The development of a vocabulary for PEEPS-SE—profiles of early expressive phonological skills for Swedish

ABSTRACT

This paper describes the development of a vocabulary for Profiles of Early Expressive Phonological Skills for Swedish (PEEPS-SE), a tool for assessment of expressive phonology in Swedish-learning children in the age range of 18–36 months. PEEPS-SE is the Swedish version of the original PEEPS, Profiles of Early Expressive Phonological Skills, which uses two age-adequate word lists—a basic word list (BWL) for the assessment of 18–24-month-old children, to which an expanded word list (EWL) is added for assessment of 24–36-month-old children, or children with more than 250 words in their expressive vocabulary.

The selection of words in PEEPS-SE is based on two types of criteria: age of acquisition and phonological complexity. The words also need to be easy to elicit in a natural way in test situations. Vocabulary data previously collected with the Swedish Early Communicative Development Inventory are used for selection of age-adequate words, where the BWL contains words acquired earlier compared to the additional words in the EWL. The latter also contains words that are more phonologically complex compared to those in the BWL. Word complexity was determined by the Swedish version of word complexity measure. PEEPS-SE has made an attempt to match the original version of PEEPS in terms of both assessment method and word selection.

KEYWORDS:

• Early language acquisition
• phonological development
• phonological complexity
• vocabulary
• speech production assessment

Introduction

Children’s individual developmental variations typically have an impact on the development of productive phonology. For example, a child engaging in social communication tends to be exposed to more linguistic input (e.g. Tomasello, 1992). Sensitivity to vocal patterns as well as motor skill development has an impact on production (MacNeilage & Davis, 1990), just as individual cognitive development influences attention and effort (Locke, 1996). Phonology is known to be related to other language areas and later literacy development (Stoel-Gammon, 2011). Phonological factors both quantitative (e.g. limited phonetic inventory, low percentage of consonants correct) (Shriberg, Austin, Lewis, McSweeny, & Wilson, 1997) and qualitative (e.g. atypical error patterns, greater sound variability) have been shown to be potential markers of long-term phonological delay (Peterson, Pennington, Shriberg, & Boada, 2009; Skebo et al., 2013; Williams & Elbert, 2003). Consonant use early in development has been found to be a predictor of later lexical and phonological development (McCune & Vihman, 2001; Vihman & Greenlee, 1987), and vowels are found to play a lesser role than consonants in establishing lexicon, at least in languages where the consonant-to-vowel ratio is relatively high (Nazzi, 2005).

Early detection of deviant phonological development is important in enabling early interventions and tailoring parent-based intervention (Ha, 2015; Roberts & Kaiser, 2011), as well as detecting potential co-existing neuropsychiatric disorders (Miniscalco, Nygren, Hagberg, Kadesjö, & Gillberg, 2006). Measuring early phonological development by studying the child’s production of speech segments, word forms and syllable structures—either independently or in relation to adult targets—can therefore play a critical role in assessing a child’s linguistic status. A tool for the measurement of early phonological development is the above-mentioned US English test Profiles of Early Expressive Phonological Skills (PEEPS) (Stoel-Gammon & Williams, 2013). PEEPS uses age-adequate test words strategically selected to reflect expressive phonology by targeting specific US English sound classes, word forms, and syllable structures. To assess the child’s developing expressive phonological system, words are appropriate items since phonological representations can be assumed to emerge from the combinatorial pressure imposed by increasing numbers of lexical representations (Lacerda, 2003). Speech articulation skills might also be assessed by using non-words or lexicalized items, but non-words can be expected to be difficult to elicit in children in the age ranges targeted by PEEPS, and lexicalized items are more variable than established words. Parental reports like the communicative developmental inventory (CDI) provide reliable insight on the child’s lexical development (e.g. Heilmann, Weismer, Evans, & Hollar, 2005; Thal, DesJardin, & Eisenberg, 2007) and therefore offer a plausible basis for finding words that are familiar to children. In addition, using CDI words is convenient for the ease of elicitation in a test situation, because they necessarily refer to events or objects in the child’s ecological setting. The Profiles of Early Expressive Phonological Skills for Swedish (PEEPS-SE) attempts to maintain compatibility with the original PEEPS by applying independently motivated universal phonetic principles that capture the physiological and aerodynamic constraints of speech sound production (Ohala, 1983, 1993). The universal phonetic principles are seen here as the potential that a newborn child has to produce any speech sound occurring in a human language, as a result of the interaction between aerodynamic entities and physiological affordances (Lindblom & Maddieson, 1988; Maddieson, Hardcastle, & Laver, 1996). However, because the child develops in an ecological context, communicative biases will rapidly introduce different weights on specific aspects of the child’s sound production, although the core developmental path will inevitably be grounded on universal phonetic principles. In other words, PEEPS-SE is based on the notion that the phonological diversity of natural languages must be compatible with a universal phonetic developmental path that starts with general aerodynamic and physiological principles and progressively converges towards the child’s ambient language. By adopting such a universal approach, further work to create language-adjusted versions of PEEPS will be possible while retaining cross-language compatibility.

PEEPS original

The original version of PEEPS is a clinical test for expressive phonology in English-speaking children aged 18–36 months (Stoel-Gammon & Williams, 2013). No norm data are yet published. PEEPS targets sound classes, word patterns and syllable structures. Words are elicited by exposing the child to toys and familiar objects and actions in a playful manner.

Stoel-Gammon and Williams suggest that tests like PEEPS make more valid and thorough assessments of expressive phonology because the selection of target words is based on age-of-acquisition data from children’s vocabularies, as measured by parent reports like the McArthur–Bates CDI (Fenson et al., 1993). Production of target phonology is therefore less likely to remain undetected since the selected words are age adequate and familiar to the children and thus are more easily elicited. In addition, word selection based on phonological complexity sets a complexity goal against which the child’s production can be measured. This enables a graded assessment of phonological complexity rather than just a production matching the target word.

The word complexity measure (WCM, Stoel-Gammon, 2010) quantifies both the articulatory complexity of target production and the child’s actual word production—reflecting the notion that late acquired structures are typically more complex than those involved in early production. Thus, the WCM-based selection of words for PEEPS enables measurements of phonological production maturity and comparisons, both over time and between children. The different articulatory demands associated with different phonological parameters can be used as bases for measuring the phonological complexity of words (see example below).

Selection of words based on vocabulary data

The words included in the PEEPS test material are part of the vocabulary of typically developed children at 24 months of age (Stoel-Gammon & Williams, 2013) and are organized in two word lists. The basic word list (BWL) contains 40 words selected on the basis of normative data for age of acquisition (AOA) (Dale & Fenson, 1996). At the age of 24 months, all 40 words in the BWL are present in the expressive vocabularies of at least 50% of children, and 35 of the 40 words (87.5%) occur in the vocabularies of at least 75% of children. At the age of 21 months, 39 of the 40 words (97.5%) are present in the vocabularies of at least 50% of children, and 18 (45%) are found in the vocabularies of at least 75% of children. At 18 months, 20 of the 40 words (50%) are found in the production of at least 50% of children, and 5 of the words (12.5%) are present in the expressive vocabulary of at least 75% of children.

In addition to the BWL, an expanded word list (EWL), composed of the BWL and an additional 20 words, is included in PEEPS. At 24 months, all 60 words are present in the vocabularies of 50% of children and 39 of the 60 words (65%) are present in the vocabularies of 75% of children. At age 21 months, 50 of the 60 words (83%) in the EWL are present in the vocabularies of at least 50% of children, and 20 of the words (33%) are present in the vocabularies of at least 75% of children. At 18 months, 21 words are present in the vocabularies of at least 50% of children and 5 words (8.3%) are present in the vocabularies of least 75% of children. This is summarized in Table 1.

Table 1. Percentage of children aged 18, 21 and 24 months expected to have PEEPS words in their expressive vocabularies.

	Basic word list (n = 40)			Expanded word list (n = 60)
Percentage thresholds	18 months	21 months	24 months	18 months	21 months	24 months
≥50%	20 words	39 words	40 words	21 words	50 words	60 words
50–74%	15 words	21 words	5 words	16 words	30 words	21 words
≥75%	5 words	18 words	35 words	5 words	20 words	39 words

Based on CDI age-of-acquisition norms (Dale & Fenson, 1996). Data from Stoel-Gammon & Willams, (2013).

Phonological complexity of the test material

The phonological complexity of the words in the PEEPS lists is the sum of eight binary phonological parameters (zero or one) associated with each word, according to the WCM (Stoel-Gammon, 2010). In this way, each word gets a complexity score that is proportional to the presumed difficulty of articulating the word—a high-score word is considered to be more difficult to produce than a low-score one.

These phonological parameters are of various types. There are two parameters concerning word patterns (productions with more than two syllables or with stress on any syllable but the first), two parameters concerning syllable structures (productions containing word-final consonants or consonant clusters) and four parameters concerning sound classes (productions containing velar consonants, liquids/syllabic liquids/rhotic vowels, fricatives/affricates or voiced fricatives/affricates).

For example, if the word butterfly is uttered as [bʌdəfwaɪ], its production results in a complexity score of 3 because one point is given for the occurrence of more than two syllables, one for the occurrence of a consonant cluster and one for the occurrence of a fricative. In contrast, the target production of the same word [bʌɾɚflaɪ] results in complexity score of 5. In addition to the three parameters present in the first example, the target production also contains a liquid and a rhotic vowel, which earns two more points. In PEEPS, the BWL includes less complex words than those included in the EWL. Scores of the words in the BWL range from 0 to 4 and have a mean complexity score of 1.3. The words in the EWL have scores ranging from 2 to 5 with a mean complexity score of 3.0 (Stoel-Gammon & Williams, 2013).

A universal perspective on PEEPS

PEEPS is proposed as a tool for the assessment of phonological development in children who speak US English. Its test materials are based on data from parental reports and on the assumption that late acquired words tend to be phonologically more complex than earlier words. While this is an intuitively acceptable notion, the original PEEPS tool still lacks the principled independent phonetic and physiological motivations necessary for cross-linguistic comparisons.

The present work is a first step towards the expansion of the scope of the original PEEPS by adapting the test to a new language, while grounding the notion of phonological complexity on universal phonetic, biological and physical principles. It proposes for the creation and implementation of PEEPS-SE, a Swedish version of PEEPS, which is expected to hold the potential of detecting aspects of phonological development based on universal principles. It assumes that the child is born with certain biological affordances (vocal cords, lips, tongue, breathing muscles etc.) that can be used to explore aerodynamics for the production of human speech sounds. By definition, these universal principles are applicable to any natural spoken language in the world, but their manifestation in the developing child will inevitably be modulated by the child’s actual language(s) setting. PEEPS-SE is thus proposed as a test for the assessment of expressive phonology in Swedish-speaking children while enabling cross-linguistic comparisons based on universal phonetic principles.

Phonological development in Swedish children

Phonological development between ages 1:6 and 4:0 years is critically linked to simultaneous lexical and morphological development. During this period, the child’s emerging ability to analyse segments interacts with the development of the phonological system (e.g. Nettelbladt & Salameh, 2007). Swedish has a rich vowel system that uses length as a distinctive prosodic property. In Swedish-speaking children, most vowel phonemes are acquired before 3 years of age (Blumenthal & Jacobsson, 2013), but the front rounded vowels appear to be more difficult to produce and are acquired later (Linell & Jennische, 1980)—a developmental path following the frequency of these vowels in natural language’s vowel systems (Liljencrants & Lindblom, 1972). Fricatives demand critical airflow and articulatory control and are therefore expected to be acquired late in the developmental process. This is also observed for Swedish children, who may not master the production of fricatives in continuous speech even by about 4:6 years of age (Blumenthal & Jacobsson, 2013). Among Swedish children, /r/ is another phoneme that is acquired late and distinguishing between dental and supradental production of consonants is reportedly difficult (Linell & Jennische, 1980).

Purpose and aim

Adapting PEEPS to Swedish is a meaningful step towards creating materials for the assessment of early expressive phonology. There are currently no materials for assessing expressive phonology in children under 3 years of age. PEEPS-SE can therefore be expected to make a valuable contribution to the Swedish test materials for clinical use. The purpose of this work is to take the first step towards the development of a Swedish version of PEEPS, PEEPS-SE. The aim is to create the word lists for PEEPS-SE—a BWL and an EWL—and stay as close as possible to the original version of PEEPS by using the same sort of criteria for word selection, with respect to the vocabulary development of Swedish children and to Swedish phonotax. The two types of criteria are based on AOA and phonological complexity. In addition, general suitability and balance between word types is taken into account. To obtain Swedish words graded in AOA and complexity, which are also easy to elicit, the PEEPS-SE materials are based on Swedish CDI (SECDI) (Berglund & Eriksson, 2000) reports.

Method

First, the words were selected with respect to expressive vocabulary data for relevant age groups. Second, words were selected with respect to their phonological complexity. In addition, the selection of words was done in such a way that they would represent a wide variety of phonological forms, that there would be a balance between verbs and nouns convenient for a clinical test situation and finally that the words would be easy to elicit.

Selection of words based on vocabulary data

Vocabulary data from the SPRINT project (Språkinterventionsprojektet, a Swedish language intervention project, Swedish Research Council 2008-5094) (e.g. Eriksson, 2014; Marklund, Marklund, Lacerda, & Schwarz, 2015) were used for the selection of words based on AOA. Longitudinal data at ages 17–18, 20–21 and 23–24 months were collected using an online version (Marklund, 2009) of the Swedish version of the McArthur–Bates communicative developmental inventory (SECDI, Berglund & Eriksson, 2000; Fenson et al., 1993). See Table 2 for details.

Table 2. Description of participants contributing with Swedish vocabulary data.

	Age group
	17–18 months	20–21 months	23–24 months
Number of children	331	282	233
Age range in days	520–575	620–668	703–757
Mean age in days (SD)	545 (12.8)	646 (12.5)	728 (12.9)
Number of boys/girls	172/159	157/125	123/110
Number of children with hearing problems	1	2	1
Number of children with both native Swedish-speaking parents	272	234	191
Number of children with parent(s) having language difficulties	20	20	16
Number of children with sibling(s) having language difficulties	8	6	5

Data from the SPRINT project (Swedish Research Council 2008–5094).

To select words according to the AOA of the words, acquisition criteria analogous to those for original PEEPS were used (Table 3). There were not enough words with low complexity (see section below) fulfilling criterion 2 (words reported as present in at least 50% of the children at 18 months) in the Swedish material, so an additional criterion with lower percentage threshold was included (criterion 3). This resulted in seven lists, one per criterion, with a total of 387 commonly reported words, all reported as being present at ages 23–24 months. Out of these 387 words, 126 were reported as being present at ages 20–21 months and 36 words at the ages of 17–18 months. By the nature of the criteria, several words matched more than one criterion within each age group.

Table 3. Acquisition criteria for word selection to the BWL and the EWL, target number of words needed in order to match PEEPS original and number of words in the Swedish material that meet the criteria.

Age in months	Criterion number	Age of acquisition criteria from PEEPS original	Target n of words for the BWL	Target n of words for the EWL	Number of words meeting criterion
18	1	Words reported as present in ≥75% of the children	5	5	8
	2	Words reported as present in ≥50% of the children	20	21	22 additional (30 in total)
	3	Words reported as present in ≥45% of the children*	–	–	6 additional (36 in total)
21	4	Words reported as present in ≥75% of the children	18	20	25
	5	Words reported as present in ≥50% of the children	39	50	101 additional (126 in total)
24	6	Words reported as present in ≥75% of the children	35	39	162
	7	Words reported as present in ≥50% of the children	40	60	225 additional (387 in total)

*Not a criterion in original PEEPS.

BWL: Basic word list; EWL: expanded word list.

Selection of words based on phonological complexity

The next step in the selection process involved calculating the phonological complexity of the words matching the AOA criteria (see above). At this stage, 13 words were excluded (from the 387 above) since consistent complexity calculations are not possible, either because there are multiple valid outcomes of that word in the SECDI questionnaire (e.g. ‘barnets namn’ the child’s own name; ‘fil/filmjölk’, sour milk) or because there were several words representing an item in the questionnaire (e.g. ‘sitta pottan’, sit potty).

The phonological complexity of the remaining 374 words was calculated using the Swedish adaptation of the WCM measure, WCM-SE (Marklund, Marklund, Schwarz & Lacerda, in preparation) (Lacerda & Marklund, in preparation). There are 10 WCM-SE parameters that all refer to the phonotactic structure of the word, but the criterion stress on any syllable but the first specifically concerns prosodic contrast. For word patterns and syllable structures, the measures in WCM-SE are the same as in the original WCM (see above). The differences between the original WCM and WCM-SE concern sound class parameters. Affricates, rhotic vowels and syllabic liquids are not present in target pronunciation of Swedish and are therefore not included in the WCM-SE. Instead, two new parameters are added in the WCM-SE: one addressing the trill /r/ or /ʀ/, and another long front rounded vowels typical for Swedish -/ʉː/, /yː/ and /øː/. Each occurrence of either of these phonemes results in one complexity point.

The sound class parameters are used in the WCM as well as in WCM-SE to calculate phonological complexity of words, but sound classes are not specifically targeted in the criteria for word selection. In accordance to the original PEEPS, the average phonological complexity of EWL words is higher than that of BWL words. Complexity criteria and the number of words meeting them can be found in Table 4.

Table 4. Complexity criteria for word selection to the BWL and for the 20 additional words to the EWL needed in order to match the original PEEPS, and the number of words in the Swedish material that meet the criteria.

Criterion number	Complexity criteria from PEEPS original	Target n of words for the BWL	Target n of the 20 additional words for the EWL	Number of Swedish words meeting criterion
1	Consonant cluster in both initial and final position	0	1	1
2	Consonant cluster word initially	0	6	89
3	Consonant cluster word finally	0	6	26
4	At least one consonant cluster	3	8	174
5	Stress on any syllable but the first	2	0	20
6	More than two syllables	2	2	22

Final selection of words

The final selection of words for the BWL and the EWL was based on the seven criteria for AOA (Table 3) and the six complexity criteria (Table 4), taking into account a general balance between types of words and ease of elicitation. The steps of the selection procedure for the BWL and the EWL are described below.

AOA criterion 1, words reported as present in ≥75% of children at 18 months: Out of the eight words matching this criterion, six were selected for the BWL because they had low complexity and were considered easy to elicit. The two other words were rejected due to high complexity score and phonetic similarity to one of the previously selected words. The six selected words had a mean complexity measure of 0.83 (range 0–2).

AOA criterion 2, words reported as present in ≥50% of children at 18 months: Nine of the 22 words meeting this criterion were selected for the BWL based on their low complexity (range 0–4, mean 1.9) and that they were easily elicited. One of the words contained a consonant cluster in word-initial position and another had stress on the second syllable. In order to not include too many onomatopoetic words or to exceed the target number of words with consonant clusters, some of the remaining words were excluded. Other words were excluded because they were similar to previously selected words or because they were articles and thus difficult to elicit. At this point, 15 words had been selected for the BWL.

AOA criterion 3, words reported as present in ≥45% of children at 18 months: Since the target number of words meeting criteria 1 or 2 was 20, an extra condition with a lower percentage threshold was added. This criterion, 3, resulted in an additional six words, out of which five were selected for the BWL. Those had a mean complexity score of 1.2 (range 0–2) and were considered easy to elicit. The sixth word was excluded because it contained a consonant cluster. At this point, 20 words had been selected for the BWL.

For the extra 20 words in the EWL, no words were selected based on criteria 1–3.

AOA criterion 4, words reported as present in ≥75% of children at 21 months: Out of the 25 words matching this criterion, 13 were already selected for the BWL based on criteria 1–3. The target number of words in the BWL for this criterion was 18, but only 1 of the additional words was selected based on appropriate complexity level (2 points) and ease of elicitation. This word contained a medial consonant cluster. This resulted in a total of 14 words in the BWL matching criterion 4. The words that were excluded are onomatopoetic, contain consonant clusters or are similar to previously selected words. In total, 21 words had thus far been selected for the BWL.

Two words were selected for the EWL based on meeting criterion 4, each containing a consonant cluster (one in initial position). Both words had a complexity score of 2.

AOA criterion 5, words reported as present in ≥50% of children at 21 months: Out of the 101 words meeting this criterion, 7 had already been selected for the BWL based on criteria 1–3. In order to reach the target number of words (39), an additional 18 words were selected for the BWL based on low complexity scores (range 0–6, mean 1.9) and ease of elicitation, resulting in 25 words meeting criterion 5. One selected word contained a word-final cluster, meaning that the target number of words meeting the fourth complexity criterion was attained by the BWL. Two of the selected words were multisyllabic and had stress on the third syllables, meaning that the target number of words meeting the fifth complexity criterion was exceeded, and the sixth was reached for the BWL. The remaining words were excluded because the target number of words matching this criterion had been reached for the BWL. In total, 39 words had so far been selected for the BWL.

For the EWL, 16 additional words were selected based on meeting criterion 5, words reported as present in ≥75% of children at 21 months (range 0–8, mean = 3). Eight of those words contained at least one consonant cluster, meaning that the target number of words meeting the fourth complexity criterion for the additional EWL words was exceeded. Clusters were in initial position in five of those words. Two words had consonant clusters in final position, and one of those in both initial and final position, meaning that the target number of words fulfilling the first complexity criterion was reached. Two of the selected words had stress on the second syllable, so the target number of words meeting the fifth complexity criterion was exceeded. One of the selected words was multisyllabic.

AOA criterion 6, words reported as present in ≥75% of children at 24 months: All 39 words previously selected for the BWL were included in the 159 words meeting this criterion. The target number of words for the BWL based on this criterion was 35 and was thus already reached. However, one additional word (complexity score 2) was selected because it could be elicited using the same object as several previously selected words. The word in question was ‘bebis’ (baby), which can be elicited by a doll, just like, for example, previously selected ‘näsa’ and ‘öga’ (nose and eye). This means that all 40 words for the BWL had been selected and that criterion 7 had already been fulfilled.

For the EWL, one additional word based on AOA criterion 6 (words reported as present in ≥75% of children) was selected (complexity score 1). Based on AOA criterion 7 (words reported as present in ≥50% of children at 24 months), one last word was selected for the EWL (complexity score 6). It was multisyllabic and contained a consonant cluster, the fourth complexity criterion being further exceeded, and the target number of words fulfilling the fifth complexity criterion being reached. The second complexity criterion was not fully met, as five words with initial cluster were selected and the target was six. Nor was the third complexity criterion reached, as target was six words with final consonant clusters, and only two were selected.

Results

The 40 words selected for the BWL and the 20 additional words for the EWL are listed in Tables 5 and 6. Phonetic transcription, English translation and complexity scores measured with WCM-SE for each individual word are presented.

Table 5. List of the 40 words selected to the BWL: Swedish word, phonetic transcription, English translation and complexity score measured with WCM-SE.

Item	Swedish word	Phonetic transcription	English translation	Complexity WCM-SE
1	lampa	[ˈˈlampa]	Lamp	2
2	katt	[ˈkatː]	Cat	2
3	vovve	[ˈˈvʊvə]	Doggie	4
4	mu	[ˈmʉː]	Moo	1
5	hoppa	[ˈˈhɔpa]	Jump	1
6	elefant	[ɛlɛfˈant]	Elephant	6
7	apa	[ˈˈɒːpa]	Monkey	0
8	pip pip	[ˈpipip]	Tweet-tweet	1
9	docka	[ˈˈdɔka]	Doll	1
10	jacka	[ˈˈjaka]	Jacket	1
11	banan	[banˈɒːn]	Banana	2
12	kaka	[ˈˈkɒːka]	Cookie	2
13	macka	[ˈˈmaka]	Sandwich	1
14	vatten	[ˈvatːɛn]	Water	3
15	gunga	[ˈˈɡɵŋa]	Swing	2
16	mamma	[ˈˈmama]	Mommy	0
17	pappa	[ˈˈpapa]	Daddy	0
18	ajaj	[ˈajaj]	‘Ouch’	1
19	tack	[ˈtak]	Thank you	2
20	buss	[ˈbɵs]	Bus	2
21	bil	[ˈbiːl]	Car	2
22	bebis	[ˈbɛbɪs]	Baby	2
23	öga	[ˈˈøːɡa]	Eye	2
24	öra	[ˈˈœːra]	Ear	2
25	näsa	[ˈˈnɛːsa]	Nose	1
26	mun	[ˈmɵn]	Mouth	1
27	mage	[ˈˈmɒːɡə]	Tummy	1
28	blöja	[ˈˈbløja]	Diaper	2
29	nej	[ˈnej]	No	1
30	bada	[ˈˈbɒːda]	Bathe	0
31	kudde	[ˈˈkɵdə]	Pillow	1
32	napp	[ˈnap]	Pacifier	1
33	nalle	[ˈˈnalə]	Teddy	1
34	bok	[ˈbuːk]	Book	2
35	sova	[ˈˈsoːva]	Sleep	3
36	schh	[ˈɕ]	‘Hush’	1
37	pippi	[ˈˈpɪpɪ]	Birdie	0
38	tittut	[tɪtˈʉːt]	Peek-a-boo	3
39	telefon	[telefˈoːn]	Telephone	5
40	boll	[ˈbɔl]	Ball	2

In the phonetic transcription, pitch accent is marked with [ˈ] (acute) and [ˈˈ] (grave).

Table 6. List of the additional 20 words for the EWL: Swedish word, phonetic transcription, English translation and complexity score measured with WCM-SE.

Item	Swedish word	Phonetic transcription	English translation	Complexity score WCM-SE
1	sked	[ˈɧeːd]	Spoon	2
2	mjölk	[ˈmjølk]	Milk	5
3	köttbullar	[ˈˈɕøtbɵlar]	Meat balls	6
4	blomma	[ˈˈblʊma]	Flower	2
5	båt	[ˈboːt]	Boat	1
6	flygplan	[ˈˈflyːɡplɒːn]	Airplane	8
7	tåg	[ˈtoːɡ]	Train	2
8	hund	[ˈhɵnd]	Dog	2
9	ko	[ˈkuː]	Cow	1
10	kanin	[kanˈiːn]	Rabbit	3
11	gris	[ˈɡriːs]	Pig	5
12	mössa	[ˈˈmøsa]	Hat	1
13	strumpa	[ˈˈstrɵmpa]	Sock	4
14	äpple	[ˈˈɛple]	Apple	2
15	jordgubbe	[ˈˈjuːɖɡɵbə]	Strawberry	3
16	ägg	[ˈɛɡ]	Egg	2
17	fot	[ˈfuːt]	Foot	2
18	klocka	[ˈˈklɔka]	Clock	3
19	anka	[ˈˈaŋka]	Duck	3
20	ballong	[balˈɔŋ]	Ballon	4

The AOA criteria 1 and 5–7 (see Table 3) were met for the BWL in PEEPS-SE. Compared to the PEEPS original, one extra word was selected using criterion 1. Criterion 2 was not fulfilled because only 9 words were selected, resulting in a total of 15 words at this stage—compared to 20 in the BWL and 21 in the EWL in the PEEPS original (see Table 7). In order to find more words to select from, the new criterion, 3, with a percentage threshold of 45, was used. Criterion 3 made it possible to select more words and meet the target number of words (20) for the BWL, but 1 word was still lacking for the EWL. Another criterion not fully met in PEEPS-SE was 4. The target number of words was 18 in the BWL, but only 14 words were selected for PEEPS-SE. For the EWL, the target number of words was 20, but only 16 were selected. This was because the other words available in this criterion were not considered as suitable for either list (see ‘Final selection of words’ in section above).

Table 7. Number of words meeting age of acquisition criteria in the PEEPS-SE and the original PEEPS (in brackets).

	Basic word list (n = 40)			Expanded word list (n = 60)
Percentage threshold	18 months	21 months	24 months	18 months	21 months	24 months
≥45%	20 words (20*)	39 words (39*)	40 words (40*)	20 words (21*)	57 words (50*)	60 words (60*)
45–50%	5 words (–)	–	–	5 words (–)	–	–
50–74%	9 words (15)	25 words (19)	0 words (5)	9 words (16)	41 words (30)	2 words (21)
≥75%	6 words (5)	14 words (18)	40 words (35)	6 words (5)	16 words (20)	58 words (39)

*Note that the lowest percentage threshold in the PEEPS original is 50.

Out of the six complexity criteria, all were reached in the BWL and the second, third and fifth criteria were also exceeded by one occurrence each. In the additional 20 words for the EWL, the third criterion was not reached, meaning that the selected words do not contain as many word-final clusters as in the original PEEPS. The fourth and fifth criteria for the EWL words were exceeded, meaning that the material in total contains three more words with consonant clusters compared to the original PEEPS and also three more words with stress on any syllable but the first. For an overview of number of selected Swedish words and target number of words meeting complexity criteria, see Table 8.

The words selected to the BWL have a mean complexity of 1.7 (range 0–6, SD 1.3). The additional 20 words for the EWL have a mean complexity of 3.0 (range 0–8, SD 1.8). Figure 1 shows the distribution of the 40 words selected for the BWL and the additional 20 EWL words, according to their complexity scores.

Figure 1. Distribution of the 40 words selected for the BWL and the additional 20 EWL words, according to their complexity scores.

Among the words selected for the word lists, all Swedish consonant phonemes are represented, and several different consonant clusters are also included (see Table 9). Swedish is a language with a complex vowel system, and the Swedish long, front, rounded vowels /ʉː/, /yː/ and /øː/ are typologically less common among the world’s languages (Liljencrants & Lindblom, 1972). This is incorporated in the WCM-SE, as these vowels add complexity points to the words where they are present. Swedish vowels present in selected words are shown in Table 10.

Table 8. Number of selected Swedish words and target number of words for BWL and additional 20 EWL-words meeting complexity criteria.

Criterion number	Complexity criteria from PEEPS original	Selected n of words to BWL (targets)	Selected n of words to the 20 additional EWL-words (targets)
1	Consonant cluster in initial and final position	0 (0)	1 (1)
2	Consonant cluster word initially	1 (0)	6 (6)
3	Consonant cluster word finally	1 (0)	2 (6)
4	At least one consonant cluster	3 (3)	11 (8)
5	Stress on any syllable but the first	3 (2)	2 (0)
6	More than two syllables	2 (2)	2 (2)

Table 9. Swedish consonant phonemes and consonant clusters in different word positions in the words selected for the BWL and for the additional 20 words to the EWL.

Word position	Consonant phonemes in BWL words	Consonant phonemes in additional 20 words for the EWL	Consonant clusters in the BWL	Consonant clusters in additional 20 words for the EWL
Initial	[p] [t] [k] [b] [d] [ɡ] [m] [n] [v] [s] [ɕ]** [h] [j] [l]	[p] [t] [k] [b] [ɡ]* [m] [f] [s]* [ɧ] [ɕ] [h] [j] [r]	/bl-/	[kl-] [ɡr-] [str-] [mj-] [fl-] [bl-]
Medial	[p] [t] [k] [b] [d] [ɡ] [m] [n] [ŋ] [f] [v] [s] [ɕ]** [j] [l] [r]	[p]* [t]* [k] [b] [ɖ]* [ɡ]* [m] [n] [ŋ] [s] [j]* [l] [r]*	/-mp-]	/-ŋk-] /-mp-] /-pl-] /-ɖɡ-]
Final	[p] [t] [k] [n] [s] [ɕ]** [j] [l]	[t] [d] [ɡ] [n] [ŋ] [s] [r]	[-nt]	[-lk] [-nd]

*Only in a consonant cluster.

**Isolated consonant phoneme.

Table 10. Swedish vowels in different word positions in the words selected for the BWL and for the additional 20 words in the EWL.

Word position	Vowels in BWL words	Vowels in the additional 20 words in the EWL
Initial	[ø] [ɛ] [œ] [a] [ɒ]	[ɛ] [a]
Medial	[i] [ɪ] [e] [ø] [ɛ] [a] [ʉ] [ɵ] [u] [ʊ] [o] [ɔ] [ɒ]	[i] [y] [e] [ø] [a] [ɵ] [u] [ʊ] [o] [ɔ] [ɒ]
Final	[ɪ] [a] [ʉ] [ə]	[u] [a] [ə]

Discussion

The words in PEEPS-SE are a selection of the SECDI items reported by parents of children in the target ages and weighed by phonological complexity criteria. The vocabulary data from the SECDI parental reports provided a good base for word selection for the PEEPS-SE. However, while trying to maintain compatibility with the original PEEPS (Stoel-Gammon & Williams, 2013), some adjustments had to be made. For instance, to match the PEEPS original, the AOA criterion 2 (words reported as present in ≥50% of children at 18 months) was not strictly fulfilled, as onomatopoetic items and words with high complexity were discarded. Onomatopoeia is known to be salient in infant-directed speech (Laing, Vihman, & Keren-Portnoy, 2017), and onomatopoetic words are often identified among the child’s first utterances (e.g. Vihman, Ferguson, & Elbert, 1986). Yet, while onomatopoeia can provide useful phonological information, such words are typically not fully conventionalized as established lexical items, and therefore a variation in pronunciation of these items is to be expected. In fact, among the words fulfilling acquisition criteria 2, there were nine selectable onomatopoetic words, but only two that are commonly used: [ˈmʉː] (moo) and [ˈpipip] (tweet tweet) were selected for the BWL to preserve phonological reliability.

In order to find enough words meeting criterion 2, it was necessary to lower the percentage threshold so that five additional words could be included in the BWL. Initially, acquisition criterion 4 could not be fulfilled because there were too many onomatopoetic words among the available items, as well as many words with consonant clusters and words with phonological structures similar to previously selected words that would not provide added value to the test material. Instead, further selection of words was made from words available at the next level of acquisition. This means that PEEPS-SE contains some words acquired by fewer children at the target age (or acquired at an older age).

With respect to complexity criteria, all criteria for the BWL, and five of the six complexity criteria for the EWL, were reached. Additionally, three complexity criteria for the BWL and two for the EWL were exceeded, which means that PEEPS-SE contains more phonological complex words than the original PEEPS (Stoel-Gammon & Williams, 2013). The fact that it was not possible to fully meet two acquisition criteria and that three complexity criteria were exceeded could mean that PEEPS-SE is slightly skewed towards older children in comparison to the original PEEPS. However, the word selection process cannot be expected to be exactly applicable across languages. Whereas at an initial stage, phonetic universals are likely to be dominant, the impact of the phonotactics of the ambient languages increases as the child develops. At the same time, it is practically impossible to capture all the phonotactic aspects of the ambient language in a necessarily limited test material. For instance, according to Linell and Jennische (1980), some Swedish phonemes seem to be acquired first in medial position and then successively in other positions. Even if all Swedish phonemes are represented in PEEPS-SE (see Tables 9 and 10), they are not represented in all possible positions. A selection of words that fully represent the Swedish phonemes in all different possible positions could increase the probability of capturing even more structures in child production, but this was not possible to accomplish during the selection procedure while still staying close to the selection criteria.

Additionally, the current WCM-SE does not take into consideration weighting different structures in sequences or different positions (see Table 9). Indeed, a consonant or vowel segment in a word is always influenced by adjacent speech sounds, but the WCM-SE measure does not take into account coarticulation (Kent & Minifie, 1977). With regard to consonant clusters, the segments that are included are not only influenced by adjacent speech segments but also by the word form. For example, the word jordgubbe [ˈˈjuːɖɡɵbə] strawberry is a Swedish word created by compounding jord and gubbe. The cluster [ɖɡ], composed of the adjacent segments of the two word parts, is treated in PEEPS-SE as a genuine cluster, taking into consideration that the word’s etymology has been integrated and that the acoustic form is what the children are, in fact, exposed to.

Finally, the sounds occurring in connected speech production will also be affected by interaction settings (Lindblom, 1990), speech style, dialects, sociolects etc., which will colour the specific linguistic experience of the child. Against the complexity of the early phonological development process, PEEPS-SE (or any other practical test) can only capture a limited part of actual language development. Nevertheless, by explicitly targeting well-defined aspects of the process, in light of clear and grounded phonetic assumptions, PEEPS-SE offers the opportunity of valid insights into the child’s phonological development, including the deviant.

Standardization of PEEPS-SE for typical Swedish children in the same age ranges is currently in progress.

Conclusion

PEEPS-SE has the potential to be a useful tool for the assessment of expressive phonology in Swedish-learning children aged 18–36 months, and it also allows for a cross-linguistic comparison with the original PEEPS.

Declaration of interest

The authors declare that there is no conflict of interest regarding the publication of this article.

Acknowledgments

This work was supported by Stiftelsen Frimurare Barnhuset in Stockholm, Department of Linguistics at Stockholm University, Speech Clinic at Danderyd’s Hospital, by grants from Swedish Research Council (2008-5094) and Kinander’s Donation Scholarship. The authors would like to thank the families that participated in the SPRINT project (VR 2008-5094) for contributing with parent reports of child vocabulary size. Many thanks go to Ellen Marklund for constructive comments during the preparation of the manuscript.