Journal of Phonetics 52 (2015) 58–69

Contents lists available at ScienceDirect

Journal of Phonetics

http://dx0095-44

⁎ CorrE-m

journal homepage: www.elsevier.com/locate/phonetics

Research Article

Effects of the growth of breath capacities on mean length of utterances:How maturing production processes influence indices oflanguage development

Victor J. Boucher⁎, Brigitte Lalonde

Laboratoire de sciences phonétiques, Département de linguistique, Université de Montréal, Montréal, Québec, Canada H3C 3J7

A R T I C L E I N F O

Article history:Received 26 June 2014Received in revised form21 April 2015Accepted 24 April 2015Available online 29 May 2015

Keywords:MLUSpeech breathingSpeech physiologyDevelopment of speech productionLanguage developmentLexical diversityLinguistic theory

.doi.org/10.1016/j.wocn.2015.04.00570 & 2015 The Authors. Published by Elsevier L

esponding author. Tel.: +1 514 343 5672.ail address: victor.boucher@umontreal.ca (V.J. B

A B S T R A C T

Measures of “mean length of utterance” (MLU) involving morpheme counts in transcripts are widely applied tospeakers of all ages and are generally interpreted as an index of developing grammar. Yet no study has examinedhow the growth of respiratory capacities influences MLU and numbers of forms in utterances. We reviewlongstanding problems of MLU counts and investigate the effects of growing breath capacities using speechsamples and measures of vital capacity (VC) of 50 speakers aged 5 to 27 years. The results show that VCcorrelates strongly with MLU, which associates with rising numbers of long lexemes. This suggests that, in normaldevelopment, the growth of VC offers the possibility of producing increasingly long utterances that can influencelexical diversity. Hence, interpreting MLU and co-varying indices of lexical development requires a considerationof the effects of maturing production processes in a perspective where developing speech and language are seento intertwine.& 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license

(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

The development of oral language presents several milestones where emerging behaviors clearly intertwine with maturationalchanges in production processes. One remarkable example of this is the emergence of babble. Research has shown that infants arenasal breathers and produce nasalized vocalizations early in life (e.g., Thom et al., 2006). By six to ten months, however, supraglottalstructures have undergone several changes including a decoupling of the naso-pharynx, and these changes which accompany a risein rhythmic behavior lead to oral vocalizations and babble (Iverson et al., 2007; Kent, 1984; Oller, 1986, 2000; van der Stelt &Koopmans-van Beinum, 1986). In interpreting such development, one should not view the growth of production processes as causingthe rise of babbling. In fact, several factors can influence the course of emerging speech. For instance, babbling is typically delayed inhearing-impaired children (Oller & Eilers, 1988), indicating that sensory stimulation is a factor. Nonetheless, maturational changes inproduction processes – not the development of hearing – basically account for the emergence of babble at 6–8 months. In otherwords, one can see that, in normal development, the growth of production structures is a necessary though not sufficient factor inexplaining the rise of babbling at a given age.

Examples such as these illustrate the point that, in building an understanding of the time course of language development, onemust consider not only the perceptual and cognitive abilities of children but also maturational changes in speech processes. This notonly relates to changes in supraglottal structures that contribute to the rise of canonical and variegated babbling. Observing subglottalprocesses is also important since the growth of such aspects as respiratory capacities may well influence the length of utterancesand the number of verbal forms they contain. Yet, such effects are most often overlooked in developmental linguistics, as illustrated inconventional measures of utterances. In particular, measures of “mean length of utterance” (MLU) involving counts of morphemes in

td. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

oucher).

V.J. Boucher, B. Lalonde / Journal of Phonetics 52 (2015) 58–69 59

transcripts of spontaneous speech are generally taken to reflect developing grammar – irrespective of the growth of speakers' breathcapacities. In fact, following the work of Brown (1973), MLU has become a standard index, often used with other tests (see, e.g.,Jalilevand & Ebrahimipour, 2014). In proposing this index, Brown referred extensively to linguistic theory (Chomsky, 1957, 1967/2006) and saw language as a “competence”, an ability to combine syntactic elements separate from speech “performance”.Accepting this division, the basic assumption underlying MLU is that, since many aspects of developing syntax imply additions ofelements in utterances, counting MLUs in morphemes captures a “cumulative complexity”, taken to reflect “knowledge” (see Brown,1973, pp. 53, 173). Such interpretation overlooks the potential effects of the growth of breathing volumes on MLU, perhaps becauseof the idea that structures of performance have little to do with numbers of morphemes in utterances. However, it should be noted thatmorpheme counts of MLU strongly correlate with numbers of syllables and “words” (at r¼ .91–.99: Arlman-Rupp, van Niekerk deHaan, & van de Sandt-Koenderman, 1976; Ekmekci, 1982; see also Hickey, 1991; Parker & Brorson, 2005; Rom & Leonard, 1990).Because syllables reflect modulations of air pressure, it seems likely that MLU would vary with speakers' growing breath capacitiesand that this would impact any unit count of MLU. Yet, no study has examined these potential effects.

As a possible explanation of this oversight, one should note that, originally, Brown's index was limited to toddlers with MLUs of nomore than four morphemes (“Stage V”), which normally appears at about 40 months (Miller & Chapman, 1981). At this young age,children are often unable to execute the maneuvers of standard measures requiring maximal respiratory effort (Desmond et al., 1997;Merkus, de Jongste, & Stocks, 2005) so effects of developing breath capacities on MLU are difficult to assess. However, applicationsof MLU currently extend to all ages (e.g., Behrens, 2006; Charness, Park, & Sabel, 2001; Justice et al., 2013; Miles, Chapman, &Sindberg, 2006; Rice, Smolik, & Perpich, 2010; Rondal & Comblain, 1996; van de Weijer, 1998). Hence, the validity of interpretingMLU as indexing knowledge separate from the growth of respiratory capacities can be evaluated. In this context, the present studyexamines how the growth of breath capacities can contribute to a developmental increase in MLU and the number of verbal units inutterances. These relationships can be clarified by using measures of MLU where “utterances” are seen as observable breath units ofspeech. This idea is not always accepted in the literature where MLU counts are often performed by reference to notions ofsentences. On this point, the following discussion exposes some longstanding problems of Brown's MLU so as to clarify ourmeasures and the relevance of data suggesting several effects of the growth of breath capacities on the length of utterances and thenumber of units they contain.

2. Measuring utterance length and effects of growing breath capacities

2.1. Defining utterances

Most studies that report MLUs follow the guidelines of Brown (1973, p. 54) who recommended that MLU be measured from 100utterances of spontaneous speech using rules of morpheme count (for slightly different rules, see Johnston, 2001; Miller & Chapman,2004; Miller & Iglesias, 2008; for varying sample sizes, see Gavin & Giles, 1996; Heilmann, Miller, & Nockerts, 2010; Hewittet al., 2005; Klee, 1992; Rondal & Defays, 1978). However, it is important to note that Brown never defined the utterance. Thisproblem of definition prevails in that available guidelines refer to conflicting criteria of utterance division leading to a lack of reliabilityin MLU, as several authors have remarked (Chabon, Kent-Udolf, & Egolf, 1982; DeThorne, Johnson, & Loeb, 2005; Eisenberg,Fersko, & Lundgren, 2001; Klee & Fitzgerald, 1985; Reed, MacMillan, & McLeod, 2001; Rice, Redmond, & Hoffman, 2006;Rollins, 1995; Rondal, Ghiotto, Bredart, & Bachelet, 1987). The difficulty essentially stems from the use of two differing concepts ofthe utterance present in the literature. The first concept is found in phonetic studies. In this work, an “utterance” is traditionally seenas a unit of speech or vocalization delimited by inspirations and bearing a declination in intensity and F0 (for a list of authorswho use this definition, see Vaissière, 1983). The second concept appears in linguistics, where authors often attempt to divideutterances in terms of assumed “sentences” and “clauses”. Both the above concepts are variably used in performing MLU counts,which points to a problem of validity. This can best be illustrated in terms of the characteristic shift in concepts that appears indevelopmental studies.

As an example of the application of the phonetic concept of utterance, Oller et al. (Oller et al., 1985; Oller & Lynch, 1992; see alsoOller, 2000) reported syllable counts of MLU for babbling infants. In this case, utterances were taken as units bordered by breathnoise and showing integrity in tone and amplitude (Oller & Lynch, 1992, p. 525). A more recent illustration is Fagan (2009) who alsorefers to utterances as breath units in observing the development of vocalization in infants. These applications differ from linguisticanalyses of older children where another notion of utterance is applied that refers to grammar. A representative example is Miller andChapman (2004) who prescribe that “When you segment the stream of speech into utterances you will use such cues as intonation,pausing, and grammatical structure to determine where one utterance ends and the next begins” (added emphasis).

In reasoning this shift in definition, it is not that utterances as breath units change. Obviously, speakers, young and old, keepproducing breath units in speech. What changes is that, at some point when children begin to produce speech that is interpretable byan adult observer, utterances are divided by reference to notions of sentences. However, this shift from an objective division ofutterances based on breath marks to one based on interpretations of grammatical units entails two questionable presumptions. Thefirst is that utterances are grammatical units when, in fact, children vocalize and babble in breath-divided units well before theyproduce any discernible grammar. In other words, the shift implies a disregard of the nature of utterance-size chunks in speech. The

V.J. Boucher, B. Lalonde / Journal of Phonetics 52 (2015) 58–6960

second presumption is that children are manipulating units like sentences. Yet attempting to divide speech into such units leads tofrequent inconsistencies, as Crystal (1974) noted in his review of Brown's (1973) division of sentences:

“…when Brown writes his chapter on coordination, decisions about whether two speech units are parts of the same sentence orare linked separate sentences will be very much dependent on utterance criteria being made explicit. It is not as if the problem ismerely an occasional one: it shouts out at the analyst all the time” (p. 295, added emphasis)

It is the case that children's speech often contains a profusion of forms like and, then, so (etc.), where dividing sentences is quite arbitrary.As an example, Crystal asked how many sentences there are in the talk of a three-year old who produces a narration like “and it goes up / upthe hill / and it goes up / in the hill / and it takes us on up the hill / and it goes up the hill / up up up /…” (1974, p. 296). But the problem is notonly that utterances as breath units do not consistently align with sentences. One must also consider that the linguistic concept of sentenceas such escapes definition (see Fries, 1952 who discusses some 200 definitions). In fact, numerous critics warn that notions of sentenceslink to training in alphabet writing (Coulmas, 1989, 2003; Harris, 1990; Kress, 1994; Miller, 1999; Miller & Weinert, 1998). In this context,guidelines of sentence division that refer to tonal contours as a backup criterion do not address the problem. Though it is common inlinguistics to view contours as mapping grammatical units, it is known that there is no isomorphic relationship between prosody and assumedsyntactic units (see Nespor & Vogel, 1986). In short, in performing MLU measures, one faces basic issues of utterance division arising from amismatch between breath units of speech and observer-based notions of sentences that critics link to a writing bias. Only the former bearsobjective marks and, for this reason, measures of MLU in the present study refer to breath-divided utterances.

2.2. Defining unit counts of utterance length

Another issue that arises is whether morpheme counts of MLU reflect elements that speakers use. Several studies have shown thatmany of the morpho-syntactic units assumed in traditional linguistic analyses may not be manipulated by speakers. For instance, it hasbeen reported that, even by three years of age, children may not divide forms categorized as verbs, pronouns, or determiners (Lieven,Pine, & Baldwin, 1997; Lieven et al., 2003; Pine & Lieven, 1997; Pine & Martindale, 1996; Tomasello, 1992, 2000a, 2000b). Proponentsof usage-based grammars also point out that countless chunks and formulaic expressions in both child and adult speech – like wanna,gi'me, shoulda, look out, take care (etc.) – may not involve separate morphemes and can function as lexical units (Bybee, 2006, 2010;Bates & Goodman, 2001; Wray, 2008). However, in failing to identify every separable element, one can surmise that morpheme countsof MLU would still index the “cumulative complexity” of utterances in terms of numbers of verbal units, as Brown (1973) suggested. Forthis to be true, though, it needs to be shown that MLU can correlate with the counts of lexicalized chunks in utterances.

2.3. The relationship between respiratory capacities and speech breathing

Contrary to interpretations that developing MLU reflects knowledge, numerous reports of age-related changes in MLU suggest theneed to weigh factors of maturing physiology (Blake, Quartaro, & Onorati, 1993; Chan, McAllister, & Wilson, 1998; Conant, 1987;Klee & Fitzgerald, 1985; Klee et al., 1989; Miller & Chapman, 1981; Parham et al., 2011; Scarborough, Wyckoff, & Davidson, 1986;Scarborough et al., 1991). Interestingly, the effect of such factors was made evident, almost accidently, in Brown's (1973, p. 55)longitudinal study of three children. As the author noted, steady increases appeared in the MLUs, except on one occasion where achild produced a sudden drop in MLU because she had a cold. It may strike one as odd that the MLU index said to reflectgrammatical knowledge can fluctuate with symptoms of a common cold. Yet it stands to reason that utterances as breath units wouldvary in length in conditions that limit normal production, such as shortness of breath, or with age-related changes in respiratorycapacities. Though such fluctuations may not undermine the value of MLU as an index, they lead one to question currentinterpretations of MLU that fail to acknowledge the effects of growing respiratory functions on utterance length.

To fathom the extent of these effects, a central measure is vital capacity (VC, or the maximal volume of air that can be forcefullyexpired after a maximal inspiration). Of course, people do not use all of their VC in speaking. But it is known that speech breathinggenerally involves a limited range of VC where breath cycles require minimal effort of expiratory muscles (Ladefoged, 1967; Mead,Bouhuys, & Proctor, 1968). Thus, because habitual speech uses a steady range or proportion of VC, age-related increases in VC (inliters) affords the use of greater volumes of air in speaking. To illustrate this, we refer to Fig. 1A which summarizes developmentalobservations of Hoit et al. (Hoit & Hixon, 1987; Hoit et al., 1990). The two dotted lines in panel A indicate the range of VC that isnormally used in producing utterances at conversational loudness. One can see that developing VCs imply a substantial increase ofthe breath volumes used in speech, and this would likely affect MLUs.

There are obviously many other maturational aspects of speech that accompany the changes illustrated in Fig. 1A, and which mayaffect MLUs. To weigh these factors, one can compare the potential effects of changes in VC and maturing speech-motor andrespiratory processes. For instance, there is a refinement of motor control throughout childhood as evidenced by articulation:children's speech is slow, highly variable, and adult-like speech rates and coordination can take up to early adolescence to develop(Kent & Forner, 1980; Kent & Vorperian, 1995; Smith, 1978; Smith & Zelaznik, 2004). Yet these changes would have minor effects onMLUs. The reason is that, even if children speak at slower syllable rates than adults (by about 15% according to Smith, 1978), theyexpend less air per syllable. For example, Hoit et al. (1990) reported that 7 year-old boys produce an average of 35 mL/syllable inspeech, while adults expend 65 mL/syllable on average (Hoit & Hixon, 1987). This is explained by the fact that children's vocal tractsrepresent smaller volumes and their vocal cords, which have a smaller mass, require less flow across the glottis to vibrate andsustain audible speech (Kent & Vorperian, 1995; Hirano, Kurita, & Nakashima, 1983). As for the changing functions of speech

Fig. 1. Data from Hoit and Hixon (1987), and Hoit et al. (1990),1. (A) Vital capacity (VC) of male speakers aged 7–25 years; the dotted lines show the average initiation volume (top line)and termination volume (bottom line) in producing utterances. (B) Proportion of VC used in producing utterances [(initiation−termination volumes)/VC].

V.J. Boucher, B. Lalonde / Journal of Phonetics 52 (2015) 58–69 61

breathing, Fig. 1A shows that children initiate and terminate utterances at lower volumes than adults (by about 5–10% according toRussell and Stathopoulos (1988) and Hoit et al. (1990)). This has been linked to changes in the compliance of the lungs and thorax,which may accompany a variation in reactivity levels of the Hering–Breuer reflex (Agostoni & Hyatt, 1986; Lanteri & Sly, 1993).However, it is important to note that, despite these changes, children and adults use similar portions of their VC in speaking, asshown in Fig. 1B. Also, if the aforementioned factors have any effect on utterance length, the magnitude of effects, such as the5–10% difference in initiation volumes between 7 and 25 years of age, can be weighed against the fact that VC rises by 330% ormore in the same period (Fig. 1A). Overall, then, it is clear that the growth of VC may have a major impact on MLUs. Moreover, sincechildren and adults use similar ratios of VC in producing utterances (Fig. 1B), measures of VC can serve to capture this influence.

3. The present study

The above data of Hoït et al. show that the growth of speakers' respiratory capacities corresponds to an increase in the breathvolumes used in producing utterances. This would likely affect MLU counts in morphemes or other units – though this remains to bedemonstrated. As such, these effects would not undermine Brown's idea that MLU can index a developing “cumulative complexity”. Infact, the value of this index finds indirect support in correlations between MLU and other indices of morphological and lexicaldevelopment (e.g., DeThorne et al., 2005; Eisenberg et al., 2001; Hickey, 1991; Jalilevand & Ebrahimipour, 2014; Nieminen, 2009;Rice et al., 2006; Rollins, Snow, & Willett, 1996; Scarborough et al., 1991). On the other hand, recognizing the effects of growingbreath capacities leads one to question the general interpretation that rising MLUs and associated changes in lexical indices reflectknowledge. In contrast to this interpretation, we submit that the growth of VC through to adulthood links to rises in MLU and that thisinfluences the number of lexical forms within utterances. On this view, it should be noted that MLU has been shown to correlate withindices of lexical diversity – though no rationale has been suggested to explain how MLU counts link to a diversification of lexical units(see DeThorne et al., 2005). One possible explanation that is explored in the present study is that developing MLU allows theproduction of increasing numbers of long lexical units. We investigate these hypotheses by examining how the growth of VC innormal speakers relates to the production of long utterances bearing rising counts of long lexemes. It should be noted that, in thepresent report, analyses of “lexemes” are limited to lexicalized units and chunks that function as nouns given the difficulty inpresuming division for other types of chunks such as expressions containing verbs (see Section 2.2).

4. Method

4.1. Subjects

The participants were 50 male speakers equally divided into five age groups of 5–7, 10–12, 15–17, 20–22, and 25–27 years. Thechoice of these groups was motivated by the need to sample equal numbers of speakers across a range of ages and by the

1 On these estimated volumes, we caution readers on possible inaccuracies in studies of speech breathing. First, a common procedure consists of measuring volumes for speech viamotions of the rib cage and abdomen using elasticometric or magnetometric techniques (following Hixon, Goldman, & Mead, 1973). With these techniques, measures of chest-walldisplacement are often converted into %VC by having subjects perform a VC maneuver and calculating volume equivalents. However, this assumes that chest-wall motion is linearlyrelated to volumes for 100% VC whereas non-linear effects occur at extremes of a VC maneuver. For instance, we used different belt systems with adults while monitoring breath flows andrepeatedly observed floor effects in motions of the chest at maximal expiration points – while air flow was still being produced (see also Stagg, Goldman, & Davis, 1978). At these points,then, continuing internal compression generates flow when the chest wall ceases to move. Consequently, calibrating the motions as if they reflect 100% VC is inaccurate, and such errormight only be corrected by using other techniques such as a full-body plethysmograph (Ohala, 1993). Finally, it is important to note that the data of Hoit et al. in Fig. 1 refer to volumes usedto produce utterances in spontaneous speech (other reports refer to the reading of passages or the recitation of slow trains of syllables, e.g. Stathopoulos & Sapienza, 1997). However, thenumber of utterances or sample sizes of Hoit and Hixon (1987), Hoit et al. (1990) are much smaller than those used to measure MLU so it is unclear whether reported volumes arerepresentative of habitual breath groups in speech.

V.J. Boucher, B. Lalonde / Journal of Phonetics 52 (2015) 58–6962

expectation that MLU and VC would level off as speakers near 21 years of age. All participants were native speakers of French withFrench as the main or only spoken language. The children in the sample were all attending regular curricula in public schools and thevast majority of adolescents and adults were students in secondary or post-secondary institutions. According to individual andparental reports, none of the individuals had a history of speech or respiratory disorders, and on the day of testing, none manifestedsigns of a condition affecting normal respiration or vocalization. Just before the tests, measures of height and weight were performedon most participants (41/50) where an access to calibrated scales was possible. Anthropometric measures (McDowell et al., 2008)showed that the stature of a 10 year-old was at the 5th percentile. This individual, however, presented a VC within a normal rangeand was retained in the analysis. All other participants were within a normal age-range of height and weight.

4.2. Procedure

4.2.1. VC measuresThese measures were performed by way of a pneumotachometer (Aerophone II, F–J Electronics) and accompanying software.

This instrument was regularly calibrated via a 1-liter syringe and set to operate at an ambient temperature of 21 1C. The procedurewas guided by the recommendations of Ferris (1978, reproduced in Kent, 1994, pp. 107–109), and standard procedures of theAmerican Thoracic Society and European Respiratory Society (ATS, 1995; Merkus et al., 2005; Miller et al., 2005). All subjects wereasked to maximally inhale and then maximally exhale, as if blowing up a balloon, through a standard disposable tube (3.2 cm indiameter; Roxon Medi-Tech) fitted to the tachometer. Prior to the maneuvers, visual demonstrations were provided involving a mimicof the task with the instrument. For the younger subjects, additional demonstrations and instructions served to illustrate maximalinspiration and expiration (blowing up balloons, storyline and mimics of the Big bad wolf, visual feedback from the spirometersoftware, etc.). Throughout the maneuvers, flows were monitored in real-time and subjects were repeatedly encouraged to inhale asdeeply as possible and to keep blowing after the initial flow peaks (“keep blowing, blow, blow, blow…”). Six maneuvers wereperformed, three before and three following the speech task, and throughout the tests the subjects were seated upright. It should alsobe noted that the ATS (1995) criteria specify three best reproducible VC measures presenting acceptable flow curves for six-secondexhalations. However, forced expiration for six seconds is unrealistic, especially for children (see Desmond et al., 1997; Miller et al.,2005). Consequently, end-of-trial was determined by the subject and in all cases this corresponded to a period of zero flow following afull VC maneuver. VCs in the present study reflect the best values obtained on six maneuvers while respecting an acceptabilitycriterion for flow patterns and a degree of reproducibility (ATS, 1995). Considering the flow patterns, post-test analyses revealed thatfour subjects produced irregular flows on one to two maneuvers, and VCs for these trials were excluded. As for reproducibility, allsubjects produced three or more reproducible values to within 400 mL and 24 subjects produced reproducible values within the rangeof 200 mL (ATS, 1995). However, the same best values were obtained across both ranges of measures.

4.2.2. Speech recording and transcriptionSpeech samples were recorded in the context of a one-to-one conversation where the interviewer and participant sat about 1 m

apart. Most of the subjects were interviewed in their home or school (two were interviewed in a laboratory setting accompanied by aparent). Uninterrupted monologue was obtained by way of open-ended questions or submitted topics bearing on the speakers'activities and interests. The recordings were performed using a 16-bit mini-disc player (Sony MZ-NH700) set to a sampling rate of44.1 kHz, and a microphone (Electro-Voice, model 635A) placed on a boom and at 3–5 cm from the subject's lips so as to captureinspiration noise. At the transcription stage, the signals were played back using software that allowed the selection of portions ofvisually displayed waveforms (Goldwave, version 5.25). Selected samples (see the rules in the Appendix) reflecting uninterruptedmonologue were transcribed in broad (“phonemic”) IPA by a trained phonetician (the co-author) using headphones (Beyerdynamic,model DT 250). The final transcripts contained no less than 71 utterances per individual. Mean sample sizes were similar across age-groups, as indicated in Table 1, and all the corpora showed stability in means in that adding 10 utterances to a sample did not alterthe cumulative MLU in syllables (MLUS) beyond 1%.

4.2.3. Counts of MLU and nominal lexical chunksThe rules used in performing morpheme and syllable counts of MLU are specified in the Appendix and it should be noted that the

guidelines for dividing morphemes approximate those used by Brown (1973, p. 56). As for counts relating to lexical diversity, “nominallexemes” in the present report are defined as free-form lexemes or formulaic expressions used as coherent units and appearing incontexts where nominal forms are expected. Identification of these chunks was straightforward and presented high inter-judge

Table 1Average sample size (no. of utterances) per age group (n¼10 per group).

Age group (yrs) Mean s.d.

5–7 92.1 15.210–12 100.9 13.315–17 104.2 8.620–22 106.4 7.925–27 100.0 11.8Overall 100.7 5.5

V.J. Boucher, B. Lalonde / Journal of Phonetics 52 (2015) 58–69 63

reliability (on a subsample, identifications of nominal chunks by two external judges showed an agreement rate of 415/416). The fewcontentious cases that arose in the analyses were resolved by referring to a data bank on French locutions (Antidote RX, version 8,Druide Informatique). Two sets of ratios were calculated based on type and token counts of lexemes categorized in terms of their size(in numbers of syllables). The first set comprised global type/utterance and token/utterance ratios calculated in terms of first 70utterances transcribed for each speaker. The second set, central to the present report, included ratios representing the proportion oftypes containing one, two, or three syllables or more. Finally, statistical analyses of these counts and measures of MLU and VC, asreported in the following section, made use of the procedures of SPSS (version 17.0).

5. Results

The reliability of the MLU measures was evaluated by comparing counts performed by the authors to those obtained by anexternal judge experienced in IPA transcription. The judge was instructed on the guidelines and asked to analyze a corpus of soundfiles using the aforementioned software and headphones. The corpus in question contained uninterrupted monologue from sixsubjects (at least one per age group) representing 15–20 utterances per speaker. Using breath noises as specified in the guidelines,inter-judge agreement on discourse-internal divisions for 100 utterances was 100%. Within these divisions, matching counts ofsyllables appeared for 89% of the utterances (in total, one judge counted 1709 syllables and the other 1699). Errors occurred mostlyin counting syllables for filled hesitations. Overall, MLU counts in syllables differed by .5% across judges (16.92 compared to 16.82).As for MLU in morphemes, matching counts were lower than for syllables and appeared for 55% of the utterances (one judge counteda total of 1664 morphemes and the other 1629). Most errors appeared in counting morphemes for irregular verbs and contractions,and MLU in morphemes differed across judges by 2.1% (16.64 compared to 16.29).

Turning to the main results, Fig. 2 illustrates the distribution of subjects' VCs. To evaluate these data, the dotted lines in the figureshow the predicted 95% range of values for healthy males (Stanojevic, Wade, & Stocks, 2008) using population statistics of heightand age (McDowell et al., 2008). One can see that the measured VCs are within a normal range, though the values of the two oldestgroups were in the middle to low regions. This slight skewing toward low values is in fact present in the population (Stanojevic et al.,2008) and did not threaten the assumptions of parametric statistical tests.

As for MLU measures, Fig. 3 gives the data plots of morpheme and syllable counts, which show similar rising trends with age. Inthis case, though, one cannot assess normality with respect to population parameters (given the absence of anthropometric scalesfor MLU). However, our specific aim was to examine how MLU co-varies with the growth of breathing capacities. On this question,Fig. 4 illustrates the strong correlations present between MLU and VC [for MLUM, r(50)¼ .81, p<.001.; for MLUS, r(50)¼ .79, p<.001].It is also useful to note that MLU counts in morphemes and syllables remain highly correlated across age groups, as shown in thecoefficients of Table 2.

To explore the possibility of independent fluctuations of MLUM, we performed partial correlations of MLUM and age (in months)while controlling for the effects of speakers' VC and numbers of syllables in utterances. As such, rising MLUM correlated with age[r(50)¼ .78, p<.001] as one could expect. However, partialling out the effects of speakers' VC substantially reduced the correlation [pr(47)¼ .35, p¼ .012]. This suggests that breath capacities present strong intervening effects. Moreover, when partialling out syllablenumbers in utterances, age and MLUM no longer correlated [or only marginally, pr(47)¼−.27, p¼ .057]. Taken together these resultsshow that rises in MLUM with age are not independent of subjects' growing breath capacities and the number of syllables that areproduced in utterances.

Finally, we checked the relationship between MLU and VC, and between MLU and subjects' stature (which was measured for 41of the 50 participants), while taking into account co-variations in age. Partial correlations showed that, when controlling for age, MLU

Fig. 2. Vital capacities of the 50 subjects. The dotted lines represent the 95% range of VC for healthy males according to predictive equations (see the text for references).

Fig. 4. Linear-regression fit of MLUM and MLUS as a function of vital capacity (VC).

Fig. 3. MLU in morphemes (MLUM) and syllables (MLUS) as a function of age.

Table 2Range of correlation between MLUS and MLUM across age groups;p< .001 throughout (n>71 utterances per speaker).

Age group (yrs) r

5–7 .89–.9410–12 .84–.9415–17 .89–.9920–22 .89–.9825–27 .90–.97Overall .92

V.J. Boucher, B. Lalonde / Journal of Phonetics 52 (2015) 58–6964

co-varies with subjects' VC [pr(47)¼ .39, p¼ .005] but not with their stature [pr(38)¼ .26, p ¼ .101]. These results indicate that, wheneffects of age are kept constant, rising MLUs strongly relate to speakers' increasing VC rather than changes in stature.

As for the hypothesized effects of rising MLUS on the diversity of lexemes, Table 3 lists the correlation coefficients obtained fortype and token ratios. To prevent error inflation of repeated tests on these ratios, a Sidak–Bonferonni adjustment of the procedure-wise alpha was applied [alphaADJ¼1−(1−alphaPW)1/p, where p is the number of procedures; Olejnik, Li, Supattathum, & Huberty,1997]. Thus, error probabilities of .05 and .01 correspond to adjusted alphas of .013 and .003, respectively. Using these test statistics,the coefficients show that ratios of type/utterance and token/utterance overall correlate significantly with MLUS. This is not surprisingsince, in producing longer utterances, speakers will use more tokens. However, one notes that the coefficients in Table 3 vary. Forinstance, type counts of 1-syll. lexemes do not correlate with MLUS whereas longer lexemes present strong correlations, whichsuggest that the length of forms can be a factor. In fact, the effects of MLU become clear when one examines the proportions ofdifferent lexemes of one, two, and three or more syllables that speakers use. Considering these proportions, Fig. 5 shows how risingMLUS relates specifically to a diversification of long lexical chunks, an effect that is not captured by global type or token counts.Hence, it appears that, as utterances become longer, speakers tend to use long lexical forms of 3-syll. or more (r¼ .64, p< .001) asopposed to 1-syll. (r¼−.60, p<.001) and 2-syll. lexemes (r¼−.13, p¼ .374). These results suggest a link between the production ofincreasingly long utterances and the tendency to produce long lexical forms.

Fig. 5. Proportions of lexemes (types) containing one, two, and three or more syllables as a function of MLUS.

Fig. 6. Mean and standard error of % occurrences of lexemes of 3-syll. or more by MLUs (average values for each of the five age groups; see also Fig. 3).

Table 3Correlation between MLUS and counts of nominal lexical chunks by type and token per utterance (n¼50).

Size of lexical chunk Type/utterance Token/utterance

1 syll. .30 .52⁎⁎

2 syll. .79⁎⁎

.87⁎⁎

3 or more syll. .81⁎⁎

.80⁎⁎

overall .81⁎⁎

.67⁎⁎

⁎⁎ p is smaller than the Sidak-adjusted alpha of .01, or .003.

V.J. Boucher, B. Lalonde / Journal of Phonetics 52 (2015) 58–69 65

To clarify this link, we examined how the correlation of r¼ .64 between MLUS and proportions of long lexemes would hold if onepartials out the effects of speakers' VC. The results showed that controlling for the effects of VC removes the correlation [pr(47)¼ .18,p¼ .184], indicating that growing VC has a strong intervening effect. This suggests that rises in MLUs, which link to the growth of VC,may influence the production of increasing numbers of long lexemes. This idea is further illustrated in Fig. 6, where one can see thatproportions of long lexemes in speech vary with group-average MLUS. Overall, one sees that speakers' growing capacity to producelong MLUs favors the use of long lexical forms.

6. Discussion and conclusion

Spoken language involves the production of pulses of air within breath units constituting utterances. Accepting this trivial principle,it may seem evident that speakers' growing breath capacities would influence the length of utterances and the number of units theycontain. The above results basically confirm this view: as predicted, there are strong correlations between VC and MLU (Fig. 3), andthe production of increasingly long utterances associates not only with a rising numbers of morphemes and syllables (Figs. 3 and 4,Table 2) but also numbers of differing lexemes (Figs. 5 and 6, Table 3). These observations call into question interpretations of MLUas simply the product of developing “knowledge”. Such interpretations have, since Brown (1976), overlooked the impact of growingbreath capacities on utterance length. In fact, it is revealing that, in proposing the MLU index, Brown (1973, p. 409) rejected a priori

V.J. Boucher, B. Lalonde / Journal of Phonetics 52 (2015) 58–6966

that MLU in morphemes could vary with numbers of syllables seen to relate to modulations of air pressure. But as the results ofTable 2 show, MLU counts in morphemes and syllables are highly correlated throughout development, much as in the case of pre-school children (Arlman-Rupp et al., 1976; Ekmekci, 1982). The analyses also indicate that, even though MLU in morphemes riseswith age, partialling out the effects of VC substantially reduces the correlation, while partialling out the number of syllables inutterances removes the correlation. Hence, contrary to Brown's opinion, MLU in morphemes does not vary independently of thenumber of syllables in utterances, and both measures correlate with the growth of speaker's breath capacities. This leads one torecognize that maturational changes in speech breathing can affect several indices of language development. In particular, thefindings suggest that the growth of VC offers the possibility of producing increasingly long utterances, which favors a diversification oflexical forms. In this perspective, aspects of language development that are indexed by MLU and type counts of lexical units appearto intertwine with the growth of production processes.

On this interpretation, one objection that may come to mind is that, in many cases and situations, speech breathing and MLU varyin ways that are unrelated to verbal proficiency. For instance, an individual with a respiratory disease might produce short MLUs butstill express age-appropriate verbal content. Conversely, speakers with a cognitive deficit may produce diminished MLUsindependently of their respiratory functions. But such cases do not undermine the observation that, in normally developingpopulations, MLU correlates with the growth of breath capacities (as in Figs. 3 and 4). In this light, the above data should not be takento suggest a causal link between VC and MLU: obviously, numerous conditions relating to an individual's cognition, perception, orproduction (etc.) can lead to values in that may not accord with the above correlations. However, as in the example of babbling wheregrowing production processes present necessary but not sufficient factors in accounting for the rise of babble at 6–8 months, thegrowth of breath capacities is a necessary (though not sufficient) factor that can explain, for instance, why MLUs of about 11 syllablesnormally arise in the speech of 12 year-olds or why MLUs cease to grow at 20–22 years. In short, the above data can be seen toreflect a relationship present in normal populations where the growth of breath capacities stands as a necessary factor that shapesthe developmental course of MLU and related indices.

In observing the effects of this factor, it could be argued that static measures of VC do not reflect the dynamics of speechbreathing or articulatory factors that could also influence MLU. In applying a capacity measure, we referred to reports showing thatchildren and adults use similar ratios of their VC in speaking (Fig. 1). While these ratios remain relatively constant, VC more thantriples as speakers enter adulthood – and this narrowly conforms to the above observations of rising MLU (see Figs. 2 and 3). Thus,measures of VC can capture the influence of growing breath volumes on MLU, and such influence would logically apply throughoutdevelopment, even for infant vocalizations as some suggest (Boliek, Hixon, Watson et al., 1996). This does not deny that articulation,respiratory dynamics, and other maturing production processes may also affect the length of utterances and hence provide furtherevidence to counter a one-sided interpretation that rising MLU reflects knowledge. The above observations simply make it clear that,among these factors, the growth of breath capacities has a major impact.

Finally, many analysts trained in linguistic theory might dismiss the preceding findings as irrelevant on the view that the object oflanguage study is syntactic competence (e.g. Chomsky, 1980/2005) or a semantic-conceptual structure of the mind (e.g. Jackendoff,2009) separate from modalities of expression. However, one can only study “language” by observing behaviors in a communicationmodality, and it needs to be asked if analyzing communicative behaviors via transcripts and assumed syntactic concepts does notneglect the structural effects of a medium of expression. For example, in contrast to gestural signs, the production of sequences ofsounds in utterances normally follows the growth of speakers' breath capacities. In analyzing speech in terms of strings of letters andwords, it may be forgotten that producing utterances requires the release of air and that brain processes serving to elaboratesequences of sounds would necessarily develop in conformity with respiratory constraints on production. Relegating such factors to abroad language faculty separate from a syntactic capacity (as in Hauser, Chomsky, & Fitch, 2002) hardly eliminates their structuraleffects on spoken language. Moreover, one cannot gain an understanding of the scope of these structural effects on languageprocesses guided by linguistic views that deny their relevance.

To illustrate the limitations of these views, consider the absence of a working explanation for observed correlations between MLUand developing lexical diversity (as noted by DeThorne et al. 2005). The tradition of analyzing speech via transcripts and conceptualunits of writing like letters, words, and sentences can thwart a view of how factors of “performance” can shape such things as the sizeof utterances and lexical forms. For instance, in linguistics, utterances are often interpreted as sentences rather than as breath unitsof speech, which precludes a consideration of the impact of growing breath capacities on verbal behavior. The present results,however, suggest the need to weigh effects of maturing production and learning processes. Specifically, the above findings suggestthat the capacity to produce increasingly long utterances can favor the production of long lexical forms. This does not imply that thegrowth of VCs is a “sufficient” factor in accounting for the rise of long forms. Certainly the development of lexical forms requires anexposure to a sociolect and marshals various cognitive capacities. Proponents of usage-based grammars make the point thatchildren acquire a grammar and a lexical repertoire via probabilistic exposure to verbal forms (for an overview, see Beckner et al.,2009; Ellis, 2002; Tomasello, 2003). However, it would be odd to claim that speakers' rising use of long lexemes which accompaniesdeveloping MLUs, as seen in Figs. 5 and 6, derives from an increasing exposure to long forms that somehow levels-off at about 21years of age. Obviously, some structural effect on the use of long forms is involved. But one should also note that the rising use oflong lexemes is not better explained by invoking an unfolding competence in manipulating grammatical concepts. For one thing, it isnot the semantic or syntactic concepts expressed by users that account for the diversification of lexemes specifically in terms of theirsize. In explaining such development, the above results suggest one account: the capacity to produce increasingly long utterancescreates rising possibilities to combine elements in long chunks, which can consolidate as lexical forms. By this account, however, the

V.J. Boucher, B. Lalonde / Journal of Phonetics 52 (2015) 58–69 67

course of developing forms refers to the contributing effects of maturing production processes, not to the separate unfolding of asyntactic competence or conceptual attributes of the mind.

Acknowledgments

This paper benefited greatly from the comments of three anonymous reviewers. We thank John Ohala for his discussion of theoriginal text especially in reference to issues of respiratory measures in young children.

Appendix

The following summarizes the guidelines that were used by the judges to perform MLU counts of French speech and counts ofnominal lexical chunks. Note that, in the present study, morpheme counts are used to illustrate a traditional linguistic analysis butthat one cannot assume that speakers are manipulating all counted morphemes (see Section 2.2). For instance, although onemay analyze chunks like He's, I'm, It's (etc.) as involving separate morphemes (pronoun+verb), it is questioned whether allspeakers are manipulating these elements (Beckner & Bybee, 2009). Such cases are equally problematic in French. For instance,it is difficult to divide Elle est which can be produced as a single long vowel [eː] or Il est produced as [je] (etc.). On the other hand,“nominal lexemes” or lexicalized chunks that function as nouns are easily divided in utterances.

Utterance segmentation and sampling1- An utterance is speech between two heard inspiration noises.2- At least 71 utterances of uninterrupted speech are used with no extreme fluctuations on the final cumulative average (see Fig. 1

and accompanying discussion).3- Close-ended responses to questions such as yes, no, I don't know (etc.) are omitted.4- Utterances containing laughter, sighs, coughs, and the like, or utterances where it is difficult to identify breath noise or numbers of

syllables are omitted.5- The last utterance in a conversational turn is omitted.

Syllable counts1- Every relative rise in intensity perceived as marking speech rate is counted as a syllable.2- Filled hesitations such as mmm, uh…(etc.) or repetitions are counted in syllables according to the speech rate of the utterance

context. For instance, if the utterance is produced at four syll./sec and a filled hesitation lasts about 250 ms, count 1.

Morpheme counts1- All IPA signs that can be commuted or omitted to create different meaningful sequences are counted. For example, [t] and [a] in “ta

fille” are counted [t+a+fɪj] because [t] can be commuted with [m+] or [s+] and [a+] with [ɛ+]. [fɪj] is the nominal lexeme. In “ilcontrôlerait” pronounced [i+ l+kõtʁol+r+ɛ], [l+ ] can be commuted with [t+], and [r+] can be omitted to create the imperfect tense.Lexemes, locutions, or expressions that are generally used as undivided chunks are not divided.

2- Liaisons are not counted as morphemes unless their omission creates a known form.3- Repetitions of lexemes or functors are counted.4- Filled hesitations and false starts are not counted.5- Utterances containing unintelligible forms or morphemes are omitted.

References

Agostoni, E., & Hyatt, R. E. (1986). Static behavior of the respiratory system Section 3. In A. P. Fishman, P. T. Macklem, J. Mead, & S. R. Geiger (Eds.), Handbook of physiology: Therespiratory system, Vol. 3 (pp. 113–130). Bethesda, MD: American Physiological Society.

Arlman-Rupp, A. J. L., van Niekerk de Haan, D., & van de Sandt-Koenderman, M. (1976). Brown's early stages: Some evidence from Dutch. Journal of Child Language, 3, 267–274.American Thoracic Society (1995). Standardization of spirometry: 1994 update. American Journal of Respiratory and Critical Care Medecine, 152, 1107–1136.Bates, E., & Goodman, J. C. (2001). On the inseparability of grammar and the lexicon: Evidence from acquisition. In M. Tomasello, & E. Bates (Eds.), Language development: The

essential readings (pp. 134–162). Malden, MA: Blackwell.Beckner, C., Blythe, R., Bybee, J., Christiansen, M. H., Croft, W., Ellis, N. C., et al. (2009). Language is a complex adaptive system: Position paper. Language Learning, 59, 1–26.Beckner, C., & Bybee, J. (2009). A usage-based account of constituency and reanalysis. Language Learning, 59, 29–48.Behrens, H. (2006). The input–output relationship in first language acquisition. Language and Cognitive Proceses, 21, 2–24.Blake, J., Quartaro, G., & Onorati, S. (1993). Evaluating quantitative measures of grammatical complexity in spontaneous speech samples. Journal of Child Language, 20, 139–152.Boliek, C. A., Hixon, T. J., Watson, P. J., & Jones, P. (1996). Vocalization and breathing during the first year of life. Journal of Voice, 10, 1–22.Brown, R. (1973). A first language: The early stages. Cambridge, MA: Harvard University Press.Bybee, J. (2006). From usage to grammar: The mind's response to repetition. Language, 82, 711–733.Bybee, J. (2010). Language, Usage, and Cognition. Cambridge: Cambridge University Press.Chabon, S. S., Kent-Udolf, L., & Egolf, D. B. (1982). The temporal reliability of Brown's mean length of utterance (MLU-M) measure with post-stage V children. Journal of Speech and

Hearing Research, 25, 117–124.Chan, A., McAllister, L., & Wilson, L. (1998). An investigation of the MLU-age relationship and predictors of MLU in 2- and 3-year old Australian children. Asia Pacific Journal of Speech,

Language and Hearing, 3, 97–108.Charness, N., Park, D. C., & Sabel, B. A. (2001). Communication, technology and aging: Opportunities and challenges for the future. New York: Springer.Chomsky, N. (1957). Syntactic structures. The Hague: Mouton.

V.J. Boucher, B. Lalonde / Journal of Phonetics 52 (2015) 58–6968

Chomsky, N. (1967/2006). Language and mind. Cambridge, NY: Cambridge University Press.Chomsky, N. (1980/2005). Rules and representations. New York: Columbia University Press.Conant, S. (1987). The relationship between age and MLU in young children: A second look at Klee and Fitzgerald's data. Journal of Child Language, 14, 169–173.Coulmas, F. (1989). Units of speech and units of writing. The writing systems of the world (pp. 37–54)Oxford: Basil Blackwell37–54.Coulmas, F. (2003). Writing systems: An introduction to their linguistic analysis. Cambridge: Cambridge University Press.Crystal, D. (1974). Review of the book a first language: The early stages. Journal of Child Language, 1, 289–307.Desmond, K. J., Allen, P. D., Demizio, D. L., et al. (1997). Redefining end of test (EOT) criteria for pulmonary function testing in children. American Journal of Respiratory and Critical Care

Medecine, 156, 542–545.DeThorne, L. S., Johnson, B. W., & Loeb, J. W. (2005). A closer look at MLU: What does it really measure?. Clinical Linguistics & Phonetics, 19, 635–648.Eisenberg, S. L., Fersko, T. M., & Lundgren, C. (2001). The use of MLU for identifying language impairment in preschool children: A review. American Journal of Speech-Language

Pathology, 10, 323–342.Ekmekci, F. O. (1982). Language development of a Turkish child: A speech analysis in terms of length and complexity. Journal of Human Sciences, 1, 103–112.Ellis, N. C. (2002). Frequency effects in language processing: A review with implications for theories of implicit and explicit language acquisition studies in second language. Acquisition,

24, 143–188.Fagan, M. K. (2009). Mean length of utterance before words and grammar: Longitudinal trends and developmental implications of infant vocalizations. Journal of Child Language, 36,

495–527.Ferris, B. (1978). Epidemiology standardization project. American Review of Respiratory Disease, 118, 75–83.Fries, C. C. (1952). The structure of English: An introduction to the construction of English sentences. New York, NY: Harcourt, Brace & World.Gavin, W. J., & Giles, L. (1996). Sample size effects on temporal reliability of language sample measures of preschool children. Journal of Speech and Hearing Research, 39, 1258–1262.Harris, R. (1990). Redefining linguistics. In H. G. Taylor, & T. J. Taylor (Eds.), Redefining linguistics (pp. 18–32). London: Routledge.Hauser, M. D., Chomsky, N., & Fitch, W. T. (2002). The faculty of language: What is it, who has it, and how did it evolve?. Science, 298, 1569–1579.Heilmann, J. J., Miller, J. F., & Nockerts, A. (2010). Using language sample databases. Language, Speech, and Hearing Services in Schools, 41, 84–95.Hewitt, L. E., Scheffner-Hammer, C., Yont, K. M., & Tomblin, J. B. (2005). Language sampling for kindergarten children with and without SLI: Mean length of utterance, IPSYN, and NDW.

Journal of Communication Disorders, 38, 197–213.Hickey, T. (1991). Mean length of utterance and the acquisition of Irish. Journal of Child Language, 18, 553–569.Hirano, M., Kurita, S., & Nakashima, T. (1983). Growth, development, and aging of human vocal folds. In D. M. Bless, & J. H. Abbs (Eds.), Vocal fold physiology: Contemporary research

and clinical issues (pp. 22–43). San Diego, CA: College-Hill Press.Hixon, T. J., Goldman, M. D., & Mead, J. (1973). Kinematics of the chest wall during speech production: Volume displacements of the rib cage, abdomen, and lung. Journal of Speech and

Hearing Research, 16, 78–115.Hoit, J. D., & Hixon, T. J. (1987). Age and speech breathing. Journal of Speech and Hearing Research, 30, 351–366.Hoit, J. D., Hixon, T. J., Watson, P. J., & Morgan, W. J. (1990). Speech breathing in children and adolescents. Journal of Speech, Language, and Hearing Research, 33, 51–69.Iverson, J. M., Hall, A. J., Nickel, L., et al. (2007). The relationship between reduplicated babble onset and laterality biases in infant rhythmic arm movements. Brain and language, 101,

198–207.Jackendoff, R. (2009). Language, consciousness, culture: Essays on mental structure. Cambridge, MA: MIT Press.Jalilevand, N., & Ebrahimipour, M. (2014). Three measures often used in language samples analysis. Journal of Child Language Acquisition and Development, 2, 1–12.Johnston, J. R. (2001). An alternate MLU calculation: Magnitude and variability of effects. Journal of Speech, Language, and Hearing Research, 44, 156–164.Justice, L. J., McGinty, A. S., Zucker, T., Cabell, S. Q., & Piasta, S. B. (2013). Bi-directional dynamics underlie the complexity of talk in teacher–child play-based conversations in

classrooms serving at-risk pupils. Early Childhood Research Quarterly, 28, 496–508.Kent, R. D. (1984). Psychobiology of speech development: Coemergence of language and a movement system. American Journal of Physiology, 15, R888–R894.Kent, R. D., & Forner, L. L. (1980). Speech segment durations in sentence recitations by children and adults. Journal of Phonetics, 8, 157–168.Kent, R. D., & Vorperian, H. K. (1995). Development of the craniofacial-oral laryngeal anatomy: A review. Journal of Medical Speech-Language Pathology, 3, 145–190.Kent, R. D. (1994). Reference manual for communicative sciences and disorders: Speech and language. Austin, TX: PRO-ED.Klee, T. (1992). Developmental and diagnostic characteristics of quantitative measures of children's language production. Topics in Language Disorder, 12, 28–41.Klee, T., & Fitzgerald, M. D. (1985). The relation between grammatical development and mean length of utterance in morphemes. Journal of Child Language, 12, 251–269.Klee, T., Schaffer, M., May, S., Membrino, I., & Mougey, K. (1989). A comparison of the age-MLU relation in normal and specifically language-impaired preschool children. Journal of

Speech and Hearing Disorders, 54, 226–233.Kress, G. (1994). Learning to write (2nd ed.). London: Routledge.Ladefoged, P. (1967). Three areas of experimental phonetics. London: Oxford University Press.Lanteri, C. J., & Sly, P. D. (1993). Changes in respiratory mechanics with age. Journal of Applied Physiology, 74, 369–378.Lieven, E., Behrens, H., Speares, J., & Tomasello, M. (2003). Early syntactic creativity: A usage-based approach. Journal of Child Language, 30, 333–370.Lieven, E., Pine, J. M., & Baldwin, G. (1997). Lexically-based learning and early grammatical development. Journal of Child Language, 24, 187–220.McDowell, M. A., Fryar, C. D., Ogden, C. L., & Flegal, K. M. (2008). Anthropometric reference date for children and adults: United States, 2003–2006. National Health Statistics Reports,

10, 1–45.Mead, J., Bouhuys, A., & Proctor, D. (1968). Mechanisms generating subglottic pressure. Annals of the New York Academy of Sciences, 155, 177–181.Merkus, P. J. F. M., de Jongste, J. C., & Stocks, J. (2005). Respiratory function measurements in infants and children. Lung function testing. European Respiratory Society Monograph, 31,

166–194.Miles, S., Chapman, R., & Sindberg, H. (2006). Sampling context affects MLU in the language of adolescents with Down syndrome. Speech, Language, & Hearing Research., 49,

325–337.Miller, J. (1999). Magnasyntax and syntactic analysis. Revue Française de Linguistique Appliquée, IV, 7–20.Miller, J., & Weinert, R. (1998). Spontaneous spoken language. Oxford: Clarendon Press.Miller, J. F., & Chapman, R. S. (1981). The relation between age and mean length of utterance in morphemes. Journal of Speech and Hearing Research, 24, 154–161.Miller, J. F., & Chapman, R. S. (2004). Systematic analysis of language transcripts (SALT, Version 8.0). Madison, WI: Language Analysis Laboratory, Waisman Center, University of

Wisconsin-Madison.Miller, J., & Iglesias, A. (2008). Systematic Analysis of Language Transcripts (SALT) research version 2008. Madison, WI: SALT Software LLC.Miller, M. R., Hankinson, J., Brusasco, V., Burgos, F., Casaburi, R., Coates, A., et al. (2005). Standardisation of spirometry. European Respiratory Journal, 26, 319–338.Nespor, M., & Vogel, I. (1986). Prosodic phonology. Dordrecht: Foris.Nieminen, L. (2009). MLU and IPSyn measuring absolute complexity. Estonian Papers in Applied Linguistics (Eesti Rakenduslingvistika Ühingu aastaraamat), 5, 173–185.Ohala, J. J. (1993). The whole body plethysmograph in speech research. Journal of the Acoustical Society of America, 93, 2416.Olejnik, S., Li, J., Supattathum, S., & Huberty, C. J. (1997). Multiple testing and statistical power with modified bonferroni procedures. Journal of Educational and Behavioral Statistics, 22,

389–406.Oller, D. K. (1986). Metaphonology and infant vocalizations. In B. Lindblom, & R. Zetterström (Eds.), Precursors of early speech (pp. 21–35). New York: Stockton Press.Oller, D. K. (2000). The emergence of the speech capacity. Mahwah, NJ: Lawrence Erlbaum.Oller, D. K., & Eilers, R. E. (1988). The role of audition in infant babbling. Child Development, 59, 441–449.Oller, D. K., Eilers, R. E., Bull, D. H., & Carney, A. E. (1985). Prespeech vocalizations of a deaf infant: A comparison with normal metaphonological development. Journal of Speech and

Hearing Research, 28, 47–63.Oller, D. K., & Lynch, M. P. (1992). Infant vocalizations and innovations in infraphonology: Toward a broader theory of development and disorders. In C. A. Ferguson, L. Menn, & C. Stoel-

Gammon (Eds.), Phonological development: Models, research, implications (pp. 509–536). Timonium, MD: York Press.Parham, D. F., Buder, E. H., Oller, D. K., & Boliek, C. A. (2011). Syllable-related breathing in infants in the second year of life. Journal of Speech, Language, and Hearing Research, 54,

1039–1050.Parker, M. D., & Brorson, K. (2005). A comparative study between mean length of utterance in morphemes (MLUm) and mean length of utterance in words (MLUw). First Language, 25,

365–376.Pine, J. M., & Lieven, E. V. M. (1997). Slot and frame patterns and the development of the determiner category. Applied Psycholinguistics, 18, 123–138.Pine, J. M., & Martindale, H. (1996). Syntactic categories in the speech of young children: The case of the determiner. Journal of Child Language, 23, 369–395.Reed, V. A., MacMillan, V., & McLeod, S. (2001). Elucidating the effects of different definitions of utterance on selected syntactic measures of older children's language samples. Asia

Pacific Journal of Speech, Language and Hearing, 6, 39–45.

V.J. Boucher, B. Lalonde / Journal of Phonetics 52 (2015) 58–69 69

Rice, M. L., Redmond, S. M., & Hoffman, L. (2006). Mean length of utterance in children with specific language impairment and in younger control children shows concurrent validity andstable and parallel growth trajectories. Journal of Speech, Language, and Hearing Research, 49, 793–808.

Rice, M. L., Smolik, F., & Perpich, D. (2010). Mean length of utterance levels in 6-month intervals for children 3 to 9 years with and without language impairments. Journal of Speech,Language, and Hearing Research, 53, 333–349.

Rollins, P. R. (1995). MLU as a matching variable: Understanding its limitations. Paper presented at the Annual Convention of the American Speech-Language-Hearing Association,Orlando, FL.

Rollins, P. R., Snow, C. E., & Willett, J. B. (1996). Predictors of MLU: Semantic and morphological developments. First Language, 16, 243–259.Rom, A., & Leonard, L. (1990). Interpreting deficits in grammatical morphology in specifically language-impaired children: Preliminary evidence from Hebrew. Clinical Linguistics &

Phonetics, 4, 93–105.Rondal, J. A., & Defays, D. (1978). Reliability of mean length of utterance as a function of sample size in early language development. Journal of General Psychology, 133, 305–306.Rondal, J. A., Ghiotto, M., Bredart, S., & Bachelet, J.-F. (1987). Age-relation, reliability and grammatical validity of measures of utterance length. Journal of Child Language, 14, 433–446.Rondal, J. A., & Comblain, A. (1996). Language in adults with Down syndrome. Down Sydndrome Research and Practice, 4, 3–14.Russell, N. K., & Stathopoulos, E. (1988). Lung volume changes in children and adults during speech production. Journal of Speech and Hearing Research, 31, 146–155.Scarborough, H., Rescorla, L., Rager-Flusberg, H., Fowler, A. E., & Sudhalter, V. (1991). The relation of utterance length to grammatical complexity in normal and language disordered

groups. Applied Psycholinguistics, 12, 23–45.Scarborough, H., Wyckoff, J., & Davidson, R. (1986). A reconsideration of the relation between age and mean utterance length. Journal of Speech and Hearing Research, 29, 394–399.Smith, B. L. (1978). Temporal aspects of English speech production: A developmental perspective. Journal of Phonetics, 6, 37–67.Smith, A., & Zelaznik, H. N. (2004). Development of functional synergies for speech motor coordination in childhood and adolescence. Developmental Psychobiology, 45, 22–33.Stagg, D., Goldman, M., & Davis, J. N. (1978). Computer-aided measurement of breath volume and time components using magnetometers. Journal of Applied Physiology, 44, 623–633.Stanojevic, S., Wade, A., Stocks, J., Hankinson, J., Coates, A.L, & Pan, H. (2008). Reference ranges for spirometry across all ages: A new approach. American Journal of Respiratory and

Critical Care Medecine, 177, 253–260.Stathopoulos, E. T., & Sapienza, C. M. (1997). Developmental changes in laryngeal and respiratory function with variations in sound pressure level. Journal of Speech and Hearing

Research, 40, 595–614.Thom, S. A., Hoit, J. D., Hixon, T. J., & Smith, A. E. (2006). Velopharyngeal function during vocalization in infants. Cleft Palate–Craniofacial Journal, 43, 539–546.Tomasello, M. (1992). First verbs: A case study of early grammatical development. Cambridge: Cambridge University Press.Tomasello, M. (2000a). Do young children have adult syntactic competence?. Cognition, 74, 209–253.Tomasello, M. (2000b). The item-based nature of children's early syntactic development. Trends in Cognitive Sciences, 4, 156–163.Tomasello, M. (2003). Constructing a language. A usage-based theory of language acquisition. Cambridge: Harvard University Press.Vaissière, J. (1983). Language-independent prosodic features. In A. Cutler, & D. R. Ladd (Eds.), Prosody: Models and Measurements (pp. 53–66). New York: Springer-Verlag.van de Weijer, J. (1998). Language input for word discovery Doctoral dissertation. Nijmegen: Katholieke Universiteit (MPI Series in Psycholinguistics 9, Max-Planck-Institute for

Psycholinguistics, Nijmegen).van der Stelt, J. M., & Koopmans-van Beinum, F. J. (1986). Early stages in the development of speech movements. In B. Lindblom, & R. Zetterström (Eds.), Precursors of early speech (pp.

37–50). New York: Stockton Press.Wray, A. (2008). Formulaic language: Pushing the boundaries. New York: Oxford University Press.