GLICKMAN, Associate Judge:
The government filed neglect petitions for seven-year-old Jas.J. and her five-year-old brother Jam.J. on March 17, 1999. The petitions alleged that their mother B.A. and her boyfriend K.C. had physically abused Jas.J. on March 10,1999 by administering excessive corporal punishment in the form of spankings, and that Jam.J. was in imminent danger of similar abuse. See D.C.Code § 16-2301(9)(A) & (E) (2001). After a fact finding hearing, the trial court found that K.C. had subjected both children to a pattern of physical abuse from which B.A. had failed to protect them. As a result, the court found that both Jas.J. and Jam.J. were neglected under D.C.Code § 16-2301(9).
On appeal from the trial court’s findings of neglect, B.A. and K.C. make three claims. First, B.A. disputes the sufficiency of the evidence of neglect. Second, B.A. argues that the trial court abused its discretion when, out of concern for the children’s welfare, it refused to let her cross-examine Jas.J. and Jam.J. after their hearsay statements were introduced as the primary evidence against her. Third, K.C. argues that the trial court erred in not allowing him to present expert testimony because he had not disclosed the expert to the government and the guardian ad litem before trial.
We reject B.A.’s first claim. The evidence heard by the trial court was sufficient to support its conclusion that the children were neglected. We agree, however, that B.A. and K.C. are entitled to relief on their other claims. In the absence of factual findings with support in the record that the probable harm to the children from having to testify substantially outweighed B.A.’s need for their testimony, the trial court should not have prevented B.A. from calling the children as witnesses. In addition, because K.C. had [907]*907no duty to disclose his expert witness before trial in the absence of a discovery request or a pretrial order requiring such disclosure, the trial court should not have precluded him from calling that witness to testify. In view of these determinations, we vacate the findings of neglect and remand for further proceedings.1
I.
A.
At trial, the government relied on the testimony of Detective Cheryl Wright-Taylor of the Metropolitan Police Department and Dr. Craig DeWolfe and Dr. Mu-nisha Mehra of the Children’s National Medical Center. The guardian ad litem for Jas.J. and Jam.J. called Dr. Rosa Herring of the Children’s National Medical Center. Except for Dr. Mehra, each of these witnesses testified to statements that the children had made to them regarding their mistreatment by K.C. and B.A. These out-of-court statements were central to the government’s case because the children were not called to testify and there were no other witnesses to their alleged abuse. B.A. and K.C. each testified and denied the allegations of abuse and neglect. B.A.’s mother and two witnesses from the children’s school testified in support of B.A. However, the children’s father W.J., whom B.A. called as a hostile witness, gave further testimony that the children had described abusive treatment from K.C. and B.A.2
Detective Wright-Taylor testified that on March 15, 1999 she received and investigated a report that Jas.J. had suffered physical abuse. Detective Wright-Taylor observed discoloration and bruising on Jas. J.’s buttocks, which Jas.J. said were still sore.3 The detective asked Jas.J. how she got the bruises, and the child told the detective that K.C. had spanked her on her bare buttocks in a public park. Detective Wright-Taylor then spoke to Jam.J., who confirmed his sister’s account of the spanking. No objection was made to the admission of this testimony about Jam.J.’s or Jas.J.’s hearsay statements.
Dr. DeWolfe, a pediatric resident at Children’s National Medical Center, testified that he talked with Jas.J. about the spanking while taking a medical history from her on March 16,1999. According to Dr. DeWolfe, Jas.J. told him that K.C. had spanked her in a park and then B.A. had spanked her at home with a belt. Dr. DeWolfe reported that Jas.J. also told him that K.C. and B.A. had each spanked her the day before the spanking in the park. K.C.’s counsel objected to the admission of JasJ.’s statements as hearsay. The court overruled the objection on the ground that Jas.J.’s statements were made for the purpose of medical diagnosis.4
Dr. Mehra, a pediatrician, testified that she examined Jas.J. after Dr. DeWolfe saw her. Dr. Mehra found bruises on JasJ.’s buttocks that were tender to the touch and that were consistent with the spanking Jas.J. had described.
[908]*908Dr. Herring, a therapist and social worker who had treated Jas.J. and Jam.J., testified as a treating physician and an expert in child psychotherapy. Over objection, Dr. Herring recounted statements that Jas.J. and Jam.J. had made to her in several therapy sessions. Dr. Herring testified that Jas.J. said that K.C. and B.A. each beat her on March 10,1999 after they learned that she had misbehaved at school that day. Jas.J. said K.C. took her to the park after school, pulled down her underpants and spanked her. Jas.J. told her mother about this spanking, and B.A. allegedly said “good. I’m going to spank you again.” Dr. Herring testified that Jam.J. told her about other abusive incidents. In one therapy session, Jam. J. said that after he had accidentally urinated on the wall, K.C. had made him lick up the urine. On another occasion Jam.J. told Dr. Herring that KC. had “touched his butt”; however, Jam.J. later retracted that statement. Dr. Herring believed that JamJ.’s retraction did not affect his credibility. She testified that, in her opinion, the details and consistency of the children’s statements to her suggested truthfulness. Dr. Herring diagnosed both Jas.J. and Jam.J. as having adjustment disorders and opined that they were “traumatized” by the physical abuse they reported from B.A. and K.C.
K.C. denied spanking Jas.J. on March 10 and testified that he had never physically disciplined either child. B.A. echoed that testimony. B.A. testified that she herself did spank her children occasionally but she insisted that she did not spank Jas.J. on March 10, 1999. B.A. had no explanation for the bruises found on Jas.J. on March 15 and 16,1999.
B.A.’s mother J.A. testified that she had never seen B.A. or K.C. physically discipline the children and that Jas.J. “tells lies.” Patricia Macantoosh-Lucas, an assistant teacher in JasJ.’s classroom, testified that Jas.J. had never told her that she had been beaten by B.A. or K.C. Ms. Macantoosh-Lucas also testified that she had observed that Jas.J. and Jam.J. had a “loving” relationship with KC. and would always run to him when he picked them up from school. Kathleen Pardue, the school principal, similarly testified that she had not observed any signs of abuse and that the children had a good relationship with KC.
Although W.J., the natural father of Jas.J. and Jam.J., was called to the stand by B.A., his testimony corroborated the allegations of the neglect petitions. W.J. testified that he had seen the bruises on Jas.J.’s buttocks after March 10, 1999, and that Jas.J. had told him that K.C. had beaten her in a park. W.J. personally had never seen KC. mistreat either of his children, but, he testified, Jas.J. had told him that KC. beat them regularly and Jam.J. had told him that KC. made him lick up urine. The children’s hearsay statements recounted by W.J. were admitted without objection.
B.
On the second day of trial, B.A.’s counsel requested permission to interview Jas.J. informally to determine “whether she can recall correctly what happened” on March 10, 1999. The guardian ad litem opposed this request, stating that Jas.J. was “reluctant to talk” about the allegations of abuse, in part because B.A. “has told both children that ... [B.A.] will go to jail if they talk about this.” The court deferred to the guardian ad litem and declined to grant B.A.’s request. B.A.’s counsel then requested permission to call Jas.J. as a witness. The court responded negatively:
No. Absolutely not. If the Guardian Ad Litem has indicated that the child will [909]*909not speak informally, the Court does not require or force children to testify in proceedings of this nature. This is a civil case. This is not a criminal case. This ease is aimed towards the best interests of the children.
K.C.’s counsel then requested permission to call both children as witnesses, arguing that because the government was introducing their hearsay statements he should be allowed “to cross-examine the children as to what they say [K.C.] did.” The guardian ad litem opposed the request. The guardian ad litem argued that Jas.J. and Jam. J. already had given consistent accounts of the March 10, 1999 incident to several different people, and claimed that “[t]hey don’t want to talk about it.” She further argued that if the children were called to testify, they would suffer “an emotional impact” from “telling the story” and from “speaking in front of their parents.” B.A.’s counsel suggested that the court could mitigate any such impact by arranging the testimony so that the children could not see their parents or K.C. The court rejected that suggestion without comment and, relying exclusively on the guardian ad litem’s representations, again ruled that the children were not to be examined:
With the GAL’s opposition and [because] the GAL is the legal representative of these children, the Court is not going to require them to testify.... [T]he Court is not going to force them in opposition of their legal representative and fiduciary who is supposed to do what is in their best interests.
Later in the trial, after the guardian ad litem presented Dr. Herring, B.A. and K.C. sought to present their own expert witness testimony. B.A.’s counsel advised the court that she intended to call Dr. Lanning Moldauer, a psychologist, to rebut Dr. Herring’s testimony. The government objected on the ground that B.A. had not given notice before trial of her intent to call an expert, either informally or by filing a statement identifying Dr. Moldauer pursuant to Superior Court Civil Rule 26(b)(4)(A)(i). The government did not claim that it had requested pretrial identification of B.A.’s expert witnesses. K.C.’s counsel then stated that he too planned to rebut Dr. Herring by calling an expert, in his case a psychiatrist named Dr. Willie Hamlin who reported in a November 4, 1998 letter that he had found no evidence of abuse when he examined Jas.J. and Jam.J. in connection with B.A.’s divorce proceeding. K.C.’s counsel claimed that he had not known of Dr. Hamlin’s existence until the day of the hearing, when B.A.’s counsel gave him a copy of the November 4, 1998 letter. Both the government and the guardian ad litem objected to K.C.’s request to call Dr. Hamlin.
The trial court directed K.C. and B.A. to submit written motions in support of their requests for leave to call their experts. In their motions B.A. and K.C. each pointed out that no party to the neglect proceeding had sought to discover their expert witnesses before trial, either through formal interrogatories or otherwise. That representation was never challenged.
Thereafter, the trial court denied B.A. and KC.’s motions in a written order. Without commenting on their argument that they had never been asked whether they intended to call expert witnesses, the court analyzed their right to call experts under the multi-factor test of Weiner v. Kneller, 557 A.2d 1306 (D.C.1989).5 The [910]*910court found that the government and guardian ad litem would suffer incurable prejudice if B.A. and K.C. were allowed to call expert witnesses whom they had not disclosed before trial. The court further found that B.A. and K.C. had not shown that they would be incurably prejudiced by the preclusion of their experts’ testimony. The court acknowledged that “[t]he testimony of Dr. Hamlin seems especially probative of the issue of abuse charged on the part of [K.C.] and could conceivably do a great deal to refute these allegations.” Nevertheless, the court concluded that what it deemed K.C. and B.A.’s “willful disregard for the evidentiary rules” weighed in favor of denying their motions. The court assumed that B.A. and K.C. had failed to abide by Rule 26(b)(4), but it did not explain how B.A. or K.C. had violated this or any other “evidentiary rule.”
C.
In an October 5, 2000 order, the trial court found that Jas.J. and Jam.J. were neglected children under D.C.Code § 16-2301(9). The court explicitly credited the testimony of Dr. Herring, Dr. DeWolfe, Dr. Mehra, Detective Wright-Taylor and W.J., and implicitly credited the hearsay statements of Jas.J. and Jam.J. The court found B.A.’s testimony helpful “in understanding the family dynamics,” but not credible as to the events of March 10, 1999. The court did not believe the testimony of K.C. or J.A.6 The court found that K.C. had used excessive corporal punishment in spanking Jas.J. in a park on March 10,1999, and that B.A. had followed up this beating by spanking Jas.J. with a belt. The March 10, 1999 spanking “was part of a pattern of abusive and violent behavior [by K.C.] towards [Jas.J.] and [Jam.J.],” as shown by the children’s reports that K.C. had made Jam.J. lick up his urine and had beaten them both on prior occasions. The court found that B.A. refused to acknowledge KC.’s abuse and had failed to protect Jas.J. and Jam.J. from K.C.
II.
B.A. argues that the evidence was insufficient to support a conclusion that she neglected Jas.J. or Jam.J. We must, however, “view the evidence in the light most favorable to the District and draw every reasonable inference in the District’s favor.” In re E.H., 718 A.2d 162, 168-69 (D.C.1998); see D.C.Code § 17-305(a) (2001). Our review must also respect “the prerogative of the [trial] judge, as the trier of fact, to determine credibility and weigh the evidence.” In re S.G., 581 A.2d 771, 775 (D.C.1990). In light of these familiar principles, B.A. cannot prevail on her sufficiency challenge.
B.A. bases that challenge on the premise that the government only presented evidence that K.C. had inflicted excessive punishment, not evidence that B.A. had also abused the children. This premise is dubious, since Dr. DeWolfe and Dr. [911]*911Herring both reported JasJ.’s statements that B.A. beat her with a belt on March 10, 1999 alter K.C. had spanked her. Even if that evidence, combined with Jas.J.’s observed physical injuries, did not suffice to prove that B.A. herself had physically abused her daughter, the evidence that B.A. failed to protect her children from KC.’s abuse and could not explain her injuries after March 10 was sufficient in itself to support a finding of neglect. See D.C.Code § 16-2801(28) (2001) (providing that a child is “abused” when the child’s parent “inflicts or fails to make reasonable efforts to prevent the infliction of physical or mental injury upon the child, including excessive corporal punishment” (emphasis added)); D.C.Code § 16-2316(c) (2001) (providing that evidence of injury to a child who was in the custody of his or her parent for which the parent “can give no satisfactory explanation shall be sufficient to justify an inference of neglect”). The trial court heard B.A. and K.C. testify and determined not only that K.C. was abusing the child but also that B.A. refused to acknowledge or restrain this abuse. We will not upset such a firsthand assessment by an experienced trial judge who had the opportunity to assess the demeanor of the witnesses. See S.G., 581 A.2d at 775.
We turn to the trial court’s refusal to let B.A. and K.C. call Jas.J. and Jam.J. as witnesses and examine them. The court based this ruling on the objection by the guardian ad litem that it would be traumatic for the young children to have to testify about the abuse they had experienced, especially “in front of their parents.” This was an eminently reasonable concern, and certainly a proper and important consideration for the court. Also important, though, was the fact that the children’s hearsay reports of physical mistreatment by B.A. or K.C., which appellants denied and no one else had witnessed, constituted the core evidence of abuse. The court credited that critical evidence in finding that Jas.J. and Jam.J. had been neglected. Appellants’ due process right to challenge that evidence was impaired substantially by their inability to confront the children — their accusers — in court through live cross-examination. Arguably too, the court’s fact finding may have been less informed and rehable than it would have been if the court had seen and heard the children itself.
Until now this court has not had occasion to consider whether or under what circumstances a trial court may deny a parent’s request to call and examine his or her child as a witness in a neglect proceeding or other civil proceeding concerned with the child’s welfare. The closest we have come to the issue has been in termination of parental rights (TPR) proceedings, when we have addressed the trial court’s statutory obligation to consider, “to the extent feasible, the child’s opinion of his or her own best interests in the matter.” D.C.Code § 16 — 2353(b)(4) (2001). TPR proceedings evoke similar, if not the same, concerns about the emotional impact on children from testifying as do neglect proceedings. See, e.g., In re A.R., 679 A.2d 470, 477 (D.C.1996) (acknowledging the validity of the trial judge’s apprehension that a six-year-old child would suffer psychological harm if questioned in camera about his preferences regarding termination and adoption). Those concerns are sensible and important ones, though we have cautioned that “[i]n an area where behavioral generalizations are risky even for experts,” a court should not “too quickly assume that the only outcome from directly involving the child in that inquiry [regarding severing the parental bond] will be harmful.” In re T.W., 623 A.2d 116, 118 (D.C.1993) (footnote omitted). In [912]*912spite of the concerns, we have expressed a strong preference for judges in TPR cases “to hear directly from the children involved ... if it is at all feasible to do so,” adding that “[it] is almost always feasible when ‘the child is old enough (or otherwise competent) to voice ... an opinion’ concerning his or her own best interests.” In re I.B., 631 A.2d 1225, 1232 (D.C.1993) (quoting T.W., 623 A.2d at 117). The I.B. court thought that two boys who were eight and twelve years old “were certainly old enough.” Id. at 1232; cf. Shelton v. Bradley, 526 A.2d 579, 581 (D.C.1987) (characterizing the trial court’s refusal in a child custody case to hear from an “undoubtedly capable” child, who was nine-years-old, as “difficult for us to fathom”).
Our TPR cases also have recognized that the trial court may place appropriate limitations on the examination of the child witness in order to protect the child from emotional or psychological injury. Thus we have said that the trial court has discretion in TPR cases to choose between interviewing the child informally in chambers and “putting that child on the witness stand.” I.B., 631 A.2d at 1232 n. 12; accord, T.W., 623 A.2d at 118 (“At least part of the concern a judge may have about a child’s being asked to ‘take sides’ can be alleviated by a nonadversarial inquiry in camera, without necessarily shrinking from questions about the child’s preference.”); cf. A.R., 679 A.2d at 473, 475, 477 (upholding trial judge’s decision not to interview the child in camera, where the judge instructed the parties that they were “perfectly free” to call the child as a witness if they wished). In T.W., 623 A.2d at 117-18, we approved reasonably tailored restrictions that the trial judge had placed on the scope of questioning to avoid causing the child witness lasting distress. The judge had barred counsel for the mother from asking T.W. directly whether she wanted to be adopted, explaining that such direct questions “could have a very detrimental effect on this child” (emphasis added) by “put[ting her] in a position where she feels that she has to choose between adoption and a mother ... [whom] she loves and cares for a lot.” Id. at 117. The judge had emphasized that he was “particularly” apprehensive about T.W., “a child in every sense of the term” about whom he had heard “evidence of conflicts and anxiety.” Id. at 118. In upholding the trial judge’s restrictions on T.W.’s examination as reasonable, we deemed it significant both that the judge based his ruling on an individualized assessment of T.W. and that counsel was free to explore the child’s feelings through less highly charged questioning. See id. We rejected the mother’s contention that the restrictions were impermissible “in the absence of expert psychological testimony about the harmful effect such questioning likely would have [had] on this particular child.” Id. at 117. I.B. and T.W. suggest that while the child’s testimony is not necessarily “indispensable,” T.W., 623 A.2d at 117, the trial court should strive whenever possible to accommodate the conflicting interests of child and parent by limiting and controlling the child’s examination rather than precluding it entirely.
TPR cases are not necessarily on all fours with neglect cases when it comes to the issue of the parent’s right to call the allegedly neglected child as a witness. The TPR statute requires that the court consider the child’s preferences if possible; there is no comparable statutory requirement in neglect cases. The child’s own interest in being heard may be different in the two types of proceeding. Testifying may expose the child to a greater emotional conflict in a neglect proceeding, when the mistreatment is fresher in the child’s mind but the child may be closer to and more dependent on the parent, than in a [913]*913TPR proceeding, when more time has passed and the strength of the child’s relationship with the parent may have diminished. (In many neglect proceedings, moreover, the child may have no material testimony to offer, or the parent may not want to examine the child — a tactic that can backfire and that many parents would eschew regardless of the need.) The nature of the child’s potential testimony differs in TPR and neglect cases as well. In a TPR case, the child’s testimony will usually be introduced to determine the child’s opinion of his or her best interests, while in a neglect case the child’s testimony will usually be introduced to determine the truth of the allegations in the neglect petition. Nonetheless, TPR and neglect cases are comparable in terms of both the potential need to hear from the child and the need to protect the child if testifying might be traumatic. Parents have fundamental rights at stake in both types of proceeding. See Santosky v. Kramer, 455 U.S. 745, 753, 102 S.Ct. 1388, 71 L.Ed.2d 599 (1982). In a TPR, the parent faces the possibility of a permanent termination of the parent-child relationship; in a neglect proceeding, the parent faces the possibility of serious state interference in that relationship, interference that could lead to a temporary or permanent loss of custody. Both types of proceedings also require the trial court to act as parens patriae to protect the child from harm. See A.R., 679 A.2d at 476 (TPR proceeding); In re S.L.E., 677 A.2d 514, 519 (D.C.1996) (neglect proceeding). Both TPRs and neglect proceedings, therefore, similarly require the trial court to consider the rights of the parent and the well-being of the child.
Appellate court decisions from other jurisdictions offer us further guidance. Several state courts have addressed the scope of a parent’s right to compel a child to testify in a neglect or other civil proceeding concerning that child’s welfare when it is claimed that testifying would be contrary to the child’s best interests. Broadly speaking, the decisions fall into two groups. Some courts have held that trial judges have no authority to preclude a child’s otherwise competent testimony on the ground that testifying might injure the child. See, e.g., Bebee v. Hargrove, 607 So.2d 1270, 1272 (Ala.Civ.App.1992); White v. White, 655 N.E.2d 523, 529 (Ind.Ct.App.1995); In re Faircloth, 137 N.C.App. 311, 527 S.E.2d 679, 683 (2000); Callicott v. Callicott, 364 S.W.2d 455, 458 (Tex.Ct.Civ.App.1963). Other courts have held that judges do have the authority to subordinate the parent’s interest and exclude a child’s testimony when there is substantial evidence that the child would be harmed by having to testify. See, e.g., In re Jennifer J., 8 Cal.App.4th 1080, 10 Cal.Rptr.2d 813, 819 (1992); In re Brandon W., 56 Conn.App. 418, 747 A.2d 526, 531-32 (2000); R.S.M. v. J.D.M., 542 S.W.2d 361, 363 (Mo.Ct.App.1976); In re Beasley, 314 Or. 444, 840 P.2d 78, 84 (1992); In re Child M., 452 Pa.Super. 230, 681 A.2d 793, 798 (1996).
The opinion of the North Carolina Court of Appeals in Faircloth typifies the approach of the courts in the first group. In Faircloth, the trial court refused to permit a father to cross-examine his children in spite of the father’s claim that they would testify that he had not abused them. 527 S.E.2d at 680-81. The trial court credited the testimony of the children’s therapists that testifying would be “extremely detrimental” to their emotional health and found that the children were “unavailable and unable to testify at this hearing due to [their] current mental status and the harm to [them] which would occur were [they] to be forced to testify.” Id. at 681. The court of appeals held that the trial court erred in focusing on “the effect that the children’s testifying would have on their [914]*914mental health, rather than upon the ability of the children to understand their obligation to tell the truth and their ability to relate events which they may have seen, heard, or experienced.” Id. at 683. The court acknowledged “the troubling aspects” of forcing a child to testify against an alleged abuser, but it held that the parent’s right to confront witnesses required the trial court to allow the children to testify if it found them competent. Id. The court noted that the trial court still had the discretion to take “appropriate measures” to protect the children during their testimony. Id. at 684.
The courts in the second group have rejected the absolutist approach exemplified by the opinion in Faircloth in favor of balancing the parent’s need to examine the child against the demonstrated risk that the child will be harmed by such examination. This alternative balancing approach is grounded in “a recognition of the overriding objective of the dependency hearing — to preserve and promote the best interests of the child.” Jennifer J., 10 Cal.Rptr.2d at 819 (holding that the trial court had the power to preclude the parents from examining their child in a termination of parental rights hearing “where the issues to be resolved would not be materially affected by the child’s testimony, and where it is shown that the child would be psychologically damaged by being required to testify”). In a leading opinion, the Oregon Supreme Court articulated the balancing test as follows:
Unless the probative value of the child’s testimony is substantially outweighed by the risk of severe emotional or psychological harm to the child from testifying, there is no discretion to exclude the child’s testimony; the child’s testimony must be admitted. If, however, the balance goes against probative worth, the trial court may exclude the evidence.
Beasley, 840 P.2d at 84. Elaborating on this test, the court identified four factors for a trial judge to consider in assessing the risk to the child from testifying in a given case: “(1) the probability of severe emotional or psychological injury to the child as a result of testifying; (2) the degree of anticipated injury; (3) the expected duration of injury; and (4) whether the expected psychological injury is substantially greater than the reaction of an average child who testifies.” Id.
Like the other courts in the second group, the Beasley court emphasized that any determination that the potential harm to the child outweighs the probative value of the child’s testimony must rest on a sufficient evidentiary foundation — “usually,” the court said, the opinions of mental health experts, and “commonly” including an examination of the child in camera. Id. at 84, 85; see, e.g., R.S.M., 542 S.W.2d at 363 (holding that the trial court did not have an adequate basis for excluding the child’s testimony as detrimental to his welfare where the court did not conduct a voir dire or interview the child). This insistence on a firm evidentiary foundation for the conclusion that testifying would be injurious to the child’s welfare reflects the high importance that even the non-absolutist courts attach to the parent’s fundamental right to present relevant testimony in defense of a charge of neglect. That right may not be curtailed, the courts in this second group hold, unless compelling reason to do so is shown to exist.
We are not persuaded to adopt the absolutist approach of cases such as Faircloth. Two vitally important interests are at stake when a parent seeks to compel a child’s testimony in a neglect proceeding. On the one hand, as the courts in both the Faircloth and the Beasley lines of cases appreciate, parents have a “fundamental liberty interest ... in the care, [915]*915custody, and management of their child.” Santosky, 455 U.S. at 753, 102 S.Ct. 1388. Neglect proceedings intrude on this fundamental interest. See In re J.J.Z., 630 A.2d 186, 192 (D.C.1993); In re B.K, 429 A.2d 1331, 1334 (D.C.1981). Such proceedings are the “point of entry” for state intervention in the family relationship, the starting point of a process that may eventuate in erosion and even termination of the bonds between parent and child. The “natural parent’s interest in preserving the parent-child relationship is [therefore] a ‘commanding’ factor in the determination of what process is due to the natural parent” in a neglect proceeding. In re M.M.M., 485 A.2d 180, 184 (D.C.1984) (quoting Lassiter v. Dep’t of Social Servs., 452 U.S. 18, 27, 101 S.Ct. 2153, 68 L.Ed.2d 640 (1981)). Our neglect statute, which sets forth the process that is due, accordingly “affords protections to the parent and child similar to those made available to a criminal defendant.” J.J.Z., 630 A.2d at 192. Among those protections, the statute specifically provides that “[ejvidence which is competent, material, and relevant shall be admissible at factfinding hearings” in neglect cases. D.C.Code § 16-2316(b) (2001) (emphasis added). The statute does not provide explicitly that children are to be protected by excluding their testimony. At a minimum, § 16-2316(b) means that the parent must be afforded a fair opportunity to present evidence to refute the charges of neglect — including probative live testimony from the child who is the subject of the charge, unless there is a compelling reason to preclude such testimony. The potential importance to the parent of being able to elicit the child’s live testimony is heightened where, as in the present case, the proof of neglect depends critically on the admission of accusatory statements that the child herself made outside of the courtroom.7 Cf. In re Ko.W., 774 A.2d 296, 306 n. 12 (D.C.2001) (finding it “significant that none of these [hearsay] accusations [by a child] has been tested by cross-examination, which is ‘the greatest legal engine ever invented for the discovery of truth.’ ” (quoting Tyree v. Evans, 728 A.2d 101, 103 (D.C.1999))).8
On the other hand, the “paramount consideration” in neglect proceedings is the best interest of the child. In re T.W., 732 A.2d 254, 258 (D.C.1999). In some neglect proceedings the child’s best interest may conflict with the parent’s interest in calling the child as a witness. Compelling reasons indeed may exist to [916]*916protect the allegedly abused child from having to testify in “the harsh atmosphere of the typical courtroom.” Coy v. Iowa, 487 U.S. 1012, 1022, 108 S.Ct. 2798, 101 L.Ed.2d 857 (1988) (O’Connor, J., concurring). There is substantial empirical support for the proposition that “a significant amount of trauma and upset experienced by the child witness is due to the presence and proximity of the accused [abuser].” Brief for Amicus Curiae American Psychological Ass’n at n. 19, Maryland v. Craig, 497 U.S. 836, 110 S.Ct. 3157, 111 L.Ed.2d 666 (1990) (No. 89-478); see Christine Brannon, The Trauma of Testifying in Court for Child Victims of Sexual Assault v. The Accused’s Right to Confrontation, 18 Law & Psychol. Rev. 439, 440 (1994); Maryland v. Craig, 497 U.S. 836, 855, 110 S.Ct. 3157, 111 L.Ed.2d 666 (1990) (recognizing “the growing body of academic literature documenting the psychological trauma suffered by child abuse victims who must testify in court”). For some children, evidence indicates, the emotional consequences of testifying in open court about their abuse can be dire and long lasting.9 It would be a “perversion” indeed of the purposes of the neglect proceeding to require allegedly abused children to testify “regardless of the trauma resulting to the child therefrom, and regardless of the necessity of such testimony in the resolution of the issues before the court.” Jennifer J., 10 Cal.Rptr.2d at 819. As parens patriae, the court must act “to protect the child’s welfare” and to protect the child from “serious risk of harm.” In re T.W., 732 A.2d 254, 258 (D.C.1999); S.L.E., 677 A.2d at 519. The court’s obligation to protect the child “begins well in advance of any adjudication of neglect,” In re M.D., 758 A.2d 27, 33 (D.C.2000) (quoting J.J.Z., 630 A.2d at 193), and requires the court to “monitor the situation carefully and promptly take steps to protect [the child] if her physical or emotional welfare is endangered.” S.L.E., 677 A.2d at 519 (quoting In re S.C.M., 653 A.2d 398, 406 n. 12 (D.C.1995)). In extreme cases, the danger to the child’s welfare may be so great that it does outweigh the parent’s right to call and confront the child in court. Cf. Craig, 497 U.S. at 853, 110 S.Ct. 3157 (holding that the state’s interest in protecting the complaining witness in a child abuse prosecution sometimes may override the defendant’s Sixth Amendment right of face-to-face confrontation, so as to justify receiving the child’s testimony in a separate room from the defendant over one-way closed circuit television).
In light of the strength of the competing interests at stake, we agree with the approach taken in Jennifer J. and Beasley. That approach is consistent with what we have done in our TPR cases. We hold that the trial court does have power to protect a child from the harmful effects of forced testimony. In most cases the court can exercise that power effectively and appropriately by imposing reasonable conditions and restrictions on the conduct and scope of the child’s examination. When such conditions and restrictions are [917]*917sufficient to safeguard the child’s welfare, there is no warrant for the court to go further and preclude the child’s examination altogether. In the extreme case, however, where a demonstrated risk of serious psychological or emotional harm to the child is not adequately mitigable by other means and substantially outweighs the parent’s need for the child’s testimony, the trial court has discretion to exclude the child as a witness.10 Cf. (William) Johnson v. United States, 683 A.2d 1087, 1099-1100 (D.C.1996) (en banc) (adopting Federal Rule of Evidence 403, which provides that relevant evidence may be excluded only if its probative value is substantially outweighed by the danger of unfair prejudice or other countervailing considerations).
The balancing test that we adopt commits the decision whether to preclude examination of a competent child witness in a neglect proceeding to the trial court’s informed discretion. A sound exercise of that discretion “requires that the trial court’s determination be based upon and drawn from a firm factual foundation.” (James) Johnson v. United States, 398 A.2d 354, 364 (D.C.1979). Conceptually, the test calls for three case-specific factual determinations that must be made to justify excluding the child’s testimony. First, the trial court must make a finding on the record that testifying would create a risk of serious harm to the child. This finding must not be based solely on generic concerns. Rather, the finding of a risk of serious harm must rest upon concrete evidence individualized to the particular child. See id. at 365 (if necessary, trial court should “undertake a specific factual inquiry ... prior to rendering a discretionary decision”). Although expert testimony may not always be required, it will often be the best evidence of such a risk. See Beasley, 840 P.2d at 85.
Second, if the trial court finds that the child is at risk of serious harm from having to testify, the court must consider whether the risk can be alleviated by means short of prohibiting the testimony altogether. A “trial court has inherent authority, unless otherwise specifically precluded, to control the conduct of the proceedings before it, in order to ensure that the proper decorum and appropriate atmosphere are established, that all parties are treated fairly, and that justice is done.” Hicks-Bey v. United States, 649 A.2d 569, 575 (D.C.1994). In neglect proceedings in particular, “[t]he judges of the trial court have broad authority to take whatever steps are necessary to protect the well-being of neglected and abused children.” S.L.E., 677 A.2d at 522. Atrial court should use this power to explore avenues to protect a child witness from the negative effect of testimony. The court could, for example, consider using closed-circuit cameras to receive the child’s testimony out of the physical presence of the parent accused of neglect. See Craig, 497 U.S. at 853-54, 110 S.Ct. 3157; Hicks-Bey, 649 A.2d at 575. The court should also consider whether its “considerable discretion to control cross-examination” (as well as direct examination) would alleviate the risk of harm. Walls v. United States, 773 A.2d 424, 431 (D.C.2001) (quoting Smith v. United States, 392 A.2d 990, 991 (D.C.1978)). “In the exercise of this discretion, the court has wide latitude to ‘impose reasonable limits’ on cross-examination ‘based on concerns about, among other things, harassment, prejudice, confusion of the issues, the witness’ safety, or interrogation [918]*918that is repetitive or only marginally relevant.’” Roundtree v. United States, 581 A.2d 315, 323 (D.C.1990) (emphasis added) (quoting Delaware v. Van Arsdall, 475 U.S. 673, 679, 106 S.Ct. 1431, 89 L.Ed.2d 674 (1986)); cf. T.W., 623 A.2d at 117-18.
Third, after taMng into consideration the risk of harm to the child and the possibility of ameliorative measures, the court must evaluate the probative value of the child’s testimony and the parent’s concomitant need for it. The court should consider the materiality of the anticipated testimony in light of the allegations of neglect, the degree to which (as in this case) those allegations are grounded in the out-of-court statements of the child, the reliability of those statements, and the totality of the evidence. In an appropriate case the court may require the parent to proffer the testimony that he or she hopes to elicit from the child, and the basis for that proffer. The parent’s inability to make a good faith proffer of relevant testimony may justify more careful control over the questioning to protect the child from undue trauma. The court may also explore alternatives to taking testimony from the child that would meet the parent’s needs, such as admitting evidence of out-of-court statements made by the child or stipulations. Finally, so long as the court acts with due regard for the rights of the parent and the creation of an adequate record, we do not rule out the possibility of assessing the importance of the child’s testimony by means of an informal in camera interview.
In the present case, the trial court prohibited appellants from calling Jas.J. and Jam.J. as witnesses without undertaking the independent fact finding and balancing that we now hold was required. In its current state the record does not contain evidentiary support for a finding that the children faced a risk of serious harm if they were called to testify. The court deferred to the assertion by the guardian ad litem that the children would suffer “an emotional impact” from testifying before appellants. This “mere proffer,” however, was “not evidence.” In re R.E.G., 602 A.2d 146, 148 (D.C.1992). “A finding as to a critical disputed fact cannot be made on the basis of an unsubstantiated proffer.” Ko.W., 774 A.2d at 306 n. 13. Furthermore, while it was perfectly plausible that testifying would be at least somewhat traumatic for Jas.J. and Jam.J., the proffer was not specific enough as to the nature and severity of the emotional injury that the children might suffer to enable the court to evaluate the danger.
We emphasize that the responsibility for weighing appellants’ need for the children’s testimony against the risk of harm to the children belonged to the trial court. Although it was the duty of the guardian ad litem to represent the children’s best interests, see D.C.Code § 16-2304(b)(5), the decision whether to permit appellants to call the children as witnesses was not delegable to the guardian ad litem. In electing to “adhere to a uniform policy” of deference to the guardian ad litem, the court erred. See (James) Johnson, 398 A.2d at 363 (“Failure to exercise choice in a situation calling for choice is an abuse of discretion.”).
Assuming that being called to the witness stand would have exposed Jas.J. and Jam.J. to a risk of emotional or psychological harm, that was not the end of the inquiry. We think that the trial court should have given serious consideration on the record to B.A.’s suggestion that the risk of harm could be mitigated by taking the children’s testimony outside appellants’ presence. While this court is not in a position to evaluate that suggestion, it would appear on its face to meet one of the concerns that the guardian ad litem ex[919]*919pressed. Other ameliorative measures, such as strict control by the trial court over the form and content of any questioning of the children, also might have been considered.
Finally, we think that the trial court should have evaluated appellants’ need to examine Jas.J. and Jam.J. Again, while this court is not in a position to assess that need, the government’s heavy reliance on the children’s out of court statements would appear to place a premium on appellants’ ability to confront the children in court. Appellants made no proffer, but their intention clearly was to cross-examine Jas.J. and Jam.J. in the hope of discrediting their accusations of mistreatment. The probative value of live testimony from the children might have been high. Cf. Ko.W., 774 A.2d at 306 n. 12.
In deciding not to allow appellants to examine Jas.J. and Jam.J., the trial court commendably had the children's best interests at heart, and it was compelled to make its decision in an uncharted area without having the benefit of our exposition of the balancing that was required. Although we discern that the trial court exercised its discretion erroneously and thus are constrained to grant relief, we “do[ ] not imply any reflection on the [trial] judge.” Id. at n. 14 (quoting King v. United States, 550 A.2d 348, 353 n. 3 (D.C.1988)). We leave it to that court, in further proceedings on remand, to develop the factual record and determine whether the risk of harm to Jas.J. and Jam.J. from being called to testify would substantially outweigh appellants’ need for their testimony.
We also agree with appellant KC.’s argument that the trial court should not have excluded the testimony of Dr. Hamlin as a sanction for his failure to identify that expert before trial. Although “[t]he decision to impose discovery sanctions is left to the ‘broad discretion’ of the trial court,” Jung v. Jung, 791 A.2d 46, 49 (D.C.2002) (quoting Lyons v. Jordan, 524 A.2d 1199, 1201 (D.C.1987)), in this case B.A. and K.C. did not violate the rules of discovery.
Under Superior Court Civil Rule 26(b)(4)(A)(i), a litigant is obligated to disclose expert witnesses in discovery only when an opposing party has made a request for such disclosure “through interrogatories”:
A party may through interrogatories require any other party to identify each person whom the other party expects to call as an expert witness at trial, to state the subject matter on which the expert is expected to testify, and to state the substance of the facts and opinions to which the expert is expected to testify and a summary of the grounds for each opinion.11
It is undisputed that neither B.A. nor K.C. was ever served with an interrogatory request to identify their experts. While it appears that the parties conducted pretrial discovery informally, it is also undisputed that B.A. and K.C. were not asked even informally to identify their expert witnesses. Nor did any pretrial order issued by the court impose a duty to make such disclosure. Thus, contrary to the trial court’s supposition, appellants violated no rule by not informing the government and [920]*920the guardian ad litem in advance of their intent to call expert witnesses.
Absent any violation by B.A. and K.C. of their discovery obligations, we perceive no basis for excluding the testimony of their experts. The trial court premised its decision on Weiner v. Kneller, 557 A.2d 1306 (D.C.1989). In Weiner, we enumerated the factors that a trial court should consider to evaluate the fairness of admitting expert testimony that was not disclosed in a required Rule 26(b)(4) statement. See id. at 1311-12; see also Dada v. Children’s Nat'l Med. Ctr., 715 A.2d 904, 905 (D.C.1998) (plaintiff failed to comply with scheduling order requiring 26(b)(4) statements); Abell v. Wang, 697 A.2d 796, 799 (D.C. 1997) (same). Weiner’s analysis, however, simply does not apply to a situation in which a party has not been required by either a discovery request or a court order to disclose its experts.
We cannot disagree with the trial court’s concerns about “trial by ambush,” but the government and the guardian ad litem cannot complain about unfair surprise when they failed to take advantage of the opportunity provided by Rule 26(b)(4) to avoid being surprised. Litigants are responsible for using the tools they are given. A party that has no duty to produce a Rule 26(b)(4) statement cannot be penalized for not producing it.
According to KC.’s proffer, his expert, offered to rebut Dr. Herring, would have countered the evidence of a pattern of abuse preceding the March 10, 1999 incident. The trial court noted that “[t]he testimony of Dr. Hamlin seems especially probative of the issue of abuse charged on the part of [K.C.] and could conceivably do a great deal to refute these allegations.” We cannot say with “fair assurance” that the erroneous exclusion of KC.’s expert did not “substantially sway” the outcome. R & G Orthopedic Appliances & Prosthetics, Inc. v. Curtin, 596 A.2d 530, 540 (D.C.1991) (quoting Kotteakos v. United States, 328 U.S. 750, 765, 66 S.Ct. 1239, 90 L.Ed. 1557 (1946)). K.C. is entitled, therefore, to relief.
III.
Having concluded that the record in its current state does not support the rulings preventing B.A. from calling Jas.J. and Jam.J. as witnesses and excluding the testimony of K.C.’s expert and that these errors were not harmless, we vacate the findings of neglect and remand the case to the Superior Court for further proceedings — i.e., a new, or at least a re-opened, fact finding hearing — in accordance with this opinion.
Reversed and remanded.