It’s and that’s are examples of a phenomenon known as limited scope formulae. Limited scope formulae (LSF) are stereotypical phrases that are used by rote automatically [14]. Braine [14] described children’s initial word combinations like these as a “constant-plus-variable” providing the examples “it’s a X” and “there’s a X” (p. 27). He explained that these formulae are limited in scope in that children use them for expressing very specific meanings only, for example, identifying something. Put another way, children’s earliest uses of grammatical morphemes such as the copula are produced without a true syntactic representation [8]. Instead, they are produced via a direct connection between the referent and phonetic content [15]. Certain fixed, stereotypical uses of copula forms are frequently heard in the parent input. Very often, these LSF-like combinations are pronominal subjects such as what or that combined with contracted forms of copula is. Children may begin to produce segments such as what’s and that’s without true knowledge of the syntactic, semantic, and phonological features of the copula before an abstract representation plays a role in their productions. In this case, children are not using their own grammatical systems productively. The term productivity refers to children’s use of grammatical structures generated from their own grammatical knowledge [16,17]. To develop grammar, uses of LSF must be replaced with productive subject+copula combinations. When this occurs, children use more creative (i.e., lower frequency) subjects and predicates, marked by a shift from pronouns to lexical NPs [8].

Traditional developmental measures of tense and agreement marking have not always accounted for the transitional period from productions taken directly from the input to productions generated by a productive, abstract system. Percent accuracy is a widely used method for measuring children’s knowledge of tense and agreement marking [4,18]. Accuracy is a rather stable measure revealing gradual growth of grammatical knowledge when used with children who have demonstrated sufficient emergence of tense and agreement morphemes in a range of sentence types, particularly past age 4 [19,20]. However, stability in scores is not observed when accuracy is used to measure tense and agreement marking when omissions are still present in children’s utterances. For the average child, tense and agreement morphemes are largely absent before age 2 [10]. When children initially use the copula in LSF they have heard in the input (e.g., It’s mine), a temporary increase in tense marking accuracy is observed [11,12]. As a result of tense and agreement morpheme omissions, accuracy scores subsequently drop as toddlers generate sentences independently of fixed phrases pulled from the input (e.g., This book _ mine).

Productivity scores are an alternative for measuring grammatical development in very young children [21,22]. Measures of productivity emphasize the significance of children’s combination of subjects and finiteness morphemes instead of measuring only the functional morphemes themselves as accuracy scores do. The emphasis on the subject+morpheme combinations recognizes that accuracy scores may be inflated by finiteness morphemes that are paired with high frequency, repeated subjects (e.g., it’s, that’s). Hadley and colleagues have used productivity measures to track the early development of tense and agreement morphemes [21–23]. These measures, such as the tense and agreement productivity (TAP) score, are more conservative than percent accuracy in crediting the child only once for the first instance of a subject+morpheme or verb+morpheme combination but not for subsequent uses and by capping the total uses of any of the morphemes at five. In the TAP score, “she is” and “the teacher is” are sufficiently unique, and “cow walks” and “cow walked” are sufficiently unique. An advantage to productivity measures over accuracy measures for very young children is that they capture the onset of utterances that children generate from their own grammars as opposed to awarding credit for tense morphemes learned from LSF.

Previous explorations of children’s acquisition of copula forms revealed that parents’ provision of grammatical structures is related to children’s subsequent use of the same structures [24,25]. However, it is important to understand the influence of parent copulas beyond solely the number of occurrences in the input. Certain instances of grammatical morphemes in parent input may be more helpful for children’s grammatical growth than others. For example, child-directed auxiliaries in parents’ genuine, information-seeking questions are more useful than the same auxiliaries in test questions [26]. Likewise, children make differential use of copula input based on whether the form precedes or follows the sentence subject (i.e., questions vs. declaratives; [27,28]). Thus, children do more with input than just track the number of exposures to a structure. Their uptake from the input differs from exposure to exposure of a morpheme.

This leads us to consider in what ways parents and clinicians could provide copulas that will best take advantage of children’s learning strategies. For one, when adults use copulas with lexical NP subjects, those subjects are more likely to have concrete referents in the environment than pronominal subjects are. Further, using lexical NP subjects may yield morpheme boundaries that are easier for children to detect because the transitional probability between the subject and the copula is lower [29]. On the other hand, LSF (e.g., that’s) have relatively higher transitional probabilities between the subject and the copula, obscuring the morpheme boundary. Lastly, uses of the full form of copula is are more acoustically salient than contracted copulas because the formant frequencies of the vowel nucleus make the syllable more audibly noticeable [30]. In input resembling LSF, contracted copulas (i.e., it’s instead of it is) are lower in phonetic salience and reduced to an acoustically shorter duration.

Children who hear parent copulas that resembles LSF more often than their peers may be particularly susceptible to relying on LSF when they attempt copulas themselves. As a result of reliance on the input, these children may use fewer unique subjects with copulas and fewer copula conjugations (e.g., am, were). This possibility is important to explore for two reasons. First, if structural properties of copula input relate to children’s copula growth, we will better understand how children make use of different forms of input. Second, overestimation of grammatical knowledge may be especially likely for children whose copula use is produced by rote and not produced through grammatical encoding. Therefore, an investigation of the relationship between parents’ use of copulas in LSF-like constructions and children’s subsequent propensity for unique subject+copula combinations would not only help us understand what forms of parent copulas best facilitate uptake. Such an investigation would also shed light on whether traditional measures of grammatical development are more likely to over credit some children’s grammatical knowledge. To be clear, we recognize that parents are aware of morpheme boundaries when using input such as what’s and that’s. We refer to this input as LSF here for convenience because this input is highly reminiscent of the child utterances Braine [14] described as being formulaic and limited in their scope.

The extant literature leaves a gap in our understanding of the role of parent LSF on children’s copula acquisition. Some previous studies have either measured productivity for the entire suite of tense and agreement morphemes or have quantified only copulas in parent and children’s utterances. Rispoli et al. [31] found that a productivity score inclusive of multiple tense/agreement morphemes was a stable, valid measure of children’s grammatical growth and a predictor of subsequent tense/agreement accuracy. Hadley et al.’s [32] investigation revealed that finiteness information in parent input was related to children’s overall finiteness productivity. This suggests that a study homing in on copula is, the morpheme that initiates the tense and agreement system, may also uncover valuable relationships. Rispoli et al. [27] classified parent copulas by full or reduced form and by clausal position, finding that the intersection of full form copulas and declarative structure was positively related to children’s later copula is productivity.

The current study seeks to build upon the findings of previous studies by classifying parent copula forms into one of two categories based on both the structure of each copula and the subjects combined with each copula. This study additionally builds upon previous uses of productivity scores as a way of determining children’s progress toward mastery (i.e., true learning; [21,22,31]). This is accomplished by measuring all of children’s unique subject+copula combinations (e.g., These guys arered) with no caps on the number that can be credited and without restricting credit to only the use of is. To understand what types of parent input contribute to children’s cross-morpheme facilitation, the outcome measure should include uses of the structure in question (i.e., copula) across person and number agreement combinations (e.g., are, was) and not only forms that children used exactly as parents had used them (e.g., is, it’s). Therefore, this study is expected to contribute new information to our understanding of copula development by classifying parent copula use in a new way and by measuring children’s copula productivity in a way that will avoid the potential for ceiling effects and in a way that reveals progress across the entire copula system. Specifically, this investigation is expected to provide insight into how children transition from using copula is exactly as it is heard in parent input to using copula forms productively in original sentences. The rationale for the focus on third person singular copula is as opposed to am, are or past tense forms is three-fold. First, is occurs more frequently in parent input than other copula forms [32,33]. Second, young children productively use copula is before other tense and agreement morphemes [31]. Since copula is acquisition occurs relatively early, a measure of productive use allows us to better understand the role of input in children’s initial emergence of the tense and agreement system [5]. Third, each copula form that is contractible to a subject (i.e., is, am, are) has a unique phonetic form. Third person singular is productions are reduced in phonetic form because of contraction to the high frequency sibilant sound [s], whereas am and are when contracted, are reduced to sonorants, which are easier to perceive [30]. Each instance of is in parent input will be examined for the presence of contraction to the sentence subject (i.e., is or ’s). Next, the subjects combined with each copula is will be classified as one of two types, pronominal or lexical.

The following research questions are addressed:

Based on previous literature, we predict that parent copulas that are either contracted or combined with a high frequency pronoun (i.e., LSF copulas) will outnumber parent copulas that are either uncontracted or combined with low frequency, lexical NP subjects (i.e., non-LSF copulas). Next, LSF copulas are expected to be negatively related to children’s growth in unique subject+copula combinations (i.e., USC) just as children are beginning to use copula forms productively. On the other hand, non-LSF copulas (e.g., The tea is ready; The diamond’s yours?) are expected to be positively related to children’s growth in USC.

Participants included 11 mothers and their typically developing, monolingual, English-speaking children (6 male and 5 female children). The transcripts were acquired from the Child Language Data Exchange System (CHILDES), an online database of donated language transcripts [34]. Ten children’s transcripts were selected from the Manchester corpus of CHILDES because the research questions in the current study required dense, longitudinal sampling. Samples in the Manchester corpus were collected by Anna Theakston and colleagues for a longitudinal study of 2- to 3-year-old typically developing children acquiring British English [35]. The children whose transcripts were used were Anne, Aran, Carl, Dominic, Gail, Joel, John, Liz, Ruth, and Warren. Two children in the Manchester corpus, Becky and Nicole, were excluded because their language samples began too late for the present research questions (i.e., 24 months of age). To compensate for this exclusion, Nina, the only child in the Suppes corpus, was added because her language samples were collected beginning at a similar age to the rest of the Manchester corpus with even denser sampling and with recordings that importantly were the same duration as the Manchester recordings [36]. In both the Manchester and Suppes corpora, children were audiotaped for 1-hour language samples while engaging in normal play activities with their mothers in their homes. Families were reported to be of middle-class SES, but as parent age and educational level were not primary measures in the original studies, that information is not reported.

The present study used donated transcripts of language samples that were collected when the children were 23, 27, and 33 months old. Mothers’ utterances were analyzed when the children were 23 months old, which was the earliest measurement point for which all children had an available transcript. Because parents increase the complexity of their input in response to their children’s development, it was important to analyze input as early as possible to reduce the likelihood that parents’ copula use was influenced by their children’s utterances. Child utterances were analyzed when the children were 27 months (mean MLU=2.62) and 33 months (mean MLU=3.13) old.

The first two parent measures were the number of LSF copulas and the number of non-LSF copulas. Since the sampling time was held constant at 1-hour long, frequencies of parent copula use could be used. The overall amount of copulas heard by each child may affect his or her copula acquisition, so frequencies were preferred to proportions for the correlations described below. Using proportions, for example dividing LSF by all copula use or total utterances, could have obscured each parent’s level of general talkativity [37]. Proportions of LSF copulas out of all copulas are provided for descriptive purposes only.

The three child outcome measures were USC at 27 months, USC at 33 months, and the growth in USC between 27 and 33 months (i.e., a difference score calculated by subtracting USC at 27 months from USC at 33 months). First, all utterances coded as containing a copula were listed. From there, unique combinations of subject+copula were counted. See Figure 1 for a sample calculation with all uses of copula forms from one child, Anne. Children were only given credit for combinations that were likely to have been produced productively [cf. 31]. To receive credit as a unique combination, copula forms of am, is, are, was, and were needed to be either uncontracted (e.g., The fringe is lots of colors; Anne, 33 months) or contracted to a lexical NP subject (e.g., The light’s on; Aran, 33 months). Combinations were only counted once so that children were not credited multiple times for repeated identical combinations (“There you are” and “Here you are”; Liz, 33 months). A subject could be repeated if a different form of copula was used (e.g., It is vs. It was; Nina, 33 months). A copula form could be repeated if combined with a different subject (e.g., train is vs. elephant is; Carl, 33 months). The unique uses were then totaled at each of the two measurement points. Finally, the difference in USC between the two measurement points was calculated to measure the growth in children’s copula productivity.

Reliability was established for parent LSF copulas, parent non-LSF copulas, and child copulas. Reliability was established for parent coding through a calculation of inter-rater agreement. After the second author coded all parent utterances, the first author randomly selected three of the 11 transcripts and independently coded them resulting in percent agreement values of 99.4%, 97.1%, and 97.5%. To establish coding reliability for children’s uses of copulas, the first author checked every copula code in child utterances and examined all child utterances without a copula code to ensure no instances had been missed. Three changes were made across the 27-month transcripts and eight changes were made across the 33-month transcripts out of 52 and 161 total copula uses, respectively. All changes were instances of auxiliaries that had been incorrectly coded as copulas. Lastly, the first author examined the scores that the second author had assigned to each child for USC at both 27 and 33 months. Disagreements on the copula uses credited as USC were resolved through consensus.

To address the first research question, variability in the structure of the copulas parents directed to their children was characterized. Mean, standard deviation, and range were determined for each of the three parent measures and each of the three child measures.

To address the second research question, how parent copulas related to children’s subsequent growth in subject-copula combinations, two Pearson product-moment correlations were run. The first correlation was between parents’ use of LSF copulas and children’s six-month growth in USC. The second was between parents’ use of all other third person singular copulas (i.e., non-LSF) and children’s six-month growth in USC. To confirm that a Pearson correlation was appropriate for detecting a relationship between parents’ copula input and children’s growth in USC over six months, we first determined that each variable followed a normal distribution using a Q-Q probability plot. We then confirmed that the assumption of bivariate normality was not violated [38,39]. For the two-tailed tests, the correlation significance level was set at 0.05.

First, the results of this study contribute new information toward explaining previous methodological challenges in the measurement of children’s grammatical growth. Specifically, the findings shed light on why some children’s tense/agreement marking accuracy appears to decline over time then increase again. Pine et al. [40] measured the accuracy of tense and agreement morphemes in the Manchester corpus (i.e., the majority of participants in the present study). Of the children included in both the present study and in Pine et al. [40], Anne, Aran, and Warren demonstrated such a pattern of temporary decline in copula accuracy scores, yet all three demonstrated growth in USC, a productivity score, as did seven of the other eight participants (see also [31]). Scores dropping over time then later increasing is inconsistent with the gradual development observed in productivity scores. It is possible that the children whose parents used the highly frequent pronominally contracted LSF copula (e.g., it’s, that’s) are susceptible to measurement error [11,31]. Specifically, children’s grammatical knowledge may be more likely to be overestimated, as indicated by inflated accuracy scores, when their parents provide tense information that is highly formulaic and phonetically reduced. Therefore, the findings of this study could help identify children for whom this type of measurement error is likely to be observed.

The findings of the current study also hold implications for both parents and clinicians working with children at risk for delayed acquisition of tense and agreement morphemes. We suspect that LSF copulas are especially likely to appear in test questions (e.g., What’s this called?). If a future study confirmed this, parents and clinicians could be encouraged to comment on the environment. Commenting should have the effect of proportionally reducing test questions and increasing the combination of names for things with third person copulas (e.g., The bowl is full; [33,42]). Adults’ input of copula forms is more likely to resemble LSF simply by virtue of using pronouns as subjects because copulas are more likely to be contracted when the subject is a pronoun such as it or that [43,44]. On the other hand, providing explicit information about objects and surroundings with lexical NP subjects could promote growth in USC by decreasing children’s reliance on rote memorization. Recent work has provided evidence that when parents are taught to increase the diversity of their own sentence subjects through other-focused commenting, their children later use more diverse third person subjects themselves and use more adult-like tense and agreement marking compared to peers [45,46].

There is room for further refinement of our parent measure. First, we do not know if one type of the non-LSF copula (i.e., uncontracted or lexically contracted) drove the correlation between parent input and child outcomes. A finer-grained coding scheme could detect a more precise relationship. Second, future research could explore specifically which types of uncontracted (i.e., full form) copula is are most helpful for children learning grammar. In the present analyses, we collapsed all full forms together. It is possible that some full forms are more helpful than others based on word order or sentence position (cf. [28]). To our knowledge, no researchers have quantified the relative frequency of different types of full form copula is in the input of British English-speaking parents, which may significantly differ from American English in copula provision because of highly frequent tag questions.

A future study could take advantage of the methodological benefits of productivity scores to address the debate regarding whether children’s initial knowledge of copulas and auxiliaries is restricted to lexically-specific constructions. Wilson [44] reported that Nina [36] and Adam [24] each demonstrated very different rates of provision for morphemes across pronoun subjects (i.e., He’s vs. I’m). Pine et al. [40] addressed this discrepancy taking participants’ knowledge of individual morphemes into account. They reported that children in the Manchester corpus had higher copula accuracy with it and he subjects than with I subjects. A future investigation could explore the stability of specific subject+copula constructions with a productivity-like score to address this question without the pitfalls of accuracy scores. Such an investigation could uncover whether cross-morpheme facilitation is present or if instead children’s use of copula morphemes relates to parent provision of those copula forms specifically.

Lastly, the children in the current study were typically developing and the sample size was relatively small, so future work is needed to explore how parent copulas relate to children’s growth in USC in at-risk populations. Furthermore, socioeconomic status was not a primary consideration during recruitment of the Manchester or Suppes corpora. As families were middle-class, we are limited in generalizing these results to other family backgrounds.