CHAPTER 1: INTRODUCTION
1.1 TOPIC AREA
Child-directed speech (CDS) has been central to research ever since Noam Chomsky declared it to be a ‘degenerate’, ‘deficient’, ‘impoverished’ form, (Fletcher & MacWhinney, 1995) stating children could not learn the rules of a language by hearing such complex input. Other studies have shown that adult input is by no means as complex as Chomskyan theories had assumed. Such studies have observed that adult-child interaction is somewhat different from adult-adult interaction, giving rise to the finding that adults generally adapt their speech when talking to children, which is termed ‘CDS’ or ‘motherese’ as it is otherwise known. Some common features have been attributed to this unique speech register. These features are said to include shorter sentences, clearly segmented slower speech, phonologically simplified utterances, restricted vocabulary, exaggerated prosody, repetitions and expansions. The language used is said to be constrained to ‘the here and now’ and related to the child’s focus of attention and ongoing activity (Harley, 2008), which all in all result in effective communication between parents and their children and also contribute to the speed and ease of a child’s language acquisition (Snow 1972).
1.2 FOCUS OF STUDY
As child-directed speech is often termed ‘motherese’ it gives a misleading impression that fathers have a negligible impact upon child language development. Hence, why the verbal environment provided by the father has been largely ignored until recent years. However, the ever-changing family roles and changes in typical male-female stereotypes in western society have influenced a change in the nature of parenting, which has given rise to the introduction of research into paternal input to children. The late twentieth century has seen an increase in fathers adopting the primary caregiver role, which has led to the popularity of ‘stay-at-home dads’. While tending to the immediate needs of children was traditionally considered to be a female responsibility, nowadays that is not the case as it is becoming increasingly popular for mothers to be in employment. Therefore, a number of studies since the 1970’s have discovered fathers as well as mothers produce the typical modifications of CDS in their speech to children, hence the suggestion that males provide an equally large facilitation to child language development as females (Berko-Gleason 1975). The scope of the literature in this area is somewhat limited, however research has indicated that the most important features of CDS are maintained by paternal input; simplicity, well-formedness, repetition and immediacy, (Berko-Gleason 1975) which has given rise to the newly-coined term ‘fatherese’. Nevertheless, there is an inconsistency in the findings of the studies in this domain.
1.3 RESEARCH QUESTION
The research question central to this dissertation is do fathers use the same features of child-language as mothers and how does parental usage of CDS compare. The focus will consider the parental input to two language-learning siblings, at different stages of language development.
1.4 STRUCTURE OF STUDY
Following this introduction, a literature review addresses the findings of numerous existing studies in the field of gender-specific child-directed speech. The methodology section explains how this investigation was carried out, including a description of the subjects observed, the methods of data capture, transcription and the variables used for analysis, followed by a description of the results gathered in the investigation and a discussion of the findings and problems encountered throughout the study. To conclude the investigation, the outcome of the study will be related back to the review of literature in order to address how the findings fit in with what is already known in the field of gender-specific CDS.
CHAPTER 2: LITERATURE REVIEW
2.1 FIRST STUDY IN THE FIELD
The research of Jean Berko-Gleason (1975) was the first in the area of CDS to consider a paternal contribution. She conducted a study to determine whether the defined features of CDS were limited to the speech of mothers or if they could be characterised as a function of adult language to children. Before the conduction of the study it was of question whether there was such a thing as men’s speech to children at all, as a previous study made a bizarre statement that men only spend an average of 37.7 seconds per day engaged in speech interaction with their children (Berko-Gleason 1975). Berko-Gleason overruled this finding stating fathers do talk to their children, but her research was not solely orientated around paternal input. She discussed unpublished studies into the speech of mothers and fathers in interactions in their home settings and also reported upon studies of interactions in a day-care setting, exploring more broadly the speech of ‘non-mothers’. When addressing children in a home setting, the research indicated that there are some similarities in the speech styles of male and female adults, but it also asserts that differences arise due to the father’s role. It was asserted that fathers perform many of the characteristics attributed to “motherese” such as restraining their speech to the ‘hear and now’, and considerably simplifying the length of their speech, as fathers were found to use a similar mean length of utterance (MLU) to mothers. However, it is asserted that mothers are more sensitive to their child’s ages in families of more than one child, stating they directed less complex utterances to their younger children and more complex utterances to the older of the siblings (Berko-Gleason 1975). An instance where a father addressed the younger of his children with a more complex utterance is mentioned, suggesting a lack of sensitivity on paternal behalf. The study concerned also distinguished between the types of sentences used by each of the parents, generalising in a qualitative sense that fathers use more direct imperatives and produced more threats as well as rarer vocabulary. This more frequent use of rare vocabulary could also suggestively contribute to the judgement that fathers are less sensitive than mothers. The research gathered in the interactions of day-care teachers further supported the findings regarding sensitivity, as the data observed an unexpected lexical usage by a male teacher towards a three year old. This suggests that weaker sensitivity is characteristic of ‘male’ language towards children as appose to the more constricted ‘father’ language. Berko-Gleason asserts that even though “fathers are not as well tuned-in to their children as mothers are in the traditional family situation: they do not have to learn to attend to subtle signals from the child, and frequently have no penalty to pay for any lack of attention…there are probably serious and far-reaching effects that result from the fact” (1975; 293). She also examined a study of gender-specific CDS in a storytelling situation, in which fathers were said to concentrate on the activity of telling a story rather than using the story to facilitate interaction with the child, which was characteristic behaviour of mothers. The mothers in this study were reported to ask a lot of questions to ensure their child fully understood the activity. Berko-Gleason maintained that “the children had to exert themselves more for the fathers, and try harder to make themselves both heard and understood. In this way, fathers can be seen as a bridge to the outside world, leading the child to change his or her language in order to be understood” (1975; 293). This gave rise to the ‘Bridge Hypothesis’ proposed by Berko-Gleason; (Dato 1975; 294) maintaining that speaking to fathers who are less sensitive than mothers in terms of language use, provides children with the linguistic skills required for talking to strangers and people in more abstract contexts.
Contextually speaking, Berko-Gleason notes that “the fathers’ language clearly demarked their role within a family: a father playing with his small son might break off the game to send the child to his mother to have his diaper changed” (Dato 1975: 291).
2.2 – DIFFERENTIAL EXPERIENCE HYPOTHESIS
Similar to the Bridge Hypothesis in terms of sensitivity is what is known as the ‘Differential Experience Hypothesis’, (McLaughlin, White, McDevitt & Raskin 1983, Lewis & Gregory 1987) which is theorised on the basis of findings that mothers provide more linguistic support for their children due to the fact they are more attuned to the child’s needs and abilities. Fathers, on the other hand, are seen to be less sensitive to children’s’ capabilities, which sees them being more linguistically demanding than mothers. This hypothesis maintains that fathers instigate a greater performance from children due to their lack of sensitivity. However it does not insinuate that fathers are better language facilitators than mothers, on the contrary, that the functions of each of the speech styles give equal contributions to child language development, in the sense that they offer experiences of a differing nature.
In sum of the above hypotheses, mothers and fathers are suggested to engage in different kinds of interactions with their children. It is not to be believed that one of these speech styles is in any way superior to the other, they are viewed in a complementary manner to one another and interpreted to manifest and reflect each of the parental roles (Chanu & Marcos 1994).
“The mother’s specific role is to provide a feeling of security by avoiding situations where the child’s established acquisitions would be challenged, while still stimulating the child. The father’s specific role is to prompt the child to attain higher levels of success, even if it means momentarily destabilising the child” (Chanu & Marcos 1994; 3).
Due to these observed differences in parental speech behaviour in terms of CDS, the communicative behaviour of children should also be expected to differ when conversing with mothers and fathers.
2.3 – FINE-TUNING HYPOTHESIS
Many studies (Snow 1972, Berko-Gleason 1975, Sokolov 1993) have found that mothers seem to ‘fine-tune’ their speech when talking to young children. Cross (1977) proposed the ‘Fine-Tuning hypothesis’ based on correlations between the measures of maternal input structure and child competence. It has been theorised that mothers adjust the length and complexity of their utterances in line with the increase in their child’s mastery of linguistic competence. This implies that parent’s decrease their use of CDS as their child’s linguistic ability develops. This is observable in terms of mean length of utterance (MLU) as it is expected that parental growth in the use of word classes and word order will occur in accordance with the growth of child comprehension and production levels. Cross observed that individual differences were found to reflect the speech styles of mothers in some cases; however, statements have been made that a mother more closely ‘fine-tunes’ her language to the child than any other family member. It is a possible point of analysis in this study to test these statements in order to see how the MLU of mothers and fathers compare.
It has been noted that mothers ‘fine-tune’ their speech to young children in more ways than one. As well as lexical and structural adjustments, prosodic adjustments are also said to be found. Prosodic fine-tuning is said to be marked by higher pitch and exaggerated intonational patterns which appeal to infants’ attention patterns (Fletcher & MacWhinney, 1995). “Manipulation of these prosodic characteristics is very high at precisely the age when infants are most responsive and by age five children receive almost no prosodic adjustments” (Fletcher & MacWhinney 1995, p.182). Such adjustments are said to be tuned to the child’s responsiveness and attentiveness whereas phonological and syntactic adjustments are tuned to the child’s production and comprehension levels respectively. Phonetics are said to be adjusted from the one-word stage onwards and include enhanced clarity of vowels and full production of often-reduced consonants. (Fletcher & MacWhinney, 1995).
2.4 – TOTAL LANGUAGE PRODUCED
In terms of analysing how much mothers and fathers speak to their young children in mean number of utterances, there is a general agreement that mothers speak more than fathers on the whole (Golinkoff & Ames 1979, Rondal 1980, Davidson & Snow 1996, Pancsofar & Vernon-Feagans 2006). However, McLaughlin et al (1983) and Lewis and Gregory (1987) found no significance in mean number of utterances.
Golinkoff and Ames (1979) found situation to have a bearing on conversational turns. They recorded parents in dyadic and triadic situations, reporting fathers to produce half as many utterances and take fewer conversational turns in a free-play situation with the mother present. However, in a dyadic play situation, mothers and fathers were reported to produce the same number of utterances and take the same number of turns. McLaughlin et al (1983) found parents to take relatively equal conversational turns while Rondal (1980) proposed that mothers take more turns.
2.5 – STRUCTURAL AND LEXICAL ASPECTS
The complexity of the parents’ sentences can be measured by making comparisons between their mean length of utterance (MLU) and number of verbs per utterance. However, there is large differentiation in the results regarding their MLU. Giattino and Hogan (1975) carried out the first published study in the field of ‘fatherese’. They provided a father-only speech analysis with which they made comparisons to previously reported investigations of mother-child data of the same nature. For the means of comparison for MLU, they recorded the father in adult-adult interaction, in which his MLU was recorded as 9.7 words. In his interaction with the child, his MLU was recorded as 5.2 words which was found to be closely correlated to the child’s MLU of 4.5 words. This evidence supports the finding that the father was aware of the child’s level of comprehension, which in turn influenced his language as he directed considerably shorter sentences to her than he did in adult-adult conversation. Discrepancies occurred in the conflicting results regarding MLU. Some studies found that mothers and fathers have similar MLU (Golinkoff & Ames 1979, Lewis & Gregory 1987, Pancsofar & Vernon-Feagans 2006) while other studies found that mothers produce a significantly longer MLU (McLaughlin et al 1983, Davidson & Snow 1996). Rondal (1980) supported the ‘Bridge Hypothesis’ and the ‘Differential Experience Hypothesis’ with the finding that although fathers’ speech was found to be shorter in length, the longest utterance in the study was also addressed by a father, portraying the lack of sensitivity central to the hypotheses. McLaughlin et al (1983) also reported that although the utterances spoken by mothers were significantly longer, they were more ‘well-tuned’ into the child’s abilities, also in support of the hypotheses.
Lewis and Gregory (1987) and Pancsofar & Vernon-Feagans (2005) are in agreement that fathers use fewer verbs per utterance. This is troublesome evidence as this variable is said to contribute towards complexity as it is evidence of low sensitivity, meaning it shows conflict with the ‘Bridge Hypothesis’ that fathers are less sensitive than mothers.
In the means of vocabulary, Davidson and Snow (1996) asserted that mothers talked more complexly, in that they used more low frequency words. They also stated that children spoke more complexly themselves in maternal dyads, showing a greater use of low-frequency vocabulary than in paternal dyads. Previous studies, the ‘Bridge Hypothesis’ and the ‘Differential Experience Hypothesis’ all undertake the belief that fathers create a more linguistically challenging environment for the child, however this study has shown that this is not always the case as it has proven an instance where mothers have provided a more sophisticated input than fathers. It was assumed that the mothers’ linguistically challenging behaviour in this study had prevailed over the stereotypically female behaviour of ‘fine-tuning’. This was attributed to the mothers’ advanced scholarly background as they were said to be as highly educated as the fathers (Davidson & Snow 1996).
Lexically speaking, there is said to be little difference in the speech of mothers and fathers measured by the type token ratio (TTR) (McLaughlin, Schutz and White 1980, Ratner 1988, Pancsofar & Vernon-Feagans 2006). Rondal (1980) found the speech of fathers to be more diverse, and Ratner (1988) in a more detailed analysis of vocabulary, found fathers to be more lexically demanding through their frequent use of rare nominal words and infrequent use of common nouns. Both of these findings are in support of Berko-Gleason’s theory that the linguistic style of fathers provides children with a ‘bridge to the outside world’.
Giattino and Hogan (1975) stated that declaratives were used in 35% of the corpus, interrogatives 34%, exclamatory sentences 9% and imperatives 6%. Giattino and Hogan (1975) and Golinkoff and Ames (1979) are in agreement that mothers and fathers use these sentence types to similar proportions.
Further conflicting evidence has been found in the area of questions. Some studies found that mothers and fathers ask the same number of questions, (Davidson & Snow 1996, Pancsofar & Vernon-Feagans 2006) whereas other studies found that mothers ask more questions, (Lewis & Gregory 1987) although other studies concluded that fathers ask more questions (McLaughlin et al 1983). Such contradictory findings are difficult to deduce an inference from. The ‘Bridge Hypothesis’ maintains that fathers are more challenging interlocutors than mothers, therefore in the means of interrogatives it is to be expected that fathers ask more wh-questions (questions that require a more elaborate response) than yes/no questions (questions that require the child to answer with a one-word answer ‘yes’ or ‘no’). Several studies (Giattino & Hogan 1975, McLaughlin et al 1983) support this finding, however there are studies opposing this evidence (Lewis & Gregory 1987, Pancsofar & Vernon-Feagans 2006). Wh-questions are said to be a challenge for children as they require the child to construct a lexical response rather than repeat the parent’s structure or simply give a non-verbal response (Berko-Gleason 1975). In terms of question types, Golinkoff and Ames (1979) and McLaughlin et al (1983) agreed that yes/no questions are asked more frequently overall than wh-questions. This makes sense in the respects that mothers speak more than fathers and mothers are more inclined to ask yes/no rather than wh-questions.
“Many researchers have studied ‘language-teaching’ aspects of parental speech. These include explicit educative behaviours such as corrections, expansions and self repetitions. Although all parental communicative behaviours are ‘educative’, considering a child can learn by observation and imitation, these specific behaviours manifest the parents’ intentional effort to teach their child” (Chanu & Marcos 1994; 7).
Some research has concluded that fathers use repetitions more frequently than mothers (McLaughlin et al 1983, Lewis & Gregory 1987) while others have concluded the reverse (Ratner 1988). Giattino and Hogan (1975) found that repetitions made up 9% of their corpus. These were said to always be repetitions of the child’s preceding utterance, not self-repetitions. When compared to a set of previously recorded female-child data, a difference was realised in the respect that mothers’ repetitions are repetitions of themselves. The conflicting findings of Golinkoff and Ames (1979) recorded that both genders use repetitions to the same frequency, and that when they occur they are always repetitions of themselves not their children. They stated repetitions are more likely to be found when requesting action rather than giving information. Giattino and Hogan (1975) found very few instances where the father used corrections and Rondal (1980) found that mothers correct their children more than fathers. While comparing their data to data from previous investigations, Giattino and Hogan (1975) found that fathers rarely used grammatically incomplete sentences where as mothers are far more likely to do so. They found very few instances where completion sentences were used and said that expansions made up a mere 0.5% of the corpus whereas they contributed 30% to the previously conducted investigation of female CDS. The explanation attributed to the low frequency of expansions in the male corpus regarded the child’s production level. As the child was considered to be linguistically fluent, the need for the father to expand her utterances was eliminated.
2.6 – FUNCTIONAL AND CONVERSATIONAL ASPECTS
Berko-Gleason (1975) found trends in the studies he examined, in that fathers produce more requests for clarification. This finding was later supported by Rondal (1980). This suggests that fathers do not understand their children as well as mothers, possibly a consequence of fathers who assume secondary caregiver position due to their employment status.
Research which has focussed on directives separates imperatives; the most direct form of directives, from interrogatives; an indirect form.
“The use of more direct or indirect forms of directives challenges the child’s comprehension level to differing degrees. When a parent uses a direct form (‘be quiet’) it is much easier to understand the communicative intention than when a parent uses an indirect form (‘can we reduce the noise level in here?’)” (Chanu & Marcos 1994; 8).
In agreement with Berko-Gleason (1975), several studies found that fathers use more direct imperatives than mothers (Rondal 1980, McLaughlin et al 1980, McLaughlin et al 1983). Interestingly, McLaughlin et al (1980) found that fathers directed more imperatives at their sons than their daughters. Berko-Gleason attributes the finding that fathers use more direct imperatives to the fact that fathers cast themselves into the role of disciplinarian in the home setting, and he states the finding that fathers direct more imperatives to their sons than their daughters “gives the impression that in our society males become accustomed early on to taking orders, and, if their fathers provide role models, to giving them” (1975; 294). McLaughlin et al (1980) found mothers use more indirectly controlling language whereas fathers use more directly controlling language. Berko-Gleason proposes that “mothers tend to couch their imperative intent in question form” (1975; 295) which conflicts with research that has evidence that fathers ask more questions overall. Golinkoff and Ames (1979) found that the situation has a bearing on parental use of directives as the amount found in dyadic situations increased from the amount used in triadic situations regardless of the gender of the parent. Parents were said to fall into a directive mode in dyadic situations.
2.7 – SUMMARY AND RESEARCH QUESTIONS
The examination of the literature in the area of caregiver input shows that there are a number of similarities and differences between mothers’ and fathers’ speech. Parental interactive styles in dyadic behaviours have been the primary focus of research, and it has been proven, despite the many discrepancies in the research, that both parents play an important effective role in child language development. Berko-Gleason asserts that “when men occupy a nurturant role they become increasingly sensitive to the needs and intentions of the child” (1975; 296), suggesting that fathers who adopt the primary caregiver role because their female partners are in employment, are more sensitive to their children’s needs, assumingly so because they spend more time with them. The ‘Bridge Hypothesis’ and the ‘Differential Experience Hypothesis’ have been theorised based on the notion of sensitivity, proposing that mothers are more sensitive of a child’s needs as they are found to ‘scaffold’ children’s utterances more often. Fathers on the other hand, are seen as more insensitive interlocutors in comparison, as they are generally found to provide children with a bigger linguistic challenge.
Generalisations have been made in summary of the variables recorded by previous studies. It has been found that mothers address more speech to their young children than fathers, both in terms of mean number of utterances and conversational turns, asserting that mothers are more talkative. Dependant on context, it has been found that fathers are capable of producing the same number of utterances and turns in dyadic situations with a child, however there is consistency in the results in the fact that it has never been proven for fathers to speak more than mothers, neither in terms of mean number of utterances nor conversational turns, withholding the hypotheses mentioned above in the respect that fathers provide the child with a more challenging conversational partner as a result of not making themselves as linguistically dependable as mothers. Mothers seem to take more responsibility for sustaining a conversation through their more frequent vocalisations.
Differences between mothers and fathers have appeared in a number of areas of research, including vocabulary and use of directives. The vocabulary of fathers is said to be more diverse and lexically demanding which contributes to the challenging linguistic behaviour fathers demonstrate towards children. Fathers are said to direct more imperatives at children than mothers, and fathers are said to direct more imperatives to their sons than their daughters. This distinction between the behaviour directed at children is attributed in relation to the socialisation of gender as males in society are said to need to become accustomed to giving and taking orders. Fathers are more likely to use an imperative whereas mothers are said to frame their directives in interrogative form. Fathers are said to engage in such usage because they adopt the role of disciplinarian.
Research shows that mothers and fathers use sentence types to relatively the same proportions, using declarative and interrogative sentences most frequently. It is commonly postulated that when repetitions occur in parental input, fathers are more likely to repeat their child’s preceding utterance whereas mothers are more likely to repeat themselves. It is generalised that mothers make more corrections to their children’s speech than fathers and mothers are significantly more probable to produce grammatically incomplete utterances. The speech of mothers is also expected to contain more expansions: a notable contributor in aiding the ‘scaffolding’ of children’s utterances, therefore showing support for the ‘Bridge Hypothesis’.
Inconsistencies have occurred across numerous variables that have been tested, including questions and MLU. Given the somewhat sceptical findings of previous studies in the area, the aim of this study is to provide a clearer insight through my own investigation of gender-specific CDS, which will hopefully shed light on the discrepancies that have occurred. Previous research has shown that findings between parental input can largely differ based on the situation they occur in (dyadic or triadic) and the context in which the interaction is held (free-play or structured play), therefore these factors will remain constant in this study. The nucleus of this analysis is the difference in gender-specific CDS styles. A mother and father each in dyadic interactions with a child of approximately two years old will be recorded and then the study will be extended in order to observe the same parents in a dyadic interaction with an older sibling. Few studies in the existing research have explored the nature of gender-specific CDS in this way; however Broen (1972) found that when mothers spoke to younger in comparison to older children they used a lower rate, fewer disfluencies, and smaller type-token ratios. They also used smaller vocabularies, but they repeated their utterances more frequently (Giattino & Hogan 1975). Davidson and Snow (1996) suggest that fathers become better conversational partners as children get older. This is an area for examination in this study.
Reviewing the variables of a number of previous investigations in order to highlight comparisons and discrepancies in their findings regarding parental speech styles has allowed me to establish a set of variables for analysis in my own investigation. Since very few conclusive results have been established by previous studies, my analysis will provide a clearer explanation to these somewhat ambiguous generalisations.
Due to the inconsistent results of previous studies, the following research questions will be attempted:
Do fathers use the same features of CDS as mothers
How, if at all, do the parental speech styles differ
and due to the lack of information regarding the differences in parental speech styles in families with more than one language-learning child, the following question will be aimed at:
Do parents direct the same linguistic behaviour towards an older and younger sibling
In line with the ‘Fine-Tuning’ hypothesis, it is expected to find that parents ‘fine-tune’ their speech more towards the younger of the siblings. Prosodic features are expected to become seldom used to the older of the siblings and it can also be hypothesised that the parent’s MLU will increase with the older child. Since the ‘Bridge Hypothesis’ and ‘Differential Experience Hypothesis’ entail that mothers are more sensitive interlocutors, my experimental hypothesis is that mothers will ‘fine-tune’ their speech more than fathers. If this is the case, it will entail the mother having a MLU score lower than the father’s and closer to the MLU of her children. The null hypothesis is identifiable if the mother and father do not produce significantly different measures of ‘fine-tuning’ or MLU. It is important to note that individual differences may arise in the study and have a considerable bearing on the results, e.g. culture, socioeconomic class or parent’s level of education.
CHAPTER 3: METHODOLOGY
The purpose of this study was to analyse the speech of a male and female parent in dyadic interactions with their language-learning children, in order to produce a cross-sectional analysis which will lead to answering the central question of whether fathers as well as mothers use the features of CDS. The following general observation questions were considered:
Do fathers use the same features of CDS as mothers
How, if at all, do the parental speech styles differ
Do parents direct the same linguistic behaviour towards an older and younger sibling
3.1 – PILOT STUDY
Prior to the actual recordings, I conducted a small preliminary investigation using only two of the subjects (the mother and child O) in order to test whether the ‘Observer’s Paradox’ would arise. I wanted to be present for the recording; however, as a consequence of this child Y was not responsive to her mother’s speech. I vacated the room, leaving the mother to start the recording after my departure. In my absence the mother described child O’s language as typical entailing that my presence gave rise to the ‘Observer’s Paradox’. I therefore ensured that only the subjects involved in any one interaction were in the play-room at the time of the recording in order to eliminate the effects of the ‘Observer’s Paradox’ altering the authenticity of my main study recordings. I attempted a preliminary transcription exercise from the results of my pilot study, in which I was exposed to the problems associated with the calculation of mean length of utterance (MLU). I carried out online research in order to obtain a protocol for such calculations (see variables below) and followed this protocol when transcribing the data.
3.2 – THE SUBJECTS
There were four subjects in the main study: the mother, the father, the older child who I will refer to as child O and the younger child who I will refer to as child Y. At the time of the recording, child Y was 1;10 and child O was 4;4. I chose the children in my investigation not to be of similar ages so that they were not at similar stages of language development. Child Y was at the one-word stage, occasionally using two-word utterances and child O was at the grammatical stage, producing utterances relatively adult-like. Both of the parents in the study were of similar ages, there was an eighteen month age gap between them. In order to make accurate comparisons with the existing results identified in the literature review section, I ensured I maintained the social class variable of the majority of those studies by recording a family representative of middle class. In terms of occupation, the mother is a part-time psychiatric consultant and the father is a full-time college lecturer, therefore both parents are university-educated. In the means of family structure, I ensured that the parents in the family I observed were the biological parents of both children, i.e. to ensure that neither child O nor child Y were step children to either parent through divorce and remarriage as atypical linguistic behaviour may be expected from a non-parent towards a child and vice versa.
3.3 – DATA COLLECTION
The speech of the family was recorded on a digital voice recorder. All recording was done during periods of free-play and all interactions took place in the family play-room, as this is the area where the subjects engage in free-play on a daily basis. I obtained four recordings in total: the mother and father both in a dyadic interaction with each child. Each separate recording consisted of at least twenty minutes of speech. The lengths of the recordings slightly differed in total as the children terminated their play sessions at different times; however, I extracted exactly fifteen minutes from each recording for analysis. I anticipated that the adult subjects may have been inhibited to behave differently with anyone other than the subjects in the room; therefore the parents began the recording after I vacated the room. The inherent problem in such a recording is the subjects’ awareness of the recorder as the investigation had to be carried out obtrusively. This could have had a possible bearing on the naturalistic nature of the data; however none of the subjects seemed to be concerned that they were being recorded. In order to eliminate the possibility of the presence of the recorder having a bearing on my results, I overlooked the first two minutes of each recording as literature advises that most people forget about the recording as they engage in activities (Wray & Bloomer 2006). Neither of the children were unfamiliar to the tape recorder as they had been recorded by their father in this way previously. The subjects were not given any special instructions in the means of expected behaviour and they were encouraged to ignore the presence of the tape recorder. Throughout the session the parents and children engaged in spontaneous play: in activities such as a scrabble board-game, an ‘etch-sketch’ drawing toy, a wooden shapes toy, a plastic utensils game, an ‘aqua beads’ shape game, a ‘guess-who’ game, playing with a ball, building blocks and making a cup of tea. After the recording I gave the adult subjects a self-completion questionnaire (see appendix 2). Questionnaires are advantageous in the fact that they are efficient to administer, they eliminate interviewer effects and they are convenient for the respondents to complete. In order to eliminate respondent fatigue, I limited the questionnaire to eight questions and ensured that they were simple and unambiguous. I asked two open questions in order to obtain qualitative data.
3.4 – DATA TRANSCRIPTION
After obtaining the four separate recordings of conversational data, I made a copy of the original recordings. I discarded the first two minutes of each recording before analysis, in order to eliminate possible effects caused by the ‘Observer’s Paradox’. I decided to transcribe exactly fifteen minutes of each recording in order to ensure a fair test overall. I orthographically transcribed the data so that the speech could be represented in order to be analysed structurally and accurately (see appendix 1). Using the set of variables below, I then analysed the data.
3.5 – VARIABLES
In order to answer the observation questions mentioned above, the following variables were measured:
Communicative Turn (CT) this is analysed as everything a speaker says before the next speaker begins. This could be one word, one sentence or several sentences.
Total number of utterance – number of utterances produced.
Mean length of utterance (MLU) – measured by the total number of morphemes divided by the total number of utterances in the dyad. An utterance is a word or a string of words identified by a pause, grammatical completeness (Golinkoff and Ames 1979) or other indication of new thought.
When counting MLU the following are counted as a single morpheme:
a) -s plural marker e.g. letter-s
b) -ed past tense marker e.g. finish-ed
c)-ing present participle marker e.g. smil-ing
d)-s 3rd person regular tense marker e.g. plays-s
e) Possessive -‘s marker e.g. daddy’s bike
f) Compound words e.g. teapot
g) Proper names e.g. Hazel
h) Irregular past tense verbs e.g. went
i) Irregular plurals e.g. children
j) Diminutives e.g. horsy
k)Catenatives e.g. wanna
l) Contractions e.g. let’s, don’t and won’t (but the following contractions are counted as two morphemes e.g. she’s, he’ll, they’re, what’s, she’d, we’ve, can’t, aren’t)
m)Reduplications e.g. daddy daddy daddy are counted as one morpheme unless the repetition is for emphasis (Speech Therapy Information and Resources 2009-2010).
n) Fillers e.g. mm, ah, oh are not counted as a morpheme
Declarative sentence – (including one-word declaratives) used to make a statement.
Interrogative sentence – used to ask a question.
Wh-questions – questions that employ the use of: what, when, where, why, who, whose, which or how.
Yes/no questions – questions that require a yes or no answer from the hearer.
Intonation questions – questions marked by a rise in intonation.
Tag questions – a question attached to the end of a statement, usually seeking confirmation.
Imperative sentence – used to give command, request or give instructions of some kind – orders, warnings advice etc.
Exclamatory sentence – emphatic sentences used to express strong emotion.
Repetition of child
Grammatically incomplete sentence – sentences involving the deletion of some words.
Sentences which could be categorised in more than one way were placed in the highest category in the order of priority list: repetition, interrogative, declarative, imperative, exclamatory, and grammatically incomplete (Giattino & Hogan 1975).
CHAPTER 4: RESULTS
CHILD YCHILD OCHILD YCHILD OMOTHERFATHERMOTHERFATHER
Turns (CT)729463134776294132Total utterances1821751632118462104152Total number of morphemes794104954995012593571660Mean length of utterance (MLU)4.3663.374.51.491.55.494.34Declaratives42422469 Wh-questions3234916 Yes/no questions12321136 Intonation
Questions7291130 Tag questions5503 Imperatives1585726 Exclamations 35162714 Parental self repetitions101164 Repetitions of child 227612 Grammatical incompletions2121
TABLE 4.1: Summary of means and amounts of parental speech and child vocalisation.
Most of the measures used in the speech analysis were simple counting procedures, using relatively straightforward criteria. In order to ensure reliability in the findings, I recalculated the data twice which removed any data verification errors that had occurred the first time.
4.1 – THE AMOUNT OF PARENTAL SPEECH
The parents‘ total utterances to child Y suggest that the mother and father direct a similar amount of utterances to the younger, in comparison to the older sibling. The results show that the mother produced approx 12 utterances per minute to child Y, while the father produced approx 11 utterances per minute to the same child. The parents’ total utterances to child O show a significant difference. The father was found to produce almost 2.5 more utterances per minute to child O than the mother. In terms of conversational exchange, there are conflicting findings in the means of conversational turns (CT). The mother was found to take 9 more CT’s than the father in the dyad with child Y, however she was found to take 30 fewer turns in the dyad with child O. Though it is not to assume that the father spoke more to child O on the whole, as although his total of utterances was greater, his turns were also shorter, shown by his MLU in the dyad with child O (4.5), which was 25% less than the mother’s rate of MLU to the same child (6), as show in FIGURE 4.1a. The father also made fewer utterances (19 less than his partner) and addressed shorter utterances to child Y, which is again evidential in his MLU. This is contrary to existing research and hypotheses that propose fathers to be more challenging, demanding interlocutors than mothers. All in all, the mother consistently talked more than the father, speaking for 68% of her dyadic interaction with child Y and 63% of her dyadic interaction with child O.
The complexity of parental speech
In terms of MLU, the mother and father were found to show a significant difference. The father’s MLU to child O is almost equal to the mothers MLU to child Y (as shown in FIGURE 4.1a). Although the MLU of the parents conflicts with evidence from previous research, the MLU of each parent in the separate dyads correlates with the MLU of each child in the specific dyad (as shown in FIGURE 4.2a). Albeit the exception to this finding is the dyadic interaction involving the mother and child Y, which stipulates the mother’s MLU to be triple the child’s measure. This finding contradicts the ‘Fine-Tuning’ hypothesis.
The complexity of children’s speech
Although child Y made more conversational turns with the mother than the father, her MLU value was the same with both parents. It is plausible to say from this finding that the father elicited more complex speech from child Y as although she measured the same MLU with both parents, she made 22 fewer utterances with her father. The reverse can be said for child O. The findings show that the mother elicited more complex, longer speech from child O than the father, due to the fact child O’s MLU value with her mother is over 1 morpheme per utterance longer than with her father and she produced 48 less utterances in total with her mother.
FIGURE 4.2b shows the percentages of repetitions used in each of the dyads. The repetitions made by the mother were more consistent than the repetitions made by the father. The majority of the mother’s repetitions were repetitions of the child rather than herself. The findings regarding the repetitions made by the father show an inconsistency as it was found that the father made 3 times as many self repetitions with child Y than he did with child O.
4.3 – THE FUNCTIONS OF PARENTAL SPEECH
TABLE 4.3a presents the proportions of utterance types employed in the parental speech. It shows the proportion of each utterance type as a percentage of the total utterances in each dyad. Totals add up to 100 per cent as sentences which could be categorised in more than one way were placed in the highest category in the order of a priority list (see methodology). FIGURE 4.3b shows the findings in table 2. FIGURE 4.3a shows the frequency of the utterance types in each dyadic interaction in the form of a clustered graph.
questions4%16.5%7%14%Tag questions3%3%0%1%Imperatives8%4.5%35%12%Exclamations19%9%16%7%Parental self repetitions5%1%10%2%Repetitions of child12%4%4%6%Grammatical incompletions1%1%1%0.5%
TABLE 4.3a: Summary of distribution of utterance types in parental speech (% of
The mother used the same proportion of declaratives in both of her dyads with the children. These included one-word declaratives and statements. In contrast, 15% of the total utterances produced to child Y by the father were declaratives, as were 33% of the total utterances he produced to child O.
Examples produced by the father included:
“I’ve got three now”
Examples produced by the mother included:
“I didn’t hear you properly that time”
Number of questions asked
Percentage of total of questions asked in dyad (%)
Number of questions asked
Percentage of total of questions asked in dyad (%)
Number of questions asked
Percentage of total of questions asked in dyad (%)
Number of questions asked
Percentage of total of questions asked in dyad (%)