Augmentative and alternative communication (AAC) systems, such as those representing words and messages with graphic symbols, can support communication for persons with little or no functional speech who are not (yet) fully literate. Graphic symbol-based systems require the preselection of vocabulary. The selected vocabulary should enable the person using the system to meet all their communication needs across different contexts and with different partners (Fallon, Light & Paige, 2001), but should be small enough to minimise the cognitive and motor demands in memorising and navigating to the location of the vocabulary. Vocabulary selection typically requires people other than the individual who will use the vocabulary to predict the words this person will require in all the communication situations that he or she encounters (Dark & Balandin, 2007). This is a complex process, often leading to the inclusion of irrelevant and seldom-used words (Dark & Balandin, 2007). Various ways to select the most relevant and appropriate vocabulary have been suggested and researched (Thistle & Wilkinson, 2015). One such method is the use of core vocabulary lists as a resource for vocabulary selection.

Previous studies in English and other European languages have indicated that 200–250 spoken words account for approximately 80% of a person’s spoken communication (Robillard, Mayer-Crittenden, Minor-Corriveau & Bélanger, 2014; Trembath, Balandin, & Togher, 2007). Spoken language sample analyses in Korean, Mandarin Chinese and isiZulu have also indicated that it is possible to isolate a limited number of lexical items (although these may not always be words but may also include morphemes) that are used with a high frequency and are reused continuously in conversations (Liu & Sloane, 2006; Mngomezulu, Tönsing, Dada & Bokaba, 2019; Shin & Hill, 2016). These frequently- and commonly-used words/morphemes have been termed ‘core vocabulary’. Core vocabulary may be useful for inclusion on AAC systems because of its small size but high potential for reusability across contexts and partners (Baker & Chang, 2006). Core vocabularies typically contain a number of structure words – words that have little semantic meaning but perform a grammatical function and therefore contribute to the grammatical correctness of sentences (Witkowski & Baker, 2012). Certain parts of speech, such as conjunctions and prepositions, are typically classified as structure words. Content words, such as nouns, verbs and adjectives, on the other hand, do have a semantic meaning.

Although Afrikaans vocabulary frequency lists have been published, these are either derived from written sources or the source of origin is not mentioned (Barnes, 1995; Malan, 1943). Since written language samples have been shown to differ from spoken language samples, these lists may not be completely representative of the spoken core vocabulary used by children (Liu & Sloane, 2006). Similarly, the translation of existing core vocabulary lists derived from samples in languages other than Afrikaans is also unlikely to yield an appropriate representative core vocabulary set, because morphological differences between different languages result in (at times substantial) differences in vocabulary lists of specific languages (Mngomezulu et al., 2019). These morphological differences would affect especially structure vocabulary. Sociocultural and geographic influences have also been noted in the core vocabulary, affecting specifically nouns, slang words and interjections (Balandin & Iacono, 1999). These differences have been noted even between Australian and American English adult-based core vocabulary lists, and can therefore be expected to be found in the lists generated for different languages. Language-specific frequency lists that are based on original spoken language samples are therefore required.

The main aim of this study was to determine the Afrikaans core vocabulary of Grade R learners without disabilities. The study had the following objectives: (1) to analyse language samples of Afrikaans-speaking Grade R learners obtained during regular preschool activities in order to determine the total number of words, the number of different/unique words (NDW), and the most frequently- and commonly-used words (core vocabulary); and (2) to further describe the core vocabulary by classifying words into content and structure words.

After unintelligible units were excluded, the words collected per participant ranged from 3108 to 3419 words. The composite sample across all 12 participants consisted of 39 645 orthographic words. The NDW obtained from the composite sample amounted to 3304 orthographically distinct words, with an overall TTR of 0.08 or 1:12.

A core vocabulary list consisting of 239 words was determined (see Appendix 1) by using the three-step process described in the methods section. The average commonality score of the 239 core words was 10. The remaining words made up the fringe vocabulary. Core words had a coverage of 79.4% of the entire composite sample. This means that, of the 39 645 orthographic words that made up the composite sample, 79.4% were core words. Fringe words therefore covered the remaining 20.6% of the sample.

The 239 words identified as core vocabulary were categorised into two broad categories consisting of content words (carrying a semantic meaning) and structure words (words that have little semantic meaning and perform a grammatical function) (Van Rooy, 2017). Of the 239 words in the vocabulary list, 76 were structure words and 163 were content words.

The percentage of content versus structure words in the most frequently-used 20, 50, 100, 200 and total number of core words is provided in Figure 1. From the figure it can be seen that the percentage of content words increases as the number of words increases.

Although there were more content than structure words (in number) in the core vocabulary, content words were used only marginally more often than structure words. Whilst the total coverage of the core was 79.4%, 41.0% of words used during conversations were content core words, whilst 38.4% of words were structure core words. The TTR for content words was about 0.01 (or 1:100), whilst the TTR for structure words was about 0.005 (or 1:200). This means that, on average, a content word occurred 100 times in the sample, whilst a structure word occurred 200 times on average.

The core vocabulary was categorised into 14 different parts of speech. Results are represented in Table 1. Verbs had the highest NDW (n = 53), followed by adverbs (n = 31) and nouns and pronouns (n = 29 each). The categories miscellaneous words (n = 1), articles (n = 2) and enclitic words (n = 2) had the lowest NDW. Regarding frequency of use, the part of speech most frequently used was pronouns, constituting 20.2% of the total sample. This was followed by verbs and adverbs, representing 18.3% and 11.3% of the total sample, respectively. The NDW contained per category does not necessarily predict the frequency of use of that category. For example, 29 pronouns covered 20.2% of the sample, whereas 53 verbs covered 18.3%. The category ‘articles’, on the contrary, contained only two words (’n and die), but accounted for 4.0% of the sample.

The language sample of 39 645 words obtained in this study yielded 3304 unique words (NDW). The core vocabulary established by the three-step process consisted of 239 words, which accounted for 79.4% of the composite sample. The results of the study show vocabulary frequency patterns that are in many ways similar to those found in studies done in other languages, where the total number of words or linguistic units used typically outnumbered the unique number of words or linguistic units by a considerable margin (Boenisch, 2014; Liu & Sloane, 2006). As reported in other studies, a considerable proportion of the total words is represented by a relatively small set of unique words. These results are consistent with those from previous literature, indicating that a small set of linguistic units are repeatedly reused with a high frequency during conversations.

The proportion of structure versus content words in the Afrikaans core vocabulary (which amounted to 32% and 68% respectively) was found to be identical to that found in the English core vocabulary established by Boenisch and Soto (2015). These authors found that in the top 300 words used most frequently by native English speakers, 68% were designated as content vocabulary, whereas 32% were designated as structure vocabulary. A similar pattern was also discerned by Mngomezulu et al. (2019), who found that 68% of the isiZulu core vocabulary established, consisted of content formatives, whereas 29% consisted of structure formatives (with 3% consisting of words classified neither as content nor core).

Although there were less structure words than content words in the core vocabulary, the coverage of content and structure core words differed only marginally. The TTRs show that, on average, a structure core word occurred twice as often as a content core word in the sample. Mngomezulu et al. (2019) also found that there were more content than structure formatives in the isiZulu core vocabulary, but that structure formatives occurred with a higher frequency. In their study, structure formatives occurred about 3.3 times as often (on average) as content formatives.

Similarities were found in terms of the parts of speech making up the core vocabulary. Pronouns, verbs and adverbs made up a large proportion of the core vocabulary in this and other studies (Boenisch & Soto, 2015; Mngomezulu et al., 2019). These parts of speech tended to be used with a high frequency; for example, many different verbs were found in the core vocabulary whilst also being used with a high frequency. Similarly, pronouns and adverbs featured prominently, both in terms of NDW and in terms of frequency of use. These findings suggest some similarities in the way parts of speech are used across languages. Verbs, for example, seem to be central to sentence construction across languages (Sutton, Soto & Blockberger, 2002).

The Afrikaans core vocabulary established in this study is based on language samples collected across a variety of preschool activities, resulting in a vocabulary that is not limited to, for example, playtime or snack time. Furthermore, excluding the first 20 min of each sample reduced the novelty effect that the recorders might have had on the participants’ speech. This study is one of few where homonyms and polysemes were rigorously separated. The aim thereof was to ensure that the meaning of each word in the core vocabulary list was unambiguous, thus enabling appropriate graphic representation of the words in an AAC system.

Whilst the inflected morphological variations of nouns, verbs, numerals, pronouns and adjectives were counted together under the root form of the word in this study, the codes and tags added to the transcription also allowed for frequency information for each variation to be obtained (Hattingh, 2018). This information can assist practitioners to make informed decisions about the necessity of including access to inflected forms on specific AAC systems.

The relatively small sample size in combination with participants from similar ages across three fairly similar sites, as well as the limited time span taken for data collection, influenced the generalisability of the data. The criteria for determining core vocabulary consisting of frequency and commonality scores have also been criticised as arbitrary (Shin & Hill, 2016). In the present study, vocabulary was included based on a commonality score of at least six (i.e., at least 50%) and a frequency count of equal to or more than 0.05%. There is no scientific reason why words with a frequency of 0.05% are designated as core vocabulary, whilst words with a frequency count of 0.049% are designated as fringe. The commonality criterion is equally arbitrary. Shin and Hill (2016) suggest that grouped frequency counts are a more rigorous and defensible way to distinguish between core and fringe words.

The established Afrikaans core vocabulary list (Hattingh, 2018) can be used as a vocabulary source for Afrikaans-speaking children who require AAC. Words can be judiciously selected from this list for inclusion on AAC systems. The core word list can be particularly helpful to alert team members to the importance of including structure words, since these words provide access to morphology and syntax. However, the provided core list is not intended to be used in isolation, but in combination with a child-specific fringe vocabulary. Environmental inventories and informants can assist in determining these fringe words that should reflect the interests, personality, and contexts of the child in need of the system (Robillard et al., 2014). An appropriate combination of core and fringe vocabulary should allow children to express themselves appropriately in various communication situations.

Since this is the first core word list established in Afrikaans based on a sample of spoken words, clinicians may also consider referring to it when selecting words for Afrikaans individuals of other ages or for use across other settings, as previous research has indicated that core vocabularies are, to some extent, useful across settings and individuals (Van Tilborg & Deckers, 2016). However, this should be done judiciously, as certain words in the core vocabulary list (specifically nouns and verbs such as blok [‘block’] and inkleur [‘to colour’]) are clearly reflective of the preschool context and specific to the population.

The vocabulary list may also be useful for purposes outside the realm of AAC. Representing words that are frequently used by Afrikaans-speaking Grade R learners, the list could be consulted in the development of formal and informal language and vocabulary assessments. It could also be consulted when developing intervention tasks targeting syntax and morphology, in an effort to choose familiar vocabulary that does not add a linguistic processing burden. However, the small and homogenous participant group has to be kept in mind, and caution should be exercised to avoid overgeneralisation of the results.

The current core vocabulary list was generated based on conversational samples from a small and homogenous group of participants. Since the way persons use language and vocabulary is intricately related to their age, as well as to the social and physical micro- and macro-context and the activity settings in which they find themselves, collecting further spoken Afrikaans language samples from persons of different age groups, various geographical regions, and during various communication situations could help to establish a more representative core vocabulary list. Furthermore, studies are needed to determine if design conventions for core–fringe based AAC systems that have been developed for other languages (e.g., English systems such as Tobii Snap Core First^TM,¹ the Super Core grid set for Grid 3^TM,² and the Pixon^TM communication boards^TM3) are appropriate for Afrikaans core-fringe-based systems. Research to determine effective methods for introducing the use of such systems to persons in need of AAC and their partners is also urgently needed.

This study contributes to the growing body of research aiming to support professionals and other team members with the challenging task of selecting the appropriate vocabulary for not (yet) literate individuals. The identified list of Afrikaans vocabulary core words can be used as a source for guiding vocabulary selection for young children in need of an Afrikaans AAC system. Specifically, the variety and proportion of content and structure words, as well as words from different parts of speech found in the core vocabulary, point to the necessity of including such a variety on a system that is intended to encourage novel utterance generation.