Is a ‘Snowman’ a ‘Man Made of Snow’? Morphology Teaching through Children’s Stories

The acquisition of word formation processes is considered to be the necessary prerequisite for the mastery of the morphology of the mother language as well as vocabulary development and vocabulary learning and teaching (cf. Nagy et al. 2006; Nagy & Herman 1987; Templeton 1989). In addition, the acquisition of the morphological component of a language makes predictions regarding the acquisition of other components, such as the syntax and/or the semantics. Aim of this paper is to describe the main axes of a program of teaching the morphology of Greek, more specifically, compounding, through children’s stories and the results of its application in class. The main core of the program is a story accompanied by consolidation exercises. The results stemming from the application of the compounding story in Greek classes underline the fact that the experimental group scores as well as the control group with respect to the assimilation of compounding rules and principles. This entails that targeted children’s stories provide an effective and fast way of teaching the morphology of Greek as a mother language.


Introduction
The acquisition of word formation processes is considered to be the essential tool for the accurate use of the morphological component of a language, its vocabulary development and learning (cf. Nagy et al. 2006;Nagy & Herman, 1987;Templeton, 1989). This importance lies in the fact that the acquisition of morphology makes predictions regarding the acquisition of other grammatical components, such as the phonology, the syntax and the semantics of the language. Except for the acquisition of the morphophonological component, story reading and story retelling are thought to be suitable for oral and verbal language development in L1 (Brice-Heath, 1982;Connelly & Clandinin, 1990;Egan, 1989;Isbell et al. 2004;Morrow, 1985;Penno et al. 2002;Read, 2008) but also L2 (cf. Sinclair-Bell, 2002;Shyu, 2008;Tsou et al. 2006). However, none of the aforementioned studies target a specific vocabulary/ grammatical category.
In the present study, we explore how Greek word formation mechanisms are activated through a targeted teaching program whose main axe is a story made up of compound forms. In other words, we investigate the linguistic performance of native speakers with respect to a) the internal structure of compound forms as well as the relation holding between distinct compound constituents, b) the role of headedness and the linking vowel, c) the degree of compound constituent variation, i.e. the degree to which compound constituents positional alternation does not affect compound meaning. Compound mechanisms of Greek will be examined in combination to cross-linguistic factors affecting compound formation and which are strong enough to facilitate the production of compound forms. It is important to mention that, although we do not underestimate or undervalue the importance of linguistic input in compound formation, we will not take into consideration here.
The paper is organized as follows; section 2 discusses the central rules of Greek compounding as well as the main findings of research in Greek compounding. Section 3 presents the main axes of the Dyonomasia teaching material applied in class, while sections 4 and 5 discuss the findings from the implementation of Dyonomasia, the research tool used in this study. Finally, section 6 concludes the paper.
The major morphological distinction among compound forms is the one which refers to lexical and morphosyntactic compounds (cf. Drachman & Malikouti-Drachman, 1994 (2) a. pali-ό-filοs 'good old friend -NOM. SG. MASC.' b. pali-ο-fίλοs 'good old friend -NOM. SG. MASC.' 6. peδί -θάvma 'whizz kid -NOM. SG. NEUT.' Compounds of the [[stem + stem] + der. Suffix] and [word + word] types are lexical because they are one-word forms and are the product of word formation processes. On the other hand, compounds of the [word + word] type are morphosyntactic because they are the product of phrasal processes (cf. Malikouti-Drachman, 1995;Nespor & Ralli, 1994, 1996. As far as one-word lexical compounds are concerned, the only distinctive feature of the data in (2a) and (2b) is the position of stress. However, the data in (3) underline the fact that there are deeper differences.
More specifically, the compound head loses its lexical stress as well as its inflectional properties (inflectional ending) in (3a). In (3b) only stress assignment differs on the surface. However, the newly formed word in (3b) loses its inflectional ending -i and the new inflectional ending -i is assigned as the unmarked inflectional form (cf. Ralli, 2005). Finally, the head retains all its morphophonological characteristics in (3c). Therefore, the [[stem+ stem] + der. Suffix] compounds receive fully unmarked properties (cf. Ralli 2005Ralli , 2007. The head in Greek is located at the right edge of the compound form, consequently, it is the second element of the compound word. Compound right headedness is a cross-linguistic bias (cf. Williams, 1981). Compound elements are distinguished, first, in coordinate and subordinate as far as the order of compound elements is concerned, and, second, in endocentric and exocentric, as far as the relation of compound elements is concerned (cf. Aronoff & Miller, 2001;Ralli, 2007). In coordinate and exocentric compounds there is no clear head and this is the reason why the position of compound elements may vary (4a, 4b), while in subordinate and endocentric compounds the head of the word defines the semantics of the newly formed word (4c, 4d) and the position of the compound constituents cannot vary. For this reason, (4d) is an ungrammatical, i.e. non-acceptable, word.
Finally, the linking vowel, which mainly takes the shape of vocalic -o -, is thought of as a relic from ancient Greek (Anastasiadi-Simeonidi 1983; Ralli & Raftopoulou 1999;Ralli 2005Ralli , 2007). It appears across-the-board when the second compound constituent starts with a consonant (5a) and is absent when the second compound constituent starts with a vowel (5b). It always appears when it is stressed (5c).
'red and white -NOM. SG. MASC. ADJ.' In an experimental task which tests the formation of existing and non-existing compound forms in Greek through two off-line questionnaire it has been shown that native speakers do not make errors in the formation of existing compounds (T1) with respect to the major compound properties discussed above, namely, compound internal structure, headedness and the linking vowel, while variation is attested in the formation of non-existing compounds (cf. Tzakosta, 2009). More specifically, adults native speakers, equally produce [[stem + stem] + der. Suffix] and [stem + stem] existing compounds, while they prefer to form [stem + word] non-existing compound types. Variability in the selection of compound type is also reported in Kalligiannaki & Tzakosta (2013); Tzakosta & Manola (2012). More specifically, Kalligiannaki & Tzakosta (2013) show that preschool native speakers of Greek prefer [stem + word] rather than [[stem + stem] + der. Suffix] compounds, while Tzakosta & Manola (2012) claim that preschool and early primary school children prefer [[stem + stem] + der. Suffix]. This implies that these two compound types carry equal perceptual loads, therefore, they can be used interchangeably and without difference in meaning.
Data from Albanian, Bulgarian Romanian, Russian, Swedish and Turkish second language learners of Greek, on the other hand, exhibit patterns to those of native speakers, in principle, irrespective of whether Greek (L2) is typologically adjacent to the native language (L1) (Tzakosta, 2011a, b;Kalligiannaki & Tzakosta, 2013;, 2020Tzakosta & Mamadaki, 2013;Tzakosta, 2017). It has been discussed that compound perception and production are highly influence by compound formation rules (cf. Tzakosta, 2017). It has also been shown that the major properties of compound formation are important perceptual cues, which, in turn, drive production, and need to be the main axes of the teaching of compounding in order to avoid the activation mnemonic mechanisms and mnemonic knowledge. Moreover, the order of teaching is suggested to rely on the frequency of emergence of these cues (Kalligiannaki & Tzakosta, 2013;Tzakosta, 2011a, b;, 2020Tzakosta & Mamadaki, 2013;Tzakosta, 2017). In chapters 3 and 4 we test the applicability of this claim.

Dyonomasia & Akiarositsa: The Journeys of a Centipede in Grammarland
Aim of this chapter is to discuss the first findings stemming from the application of the teaching program which was created by Sinodi and Tzakosta (2014a, b) with the aim to target learning and teaching of specific word formation mechanisms. Sinodi and Tzakosta's work is based on the findings and claims made by Kalligiannaki & Tzakosta (2013); Tzakosta (2009Tzakosta ( , 2011aTzakosta ( , 2011bTzakosta ( , 2017; Tzakosta & Mamadaki (2013) according to which preference for specific forms is defined by mnemonic rather than word formation mechanisms in existing words, whereas, variation is attested in non-existing newly formed words. More specifically, Sinodi & Tzakosta (2014a, 2014b produced two distinct teaching materials with which are intended to accurately evaluate the factors, the principles and the conditions which govern and, at the same time, facilitate language learning and teaching. The teaching materials considers the fundamental aim of the preschool curriculum, i.e. that teaching is expected to take place in a playful manner (cf. Read, 2008).
The two teaching materials created by Sinodi and Tzakosta (2014a, b) are called Dyonomasia and Aki-aros-tsa. Each of them describes and targets a morphological aspect of Greek. More specifically, Dyonomasia is a teaching tool for word compounding and Aki-aros-itsa is a teaching tool for word derivation. Compounding and derivation are two distinct "journeys of a centipede in Grammarland". Grammarland is supposed to be a land where a traveler can make several stops. These stops are the different grammatical phenomena. Both teaching materials are driven by the same (teaching) 'philosophy'. In other words, each program is based on a story made up of real and novel words in order to test the degree to which word formation is due to mnemonic strategies or the productive application of word formation rules.
The central heroes of the story are two little animals, Roussa, a centipede, and Νoula, a kitten. Roussa and Noula become friends and they travel around Grammarland, they make new friends and enrich their linguistic knowledge. The story has been chosen as the best tool for the evaluation of language knowledge evaluation since it offers a context of meaningful communication. Story reading and story retelling are suitable for oral and verbal language development (Egan, 1989;Isbell et al., 2004;Morrow, 1985). The story is accompanied by a guide addressed to parents and language instructors as well as a set of 10 suggested linguistic activities (see pictures 1 and 2). The linguistic activities take the shape of word matching, word combination, filling in the blanks, choosing the correct form. Dyonomasia and Akiarositsa are directed to children with an age range between 4 -8 years who develop their skills in morphological and vocabulary development.

Some Preliminary Findings
Dyonomasia was first implemented in two groups of monolingual native speakers of Greek. The first group (Group 1 -G1) consisted of 54 monolingual preschool children who are native speakers with typical language development, and the second group (Group 2 -G2) consisted of 27 monolingual preschool children who are native speakers with atypical development. Dyonomasia was implemented in three phases. During Phase 1, the kindergarten teacher first read the story aloud and then the whole class talked about compound and derived words, their definition and their characteristics. During Phase 2, children were asked to draw a scene from the story in which compound forms should be central (cf. pic 3-5). This way we would be able to test whether compound forms were actually correctly comprehended and perceived. Children then made a poster with their compound forms drawings (pic. 6). Finally, during Phase 3, children were asked to form compound forms by means of a picture naming task (pic. 7). The picture is quite the opposite after Phase 3 of the application of Dyonomasia. More specifically, the rate of unsuccessfully formed compounds drops for both G1 (32%) and G2 (31%), while successfully formed compounds is quite satisfying (G1, 68% -G2, 69%). It is important to note that atypically developing children score better than typically developing ones. As discussed in earlier work of Tzakosta & Stavgiannoudaki (2013), atypically developing children play serious attention to rules, in that case, compound formation rules, and, as a result, they score slightly better than typically developing children. Such results highlight the role that rules may play in the language teaching process.

The Present Study The Tool
Dyonomasia is the teaching material also implemented in this study since its aim is the evaluation of the degree of accurate application of compounding rules. Dyonomasia tests the productive use of existing as well as non-existing compound forms. Dyonomasia's name is made of the fundamental rule of compounding; a compound form is word made up of at least two lexical constituents. Dyonomasia consists of three parts: a story, a set of 10 representative teaching/ practice activities and a guide directed to parents and educators. Given that the use of real words may be attributed to mnemonic strategies, the accurate formation of novel words certifies the acquisition and productive use of word formation rules. The program was realized in three phases. In Phase 1, participants had to a) read the story, b) talk about its content, the derived words found in it and their properties, and, c) make a list of the derived forms of the story. In Phase 2, children participants were asked to a scene from the story (Pic. 8-11).

Pic. 8. A drawing of the village of Dyonomasia
Pic. 9. A drawing of a hotel Pic. 10. A drawing of a centipede Pic. 11. A drawing of a centipede Finally, in Phase 3, participants took part in various linguistic activities which took the form of picture naming tasks, close test and word-matching (Pic. 12).

Pic. 12. A word-matching activity of Dyonomasia
The three intervention phases were preceded by a pre-intervention phase (Pre-IP) during which the participants were indirectly asked to recognize and form compound forms by answering questions like 'what is a snowman?' or 'or how is a man made of snow called?'. The three intervention phases were followed by a post-intervention phase (Post-IP) during which the EG2 participants were tested in the formation of derived forms through filling in a questionnaire. Participants were tested individually. Data were analyzed using the IBM SPSS statistics packet.

Participants
In this study, Dyonomasia was implemented in a group of 147 monolingual preschool children (age range: 5-6 years) (Experimental Group, EG), because we aimed to test the effect of the program before the critical period age limit (7 years) (cf. Lenneberg, 1967). Α group of 40 adults (age range 18-50 years) all of whom had attended tertiary education served as the control group (Control Group, CG). The CG participated only in the pre-IP.

Results and Discussion
In graphs 1-8 we present the results of compound formation and the accurate application of compounding rules by adults and children in the pre-IP. We observe that both adults and children do almost equally well in the formation of existing and non-existing compounds, though the scores are not very high for either group (Graphs 1 and 2). Scores are higher for existing compounds (52,56%, for adults, 47,66%, for children) compared to non-existing compounds (45,04%, for adults, 41,07%, for children). Higher scores of existing compounds are due to the activation of mnemonic mechanisms, while purely word formation mechanisms are activated in novel compounds (cf. Tzakosta, 2017 and more references therein). However, though comparable, scores for children and adults have different explanations. More specifically, adults seem to avoid the formation of compound words, while preschool children show satisfactory results in compound formation during the process of acquisition/ full development of the mother language (cf. Tzakosta & Manola, 2012). In this context, children's failure to produce compound words is also reported (7,90%).

Graph 1. Pre-IP adults -compound formation
Graph 2. Pre-IP Child-compound formation Graphs 3 and 4 illustrate adults' and children's preference for certain compound types. More specifically, adults seem to show slight preference for [[stem + stem] + der. Suffix]], like 'lemon-ό-dasos' forms (54,67%), in which stress is readjusted (in accordance with Tzakosta, 2009), while children more clearly prefer [stem + word] compounds (67,36%), like 'lemon-odάsos' (in accordance with Tzakosta & Manola, 2012). This preference is expected for children, since children prefer to produce a form in which the morphophonological characteristics of at least one of the compound constituent, i.e. the second one which also happens to be the head of the word, are fully retained (this happens in the case of 'lemon-o-dάsos').
Graph 3. Pre-IP adults -compound types Graph 4. Pre-IP children -compound types The findings illustrated in graphs 3 and 4 are further supported by the graphs 5 and 6 in which it becomes more obvious that children tend to produce compound words in which the second constituent is stressed (75,61%, as opposed to 52,57%, for the adults). In other words, stress readjustment/ relocation is not preferred (38,34%, for adults, 16,76%, for children).
Graph 5. Pre -IP adults -compound stress Graph 6. Pre -IP children -compound stress With respect to headedness, both adults and children assign headedness to the right constituent of the compound form (87,74%, for adults, 82,16%, for children). Therefore, the rightmost constituent carries the semantic load of the compound form. In general, both the CG and the EG accurately implement the rule of right headedness. In addition, children produce more coordinate compounds compared to adults (4,16%, for adults, 8,53%, for children).
Graph 7. Pre-IP adults -compound heads In the above as well as the following graphs, there is none which depicts the emergence of the linking vowel in compound forms. Linking vowels are used without exception in all produced forms. Total results and graphs for both CG and EG are provided in appendix 1.
Graphs 9-12 depict the pre-IP scores of the EG and CG with respect to the above tested categories, i.e. compound formation (Graph 9), compound types (Graph 10), compound stress (Graph 11) and compound headedness (Graph 12). Both groups' scores are in the same direction for the compound formation and the compound headedness categories (Graph 9, Graph 12). Children's scores are lower than those of the adults but this is expected since the process of acquisition of the mother language is not completed for children.
Graph 8. Comparative per group performance on compound forms Graph 9. Comparative per group performance on compound types Graph 10. Comparative per group performance on compound stress Graph 11. Comparative per group performance on compound heads In Graphs 13-20, we compare the pre-IP and post-IP scores for EG. For the ease of reading, the pre-IP Graphs 2, 4, 6 and 8 are re-written as Graphs 13, 15, 17 and 19. It appears that there is improvement in EG's scores for most tested compound categories in the post-IP. More specifically, the scores for the formation of existing compounds (Graphs 13, 14) raise in the post-IP, even though children seem to be more hesitant in the formation of non-existing compounds after the intervention. However, it is important to note that the rate of idiosyncratic compounds is slightly higher in post-IP. We assume that this happens because children seek various ways of forming compound words. One type of idiosyncratic compounds is periphrastic compounds, like 'άnthropos apό xiόni = a man made of snow' instead of 'xionάnthropos = snowman'.
Graph 12. Pre-interv. Child -compound formation Graph 13. Post-interv. Child -compound formation The above findings verify our initial hypotheses that a) targeted language teaching material raise children's learning interest and b) children's stories provide an effective way of language teaching. The results showed that targeted teaching materials successfully evaluate the level of the speakers' language knowledge, locate speakers' weaknesses and provide a satisfactory way to restore them.

Conclusions
In this paper we tested the results stemming from the application of a teaching program which places emphasis on the formation of compound forms. The results displayed that teaching materials which target specific grammatical phenomena get clear results with respect to the participants' understanding of the phenomena and the degree of rule learning which underlie the tested grammatical phenomena. Our findings further underline the fact that children's stories provide a natural and effective way of language teaching, in general, and teaching the morphology of Greek as L1, in particular. The presented teaching program highlights the importance of specialized and focused children's stories for the acquisition of the morphological component, vocabulary development and development of the mother tongue, in general.