to combine this symbol with another symbol to communicate the specific animal DOG versus combining the lamb symbol with a different unique symbol to represent the specific animal CAT. In the current manuscript, we will focus on the first option described above—that is, word- and morpheme-level graphic symbols, as opposed to symbols representing phrases or sentences. Conceptually, the approaches discussed in this research note also are highly relevant to iconic encoding approaches to graphic symbol use; however, these applications will not be specifically discussed in this research note as they have not yet been explored using the reported measures.

Building Sentences Using Graphic Symbols

Graphic symbols have been in popular use for communication purposes with children using AAC since the dawn of the AAC discipline as we know it. The research literature is replete with studies demonstrating that children can effectively use graphic symbols to meet basic expressive pragmatic and semantic needs, such as the syntactically simple task of requesting desired objects (Schlosser & Sigafoos, 2002). However, like children who primarily rely on spoken language, children who use aided AAC (i.e., communication tools or devices) must move beyond early pragmatic and semantic development and acquire grammatical skills in order to become sophisticated communicators. That is, even though graphic symbols may contain rich depictions that could represent more than one word, children who are to achieve linguistic competence must learn to construct sentences word by word and morpheme by morpheme. These constructed utterances must adhere to the linguistic rules of whatever language they would speak if they were able to rely solely on their natural speech for communication.

In children who are typically developing, first words appear at approximately 12 months of age. The onset of English syntax begins soon after, with word combinations typically appearing around 18 months of age. Additionally, many children begin to acquire bound grammatical morphemes, such as progressive –ing, plural –s, and possessive 's, before their second birthday (Brown, 1973). By the time children are 3 and 4 years old, children who are typically developing speak using complex sentences with embedded phrases and clauses—that is, well before entering kindergarten and far earlier than most children achieve functional literacy skills. Thus, for children with complex communication needs who are struggling with speech and language development, graphic symbol interventions focused on more than just pragmatic and semantic goals for functional use of one- and two-symbol utterances are critical considerations; that is, grammar-focused goals and interventions can help to ensure that children achieve their communicative potential, which in turn can increase the likelihood of appropriately challenging educational placements.

Typical spoken language developmental norms and trajectories can be used as models for aided language development and can help guide clinical decision making. These guidelines inherently help to ensure that clinicians promote all aspects of language development with children who have complex communication needs—not just pragmatics and semantics. Young children who are typically developing learn to use grammar early in development. This is not a lofty goal; rather, grammaticality is essential for clear communication and is foundational to attaining basic educational—and eventually employment—skills and success.

A growing body of evidence indicates that young children can use graphic symbols organized on grids to construct multisymbol utterances, including utterances that adhere to spoken word order (Binger, Kent-Walsh, King, & Mansfield, 2017; Binger, Kent-Walsh, King, Webb, & Buenviaje, 2017; Kent-Walsh et al., 2015; Tönsing et al., 2014; Wilkinson et al., 1994), prior to developmental stages when they could reasonably be expected to read and write. That is, the high-level literacy skills that would be required to orthographically produce the types of grammatical sentences that preschoolers use in their spoken language are simply not feasible for most preschoolers. Certainly, achieving full literacy skills is ultimately required to attain full linguistic competence, and pressing for early exposure to—and instruction in—early literacy skills is warranted for children who require AAC (e.g., Caron et al., 2018). At the same time, graphic symbol communication provides a crucial evidence-based, intermediary step that ultimately can help children with complex communication needs keep pace with the expressive language skills of their age-matched peers prior to developing fully functional literacy skills.

Assuming, then, that clinicians and researchers should target sentence building via graphic symbol communication, they also must measure progress. However, relying only on classic measures such as number of different words or mean length of utterance (MLU) may not account for differences inherent in learning to communicate via graphic symbols and fail to capture many of the intermediate steps along the path of developmental progress. Thus, accurately documenting graphic symbol development presents a critical and pressing need for both clinical and research purposes.

Novel Measurement Challenges in Graphic Symbol Communication

Graphic symbol communication presents a range of unique challenges for early communicators. One compelling example relates to the sheer volume of words and utterances that are relevant within early communication. Children who are typically developing use approximately 1,000 words and a wide range of grammatical markers by the time they reach their third birthday (Owens, 2016), with thousands more words soon to follow. Finding ways to effectively and efficiently store and retrieve such a corpus of vocabulary via graphic symbols has presented an ongoing challenge in the AAC discipline. Although a few researchers have examined this issue on a “micro” level—for example, examining accurate retrieval of a small corpus of words with limited learning opportunities (Drager et al., 2004)—systematic research in this realm is in its infancy, and there is no widely accepted standard of care; even AAC specialists may disagree on the “best” approach for providing a given client with access to a developmentally appropriate full range of vocabulary (Dietz et al., 2012). The implications of decisions on vocabulary provision are considerable given that available vocabulary on an AAC device, or lack thereof, can influence opportunities to develop grammar skills.

Another issue unique to aided AAC is message co-construction; that is, messages created by both the person using AAC and the communication partner using the AAC system over a series of turn exchanges, with the partner asking questions and providing expansions (Clarke et al., 2017). When this occurs, determining not only utterance boundaries but also what counts as an utterance presents challenges. Use of recasting as a potentially useful feature of adult scaffolding of child language acquisition has been explored in the AAC literature (e.g., Clarke et al., 2017); however, significant questions still remain as to how much children require feedback on their errors, the most effective types of feedback that can be provided, and what the role of child repair is in language learning through aided AAC. In the current article, the measures discussed are specific to independently produced utterances and, therefore, assume utterance boundaries have indeed been established.

One additional challenge that is not conceptually unique to aided AAC, although it takes a novel form, is determining the child's true intent—particularly for agrammatical utterances. For example, if the child selects a single symbol such as TREE, what is the intended pragmatic and linguistic meaning? Is the child commenting or labeling, or does the child have some other intent? Does the child mean to say, Tree, That tree is beautiful, I wish I could climb that tree, or something else? These issues also exist in early spoken language when children first begin to use and combine words; however, children who use graphic symbols may lack access to relevant vocabulary or instruction in how to move beyond the single-word stage of expressive language development. We cannot tell, then, if given the chance, the child using graphic symbols would produce a more grammatically complete utterance or not, thus clarifying communicative intent.

Regardless of the child's underlying intent, we argue that the clinical objective should be to teach the child to adhere to spoken language grammar in their productions, as this supports the long-term goal of achieving linguistic competence. The measures we propose, therefore, assume that this is the goal. Given this assumption, at least some of the common measures used to track spoken language development should, with minor adjustments as discussed below, apply to graphic symbol messages. Additionally, measures unique to building grammatical graphic symbol utterances also are warranted, given the challenges and differences across communication modes. These combined approaches should encourage clinicians to focus on sentence building and allow clinicians to track progress over time.

The Graphic Symbol Utterance and Sentence Development Framework

Figure 1 depicts our proposed Graphic Symbol Utterance and Sentence Development Framework. This framework incorporates components of Hadley's (2014) sentence-focused framework—namely, a progression from early symbol productions (Phase 1; typically one to two symbols in length) to early symbol combinations (Phase 2) to childlike sentences (Phase 3) to adultlike sentences (Phase 4). Notably, Phases 1 and 2 describe early utterances, which do not necessarily convey a complete thought or clear meaning, compared with Phases 3 and 4, which involve productions of true sentences, which are the basic unit of grammar (Hadley, 2014) and typically contain both a subject and a predicate (and therefore require a noun or pronoun as well a main verb). Consistent with Hadley's original framework, Phase 3 childlike sentences must contain explicit subjects and lexical verbs, but they also are characterized by omissions and errors in grammatical structure. In comparison, Phase 4 sentences more closely match adult sentence productions, although not all grammatical markers will necessarily be in place.

Graphic Symbol Utterance and Sentence Development Framework. NP = noun phrase; VP = verb phrase.

Figure 1 highlights the broad movement from pragmatics and semantics to also include syntax and grammar as language development progresses and incorporates challenges unique to learning to produce sentences with graphic symbols—such as early issues with selecting relevant symbols and placing those symbols in correct spoken word order. Figure 1 also highlights, via the arrow on the left, the constant, ongoing need to attend to both word class diversity (e.g., nouns vs. verbs vs. prepositions) and lexical diversity (i.e., within each word class, such as diversity of verbs). The Graphic Symbol Utterance and Sentence Development Framework was informed by years of graphic symbol intervention research with young children who have profound speech impairments (which we have typically defined as being less than 50% intelligible at the single-word level; e.g., Binger, Kent-Walsh, & King, 2017; Binger, Kent-Walsh, King, & Mansfield, 2017; Kent-Walsh et al., 2015), including two relatively large randomized controlled trials currently underway in our labs.

Additionally, we have developed two tables that highlight key clinical questions and measures to guide clinical decision making. Table 1 includes examples of suggested methods for analyzing individual utterances. Here, we examine two different target sentences across four different children, whose productions are illustrative of each of the four different phases from Figure 1. Table 2 includes analyses for these same four children that can be used to characterize a larger group of utterances, which most commonly would be collected via a language sample. Notably, actual target utterances are listed in both tables, which typically is not possible during language sampling—that is, clinicians typically are not trying to elicit highly specific structures during language sampling. We chose to include specific sentence targets here for two reasons: First, we often attempt to elicit these types of sentences in our research, and second, they work well for the descriptive purposes of this research note. We refer to Figure 1 and Tables 1 and ​2 throughout the remainder of this research note as we highlight various measures.

Table 1.

Suggested utterance-level analyses for two example target sentences across four children.

Table 2.

Suggested analyses to characterize a corpus of graphic symbol utterances. a

It is also important to note that most young children who can benefit from aided AAC also use speech to communicate to varying degrees, with many—depending on the nature and severity of their speech disorder—eventually transitioning away from aided AAC altogether. How the proposed graphic symbol framework accommodates the simultaneous development of speech is not well understood at the moment but is a subject of high interest that we plan to explore as part of our current randomized controlled trials that are underway.

Measuring Aided Utterance and Sentence Development

Communicative Intent

In typical development, communicative intents or functions—such as requesting, rejecting, obtaining attention, and commenting—emerge early in development, before children say their first words. Once children do begin to speak, they then map their words onto these existing intents. For example, a child will initially vocalize and reach out to a parent to gain attention, and in later stages of development, the child will map the word Daddy or Mama to this interaction. As numerous existing checklists, charts, and tables that allow clinicians to describe communicative intents exist in the literature, we refer readers to other resources to describe this component of communication development. These tools can be used to describe any communication modes, that is, graphic symbols, spoken words, manual signs, or nonverbal communication (Downing, 2005, pp. 39–41; Paul et al., 2018, p. 56).

Vocabulary and Meaning

Relevant Vocabulary

When children use graphic symbols to communicate, one of the most basic issues is ensuring that the vocabulary that the child selects is relevant to the context. When children speak or sign, they generally use relevant language, even if the underlying communicative intent is unclear, so vocabulary relevance is seldom a point of discussion. In contrast, aided language requires that children locate and select their words from a corpus of preselected (and, in the case of speech-generating devices, preprogrammed) symbols. A range of operational demands can impact a child's selection, which can lead to selecting irrelevant symbols. Thus, ensuring relevant vocabulary selection is included in Phase 1, at the earliest phase of symbol selection.

Selecting relevant symbols is a precursor to higher levels of expressive symbolic communication; as we see with Child A in Tables 1 and ​2, not even pragmatic intent is clear when the child fails to select relevant vocabulary. For example, the meaning behind Child A's utterance PIG BED when playing with the dogs and the bed is unclear, in part due to the inclusion of the seemingly irrelevant symbol PIG. As discussed above, the complexities of storage and retrieval of graphic symbols make this element particularly challenging. One intermediate option to address these types of challenges is to provide access to graphic symbol displays designed to minimize the demands of locating vocabulary—that is, creating topic, or activity-based, displays. This approach helps to ensure that children have early, positive experiences in accessing a range of vocabulary and with early sentence building. Although a child may have a complex communication system with hundreds, or even thousands, of words that are stored within an organizational system designed to serve the child well in the long run, also providing access to more limited displays that do not necessitate complex navigation to build sentences is warranted. This way, the child can focus on utterance and sentence building—that is, true expressive language learning—without being overwhelmed by navigational demands. This approach can be implemented in conjunction with efforts to transition the child to accessing these same symbols within the more complex organizational system as the child's level of mastery increases. Such an approach enables clinicians to embrace the long-term goal of accessing a large lexicon with consistently placed vocabulary without sacrificing the basic principles of early child language development (i.e., the simultaneous development of pragmatics, semantics, and grammar).

To track increases in children's use of relevant symbols in the context of a large-scale clinical trial (e.g., Kent-Walsh et al., 2019), we currently are exploring the use of a new measure: percentage of relevant symbols or PRSym (see Table 2). PRSym is a simple calculation of the number of symbols selected that are relevant to the known communicative context divided by the total number of selected symbols (i.e., PRSym = # relevant symbols/total # symbols). For example, for Child A's five utterance productions in Table 2, six of the 10 symbols this child used were relevant (60%), compared with 10 of 10 symbols selected by Child B (100%). Although we do not yet know when and how well this measure accurately captures change over time, this measure has strong face validity (i.e., it makes sense that we would want this percentage to increase over time), and ongoing research in our labs indicates that this measure appears to warrant exploration. For example, in Table 2, both Child A and Child B are consistently producing two-symbol utterances, but Child A's PRSym is 60%, compared with Child B's PRSym of 100%. Child A's inconsistent use of relevant symbols results in uncertain communicative intents (is she commenting, requesting, or something else?), compared with the comments evident for Child B. PRSym, then, appears to help differentiate children functioning at an early symbol production level (Phase 1 in Figure 1), compared with children functioning at the early symbol combination level (Phase 2).

Semantic Relations

Early two-word utterances mark the beginning of expressive syntax development and occur in typical development at approximately 18 months of age, when children have approximately 50 words in their expressive lexicons (Owens, 2016). Early two- to three-word utterances in any communication mode can be coded using semantic relations (e.g., agent–action–object, possessor–entity). This coding system focuses on semantic meaning and provides a useful way to determine how flexibly a child uses her still-relatively-limited lexicon. For example, for the target utterance The dog –s jump –ed on the bed, Children B, C, and D all use the symbol DOG as both an object (e.g., HIDE DOG) and an agent (DOG JUMP). However, in both spoken and aided language, the inherent lack of grammaticality makes these utterances difficult to reliably code. For example, the meaning behind Child A's production of PIG BED would be unclear even if the child had produced the relevant target words DOG BED. This could be meant to convey possessor–entity (Dog's bed) or agent–locative (The dog is [jumping] on the bed) or to simply label the two entities (I see a dog and a bed). At times, the context of an utterance clarifies the relationship, but often, it does not. In typical spoken language development, this phase of expressive language development is relatively brief, with children generally producing childlike sentences—that is, utterances with an explicit subject and a lexical verb but that still contain omissions and grammatical errors (Hadley, 2014)—by approximately 27 months of age (Hadley, 2006). Thus, semantic relations are most useful for only the earliest portion of word (or symbol) combinations.

Additionally, clinicians can and should note the various types of errors children make in their early symbol combinations (and beyond). Word order issues (i.e., inversions) have long been noted as a unique aided language issue (Binger & Light, 2008) and are discussed in more detail in the Syntax, Morphology, and Grammar section below. However, recent research demonstrates that children tend to make different types of errors with different types of structures, such as higher rates of inversions with certain types of agent–action–object utterances but more omissions and substitutions for other types of word combinations (Binger et al., 2019). Also, children may demonstrate different, individualized learning patterns, which can change over time (Binger, Kent-Walsh, King, Webb, et al., 2017; Kent-Walsh et al., 2015). Tracking error patterns can help clinicians better tailor their interventions—that is, working to highlight the importance of word order for certain utterance types (if inversions are present) versus selecting the correct symbols (if substitutions or omissions are present).

Word Class and Lexical Diversity

In spoken language development, both word class diversity (verbs vs. nouns vs. prepositions) and lexical diversity (e.g., diversity of verbs) are present from the very beginning of expressive language; children's first 50 words include not only nouns but also verbs, prepositions, adjectives, and personal–social terms such as bye-bye (Owens, 2016). Producing a range of word classes is foundational to sentence building; all word classes are necessary to build grammatically complete sentences. Additionally, of course, lexical diversity expands rapidly in typical development starting at approximately eighteen months, including expansion of both closed word classes (determiners, pronouns, prepositions, and conjunctions) and open word classes (nouns, verbs, adjectives, and adverbs). One frequently discussed challenge with graphic symbol-based communication is the inherently abstract nature of the symbols used to represent most non-noun words. Researchers and clinicians have struggled to find the best approaches to graphically represent abstract concepts (e.g., Worah et al., 2015), with no current consensus. Regardless of the approach that the AAC team ultimately chooses, ensuring access to a wide array of vocabulary is essential to language development; both word class and lexical diversity are required for ongoing language development, since children cannot build sentences if all they have access to are nouns (Adamson et al., 1992).

Word Class Diversity

Given the importance of word class diversity and the overuse of nouns often seen in graphic symbol communication, we are introducing a new measure that highlights the need to consciously ensure access to—and instruction in—use of various word classes for children using aided language: number of different word classes (NDWC). This can be used as a simple count of the NDWC observed in a corpus of aided utterances. Table 2 illustrates the increasing number of word classes that may appear as a child's aided language develops, that is, three for Children A and B, five for Child C, and six for Child D, with the latter missing only adverbs and conjunctions in this small sample of five utterances. Like PRSym, this measure has strong face validity but requires validation to see if and when it correlates with increases in utterance and sentence complexity.

Lexical Diversity

Historically, the most widely used measures of lexical diversity used to characterize preschoolers' spoken language are total number of words, number of different words, and type–token ratio (TTR), all of which commonly are derived from language samples. In spoken language, total number of words and number of different words increase significantly with age (Leadholm & Miller, 1992) and are more sensitive estimates of lexical diversity than TTR (Klee, 1992). These measures can easily be adapted for graphic symbol communication, that is, total number of symbols, number of different symbols, and TTR for symbols. More recently, researchers have proposed compelling measures that focus on the complexity and diversity of verbs (Hadley, 2014), which ultimately may prove more useful in tracking sentence development. Spoken language research demonstrates the importance of verbs, including their crucial role in learning sentence frames (Hadley, 2006) and in moving from early word combinations to childlike sentences to adult sentences (Hadley, 2014). Typically developing children “should have a diverse verb lexicon [and] produce frequent and diverse simple sentences…by 3 years of age” (Hadley, 2006, p. 175). Importantly, recent research has demonstrated that verb lexicon measures better predicted grammatical outcomes than noun lexicon measures for young typically developing children during the early stages of grammatical learning (Hadley et al., 2016). These findings illustrate the pivotal role of verbs in sentence development and indicate the need to pay careful attention to verb acquisition. The examples in Table 2 illustrate the differences in verb diversity across children, from Child A's use of only one verb (a transitive verb) through the relatively diverse array of transitive and intransitive lexical verbs as well as auxiliary verbs and a copula for Child D. These examples highlight the need to track not only the number of different verbs but also the type (e.g., transitive vs. intransitive, main V vs. auxiliary V) to ensure that children are acquiring the basic elements of sentence building.

Syntax, Morphology, and Grammar

Word Order Accuracy

Although inversions are not the only types of errors that children make when learning to produce graphic symbol utterances—that is, omissions and substitutions also are common, depending on the utterance type (Binger, Kent-Walsh, King, Webb, et al., 2017; Binger et al., 2019)—word order presents particular challenges to graphic symbol communication. In aided communication, children are faced with an array of symbols and can select whichever symbols they like, a task that is inherently different from speaking or signing. Even children with severe speech disorders who have intact receptive language and presumably would have intact expressive language if their speech were intelligible make word order errors as they learn to map spoken word order onto graphic symbol utterances (Binger, Kent-Walsh, King, & Mansfield, 2017; Binger, Kent-Walsh, King, Webb, et al., 2017; Binger et al., 2019; Kent-Walsh et al., 2015).

Figure 1 reflects the word order patterns we see in children at both the early symbol combination (Phase 2) and childlike sentence (Phase 3) phases as they learn to produce increasingly complex arrays of symbols (cf. word order issues are not included in Phase 4, as clinical experience indicates that this problem decreases as grammatical accuracy increases). Child C, who is largely at the childlike sentence phase, provides an example of word order issues with the sentence DOG JUMP BED ON (vs. the correct DOG JUMP ON BED; see Table 2). Although some researchers have argued that metalinguistic demands are likely to blame for these kinds of challenges (Light & McNaughton, 2012), a growing body of evidence indicates that many children readily learn to use correct spoken word order with relatively brief instruction (Binger, Kent-Walsh, King, Webb, et al., 2017; Tönsing et al., 2014) and sometimes with very little instruction at all (Binger, Kent-Walsh, King, Webb, et al., 2017).

Our own working theory is that, once children truly learn that their task is to map spoken word order onto graphic symbols, we see fewer and fewer of these types of errors. We have not included any specific measures of inversions per se; rather, we encourage clinicians to be aware of this issue as well as other error types (substitutions, omissions) and adjust intervention targets accordingly. One of our current projects indicates that a promising approach for teaching children to attend to word order is to teach children to produce contrastive sentences that contain all of the same words but have different meanings depending on the word order, such as contrasting DOG HUG PIG versus PIG HUG DOG (Binger, Kent-Walsh, King, Webb, et al., 2017) and IS DOG CRY –ING versus DOG IS CRY –ING (Kent-Walsh et al., 2015), assuming the child is at an appropriate developmental level for such targets (for a discussion of this, see Binger et al., 2019).

Grammatical Intent

As children move beyond early, agrammatical word combinations and begin producing longer graphic symbol utterances, they move closer to the production of true sentences. In Figure 1, we use the term grammatical intent to refer to the listener clearly understanding the underlying sentence, even though the sentence lacks full grammaticality. For example, Child C's production I HIDE DOG (see Table 1, Utterance 1) has clear grammatical intent, with the full grammatical expression being some past, present, or future tense version of I HIDE THE DOG. Grammatical intent is negatively impacted by inversions; for example, I DOG HIDE has less certain grammatical intent (is the child hiding the dog, or are both the child and dog hiding?). If the utterance has clear grammatical intent and correct word order but is not yet largely grammatical, we would categorize the utterance as a childlike sentence.

MLU and Other Traditional Measures of Grammar Development

Mean length of utterance in morphemes and words are longstanding measures that have been used to reliably track developing grammatical skills of young children (Rice et al., 2010). However, MLU for graphic symbol measurement—which we are calling MLUSym—has long been acknowledged as problematic for tracking graphic symbol development. This is due to a range of factors such as high rates of message co-construction as well as a lack of access to—and instruction in the use of—grammatical markers (Binger & Light, 2008; Soto, 1999). We addressed the issue of co-construction early in this research note; essentially, co-constructed messages cannot be part of any analysis that uses MLU or similar measures. Providing access to grammatical markers and teaching children to use them (e.g., Binger et al., 2011) addresses the second concern. Additionally, some researchers have argued that the metalinguistic demands of using grammatical markers—or even graphic symbols more generally—may be too high for young children (Light & McNaughton, 2012). This concern is challenged by the growing body of evidence that refutes this claim; that is, multiple experimentally controlled studies have demonstrated that at least some young children can learn quickly to accurately use rule-based graphic symbol utterances (Binger, Kent-Walsh, King, Webb, et al., 2017; Tönsing & Dada, 2016), including accurate use of a range of grammatical markers (Binger, Kent-Walsh, King, & Mansfield, 2017). Given the viability and importance of these goals, finding ways to accurately measure progress is a clinical priority.

That said, MLUSym is still problematic when used on its own. Children A and B in Table 2 illustrate this point: Although both have an MLU of 2.0 in this brief sample, they are in two distinctly different stages, as discussed in the Relevant Vocabulary section above. That is, Child A is largely functioning at the early symbol production stage (i.e., Phase 1 in Figure 1), with frequent productions of irrelevant symbols and questionable communicative intent, while Child B is at the early symbol combination stage (Phase 2), with discernable semantic relations that adhere to spoken word order rules. MLUSym alone does not, then, differentiate children at these two phases. PRSym does, however, with Child A's PRSym of 60% compared with Child B's of 100%, indicating that a combination of MLUSym and PRSym could be quite clinically useful. Taken together, these measures indicate that Child A would benefit from instruction in selecting relevant symbols before focusing on longer productions, while Child B's goals should focus on expanding utterances and moving into childlike sentence productions (with, as the additional measures discussed below indicate, a syntactic focus on noun phrase and verb phrase development). Both preliminary work in our labs (e.g., Kent-Walsh et al., 2019) and clinical experience support this approach at these early symbol combination phases; additional rigorous, controlled research with larger numbers of children will further confirm this approach.

Two additional and particularly compelling measures, given known issues of graphic symbol utterance development, that may be used to characterize graphic symbol development are the percentage of grammatical utterances (Eisenberg & Guo, 2016, 2018) and the percentage of complex sentences (Paul et al., 2018). Productions of these more sophisticated types of sentences must be encouraged for children who have the potential to produce them and should be explored. Additional measures that have the potential to offer clinical assistance with setting specific goals (vs. extremely broad goals such as increasing the percentage of grammatical utterances) are explored below.

An additional consideration is the development of sentence forms, with the following included in Brown's (1973) stages of sentence development: declaratives, interrogatives, imperatives, and exclamations. The sentence forms classically considered within Brown's stages include declarative, interrogative, negative, conjoining, and embedding. Clinicians are encouraged to track the development of these sentence forms with children who are using graphic symbols to communicate, just as they would with children relying on speech, to ensure that children learn to flexibly use the various forms. Focus on interrogatives, and providing children with question words such as who, what, where, and how, is of particular importance. Children who use AAC typically are given relatively few opportunities to initiate conversation and tend to be passive communicators (Light et al., 1985), so it is essential to ensure access to words that allow and encourage children to initiate communication and inquire about their worlds.

Noun Phrases, Verb Phrases, and Unique Symbol Combinations (Unique Subject–Verb)

Noun phrase and verb phrase development are fundamental to sentence building, and therefore, both are included in utterance-level analyses of individual utterances in Table 1. Common noun phrase elements include nouns, pronouns, determiners, adjectives, prepositional phrases, relative clauses (e.g., the dog that scared the man), and participle phrases (e.g., the dog whining for a walk). Common verb phrase elements include modal auxiliaries (may, can, could, should, etc.), perfective auxiliaries (have, has), main verbs, negatives, prepositional phrases (the dog ate by the doghouse), noun phrases (the dog ate a T-bone steak), noun complements (the dog is handsome/a handsome fellow), and adverbial phrases (the dog ate quickly/as quickly as possible). We see clear examples of verb phrase growth across Children A, B, C, and D in Table 1. For example, for the verb phrase in the second target (JUMP –ED ON THE BED), we see movement from no verb phrase at all for Child A, to the single lexical verb JUMP from Child B, to the addition of a partial and then complete prepositional phrase for Children C and D, respectively. To support syntax development, then, we must not only attend to a breadth of word classes (measured via NDWC) and individual words within those word classes (such as measuring verb diversity) but also track the relationships across words, including the relationships within English noun phrases and verb phrases.

Another highly promising approach that considers relationships across words is unique subject–verb (USV) combinations, which may help map a path from early word combinations to childlike sentences (Hadley, 2014; McKenna & Hadley, 2014). To count as a USV, a sentence must (a) include an explicit noun or pronoun in the subject noun phrase position, (b) include a lexical verb, and (c) reflect a sufficiently different subject–verb (SV) combination (McKenna & Hadley, 2014, p. 161). For example, the USVs for the five targeted sentences in Table 2 include I HIDE, DOG JUMP, PIG SLEEP, I WASH, and LION BED, with the total USV counts listed for each of the four children (i.e., two, four, and four for Children B, C, and D, respectively). The authors suggest that, in spoken language, USVs “may be particularly useful for children who are combining words on a regular basis (i.e., MLU 1.50–2.50)” (McKenna & Hadley, 2014, p. 167). In our Graphic Symbol Utterance and Sentence Development Framework, then, Unique subject-verb symbol combinations (USV-Sym) are likely most useful for children entering the early symbol combination phase (Phase 2) through the end of the childlike sentence phase (Phase 3).

Although USV-Sym has, to our knowledge, been untested with graphic symbol utterances, we nonetheless suggest this as an exploratory measure, as it holds promise based on both the spoken language research that supports this measure and its specific implications for aided language goals and intervention. First, SV and subject–verb–object utterances are present for most typically developing children by 24–26 months of age; thus, this measure truly focuses on early sentence development. Second, in McKenna and Hadley's research, USV—but not MLU—was related to grammatical complexity for typically developing children with MLUs of less than 3.25 who were aged 30 months (i.e., children generally functioning in the childlike sentence phase). Relatedly, USV increases over time for typically developing toddlers (Hadley et al., 2018 , 2017) and expands in a developmental sequence across subject types (e.g., first vs. second vs. third person). This measure of sentence diversity, then, holds promise for tracking the growth of childlike sentences and could assist with writing goals (for more information on goal writing, see Hadley, 2014). Second, McKenna and Hadley (2014) note that this early SV sentence diversity may assist with the subsequent acquisition of grammatical marking, which is required for productions of adultlike sentences; most grammatical markers in English are focused on verbs, so lexical verb diversity is a prerequisite to fully acquiring grammatical markers. Third, USVs may be tracked in real time using a structure-specific approach (Hadley et al., 2018; McKenna & Hadley, 2014). This can be accomplished when taking a language sample—or over a longer period of time—by recording only the core subject (i.e., noun or pronoun) and main verb of the child's early sentences (Children C and D's productions of I HIDE, DOG JUMP, etc. in Table 2; also see the appendix in McKenna & Hadley, 2014).

Using USV-Sym will also draw attention to issues unique to aided language, especially the need to promote diverse vocabulary use. Clinicians working with children who use AAC are often encouraged to take a “core vocabulary” approach to aided communication (Boenisch & Soto, 2015), which is designed to reduce the complex navigational demands involved in trying to locate hundreds or even thousands of different words and symbols in an aided system by stressing use of the most commonly occurring words. Although a basic level of grammaticality is often stressed (many words that express grammatical functions such as is, are, and, a, and the are core vocabulary words), this approach can limit vocabulary diversity and growth, with an overreliance on simple words (e.g., go never expanding to related verbs such as walk, leave, or depart). This tension between navigational demands and vocabulary diversity presents genuine challenges, particularly for children with severe motor impairments and those with intellectual disabilities. We encourage clinicians, therefore, to use measures such as verb diversity, NDWC, and USV-Sym to draw attention to the need to expand vocabulary to support ongoing sentence development even in the face of these challenges. Put another way, the team needs to be aware of the various options to make the best informed choices in the face of complex AAC decision making. We must be vigilant in our support of ongoing vocabulary diversity, regardless of the chosen communication modes, always remembering that “complex talk reflects complex thought” (Nippold, 2014, pg. 154) and prevent the corollary: limited words reflecting limited thoughts.

Grammatical Morphemes

In aided language, use of grammatical morphemes is an area of particular concern. These emerge early in spoken language development (Brown, 1973) but nonetheless are often ignored in early aided language development. Although these markers may seem superfluous for young graphic symbol communicators, we must remember that children begin using these markers early in development for good reason: Without these markers, the intended message is unclear. For example, in Table 2, Child B produced LION BED in the fifth utterance. The intent of this utterance is unclear; the child could simply be labeling a lion and a bed, explaining that the lion needs a bed, indicating that the lion is (or should be) on the bed, and so forth. However, simply adding –'S to produce LION –'S BED unambiguously defines this as a possessive relationship. Some preschoolers with severe speech impairments readily learn to use grammatical markers in their graphic symbol utterances (Binger, Kent-Walsh, King, & Mansfield, 2017), and even children with developmental delays can efficiently learn to use these markers (Binger et al., 2011). Given their importance in achieving linguistic competence, grammatical markers should be considered early in development. For example, Tables 1 and ​2 illustrate a range of missing grammatical morphemes for Child B's early symbol combinations and Child C's childlike sentences, with Child D demonstrating more proficiency with these morphemes in his adultlike sentences. These markers can be tracked more broadly with samples of utterances using our new number of different inflectional morphemes measure, as illustrated in Table 2. Like NDWC, number of different inflectional morphemes is a simple count of the number of inflectional morphemes and can then be characterized by what is present and what is missing. English has only eight inflectional morphemes, relatively few compared with many other languages. Research is required to see if and when this measure changes over time and predicts sentence complexity, as well as what should be required to give the child credit (e.g., Is one production of possessive –'s enough? Two or more?) for acquisition.

Conclusion

Graphic symbol-based communication is an evidence-based approach to support young children who require AAC. Using graphic symbols can help children to move beyond early pragmatic and semantic development and acquire grammatical skills along the path to meeting typical language milestones and becoming sophisticated communicators. In supporting language development across all domains, clinicians and researchers encounter a variety of measurement challenges specific to graphic symbol communication. This research note presented a framework of the phases of English graphic symbol utterance and sentence development and suggested approaches for measuring and analyzing graphic symbol use and development to assist clinicians in heightening awareness of the path toward grammaticality as well as tracking progress. Future research is warranted to validate the novel measures that are presented and map a path for setting goals that embrace all aspects of grammatical graphic symbol sentence development.

Acknowledgments

This research note is primarily based on current, ongoing research supported by National Institutes of Health Grant 5R01DC016321 (awarded to Cathy Binger, Jennifer Kent-Walsh, and Nancy Harrington) and is also highly influenced by results from research supported by National Institutes of Health Grants 1R03DC011610 (awarded to Cathy Binger) and 1R15DC014585 (awarded to Jennifer Kent-Walsh and Nancy Harrington). We are greatly indebted to the editors and reviewers for their thoughtful and thorough reviews. Their reviews helped us to clarify our thinking and strengthened the final product immeasurably. We are also grateful to the families who have participated in our research, making the bases for this work possible.

Funding Statement

This research note is primarily based on current, ongoing research supported by National Institutes of Health Grant 5R01DC016321 (awarded to Cathy Binger, Jennifer Kent-Walsh, and Nancy Harrington) and is also highly influenced by results from research supported by National Institutes of Health Grants 1R03DC011610 (awarded to Cathy Binger) and 1R15DC014585 (awarded to Jennifer Kent-Walsh and Nancy Harrington).

References

• Adamson L. B., Romski M. A., Deffebach K., & Sevcik R. A. (1992). Symbol vocabulary and the focus of conversations: Augmenting language development for youth with mental retardation. Journal of Speech and Hearing Research, 35(6), 1333–1343. https://doi.org/10.1044/jshr.3506.1333 [PubMed] [Google Scholar]
• Binger C., Kent-Walsh J., & King M. (2017). Dynamic assessment for three- and four-year old children who use augmentative and alternative communication: Evaluating expressive syntax. Journal of Speech, Language, and Hearing Research, 60(7), 1946–1958. https://doi.org/10.1044/2017_JSLHR-L-15-0269 [PMC free article] [PubMed] [Google Scholar]
• Binger C., Kent-Walsh J., King M., & Mansfield L. (2017). Early sentence productions of three- and four-year-old children who use augmentative and alternative communication. Journal of Speech, Language, and Hearing Research, 60(7), 1930–1945. https://doi.org/10.1044/2017_JSLHR-L-15-0408 [PMC free article] [PubMed] [Google Scholar]
• Binger C., Kent-Walsh J., King M., Webb E., & Buenviaje E. (2017). Early sentence productions of 5-year-old children who use augmentative and alternative communication. Communication Disorders Quarterly, 38(3), 131–142. https://doi.org/10.1177/1525740116655804 [Google Scholar]
• Binger C., & Light J. (2008). The morphology and syntax of individuals who use AAC: Research review and implications for effective practice. Augmentative and Alternative Communication, 24(2), 123–138. https://doi.org/10.1080/07434610701830587 [PubMed] [Google Scholar]
• Binger C., Maguire-Marshall M., & Kent-Walsh J. (2011). Using aided AAC models and contrastive targets to teach grammatical morphemes to children who use AAC. Journal of Speech, Language, and Hearing Research, 54(1), 160–176. https://doi.org/10.1044/1092-4388(2010/09-0163) [PubMed] [Google Scholar]
• Binger C., Richter K., Taylor A., Williams E. K., & Willman A. (2019). Error patterns and revisions in the graphic symbol utterances of 3- and 4-year-old children who need augmentative and alternative communication. Augmentative and Alternative Communication, 35(2), 95–108. https://doi.org/10.1080/07434618.2019.1576224 [PMC free article] [PubMed] [Google Scholar]
• Boenisch J., & Soto G. (2015). The oral core vocabulary of typically developing English-speaking school-aged children: Implications for AAC practice. Augmentative and Alternative Communication, 31(1), 77–84. https://doi.org/10.3109/07434618.2014.1001521 [PubMed] [Google Scholar]
• Brown R. (1973). A first language: The early stages. Harvard University Press; https://doi.org/10.4159/harvard.9780674732469 [Google Scholar]
• Caron J., Light J., Holyfield C., & McNaughton D. (2018). Effects of dynamic text in an AAC app on sight word reading for individuals with autism spectrum disorder. Augmentative and Alternative Communication, 34(2), 143–154. https://doi.org/10.1080/07434618.2018.1457715 [PubMed] [Google Scholar]
• Clarke M. T., Soto G., & Nelson K. (2017). Language learning, recasts, and interaction involving AAC: Background and potential for intervention. Augmentative and Alternative Communication, 33(1), 42–50. https://doi.org/10.1080/07434618.2016.1278130 [PubMed] [Google Scholar]
• Dietz A., Quach W., Lund S. K., & McKelvey M. (2012). AAC assessment and clinical-decision making: The impact of experience. Augmentative and Alternative Communication, 28(3), 148–159. https://doi.org/10.3109/07434618.2012.704521 [PubMed] [Google Scholar]
• Downing J. E. (2005). Teaching communication skills to students with severe disabilities (2nd ed.). Brookes. [Google Scholar]
• Drager K. D. R., Light J. C., Fallon K. A., & Jeffries L. Z. (2004). Learning of dynamic display AAC technologies by typically developing 3-year-olds: Effect of different layouts and menu approaches. Journal of Speech, Language, and Hearing Research, 47(5), 1133–1148. https://doi.org/10.1044/1092-4388(2004/084) [PubMed] [Google Scholar]
• Eisenberg S., & Guo L. Y. (2016). Using language sample analysis in clinical practice: Measures of grammatical accuracy for identifying language impairment in preschool and school-aged children. Seminars in Speech and Language, 37(2), 106–116. https://doi.org/10.1055/s-0036-1580740 [PubMed] [Google Scholar]
• Eisenberg S. L., & Guo L.-Y. (2018). Percent grammatical responses as a general outcome measure: Initial validity. Language, Speech, and Hearing Services in Schools, 49(1), 98–107. https://doi.org/10.1044/2017_LSHSS-16-0070 [PMC free article] [PubMed] [Google Scholar]
• Hadley P. A. (2006). Assessing the emergence of grammar in toddlers at risk for specific language impairment. Seminars in Speech and Language, 27(3), 173–186. https://doi.org/10.1055/s-2006-948228 [PubMed] [Google Scholar]
• Hadley P. A. (2014). Approaching early grammatical intervention from a sentence-focused framework. Language, Speech, and Hearing Services in Schools, 45(2), 110–116. https://doi.org/10.1044/2014_LSHSS-14-0017 [PubMed] [Google Scholar]
• Hadley P. A., McKenna M. M., & Rispoli M. (2018). Sentence diversity in early language development: Recommendations for target selection and progress monitoring. American Journal of Speech-Language Pathology, 27(2), 553–565. https://doi.org/10.1044/2017_AJSLP-17-0098 [PubMed] [Google Scholar]
• Hadley P. A., Rispoli M., Holt J. K., Papastratakos T., Hsu N., Kubalanza M., & McKenna M. M. (2017). Input subject diversity enhances early grammatical growth: Evidence from a parent-implemented intervention. Language Learning and Development, 13(1), 54–79. https://doi.org/10.1080/15475441.2016.1193020 [PMC free article] [PubMed] [Google Scholar]
• Hadley P. A., Rispoli M., & Hsu N. (2016). Toddlers' verb lexicon diversity and grammatical outcomes. Language, Speech, and Hearing Services in Schools, 47(1), 44–58. https://doi.org/10.1044/2015_LSHSS-15-0018 [PubMed] [Google Scholar]
• Kent-Walsh J., Binger C., & Buchanan C. (2015). Teaching children who use augmentative and alternative communication to ask inverted yes–no questions using aided modeling. American Journal of Speech-Language Pathology, 24(2), 222–236. https://doi.org/10.1044/2015_AJSLP-14-0066 [PubMed] [Google Scholar]
• Kent-Walsh J., Binger C., Harrington N., Evelyn S., Feliciani S., Henerson B., Rice A., & Van Gampelaere J. (2019). Targeting semantics & syntax with preschoolers using AAC: Case studies. Poster presented at the Annual Convention of the American Speech-Language-Hearing Association, Orlando, FL, United States. [Google Scholar]
• Klee T. (1992). Developmental and diagnostic characteristics of quantitative measures of children's language production. Topics in Language Disorders, 12(2), 28–41. https://doi.org/10.1097/00011363-199202000-00005 [Google Scholar]
• Leadholm B. J., & Miller J. F. (1992). Language sample analysis: The Wisconsin guide. Wisconsin Department of Public Instruction; http://files.eric.ed.gov/fulltext/ED371528.pdf [Google Scholar]
• Light J., Collier B., & Parnes P. (1985). Communicative interaction between young nonspeaking physically disabled children and their primary caregivers: Part I—Discourse patterns. Augmentative and Alternative Communication, 1(2), 74–83. https://doi.org/10.1080/07434618512331273561 [Google Scholar]
• Light J., & McNaughton D. (2012). Supporting the communication, language, and literacy development of children with complex communication needs: State of the science and future research priorities. Assistive Technology, 24(1), 34–44. https://doi.org/10.1080/10400435.2011.648717 [PubMed] [Google Scholar]
• McKenna M. M., & Hadley P. A. (2014). Assessing sentence diversity in toddlers at-risk for language disorders. SIG 1 Perspectives on Language Learning and Education, 21(4), 159–172. https://doi.org/10.1044/lle21.4.159 [Google Scholar]
• Nippold M. A. (2014). Language intervention at the middle school: Complex talk reflects complex thought. Language, Speech, and Hearing Services in Schools, 45(2), 153–156. https://doi.org/10.1044/2014_LSHSS-14-0027 [PubMed] [Google Scholar]
• Owens R. E. (2016). Language development: An introduction (9th ed.). Pearson. [Google Scholar]
• Paul R., Norbury C., & Gosse C. (2018). Language disorders from infancy through adolescence: Listening, speaking, writing, and communicating (5th ed.). Elsevier Mosby. [Google Scholar]
• Rice M. L., Smolik F., Rytting N., & Blossom M. (2010). Mean length of utterance levels in 6-month intervals for children 3 to 9 years with and without language impairments. Journal of Speech, Language, and Hearing Research, 53(2), 333–349. https://doi.org/10.1044/1092-4388(2009/08-0183) [PMC free article] [PubMed] [Google Scholar]
• Schlosser R. W., & Sigafoos J. (2002). Selecting graphic symbols for an initial request lexicon: Integrative review. AAC: Augmentative and Alternative Communication, 18(2), 102–123. https://doi.org/10.1080/07434610212331281201 [Google Scholar]
• Soto G. (1999). Understanding the impact of graphic sign use on the message structure. In Loncke F. T., Clibbens J., Arvidson H. H., & Lloyd L. L. (Eds.), Augmentative and alternative communication: New directions in research and practice (p. 40– 48). Whurr Publishers. [Google Scholar]
• Tönsing K. M., & Dada S. (2016). Teachers' perceptions of implementation of aided AAC to support expressive communication in South African special schools: A pilot investigation. Augmentative and Alternative Communication, 32(4), 282–304. https://doi.org/10.1080/07434618.2016.1246609 [PubMed] [Google Scholar]
• Tönsing K. M., Dada S., & Alant E. (2014). Teaching graphic symbol combinations to children with limited speech during shared story reading. Augmentative and Alternative Communication, 30(4), 279–297. https://doi.org/10.3109/07434618.2014.965846 [PubMed] [Google Scholar]
• Wilkinson K. M., Romski M. A., & Sevcik R. A. (1994). Emergence of visual-graphic symbol combinations by youth with moderate or severe mental retardation. Journal of Speech and Hearing Research, 37(4), 883–895. https://doi.org/10.1044/jshr.3704.883 [PubMed] [Google Scholar]
• Worah S., McNaughton D., Light J., & Benedek-Wood E. (2015). A comparison of two approaches for representing AAC vocabulary for young children. International Journal of Speech-Language Pathology, 17(5), 460–469. https://doi.org/10.3109/17549507.2014.987817 [PubMed] [Google Scholar]

Articles from Language, Speech, and Hearing Services in Schools are provided here courtesy of American Speech-Language-Hearing Association